Volume 91 Number 6 Published monthly by the American Psychological Association
December 2006
ISSN 0022-3514
Journal of
Personality and Social Psychology ATTITUDES AND SOCIAL COGNITION
Charles M. Judd, Editor Dacher Keltner, Associate Editor Anne Maass, Associate Editor Bernd Wittenbrink, Associate Editor Vincent Yzerbyt, Associate Editor INTERPERSONAL RELATIONS AND GROUP PROCESSES
John F. Dovidio, Editor Daphne Blunt Bugental, Associate Editor Beverley Fehr, Associate Editor Jacques-Philippe Leyens, Associate Editor Antony Manstead, Associate Editor Cynthia L. Pickett, Associate Editor Jeffry A. Simpson, Associate Editor Scott Tindale, Associate Editor Jacquie D. Vorauer, Associate Editor PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES
www.apa.org/journals/psp
Charles S. Carver, Editor Tim Kasser, Associate Editor Mario Mikulincer, Associate Editor Eva M. Pomerantz, Associate Editor Richard W. Robins, Associate Editor Gerard Saucier, Associate Editor Thomas A. Widiger, Associate Editor
The Journal of Personality and Social Psychology publishes original papers in all areas of personality and social psychology. It emphasizes empirical reports but may include specialized theoretical, methodological, and review papers. The journal is divided into three independently edited sections: f ATTITUDES AND SOCIAL COGNITION addresses those domains of social behavior in which cognition plays a major role, including the interface of cognition with overt behavior, affect, and motivation. Among topics covered are the formation, change, and utilization of attitudes, attributions, and stereotypes, person memory, self-regulation, and the origins and consequences of moods and emotions insofar as these interact with cognition. Of interest also is the influence of cognition and its various interfaces on significant social phenomena such as persuasion, communication, prejudice, social development, and cultural trends. f INTERPERSONAL RELATIONS AND GROUP PROCESSES focuses on psychological and structural features of interaction in dyads and groups. Appropriate to this section are papers on the nature and dynamics of interactions and social relationships, including interpersonal attraction, communication, emotion, and relationship development, and on group and organizational processes such as social influence, group decision making and task performance, intergroup relations, and aggression, prosocial behavior and other types of social behavior. f PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES publishes research on all aspects of personality psychology. It includes studies of individual differences and basic processes in behavior, emotions, coping, health, motivation, and other phenomena that reflect personality. Articles in areas such as personality structure, personality development, and personality assessment are also appropriate to this section of the journal, as are studies of the interplay of culture and personality and manifestations of personality in everyday behavior. Manuscripts: Submit manuscripts to the appropriate section editor according to the above definitions and according to the Instructions to Authors. Section editors reserve the right to redirect papers among themselves as appropriate unless an author specifically requests otherwise. Rejection by one section editor is considered rejection by all; therefore a manuscript rejected by one section editor should not be submitted to another. The opinions and statements published are the responsibility of the authors, and such opinions and statements do not necessarily represent the policies of APA or the views of the editors. Section editors’ addresses appear below:
ATTITUDES AND SOCIAL COGNITION Charles M. Judd, Editor c/o Laurie Hawkins Department of Psychology University of Colorado UCB 345 Boulder, CO 80309
INTERPERSONAL RELATIONS AND GROUP PROCESSES John F. Dovidio, Editor Department of Psychology University of Connecticut 406 Babbidge Road Storrs, CT 06269-1020
PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES Charles S. Carver, Editor ATTN: JPSP: PPID Department of Psychology University of Miami P.O. Box 248185 Coral Gables, FL 33124-0751 Change of Address: Send change of address notice and a recent mailing label to the attention of the Subscriptions Department, American Psychological Association, 30 days prior to the actual change of address. APA will not replace undelivered copies resulting from address changes;
journals will be forwarded only if subscribers notify the local post office in writing that they will guarantee periodicals forwarding postage. Electronic access: APA members who subscribe to this journal have automatic access to a 3-year file of the journal in the PsycARTICLES姞 full-text database. See http://members.apa.org/access. Reprints: Authors may order reprints of their articles from the printer when they receive proofs. Single Issues, Back Issues, and Back Volumes: For information regarding back issues or back volumes write to Order Department, American Psychological Association, 750 First Street, NE, Washington, DC 20002-4242. Microform Editions: For information regarding microform editions, write to University Microfilms, Ann Arbor, MI 48106. Copyright and Permission: Those who wish to reuse APAcopyrighted material in a non-APA publication must secure from APA and the author of reproduced material written permission to reproduce a journal article in full or journal text of more than 500 words. APA normally grants permission contingent upon like permission of the author, inclusion of the APA copyright notice on the first page of reproduced material, and payment of a fee of $20 per page. Permission from APA and fees are waived for those who wish to reproduce a single table or figure from a journal for use in a print product, provided the author’s permission is obtained and full credit is given to APA as copyright holder and to the author through a complete citation. (Requesters requiring written permission for commercial use of a single table or figure will be assessed a $25 service fee.) Permission and fees are waived for authors who wish to reproduce their own material for personal use; fees only are waived for authors who wish to use more than a single table or figure of their own material commercially (but for use in edited books, fees are waived for the author only if serving as the book editor). Permission and fees are waived for the photocopying of isolated journal articles for nonprofit classroom or library reserve use by instructors and educational institutions. A permission fee may be charged to the requester if students are charged for the material, multiple articles are copied, or large-scale copying is involved (e.g., for course packs). Access services may use unedited abstracts without the permission of APA or the author. Libraries are permitted to photocopy beyond the limits of U.S. copyright law: (1) post-1977 articles, provided the per-copy fee in the code for this journal (0022-3514/06/ $12.00) is paid through the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923; (2) pre-1978 articles, provided that the per-copy fee stated in the Publishers’ Fee List is paid through the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923. Address requests for reprint permission to the Permissions Office, American Psychological Association, 750 First Street, NE, Washington, DC 20002-4242. APA Journal Staff: Susan J. A. Harris, Senior Director, Journals Program; Skip Maier, Director, Journal Services; Paige W. Jackson, Director, Editorial Services; Becky Shaw, Account Manager; Jodi Ashcraft, Advertising Sales Manager.
Journal of Personality and Social Psychology (ISSN 0022-3514) is published monthly in two volumes per year by the American Psychological Association, 750 First Street, NE, Washington, DC 20002-4242. Subscriptions are available on a calendar year basis only (January through December). The 2007 rates follow: Nonmember Individual: $442 Domestic, $488 Foreign, $515 Air Mail. Institutional: $1,349 Domestic, $1,443 Foreign, $1,470 Air Mail. APA Member: $208. Write to Subscriptions Department, American Psychological Association, 750 First Street, NE, Washington, DC 20002-4242. Printed in the U.S.A. Periodicals postage paid at Washington, DC, and at additional mailing offices. POSTMASTER: Send address changes to Journal of Personality and Social Psychology, 750 First Street, NE, Washington, DC 20002-4242.
The paper in this journal meets or exceeds EPA guidelines for recycled paper. Since 1986, this journal has been printed on acid-free paper.
NEW RELEASES from the American Psychological Association
Psychological Treatment of Obsessive-Compulsive Disorder Fundamentals and Beyond
Edited by Martin M. Antony, Christine Purdon, and Laura J. Summerfeldt 2007. 328 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-484-1 ISBN-13: 978-1-59147-484-5 Item # 4317123
Termination in Psychotherapy A Psychodynamic Model of Processes and Outcomes
Anthony S. Joyce, William E. Piper, John S. Ogrodniczuk, and Robert H. Klein 2007. 224 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-730-1 ISBN-13: 978-1-59147-730-3 Item # 4316086
Why Aren’t More Women in Science? Top Gender Researchers Debate the Evidence
Edited by Stephen J. Ceci and Wendy M. Williams 2007. 248 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-485-X ISBN-13: 978-1-59147-485-2 Item # 4316085
Psychology and the Department of Veterans Affairs A Historical Analysis of Training, Research, Practice, and Advocacy
Rodney Baker and Wade Pickren 2007. 200 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-453-1 ISBN-13: 978-1-59147-453-1 Item # 4316084
Sensation Seeking and Risky Behavior Marvin Zuckerman 2007. 326 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-738-7 ISBN-13: 978-1-59147-738-9 Item # 4317124
Career Paths in Psychology Where Your Degree Can Take You Second Edition
Edited by Robert J. Sternberg 2007. 443 pages. Paperback. List: $29.95 APA Member/Affiliate: $24.95 ISBN 1-59147-732-8 ISBN-13: 978-1-59147-732-7 Item # 4313008
AD0485
To O r d e r : 8 0 0 - 374 -2721 • w w w. a p a .or g / b o o ks
Journal of
Personality Social Psychology and
www.apa.org/journals/psp December 2006 VOLUME 91 NUMBER 6
Copyright © 2006 by the American Psychological Association
Attitudes and Social Cognition 995
Understanding Implicit and Explicit Attitude Change: A Systems of Reasoning Analysis Robert J. Rydell and Allen R. McConnell
1009
Unpriming: The Deactivation of Thoughts Through Expression Betsy Sparrow and Daniel M. Wegner
1020
Framing Discrimination: Effects of Inclusion Versus Exclusion Mind-Sets on Stereotypic Judgments Kurt Hugenberg, Galen V. Bodenhausen, and Melissa McLain
Interpersonal Relations and Group Processes 1032
The Relative Deprivation–Gratification Continuum and the Attitudes of South Africans Toward Immigrants: A Test of the V-Curve Hypothesis Michae¨l Dambrun, Donald M. Taylor, David A. McDonald, Jonathan Crush, and Alain Me´ot
1045
Relationship Perceptions and Persistence: Do Fluctuations in Perceived Partner Commitment Undermine Dating Relationships? Ximena B. Arriaga, Jason T. Reed, Wind Goodfriend, and Christopher R. Agnew
1066
Perceiving Outgroup Members as Unresponsive: Implications for Approach-Related Emotions, Intentions, and Behavior David A. Butz and E. Ashby Plant
1080
Group Decision Making in Hidden Profile Situations: Dissent as a Facilitator for Decision Quality Stefan Schulz-Hardt, Felix C. Brodbeck, Andreas Mojzisch, Rudolf Kerschreiter, and Dieter Frey
1094
Knowing Your Place: Self-Perceptions of Status in Face-to-Face Groups Cameron Anderson, Sanjay Srivastava, Jennifer S. Beer, Sandra E. Spataro, and Jennifer A. Chatman
Personality Processes and Individual Differences 1111
Emotion Without a Word: Shame and Guilt Among Rara´muri Indians and Rural Javanese Seger M. Breugelmans and Ype H. Poortinga
1123
Helping One’s Way to the Top: Self-Monitors Achieve Status by Helping Others and Knowing Who Helps Whom Francis J. Flynn, Ray E. Reagans, Emily T. Amanatullah, and Daniel R. Ames
(contents continue)
1138
Higher-Order Factors of the Big Five in a Multi-Informant Sample Colin G. DeYoung
1152
Love, Work, and Changes in Extraversion and Neuroticism Over Time Christie Napa Scollon and Ed Diener
Other 1166 1031 ii
Acknowledgment E-Mail Notification of Your Latest Issue Online! Subscription Order Form
This issue completes Volume 91 and contains the author index to the volume.
ii
1138
Higher-Order Factors of the Big Five in a Multi-Informant Sample Colin G. DeYoung
1152
Love, Work, and Changes in Extraversion and Neuroticism Over Time Christie Napa Scollon and Ed Diener
Other 1166 1031 ii
Acknowledgment E-Mail Notification of Your Latest Issue Online! Subscription Order Form
This issue completes Volume 91 and contains the author index to the volume.
ii
ATTITUDES AND SOCIAL COGNITION CHARLES M. JUDD, Editor University of Colorado at Boulder ASSOCIATE EDITORS DACHER KELTNER University of California, Berkeley ANNE MAASS Universita` di Padova, Padova, Italy BERND WITTENBRINK University of Chicago VINCENT YZERBYT Catholic University of Louvain, Louvain-la-Neuve, Belgium CONSULTING EDITORS ICEK AJZEN University of Massachusetts
ALICE H. EAGLY Northwestern University
NIRA LIBERMAN Tel Aviv University, Tel Aviv, Israel
LINDA SKITKA University of Illinois at Chicago
NICHOLAS EPLEY University of Chicago
DIANE M. MACKIE University of California, Santa Barbara
JOHN SKOWRONSKI Northern Illinois University
RUSSELL H. FAZIO Ohio State University
NEIL MACRAE Dartmouth College
ELIOT R. SMITH Indiana University Bloomington
LISA FELDMAN BARRETT Boston College
TONY MANSTEAD Cardiff University, Cardiff, Wales
SUSAN T. FISKE Princeton University
THOMAS MUSSWEILER Universita¨t Ko¨ln, Cologne, Germany
DIEDERIK STAPEL University of Groningen, Groningen, the Netherlands
BARBARA L. FREDRICKSON University of Michigan
JAMES M. OLSON University of Western Ontario, London, Ontario, Canada
WENDI GARDNER Northwestern University
MAHZARIN BANAJI Harvard University
BERNADETTE M. PARK University of Colorado at Boulder
DANIEL GILBERT Harvard University
MONICA BIERNAT University of Kansas
RICHARD E. PETTY Ohio State University
THOMAS GILOVICH Cornell University
IRENE V. BLAIR University of Colorado at Boulder
NEAL J. ROESE University of Illinois at Urbana– Champaign
ANTHONY G. GREENWALD University of Washington
GALEN V. BODENHAUSEN Northwestern University
DAVID L. HAMILTON University of California, Santa Barbara
MARKUS BRAUER LAPSCO, Universite´ Blaise Pascal Clermont-Ferrand, France
EDWARD R. HIRT Indiana University Bloomington
MARILYNN B. BREWER Ohio State University
TIFFANY ITO University of Colorado at Boulder
JOHN T. CACIOPPO University of Chicago
YOSHIHISA KASHIMA University of Melbourne, Victoria, Australia
OLIVIER CORNEILLE Catholic University of Louvain, Louvain-la-Neuve, Belgium
KARLE CHRISTOPHE KLAUER Albrecht-Ludwigs-Universita¨t Freiburg, Freiburg, Germany
PATRICIA DEVINE University of Wisconsin—Madison AP DIJKSTERHUIS University of Amsterdam, Amsterdam, the Netherlands DAVID DUNNING Cornell University
MYRON ROTHBART University of Oregon LAURIE RUDMAN Rutgers, The State University of New Jersey MARK SCHALLER University of British Columbia, Vancouver, British Columbia, Canada TONI SCHMADER University of Arizona NORBERT SCHWARZ University of Michigan
ARIE W. KRUGLANSKI University of Maryland
GU¨N R. SEMIN Free University, Amsterdam, the Netherlands
ALAN LAMBERT Washington University in St. Louis
JEFFREY W. SHERMAN University of California, Davis
JENNIFER LERNER Carnegie Mellon University
STEVEN J. SHERMAN Indiana University Bloomington
FRITZ STRACK Universita¨t Wu¨rzburg, Wu¨rzburg, Germany ABRAHAM TESSER University of Georgia YAACOV TROPE New York University THERESA K. VESCIO Pennsylvania State University WILLIAM VON HIPPEL University of New South Wales, Sydney, Australia DUANE T. WEGENER Purdue University DANIEL M. WEGNER Harvard University DIRK WENTURA Saarland University, Saarbru¨cken, Germany DANIEL WIGBOLDUS Radboud University Nijmegen, Nijmegen, the Netherlands TIMOTHY D. WILSON University of Virginia PIOTR WINKIELMEN University of California, San Diego MARK P. ZANNA University of Waterloo, Waterloo, Ontario, Canada
ASSISTANT TO THE EDITOR—LAURIE HAWKINS
INTERPERSONAL RELATIONS AND GROUP PROCESSES JOHN F. DOVIDIO, Editor University of Connecticut ASSOCIATE EDITORS DAPHNE BLUNT BUGENTAL University of California, Santa Barbara
ARTHUR ARON State University of New York at Stony Brook
RUPERT BROWN The University of Kent at Canterbury, Canterbury, England
BEVERLEY FEHR University of Winnipeg, Winnipeg, Manitoba, Canada
XIMENA ARRIAGA Purdue University
LORNE CAMPBELL University of Western Ontario, London, Ontario, Canada
JACQUES-PHILIPPE LEYENS Catholic University of Louvain, Louvain-la-Neuve, Belgium ANTONY MANSTEAD Cardiff University, Cardiff, United Kingdom
WINTON W. T. AU The Chinese University of Hong Kong, Shatin, Hong Kong MARK BALDWIN McGill University, Montreal, Quebec, Canada
CYNTHIA L. PICKETT University of California, Davis
KIM BARTHOLOMEW Simon Fraser University, Burnaby, British Columbia, Canada
JEFFRY A. SIMPSON University of Minnesota, Twin Cities Campus
C. DANIEL BATSON University of Kansas
SCOTT TINDALE Loyola University Chicago JACQUIE D. VORAUER University of Manitoba, Winnipeg, Manitoba, Canada CONSULTING EDITORS DOMINIC ABRAMS University of Kent at Canterbury, Canterbury, England CHRIS AGNEW Purdue University
B. ANNE BETTENCOURT University of Missouri—Columbia GERD BOHNER Universita¨t Bielefeld, Bielefeld, Germany NIALL BOLGER Columbia University NYLA R. BRANSCOMBE University of Kansas JONATHON D. BROWN University of Washington
SERENA CHEN University of California, Berkeley MARGARET CLARK Yale University CARSTEN DE DREU University of Amsterdam, Amsterdam, the Netherlands STE´PHANIE DEMOULIN Catholic University of Louvain Louvain-la-Neuve, Belgium, and Belgan National Fund for Scientific Research, Brussels, Belgium
KLAUS FIEDLER University of Heidelberg, Heidelberg, Germany GARTH FLETCHER University of Canterbury, Christchurch, New Zealand SHELLY GABLE University of California, Los Angeles LOWELL GAERTNER University of Tennessee, Knoxville SAMUEL L. GAERTNER University of Delaware ADAM GALINSKY Northwestern University PETER GLICK Lawrence University STEPHANIE A. GOODWIN Purdue University
DAVID DESTENO Northeastern University
MARTIE G. HASSELTON University of California, Los Angeles
STEVE DRIGOTAS Johns Hopkins University
S. ALEXANDER HASLAM University of Exeter, Exeter, United Kingdom
ELISSA S. EPEL University of California, San Francisco VICTORIA ESSES University of Western Ontario, London, Ontario, Canada
(editors continue)
VERLIN HINSZ North Dakota State University GORDON HODSON Brock University, St. Catherine’s, Ontario, Canada
MICHAEL A. HOGG University of Queensland, Brisbane, Australia
LAURA J. KRAY University of California, Berkeley
ANDREA B. HOLLINGSHEAD University of Southern California JOHN G. HOLMES University of Waterloo, Waterloo, Ontario, Canada RICK H. HOYLE University of Kentucky
JAMES R. LARSON JR. University of Illinois at Chicago COLIN WAYNE LEACH University of Sussex, Sussex, United Kingdom JOHN LEVINE University of Pittsburgh JOHN E. LYDON McGill University, Montreal, Quebec, Canada
JOLANDA JETTEN University of Exeter, Exeter, United Kingdom
JON K. MANER Florida State University
JAMES D. JOHNSON University of North Carolina at Wilmington TATSUYA KAMEDA Hokkaido University, Sapporo, Japan BENJAMIN R. KARNEY RAND Corporation, Santa Monica, California YOSHI KASHIMA University of Melbourne, Victoria, Australia
BRENDA MAJOR University of California, Santa Barbara CRAIG MCGARTY Australian National University, Canberra, Australia WENDY BERRY MENDES Harvard University RICHARD MORELAND University of Pittsburgh
DEBORAH A. KASHY Michigan State University
SABINE OTTEN University of Gro¨ningen, Gro¨ningen, the Netherlands CRAIG D. PARKS Washington State University LOUIS A. PENNER Wayne State University PAULA PIETROMONACO University of Massachusetts at Amherst
CHRISTINE SMITH Grand Valley State University HEATHER J. SMITH Sonoma State University RUSSELL SPEARS Cardiff University, Cardiff, Wales CHARLES STANGOR University of Maryland GARY L. STASSER Miami University—Ohio
TOM POSTMES University of Exeter, Exeter, United Kingdom
WALTER STEPHAN New Mexico State University
FELICIA PRATTO University of Connecticut
WILLIAM B. SWANN JR. University of Texas at Austin
HARRY T. REIS University of Rochester
JANET SWIM Pennsylvania State University
W. STEVEN RHOLES Texas A&M University
LEIGH L. THOMPSON Northwestern University
JENNIFER A. RICHESON Northwestern University
TOM TYLER New York University
MARK SCHALLER University of British Columbia, Vancouver, British Columbia, Canada
JEROEN VAES University of Padova, Padova, Italy
BRIAN MULLEN KERRY KAWAKAMI University of Kent at Canterbury, York University, Toronto, Ontario, Canada Canterbury, England JANICE R. KELLY AME´LIE MUMMENDEY Purdue University Friedrich-Schiller-Universita¨t, Jena, DACHER KELTNER Jena, Germany University of California, Berkeley MARK MURAVEN DAVID A. KENNY University at Albany, State University University of Connecticut of New York
DAVID A. SCHROEDER University of Arkansas
KEES VAN DEN BOS University of Utrecht, Utrecht, the Netherlands
CONSTANTINE SEDIKIDES University of Southampton, Southampton, England
PAUL A. M. VAN LANGE Free University, Amsterdam, Amsterdam, the Netherlands
PHILLIP R. SHAVER University of California, Davis
LAURIE R. WEINGART Carnegie Mellon University
J. NICOLE SHELTON Princeton University
GWEN M. WITTENBAUM Michigan State University
DOUGLAS T. KENRICK Arizona State University
SANDRA L. MURRAY State University of New York at Buffalo
MARGARET SHIH University of Michigan
NORBERT L. KERR Michigan State University
STACEY SINCLAIR LISA A. NEFF University of Virginia University of Toledo ASSISTANT TO THE EDITOR—CHRISTINE KELLY
WENDY L. WOOD Texas A&M University MICHAEL ZA´RATE University of Texas at El Paso
PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES CHARLES S. CARVER, Editor University of Miami ASSOCIATE EDITORS TIM KASSER Knox College
GEORGE A. BONANNO Teachers College, Columbia University
AVSHALOM CASPI MARIO MIKULINCER Bar-Ilan University, Ramat-Gan, Israel King’s College, London EDWARD C. CHANG EVA M. POMERANTZ University of Michigan University of Illinois at Urbana– Champaign RICHARD W. ROBINS University of California, Davis GERARD SAUCIER University of Oregon THOMAS A. WIDIGER University of Kentucky
SERENA CHEN University of California, Berkeley A. TIMOTHY CHURCH Washington State University JAMES COAN University of Wisconsin—Madison M. LYNNE COOPER University of Missouri—Columbia
EDDIE HARMON-JONES Texas A&M University
DANIEL W. RUSSELL Iowa State University
TODD HEATHERTON Dartmouth College
OLIVER C. SCHULTHEISS University of Michigan
JUTTA HECKHAUSEN University of California, Irvine
SUZANNE C. SEGERSTROM University of Kentucky
STEVEN J. HEINE University of British Columbia, Vancouver, British Columbia, Canada
KENNON M. SHELDON University of Missouri—Columbia
RICHARD KOESTNER McGill University Montreal, Quebec, Canada
C. R. SNYDER University of Kansas SANJAY SRIVASTAVA University of Oregon
DAVID LUBINSKI Vanderbilt University
TIMOTHY STRAUMAN Duke University
MICHAEL EID University of Geneva, Geneva, Switzerland
RICHARD E. LUCAS Michigan State University
MICHAEL J. STRUBE Washington University
ROBERT R. MCCRAE National Institute on Aging, Baltimore
JERRY SULS University of Iowa
ANDREW J. ELLIOT University of Rochester
WENDY BERRY MENDES Harvard University
WILLIAM B. SWANN JR. University of Texas at Austin
LISA FELDMAN BARRETT Boston College
RODOLFO MENDOZA-DENTON University of California, Berkeley
HOWARD TENNEN University of Connecticut Health Center
WILLIAM FLEESON Wake Forest University
DANIEL K. MROCZEK Fordham University
MICHAEL C. ASHTON Brock University, St. Catherines, Ontario, Canada
SUZANNE THOMPSON Pomona College
R. CHRIS FRALEY University of Illinois at Chicago
STEPHEN A. PETRILL Pennsylvania State University
OZLEM AYDUK University of California, Berkeley
ANTONIO L. FREITAS State University of New York at Stony Brook
RALPH L. PIEDMONT Loyola College in Maryland
ROBERT J. VALLERAND Universite´ du Que´bec a` Montre´al Montreal, Quebec, Canada
CONSULTING EDITORS STEPHAN A. AHADI American Institutes for Research, Washington, DC JAMIE ARNDT University of Missouri—Columbia JENS B. ASENDORPF Humboldt-Universita¨t Berlin Berlin, Germany
E. ASHBY PLANT Florida State University
ROY F. BAUMEISTER Florida State University VERO´NICA BENET-MARTI´NEZ University of California, Riverside
DAVID C. FUNDER University of California, Riverside STEVEN W. GANGESTAD University of New Mexico
BRENT ROBERTS University of Illinois at Urbana–Champaign
APRIL L. BLESKE-RECHEK University of Wisconsin—Eau Claire
CAROL L. GOHM University of Mississippi
MICHAEL D. ROBINSON North Dakota State University
ASSISTANT TO THE EDITOR—BARBARA ADEWUSI
KATHLEEN D. VOHS University of Minnesota DAVID WATSON University of Iowa BARBARA WOIKE Columbia University REX A. WRIGHT University of Alabama at Birmingham
Volume 91, Numbers 1–6 July–December 2006
Journal of
Personality and Social Psychology ATTITUDES AND SOCIAL COGNITION
Charles M. Judd, Editor Dacher Keltner, Associate Editor Anne Maass, Associate Editor Bernd Wittenbrink, Associate Editor Vincent Yzerbyt, Associate Editor INTERPERSONAL RELATIONS AND GROUP PROCESSES
John F. Dovidio, Editor Daphne Blunt Bugental, Associate Editor Beverley Fehr, Associate Editor Jacques-Phillippe Leyens, Associate Editor Antony Manstead, Associate Editor Cynthia L. Pickett, Associate Editor Jeffry A. Simpson, Associate Editor Scott Tindale, Associate Editor Jacquie D. Vorauer, Associate Editor PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES
Charles S. Carver, Editor Tim Kasser, Associate Editor Mario Mikulincer, Associate Editor Eva M. Pomerantz, Associate Editor Richard W. Robins, Associate Editor Gerard Saucier, Associate Editor Thomas A. Widiger, Associate Editor ISSN 0022-3514 Published monthly by the American Psychological Association 750 First Street, NE, Washington, DC 20002-4242 Copyright © 2006 by the American Psychological Association
ATTITUDES AND SOCIAL COGNITION Editor Charles M. Judd, University of Colorado at Boulder
Alan Lambert, Washington University in St. Louis Jennifer Lerner, Carnegie Mellon University Nira Liberman, Tel Aviv University, Tel Aviv, Israel Diane M. Mackie, University of California, Santa Barbara Neil Macrae, Dartmouth College Tony Manstead, Cardiff University, Cardiff, Wales Thomas Mussweiler, Universita¨t Ko¨ln, Cologne, Germany James M. Olson, University of Western Ontario, London, Ontario, Canada Bernadette M. Park, University of Colorado at Boulder Richard E. Petty, Ohio State University Neal J. Roese, University of Illinois at Urbana–Champaign Myron Rothbart, University of Oregon Laurie Rudman, Rutgers, The State University of New Jersey Mark Schaller, University of British Columbia, Vancouver, British Columbia, Canada Toni Schmader, University of Arizona Norbert Schwarz, University of Michigan Gu¨n R. Semin, Free University, Amsterdam, the Netherlands Jeffrey W. Sherman, University of California, Davis Steven J. Sherman, Indiana University Bloomington Linda Skitka, University of Illinois at Chicago John Skowronski, Northern Illinois University Eliot R. Smith, Indiana University Bloomington Diederik Stapel, University of Groningen, Groningen, the Netherlands Fritz Strack, Universita¨t Wu¨rzburg, Wu¨rzburg, Germany Abraham Tesser, University of Georgia Yaacov Trope, New York University Theresa K. Vescio, Pennsylvania State University William Von Hippel, University of New South Wales, Sydney, Australia Duane T. Wegener, Purdue University Daniel M. Wegner, Harvard University Dirk Wentura, Saarland University, Saarbru¨cken, Germany Daniel Wigboldus, Radboud University Nijmegen, Nijmegen, the Netherlands Timothy D. Wilson, University of Virginia Piotr Winkielmen, University of California, San Diego Mark P. Zanna, University of Waterloo, Waterloo, Ontario, Canada Assistant to the Editor—Laurie Hawkins
Associate Editors Dacher Keltner, University of California, Berkeley Anne Maass, Universita` di Padova, Padova, Italy Bernd Wittenbrink, University of Chicago Vincent Yzerbyt, Catholic University of Louvain, Louvain-la-Neuve, Belgium Consulting Editors Icek Ajzen, University of Massachusetts Mahzarin Banaji, Harvard University Monica Biernat, University of Kansas Irene V. Blair, University of Colorado at Boulder Galen V. Bodenhausen, Northwestern University Markus Brauer, LAPSCO, Universite´ Blaise Pascal Clermont-Ferrand, France Marilynn B. Brewer, Ohio State University John T. Cacioppo, University of Chicago Olivier Corneille, Catholic University of Louvain, Louvain-la-Neuve, Belgium Patricia Devine, University of Wisconsin—Madison Ap Dijksterhuis, University of Amsterdam, Amsterdam, the Netherlands David Dunning, Cornell University Alice H. Eagly, Northwestern University Nicholas Epley, University of Chicago Russell H. Fazio, Ohio State University Lisa Feldman Barrett, Boston College Susan T. Fiske, Princeton University Barbara L. Fredrickson, University of Michigan Wendi Gardner, Northwestern University Daniel Gilbert, Harvard University Thomas Gilovich, Cornell University Anthony G. Greenwald, University of Washington David L. Hamilton, University of California, Santa Barbara Edward R. Hirt, Indiana University Bloomington Tiffany Ito, University of Colorado at Boulder Yoshihisa Kashima, University of Melbourne, Victoria, Australia Karle Christophe Klauer, Albrecht-Ludwigs-Universita¨t Freiburg, Freiburg, Germany Arie W. Kruglanski, University of Maryland
INTERPERSONAL RELATIONS AND GROUP PROCESSES Editor John F. Dovidio, University of Connecticut
Ste´phanie Demoulin, Catholic University of Louvian, Louvain-la-Neuve, Belgium, and Belgan National Fund for Scientific Research, Brussels, Belgium David DeSteno, Northeastern University Steve Drigotas, Johns Hopkins University Muriel Dumont, Catholic University of Louvain, Louvain-la-Neuve, Belgium Elissa S. Epel, University of California, San Francisco Victoria Esses, University of Western Ontario, London, Ontario, Canada Klaus Fiedler, University of Heidelberg, Heidelberg, Germany Garth O. Fletcher, University of Canterbury, Christchurch, New Zealand Shelly Gable, University of California, Los Angeles Lowell Gaertner, University of Tennessee, Knoxville Samuel L. Gaertner, University of Delaware Adam D. Galinsky, Northwestern University Peter Glick, Lawrence University Stephanie A. Goodwin, Purdue University Martie G. Hasselton, University of California, Los Angeles S. Alexander Haslam, University of Exeter, Exeter, United Kingdom Verlin Hinsz, North Dakota State University Gordon Hodson, Brock University, St. Catherine’s, Ontario, Canada Michael A. Hogg, University of Queensland, Brisbane, Australia Andrea B. Hollingshead, University of Southern California John G. Holmes, University of Waterloo, Waterloo, Ontario, Canada Rick H. Hoyle, University of Kentucky Jolanda Jetten, University of Exeter, Exeter, United Kingdom James D. Johnson, University of North Carolina at Wilmington Tatsuya Kameda, Hokkaido University, Sapporo, Japan Benjamin R. Karney, RAND Corporation, Santa Monica, California Yoshi Kashima, University of Melbourne, Victoria, Australia Deborah A. Kashy, Michigan State University Kerry Kawakami, York University, Toronto, Ontario, Canada Janice R. Kelly, Purdue University Dacher Keltner, University of California, Berkeley David A. Kenny, University of Connecticut
Associate Editors Daphne Blunt Bugental, University of California, Santa Barbara Beverley Fehr, University of Winnipeg, Winnipeg, Manitoba, Canada Jacques-Philippe Leyens, Catholic University of Louvain, Louvain-la-Neuve, Belgium Antony Manstead, Cardiff University, Cardiff, United Kingdom Cynthia L. Pickett, University of California, Davis Jeffry A. Simpson, University of Minnesota, Twin Cities Campus Scott Tindale, Loyola University Chicago Jacquie D. Vorauer, University of Manitoba, Winnipeg, Manitoba, Canada Consulting Editors Dominic Abrams, University of Kent at Canterbury, Canterbury, England Chris Agnew, Purdue University Arthur Aron, State University of New York at Stony Brook Ximena Arriaga, Purdue University Winton W. T. Au, The Chinese University of Hong Kong, Shatin, Hong Kong Mark Baldwin, McGill University, Montreal, Quebec, Canada Kim Bartholomew, Simon Fraser University, Burnaby, British Columbia, Canada C. Daniel Batson, University of Kansas B. Anne Bettencourt, University of Missouri—Columbia Gerd Bohner, Universita¨t Bielefeld, Bielefeld, Germany Nyla R. Branscombe, University of Kansas Jonathon D. Brown, University of Washington Rupert Brown, The University of Kent at Canterbury, Canterbury, England Lorne Campbell, University of Western Ontario, London, Ontario, Canada Serena Chen, University of California, Berkeley Margaret Clark, Yale University Carsten de Dreu, University of Amsterdam, Amsterdam, the Netherlands x
Douglas T. Kenrick, Arizona State University Norbert L. Kerr, Michigan State University Laura J. Kray, University of California, Berkeley James R. Larson Jr., University of Illinois at Chicago Colin Wayne Leach, University of Sussex, Sussex, United Kingdom John Levine, University of Pittsburgh John E. Lydon, McGill University, Montreal, Quebec, Canada Brenda Major, University of California, Santa Barbara Jon K. Maner, Florida State University Craig McGarty, Australian National University, Canberra, Australia Wendy Berry Mendes, Harvard University Richard Moreland, University of Pittsburgh Brian Mullen, University of Kent at Canterbury, Canterbury, England Ame´lie Mummendey, Friedrich-Schiller-Universita¨t, Jena, Jena, Germany Mark Muraven, University at Albany, State University of New York Sandra L. Murray, State University of New York at Buffalo Lisa A. Neff, University of Toledo Sabine Otten, University of Gro¨ningen, Gro¨ningen, the Netherlands Craig D. Parks, Washington State University Louis A. Penner, Wayne State University Paula Pietromonaco, University of Massachusetts at Amherst Tom Postmes, University of Exeter, Exeter, United Kingdom Felicia Pratto, University of Connecticut Harry T. Reis, University of Rochester W. Steven Rholes, Texas A&M University Jennifer A. Richeson, Northwestern University
Mark Schaller, University of British Columbia, Vancouver, British Columbia, Canada David A. Schroeder, University of Arkansas Constantine Sedikides, University of Southampton, Southampton, England Phillip R. Shaver, University of California, Davis J. Nicole Shelton, Princeton University Margaret Shih, University of Michigan Stacey Sinclair, University of Virginia Christine Smith, Grand Valley State University Heather J. Smith, Sonoma State University Russell Spears, Cardiff University, Cardiff, Wales Charles Stangor, University of Maryland Gary L. Stasser, Miami University—Ohio Walter Stephan, New Mexico State University William B. Swann Jr., University of Texas at Austin Janet Swim, Pennsylvania State University Leigh L. Thompson, Northwestern University Tom Tyler, New York University Jeroen Vaes, University of Padova, Padova, Italy Kees van den Bos, University of Utrecht, Utrecht, the Netherlands Paul A. M. van Lange, Free University, Amsterdam, Amsterdam, the Netherlands Laurie R. Weingart, Carnegie Mellon University Gwen M. Wittenbaum, Michigan State University Wendy L. Wood, Texas A&M University Michael Za´rate, University of Texas at El Paso Assistant to the Editor—Christine Kelly
PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES Editor Charles S. Carver, University of Miami
Todd Heatherton, Dartmouth College Jutta Heckhausen, University of California, Irvine Steven J. Heine, University of British Columbia, Vancouver, British Columbia, Canada Richard Koestner, McGill University, Montreal, Quebec, Canada David Lubinski, Vanderbilt University Richard E. Lucas, Michigan State University Robert R. McCrae, National Institute on Aging, Baltimore Wendy Berry Mendes, Harvard University Rodolfo Mendoza-Denton, University of California, Berkeley Daniel K. Mroczek, Fordham University Stephen A. Petrill, Pennsylvania State University Ralph L. Piedmont, Loyola College in Maryland E. Ashby Plant, Florida State University Brent Roberts, University of Illinois at Urbana–Champaign Michael D. Robinson, North Dakota State University Daniel W. Russell, Iowa State University Oliver C. Schultheiss, University of Michigan Suzanne C. Segerstrom, University of Kentucky Kennon M. Sheldon, University of Missouri—Columbia C. R. Snyder, University of Kansas Sanjay Srivastava, University of Oregon Timothy Strauman, Duke University Jerry Suls, University of Iowa William B. Swann Jr., University of Texas at Austin Howard Tennen, University of Connecticut Health Center Suzanne Thompson, Pomona College Robert J. Vallerand, Universite´ du Que´bec a` Montre´al, Montreal, Quebec, Canada Kathleen D. Vohs, University of Minnesota David Watson, University of Iowa Barbara Woike, Columbia University Rex A. Wright, University of Alabama at Birmingham Assistant to the Editor—Barbara Adewusi
Associate Editors Tim Kasser, Knox College Mario Mikulincer, Bar-Ilan University, Ramat-Gan, Israel Eva M. Pomerantz, University of Illinois at Urbana–Champaign Richard W. Robins, University of California, Davis Gerard Saucier, University of Oregon Thomas A. Widiger, University of Kentucky Consulting Editors Stephan A. Ahadi, American Institutes for Research, Washington, DC Jamie Arndt, University of Missouri—Columbia Jens B. Asendorpf, Humboldt-Universita¨t Berlin, Berlin, Germany Michael C. Ashton, Brock University, St. Catherines, Ontario, Canada Ozlem Ayduk, University of California, Berkeley Roy F. Baumeister, Florida State University Vero´nica Benet-Martı´nez, University of California, Riverside April L. Bleske-Rechek, University of Wisconsin—Eau Claire George A. Bonanno, Teachers College, Columbia University Avshalom Caspi, King’s College, London Edward C. Chang, University of Michigan Serena Chen, University of California, Berkeley A. Timothy Church, Washington State University James Coan, University of Wisconsin—Madison M. Lynne Cooper, University of Missouri—Columbia Michael Eid, University of Geneva, Geneva, Switzerland Andrew J. Elliot, University of Rochester Lisa Feldman Barrett, Boston College William Fleeson, Wake Forest University R. Chris Fraley, University of Illinois at Chicago Antonio L. Freitas, State University of New York at Stony Brook David C. Funder, University of California, Riverside Steven W. Gangestad, University of New Mexico Carol L. Gohm, University of Mississippi Eddie Harmon-Jones, Texas A&M University
APA Journal Staff Susan J. A. Harris, Senior Director, Journals Program; Skip Maier, Director, Journal Services; Paige W. Jackson, Director, Editorial Services; Becky Shaw, Account Manager, Julie Palmer-Hoffman, Editorial Manager, Melissa Shella, Lead Editor, Patricia Beck, Marla Bonner, Stefanie Lazer, Amy Myers, Lisa O’Hearn, Charles Rhoads, Kimberly Till, Manuscript Editors; Aysha Y. Longshore, Editorial Production Manager xi
Author Index to Volume 91 Key to Pagination Issue No. 1 2 3
Month July August September
Pages 1–204 205–368 369–582
Issue No. 4 5 6
ARTICLES
Pages 583–796 797–994 995–1166
Brackett, Marc A., Rivers, Susan E., Shiffman, Sara, Lerner, Nicole, and Salovey, Peter—Relating Emotional Abilities to Social Functioning: A Comparison of Self-Report and Performance Measures of Emotional Intelligence. . . . . . . . . . . . . 780 Breugelmans, Seger M., and Poortinga, Ype H.—Emotion Without a Word: Shame and Guilt Among Rara´muri Indians and Rural Javanese. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1111 Brin˜ol, Pablo, Petty, Richard E., and Wheeler, S. Christian—Discrepancies Between Explicit and Implicit Self-Concepts: Consequences for Information Processing. . . . . . . . . . . . . . . . . 154 Brodbeck, Felix C.—see Schulz-Hardt, Stefan Brunell, Amy B.—see Finkel, Eli J. Burke, Christopher T.—see Carnelley, Katherine B. Burkley, Edward—see Muraven, Mark Burnham, Terence C.—see DeSteno, David Burton, Kimberly D., Lydon, John E., D’Alessandro, David U., and Koestner, Richard—The Differential Effects of Intrinsic and Identified Motivation on Well-Being and Performance: Prospective, Experimental, and Implicit Approaches to SelfDetermination Theory. . . . . . . . . . . . . . . . . . . . . . . . . . . 750 Butler, Emily A.—see Srivastava, Sanjay Butz, David A., and Plant, E. Ashby—Perceiving Outgroup Members as Unresponsive: Implications for Approach-Related Emotions, Intentions, and Behavior. . . . . . . . . . . . . . . . . . 1066 Bybee, Deborah—see Oyserman, Daphna
Agnew, Christopher R.—see Arriaga, Ximena B. Amanatullah, Emily T.—see Flynn, Francis J. Ames, Daniel R.—see Flynn, Francis J. Amodio, David M., and Devine, Patricia G.—Stereotyping and Evaluation in Implicit Race Bias: Evidence for Independent Constructs and Unique Effects on Behavior. . . . . . . . . . . . . 652 Anderson, Cameron, Srivastava, Sanjay, Beer, Jennifer S., Spataro, Sandra E., and Chatman, Jennifer A.—Knowing Your Place: Self-Perceptions of Status in Face-to-Face Groups. . . . . . . . 1094 Armor, David A., and Sackett, Aaron M.—Accuracy, Error, and Bias in Predictions for Real Versus Hypothetical Events. . . . 583 Arndt, Jamie—see Wildschut, Tim Arriaga, Ximena B., Reed, Jason T., Goodfriend, Wind, and Agnew, Christopher R.—Relationship Perceptions and Persistence: Do Fluctuations in Perceived Partner Commitment Undermine Dating Relationships?. . . . . . . . . . . . . . . . . . . . . 1045 Bain, Paul G., Kashima, Yoshihisa, and Haslam, Nick—Conceptual Beliefs About Human Values and Their Implications: Human Nature Beliefs Predict Value Importance, Value Trade-Offs, and Responses to Value-Laden Rhetoric. . . . . . . . . . . . . . . Balcetis, Emily, and Dunning, David—See What You Want to See: Motivational Influences on Visual Perception. . . . . . . . . . . Barrett, H. Clark, Frederick, David A., Haselton, Martie G., and Kurzban, Robert—Can Manipulations of Cognitive Load Be Used to Test Evolutionary Hypotheses?. . . . . . . . . . . . . . . Bartels, Meike—see De Fruyt, Filip Bartlett, Monica Y.—see DeSteno, David (two entries) Bartz, Jennifer A., and Lydon, John E.—Navigating the Interdependence Dilemma: Attachment Goals and the Use of Communal Norms With Potential Close Others. . . . . . . . . . . . . Baumeister, Roy F.—see DeWall, C. Nathan Baumeister, Roy F.—see Gailliot, Matthew T. Bazerman, Max H.—see Caruso, Eugene M. Bazerman, Max H.—see Epley, Nicholas Beer, Jennifer S.—see Anderson, Cameron Benham, Grant, Woody, Erik Z., Wilson, K. Shannon, and Nash, Michael R.—Expect the Unexpected: Ability, Attitude, and Responsiveness to Hypnosis. . . . . . . . . . . . . . . . . . . . . . . Bergeman, C. S.—see Ong, Anthony D. Birnbaum, Gurit E., Reis, Harry T., Mikulincer, Mario, Gillath, Omri, and Orpaz, Ayala—When Sex Is More Than Just Sex: Attachment Orientations, Sexual Experience, and Relationship Quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bisconti, Toni L.—see Ong, Anthony D. Bodenhausen, Galen V.—see Hugenberg, Kurt Bolger, Niall—see Carnelley, Katherine B. Bo¨rdgen, Sandra—see Reisenzein, Rainer
Month October November December
351 612
Campbell, W. Keith—see Finkel, Eli J. Carnelley, Katherine B., Wortman, Camille B., Bolger, Niall, and Burke, Christopher T.—The Time Course of Grief Reactions to Spousal Loss: Evidence From a National Probability Sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carnevale, Peter J.—see Henderson, Marlone D. Caruso, Eugene M., Epley, Nicholas, and Bazerman, Max H.—The Costs and Benefits of Undoing Egocentric Responsibility Assessments in Groups. . . . . . . . . . . . . . . . . . . . . . . . . . . . Caruso, Eugene M.—see Epley, Nicholas Chapman, Judith Flynn—see DeSteno, David Chartrand, Tanya L.—see Finkel, Eli J. Chatman, Jennifer A.—see Anderson, Cameron Chavanon, Mira-Lynn—see Wacker, Jan Chen, Henian—see Kasen, Stephanie Chun, Woo Young, and Kruglanski, Arie W.—The Role of Task Demands and Processing Resources in the Use of Base-Rate and Individuating Information. . . . . . . . . . . . . . . . . . . . . Cialdini, Robert B.—see Griskevicius, Vladas (two entries) Clarkson, Joshua J.—see Tormala, Zakary L. Cohen, Patricia—see Kasen, Stephanie Crawford, Thomas—see Kasen, Stephanie Crush, Jonathan—see Dambrun, Michae¨l Curhan, Jared R., Elfenbein, Hillary Anger, and Xu, Heng—What Do People Value When They Negotiate? Mapping the Domain of Subjective Value in Negotiation. . . . . . . . . . . . . . . . . .
513
77
342
929
xii
476
857
205
493
AUTHOR INDEX TO VOLUME 91 D’Alessandro, David U.—see Burton, Kimberly D. Dalton, Amy N.—see Finkel, Eli J. Dambrun, Michae¨l, Taylor, Donald M., McDonald, David A., Crush, Jonathan, and Me´ot, Alain—The Relative Deprivation Gratification Continuum and the Attitudes of South Africans Toward Immigrants: A Test of the V-Curve Hypothesis. . . . . 1032 Dasgupta, Nilanjana, and Rivera, Luis M.—From Automatic Antigay Prejudice to Behavior: The Moderating Role of Conscious Beliefs About Gender and Behavioral Control. . . . . . . . . . . 268 De Clercq, Barbara—see De Fruyt, Filip Decuyper, Mieke—see De Fruyt, Filip De Dreu, Carsten K. W.—see Van Kleef, Gerben A. De Fruyt, Filip, Bartels, Meike, Van Leeuwen, Karla G., De Clercq, Barbara, Decuyper, Mieke, and Mervielde, Ivan—Five Types of Personality Continuity in Childhood and Adolescence. . . . 538 DeSteno, David, Bartlett, Monica Y., and Salovey, Peter—Constraining Accommodative Homunculi in Evolutionary Explorations of Jealousy: A Reply to Barrett et al. (2006). . . . . . . 519 DeSteno, David, Valdesolo, Piercarlo, and Bartlett, Monica Y. —Jealousy and the Threatened Self: Getting to the Heart of the Green-Eyed Monster. . . . . . . . . . . . . . . . . . . . . . . . . . . . 626 Deutsch, Roland, Gawronski, Bertram, and Strack, Fritz—At the Boundaries of Automaticity: Negation as Reflective Operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385 Devine, Patricia G.—see Amodio, David M. DeWall, C. Nathan, and Baumeister, Roy F.—Alone but Feeling No Pain: Effects of Social Exclusion on Physical Pain Tolerance and Pain Threshold, Affective Forecasting, and Interpersonal Empathy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 DeYoung, Colin G.—Higher-Order Factors of the Big Five in a Multi-Informant Sample. . . . . . . . . . . . . . . . . . . . . . . . . 1138 Dhar, Ravi—see Fishbach, Ayelet Diener, Ed—see Scollon, Christie Napa Dunning, David—see Balcetis, Emily Elfenbein, Hillary Anger—see Curhan, Jared R. Epley, Nicholas, Caruso, Eugene M., and Bazerman, Max H.—When Perspective Taking Increases Taking: Reactive Egoism in Social Interaction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Epley, Nicholas—see Caruso, Eugene M.
872
Ferreira, Mario B., Garcia-Marques, Leonel, Sherman, Steven J., and Sherman, Jeffrey W.—Automatic and Controlled Components of Judgment and Decision Making. . . . . . . . . . . . . . . 797 Finkel, Eli J., Campbell, W. Keith, Brunell, Amy B., Dalton, Amy N., Scarbeck, Sarah J., and Chartrand, Tanya L.—HighMaintenance Interaction: Inefficient Social Coordination Impairs Self-Regulation. . . . . . . . . . . . . . . . . . . . . . . . . . . 456 Fischer, Peter—see Jonas, Eva Fishbach, Ayelet, Dhar, Ravi, and Zhang, Ying—Subgoals as Substitutes or Complements: The Role of Goal Accessibility. . . . 232 Fletcher, Garth J. O.—see Overall, Nickola C. Flynn, Francis J., Reagans, Ray E., Amanatullah, Emily T., and Ames, Daniel R.—Helping One’s Way to the Top: SelfMonitors Achieve Status by Helping Others and Knowing Who Helps Whom. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1123 Frederick, David A.—see Barrett, H. Clark Frey, Dieter—see Schulz-Hardt, Stefan Fujita, Kentaro—see Henderson, Marlone D. Funder, David C.—see Letzring, Tera D. Gable, Shelly L., Gonzaga, Gian C., and Strachman, Amy—Will You Be There for Me When Things Go Right? Supportive Responses to Positive Event Disclosures. . . . . . . . . . . . . . .
904
Gailliot, Matthew T., Schmeichel, Brandon J., and Baumeister, Roy F.—Self-Regulatory Processes Defend Against the Threat of Death: Effects of Self-Control Depletion and Trait SelfControl on Thoughts and Fears of Dying. . . . . . . . . . . . . . Galinsky, Adam D.—see Kray, Laura J. Gangestad, Steven W.—see DeSteno, David Garcia-Marques, Leonel, Santos, A. Sofia C., and Mackie, Diane M.—Stereotypes: Static Abstractions or Dynamic Knowledge Structures?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Garcia-Marques, Leonel—see Ferreira, Mario B. Gawronski, Bertram—see Deutsch, Roland Gillath, Omri—see Birnbaum, Gurit E. Goldberg, Lewis R.—see Hampson, Sarah E. Goldstein, Noah J.—see Griskevicius, Vladas Gonzaga, Gian C.—see Gable, Shelly L. Goodfriend, Wind—see Arriaga, Ximena B. Gray, Peter B.—see DeSteno, David Griskevicius, Vladas, Cialdini, Robert B., and Kenrick, Douglas T.— Peacocks, Picasso, and Parental Investment: The Effects of Romantic Motives on Creativity. . . . . . . . . . . . . . . . . . . . Griskevicius, Vladas, Goldstein, Noah J., Mortensen, Chad R., Cialdini, Robert B., and Kenrick, Douglas T.—Going Along Versus Going Alone: When Fundamental Motives Facilitate Strategic (Non)Conformity. . . . . . . . . . . . . . . . . . . . . . . . Gross, James J.—see Srivastava, Sanjay
xiii
49
814
63
281
Halabi, Samer—see Nadler, Arie Hampson, Sarah E., and Goldberg, Lewis R.—A First Large Cohort Study of Personality Trait Stability Over the 40 Years Between Elementary School and Midlife. . . . . . . . . . . . . . . . 763 Harkins, Stephen G.—Mere Effort as the Mediator of the Evaluation–Performance Relationship. . . . . . . . . . . . . . . . . 436 Haselton, Martie G.—see Barrett, H. Clark Haslam, Nick—see Bain, Paul G. Henderson, Marlone D., Fujita, Kentaro, Trope, Yaacov, and Liberman, Nira—Transcending the “Here”: The Effect of Spatial Distance on Social Judgment. . . . . . . . . . . . . . . . . . . . . . 845 Henderson, Marlone D., Trope, Yaacov, and Carnevale, Peter J.— Negotiation From a Near and Distant Time Perspective. . . . . 712 Hirschberger, Gilad—Terror Management and Attributions of Blame to Innocent Victims: Reconciling Compassionate and Defensive Responses. . . . . . . . . . . . . . . . . . . . . . . . . . . 832 Holtbernd, Thomas—see Reisenzein, Rainer Hugenberg, Kurt, Bodenhausen, Galen V., and McLain, Melissa— Framing Discrimination: Effects of Inclusion Versus Exclusion Mind-Sets on Stereotypic Judgments. . . . . . . . . . . . . . 1020 Humrichouse, John—see Watson, David Imada, Toshie—see Kitayama, Shinobu Ishii, Keiko—see Kitayama, Shinobu Jonas, Eva, and Fischer, Peter—Terror Management and Religion: Evidence That Intrinsic Religiousness Mitigates Worldview Defense Following Mortality Salience. . . . . . . . . . . . . . . . Karasawa, Mayumi—see Kitayama, Shinobu Karpinski, Andrew, and Steinman, Ross B.—The Single Category Implicit Association Test as a Measure of Implicit Social Cognition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Karremans, Johan C.—see van Prooijen, Jan-Willem Kasen, Stephanie, Chen, Henian, Sneed, Joel, Crawford, Thomas, and Cohen, Patricia—Social Role and Birth Cohort Influences on Gender-Linked Personality Traits in Women: A 20-Year Longitudinal Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . .
553
16
944
xiv
AUTHOR INDEX TO VOLUME 91
Kashima, Yoshihisa—see Bain, Paul G. Keller, Matthew C., and Nesse, Randolph M.—The Evolutionary Significance of Depressive Symptoms: Different Adverse Situations Lead to Different Depressive Symptom Patterns. . . . Kenrick, Douglas T.—see Griskevicius, Vladas (two entries) Kerschreiter, Rudolf—see Schulz-Hardt, Stefan Kitayama, Shinobu, Ishii, Keiko, Imada, Toshie, Takemura, Kosuke, and Ramaswamy, Jenny—Voluntary Settlement and the Spirit of Independence: Evidence from Japan’s “Northern Frontier”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kitayama, Shinobu, Mesquita, Batja, and Karasawa, Mayumi— Cultural Affordances and Emotional Experience: Socially Engaging and Disengaging Emotions in Japan and the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Klar, Yechiel—see Roccas, Sonia Koestner, Richard—see Burton, Kimberly D. Kray, Laura J., Galinsky, Adam D., and Wong, Elaine M.—Thinking Within the Box: The Relational Processing Style Elicited by Counterfactual Mind-Sets. . . . . . . . . . . . . . . . . . . . . . Kruglanski, Arie W.—see Chun, Woo Young Kurzban, Robert—see Barrett, H. Clark Lerner, Nicole—see Brackett, Marc A. Letzring, Tera D., Wells, Shannon M., and Funder, David C.— Information Quantity and Quality Affect the Realistic Accuracy of Personality Judgment. . . . . . . . . . . . . . . . . . . . . . Liberman, Nira—see Henderson, Marlone D. Liviatan, Ido—see Roccas, Sonia Lydon, John E.—see Bartz, Jennifer A. Lydon, John E.—see Burton, Kimberly D. Mackie, Diane M.—see Garcia-Marques, Leonel Malik, Jill—see Vaughn, Leigh Ann Manstead, Antony S. R.—see Van Kleef, Gerben A. Marx, David M., and Stapel, Diederik A.—Distinguishing Stereotype Threat From Priming Effects: On the Role of the Social Self and Threat-Based Concerns. . . . . . . . . . . . . . . . . . . . Mason, Winter—see Queller, Sarah Matsumoto, David, and Willingham, Bob—The Thrill of Victory and the Agony of Defeat: Spontaneous Expressions of Medal Winners of the 2004 Athens Olympic Games. . . . . . . . . . . Matthews, Karen A.—see Ruiz, John M. Matz, Denise—see Reisenzein, Rainer McCarthy, Kimberly—see Pronin, Emily McConnell, Allen R.—see Rydell, Robert J. McDonald, David A.—see Dambrun, Michae¨l McGonigal, Kelly M.—see Srivastava, Sanjay McIntyre, Matthew, Gangestad, Steven W., Gray, Peter B., Chapman, Judith Flynn, Burnham, Terence C., O’Rourke, Mary T., and Thornhill, Randy—Romantic Involvement Often Reduces Men’s Testosterone Levels—But Not Always: The Moderating Role of Extrapair Sexual Interest. . . . . . . . . . . . . . . . . McLain, Melissa—see Hugenberg, Kurt Me´ot, Alain—see Dambrun, Michae¨l Mervielde, Ivan—see De Fruyt, Filip Mesquita, Batja—see Kitayama, Shinobu Mikulincer, Mario—see Birnbaum, Gurit E. Mojzisch, Andreas—see Schulz-Hardt, Stefan Mortensen, Chad R.—see Griskevicius, Vladas Muraven, Mark, Shmueli, Dikla, and Burkley, Edward—Conserving Self-Control Strength. . . . . . . . . . . . . . . . . . . . . . . . .
316
369
890
33
111
243
568
642
524
Nadler, Arie, and Halabi, Samer—Intergroup Helping as Status Relations: Effects of Status Stability, Identification, and Type of Help on Receptivity to High-Status Group’s Help. . . . . . . Nash, Michael R.—see Benham, Grant Nesse, Randolph M.—see Keller, Matthew C. Niemiec, Christopher P.—see Sheldon, Kennon M.
97
Ong, Anthony D., Bergeman, C. S., Bisconti, Toni L., and Wallace, Kimberly A.—Psychological Resilience, Positive Emotions, and Successful Adaptation to Stress in Later Life. . . . . . . . . O’Rourke, Mary T.—see DeSteno, David Orpaz, Ayala—see Birnbaum, Gurit E. Overall, Nickola C., Fletcher, Garth J. O., and Simpson, Jeffry A.— Regulation Processes in Intimate Relationships: The Role of Ideal Standards. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oyserman, Daphna, Bybee, Deborah, and Terry, Kathy—Possible Selves and Academic Outcomes: How and When Possible Selves Impel Action. . . . . . . . . . . . . . . . . . . . . . . . . . . .
188
Petkova, Zhivka—see Vaughn, Leigh Ann Petty, Richard E.—see Brin˜ol, Pablo Petty, Richard E.—see Tormala, Zakary L. Plant, E. Ashby—see Butz, David A. Poortinga, Ype H.—see Breugelmans, Seger M. Pronin, Emily, Wegner, Daniel M., McCarthy, Kimberly, and Rodriguez, Sylvia—Everyday Magical Powers: The Role of Apparent Mental Causation in the Overestimation of Personal Influence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
218
Queller, Sarah, Schell, Terry, and Mason, Winter—A Novel View of Between-Categories Contrast and Within-Category Assimilation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
406
Ramaswamy, Jenny—see Kitayama, Shinobu Reagans, Ray E.—see Flynn, Francis J. Reed, Jason T.—see Arriaga, Ximena B. Reis, Harry T.—see Birnbaum, Gurit E. Reisenzein, Rainer, Bo¨rdgen, Sandra, Holtbernd, Thomas, and Matz, Denise—Evidence for Strong Dissociation Between Emotion and Facial Displays: The Case of Surprise. . . . . . . Richards, Jane M.—see Srivastava, Sanjay Rivera, Luis M.—see Dasgupta, Nilanjana Rivers, Susan E.—see Brackett, Marc A. Roccas, Sonia, Klar, Yechiel, and Liviatan, Ido—The Paradox of Group-Based Guilt: Modes of National Identification, Conflict Vehemence, and Reactions to the In-Group’s Moral Violations Rodriguez, Sylvia—see Pronin, Emily Routledge, Clay—see Wildschut, Tim Ruiz, John M., Matthews, Karen A., Scheier, Michael F., and Schulz, Richard—Does Who You Marry Matter for Your Health? Influence of Patients’ and Spouses’ Personality on Their Partners’ Psychological Well-Being Following Coronary Artery Bypass Surgery. . . . . . . . . . . . . . . . . . . . . . . Rydell, Robert J., and McConnell, Allen R.—Understanding Implicit and Explicit Attitude Change: A Systems of Reasoning Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sackett, Aaron M.—see Armor, David A. Salovey, Peter—see Brackett, Marc A. Salovey, Peter—see DeSteno, David Santos, A. Sofia C.—see Garcia-Marques, Leonel Scarbeck, Sarah J.—see Finkel, Eli J. Scheier, Michael F.—see Ruiz, John M. Schell, Terry—see Queller, Sarah
730
662
295
698
255
995
AUTHOR INDEX TO VOLUME 91 Schmeichel, Brandon J.—see Gailliot, Matthew T. Schulz, Richard—see Ruiz, John M. Schulz-Hardt, Stefan, Brodbeck, Felix C., Mojzisch, Andreas, Kerschreiter, Rudolf, and Frey, Dieter—Group Decision Making in Hidden Profile Situations: Dissent as a Facilitator for Decision Quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1080 Schwartz, Sandra—see Vaughn, Leigh Ann Scollon, Christie Napa, and Diener, Ed—Love, Work, and Changes in Extraversion and Neuroticism Over Time. . . . . . . . . . . . 1152 Sedikides, Constantine—see Wildschut, Tim Sheldon, Kennon M., and Niemiec, Christopher P.—It’s Not Just the Amount That Counts: Balanced Need Satisfaction Also Affects Well-Being. . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 Sherman, Jeffrey W.—see Ferreira, Mario B. Sherman, Steven J.—see Ferreira, Mario B. Shiffman, Sara—see Brackett, Marc A. Shmueli, Dikla—see Muraven, Mark Simpson, Jeffry A.—see Overall, Nickola C. Sneed, Joel—see Kasen, Stephanie Sparrow, Betsy, and Wegner, Daniel M.—Unpriming: The Deactivation of Thoughts Through Expression. . . . . . . . . . . . . . 1009 Spataro, Sandra E.—see Anderson, Cameron Srivastava, Sanjay, McGonigal, Kelly M., Richards, Jane M., Butler, Emily A., and Gross, James J.—Optimism in Close Relationships: How Seeing Things in a Positive Light Makes Them So. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Srivastava, Sanjay—see Anderson, Cameron Stapel, Diederik A.—see Marx, David M. Steinman, Ross B.—see Karpinski, Andrew Stemmler, Gerhard—see Wacker, Jan Strachman, Amy—see Gable, Shelly L. Strack, Fritz—see Deutsch, Roland
xv
Van Kleef, Gerben A., De Dreu, Carsten K. W., and Manstead, Antony S. R.—Supplication and Appeasement in Conflict and Negotiation: The Interpersonal Effects of Disappointment, Worry, Guilt, and Regret. . . . . . . . . . . . . . . . . . . . . . . . . Van Leeuwen, Karla G.—see De Fruyt, Filip van Prooijen, Jan-Willem, Karremans, Johan C., and van Beest, Ilja— Procedural Justice and the Hedonic Principle: How Approach Versus Avoidance Motivation Influences the Psychology of Voice. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vaughn, Leigh Ann, Malik, Jill, Schwartz, Sandra, Petkova, Zhivka, and Trudeau, Lindsay—Regulatory Fit as Input for Stop Rules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wacker, Jan, Chavanon, Mira-Lynn, and Stemmler, Gerhard—Investigating the Dopaminergic Basis of Extraversion in Humans: A Multilevel Approach. . . . . . . . . . . . . . . . . . . . . . Wallace, Kimberly A.—see Ong, Anthony D. Watson, David, and Humrichouse, John—Personality Development in Emerging Adulthood: Integrating Evidence From SelfRatings and Spouse Ratings. . . . . . . . . . . . . . . . . . . . . . . Wegner, Daniel M.—see Pronin, Emily Wegner, Daniel M.—see Sparrow, Betsy Wells, Shannon M.—see Letzring, Tera D. Wheeler, S. Christian—see Brin˜ol, Pablo Wildschut, Tim, Sedikides, Constantine, Arndt, Jamie, and Routledge, Clay—Nostalgia: Content, Triggers, Functions. . . . . . Williams, Kipling D.—see van Beest, Ilja Willingham, Bob—see Matsumoto, David Wilson, K. Shannon—see Benham, Grant Wong, Elaine M.—see Kray, Laura J. Woody, Erik Z.—see Benham, Grant Wortman, Camille B.—see Carnelley, Katherine B.
124
686
601
171
959
975
Xu, Heng—see Curhan, Jared R. Takemura, Kosuke—see Kitayama, Shinobu Taylor, Donald M.—see Dambrun, Michae¨l Terry, Kathy—see Oyserman, Daphna Thornhill, Randy—see DeSteno, David Tormala, Zakary L., Clarkson, Joshua J., and Petty, Richard E.— Resisting Persuasion by the Skin of One’s Teeth: The Hidden Success of Resisted Persuasive Messages. . . . . . . . . . . . . . Trope, Yaacov—see Henderson, Marlone D. (two entries) Trudeau, Lindsay—see Vaughn, Leigh Ann Valdesolo, Piercarlo—see DeSteno, David van Beest, Ilja, and Williams, Kipling D.—When Inclusion Costs and Ostracism Pays, Ostracism Still Hurts. . . . . . . . . . . . . . van Beest, Ilja—see van Prooijen, Jan-Willem
Zhang, Ying—see Fishbach, Ayelet
OTHER
423
918
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1166 American Psychological Association Subscription Claims Information . . . . . . . . . . . . . . . . . 15, 217, 475, 697 Call for Nominations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 993 E-Mail Notification of Your Latest Issue Online! . . . . . . . . . . . . . ii (July), 315, 435, 600, 1031 Instructions to Authors . . . . . . . . . . . . . . . . . . 96, 242, 384, 779 Low Publication Prices for APA Members and Affiliates . . . . . 749 New Editors Appointed, 2008 –2013 . . . . . . . . . . . . . . . . . . 871 Subscription Order Form . . . . . . 62, ii (August), 958, ii (December)
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1009 –1019
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1009
Unpriming: The Deactivation of Thoughts Through Expression Betsy Sparrow and Daniel M. Wegner Harvard University Unpriming is a decrease in the influence of primed knowledge following a behavior expressing that knowledge. The authors investigated strategies for unpriming the knowledge of an answer that is activated when people are asked to consider a simple question. Experiment 1 found that prior correct answering eliminated the bias people normally show toward correct responding when asked to answer yes–no questions randomly. Experiment 2 revealed that prior answering intended to be random did not unprime knowledge on subsequent attempts to answer randomly. Experiment 3 found that exposure to the correct answer did not influence the knowledge bias but that exposure to the incorrect answer increased bias. Experiment 4 revealed that merely expressing the answer for oneself was sufficient to unprime knowledge. Experiment 5 found that each item of activated knowledge needs to be unprimed specifically, in that correctly answering 1 question does not reduce the knowledge bias in randomly answering another. Keywords: priming, intelligence, knowledge, control, answering questions
influence (Toth & Reingold, 1996). Conceptual and perceptual primes presented subliminally or surreptitiously can influence the ease with which the prime itself is later recognized (e.g., Jacoby & Dallas, 1981), can facilitate use of the prime in answers to questions such as word completions (e.g., Tulving, Schacter, & Stark, 1982), and can enhance the use of associates of the prime in other tasks (e.g., Higgins & King, 1981)—all without the person’s explicit awareness of the prime’s influence (for reviews of the literature, see Roediger & McDermott, 1993; Schacter, Chiu, & Ochsner, 1993). Conceptual primes can operate across modalities (e.g., from auditory to visual) and involve the encoding of semantic meaning so that associates of the prime are also activated. Perceptual primes are modality specific and do not involve elaborative encoding (see Blaxton, 1989, for a discussion of the distinctions). Explicit memory systems may play more of a role in conceptual memory tasks (Mulligan, 1997), but it has been argued that priming tasks generally involve both conceptual and perceptual processing, which involve deactivations in differential brain areas (Schacter & Buckner, 1998). When a prime is relevant to a possible behavior, it can increase the likelihood of that behavior (Bargh & Chartrand, 2000; Dijksterhuis & Bargh, 2001). Participants primed with the concept of hostility, for example, delivered more intense shocks to a person than did participants who were not primed (Carver, Ganellen, Froming, & Chambers, 1983). Similarly, participants primed with rudeness were more likely than others to interrupt someone engaged in conversation, and those primed with thoughts of the elderly were more likely to walk slowly while exiting from an experiment (Bargh, Chen, & Burrows, 1996). Other priming studies have found behavioral effects for primes of helpfulness (Macrae & Johnston, 1998), conformity (Epley & Gilovich, 1999), and even intelligence (Dijksterhuis & van Knippenberg, 1998). Priming can also unconsciously influence evaluations (Bargh, Chaiken, Raymond, & Hymes, 1996) and can prompt the nonconscious pursuit of goals (Chartrand & Bargh, 1996). These various priming effects suggest a model of human behavior in which people are controlled by the happenstance array of
When an experience primes a person to think about something, the person’s behavior and judgment may be influenced by the prime. The passing mention of a burrito in conversation one day, for example, might incline the listener to seek out a Mexican restaurant for lunch— even though the listener has no explicit memory of the word or the mentioning. This effect is well-known in psychology and the focus of many studies demonstrating the welter of subtle primings that guide human thought and behavior every day. What is not clearly understood, however, is what draws the influence of a prime to a close. The question addressed in our studies was whether, when the primed thought is expressed in some way, it then becomes less likely to have such cascading effects. We tested whether unpriming occurs when a primed thought is acted on.
Priming Effects The influence of priming was first observed experimentally by Storms (1958), who found that words presented for a subsequent memory test were often used by participants in an intervening indirect word-association task. This effect was named priming in a replication by Segal and Cofer (1960) and led to the seminal work of Neely (1977), who found that lexical decisions were completed with faster reaction times when a decision was preceded with a primed word that was semantically related. Ever since, many researchers have examined the properties of such unconscious
Betsy Sparrow and Daniel M. Wegner, Department of Psychology, Harvard University. This research was supported in part by National Institute of Mental Health Grant 49127 to Daniel M. Wegner. For serving as experimenters we thank Benjamin Chen, Jeffrey Fernandez, Myu Kulathungam, Katharina Kurkanski, Jonathan Spiker, and James Song. Correspondence concerning this article should be addressed to Betsy Sparrow, Department of Psychology, 33 Kirkland Street, 1468 William James Hall, Harvard University, Cambridge, MA 02138. E-mail:
[email protected] 1009
SPARROW AND WEGNER
1010
environmental influences they encounter (cf. Bargh & Ferguson, 2000). In the course of a day, a person is exposed to a sequence of primes, each of which may or may not achieve some influence, and the person’s primed path through the environment then determines which further primes will be encountered. This sounds like a workable model of human behavior— until the person reaches an environment that only primes one thing. The poor soul primed to get a burrito, for example, arrives at the Mexican restaurant and, further primed by the environment, ends up staying all day, unable to do anything but stand, transfixed and drooling, at the door of the cocina. Priming in experiments can sometimes indeed last for days (Squire, Shimamura, & Graf, 1987), so behavior and thought that reflect only the environment would regularly bring behavior to a perseverative cycle whenever priming occurs. The realization prompted by this example is that an adaptive process of priming would need to have a natural endpoint to draw it to a close. This endpoint may be the occurrence of an influence of the prime. For priming to be an adaptive process, in other words, it should come to an end when the prime has been used. Once a primed thought or behavior occurs, there should be a relatively rapid reduction in the influence of the prime on subsequent thought or behavior. Just as a person playing the piano must rapidly overcome the influence of each past note on the sheet music for the song to move forward, a person responding (even unconsciously) to a changing array of environmental primes would perform most effectively if each prime’s potential influence was curtailed when that prime had indeed registered a change in the person’s thought or action. To be effective in guiding behavior, priming should no longer guide behavior once the behavior has occurred. Primebased behavior should result in unpriming.
Varieties of Unpriming The history of psychology features many descriptions of unpriming, when unpriming is defined broadly as the reduction in the influence of a prime that occurs when the prime has been acted on. Versions of the idea of unpriming can be found in literatures focusing on catharsis, completion, and updating— emphasizing, respectively, the emotional, motivational, and cognitive ways of understanding this phenomenon. Emotional catharsis was perhaps the earliest of these ideas and can be found in Aristotle’s (trans. 1961) Poetics in the hypothesis that the negative emotions of pity and fear are reduced through the viewing of tragedy. The related view rendered by Breuer and Freud (1893–1895/ 1955) was that the expression of pathological emotions might serve to dissipate those emotions. Modern research on catharsis involving a simple venting of emotion—such as expressing aggression through hitting a punching bag— has not found that the emotion is deactivated by such activities (e.g., Bushman, Baumeister, & Stack, 1999). However, other studies of catharsis, such as Pennebaker’s diary studies (Pennebaker, 1989; Pennebaker & Beall, 1986), have found that expressing thoughts and feelings accrues health benefits and decreases subsequent thoughts of trauma. The motivational approach to unpriming suggests that behaving on the basis of a prime can achieve a sense of goal completion that renders the prime less active. Much has been written about people’s tendency for enhanced memory for incomplete over completed intentions (Zeigarnik, 1935) and the motivating quality of incomplete intentions (Lewin, 1939, 1951; Martin, Tesser, &
McIntosh, 1993). Indeed, much of the work on Zeigarnik-type effects has focused on the importance of completing important tasks (Martin et al., 1993) or goals specifically relevant to the self (Gollwitzer & Wicklund, 1982, 1985). Such work has suggested that intention is itself a prime, an urge toward behavior that must be fulfilled before the thought can be unprimed. In this regard, Fo¨rster, Liberman, and Higgins (2005) found that conscious goals enhance the accessibility of goal-related words over nonrelated words in a lexical decision task but that this accessibility is inhibited over time once the goal has been achieved. The idea that completing the pursuit of a goal can yield unpriming is particularly evident in discussions of the reliability of the measurement of motivation. Critiques of measures of motivation such as the Thematic Apperception Test (TAT; Morgan & Murray, 1935) revolved around the repeated finding of low internal consistency of such tests (Entwistle, 1972). The fact that the motivation to achieve waxes and wanes over the course of describing a series of pictures was embraced by Atkinson, Bongort, and Price (1977), however, who viewed this inconsistency as a natural result of the reduction in a motive that results when the motive is expressed in fantasy. In this analysis, the unreliability of the TAT occurs because an answer to a TAT item relevant to a motive reduces the psychological influence of that motive and so undermines the effect of that motive on subsequent items. Behaving in response to an accessible psychological influence, in other words, unprimes that influence. Notions like that of unpriming have surfaced in the cognition literature to explain the updating of cognitive representations or memories that occurs when new information becomes available. In studies of perception, for example, researchers have examined the inhibition of return in attention (Posner & Cohen, 1984)—a reduction in attention toward previously attended visual areas. Researchers in studies of executive function, in turn, have examined how the inhibition of previous task sets can facilitate task switching (e.g., Allport, Styles, & Hsieh, 1994). In this regard, Mayr and Keele (2000) found that shifts of intention between differing goal states may reduce the previous goal state in a process they term backward inhibition. Memory researchers also have examined effects of old memory retrieval on the updating of old memories with new memories (e.g., when one parks one’s car in a different place) in studies of retrieval inhibition (e.g., Bjork & Landauer, 1978). Cognitive studies of semantic satiation also suggest a kind of updating, in that priming may occasionally be slowed as a result of multiple repeated exposures of a prime, although the conditions under which this occurs are not clear (Esposito & Pelton, 1971). Semantic satiation appears to occur only when participants are engaged in a task that requires explicit use of knowledge of a target’s category membership (Smith, 1984). The literatures on catharsis, completion, and updating suggest that psychological theorists have often recognized that behavior prompted by a stimulus can naturally reduce the propensity toward subsequent stimulus-related behavior. The question of whether a thought deactivation process ensues when people behave on the basis of a primed thought, however, remains to be tested.
The Random Answering Paradigm Our interest in unpriming was prompted by the unusually robust priming effects observed in the random answering paradigm by Wegner, Fuller, and Sparrow (2003). In this paradigm, respondents
UNPRIMING
were asked to answer a series of simple yes–no questions as randomly as they could. These studies did not pursue the problem of whether people can make response sequences that resemble random sequences in details such as run length or nonredundancy (e.g., Baddeley, Emslie, Kolodny, & Duncan, 1998; Nickerson, 2002). The focus was on participants’ response correctness when correctness was not mentioned and participants were explicitly instructed to answer randomly. Participants were instructed not to use prearranged strategies but rather to listen to each question and “flip a coin in your head” to determine whether to answer “yes” or “no.” In other words, participants were asked to respond to each question randomly, without trying to be correct or incorrect. They were not asked to attempt to produce a pattern of responses that would appear random. With the correctness of “yes” and “no” responses balanced, truly random responding would be expected to produce correct responses 50% of the time, but respondents almost never achieved this low level of correctness and instead answered questions correctly on average at rates far higher than chance. Primed by their knowledge of the correct answers, participants in these studies seemed unable to control this intelligence in the pursuit of random answering. Further evidence collected by Wegner et al. (2003) suggests that the influence of primed knowledge activation in this paradigm was indeed uncontrollable. Extra financial incentives to be random did not undermine the primed knowledge effect, and extra time to establish a random response was also ineffective. The incentive did lower participants’ postexperimental estimates of their correctness, showing that they were under the mistaken impression that they could overcome the tendency to answer questions correctly. It is also worth noting that responses made quickly (in under 1,000 ms) were correct nearly 50% of the time or random, whereas questions answered in over 1,000 ms were more likely to be answered correctly. The only strategy participants seemed to be able to use to achieve random answering, in other words, was to respond preemptively before fully registering the question they were asked. Drawing from past research, we can suggest several possible explanations for why knowledge may act as a prime to generate correct responses in the random answering paradigm. These explanations follow from the general idea that the knowledge is activated by the questioning and cannot be deactivated by attempts at voluntary mental control. Higgins (1996) theorized that knowledge stores may vary in their potential to be activated. The context may increase accessibility or a history of recent knowledge use may create activation. Beyond this, there is the possibility that people feel compelled to answer questions correctly because there are implicit rules of communication, such as those Grice (1975) described, which would prompt an answer to a question if the answer is known. Random answering may be difficult because of the deeply established social norm to provide correct information in response to questions. The idea that knowledge can be activated by a goal state (to answer a question) suggests that the difficulty of random answering may involve the activation of thoughts by the goal. In this Zeigarnik-like view, a motivational process brings the answer to mind until the goal of answering correctly is complete (Goschke & Kuhl, 1993; Marsh, Hicks, & Bink, 1998; Martin et al., 1993). Alternatively, it may be easier for the person to provide a correct answer to a question than to try to overcome thoughts of the correct answer and come up with a random answer. It has been
1011
shown that beliefs are expressed quickly and automatically, whereas the rejection of an idea is a more effortful process (Gilbert, 1991; Wegner, Coulton, & Wenzlaff, 1985). Trying to suppress the thoughts generated by the knowledge of the right answer may lead to a counterproductive increase in these same thoughts, which may block a random response (Wegner, Schneider, Carter, & White, 1987). There are yet other potential explanations for the random answering effect (see Wegner et al., 2003), and the exploration of unpriming may shed some light on these.
The Present Research In these studies, we examined the effects of answering a question correctly on the subsequent ability to answer the question randomly. Prior correct answering in the random answering paradigm involves behaving on the basis of the knowledge primed by the question and so provides a basic test of whether the use of primed knowledge can induce unpriming. Experiment 1 tested whether answering questions correctly before attempting to answer them randomly would result in successful random answers. In Experiment 2, we explored whether correct answer unpriming is only a result of answering each question twice, regardless of the correctness of the first response. In Experiment 3, we explored whether simple exposure to correct or incorrect answers prior to the random answer task might explain unpriming effects. In Experiment 4, we investigated the minimal level of response that would allow unpriming by testing whether, if participants answer questions correctly only to themselves, this expression would be sufficient to deactivate the thought and allow random responding. Experiment 5 tested whether answering any question correctly prior to random responding would allow a generalized expression of correctness that would result in unpriming.
Experiment 1: Prior Correct Answering and Random Response This study tested the unpriming effect of correctly answering a question. Participants in two conditions were asked to answer a series of easy yes–no questions randomly. Those in one condition were asked to answer each question correctly before answering it randomly, whereas those in the other condition were asked to provide only a random answer. Correctness of the random answers was assessed for both.
Method Participants. Forty-eight undergraduates at Harvard University (31 women and 17 men) participated for course credit in the psychology department study pool or for monetary compensation. All participants in this and the following experiments gave informed consent for participation. Questions. Participants answered a series of 60 questions with “yes” or “no” responses. All of the questions asked were easy (e.g., “Does 2 plus 3 equal 5?” “Does a triangle have three sides?”), so all participants were assumed to have knowledge of the correct answers. The correct answer was “yes” for half the questions and “no” for the other half (participants were unaware both of this and of the total number of questions they would answer). Questions were presented in a predetermined random order so that correct “yes” and “no” responses would be presented with no particular pattern. Design and procedure. Participants were run individually at a computer that had a pair of keys labeled yes and no. Participants in a randomonly condition were told that they were to answer a series of questions and
SPARROW AND WEGNER
1012
that they were to try to answer each question randomly. Participants in a correct–random condition were told that they were to answer a series of questions, for which each question would be presented twice in succession. For the first appearance of the question, they were instructed to try to answer the question correctly. For the second appearance of the question, they were to attempt to answer the question randomly. Participants in both conditions were then left alone to read through instructions presented on the computer and to begin the experiment when they were ready. The instructions on the computer regarding random response, shown to participants in both correct–random and random-only conditions, were as follows: Try not to generate a predictable pattern of yes/no/yes/no or yes/yes/ yes, but try to generate a random response when answering each question. One way to think about this is to try to mentally flip a coin in your head when each question is asked. Participants in the correct–random condition received the additional instruction, “Please try your best to answer the question correctly the first time it is asked, and then try to answer randomly the second time that it is asked.” Questions were presented on a PC monitor and answers were recorded through the program DirectRT (Jarvis, 2000). Participants saw each question in the center of the screen along with the words yes and no along the bottom of the screen and heard the question read through the computer speakers. The interval between question presentations was 2 s. For participants in the correct–random condition, for each first presentation of a question, an instruction appeared at the top of the screen asking participants to “try to answer the question correctly.” For the second presentation of the question, participants were instructed to “try to answer the question randomly.” Sixty individual questions were shown twice each, for a total of 120 trials. Participants in the random-only condition saw each question once, for 60 trials, with the instruction “try to answer each question randomly” appearing on the top of the screen. Participants in this and all of the following experiments were debriefed prior to dismissal.
Results and Discussion Participants who were allowed to answer only once, randomly, exhibited a significantly higher mean proportion of correct responses (M ⫽ .58, SD ⫽ .15) than did correct–random participants (M ⫽ .49, SD ⫽ .12), t(46) ⫽ 2.07, p ⬍ .05, 2 ⫽ .09 (see Figure 1). Participants in the random-only condition answered at correctness levels significantly greater than chance (M ⫽ .58 tested
against a statistic of .50), t(24) ⫽ 2.53, p ⬍ .02. Participants in the correct–random condition (M ⫽ .49) were no different from chance, t(22) ⫽ 0.23, p ⫽ .82. Thus, it appears that allowing participants to express the correct response first significantly reduces the tendency to provide a correct response when it is not appropriate—that is, when a random response is requested. It is worth noting that for correct–random participants, the mean proportion correct when a correct response was requested was .98 (SD ⫽ .03), significantly greater than the total mean proportion for random responses from these same participants (M ⫽ .49, SD ⫽ .12), t(22) ⫽ 18.18, p ⬍ .001, 2 ⫽ .94. Perhaps participants who successfully performed near chance correctness levels had a strategy to answer systematically and thus were not actually unprimed per se. The 60 questions had half “yes” and half “no” correct responses, although participants did not know this, nor did they know how many questions they would have to answer. The best strategy, therefore, to reach .50 would be to answer all questions with “yes” or all questions with “no.” Not a single participant in either the random-only or correct–random conditions used this strategy for their random responses; the highest number of “yes” answers from any participant was 35 out of 60 and the lowest number was 26. If participants came up with a strategy for successful random responding, it would be expected that they would begin the task with such a strategy (e.g., trying to flip a coin in their head as suggested in the instructions) or that they would perfect a strategy as they went along. Thus, one might expect either a random response strategy that is more successful early, with the strategy breaking down over the course of the trials, or, alternatively, a random response strategy that is more successful later, when compared with earlier trials. However, there was no difference when looking at the first half and second half of the questions separately from mean proportions correct reported above for the responses overall. These findings suggest that responding to a question may deactivate knowledge of the answer, allowing the respondent to answer the question randomly later on. However, several other possible interpretations need to be examined, and these are taken up in turn in the subsequent experiments.
Mean proportion correct during random responding
Random-only
Correct-random
0 .0 0
0 .50
1. 0 0
Figure 1. Mean proportions correct for participants in the random-only and correct–random conditions in Experiment 1. Error bars represent standard errors of the mean.
UNPRIMING
Experiment 2: Prior Random Answering and Random Response One possible interpretation of the results of Experiment 1 is that any prior answering of a question might reduce knowledge activation and allow later random responding. This experiment assessed whether an initial random response might serve the same unpriming function as an initial correct response on subsequent random responding.
Method Forty-five participants (34 women and 11 men) were recruited as in Experiment 1 and randomly assigned to one of three conditions: random only and correct–random, as in Experiment 1, and random–random, which was added for this study. The procedures for the random-only and correct– random conditions paralleled those of the prior study. The procedure for the random–random condition departed from the procedures of the other conditions only in that participants were asked to answer each question randomly twice in a row. In the random–random condition, instructions to “try to answer the question randomly” appeared at the top of each presentation screen. Participants in all conditions were then left alone to read through a series of detailed instructions presented on the computer and to begin the experiment when they were ready. Questions and instructions otherwise replicated those of Experiment 1.
Results and Discussion Comparing the first appearance of each question (the only appearance for random-alone participants) reveals that participants asked to answer these questions correctly exhibited a significantly greater mean proportion correct (M ⫽ .94, SD ⫽ .06) than either random group (M ⫽ .64, SD ⫽ .22, for the random–random condition and M ⫽ .69, SD ⫽ .20, for the random-alone condition), F(2, 42) ⫽ 12.84, p ⬍ .001, 2 ⫽ .40. Responses in the correct– random condition significantly differed from responses in both random conditions, although the random conditions did not differ from each other ( p ⬍ .05, Newman–Keuls). Participants in neither random condition answered at chance levels, as a test value of .50 was significantly exceeded by the correctness of both random– random participants (M ⫽ .64, SD ⫽ .22), t(14) ⫽ 2.51, p ⬍ .03, and random-alone participants (M ⫽ .69, SD ⫽ .20), t(14) ⫽ 3.65, p ⬍ .005. These results indicate that participants in the correct– random conditions were answering their first question correctly and that the first response of participants in the random–random condition was like the random-only condition in that participants could not answer randomly. The key issue for this experiment was the random–random participants’ second random response. If merely answering a question randomly once unprimed the answer, we would expect there to be no difference between the correctness of the second responses in the random–random and correct–random conditions. This was not the case, however, as correctness of the second response was significantly greater in the random–random condition (M ⫽ .66, SD ⫽ .20) than in the correct–random condition (M ⫽ .48, SD ⫽ .13), t(28) ⫽ 2.85, p ⬍ .01. Second random response correctness was also significantly greater than a test value of .50, t(14) ⫽ 3.00, p ⬍ .01, whereas random response correctness after correct responses was not, t(14) ⫽ 0.63, p ⫽ .55. A paired comparison between first and second random responses for random–random participants also showed no significant differ-
1013
ence, t(14) ⫽ 0.83, p ⫽ .42. Thus, it appears that answering a question twice, in and of itself, was not sufficient to unprime knowledge of the correct answer.
Experiment 3: Exposure to Correct and Incorrect Answers and Random Response In this experiment, we manipulated exposure to knowledge independent of the participant’s response to determine whether mere knowledge activation might be effective in unpriming. One aspect of answering a question correctly, after all, is simply that the correct answer comes to mind, and we wanted to ascertain whether the participant’s correct response would yield unpriming above and beyond any effect produced by mere external reminding of the knowledge. We were also curious about the influence of exposure to incorrect answers: Might such exposure influence the effectiveness of attempts to respond randomly? To test these influences, we included in this experiment the usual random-only and correct–random conditions but also included other conditions in which participants were supraliminally primed with either the right answer to each question or the wrong answer to each question prior to their attempt to answer the question randomly. These correct-prime–random and incorrect-prime–random conditions allowed us to assess the relative influence of internally generated correct answers and externally provided correct and incorrect answers on random responses.
Method One hundred three participants (61 women and 42 men) were recruited as in the prior experiments and randomly assigned to one of four conditions: the random-only, correct–random, correct-prime–random, and incorrect-prime–random conditions. The procedures for the random-only and correct–random conditions paralleled those of the prior studies. Participants heard the questions presented by the computer through DirectRT as before (Jarvis, 2000), but, in this experiment, the question did not appear on the screen so that correct and incorrect answers for the two answer presentations could appear in the middle of the screen when participants heard the question. Instructions for answering the question appeared at the top of the screen, and yes and no appeared on the bottom of the screen. For the correct-prime–random condition, participants heard each question only once and were asked to answer each question randomly. Included in the task instructions shown before the question trials began was a request for participants to “please keep your eyes on the computer screen at all times.” This was because as each question was being heard through the speakers, the correct answer (either yes or no) to each question was displayed in the middle of the screen for 1,000 ms. In the incorrect-prime–random condition, participants heard each question only once and were asked to answer each question randomly. Participants received the same request to “please keep your eyes on the computer screen at all times.” As each question was being heard through the speakers, the incorrect answer to the question (either yes or no) was displayed in the middle of the screen for 1,000 ms.
Results and Discussion There was a main effect of condition on mean proportion correct, F(3, 99) ⫽ 8.93, p ⬍ .001, 2 ⫽ .21. Participants who answered questions correctly first in the correct–random condition showed a significantly lower (M ⫽ .47, SD ⫽ .07) mean proportion correct on random responses than did participants who answered randomly only (M ⫽ .59, SD ⫽ .19), randomly with the correct answer prime (M ⫽
1014
SPARROW AND WEGNER
.59, SD ⫽ .18), and randomly with the incorrect answer prime (M ⫽ .70, SD ⫽ .18), p ⬍ .05 in each case (Newman–Keuls; see Figure 2). Participants with the incorrect answer prime showed a significantly greater mean proportion correct on random response (M ⫽ .70, SD ⫽ .18) than participants in every other condition, p ⬍ .05 in each case (Newman–Keuls). The random-only (M ⫽ .59, SD ⫽ .19) and correct-prime–random conditions (M ⫽ .59, SD ⫽ .18) did not differ significantly, indicating that simple exposure to the correct answer does not yield any notable degree of unpriming. This observation suggests that the mere thought of the correct response is not sufficient to initiate unpriming and that the actual response may be required to produce an unpriming effect. Random responses in all conditions were also significantly greater than a test value of .50: For the random-only condition, t(25) ⫽ 2.51, p ⬍ .02; for the correct–random condition, t(26) ⫽ 2.38, p ⬍ .05; for the correct-prime–random condition, t(24) ⫽ 5.70, p ⬍ .001; for the incorrect–prime–random condition, t(25) ⫽ 2.57, p ⬍ .02. Correctness for random responses following correct responses (M ⫽ .47) was not significantly different from .50. It is interesting that providing the incorrect answers to participants increases their subsequent tendency to provide nonrandom, correct responses to questions. It may be that when participants see the incorrect answer, their own knowledge of the correct answer is made particularly salient and more likely to defeat their subsequent attempt to answer the question randomly. To be sure, the exposure to the incorrect answer does not unprime the answer. Rather, such exposure may prompt motivations such as psychological reactance (Brehm, 1966) or a desire to suppress thoughts of the incorrect answer (Wegner, 1994; Wenzlaff & Wegner, 2000) that may then increase the influence of the known correct answer on subsequent attempts to produce random responses.
Experiment 4: Components of Correct Answering and Random Response This experiment was designed to determine whether some subcomponent of correctly answering a question suffices to
produce the unpriming of the answer. To understand such a subcomponent, it is useful to recognize that the correct answering of a question has several consequences in this experimental context. The most encompassing consequence is that the correct answer is a full communication from the participant to the experimenter. In such full communication, the participant expresses the answer, this expression communicates the content of the answer to the experimenter, and the communication also serves as a self-presentation to the experimenter that the participant indeed knows the answer. By this analysis, full communication might not be necessary for unpriming, because either the self-presentation of knowledgeableness (without communication of the answer to the experimenter) or the mere expression of the answer for oneself (without communication of the answer to the experimenter and also without selfpresentation of knowledgeableness to the experimenter) might be sufficient to produce the unpriming effect. This study was designed to determine whether full communication; self-presentation; or, at minimum, mere expression of the answer for oneself is sufficient for the unpriming of knowledge in the random answering paradigm.
Method Participants. Eighty-one participants were recruited as in the prior experiments. A programming error rendered results for 3 incomplete, so the final sample consisted of 78 participants (52 women and 26 men). Design and procedure. Participants were run in one of four conditions. Two of these replicated Experiment 1: a random-only condition and a correct–random condition that involved full communication of the participants’ answers to the experimenter. The additional conditions of selfpresentation and mere expression involved decompositions of the correct– random condition that limited communication of the participants’ answers to the experimenter. In the self-presentation condition, participants were shown each question twice in succession. For the first presentation, participants were asked to indicate whether they knew the answer to the question. On the bottom of the screen for the first presentation of each question
Figure 2. Mean proportions correct for participants in the random-only, correct–random, correct-prime–random, and incorrect-prime–random conditions in Experiment 3. Error bars represent standard errors of the mean.
UNPRIMING were the response options Know and Don’t Know corresponding to the keys to be pressed. For the second presentation, participants were asked to respond randomly, following the usual random response instructions. For this presentation, the bottom of the screen indicated the response options yes and no. In the mere-expression condition, participants were also shown each question twice in a row. For both the first and the second presentations, on the bottom of the screen were the response options yes and no. For the first presentation of the question, participants were asked to press underneath the table below the keys designated yes and no to answer the question correctly (but privately in a way only they would know). They were then asked to press the space bar to move on to the second presentation of each question. For the second presentation, participants were asked to answer the question randomly using the keys designated yes and no on the computer keyboard, following the usual random response instructions. Participants in all conditions answered 56 questions. To make the self-presentation issue (of whether the participant knew the answer to the question) a bit more equivocal than in the prior studies, these questions included both easy items and hard ones. Forty questions were easy (e.g., “Does 2 plus 3 equal 5?” “Does a triangle have 4 sides?”) and 16 were hard (e.g., “Are more babies born in February than in any other month?” “Did Alfred Hitchcock eat meat?”). The correct answer was “yes” for half of the total questions and “no” for the other half.
Results and Discussion Responses were computed for easy questions only. (The experimental groups did not differ from one another significantly in the mean proportion of hard questions answered correctly.) Analysis of response correctness during the random answering portion of the experiment revealed a significant main effect for condition, F(3, 74) ⫽ 3.67, p ⬍ .02, 2 ⫽ .13. Participants in the randomonly condition were less able to overcome primed correctness (M ⫽ .62, SD ⫽ .16) than were those in either the correct–random (communication) condition (M ⫽ .47, SD ⫽ .11) or the mereexpression condition (M ⫽ .51, SD ⫽ .17), both ps ⬍ .05 (Newman–Keuls). Participants in the random-only condition (M ⫽ .62, SD ⫽ .16) were also marginally more correct than were self-presentation participants (M ⫽ .55, SD ⫽ .16), p ⬍ .07, and participants in the self-presentation condition were marginally
1015
more correct (M ⫽ .55, SD ⫽ .16) than were those in the correct– random (communication) condition (M ⫽ .47, SD ⫽ .11), p ⬍ .07 (see Figure 3). Mean correctness during random responding in the various conditions was also compared with a test value of .50. Random responses in the random-only condition were significantly more correct than a test value of .50, t(21) ⫽ 3.55, p ⬍ .003. Correctness levels were not significantly greater than .50 for random response in any of the conditions in which participants offered a prior expression of the answer: correct–random, t(17) ⫽ 1.24, p ⫽ .24; self-presentation, t(19) ⫽ 1.30, p ⫽ .20; mere expression, t(18) ⫽ .27, p ⫽ .79. These results thus do not provide a strong conclusion regarding the influence of differing forms of expression on the effectiveness of unpriming. It appears that participants in the two groups who provided a correct answer (the correct–random and mereexpression conditions), whether the answer was conveyed to the experimenter through the computer or simply expressed privately through pressing their fingers under the table, were more successful at unpriming knowledge-based random responses than were participants in the other groups, although this comparison was only marginally significant for the mere-expression group. The actual correct answer should be expressed, whether to the self or to the experimenter, to provide maximum unpriming of activated knowledge. The action used to fully deactivate a prime may have to be specific to the knowledge activated—such as a report of that knowledge—rather than a more global indication simply that the knowledge is known. Unpriming imparted by a correct response was not merely an issue of communication and presentation to the experimenter. The lowest level of expression tested—simply expressing knowledge privately to oneself—showed a tendency to remove the biasing impact of answer knowledge for subsequent random responses. The next experiment was designed to examine two possibilities raised by these results: whether each individual prime needs to be specifically deactivated and whether, as suggested by the somewhat reduced correctness levels in the self-presentation condition, a general display of knowledge might be enough to yield unpriming.
Figure 3. Mean proportions correct for participants in the random-only, correct–random, self-presentation, and mere-expression conditions in Experiment 4. Error bars represent standard errors of the mean.
1016
SPARROW AND WEGNER
Experiment 5: Correctly Answering a Different Question and Random Response It may be that the opportunity to answer some questions correctly and thus unprime the activation created by knowledge will generalize to other questions, allowing a general unpriming effect. This possibility was suggested by the (marginal) unpriming afforded in the prior experiment by the self-presentation of knowledgeableness. However, it may be that deactivation has to be specific for each subsequent random response. Each answer generated by each question may yield the priming of that answer knowledge, and each such prime may thus need to be to be deactivated through expression to allow for an overall proportion correct to be at chance or unprimed levels. This experiment gauged whether unpriming in the random answering paradigm is general or item specific.
Method Forty participants (25 women and 15 men) recruited as in the prior studies answered the series of 60 easy questions with “yes” or “no” responses as in Experiment 1. Participants were run in one of two conditions: a random-only condition, conducted as in the prior studies, and a correct-unrelated-random condition. For participants in the correctunrelated-random condition, for the presentation of the first 30 questions (Questions 1–30), an instruction appeared at the top of the screen asking participants to “try to answer the question correctly.” Participants answered the first 30 questions correctly but did not also answer these same questions randomly. Subsequently, for the second 30 questions (Questions 31– 60), participants were instructed to “try to answer the question randomly.” There were an equal number of “yes” and “no” correct responses across the two blocks, presented in the same randomly determined order as in previous experiments. Participants in the random-only condition answered 30 questions (Questions 31– 60), the same 30 questions to which participants in Condition 1 responded randomly. They received detailed instructions for how to answer randomly prior to beginning the task, as did participants in the previous studies. With each question presentation, participants in the random-only condition received the instruction to “try to answer randomly” on the top of the screen. Thus, participants in both conditions responded randomly to the same questions, but participants in the correct-unrelated-random condition responded correctly to other questions first.
Results and Discussion Random responses from participants who had previously answered other questions correctly (M ⫽ .57, SD ⫽ .19) and from participants who answered questions randomly only (M ⫽ .61, SD ⫽ .18) did not significantly differ from one another in mean proportion correct, t(38) ⫽ 0.92, p ⫽ .36. However, the mean for each group of participants was significantly different from a test statistic of .50 (the mean proportion expected from random response) when compared individually: For the correct-unrelatedrandom condition, t(19) ⫽ 2.02, p ⬍ .05; for the random-only condition, t(19) ⫽ 2.54, p ⬍ 03. Thus, it appears that answering unrelated questions correctly does not lead to successful random responding any more than answering randomly only does. Deactivation of a knowledge prime may need to be specific before knowledge can be overcome. This finding also suggests that the self-presentation of knowledgeableness is not critical for the production of unpriming. Participants in the correct-unrelated-random condition achieved the usual high rate of correctness in answering
the first 30 questions (M ⫽ .96), but this display of knowledgeableness did not significantly reduce their knowledge-primed responding to the subsequent different set of questions.
General Discussion In these studies, we found that expressing the answer to a question can help a person overcome the unwanted influence of that answer on subsequent responding. In each study, we examined such influence in the random answering paradigm: People were asked to make random responses to simple yes–no questions, for which the correctness of the two answers had been balanced at 50%. Correct responses indicating bias toward the known answer occur commonly and appear to be uncontrollable (Wegner et al., 2003), but, in our studies, this bias was easily overcome when participants followed instructions to answer each question correctly before attempting to give their random answer. In Experiment 1, participants who were allowed to answer easy yes–no questions correctly first before answering each one randomly had a significantly lower mean proportion correct for random responses than did participants who responded randomly alone. This unpriming effect was not attributable to mere repetition of answering, as participants in Experiment 2 who answered each question randomly twice in a row exhibited mean proportions correct that were significantly greater than chance for both first and second random responses. These proportions were comparable to single random responses and significantly greater than random responses that occurred after correct responses. Experiment 3 exposed participants to right and wrong answers provided by the computer to see whether such exposure might underlie the unpriming produced by correct responding. Having the computer supply the right answer to each question with a supraliminal prime, however, was not sufficient to unprime answer knowledge in subsequent random responding. External generation of the incorrect answer even had the curious effect of enhancing correctness during subsequently attempted random answering rather than acting to unprime the answer and decrease correctness. In Experiment 4, we examined three versions of correct answering to see what the minimal circumstances might be to produce unpriming. The study was designed to test whether unpriming requires full communication of the question’s answer, only the self-presentation that one knows the answer (not the content of the answer itself), or only the mere expression of the answer (without any communication of the answer or self-presentation of knowledgeableness). In line with the prior experiments, full communication of correct answers (to the experimenter via the computer) successfully unprimed answer knowledge. However, mere expression of the answer to oneself (by tapping beneath the desk) had the same unpriming effect. Participants who reported that they had the knowledge of the correct responses (self-presentation) but did not use the knowledge itself by reporting the correct answer achieved a modicum of unpriming but were only marginally more successful at answering randomly than were participants given no unpriming manipulation. This moderate success, however, suggested that self-presentation of knowledgeableness might play some role in unpriming knowledge, so Experiment 5 was conducted to examine the influence of selfpresentation of knowledge more completely. In this study, it was found that each piece of activated knowledge needs to be expressed
UNPRIMING
specifically, through answering the same questions correctly prior to answering randomly, for random responses to be free of answerknowledge influence. Answering some questions correctly first and then other questions randomly does not provide a generalized expression of correctness or self-presentation of knowledgeableness that allows subsequent behavior to be free of the influence of answer knowledge. Taken as a whole, these findings indicate that the expression of the answers to questions can often eliminate the influence of answer knowledge on a later response. This is not a trivial achievement, because people who are given time and incentives to try to eliminate such influence cannot do it voluntarily (Wegner et al., 2003). Our observations of unpriming suggest that it is important to consider how unpriming may operate, how general such effects may be beyond the random answering paradigm, and how useful these results may be for psychological research or application.
Explanations of Unpriming Our introductory comments on the variety of concepts that resemble unpriming in the history of psychology suggest that settling on one satisfactory explanatory framework for the present findings may be something of a challenge. Does unpriming result from processes of catharsis, completion, updating, or yet something else? The present results do not arbitrate among these broad classes of explanation, as our studies were conducted to establish the characteristics of the unpriming phenomenon rather than to test explanations for it. However, the findings do offer some helpful insight that can inform attempts at explanation. For example, one of our initial ideas about the unpriming effect was to explain it in terms of Grice’s (1975) principle of cooperation in conversation. According to this principle, when a speaker asks a question of a listener, the listener normally tries to cooperate by making the conversational contribution that is required—in this case, answering the question. It is interesting, after all, how powerfully a question brings to mind an answer, whether one is actively interested in providing the answer or not (Swann, Giuliano, & Wegner, 1982; Wegner, Wenzlaff, Kerker, & Beattie, 1981). This principle suggests that once a correct answer has been given, the impetus to continue cooperation is eliminated and any requirement to rehearse the answer or hold it in mind is relaxed. Our finding that unpriming occurs even when the participant merely expresses the answer privately (Experiment 4), however, suggests that analyses of unpriming based in norms of conversation might not be fruitful. Another way of conceptualizing unpriming would be to say that the self-presentation of knowledge is important for the effect. Correctly answering a question involves the self-presentation of knowing, and the reduction of this concern due to providing the correct answer might create unpriming by allowing the person to move on to other concerns. This possibility was given some marginal support by our finding that simply reporting that one knows the answer can be partially effective in unpriming the answer (Experiment 4). The self-presentation hypothesis was undermined, however, by the finding that correct answering only unprimed that specific answer rather than producing a general sense of knowledgeableness that could unprime yet other answers (Experiment 5). The finding of Wegner et al. (2003) that random answering was just as biased by knowledge in a sample of partic-
1017
ipants drawn from outside a university setting as it was among participants in a university also suggests that the self-presentation of knowledgeableness is not an attractive explanation for effects in this paradigm. Two other accounts of unpriming were noted earlier. These included the idea that expression satisfies a Zeigarnik-like motive instantiated by the question to express the correct answer and the idea that expression might reduce the effort the person must exert to keep the correct answer out of mind and therefore reduce the suppression-induced activation of the answer. The observed findings are generally consistent with both of these accounts and so do not aid in determining which might be a more satisfactory explanation.
Generality of Unpriming Effects To what degree might the unpriming effects observed in these studies portend similar phenomena in other priming paradigms? Perhaps the most direct parallels might be found in other paradigms that yield knowledge-priming effects people find difficult to control. The difficulty of overcoming the Stroop (1935) interference effect has been well documented (MacLeod, 1991), for example, and it is also widely appreciated that responses to the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998) are hard to control. Might some expression of the uncontrollable knowledge in these paradigms release respondents from the biases that normally influence their responses? Expression after a period of suppression has been shown to reduce the suppressed thought’s subsequent accessibility (Liberman & Fo¨rster, 2000), and Fo¨rster et al. (2005) found that goalrelated words are enhanced over non– goal-related words, but, over time, this accessibility is diminished once the goal has been achieved. One may unprime a thought without an explicit intention or goal, as may be seen in the simple satiation effects found by Smith (Smith, 1984; Smith & Klein, 1990), although the expression takes many trials before unpriming is achieved. It is not clear that expression or action would necessarily have the same influence in these paradigms or that it might serve the purpose of unpriming in other behavior-priming paradigms (e.g., Bargh, Chen, & Burrows, 1996). What differentiates priming effects that linger for days or weeks from the unpriming effects found in our studies? Tulving, Schacter, and Stark (1982) showed that recognition memory was diminished for participants studied 7 days later as opposed to 1 hr later but that priming effects lingered. What if the participants had completed the same word fragments both 1 hr later and 7 days later in a within-subject design? If the unpriming effects found in the random response paradigm were to generalize to direct priming, the already completed word fragments would be expected to be unprimed. The random answering paradigm presents a unique situation in which knowledge expression acts immediately to eliminate an otherwise hard-to-overcome priming effect of prior knowledge, and it is unclear whether expressive action might have similar effects across other priming paradigms. To the degree that our general account of unpriming is correct—and people indeed need some way to overcome the influence of primes if they are to move from one behavior setting to another and not get stuck in a primed behavior loop—it may well be that expression of some kind could foster unpriming in many circumstances.
SPARROW AND WEGNER
1018
Some limits on the generality of unpriming are suggested by our results. Unpriming does not occur when the action merely has surface features that resemble those of the knowledge-expressing action— merely responding in a random way, for example, does not eliminate subsequent bias toward knowledge (Experiment 2). Unpriming seems to require action based on the prime, not only reexposure to the priming influence (Experiment 3). Unpriming seems to require, at a minimum, some private expression of the knowledge (Experiment 4), and actions that will unprime knowledge must be specific to the knowledge (Experiment 5). Our findings thus circumscribe the generality of unpriming in a number of ways.
Implications and Applications How could unpriming serve to reduce unwanted knowledge influences in everyday life or in conditions of psychopathology? People are often drawn to the rehearsal of undesired thoughts or to the intrusive recurrence of images or ideas that they cannot control (Clark, 2005). It may be that there is some role for procedures like unpriming in therapies designed to help people overcome such unwanted thoughts. Psychotherapeutic approaches based on exposure to unpleasant memories (Foa & Meadows, 1997), expression of traumatic experiences (Pennebaker, 1997), or acceptance of difficult circumstances (Hayes, Strosahl, & Wilson, 1999) might be useful because they encourage people to address unwanted thoughts by expressing these thoughts to themselves or to others. Expression might also aid people in overcoming unwanted prejudices, perhaps unpriming knowledge that is held in mind but that is inconsistent with the person’s avowed explicit attitudes (Monteith, Sherman, & Devine, 1998; Sherman, in press). The range of potential uses for unpriming is substantial because behavior primed by a person’s own knowledge may not always be the kind of behavior that person wants to perform.
References Allport, D. A., Styles, E. A., & Hsieh, S. (1994). Shifting intentional set: Exploring the dynamic control of tasks. In C. Umilta` & M. Moscovitch (Eds.), Attention and performance XV: Conscious and nonconscious information processing (pp. 421– 452). Cambridge, MA: MIT Press. Aristotle. (1961). Poetics (S. H. Butcher, Trans.). New York: Hill & Wang. Atkinson, J., Bongort, K., & Price, L. H. (1977). Explorations using computer simulation to comprehend thematic apperceptive measurement of motivation. Motivation & Emotion, 1, 1–27. Baddeley, A., Emslie, H., Kolodny, J., & Duncan, J. (1998). Random generation and the executive control of working memory. Quarterly Journal of Experimental Psychology: Human Experimental Psychology, 51(A), 819 – 852. Bargh, J. A., Chaiken, S., Raymond, P., & Hymes, C. (1996). The automatic evaluation effect: Unconditionally automatic attitude activation with a pronunciation task. Journal of Experimental Social Psychology, 32, 185–210. Bargh, J. A., & Chartrand, T. L. (2000). The mind in the middle: A practical guide to priming and automaticity research. In H. Reis & C. Judd (Eds.), Handbook of research methods in social psychology (pp. 253–285). New York: Cambridge University Press. Bargh, J., A. Chen, M., & Burrows, L. (1996). The automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230 –244. Bargh, J., & Ferguson, M. J. (2000). Beyond behaviorism: On the automaticity of higher mental processes. Psychological Bulletin, 126, 925–945.
Bjork, R. A., & Landauer, T. K. (1978). On keeping track of the present status of people and things. In M. M. Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical aspects of memory (pp. 52– 60). London: Academic Press. Blaxton, T. A. (1989). Investigating dissociations among memory measures: Support for a transfer-appropriate processing framework. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 657– 668. Brehm, J. W. (1966). A theory of psychological reactance. New York: Academic Press. Breuer, J., & Freud, S. (1955). Studies on hysteria. In J. Strachey (Ed. & Trans.), The standard edition of the complete psychological works of Sigmund Freud (Vol. 2). London: Hogarth Press. (Original work published 1893–1895) Bushman, B. J., Baumeister, R. F., & Stack, A. D. (1999). Catharsis, aggression, and persuasive influence: Self-fulfilling or self-defeating prophecies? Journal of Personality and Social Psychology, 76, 367–376. Carver, C. S., Ganellen, R. J., Froming, W. J., & Chambers, W. (1983). Modeling: An analysis in terms of category accessibility. Journal of Experimental Social Psychology, 19, 403– 421. Chartrand, T. L., & Bargh, J. A. (1996). Automatic activation of impression formation and memorization goals: Nonconscious goal priming reproduces effects of explicit task instructions. Journal of Personality and Social Psychology, 71, 464 – 478. Clark, D. A. (Ed.). (2005). Intrusive thoughts in clinical disorders: Theory, research, and treatment. New York: Guilford Press. Dijksterhuis, A., & Bargh, J. A. (2001). The perception– behavior expressway: Automatic effects of social perception on social behavior. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 33, pp. 1– 40). San Diego, CA: Academic Press. Dijksterhuis, A., & van Knippenberg, A. (1998). The relation between perception and behavior, or how to win a game of Trivial Pursuit. Journal of Personality and Social Psychology, 74, 865– 877. Entwistle, D. R. (1972). To dispel fantasies about fantasy-based measures of achievement motivation. Psychological Bulletin, 77, 377–391. Epley, N., & Gilovich, T. (1999). Just going along: Nonconscious priming and conformity to social pressure. Journal of Experimental Social Psychology, 35, 578 –589. Esposito, N. J., & Pelton, L. H. (1971). Review of the measurement of semantic satiation. Psychological Bulletin, 75, 330 –346. Foa, E. B., & Meadows, E. A. (1997). Psychosocial treatments for posttraumatic stress disorder: A critical review. Annual Review of Psychology, 48, 449 – 480. Fo¨rster, J., Liberman, N., & Higgins, E. T. (2005). Accessibility from active and fulfilled goals. Journal of Experimental Social Psychology, 41, 220 –239. Gilbert, D. T. (1991). How mental systems believe. American Psychologist, 46, 107–119. Gollwitzer, P. M., & Wicklund, R. A. (1982). Admission of failure and symbolic self-completion: Extending Lewinian theory. Journal of Personality and Social Psychology, 43, 358 –371. Gollwitzer, P. M., & Wicklund, R. A. (1985). Self-symbolizing and the neglect of others’ perspectives. Journal of Personality and Social Psychology, 48, 702–715. Goschke, T., & Kuhl, J. (1993). Representation of intentions: Persisting activation in memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1211–1226. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464 –1480. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Sytnax and semantics: Vol. 3. Speech acts (pp. 41–58). New York: Academic Press. Hayes, S. C., Strosahl, K., & Wilson, K. G. (1999). Acceptance and commitment therapy: An experiential approach to behavior change. New York: Guilford Press.
UNPRIMING Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins & A. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133–168). New York: Guilford Press. Higgins, E. T., & King, G. (1981). Accessibility of social constructs: Information processing consequences of individual and contextual variability. In N. Cantor & J. Kihlstrom (Eds.), Personality, cognition, and social interaction (pp. 69 –121). Hillsdale, NJ: Erlbaum. Jacoby, L. L., & Dallas, M. (1981). On the relationship between autobiographical memory and perceptual learning. Journal of Experimental Psychology: General, 3, 306 –340. Jarvis, B. G. (2000). DirectRT research software (Version 2000) 关Computer software兴. New York: Empirisoft. Lewin, K. (1939). Will and needs. In W. D. Ellis (Ed.), A source book of Gestalt psychology (pp. 283–299). New York: Harcourt, Brace. Lewin, K. (1951). Behavior and development as a function of the total situation. In D. Cartwright (Ed.), Field theory in social science: Selected theoretical papers (pp. 791– 843). New York: Harper & Row. Liberman, N., & Forster, J. (2000). Expression after suppression: A motivational explanation of postsuppressional rebound. Journal of Personality and Social Psychology, 79, 190 –203. MacLeod, C. M. (1991). Half a century of research on the Stroop effect: An integrative review. Psychological Bulletin, 109, 163–203. Macrae, C. N., & Johnston, L. (1998). Help, I need somebody: Automatic action and inaction. Social Cognition, 16, 400 – 417. Marsh, R. L., Hicks, J. L., & Bink, M. L. (1998). Activation of completed, uncompleted, and partially completed intentions. Journal of Experimental Psychology: Learning, Memory, and Cognition, 24, 350 –361. Martin, L. L., Tesser, A., & McIntosh, W. D. (1993). Wanting but not having: The effects of unattained goals on thoughts and feelings. In D. M. Wegner & J. W. Pennebaker (Eds.), The handbook of mental control (pp. 552–572). New York: Prentice Hall. Mayr, U., & Keele, S. W. (2000). Changing internal constraints on action: The role of backward inhibition. Journal of Experimental Psychology: General, 129, 4 –26. Monteith, M. J., Sherman, J. W., & Devine, P. G. (1998). Suppression as a stereotype control strategy. Personality and Social Psychology Review, 2, 63– 82. Morgan, C. D., & Murray, H. A. (1935). A method of investigating fantasies: The Thematic Apperception Test. Archives of Neurology and Psychiatry, 34, 289 –306. Mulligan, N. W. (1997). Attention and implicit memory tests: The effects of varying attentional load on cognitive priming. Memory & Cognition, 25, 11–17. Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-category attention. Journal of Experimental Psychology: General, 106, 226 –254. Nickerson, R. S. (2002). The production and perception of randomness. Psychological Review, 109, 330 –357. Pennebaker, J. W. (1989). Confession, inhibition, and disease. Advances in Experimental Social Psychology, 22, 211–244. Pennebaker, J. W. (1997). Writing about emotional experiences as a therapeutic process. Psychological Science, 8, 162–166. Pennebaker, J. W., & Beall, S. K. (1986). Confronting a traumatic event: Toward an understanding of inhibition and disease. Journal of Abnormal Psychology, 95, 274 –281. Posner, M. I., & Cohen, Y. P. C. (1984). Components of visual orienting. In H. Bouma & D. Bouwhuis (Eds.), Attention and performance X: Control of language processes (pp. 531–556). Hillsdale, NJ: Erlbaum. Roediger, H. L., & McDermott, K. B. (1993). Implicit memory in normal
1019
human subjects. In F. Boller & J. Grafman (Eds.), Handbook of neuropsychology (Vol. 8, pp. 63–131). Amsterdam: Elsevier. Schacter, D. L., & Buckner, R. L. (1998). Priming and the brain. Neuron, 20, 185–195. Schacter, D. L., Chiu, C. Y. P., & Ochsner, K. N. (1993). Implicit memory: A selective review. Annual Review of Neuroscience, 16, 159 –182. Segal, S. J., & Cofer, C. N. (1960). The effects of recency and recall on free association. American Psychologist, 15, 451. Sherman, J. W. (in press). Controlled influences on implicit measures: Confronting the myth of process-purity and taming the cognitive monster. In R. E. Petty, R. H. Fazon, & P. Brinol (Eds.), Attitudes: Insights from the new wave of implicit measures. Mahwah, NJ: Erlbaum. Smith, L. C. (1984). Semantic satiation affects category membership decision time but not lexical priming. Memory & Cognition, 12, 483– 488. Smith, L. C., & Klein, R. (1990). Evidence for semantic satiation: Repeating a category slows subsequent semantic processing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16, 852– 861. Squire, L. R., Shimamura, A. P., & Graf, P. (1987). Strength and duration of priming effects in normal subjects and amnesic patients. Neuropsychologia, 25, 195–210. Storms, L. H. (1958). Apparent backward association: A situational effect. Journal of Experimental Psychology, 55, 390 –395. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 12, 242–248. Swann, W. B., Jr., Giuliano, T., & Wegner, D. M. (1982). Where leading questions can lead: The power of conjecture in social interaction. Journal of Personality and Social Psychology, 44, 1025–1035. Toth, J. P., & Reingold, E. M. (1996). Beyond perception: Conceptual contributions to unconscious influences of memory. In G. Underwood (Ed.), Implicit cognition (pp. 41– 84). Oxford, England: Oxford University Press. Tulving, E., Schacter, D. L., & Stark, H. A. (1982). Priming effects in word-fragment completion are independent of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 336 –342. Wegner, D. M. (1994). Ironic processes of mental control. Psychological Review, 101, 34 –52. Wegner, D. M., Coulton, G., & Wenzlaff, R. (1985). The transparency of denial: Briefing in the debriefing paradigm. Journal of Personality and Social Psychology, 49, 382–391. Wegner, D. M., Fuller, V., & Sparrow, B. (2003). Clever hands: Uncontrolled intelligence in facilitated communication. Journal of Personality and Social Psychology, 85, 5–19. Wegner, D. M., Schneider, D. J., Carter, S., & White, T. (1987). Paradoxical effects of thought suppression. Journal of Personality and Social Psychology, 53, 5–13. Wegner, D. M., Wenzlaff, R. M., Kerker, R. M., & Beattie, A. E. (1981). Incrimination through innuendo: Can media questions become public answers? Journal of Personality and Social Psychology, 40, 822– 832. Wenzlaff, R. M., & Wegner, D. M. (2000). Thought suppression. Annual Review of Psychology, 51, 51–91. Zeigarnik, B. (1935). On finished and unfinished tasks. In K. Lewin (Ed.), A dynamic theory of personality (pp. 300 –314). New York: McGrawHill.
Received September 3, 2005 Revision received March 11, 2006 Accepted March 28, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1020 –1031
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1020
Framing Discrimination: Effects of Inclusion Versus Exclusion Mind-Sets on Stereotypic Judgments Kurt Hugenberg
Galen V. Bodenhausen
Miami University
Northwestern University
Melissa McLain University of Southern California Three studies investigated how inclusion versus exclusion strategies differentially lead to stereotypic decisions. In inclusion strategies, suitable targets are selected from a list of candidates, whereas in exclusion strategies, unsuitable candidates are eliminated. Across 2 separate target domains (Study 1: male and female politicians; Studies 2 and 3: African American and European American basketball players), exclusion strategies, as compared with inclusion strategies, elicited higher levels of both sensitivity stereotyping (i.e., greater difficulty distinguishing among members of stereotyped groups) and criterion stereotyping (i.e., setting different decision thresholds for judging members of different groups; see M. R. Banaji & A. G. Greenwald, 1995). Thus, the strategy used during decision making can influence the final decision via 2 theoretically distinct stereotyping mechanisms. Keywords: stereotypes/stereotyping, decision making, task framing, mind-sets, inclusion– exclusion discrepancy
two strategies should result in a final choice set of individuals who are deemed well qualified with respect to the relevant criteria. Indeed, eliminating everyone who is not well qualified should produce the same final choice set as selecting everyone who is well qualified. Normatively, these strategies should be the perfect converse of one another, leading to identical choice sets; however, an increasing body of research indicates that decision makers make substantially different decisions under exclusion and inclusion mind-sets. Perhaps the most robust difference in decision making is that individuals in an exclusion mind-set tend to produce substantially larger final choice sets than do individuals in an inclusion mindset, an effect known as the inclusion– exclusion discrepancy (IED; Levin, Huneke, & Jasper, 2000; Levin, Jasper, & Forbes, 1998; Yaniv & Schul, 1997, 2000; Yaniv, Schul, Raphaelli-Hirsch, & Maoz, 2002). In one demonstration of the IED, Yaniv et al. (2002) had Israeli participants predict which of the many Israeli political parties would gain seats in the Knesset in an upcoming election. Yaniv and colleagues showed participants a list of political parties that were fielding candidates in the election and manipulated whether participants were required to (a) select the parties that they believed would gain seats in the election (i.e., an inclusion mindset) or (b) eliminate those parties that they believed would not gain seats in the election (i.e., an exclusion mind-set). In line with mounting evidence for the IED, Yaniv et al. found that participants tended to have much larger final choice sets when operating under an exclusion rather than an inclusion mind-set. According to Yaniv and Schul (1997, 2000), the IED arises from a general tendency toward inaction under uncertainty (see also Heller et al., 2002), with the discrepancy arising because inaction in inclusion and exclusion mind-sets has differing consequences. For individuals in an inclusion mind-set, failing to act on an option
A substantial literature attests to the fact that decision makers often arrive at substantially different kinds of judgments depending on how a choice option is framed (Ku¨hberger, 1998; Tversky & Kahneman, 1986). For example, if a medical treatment is said to have a 25% mortality rate, it will be judged less favorably than if it is said to have a 75% survival rate. When stated abstractly, the logical equivalence of these two alternative frames is obvious, but psychologically, the different descriptions produce reliably different perceptions and different choices. It is also possible to frame a decision-making process in different ways. For example, every year, graduate programs receive a glut of applications for a limited number of positions, and every year, admissions committees must decide how to reduce a large pool of applicants to a final list of accepted candidates. There are two quite different ways to approach this potentially daunting task of reducing a relatively large set of options to a small, delimited choice set (Heller, Levin, & Goransson, 2002). One strategy is to select all of the well-qualified candidates from the broader pool of applicants; this approach constitutes an inclusion strategy. An alternative strategy is to eliminate all candidates who are not well qualified, with the remaining individuals constituting the chosen set; this approach constitutes an exclusion strategy. Either of these
Kurt Hugenberg, Department of Psychology, Miami University; Galen V. Bodenhausen, Department of Psychology, Northwestern University; Melissa McLain, Rossier School of Education, University of Southern California. We are grateful to Bill von Hippel for helpful comments on an earlier version of this article. Correspondence concerning this article should be addressed to Kurt Hugenberg, Miami University, Department of Psychology, Oxford, OH 45056. E-mail:
[email protected] 1020
DISCRIMINATION AND MIND-SETS
leads to that particular option not being part of the final choice set. Conversely, for individuals in an exclusion mind-set, the consequence of inaction is a failure to exclude a particular option, leading to that particular option being retained in the final choice set. Yaniv and Schul (2000) found that decision makers are particularly likely not to act on what they called middling cases. For extremely well-qualified or extremely unqualified candidates, there should be no effect of mind-set, but when there is uncertainty, the tendency not to act produces larger final choice sets under exclusion, as compared with inclusion, task framing. In the present studies, we sought to investigate how stereotypes might influence decision makers’ choices under inclusion versus exclusion frames. For example, if candidates are being considered for admission to a graduate program in engineering and if common social stereotypes suggest that women are generally not well qualified to become engineers, would this bias be more evident under an inclusion or an exclusion framing of the choice process? To derive predictions about this question, it was necessary to consider in greater detail the psychological processes at work in the IED. Yaniv et al. (2002) argued that, in signal-detection terms, decision makers set a lower criterion (also known as bias) for membership in the final choice set when in an exclusion mind-set. For example, in their study of the Israeli elections, Yaniv and colleagues were able to check participants’ predictions against the subsequent election results, allowing them to calculate hit and false-alarm rates for participants in both the inclusion and the exclusion mind-sets. Signal-detection analyses indicated that participants set a relatively lax criterion for retaining an option in the final choice set in the exclusion condition, compared with the inclusion condition. However, Yaniv et al. found that inclusion and exclusion mind-sets did not lead to differences in sensitivity for the response options, indicating that task framing had no effect on participants’ capacity to discriminate between parties who would and would not gain seats in the Knesset. In related research, Levin et al. (2000) found evidence that the psychological effects of mind-sets on decision making can go beyond simple differences in criterion setting. Specifically, they found that participants making decisions in inclusion mind-sets tended to return to each of the possible options more frequently and also seriously considered more of the choice options than did participants in the exclusion mind-set. Thus, participants in an inclusion mind-set seemed to engage in deeper deliberation about each of the response options than did participants in an exclusion mind-set. This latter finding suggests that under some conditions, sensitivity differences in decision makers’ choices may in fact be elicited depending upon the mind-set under which they construct their choice set. If inclusion and exclusion mind-sets can indeed elicit different levels of processing depth, then one can readily expect that reliance on stereotypes will also covary with these mind-sets. A great deal of research suggests that stereotypic effects tend to be at their most powerful when motivation and capacity to process social targets (i.e., individuate) are reduced (e.g., Fiske & Neuberg, 1990; Macrae & Bodenhausen, 2000). Thus, insofar as inclusion and exclusion mind-sets have different implications for how deeply targets are processed, the IED may prove to be an interesting domain in which to investigate stereotyping processes.
1021
Detecting Stereotypes Via Signal Detection As previously noted, past work by Yaniv et al. (2002) used signal-detection analysis to examine the processes underlying the IED. Within the stereotyping literature, prior work by Banaji and colleagues (e.g., Banaji & Greenwald, 1995; Park & Banaji, 2000) has used these same signal-detection methods (see Green & Swets, 1974) to explicate the processes underlying stereotyping effects. The sensitivity parameter (d⬘ or A⬘) has been linked to what Banaji and Greenwald (1995) called sensitivity stereotyping, or the tendency for perceivers to have difficulty distinguishing or differentiating among the members of a social group. In other words, this parameter relates to the degree of perceived homogeneity, and thus confusability or interchangeability, of the members of stereotyped groups (e.g., Judd & Park, 1988; Quattrone & Jones, 1980). In contrast, the signal-detection criterion parameter ( or B⬙) has been mapped onto a separate mechanism by which stereotypes operate. Specifically, this criterion stereotyping involves the adoption of different criteria in judging members of different groups (e.g., Biernat, 2003). Consider an example drawn from the work of Park and Banaji (2000), whose experiments focused on the stereotype of African Americans as basketball players. They had participants decide which African American and European American targets were basketball players. In this case, criterion stereotyping manifests as a tendency to set a lower criterion for African Americans to be considered basketball players than for European Americans. Sensitivity stereotyping, however, manifests as a tendency to confuse African Americans who actually have athletic prowess with those who do not. When sensitivity stereotyping is operating, members of a stereotyped group seem more homogeneous and thus are difficult to discriminate from one another. Past work has found that sensitivity stereotyping and criterion stereotyping are responsive to different manipulations in different settings. Thus, not all instances of stereotyping necessarily engage both of these stereotyping processes. For example, in their initial demonstration of criterion stereotyping, Banaji and Greenwald (1995) used a false fame paradigm (see Jacoby, Kelley, Brown, & Jasechko, 1989) in which participants were exposed to a list of male and female, famous and nonfamous names. After a 24-hr delay, participants returned and were provided with a list of twice as many names, including all of the names observed in the first session. The typical finding, which was replicated by Banaji and Greenwald, is that nonfamous male names in the original list were more likely than nonfamous female names to seem famous when reencountered in the second session. Additionally, Banaji and Greenwald used the hits and false alarms to male and female targets in this false fame task to compute indices of sensitivity and criterion stereotyping. In this case, Banaji and Greenwald found that the false fame effect was due to a robust criterion stereotyping effect, but they discovered no reliable evidence of sensitivity stereotyping. Thus, although participants showed no difference in perceptions of homogeneity for male and female targets, they set the bar for deciding a woman was famous higher than they did for male targets. In a more recent demonstration, Park and Banaji (2000) found that both sensitivity stereotyping and criterion stereotyping are at play when perceivers are in happy mood states. In these studies, Park and Banaji presented participants with a list of African American and European American names of professional
HUGENBERG, BODENHAUSEN, AND MCLAIN
1022
basketball players intermingled with African American and European American distractor names. Although participants in a neutral mood showed neither reliable sensitivity stereotyping nor criterion stereotyping, participants in a happy mood showed both forms of stereotyping. Thus, stereotyping effects observed in happy-mood participants were due to a combination of increased perceptions of homogeneity of African Americans and a decrease in participants’ threshold for including African Americans in a stereotype-relevant group (i.e., basketball players). The present studies sought to understand how decision-making mind-sets might promote or undermine these distinct forms of stereotyping. Given the past work on the IED, hypotheses for sensitivity stereotyping seemed clear. The findings of Levin et al. (2000) suggested that inclusion mind-sets facilitate higher levels of elaboration than exclusion mind-sets. Insofar as relatively extensive processing is a necessary precursor for individuation (e.g., Fiske & Neuberg, 1990), we hypothesized that sensitivity stereotyping would be less likely to be observed in inclusion conditions compared with an exclusion mind-set, in which decision makers would appear to process individual targets less extensively. Park and Banaji’s (2000) study used a de facto inclusion mind-set, and they found little evidence of sensitivity stereotyping under neutral mood conditions. However, the psychological tendencies associated with the exclusion mind-set might result in sensitivity stereotyping even in a neutral mood state. Our hypotheses for criterion stereotyping, however, were more tentative. The processing depth effects of the different mind-sets did not necessarily have clear implications for criterion stereotyping, making our investigation of the effects of mind-set on criterion stereotyping a more exploratory one. The findings of Yaniv et al. (2002) suggested that decision makers set a lower criterion for membership in the target category when operating under an exclusion mind-set. On the basis of their principle of inaction under uncertainty, one might expect this criterion shift to be a general tendency that operates under an exclusion mind-set independently of any additional effect of criterion stereotyping that might be observed. Alternatively, it could be that exclusion mind-sets might also elicit a relatively high degree of criterion stereotyping, as compared with the inclusion mind-sets. Study 1 was designed to test the extent to which inclusion and exclusion mind-sets differentially engaged sensitivity and criterion stereotyping.
Study 1 In Study 1, participants were presented with a list of names and were asked to determine which ones were politicians. The names on this list included both actual politicians and nonpoliticians, and both of these subgroups included both male and female names. Women are culturally viewed as being better suited for communal roles rather than agentic roles (Eagly & Karau, 2002; Eagly, Wood, & Diekman, 2000). As such, it is counterstereotypic for women to hold occupations that are strongly associated with agentic qualities, such as politicians and civic leaders. To determine whether a man has the qualities necessary for politics, one must consider the individual’s personal characteristics and attributes, whereas the stereotype of women permits perceivers to categorically assume that women are generally unsuitable for this kind of agentic role. In the present context, sensitivity stereotyping would be evident if people were less able to differentiate politi-
cians from nonpoliticians when considering female (compared with male) names, and criterion stereotyping would be evident if a higher criterion for membership in the category “politicians” were set for female as opposed to male names (see Banaji & Greenwald, 1995). On the basis of the findings of Levin et al. (2000) showing more extensive processing of the choice set under inclusion conditions, we expected to find greater sensitivity stereotyping under an exclusion mind-set than under an inclusion mind-set. We also expected to replicate the general IED effect and sought to determine whether it would be based on the establishment of a lower criterion in the exclusion condition and, if so, whether this criterion shift would interact with target sex at all.
Method Participants and Design One hundred nineteen undergraduates (67 female) from Miami University participated in this research. Thirteen participants did not complete the task as instructed (e.g., they explicitly both included and excluded targets) and were eliminated from all analyses. Participants either completed the research in a laboratory setting in exchange for partial course credit or were approached at public locations on the university campus and asked to participate in exchange for a piece of candy. Preliminary analyses included location (laboratory vs. public locations) as a between-subjects factor; however, this factor showed neither reliable main effects nor interactions and is not discussed further. The experimental design was a 2 (mind-set: inclusion vs. exclusion) ⫻ 2 (target category: politician vs. nonpolitician) ⫻ 2 (target sex: male vs. female) mixed-model design with repeated measures on the latter two factors.
Materials and Procedure After giving informed consent, participants were asked to complete a short questionnaire regarding their knowledge of politicians and judges. This questionnaire involved a form entitled People in Politics and Law as well as a demographics questionnaire, in that order. The People in Politics and Law questionnaire served as the primary dependent measure. It contained a set of 40 names: 20 male names, 10 of which belonged to actual politicians and judges, and 20 female names, 10 of which belonged to actual politicians and judges (see the Appendix). All names were presented on the same page. The 40 names appeared in a single predetermined random order on the page, in four columns of 10 names each. The 20 politicians were selected on the basis of pretesting to ensure that general knowledge about the male and female politicians was closely equated. We generated a list of 122 politicians and judges (61 female) of current or historical note. Two random orders of this list were generated in a survey called the Political Knowledge Pretest, in which participants were asked to complete a recognition test of the 122 names. One-hundred nine Miami University undergraduates who completed this pretesting were asked to place a check mark beside the names that they recognized as politicians or judges. For names that were checked as belonging to the category “politician or judge,” participants were asked to rate their certainty that the target was, indeed, a politician or judge on a 7-point scale (1 ⫽ not at all certain; 7 ⫽ very certain). Certainty was included to ensure that if participants were using different certainty thresholds to differentially respond to the male and female names, we could select male and female politicians who elicited identical levels of recognition and certainty. The frequency with which each of the 122 names was recognized as a politician or judge and the self-reported certainty of the recognitions were both calculated. Male and female names most closely matched in recognition frequency and recognition certainty were selected. In all cases, the matched male and female targets differed by less than 4% in recognition frequency,
DISCRIMINATION AND MIND-SETS with no mean recognition frequency differences for female (M ⫽ 39.9%) and male targets (M ⫽ 39.0%; p ⬎ .9). Similarly, the matched male and female targets never differed by more than 0.40 in recognition certainty (on a 7-point scale), with no mean recognition certainty differences for female (M ⫽ 5.49) and male (M ⫽ 5.56) targets selected for use in this study ( p ⬎ .8). There were no participant sex effects in recognition frequency or certainty for the targets selected for use in this study ( ps ⬎ .7). To avoid floor and ceiling effects, we also made an effort to select names recognized by more than one fifth but fewer than four fifths of pretest participants. The instructions at the top of the People in Politics and Law questionnaire manipulated the task framing, placing participants into either an inclusion or an exclusion mind-set. Specifically, participants assigned to the inclusion mind-set condition were instructed to “circle the names of the individuals who ARE politicians or judges” (emphasis in original), whereas participants in the exclusion mind-set condition were instructed to “cross off the names of individuals who ARE NOT politicians or judges” (emphasis in original). After completing this task, participants completed a brief demographics questionnaire and then were debriefed.
Results and Discussion Preliminary Analyses Preliminary analyses were conducted on the size of the choice set for both male and female targets (i.e., the number of targets who were considered to be politicians or judges). To investigate how the size of the choice set for male versus female targets varied as a function of inclusion versus exclusion framing, we subjected the data to a 2 (mind-set) ⫻ 2 (participant sex) ⫻ 2 (target sex) mixed analysis of variance (ANOVA), with repeated measures on the third factor. Replicating the standard IED, the average choice set was substantially larger in the exclusion (M ⫽ 17.69) than the inclusion (M ⫽ 6.91) condition, F(1, 101) ⫽ 97.27, p ⬍ .001. The ANOVA also yielded a main effect of target sex, with male targets being retained in the final choice set (M ⫽ 16.55) with greater frequency than female targets (M ⫽ 8.04), F(1, 101) ⫽ 555.47, p ⬍ .001. These main effects were also qualified by a significant two-way Mind-Set ⫻ Target Sex interaction, F(1, 101) ⫽ 100.29, p ⬍ .001, indicating that the tendency to retain more male than female targets in the final choice set was stronger in the exclusion (Mdif ⫽ 12.13) than in the inclusion (Mdif ⫽ 4.89) conditions. Additionally, the ANOVA yielded a Participant Sex ⫻ Target Sex interaction, F(1, 101) ⫽ 5.22, p ⫽ .024, indicating that the tendency to retain more male than female targets in the final choice set was stronger for male (Mdif ⫽ 8.69) than for female participants (Mdif ⫽ 7.69). No other main effects or interactions achieved statistical significance ( ps ⬎ .10).
Signal-Detection Analyses Although useful in connecting the current investigation to the broader context of research on the IED, investigations of raw choice-set size are less informative in investigating how stereotypic biases may differentially play out in inclusion and exclusion mind-sets. As previously noted, past work by Yaniv and colleagues (e.g., Yaniv et al., 2002) has found that the IED in set size is due primarily to strategic shifts in the criterion for what is considered an acceptable candidate for the final choice set, whereas sensitivity seems invariant to task framing. Past work in the domain of sensitivity and criterion stereotyping, however, has
1023
found that different manipulations have differential effects on sensitivity and criterion stereotyping. To test whether the IED is related to sensitivity and/or criterion stereotyping, we performed signal-detection analyses (Green & Swets, 1974). In line with the previous work on the IED (e.g., Yaniv et al., 2002), we decomposed hit and false-alarm rates into nonparametric estimates of sensitivity (A⬘) and criterion (B⬙) separately for male and female targets (see Grier, 1971).1 Of particular interest is how criterion and sensitivity stereotyping differ as a function of inclusion versus exclusion mind-sets. Results for sensitivity. If, as Levin et al. (2000) found, individuals consider their response options more extensively under inclusion task framing, then sensitivity stereotyping should be less evident in an inclusion as compared with an exclusion mind-set. This pattern would be reflected as a smaller difference in A⬘ between male and female targets in the inclusion condition, relative to the exclusion condition. Given that men cannot be assumed to be politicians simply on the basis of their sex, stereotypes are not particularly informative for male targets, and sensitivity for these targets may be comparable under inclusion and exclusion conditions. That is, the relatively high levels of sensitivity in the inclusion mind-set should be maintained in exclusion conditions for male targets (similar to the findings of Yaniv et al., 2002, in the absence of stereotypes). Insofar as women can be stereotypically excluded from agentic roles, however, sensitivity stereotyping of women should be reflected in a lower A⬘ for female targets in the exclusion condition, relative to the inclusion condition. The indices of sensitivity (A⬘) for both male and female targets were submitted to a 2 (mind-set) ⫻ 2 (participant sex) ⫻ 2 (target sex) ANOVA, with repeated measures on the third factor. The ANOVA revealed both a main effect of mind-set, F(1, 101) ⫽ 9.17, p ⫽ .003, and a main effect of target sex, F(1, 101) ⫽ 5.92, p ⫽ .017; however, both of these main effects were qualified by the predicted Mind-Set ⫻ Target Sex interaction, F(1, 101) ⫽ 7.37, p ⫽ .008. As can be seen in Table 1, comparing across mind-sets, it is clear that the Mind-Set ⫻ Target Sex interaction was driven by the reduction in sensitivity to female targets under the exclusion mind-set. Although sensitivity for male targets did not differ across the mind-set conditions, t(104) ⫽ 1.10, ns, sensitivity dropped much more precipitously for female targets in the exclusion as compared with the inclusion condition, t(104) ⫽ 3.28, p ⫽ .001. Notably, the three-way Mind-Set ⫻ Target Sex ⫻ Participant Sex interaction failed to approach statistical significance, F(1, 101) ⫽ 0.04, p ⬎ .8, indicating that the key results with respect to sensitivity stereotyping were similar for male and female decision makers. Results for criterion. Yaniv et al. (2002) found that decision makers set a generally lower criterion under exclusion (compared with inclusion) mind-sets. It was an open question, however, whether mind-sets would moderate observed levels of criterion stereotyping. Separate indices of criterion (B⬙) for male and female targets were submitted to a 2 (mind-set) ⫻ 2 (participant sex) ⫻ 2 1 As is common in signal-detection analyses, adjustments were made to the data to address the problem of empty cells (e.g., 0% false alarms or 100% hits in one or more conditions for a participant), such that 0% was adjusted to 5% and 100% was adjusted to 95%. Identical adjustments were made for data in all studies.
HUGENBERG, BODENHAUSEN, AND MCLAIN
1024
Table 1 Mean Sensitivity and Criterion to Male and Female Targets as a Function of Mind-Set in Study 1 Decision parameter Sensitivity (A⬘) Male targets Female targets Criterion (B⬙) Male targets Female targets
Inclusion mind-set
Exclusion mind-set
.77 .78
.73 .56
.43 .47
⫺.08 .11
(target sex) ANOVA, with repeated measures on the third factor. Replicating Yaniv et al. (2002), the ANOVA yielded a very robust main effect of mind-set on criterion, F(1, 101) ⫽ 70.49, p ⬍ .001, such that the criterion was placed much higher in inclusion (M ⫽ 0.45) than in exclusion (M ⫽ ⫺0.01) task frames. The ANOVA also yielded a main effect of target sex, F(1, 101) ⫽ 10.16, p ⫽ .002, indicating a higher criterion for female targets (M ⫽ 0.29) than for male targets (M ⫽ 0.18), reflecting criterion stereotyping. These main effects, however, were qualified by a Mind-Set ⫻ Target Sex interaction, F(1, 101) ⫽ 5.09, p ⫽ .026. As can be seen in Table 1, the interaction pattern indicates that although the criterion did not differ for male and female targets in the inclusion condition, t(58) ⫽ 1.33, p ⬎ .15, the criterion for membership in the category “politicians” was higher for female than for male targets in the exclusion condition, t(46) ⫽ 2.84, p ⫽ .007. Just as participants set the bar higher for women than men to be considered famous in Banaji and Greenwald’s (1995) false fame studies, so too here did participants in the exclusion condition set the bar higher for women than men to be considered politicians. Implications. These results confirm that the way a choice task is framed (inclusion vs. exclusion) has a notable influence on the emergence of stereotypic bias. Specifically, when participants were asked to exclude individuals who did not belong in the category “politicians,” they showed greater sensitivity stereotyping and criterion stereotyping with respect to female targets, who were stereotypically thought of as not being politicians. That is, both male and female decision makers were less sensitive in distinguishing among the female targets under the exclusion mind-set, and they set a higher criterion for designation as a category member for women than men under the exclusion mind-set. To return to our initial example, these results suggest that it may very well be the case that decisions about candidates for admission to an engineering program would be more contaminated by sexist stereotypes if the admissions committee approached the task with an orientation to eliminate unqualified candidates than if it approached the task with an orientation to select the qualified candidates.
are not politicians but are instead likely to be found in more communal roles. As such, women can be categorically excluded from the category “politician” if one relies on common social stereotypes. In contrast to this situation, Study 2 investigated cultural stereotypes implying that members of a stereotyped group in fact do belong in the target category. Specifically, we investigated the stereotype of African American athleticism. Whereas there is no general expectation that White men play basketball, there is a general stereotype that Black men do play basketball. In Study 2, we investigated the effects of inclusion versus exclusion mind-sets on determining which of a set of African American versus European American names belonged to the category “basketball players.” This task allowed us to examine whether the same patterns seen in Study 1 (in which stereotypes implied exclusion from the choice set) would also apply under conditions in which stereotypes implied inclusion in the choice set. It also allowed us to check for generalization from the domain of sex stereotypes to that of racial stereotypes. Another extension of Study 1 involved examining the effects of expertise on the observed biases. It might be expected that individuals who possess considerable knowledge of the target domain are able to rely more on direct recollection and thus should show generally greater sensitivity that is not particularly affected by task framing or by social stereotypes. Along these lines, Yaniv et al. (2002) confirmed that expertise tends to attenuate the IED; however, Yaniv et al. found that expertise influences only sensitivity, not criterion. They sensibly argued that this outcome occurs because expertise influences knowledge, which is one of the constructs responsible for sensitivity. Just as a more experienced radiologist might be expected to better discriminate between cancerous and noncancerous growths (Swets, Dawes, & Monahan, 2000) and expert airport screeners to better discriminate bombs from normal luggage (Schwaninger, Hardmeier, & Hofer, 2005), so too should a more knowledgeable basketball viewer be expected to more accurately discriminate between players and nonplayers. Because stereotypes have their strongest influence in ambiguous situations, however, we hypothesized that only individuals with relatively low levels of basketball expertise would show evidence of sensitivity stereotyping in an exclusion mind-set. Criterion or decision threshold, however, is a strategically controllable parameter unrelated to the ability to discriminate between a player and a nonplayer. Indeed, mere knowledge of who is a player or not has no influence on the costs and benefits of hits and false alarms. Although expertise can be related to decision threshold, this relationship typically holds when expertise offers additional information as to the benefits and costs of errors (Swets et al., 2000). Therefore, in Study 2, we hypothesized that whereas relative domain expertise would moderate the sensitivity stereotyping effects, it would be unrelated to criterion stereotyping. To examine this issue, we also included a measure of relative basketball expertise in Study 2.
Study 2 In Study 2, we sought to replicate and extend these basic findings. One modification involved changing the relationship between the stereotype and the target category. In Study 1, we investigated cultural stereotypes implying that women are not likely to be politicians. Whereas there is no general expectation that men are politicians, there is a general expectation that women
Method Participants and Design Eighty Miami University undergraduates (23 female) participated in this research. Seventy-four of the participants were European American, and 4 were Asian American. Four participants who did not complete the primary
DISCRIMINATION AND MIND-SETS dependent measure were removed from all analyses. The experimental design was a 2 (mind-set: inclusion vs. exclusion) ⫻ 2 (domain expertise: high vs. low) ⫻ 2 (target category: basketball player vs. nonplayer) ⫻ 2 (target race: African American vs. European American) mixed-model design with repeated measures on the latter two factors.
Materials and Procedure Participants were approached in public places on a university campus and were asked to complete a short questionnaire regarding their knowledge of popular culture. Persons who agreed to participate were given a brief packet of measures containing a form entitled Basketball Knowledge Questionnaire, in addition to a brief demographics questionnaire, in that order. The Basketball Knowledge Questionnaire served as the primary dependent measure in this study and was adapted from the questionnaire used by Park and Banaji (2000, Experiment 2). This questionnaire contained a set of 40 names: 20 African American names, 10 of which belonged to actual basketball players, and also 20 European American names, 10 of which belonged to actual basketball players. The 40 names appeared in a single predetermined random order on a single page, in four columns of 10 names each. The instructions at the top of the Basketball Knowledge Questionnaire manipulated the task framing, placing participants into either an inclusion or an exclusion mind-set. Similar to Study 1, participants assigned to the inclusion mind-set condition were instructed to “circle the names of the individuals who ARE professional basketball players” (emphasis in original). Participants in the exclusion mind-set condition were instructed to “cross off the names of individuals who ARE NOT professional basketball players” (emphasis in original). After completing this task, participants recorded how many hours per week they spent watching professional basketball games during basketball season. This measure was designed to serve as a proxy for their relative domain expertise; individuals who watched more basketball should have been more familiar with who is and is not a professional basketball player. Participants then completed a brief demographics questionnaire. After the survey was completed, participants were debriefed and were given a candy bar in return for their participation.
Results and Discussion Preliminary Analyses Preliminary analyses were conducted on the size of the choice set for both Black and White targets (i.e., the number of targets who were considered to be basketball players). The data were subjected to a 2 (mind-set) ⫻ 2 (expertise) ⫻ 2 (target race) mixed ANOVA, with repeated measures on the third factor. Participants who did not watch any basketball at all (0 hr) were considered to be of lower expertise (n ⫽ 33), whereas participants who watched at least some basketball each week (⬎1⁄2 hr) were considered to be of higher expertise (n ⫽ 43). Thus, the measure of expertise as used in these analyses generally reflects no regular exposure versus some regular exposure to professional basketball. The results again replicated the standard IED; the final choice set was larger in the exclusion (M ⫽ 14.54) than the inclusion (M ⫽ 10.73) condition, F(1, 72) ⫽ 6.45, p ⫽ .01. Individuals higher in expertise also retained more targets in the final choice set (M ⫽ 14.99) than did individuals lower in expertise (M ⫽ 10.28), F(1, 72) ⫽ 9.82, p ⬍ .01. The ANOVA also yielded a three-way Mind-Set ⫻ Expertise ⫻ Target Race interaction, F(1, 72) ⫽ 10.92, p ⫽ .001. This interaction revealed that individuals low in basketball expertise tended to retain more African Americans than European Americans in the final choice set when under an inclu-
1025
sion mind-set, an effect that reversed when under an exclusion mind-set. Individuals high in basketball expertise showed no such effects. The ANOVA yielded no other significant main effects or interactions.
Signal-Detection Analyses Results for sensitivity. An index of sensitivity (A⬘) was calculated for African American and European American names separately, for each participant. These indices of sensitivity were then submitted to a 2 (mind-set) ⫻ 2 (expertise) ⫻ 2 (target race) ANOVA, with repeated measures on the third factor. This analysis revealed an expected main effect of expertise, F(1, 72) ⫽ 15.71, p ⬍ .001, replicating both Yaniv et al. (2002) and Park and Banaji (2000) in showing that individuals with high expertise (M ⫽ 0.82) had a better capacity to distinguish between the players and nonplayers than did those low in expertise (M ⫽ 0.63). The analysis also revealed both a main effect of task framing, F(1, 72) ⫽ 5.60, p ⫽ .02, and a marginal main effect of target race, F(1, 72) ⫽ 3.17, p ⫽ .08; however, both of these main effects were qualified by the predicted Task Framing ⫻ Target Race interaction, F(1, 72) ⫽ 5.59, p ⫽ .02. In the inclusion mind-set, participants were equally sensitive to African American (M ⫽ 0.79) and European American (M ⫽ 0.78) targets, t(38) ⫽ ⫺0.50, p ⬎ .6. In the exclusion mind-set, however, participants were marginally worse at distinguishing among African American targets (M ⫽ 0.64) than among European American targets (M ⫽ 0.73), t(36) ⫽ 1.70, p ⬍ .10. Importantly, this Mind-Set ⫻ Target Race interaction was further qualified by the predicted three-way Mind-Set ⫻ Expertise ⫻ Target Race interaction, F(1, 72) ⫽ 10.80, p ⬍ .01 (see Figure 1). To further investigate the nature of this three-way interaction, we decomposed it into two 2 (mind-set) ⫻ 2 (target race) interactions, one at each level of expertise. For those with relatively low expertise, the predicted Mind-Set ⫻ Target Race interaction emerged, F(1, 31) ⫽ 6.96, p ⫽ .01, such that decision makers operating under an inclusion frame showed no difference in sensitivity for African American and European American targets, t(16) ⫽ ⫺1.23, p ⬎ .2, whereas decision makers operating under an exclusion frame showed worse sensitivity for African American than for European American targets, t(15) ⫽ 2.28, p ⫽ .037. For more expert participants, neither main effects ( ps ⬎ .25) nor the Mind-Set ⫻ Target Race interaction ( p ⫽ .15) emerged. The pattern of this three-way interaction suggests that, as predicted, sensitivity is generally high in an inclusion mind-set, wherein decision makers have been shown to consider each option more extensively (Levin et al., 2000). When the mind-set elicits a relatively high amount of processing of the alternatives, both majority and minority group targets are equally individuated. This pattern replicated Park and Banaji’s (2000, Experiments 2 and 3) control (neutral mood) condition, in which they found roughly equal sensitivity for African American and European American targets. However, as predicted, in an exclusion mind-set, sensitivity to African American targets dropped as compared with sensitivity to European American targets. Thus, when the mind-set itself elicited less engagement with and processing of the alternatives, individuals were less likely to engage in the effort required to distinguish among African Americans. These effects were moderated by expertise: Although the sensitivity of nonexperts was subject to the moderating effects of decision-making mind-sets, the
1026
HUGENBERG, BODENHAUSEN, AND MCLAIN
Figure 1. Mean sensitivity (A⬘) to both African American and European American targets as a function of task framing and level of expertise for Study 2.
effects of mind-set and target race were eliminated for persons higher in basketball expertise. When respondents have relatively more elaborate knowledge of the relevant domain, there is correspondingly less room for the biasing effects of mind-set or target race. Results for criterion. Separate indices for criterion (B⬙) set for African American and European American targets were calculated for each participant. These indices of criterion were then submitted to a 2 (mind-set) ⫻ 2 (expertise) ⫻ 2 (target race) ANOVA, with repeated measures on the third factor. Again, we replicated the main effect of task framing, F(1, 72) ⫽ 28.85, p ⬍ .001, such that the criterion was placed higher in the inclusion (M ⫽ 0.37) than in the exclusion (M ⫽ 0.02) mind-set. Mirroring the findings from Study 1, this main effect of task framing was qualified by an interaction with target race, F(1, 72) ⫽ 12.28, p ⫽ .001. Replicating the pattern of data observed in Park and Banaji’s (2000) control conditions (their Studies 2 and 3), in the inclusion condition, the criterion for African American targets (M ⫽ 0.41) was set marginally higher than that set for European American targets (M ⫽ 0.32), t(38) ⫽ ⫺1.83, p ⫽ .08. As predicted, however, in the exclusion mind-set, criterion stereotyping of African American (M ⫽ ⫺0.06) as compared with European American (M ⫽ 0.08) targets was observed, t(36) ⫽ 2.75, p ⬍ .01. Again replicating Yaniv et al. (2002), expertise level did not influence criterion. In this study, neither the main effect ( p ⬎ .45) nor interactions including expertise ( ps ⬎ .13) achieved significance. The results with respect to criterion levels in Study 2 jibe well with the pattern observed in Study 1. In Study 1, participants in the exclusion mind-set set the bar higher for women to gain entre´e into the counterstereotypic groups of politicians and judges. In Study 2, the converse occurred, with the exclusion mind-set eliciting a lower criterion for African Americans to be considered members of the stereotypic group of basketball player. Implications. As was the case in Study 1, we again found evidence that an exclusion mind-set breeds both greater sensitivity stereotyping and greater criterion stereotyping. This pattern was replicated in a new domain (racial stereotypes) and in a context in
which the stereotype implied membership in the target category. These results were also moderated by expertise in a meaningful and predictable way. For sensitivity stereotyping, expertise had clear effects. In this case, individuals high in expertise showed equally high levels of sensitivity to African American and European American targets in both mind-sets, but under uncertainty, biases had room to operate. Stereotypic biases are known to be at their most powerful when situations are ambiguous (Bodenhausen & Macrae, 1998; Hugenberg & Bodenhausen, 2003, 2004). For low-expertise participants, the situation was a much more ambiguous one, and as such, stereotypic biases tended to influence participants’ choices. When participants had less expertise and, thus, less ability to distinguish between exemplars of a particular racial category, mind-set played a large role in the extent to which they perceived African Americans (vs. European Americans) to be a homogeneous group. For criterion stereotyping, however, expertise was neither predicted nor found to be a reliable moderator. Replicating previous research (Yaniv et al., 2002), expertise was related only to sensitivity and not to criterion. Given the nature of these two separate parameters, this is a quite sensible outcome. Whereas sensitivity is a reflection of predictive ability and is thus sensibly related to domain expertise, criterion is a strategically controlled parameter, affected not by knowledge but rather by strategic concerns such as the perceived cost of errors.
Study 3 Thus far, exclusion mind-sets in decision making have been shown to elicit low levels of sensitivity toward members of stereotyped groups. The exclusion mind-set also elicits criterion stereotyping as well, with the bar being set comparatively high for stereotyped individuals to gain entre´e into counterstereotypic categories (i.e., women vis-a`-vis the category “politician”) but being set comparatively low for access to stereotype-consistent categories (i.e., African Americans vis-a`-vis the category “basketball player”). As found in Study 2, however, sensitivity, but not crite-
DISCRIMINATION AND MIND-SETS
rion stereotyping, was moderated by participants’ relative expertise. Study 3 was designed to replicate and extend the results of the previous studies. Whereas Study 2 investigated a variable that moderated the effects of exclusive thinking on sensitivity stereotyping (i.e., expertise), Study 3 was designed to investigate a variable hypothesized to moderate the effects of exclusive thinking on criterion stereotyping. As previously noted, unlike sensitivity, criterion cutoff is a strategically controllable decision parameter (Swets, 1992; Swets et al., 2000). As such, we hypothesized it would be moderated by relevant motivational constructs and in particular by motivation to control prejudiced responses. Specifically, we hypothesized that only individuals relatively unmotivated to appear nonprejudiced would show a strong pattern of criterion stereotyping in the exclusion mind-set. That is, we hypothesized that the criterion stereotyping in the exclusion mind-set observed in Study 2 was primarily due to individuals low in motivation to control prejudice. In contrast, we hypothesized that situations tending to elicit criterion stereotyping (i.e., an exclusion mind-set) would not elicit such a response from individuals high in motivation to control prejudice. Instead, those high in motivation to control prejudice would make different strategic decisions regarding the criterion for category membership in an effort to avoid ostensibly stereotypic judgments. In particular, they would be expected to hold an equal or even higher criterion for Black as compared with White targets. In essence, participants very high in motivation to control prejudice might overcorrect for their presumed criterion stereotyping by holding an artificially high criterion for Black as compared with White targets. In fact, in Study 2, we observed a tendency to hold a higher threshold for Black targets in the inclusion condition. Thus, whereas we hypothesized that individuals with high motivation to control would have an equally high threshold for Black (as compared with White) targets across conditions, only individuals with low motivation to control would show the drop in criterion in an exclusion mind-set. To test these hypotheses, we used the same procedure as in Study 2 but collected Plant and Devine’s (1998) Internal and External Motivation to Respond Without Prejudice Scales instead of a measure of participants’ relative domain expertise.
1027
demographics questionnaire. The order of the Basketball Knowledge Questionnaire and the Internal and External Motivation to Respond Without Prejudice Scales was counterbalanced on a between-subjects basis; the demographics questionnaire always appeared last in the experimental packet. Preliminary analyses found that the counterbalancing factor yielded neither main effects nor interactions and thus is not discussed further. As in Study 2, the Basketball Knowledge Questionnaire served as the primary dependent measure in this study, and instructions at the top of the questionnaire manipulated participants’ mind-set. After the survey was completed, participants were thanked and debriefed.
Results and Discussion Preliminary Analyses Motivation to control prejudiced reactions. As is typical with analyses using the Internal and External Motivation to Respond Without Prejudice Scales, the internal motivation to respond without prejudice (IMS) and external motivation to respond without prejudice (EMS) were treated as separate predictors. In no analysis did IMS interact with any of the other factors, and thus, it is not discussed further. EMS did qualify a number of the observed effects; thus, analyses using EMS as a continuous, individualdifferences predictor variable are reported where appropriate. Choice set. As in the previous studies, preliminary analyses were conducted on the size of the choice set for both Black and White targets (i.e., the number of targets who were considered to be basketball players). The data were subjected to a 2 (mind-set: inclusion vs. exclusion) ⫻ 2 (target race: Black vs. White) mixedmodel ANOVA, with repeated measures on the third factor. Preliminary analyses, which included IMS and EMS scores, yielded no significant effects of IMS or EMS; thus, these were dropped from all subsequent analyses of choice-set size. The results again strongly replicated the standard IED; the average choice set was substantially larger in the exclusion (M ⫽ 27.06) than the inclusion (M ⫽ 7.07) condition, F(1, 108) ⫽ 278.67, p ⬍ .001. The results also showed a main effect of target race, indicating that participants retained more African American targets (M ⫽ 9.74) than European American targets (M ⫽ 7.33) in the final choice set, F(1, 108) ⫽ 63.87, p ⬍ .001. The ANOVA yielded no other significant effects.
Method Participants and Design One-hundred twenty Miami University undergraduates (91 female) participated in this research for partial course credit. Ten participants who did not complete the measures or did not follow task instructions (e.g., they explicitly both included and excluded targets) were removed from all analyses. The experimental design was a 2 (mind-set: inclusion vs. exclusion) ⫻ 2 (target category: basketball player vs. nonplayer) ⫻ 2 (target race: African American vs. European American) mixed-model factorial design with repeated measures on the latter two factors. Motivation to control prejudiced responses served as a continuous predictor variable.
Materials and Procedure Materials and procedure were identical to those used in Study 2, except as noted. After providing informed consent, participants were given a brief packet of measures containing the Basketball Knowledge Questionnaire used in Study 2, as well as Plant and Devine’s (1998) 10-item Internal and External Motivation to Respond Without Prejudice Scales and a brief
Signal-Detection Analyses Results for sensitivity. An index of sensitivity (A⬘) was calculated for African American and European American names separately, for each participant. These indices of sensitivity were then subjected to preliminary analyses using a general linear model with mind-set (inclusion vs. exclusion) as a between-subjects factor, target race (Black vs. White) as a within-subjects factor, and motivation to control prejudiced responses as a continuous predictor variable (see Judd, McClelland, & Smith, 1996). Separate analyses were conducted with IMS and EMS; as predicted, neither IMS nor EMS yielded significant interactions with target race ( ps ⬎ .18). Thus, IMS and EMS were dropped from subsequent analyses of sensitivity. Replicating Study 2, this ANOVA yielded a Task Framing ⫻ Target Race interaction, F(1, 108) ⫽ 9.09, p ⬍ .001. As with the previous studies, inclusion (M ⫽ 0.57) versus exclusion (M ⫽ 0.62) mind-set had no influence on sensitivity toward European
1028
HUGENBERG, BODENHAUSEN, AND MCLAIN
American targets, t(108) ⫽ 0.92, p ⬎ .35. Mind-set did, however, strongly influence sensitivity toward African American targets such that sensitivity showed a substantial drop from the inclusion (M ⫽ 0.71) to the exclusion (M ⫽ 0.56) mind-set, t(108) ⫽ 3.77, p ⬍ .001. Results for criterion. Separate indices of the criterion (B⬙) set for African American and European American targets were calculated for each participant. These indices of criterion were then subjected to a general linear model with mind-set (inclusion vs. exclusion) as a between-subjects factor, target race (Black vs. White) as a within-subjects factor, and motivation to control prejudiced responses as a continuous predictor variable. Separate analyses were conducted with IMS and EMS; IMS yielded no significant effects ( ps ⬎ .7). Thus, the following analyses of criterion include only EMS as a continuous predictor. These analyses again replicated the standard IED findings, showing a main effect of task framing on criterion, F(1, 106) ⫽ 9.20, p ⬍ .01, such that the cutoff criterion was placed higher in the inclusion (M ⫽ 0.30) than in the exclusion (M ⫽ ⫺0.01) mind-set. The analyses also yielded a two-way Target Race ⫻ Mind-Set interaction, F(1, 106) ⫽ 7.89, p ⬍ .01, such that a stronger drop in criterion was elicited for Black targets by the exclusion (M ⫽ ⫺0.02) versus the inclusion (M ⫽ 0.39) mind-sets, as compared with the White targets in the exclusion (M ⫽ 0.01) versus the inclusion (M ⫽ 0.39) mind-sets. Importantly, however, the predicted three-way Task Framing ⫻ Target Race ⫻ EMS interaction was also observed, F(1, 106) ⫽ 4.62, p ⫽ .034. To further investigate the nature of this three-way interaction, we decomposed it into two Target Race ⫻ EMS interactions, one for each mind-set. For participants in an inclusion mind-set, no Target Race ⫻ EMS interaction was observed, F(1, 59) ⫽ 0.86, p ⬎ .35. Instead, we observed only a marginal main effect of target race, F(1,59) ⫽ 2.91, p ⫽ .09, indicating that the criterion was set higher for Black (M ⫽ 0.39) than for White (M ⫽ 0.22) targets in the inclusion mind-set, replicating the pattern of data observed in Study 2. For participants in the exclusion mind-set, however, the predicted Target Race ⫻ EMS interaction was observed, F(1, 47) ⫽ 5.14, p ⫽ .028. As can be seen in Figure 2, for exclusioncondition participants with relatively low levels of EMS, one sees a lower criterion for Black as compared with White targets, replicating the criterion data observed under exclusion conditions in
Study 2. As external motivation to control prejudice increased, however, so did the criterion for Black as compared with White targets. Indeed, at very high levels of EMS, a higher criterion was set for Black as compared with White targets. Implications. This study generally replicated the patterns of sensitivity and criterion stereotyping observed in the previous studies. More importantly, however, the current study suggests that criterion stereotyping can be modulated by relevant motivational factors. Whereas Study 2 indicated that at relatively high levels of domain expertise, an exclusion mind-set does not elicit sensitivity stereotyping, in the current study, it seemed that the criterion stereotyping in exclusion mind-sets observed in the previous studies occurred only for individuals low in EMS. Insofar as participants were high in external motivation to control prejudiced responses (i.e., they did not wish to appear prejudiced to others), they tended to adhere to a higher criterion for Black than White targets, likely in an attempt to avoid politically incorrect responses. Indeed, across both Studies 2 and 3, there was a weak but consistent tendency for criterion to be set higher for Black as compared with White targets in inclusion conditions. Decision makers wary of responding stereotypically to African American targets (i.e., high EMS participants) retained that higher threshold for Black as compared with White in exclusion mind-sets. As this wariness of responding stereotypically waned, the tendency to engage in criterion stereotyping of Black targets increased in an exclusion mind-set. Importantly, motivation to control prejudice was not a panacea for all stereotypic responding. To the contrary, external motivation to control prejudice moderated only criterion stereotyping. Sensitivity stereotyping occurred equally, regardless of participants’ motivation to control prejudiced responses. This pattern of data fits well with previous theory and data regarding sensitivity and criterion in the IED (see Yaniv et al., 2002). As criterion is responsive to participants’ choice strategies (Swets, 1992; Swets et al., 2000), given that motivation to control prejudiced responses is a strategic motivation, the fact that EMS moderates criterion, but not sensitivity stereotyping, is sensible. Additionally, it seems that not all motivations to control prejudice are created equal. External motivation to control prejudice, but not internal motivation to control prejudice, uniquely predicts criterion stereotyping. It was only those low in EMS who showed a willingness to engage in criterion stereotyping in the third study.
General Discussion
Figure 2. Mean exclusion mind-set criterion (B⬙) to both African American and European American targets as a function of external motivation to respond without prejudice for Study 3.
Across three studies and two separate sets of stimuli, we reliably found that inclusion and exclusion mind-sets lead to different proclivities to engage in sensitivity stereotyping and criterion stereotyping for decisions about members of stereotyped social groups. In Study 1, we found that sensitivity to male and female targets was equally high in an inclusion mind-set. In an exclusion mind-set, sensitivity to male targets remained relatively high, suggesting continuing individuation of these targets. Sensitivity to stereotyped female targets, however, showed a marked drop for participants in an exclusion mind-set, indicating increased perceptions of homogeneity among these female targets. Studies 2 and 3 replicated this sensitivity stereotyping, here in the domain of racial stereotypes, with Study 2 further indicating that the biases were confined to individuals lower in domain expertise. Criterion ste-
DISCRIMINATION AND MIND-SETS
reotyping showed an equally interesting pattern. In Study 1, an inclusion mind-set elicited quite comparable criteria for women and men, but an exclusion mind-set elicited a higher threshold for retaining women than men in the final choice set. Similarly, Studies 2 and 3 found that an exclusion mind-set elicited criterion stereotyping toward African Americans as compared with European Americans such that participants in an exclusion mind-set held a lower threshold for retaining African Americans than European Americans in the final choice set. Notably, Study 3 found that this criterion stereotyping under an exclusion mind-set occurred most strongly for individuals low in EMS.
Extending the Inclusion–Exclusion Discrepancy Overall, the current results fit well with the existing constellation of data regarding the IED, but these findings also extend the IED literature in meaningful ways. For example, Yaniv et al. (2002) found that inclusive and exclusive modes of thinking induce different response criteria, with exclusion mind-sets resulting in a lower criterion than inclusion mind-sets. The current studies clearly replicated the quite robust main effect of mind-set on bias. Similarly, we also replicated the previous work of Yaniv and Schul (2000) in showing a moderating role of expertise on sensitivity. However, going beyond previous findings, these studies extend the understanding of the IED by illuminating the potential effects of decision-making mind-sets on sensitivity. Clearly, we did not find any general main effect of task framing on sensitivity, and this pattern accords with the past work finding no differences in sensitivity across mind-set (Yaniv et al., 2002). However, we did find that mind-set can influence sensitivity levels under specific circumstances, namely, when social stereotypes are relevant to the decision that is being made. Specifically, members of social categories that are particularly likely to be treated categorically (i.e., members of stereotyped groups) are subject to drops in sensitivity under an exclusion mind-set.
Mind-Set Effects on Sensitivity and Criterion Stereotyping To our knowledge, these studies constitute the first demonstration that sensitivity and criterion stereotyping can be differentially induced by different task framing in decision making. Although past work by Banaji and colleagues found that false fame judgments are the result of criterion stereotyping (Banaji & Greenwald, 1995) and that both sensitivity stereotyping and criterion stereotyping are enhanced by positive mood (Park & Banaji, 2000), the current work reveals that different strategies for coming to a judgment can elicit differential treatment of members of stereotyped groups. Given that both of the strategies that we investigated are commonly used in everyday-life contexts (see Heller et al., 2002), the practical implications may be considerable. These findings that exclusive modes of thinking lead to sensitivity stereotyping (as compared with inclusive modes of thinking) become all the more striking when one considers how difficult it is to eliminate the IED. For example, Yaniv and Schul (1997, Study 2) had participants play a multiple-choice trivia game in which the correct answer to a trivia question was embedded in a list of 20 response alternatives. Yaniv and Schul paid participants for correct answers retained in the final choice set but paid them more for smaller choice sets, thereby offering a self-interested motivation not to
1029
retain unacceptable cases in the final choice set. Despite this motive, the IED remained; the financial motivation was insufficient to eliminate the effect. Despite clearly replicating sensitivity and criterion stereotyping across two separate domains and across three separate studies, the current work does not speak definitively to the mechanisms underlying the observed results. Our own logic has rested upon previous work by Levin et al. (2000), who suggested that inclusion and exclusion mind-sets elicit different levels of deliberation regarding the response options. Specifically, we hypothesize that insofar as an inclusion mind-set elicits higher levels of deliberation about the targets than an exclusion mind-set, one will observe less stereotyping in an inclusion mind-set. Although this explanation accords well with previous research and theory suggesting that deeper processing elicits individuation and reduces stereotyping (for a recent review, see Bodenhausen, Macrae, & Hugenberg, 2003), it is certainly not the only possible mechanism underlying these effects. One potentially interesting alternative is that the mind-set used by participants may itself act as a type of information to participants, suggesting the relative ease or difficulty of the task. Heller et al. (2002) found that participants were more likely to spontaneously adopt an exclusion mind-set when faced with difficult decisions. If difficult tasks are commonly associated with an exclusion mind-set, perhaps inducing participants to use an exclusion mind-set may have made the task seem more difficult, eliciting the use of stereotypes as judgmental heuristics to simplify this seemingly complex task. Another alternative is that inclusion and exclusion tasks may differ in the extent to which they induce a need for closure (Webster & Kruglanski, 1994). Although little research exists on this possibility, there is some evidence (see Levin et al., 2000) that individuals in an exclusion mind-set spend less time processing alternatives than do individuals in an inclusion mind-set. Although certainly not definitive, such an outcome may be indicative of exclusion mind-sets eliciting a stronger motivation for closure. If this is true, it could also explain the relatively higher levels of sensitivity and criterion stereotyping in exclusion conditions.
Implications for Social Decision Making Just as it seems to fly in the face of rationality to feel better about a drug that results in a 75% survival rate compared with one that results in a 25% mortality rate, it seems strange that an instruction to eliminate the unsatisfactory options produces quite different results compared with an instruction to select the satisfactory options. Whereas it has been well documented that the former instruction results in a larger final choice set, the present findings clarify that this IED is not the only way that task framing can influence decision-making processes and choice behavior. The exclusion frame produces not only a larger choice set but one that reflects both relatively poorer sensitivity to members of stereotyped social groups (whereas sensitivity to dominant groups remains largely intact) and a differential criterion for retention in stereotype-relevant categories. Human lives are filled with important social decisions, many of which could be subject to different mind-sets. From deciding which potential partners to date, to which colleagues to promote, to which applicants to admit to graduate school, important life decisions can be approached via inclusion or exclusion frameworks.
HUGENBERG, BODENHAUSEN, AND MCLAIN
1030
The current work finds that framing a decision-making process slightly differently, despite the obvious logical equivalence of the two task frames, can lead not only to different decisions but also to different patterns of stereotyping. Understanding how these different mind-sets operate could allow decision makers to ameliorate the otherwise potent effects of stereotypes in important life decisions. Returning to our earlier example of graduate admissions, the current findings suggest that if a graduate admissions committee approaches the admissions decision under an exclusion mind-set, then neither the interests of the well-qualified female candidates nor those of the selection committee will be well served. In fact, if the selection committee uses exclusive decision processes, distinctions between qualified and less qualified female candidates will become more blurred. Additionally, if the admissions committee holds a higher threshold for the female than the male applicants under an exclusion mind-set, not only will qualified female applicants be excluded but more women than men will be excluded from the list altogether. Our results suggest that the combination of such sensitivity and criterion stereotyping and the potential discriminatory selection patterns that may result are best attenuated by a mind-set of inclusion.
References Banaji, M. R., & Greenwald, A. G. (1995). Implicit gender stereotyping in judgments of fame. Journal of Personality and Social Psychology, 68, 181–198. Biernat, M. (2003). Toward a broader view of social stereotyping. American Psychologist, 58, 1019 –1027. Bodenhausen, G. V., & Macrae, C. N. (1998). Stereotype activation and inhibition. In R. S. Wyer, Jr. (Ed.), Advances in social cognition: Vol. 11. Stereotype activation and inhibition (pp. 1–52). Mahwah, NJ: Erlbaum. Bodenhausen, G. V., Macrae, C. N., & Hugenberg, K. (2003). Social cognition. In I. B. Weiner (Series Ed.) & T. Millon & M. J. Lerner (Vol. Eds.), Handbook of psychology: Vol. 5. Personality and social psychology (pp. 257–282). Hoboken, NJ: Wiley. Eagly, A. H., & Karau, S. J. (2002). Role congruity theory of prejudice toward female leaders. Psychological Review, 109, 573–598. Eagly, A. H., Wood, W., & Diekman, A. (2000). Social role theory of sex differences and similarities: A current appraisal. In T. Eckes & H. M. Trautner (Eds.), The developmental social psychology of gender (pp. 123–174). Mahwah, NJ: Erlbaum. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York: Academic Press. Green, D. M., & Swets, J. A. (1974). Signal detection theory and psychophysics. New York: Krieger. Grier, J. B. (1971). Nonparametric indexes for sensitivity and bias: Computing formulas. Psychological Bulletin, 75, 424 – 429. Heller, D., Levin, I. P., & Goransson, M. (2002). Selection of strategies for narrowing choice options: Antecedents and consequences. Organizational Behavior and Human Decision Processes, 89, 1194 –1213.
Hugenberg, K., & Bodenhausen, G. V. (2003). Facing prejudice: Implicit prejudice and the perception of facial threat. Psychological Science, 14, 640 – 643. Hugenberg, K., & Bodenhausen, G. V. (2004). Ambiguity in social categorization: The role of prejudice and facial affect in face categorization. Psychological Science, 15, 342–345. Jacoby, L. L., Kelley, C. M., Brown, J., & Jasechko, J. (1989). Becoming famous overnight: Limits on the ability to avoid unconscious influences of the past. Journal of Personality and Social Psychology, 56, 326 –338. Judd, C. M., McClelland, G. H., & Smith, E. R. (1996). Testing treatment by covariate interactions when treatment varies within subjects. Psychological Methods, 1, 366 –378. Judd, C. M., & Park, B. (1988). Out-group homogeneity: Judgments of variability at the individual and group levels. Journal of Personality and Social Psychology, 54, 778 –788. Ku¨hberger, A. (1998). The influence of framing on risky decisions: A meta-analysis. Organizational Behavior and Human Decision Processes, 75, 23–55. Levin, I. P., Huneke, M. E., & Jasper, J. D. (2000). Information processing at successive stages of decision making: Need for cognition and inclusion– exclusion effects. Organizational Behavior and Human Decision Processes, 82, 171–193. Levin, I. P., Jasper, J. D., & Forbes, W. S. (1998). Choosing versus rejecting options at different stages of decision making. Journal of Behavioral Decision Making, 11, 193–210. Macrae, C. N., & Bodenhausen, G. V. (2000). Social cognition: Thinking categorically about others. Annual Review of Psychology, 51, 93–120. Park, J., & Banaji, M. R. (2000). Mood and heuristics: The influence of happy and sad states on sensitivity and bias in stereotyping. Journal of Personality and Social Psychology, 78, 1005–1023. Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond without prejudice. Journal of Personality and Social Psychology, 75, 811– 832. Quattrone, G. A., & Jones, E. E. (1980). The perception of variability within in-groups and out-groups: Implications for the law of small numbers. Journal of Personality and Social Psychology, 38, 141–152. Schwaninger, A., Hardmeier, D., & Hofer, F. (2005). Aviation security screeners’ visual abilities and visual knowledge measurement. IEEE Aerospace and Electronic Systems, 20, 29 –35. Swets, J. A. (1992). The science of choosing the right decision threshold in high-stakes diagnostics. American Psychologist, 47, 522–532. Swets, J. A., Dawes, R. M., & Monahan, J. (2000). Psychological science can improve diagnostic decisions. Psychological Science in the Public Interest, 1, 1–26. Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of Business, 59, 251–278. Webster, D. M., & Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. Journal of Personality and Social Psychology, 67, 1049 –1062. Yaniv, I., & Schul, Y. (1997). Elimination and inclusion procedures in judgment. Journal of Behavioral Decision Making, 10, 211–220. Yaniv, I., & Schul, Y. (2000). Acceptance and elimination procedures in choice: Noncomplementarity and the role of implied status quo. Organizational Behavior and Human Decision Processes, 82, 293–313. Yaniv, I., Schul, Y., Raphaelli-Hirsch, R., & Maoz, I. (2002). Inclusive and exclusive modes of thinking: Studies of prediction, preference and social perception during parliamentary elections. Journal of Experimental Social Psychology, 38, 352–367.
DISCRIMINATION AND MIND-SETS
1031
Appendix Names Used in the People in Politics and Law Questionnaire, Study 1 Female Politicians and Judges Elizabeth Cady Stanton Ruth Bader Ginsburg Susan B. Anthony Margaret Thatcher Madeleine Albright Sandra Day O’Connor Elizabeth Hanford Dole Janet Reno Carol Moseley-Braun Nancy Pelosi
Male Politicians and Judges Richard Gephardt Antonin Scalia Tom DeLay George Voinovich Bill Frist Tom Ridge Henry Clay William Rehnquist Paul O’Neill Tom Daschle
Female Distractors Janet Adams Miriam Wegner Angela Mitchell Janet Ann Felty Judy Smyth Mary Johanneson Kristina McLain Rebecca Ann Heckman Karen Shell Shira Gabriel
Male Distractors Jeff Anderson Jeff Valman Harold Fox Josh Muennich Michael Cutting Andrew Prior James Baldwin Mark Thomas Black Bill Tach Roy Lawrence Received March 28, 2005 Revision received March 28, 2006 Accepted March 29, 2006 䡲
DISCRIMINATION AND MIND-SETS
1031
Appendix Names Used in the People in Politics and Law Questionnaire, Study 1 Female Politicians and Judges Elizabeth Cady Stanton Ruth Bader Ginsburg Susan B. Anthony Margaret Thatcher Madeleine Albright Sandra Day O’Connor Elizabeth Hanford Dole Janet Reno Carol Moseley-Braun Nancy Pelosi
Male Politicians and Judges Richard Gephardt Antonin Scalia Tom DeLay George Voinovich Bill Frist Tom Ridge Henry Clay William Rehnquist Paul O’Neill Tom Daschle
Female Distractors Janet Adams Miriam Wegner Angela Mitchell Janet Ann Felty Judy Smyth Mary Johanneson Kristina McLain Rebecca Ann Heckman Karen Shell Shira Gabriel
Male Distractors Jeff Anderson Jeff Valman Harold Fox Josh Muennich Michael Cutting Andrew Prior James Baldwin Mark Thomas Black Bill Tach Roy Lawrence Received March 28, 2005 Revision received March 28, 2006 Accepted March 29, 2006 䡲
INTERPERSONAL RELATIONS AND GROUP PROCESSES
The Relative Deprivation–Gratification Continuum and the Attitudes of South Africans Toward Immigrants: A Test of the V-Curve Hypothesis Michae¨l Dambrun
Donald M. Taylor
McGill University and Universite´ Blaise Pascal
McGill University
David A. McDonald and Jonathan Crush
Alain Me´ot
Queen’s University
Universite´ Blaise Pascal
It has long been established that there is a linear and positive relationship between relative deprivation and prejudice. However, a recent experiment suggests that the converse of relative deprivation, relative gratification, may also be associated with prejudice (S. Guimond & M. Dambrun, 2002). Specifically, the evidence suggests that the usual test for a linear relationship between relative deprivation– gratification and prejudice might conceal the existence of a bilinear relationship. This function, labeled the V-curve hypothesis, predicts that both relative deprivation and relative gratification are associated with higher levels of prejudice. This hypothesis was tested with a representative sample of South Africans (N ⫽ 1,600). Results provide strong support for the V-curve hypothesis. Furthermore, strength of ethnic identification emerged as a partial mediator for the effect of relative gratification on prejudice. Keywords: relative gratification, relative deprivation, prejudice, V-curve hypothesis, group identification
The present research applied both relative deprivation and relative gratification theories to understand the negative attitudes that South Africans have toward immigrants to their country. A comprehensive, in depth survey initiated by the Southern African Migration Project, using a representative national survey, demonstrated that immigrants are the prime target for prejudice in postapartheid South Africa (Mattes, Taylor, Poore, & Richmond, 1999).
Relative Deprivation and Intergroup Attitudes Relative deprivation has offered a number of important insights into researchers’ understanding of intergroup attitudes (Brewer & Brown, 1998; Fiske, 1998; Pettigrew, 2002; Runciman, 1966). The concept of relative deprivation was coined by researchers who were studying the satisfaction levels of American soldiers during the Second World War (see Stouffer, Suchman, DeVinney, Starr, & Williams, 1949). The major assumption of relative deprivation theory is that a person’s or group’s satisfaction is not related to their objective circumstances but, rather, to their condition relative to other persons or groups. This implies, for example, that objectively disadvantaged people may feel less deprived than objectively advantaged people because of the chosen target for their social comparisons. Relative deprivation theory has been successfully applied to a variety of social–psychological domains including collective action (Gurr, 1970; see also Guimond & Dube´Simard, 1983), revolution (Davies, 1962), and the dynamics of intergroup hostility and prejudice (see Mummendey, Kessler, Klink, & Mielke, 1999; Runciman, 1966). In terms of prejudice, relative deprivation theory postulates that unfavorable comparisons (the cognitive component of relative deprivation) can generate feelings of deprivation (the affective component of relative deprivation) that motivate outgroup hostility (see, e.g., Grant & Brown, 1995). Relative deprivation has consistently been identified as being a strong and robust predictor of intergroup attitudes in a variety of countries including the United States (Vanneman & Pettigrew,
Michae¨l Dambrun, Department of Psychology, McGill University, Montreal, Quebec, Canada, and Laboratoire de Psychologie Sociale et Cognitive, Centre National de la Recherche Scientifique (LAPSCO CNRS), Universite´ Blaise Pascal, Clermont-Ferrand, France; Donald M. Taylor, Department of Psychology, McGill University; David A. McDonald, Faculty of Arts and Science, Queen’s University, Kingston, Ontario, Canada; Jonathan Crush, Southern African Research Centre, Queen’s University; Alain Me´ot, LAPSCO CNRS, Universite´ Blaise Pascal. This research was funded by the Southern African Migration Project and the Social Sciences and Humanities Research Council of Canada. We sincerely thank David A. Kenny for his very helpful statistical recommendations, Robert Mattes and Roxane de la Sablonniere for their assistance, and Serge Guimond and Douglas Palmer for useful comments on a previous version of this article. Correspondence concerning this article should be addressed to Michae¨l Dambrun, UFR de Psychologie, Universite´ Blaise Pascal, LAPSCO UMR CNRS, 34 avenue Carnot, 63000, Clermont-Ferrand, France. E-mail:
[email protected] Journal of Personality and Social Psychology, 2006, Vol. 91, No. 6, 1032–1044 Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1032
1032
PREJUDICE AND THE V-CURVE HYPOTHESIS
1972), India (Tripathi & Strivastava, 1981), South Africa (Appelgryn & Nieuwoudt, 1988), and Western Europe (Dambrun & Guimond, 2001; Pettigrew & Meertens, 1995). In many different cultures, higher levels of relative deprivation have been associated with greater levels of prejudice. Confirming the key function of relative deprivation in intergroup phenomena, a recent study revealed that even when one controls for other sociopsychological variables (e.g., identity threat, social dominance orientation), relative deprivation still remains one of the main predictors of extreme right-wing political attitudes (Dambrun, Maisonneuve, Duarte, & Guimond, 2002). The relative deprivation framework suggests that negative attitudes toward immigrants to South Africa may be the result of perceived relative deprivation among South Africans. However, a recent experiment suggests the possibility of a more complex relationship between South Africans’ perceptions of their circumstances and negative attitudes toward immigrants. Specifically, Guimond and Dambrun (2002) have noted that although the role of relative deprivation in the regulation of prejudice has been extensively studied, relative gratification, the converse of relative deprivation, has received little attention (see also Leach, Snider, & Iyer, 2002; Pettigrew, 2002).
Relative Gratification and Intergroup Hostility Relative deprivation theorists have long argued that when people feel better off than others, the result is a state of relative gratification, the opposite of relative deprivation (Leach et al., 2002; Martin, 1981; Smith, Spears, & Oyen, 1994; Vanneman & Pettigrew, 1972). Although relative deprivation theorists acknowledge that relative gratification is conceptually possible, its potential effects on intergroup attitudes have received little attention. Challenging the commonsense conclusion that if relative deprivation is associated with greater prejudice, relative gratification should be related to greater tolerance, recent research has revealed that relative gratification may actually lead to greater intergroup hostility. Specifically, in a series of experiments with psychology students in France, Guimond and Dambrun (2002) manipulated both relative deprivation and relative gratification by confronting participants with declining (relative deprivation) or improving (relative gratification) personal job opportunities (Study 1) and group job opportunities (Study 2). For example, in the second study, the students were led to believe that their own group (psychology students) would be much worse off in terms of job opportunities in the future (group relative deprivation condition) than an outgroup (students in law) or much better off (group relative gratification condition). Following the manipulation, in addition to measuring various intergroup attitudes, Guimond and Dambrun measured perceptions of the standing of the two groups (cognitive component) and the feelings associated with these perceptions (affective component). The results confirmed the usual effect of relative deprivation on prejudice but also revealed that relative gratification increased the level of prejudice toward stigmatized groups in France, increased ingroup bias, and led to an increase in the willingness to support and act in favor of restrictive immigration policies. Moreover, although the effect of relative deprivation on prejudice was partially mediated by the affective component (dissatisfaction), the effect of relative gratification was mediated by the cognitive component. Thus, the perception of being better off, regardless of the feelings of satisfaction it produced, led to greater prejudice.
1033
The implication of these results is that not only are negative relative outcomes (relative deprivation) associated with prejudice but relatively positive ones (relative gratification) are as well. Applied to the South African context, these findings suggest that hostility toward immigrants might be associated not only with perceptions of economic deprivation but also with perceptions of economic improvement or gratification. When the relative deprivation and relative gratification effects are combined, a V-curve relationship between perceptions of relative economic condition and intergroup hostility should result.
The V-Curve Hypothesis and Intergroup Attitudes Integrating the roles of relative deprivation and relative gratification in the regulation of intergroup attitudes may provide for a more functional theoretical framework than the classic one derived from relative deprivation on its own. Although relative deprivation theory proposes a linear relationship between the perception of relative economic conditions and negative intergroup attitudes, the relative gratification perspective suggests a V-curve or bilinear relationship. The only empirical evidence of a bilinear relationship to date arises from a single experiment (Guimond & Dambrun, 2002), raising the possibility that it reflects a laboratory artifact. Using a representative sample of South Africans should allow for a test of the predictive validity of the role of relative gratification and the resulting bilinear relationship. The impact of relative gratification on prejudice may well have been ignored in past research because of a methodological bias. In studies dealing with relative deprivation, the effect of relative deprivation on prejudice has been tested by applying a linear model of prediction (e.g. Pettigrew & Meertens, 1995). Consistently these studies have revealed that relative deprivation varies positively and linearly with the expression of prejudice toward outgroups; the more participants feel deprived, the more they express negative attitudes toward outgroups. However, the relative gratification perspective suggests a different form of relationship between the relative deprivation– gratification continuum and prejudice. Specifically, from this new perspective, both relative deprivation and relative gratification will be associated with greater prejudice. The linear model cannot test the comparative effects of relative deprivation and relative gratification, as it tests only the general effect for the relative deprivation– gratification continuum. A bilinear model of prediction, however, allows for a test of the combined effects of relative deprivation and relative gratification. This pattern of relationship has never been tested in the context of prejudice, which may explain why few studies report an effect for relative gratification. In what is probably the only existing study that has systematically considered both relative deprivation and relative gratification, Grofman and Muller (1973) suggested that both may have similar effects. In their article “The Strange Case of Relative Gratification and Potential for Political Violence: The V-Curve Hypothesis,” they reported that the greatest potential for political violence is manifested both by individuals who perceive negative change (relative deprivation) and by those who perceive positive change (relative gratification). The first objective of the present study was to test both the linear and bilinear functions of the relative deprivation– gratification continuum on attitudes toward immigrants to South Africa. Following relative deprivation theory, we hypothesized a significant linear relationship between relative deprivation and negative atti-
1034
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
tudes toward African immigrants, such that those expressing greater hostility would be those who feel the most deprived. However, on the basis of the relative gratification perspective, we also hypothesized a significant bilinear relationship, which would indicate that both relative deprivation and relative gratification are associated with greater levels of hostility toward immigrants. Thus, we hypothesized that those who feel worse off (relative deprivation) and those who feel better off (relative gratification) would display more prejudice.
Toward an Explanation of Relative Gratification on Intergroup Attitudes? Identifying Mediators According to social identity theory (Tajfel & Turner, 1979), people are motivated to maintain or achieve a positive social identity. Because the self-concept is partially derived from group membership, which positively influences self-esteem through favorable intergroup comparisons, people tend to identify more strongly with the group to which they belong than with outgroups. In the context of relative gratification, people may identify even more strongly with their social group. A favorable comparison, such as perceived economic improvement in the country or perceived improvement in group status, would underlie a state of relative gratification. In this context, we argue that people may feel more pride in their own group and more attracted to it (see Doosje, Spears, & Ellemers, 2002). This should result in stronger ingroup identification among people perceiving relative gratification. It has been well established that stronger ingroup identification is associated with increases in ingroup bias and outgroup derogation (e.g. Perreault & Bourhis, 1999). Consequently, on the basis of social identity theory, we predicted a mediational model in which group identification mediates the effect of relative gratification on intergroup attitudes. Specifically, in the South African context, the second objective of this study was to test the hypothesis that strength of ethnic identification would mediate the effect of relative gratification in terms of prejudice toward immigrants.
Group Status, Relative Gratification, and Intergroup Attitudes: Identifying Moderators A final aim of the present study was to determine whether economic relative gratification produces hostility toward all outgroups, suggesting that it generates generalized prejudice, or whether, to the contrary, it leads to hostility toward specific outgroups. We suspect that economic relative gratification produces intergroup hostility and also that the participant’s own socioeconomic status (SES) moderates which outgroup will be targeted. Specifically, we predicted that low status outgroups would be the target of low SES people perceiving economic relative gratification and that high status outgroups would be the target of high SES people perceiving relative gratification. From our relative gratification perspective, a favorable economic comparison should lead people to perceive economic relative gratification, which in turn would motivate them to support ideologies that maintain their relative advantage. Derogation of relevant immigrant groups permits people perceiving economic relative gratification to justify and maintain their advantage. This process is consistent with the major assumption of the instrumental model of group conflict (Esses, Jackson, & Armstrong, 1998),
which claims that when economic gains and losses are at stake, people are motivated by their own economic interests. Indeed, it can be argued that people perceiving economic relative gratification are motivated to maintain their economic advantage in order to foster their own interests and that this leads to greater prejudice toward relevant outgroups. However, we suggest that the specific outgroups targeted for derogation are dependent on the SES of the people perceiving relative gratification. Specifically, when people perceive economic relative gratification, they are motivated to maintain their advantaged position by derogating groups that are perceived as potential competitors. It has long been demonstrated that the perception of economic competition is associated with intergroup hostility (Campbell, 1965; Levine & Campbell, 1972; Sherif, 1966) and derogation of immigrants (Esses et al., 1998; L. M. Jackson & Esses, 2000). In the context of relative gratification, we suspect that low and high SES people do not perceive the same competitors. According to the instrumental model of group conflict, “for dimensions relevant to obtaining resources, groups that are similar to the ingroup are more likely to be seen as competitors” (Esses et al., 1998, p. 704). Thus, low SES South Africans perceiving relative gratification should perceive low status immigrants as potential competitors (e.g., African immigrants in South Africa) but would tend to disregard high status immigrants as potential competitors (e.g., Western immigrants in South Africa). Because, in this specific case, low status immigrants should be perceived as more threatening than high status immigrants, we hypothesized that among low SES South Africans, relative gratification should be more strongly associated with prejudice toward African immigrants than with prejudice toward Western immigrants. Among high SES South Africans experiencing relative gratification, we hypothesized the opposite pattern of results. We argue that high status immigrants will be perceived as more threatening than low status immigrants by high SES people perceiving economic relative gratification. Consequently, we hypothesized that among high SES South Africans, economic relative gratification should be more strongly associated with greater levels of prejudice toward high status immigrants (Western immigrants) than toward low status immigrants (African immigrants).
Method Participants A total of 1,600 South Africans were surveyed, with the sample being drawn from official census data (from the 1996 Census by Statistics South Africa; see http://www.statssa.gov.za/census01/Census96/HTML/default .htm) and information from national organizations that attempt to maintain population statistics. We used this information to draw a clustered, randomly stratified, nationally representative sample. This representative sample comprised 800 women and 800 men and 869 Blacks, 372 Whites, 208 Coloreds, and 151 Indians/Asians.1 The average age was 38.3 (SD ⫽ 16.31; range ⫽ 16 to 99).
Procedure The sampling procedure involved randomly selecting a series of primary sampling units (PSUs) from a larger list of suburbs and magisterial districts, with the chance of selection being weighted proportionately by the
1 These labels (i.e., Black, White, Colored, and Indian/Asian) are used for official categorization in South Africa and are commonly used among all people in South Africa.
PREJUDICE AND THE V-CURVE HYPOTHESIS population of the suburb or the district. Once a PSU had been established, maps were used to select, at random, a place to begin interviewing. Interviewers were then required to walk in a randomly determined direction and conduct an interview at every nth home, depending on how many interviewers were required within that designated PSU. The use of such a detailed sampling procedure ensured that no systematic bias affected the sampling procedure. Once the survey was completed, the 1,600 respondents were compared with existing population statistics, and the data were weighted according to any discrepancies. Thus, a combination of careful sample selection and postsample analyses corrections yielded a sample that accurately represented the population of South Africa. To further guard against potential bias, surveyors followed strict rules once a household had been selected for inclusion in the sample. They were first required to list all household members over the age of 18. From this list, the surveyor chose the actual person to be interviewed according to a preestablished random schedule. Once the person was selected, the interviewer made three attempts to schedule an interview. Only after three failed attempts was the interviewer allowed to replace that person following the same procedure at a predetermined randomly selected replacement household. The logistics of preparing a nationwide survey of this magnitude are formidable. For example, it was necessary to have the survey instrument translated from English into the other 10 official languages and then back-translated into English via the double-blind method for us to ensure that translations reliably communicated the intended meanings. Also, co-ethnic interviewers had to be recruited so that respondents would be interviewed by someone who could speak their language fluently. The survey instrument was designed so that respondents were required to answer questions in a standard format, but one that offered them a range of response alternatives. The interviewer, therefore, was required to pose the questions in a predetermined order. The order of questions was carefully determined to proceed from simple to complex questions and from nonpersonal to more socially sensitive questions.
Questionnaire A wide variety of measures were used in the questionnaire. Of particular relevance for the present study were items that focused on perceptions of relative deprivation and gratification, ethnic identification, and attitudes toward immigrants. At the end of the questionnaire, participants were also asked to indicate their age, gender, ethnicity, education level, and annual income. Measures of relative deprivation and gratification. Our objective was to measure a general state of relative deprivation– gratification. Thus, on
1035
the basis of previous research (Guimond & Dambrun, 2002; Guimond & Dube´-Simard, 1983; Pettigrew & Meertens, 1995; Runciman, 1966), eight items similar in content and design to previous scales were selected (see Table 1). Specifically, two components of relative deprivation– gratification were assessed: economic relative gratification– deprivation (Items 1, 2, 3, and 4) and overall relative gratification– deprivation (Items 5, 6, 7, and 8). The internal consistency of this eight-item scale was found to be satisfactory (␣ ⫽ .76). All items used 5-point rating scales ranging from 1 (very satisfied) to 5 (very dissatisfied) for Items 1 and 3 and ranging from 1 (much better) to 5 (much worse) for Items 2, 4, 5, 6, 7, and 8. Thus, higher scores indicate greater perceptions of relative deprivation. Conversely, lower scores indicate greater perceived relative gratification. Measure of ethnic identification. Participants were asked to rate on a six-item scale the extent to which they identified with their national ethnic group (i.e., Black, White, Colored, and Indian/Asian; see Footnote 1). To assess ethnic identification, we used six items: “Being a X is a very important part of how you see yourself” (Item 1); “You would want your children to think of themselves as X” (Item 2); “It makes you feel proud to be a X” (Item 3); “You feel much stronger ties to X, than to other South Africans” (Item 4); “Of all the groups in South Africa, X are the best” (Item 5); and “X are very different from other South Africans” (Item 6). Three of these items do not involve any explicit intergroup comparisons (Items 1, 2, and 3), whereas the remaining three do involve an intergroup comparison (Items 4, 5, and 6). The internal consistency of this six-item scale was found to be satisfactory (␣ ⫽ .79). All items used 5-point rating scales ranging from 1 (strongly disagree) to 5 (strongly agree). The six items were drawn mainly from three different scales designed to assess social identity (Brown, Condor, Mathews, Wade, & Williams, 1986; Garza & Herringer, 1987; S. E. Jackson, 1981). Measures of intergroup attitudes. Participants were asked to rate on a 10-point scale the extent to which they felt unfavorable (0) or favorable (10) toward their own ethnic ingroup (i.e.. Blacks, Whites, Coloreds, and Indians/ Asians; see Footnote 1; M ⫽ 8.58, SD ⫽ 1.91) and toward five immigrant groups (␣ ⫽ .88): people living in South Africa from Zimbabwe, Mozambique, Lesotho, and other African countries (i.e., African immigrants; ␣ ⫽ .90; M ⫽ 4.22, SD ⫽ 2.16) and people living in South Africa from countries in Europe and North America (i.e., Western immigrants; M ⫽ 4.89, SD ⫽ 2.38). It is important to note that participants expressed their attitudes toward their ingroup and the different outgroups and could not logically be a member of any outgroup. Only participants who were South African (i.e., having the South African nationality) were interviewed. On the basis of these different ratings, we constructed four indicators: (a) By averaging the four ratings for African immigrants, we created a measure
Table 1 Means and Standard Deviations of Relative Deprivation–Gratification Items Item 1. At the moment are you (satisfied/dissatisfied) with your personal economic conditions? 2. Do you expect that your personal economic conditions will get (better/ same/worse) one year from now? 3. At the moment are you (satisfied/dissatisfied) with economic conditions in South Africa? 4. Do you expect that economic conditions in South Africa will get (better/ same/worse) one year from now? 5. Would you say that your overall personal conditions are (better/same/worse) than those of other South Africans? 6. Would you say that your overall personal conditions are (better/same/worse) as other (members of the ingroup)? 7. Would you say that the overall conditions of people from your ingroup are (better/same/worse) than those of other groups in South Africa? 8. Would you say that the overall conditions of South Africa are (better/same/worse) than those in other Southern African countries?
M
SD
3.19
1.12
2.8
1.05
3.55
1.04
2.99
1.17
2.77
0.84
2.82
0.76
2.87
0.82
2.35
0.86
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
1036
of attitudes toward African immigrants; (b) ratings of Western immigrants provided a measure of attitudes toward Western immigrants; (c) by subtracting the ratings for African immigrants (i.e., derogation of African immigrants) from the ratings for the respondents’ own group (i.e., ingroup bias), we arrived at a measure of prejudice toward African immigrants; and finally, (d) by subtracting the ratings for Western immigrants from the ratings for their own group, we obtained a measure of prejudice toward Western immigrants. Research has provided compelling evidence for the validity of such relative measures of prejudice (see Castano, Yzerbyt, Paladino, & Sacchi, 2002; Guimond & Dambrun, 2002; Guimond & Palmer, 1993; Levin, Federico, Sidanius, & Rabinowitz, 2002; Sidanius & Pratto, 1999). Typically, the difference between the rating of the ingroup and the rating of the outgroup is a more sensitive measure of prejudice than the rating of the outgroup alone (Guimond et al., 2003). Consequently, these were the two measures retained as dependent variables: prejudice toward African immigrants and prejudice toward Western immigrants. Confirmatory factor analysis. To verify the validity of our different scales, we performed a confirmatory factor analysis. Specifically, we compared our predicted model with the null model (i.e., in which all the items load on the same latent factor). The predicted model was composed of three latent variables: relative gratification– deprivation, ethnic identification, and prejudice. Two distinct components made up both the relative gratification– deprivation variable (i.e., economic vs. overall relative deprivation– gratification) and the ethnic identification variable (i.e., ethnic identification without vs. with an intergroup comparison). Confirming the validity of our different scales, the predicted model, 2(95, N ⫽ 1,600) ⫽ 651.18, p ⬍ .001, comparative fit index (CFI) ⫽ .94, goodness-of-fit index (GFI) ⫽ .95, normed fit index (NFI) ⫽ .93, root-mean-square error of approximation (RMSEA) ⫽ .06, fitted significantly more with the data than did the null model, 2(104, N ⫽ 1,600) ⫽ 5,645.90, p ⬍ .001, CFI ⫽ .36, GFI ⫽ .65, NFI ⫽ .36, RMSEA ⫽ .18. The chi-square difference between the two models was significant, ⌬2(9, N ⫽ 1,600) ⫽ 4,994.72, p ⬍ .001.
Results The Effects of Relative Deprivation–Gratification on Intergroup Attitudes To test our main hypothesis that the bilinear function of the relative gratification– deprivation continuum will provide a better fit of the data than the linear function, we compared two regression models in which the relative gratification– deprivation scores were centered at the grand mean. The first model, labeled the reduced model, corre-
sponds to the classical linear model. It includes a single independent variable (i.e., relative gratification– deprivation). The second model, referred to as the full model, corresponds to the bilinear model. It includes two independent variables. The first (i.e., relative gratification) included only the ratings equal to or below the median (i.e., zero). The second independent variable (i.e., relative deprivation) included the rating above the median. In both cases, zeros replaced ratings of excluded participants. The full model assumes a different slope for participants who perceived relative gratification than for those who perceived relative deprivation. Testing the equality of these two slopes allowed us to choose which of the two models, the full or the reduced, is the most appropriate. The statistical procedure, for which theoretical support can be found in Brook and Arnold (1985), involves determining the extent to which using the full model (i.e., different slopes) in comparison with the reduced model (i.e., single slope) results in a significant increment of explained variance. Hence, as in a hierarchical multiple regression analysis (e.g., Cohen & Cohen, 1983) both the increment in R2 (i.e., I) and its statistical significance (i.e., FI) were computed. A significant increment in R2 means that using two different slopes (i.e., bilinear model) provides a better fit of the data than using the classical linear function (i.e., single slope). For each analysis, the eight-item relative gratification– deprivation scale was used as the independent variable and the measures of intergroup attitudes as the dependent variables. Results from these statistical analyses are presented in Table 2. First, with respect to prejudice toward African immigrants, the test of the difference between the two models was significant (I ⫽ .015, p ⬍ .001), showing that the bilinear model better fitted the data than the classical linear one. The reduced model was not significant. The two slopes included in the full model were significantly different from zero and the estimates of these slopes were of opposite signs (see Table 2). As Figure 1A illustrates, participants displayed significantly greater prejudice toward African immigrants to the extent that they felt either gratified or deprived. In terms of prejudice toward Western immigrants, both the reduced and full models were significant (see Table 2). However, the test of the difference between the two models was significant (I ⫽ .010, p ⬍ .001). The slope of the reduced model was not in
Table 2 The Effects of Relative Deprivation–Gratification on Intergroup Attitudes Hierarchical analysis Model
Cum. R
Cum. R2
F
Coefficients for the full model I
F1
Independent variable
Estimate
SE

sr22
3.927*** ⫺0.707*** 1.281***
.12 .22 .27
⫺.093 .136
.007 .015
3.221*** ⫺1.126*** 0.589*
.14 .24 .30
⫺.140 .059
.016 .003
Prejudice toward African immigrants Reduced Full
.028 .126
.001 .016
1.17 11.58***
.001 .015
1.17 21.98***
Constant XRG XRD
Prejudice toward Western immigrants Reduced Full
.076 .125
.006 .016
7.93** 10.74***
.006 .010
7.93** 13.47***
Constant XRG XRD
Note. Relative gratification– deprivation scores were centered at the grand mean for each analysis. Estimate, SE, and  for each term are controlled for the other term. I ⫽ increment in R2; F1 ⫽ test of the significance for the R2 increment; sr22 ⫽ squared semipartial correlation for one term controlling for the other term; Reduced ⫽ linear model (single slope); Full ⫽ bilinear model (two slopes); XRG ⫽ relative gratification term; XRD ⫽ relative deprivation term. * p ⬍ .05. ** p ⬍ .01. *** p ⬍ .001.
PREJUDICE AND THE V-CURVE HYPOTHESIS
The Effect of Relative Deprivation–Gratification on Ethnic Identification
A 10
9
Prejudice toward African immigrants
8
7
6
5
4
3
2
1
0 -2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
High < R. Gratification > Low < R. Deprivation > High
B
Using the same statistical procedure, we tested both the reduced and the full model using the full scale of relative deprivation– gratification2 (centered at the grand mean) as the independent variable and the ethnic identification scale as the dependent variable. Both the reduced model, R ⫽ .135, R2 ⫽ .018, F(1, 1596) ⫽ 29.81, p ⬍ .001, and the full model, R ⫽ .151, R2 ⫽ .023, F(2, 1595) ⫽ 18.51, p ⬍ .001, were statistically significant. However, the test of the difference between the two models was significant, I ⫽ .004, FI(2, 1595) ⫽ 7.10, p ⬍ .008. The slope of the reduced model reveals that the more the participants felt gratified on the relative gratification– deprivation continuum, the more they identified with their ethnic group (Estimate ⫽ ⫺.154, SE ⫽ .028,  ⫽ ⫺.135, p ⬍ .001, R2 ⫽ .018). However, the bilinear function reveals that it was only those who felt more gratified who identified more strongly with their ethnic group (Estimate ⫽ ⫺.261, SE ⫽ .049,  ⫽ ⫺.147, p ⬍ .001, sr22 ⫽ .017). The relationship between relative deprivation and ethnic identification was not significant (Estimate ⫽ ⫺.01, SE ⫽ .059,  ⫽ ⫺.007, ns; see Figure 2).
Ethnic Identification as a Mediator of the Effect of Relative Gratification on Intergroup Attitudes
10
9
Because we predicted that ethnic identification would mediate the effect of relative gratification on prejudice, we used a median split to isolate the effect of relative gratification.3 Including in the analysis respondents who perceived relative deprivation would not allow for a test of our specific hypothesis. We predicted that ethnic
8
Prejudice toward Western immigrants
1037
7
6
5
4
3
2
1
0 -2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
High < R. Gratification > Low < R. Deprivation > High
Figure 1. Bilinear relationship between the relative gratification– deprivation continuum and prejudice toward African (A) and Western (B) immigrants. R. ⫽ Relative.
the hypothesized direction. Revealing a main effect of relative gratification, the more participants felt gratified, the more they displayed prejudice toward Western immigrants (b ⫽ ⫺.398, SE ⫽ .141,  ⫽ ⫺.076, p ⬍ .01, R2 ⫽ .006). Concerning the full model, the two slopes were significantly different from zero and the estimates of these slopes were opposite in sign (see Table 2). As shown in Figure 1B, both relative deprivation and relative gratification were associated with greater levels of prejudice toward Western immigrants. Finally, additional analyses revealed that when we controlled for age, gender, ethnicity, and education, both the linear and bilinear effects of relative deprivation– gratification maintained their levels of significance.
2 Our general hypothesis suggests that perceived relative gratification results in ethnic identification. However, because South Africans have important group identities at different levels of inclusion (e.g., South African, Black, Xosa), it is reasonable to propose a matching process such that different levels of relative gratification will have the most impact on the corresponding level of identity. Future research needs to examine this possibility. Because, in the present study, the scale of relative gratification was composed of items focusing on various dimensions, we examined whether some items were related more closely with the scale of national ethnic identification than others. Of interest, except for Item 1 (personal economic satisfaction), all items assessing relative gratification were significantly related to ethnic identification. Thus, confirming our general hypothesis, the more the participants felt gratified in South Africa, the more they identified with their national ethnic group. nificantly related to ethnic identification. Thus, confirming our general hypothesis, the more the participants felt gratified in South Africa, the more they identified with their national ethnic group. 3 At this point, it is important to note that the full model (bilinear) tests the joint effect of relative gratification and relative deprivation on prejudice. Thus, testing whether the bilinear function is significantly mediated by the measure of ethnic identification would not allow us to test our specific hypothesis that ethnic identification should mediate the effect of relative gratification on prejudice. Similarly, concerning our moderation hypothesis, testing whether the bilinear function is significantly moderated by SES would not allow us to test our specific hypothesis that SES should moderate the effect of relative gratification. No hypothesis has been formulated for the effect of relative deprivation; thus including participants perceiving relative deprivation would result in a shift between our specific hypotheses and the analyses.
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
1038 5
Low < Strength of ethnic identification > High
4.5
4
3.5
3
2.5
2
1.5
1 -2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
High < R. Gratification > Low < R. Deprivation > High
Figure 2. Bilinear relationship between the relative gratification– deprivation continuum and ethnic identification. R. ⫽ Relative.
identification would mediate the effect of relative gratification but not the effect of relative deprivation. Thus, respondents perceiving relative deprivation were excluded from the mediation procedure. A score of 0 on the relative gratification– deprivation measure (centered at the grand mean) corresponded both to the median and to a neutral score. To test the linear function of the relative gratification effect, we had to include neutral scores as a baseline in the regression procedure. Thus, on the basis of their relative gratification– deprivation score, participants were split at the median into deprived (n ⫽ 661) and gratified (n ⫽ 938) groups. More participants constituted the gratified group only because a relatively large proportion of the sample scored at the median. In order to test the mediating role of ethnic identification in the relationship between relative gratification and intergroup attitudes, we followed the regression procedure advocated by Baron and Kenny (1986). The first requirement is that the independent variable (i.e., relative gratification) be related to the dependent variable (i.e., intergroup attitudes). As shown in Figure 3A, relative gratification was significantly and positively related to prejudice toward African immigrants ( ⫽ .074, p ⬍ .03, R2 ⫽ .005)4 and prejudice toward Western immigrants ( ⫽ .126, p ⬍ .001, R2 ⫽ .016). Second, the mediating variable (i.e., ethnic identification) should be related to the independent variable and the dependent variable. As Figure 3A illustrates, these two requirements were satisfied. Ethnic identification and relative gratification correlated positively and significantly ( ⫽ .199, p ⬍ .001, R2 ⫽ .04). Moreover, both prejudice toward African immigrants ( ⫽ .101, p ⬍ .003, R2 ⫽ .01) and prejudice toward Western immigrants ( ⫽ .094, p ⬍ .007, R2 ⫽ .009) were significantly and positively related to ethnic identification. The final and most basic requirement specified by Baron and Kenny (1986) is that a mediating variable should predict the dependent variable even when the independent variable is statistically controlled, whereas the effect of the independent variable on the dependent measure should be significantly reduced when the mediating variable is statistically controlled. This requirement was tested separately for the two measures of intergroup attitudes.
As expected, the effect of relative gratification on prejudice toward African immigrants was rendered nonsignificant when the measure of ethnic identification was statistically controlled ( ⫽ .056, p ⬎ .105, sr22 ⫽ .003). But the effect of ethnic identification on prejudice toward African immigrants remained significant even when relative gratification was statistically controlled ( ⫽ .090, p ⬍ .009, sr22 ⫽ .008). Consistently, the Sobel test, performed to test the significance of the mediation, was significant (z ⫽ 2.44, p ⬍ .01; see Figure 3A), and 24.3% of the effect of relative gratification on prejudice toward Africans immigrants was mediated by ethnic identification, indicating a partial mediation. Confirming the validity of our model, we found no support for the reversed model (see Figure 3B). Finally, the effect of ethnic identification on prejudice toward Western immigrants remained significant even when relative gratification was statistically controlled ( ⫽ .070, p ⬍ .05, sr22 ⫽ .005). The effect of relative gratification on prejudice toward Western immigrants was significantly reduced (z ⫽ 2.18, p ⬍ .05) but remained significant when the measure of ethnic identification was statistically controlled ( ⫽ .111, p ⬍ .002, sr22 ⫽ .012), indicating a partial mediation (see Figure 3C). A relatively small portion of the effect of relative gratification on prejudice toward Western immigrants was mediated by the measure of ethnic identification (12%). Again, confirming the validity of our model, we found no support for the reversed model (see Figure 3D).
SES as a Moderator of the Effect of Relative Gratification on the Target of Prejudice Among low SES South Africans, we hypothesized that relative gratification would be more strongly associated with prejudice toward African immigrants than with prejudice toward Western immigrants. Conversely, among high SES South Africans, we predicted that relative gratification would be more strongly associated with greater levels of prejudice toward Western immigrants than with prejudice toward African immigrants. Because we predicted this interaction only with scores of relative gratification, not with scores of relative deprivation, people perceiving relative deprivation were excluded from the moderation procedure (see Footnote 3). Thus, on the basis of the median split used in the mediation procedure, only gratified respondents were included in the moderation analysis (n ⫽ 938). To test our specific hypothesis, we performed a regression analysis using relative gratification scores as the first independent variable, scores of annual income as the second independent variable (i.e., SES), and the interaction between scores of relative gratification and scores of annual income as the third 4 In order to be sure that the significance of the effect of relative gratification on prejudice was not simply due to large sample size, we performed additional analyses. These analyses revealed that the linear effect of relative gratification on prejudice remained significant even when we used a very restrictive value of selection of participants. For example, with a very restrictive selection (i.e., selection of participants whose scores were ⱕ⫺1 on the centered relative gratification– deprivation scores; n ⫽ 121), the effect of relative gratification on prejudice toward African immigrants is significant and accounts for a larger part of the variance ( ⫽ ⫺.187, R2 ⫽ .035) than it does with a less restrictive standard selection based on a median split ( ⫽ ⫺.074; R2 ⫽ .005, n ⫽ 872).
PREJUDICE AND THE V-CURVE HYPOTHESIS
1039
Target group: AFRICAN IMMIGRANTS z = 2.44, p < .01
A: Predicted Model .199*** RELATIVE GRATIFICATION
.090** ETHNIC IDENTIFICATION
PREJUDICE TOWARD AFRICAN IMMIGRANTS
.056 ns (.074*) z = .87, ns
B: Reversed Model .074* RELATIVE GRATIFICATION
PREJUDICE TOWARD AFRICAN IMMIGRANTS
.086** ETHNIC IDENTIFICATION
.196*** (.199***)
Target group: WE STERN IMMIGRANTS z = 2.18, p < .05
C: Predicted Model .199*** RELATIVE GRATIFICATION
.070* ETHNIC IDENTIFICATION
PREJUDICE TOWARD WESTERN IMMIGRANTS
.111** (.126***) z = -1.66, ns
D: Reversed Model .126*** RELATIVE GRATIFICATION
PREJUDICE TOWARD WESTERN IMMIGRANTS
.068** ETHNIC IDENTIFICATION
.207*** (.199***) Figure 3. Ethnic identification as a mediator of the effect of relative gratification on prejudice toward both African and Western immigrants. The values in parentheses represent the beta coefficients without controlling for the mediating variable. *p ⬍ .05; **p ⬍ .01; ***p ⬍ .001.
and final independent variable. The two first variables were standardized. By subtracting the scores of prejudice toward Western immigrants from the scores of prejudice toward African immigrants, we obtained a new dependent variable.
Higher scores on this dependent variable indicate greater prejudice toward African immigrants than toward Western immigrants. Neutral scores indicate that both outgroups are derogated equally. Conversely, lower scores indicate greater
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
1040
prejudice toward Western immigrants than toward African immigrants. Results from the regression analysis are presented in Table 3. As expected, the interaction between relative gratification and SES (i.e., annual income) was significant ( ⫽ ⫺.087, p ⬍ .024, sr22 ⫽ .007). The pattern of this interaction is depicted in Figure 4. As expected, the interaction reveals that the more high SES South Africans perceived relative gratification, the more they derogated Western immigrants compared with African immigrants ( ⫽ .159, p ⬍ .005, R2 ⫽ .025). However, contrary to our expectation, low SES South Africans perceiving relative gratification were not more prejudiced toward African immigrants than toward Western immigrants ( ⫽ ⫺.01, ns). They targeted both outgroups equally.
High SES
Target(s) of prejudice: Western < both > African
2
1
0
-1
Discussion Using a large national sample from South Africa, the present research provides the first major test, in a natural setting, of a new theoretical dimension of prejudice. Over the past several decades, relative deprivation theorists have documented the role of relative deprivation in the explanation of intergroup attitudes and behaviors, suggesting that the more people feel deprived, the more likely they are to display outgroup prejudice. The theoretical implication of this legacy of research is that the less people feel deprived, the less likely they are to display negative intergroup attitudes and behaviors. In the present research, we have added a new dimension by pointing to the role that relative gratification may play in the understanding of prejudice. We predicted a bilinear rather than a linear relationship between the relative gratification– deprivation continuum (i.e., perception of economic conditions) and prejudice toward immigrants to South Africa. We hypothesized that both relative gratification and relative deprivation would be associated with higher levels of intergroup hostility. We found very strong support for a bilinear function. As hypothesized, this V-curve relationship reveals that both relative gratification and relative deprivation are associated with greater levels of prejudice toward both African and Western immigrants to South Africa. Although the general linear function was relatively poorly related to intergroup attitudes, the bilinear equation accounted for a significantly greater percentage of the explained variance with regard to prejudice and was always highly signifi-
Low SES
3
-2
-3 -2
-1.5
-1
-0.5
0
High < Relative Gratification > Low
Figure 4. Socioeconomic status (SES) as a moderator of the effect of relative gratification on the target of prejudice. Higher scores on the dependent variable indicate greater prejudice toward African immigrants than toward Western immigrants. Neutral scores indicate that both outgroups are derogated equally. Conversely, lower scores indicate greater prejudice toward Western immigrants than toward African immigrants.
cant. Moreover, the direction of the linear function was not always in the expected direction. Indeed, although the linear function was related positively but not significantly to prejudice toward African immigrants, it was significantly related negatively to prejudice toward Western immigrants. Thus, the direction of the relationship seems to depend on the status of the outgroup. A classic but nonsignificant effect of relative deprivation arises in the case of a low status outgroup, and an opposite effect arises in the case of the higher status outgroup. However, these general tendencies mask a more complex but systematic pattern of results in which both relative deprivation and relative gratification are associated with higher levels of prejudice toward both African and Western im-
Table 3 Socioeconomic Status as a Moderator of the Effect of Relative Gratification on the Target of Prejudice Independent variable
Estimate
SE

t
p
sr22
Constant Relative gratification Income Relative Gratification ⫻ Income
.411 .164 ⫺.477 ⫺.226
.087 .087 .088 .100
.073 ⫺.207 ⫺.087
4.75 1.89 5.39 2.26
.0001 .059 .0001 .024
.005 .040 .007
Note. Relative gratification and annual income scores were centered at the grand mean. Concerning the first independent variable, lower scores indicate greater perception of relative gratification. For the second independent variable, lower scores indicate greater annual income. The difference scores between the two measures of prejudice were used as the dependent variable. Higher scores on this dependent variable indicate greater prejudice toward African immigrants than toward Western immigrants. Neutral scores indicate that both outgroups are derogated equally. Conversely, lower scores indicate greater prejudice toward Western immigrants than toward African immigrants. sr22 ⫽ squared semipartial correlation for one term controlling for the other term.
PREJUDICE AND THE V-CURVE HYPOTHESIS
migrants. In fact, we suggest that the difference in the direction of the linear relationship between high and low status outgroups simply indicates the general tendency of the distribution, concealing a more complex V-curve relationship. Whereas the nonsignificant linear relative deprivation effect (i.e., positive linear slope) on prejudice toward African immigrants masks the existence of the effects of both relative gratification and relative deprivation, the linear relative gratification effect (i.e., negative linear slope) on prejudice toward Western immigrants tends to mask the existence of a significant relative deprivation effect. Thus, the results strongly suggest that analyses of the linear relationship between the relative perception of both economic and overall conditions and intergroup phenomena are not sufficient. Such a traditional analysis can reveal a significant but relatively spurious relationship while concealing a more complex one. The V-curve effect is found to be strongly consistent across measures of intergroup attitudes. As shown by Figures 1A and 1B (see also Table 2), the bilinear function (i.e., full model) significantly predicts the measures of intergroup attitudes in each of the two cases. On the basis of this V-curve relationship, both relative deprivation and relative gratification were associated with greater levels of prejudice toward immigrants. These results confirm the key role of relative deprivation but in addition reveal that the relative gratification effect on intergroup attitudes, previously largely unexplored and undocumented, is robust and not limited to the French and the laboratory context (Grofman & Muller, 1973; Guimond & Dambrun, 2002; Guimond et al., 2003).
The V-Curve Hypothesis: The Mediating Role of Ethnic Identification Previous research has documented that the effect of relative deprivation is mediated by negative feelings, the affective component of relative deprivation (see Grant & Brown, 1995; Guimond & Dube´-Simard, 1983). If negative emotions mediate the relative deprivation effect, what explains the effect of relative gratification? An important aim of the present study was to explore the potential mediating role of ethnic identification. Specifically, we hypothesized that when South Africans perceive a general improvement in both their own personal conditions and their ingroup conditions, they might feel more pride in their ingroup and more attracted to it. However, because South Africans have important group identities at different levels of inclusion (e.g., South African, Black, Xosa), it is reasonable to propose a matching process such that different levels of relative gratification will have the most impact on the corresponding level of identity. Future research needs to examine this possibility. Concerning our prediction, because stronger ingroup identification tends to be related to an increased bias against outgroups (e.g. Perreault & Bourhis, 1999), ingroup identification should act as a mediator of the effect of relative gratification on intergroup attitudes. The results of the present study provide relatively clear support for this hypothesis. The model by which ethnic identification mediates the effect of relative gratification on prejudice toward both Africans and Western immigrants received stronger support than the reverse model by which prejudice mediates the effect of relative gratification on ethnic identification (see Figure 3). Moreover, for both measures of prejudice, statistical analyses revealed that ethnic identification is a signifi-
1041
cant mediator. However, because ethnic identification mediates, respectively, 24% and 12% of the effect of relative gratification on prejudice toward Africans immigrants and on prejudice toward Western immigrants, we conclude that ethnic identification is a partial rather than a full mediator and, consequently, that other variables also underlie the effect of relative gratification. For example, when people are in a state of relative gratification, they find themselves in a privileged position (Kawakami & Dion, 1995). Greater prejudice toward outgroups may emerge in an attempt to justify and maintain such privileges. As Crocker, Major, and Steele (1998) argued, “People of higher status may stigmatize those of lower status to justify their advantages” (p. 509). Clearly, the role of justification processes in understanding the phenomenon of relative gratification needs to be investigated.
The V-Curve Hypothesis, SES, and Target Outgroup Status The results of the present study provide mixed support for our moderation hypothesis. As predicted, for high SES South Africans, relative gratification was associated with higher levels of prejudice toward Western immigrants than toward African immigrants. However, low SES South Africans perceiving relative gratification derogated African and Western immigrants to the same extent. In other words, contrary to our expectation, low status participants perceiving relative gratification do not seem more inclined to derogate low status outgroups than high status outgroups. Thus, these results seem to partially confirm our hypothesis derived from the instrumental model of group conflict (Esses et al., 1998), which predicted that relevant outgroups are not the same for low and high SES participants who perceive relative gratification. We argued that in order to maintain their perceived improving economic conditions, people derogate outgroups identified as potential competitors who pose a threat to the maintenance of their advantaged position. Results seem to suggest that both African and Western immigrants are identified as potential competitors by low SES South Africans perceiving relative gratification. Such was not the case for high SES respondents. These results are relatively consistent with the basic premise that perceptions of competition, economic competition, and threat to status lead to hostile intergroup attitudes (Brown, 1995; Campbell, 1965; Esses et al., 1998; L. M. Jackson & Esses, 2000; Sherif, Harvey, White, Hood, & Sherif, 1961; Stephan, Ybarra, & Bachman, 1999). Finally, the present results permit us to refine the manner in which relative gratification affects intergroup attitudes. The results of the present study are relatively consistent with the general hypothesis that people perceiving relative gratification are motivated to derogate potential outgroup competitors who may threaten the maintenance of their advantaged position. However, this hypothesis needs to be tested more directly in future research.
Integrating the V-Curve Perspective With Related Lines of Research Five decades of research on relative deprivation has concluded that unfavorable comparisons generate feelings of dissatisfaction, and these lead to prejudice and intergroup hostility. The results of the present research suggest that favorable comparisons can also lead to hostile intergroup attitudes. It has long been argued that
1042
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
prejudice and intergroup hostility are mainly associated with negative experiences, such as negative feedback, threats to identity and self-esteem, economic threat, and frustration (Brown, 1995; Cialdini & Richardson, 1980; Dollard, Miller, Doob, Mowrer, & Sears, 1939; Fein & Spencer, 1997). Our results do not contradict this perspective. They do suggest, however, that favorable comparisons resulting in a state of relative gratification can also play an important role in the emergence of negative intergroup phenomena. Such a conclusion is by no means inconsistent with other theoretical orientations in the field of intergroup relations. For example, it has been demonstrated that high status or privileged groups tend to be more ethnocentric in their attitudes and behavior than low status and disadvantaged groups (see Bettencourt, Dorr, Charlton, & Hume, 2001; Sachdev & Bourhis, 1987). Social dominance theorists have also shown that advantaged groups display greater ingroup bias and are more favorably disposed toward inequality and hierarchical relations favoring dominant groups than are disadvantaged groups (Sidanius & Pratto, 1999). Duckitt’s (2001) theory also seems to be compatible with the relative gratification effect. In his dual process model, Duckitt and his colleagues (see Duckitt, 2001; Duckitt, Wagner, Plessis, & Birum, 2002) argued that dual motivational and cognitive processes underlie two distinct dimensions of prejudice. Specifically, “threatdriven control and security motivation and competitively driven dominance or superiority motivation” (Duckitt et al., 2002, p. 88) correspond to two independent processes that underlie prejudice. The first process mainly refers to threat and fear. The second process is more related to dominance, status, and power. Of interest, whereas the effect of relative deprivation might be related to the first process (threat), the relative gratification effect appears to be more closely related to the second process (power). In fact, because both relative gratification and high status involve a positive position on some evaluative dimension of comparison, it could be argued that relative gratification corresponds to one of the specific processes that underlies the more general concept of high status. However, both low and high status group members can perceive relative gratification. Thus, relative gratification can be seen as a dynamic process characterizing members who perceive that their personal or group situation is improving, has improved, or will improve. However, it is probable that the dynamics of both relative gratification and high status group membership share a similar sociopsychological logic. Individuals experiencing both find themselves in a relatively privileged position. We suggest that people are motivated to maintain such a position; it allows them to maintain their own interests and to occupy a valuable social position. However, maintaining such a position implies the use of strategies that require advantaged people to derogate potential threatening competitors. Because it involves a favorable comparison, relative gratification can be defined as a “positive experience” closely related to dominance, status, and power. However, in certain contexts, it seems that relative gratification can be associated with perceived competition, threat, and defense of own interests, which are more “negative experience” by nature. Thus, it would be important to recognize that in certain situations, a positive experience and advantaged economic situation can result in defensive processes that favor derogation of relevant outgroups. The role of ethnic identification reveals a more complex structure to the relative gratification effect. If the moderation of the
effect of relative gratification by both SES and outgroup target status argues in favor of an underlying defensive process, the mediating function of ethnic identification suggests another entirely different psychological mechanism. Thus, it seems that the relative gratification effect may be rooted in relatively distinct and independent processes. According to both realistic group conflict theory (Levine & Campbell, 1972; Sherif et al., 1961) and the instrumental model of group conflict (Esses et al., 1998), defensive strategies are more likely to appear when resources are relatively limited and when the economic context is bleak. Thus, in our view, a depressed and uncertain economic context may drive the defensive mechanisms that allow people to maintain their self-interest. On the other hand, we propose that the relative gratification effect also involves a motivation for superiority– dominance. The fact that in the relative gratification context, people feel more pride in their ethnic ingroup and this pride partially mediates the effect on hostile intergroup attitudes is compatible with this view. Thus, consistent with previous theory and research (e.g., Duckitt, 2001), it seems that relative gratification may be driven by two relatively independent processes.
Limitations and Future Directions Although the results of the present research support our hypotheses, the correlations between the variables are modest, accounting for a small percentage of variance (see Footnote 4). Using an experimental design and only student participants, Guimond and Dambrun (2002) found that the relative gratification effect explained approximately 10% of the variance. Using a correlational design along with a representative sample for the present study, we have shown that the explained percentage of variance for the relative gratification effect varies between ⬍1% and 4% depending on the specific measure. Similarly, the well-established relationship between ethnic identification and prejudice explained approximately 1% of the variance in the present study. Because representative samples are more heterogeneous than student samples, it is not surprising that effect sizes tend to decrease. The fact that observed effect sizes are small indicates that we need to acknowledge that the relative gratification effect accounts for a small portion of the variance in real-world terms, but it is nevertheless a significant factor. We know little about the significance of many social–psychology variables and theories in the real world. Thus, research demonstrating the significance of a social– psychological variable, such as relative gratification, represents in our view a clear advance for social psychology. Because the present study is based on a correlational design, no strong claims about causal relations among variables can be made. For example, future research needs to examine the mediating role of ethnic identification using an experimental design. In terms of the effect of relative gratification on prejudice, an important aim of the present study was to test the ecological validity of the relative gratification effect arising from a single laboratory experiment (Guimond & Dambrun, 2002). Clearly the effect of relative gratification reflects more than a mere laboratory artifact, but new insights into its underlying processes should emerge from examining the relationship between relative gratification and prejudice in a variety of different intergroup contexts.
PREJUDICE AND THE V-CURVE HYPOTHESIS
Understanding the South African Intergroup Situation Relative deprivation theory has often been applied to the context of South Africa (Appelgryn & Nieuwoudt, 1988; De La Rey & Raju, 1996; Duckitt & Mphuthing, 2002). The results of the present study offer new considerations for understanding the intergroup dynamics in South Africa. Recent research has documented that immigrants are a prime target for discrimination in South Africa (Mattes et al., 1999). Intriguingly, this hostility seems to be generalized across contextual boundaries. Even when different variables such as social class, education, age, gender, and ethnicity are analyzed, no subgroup was found to display positive attitudes toward immigrants. Following the traditional perspective of relative deprivation theory, it could be argued that hostility toward African immigrants mainly reflects strong perceptions of economic decline among the current South African population (Hepworth & West, 1988; Hovland & Sears, 1940). However, the results of the present study force a new interpretation by suggesting that perceptions of economic improvement also tend to be associated with intergroup hostility in South Africa. This is relatively consistent with the study of Green, Glaser, and Rich (1998). In their extensive analysis of the relationship between economic downturns and negative behaviors toward stigmatized outgroups (e.g., hate crimes), they found little support for the usual claim that perceptions of economic decline (relative deprivation) produce intergroup hostility. Thus, testing the linear relationship between the perception of economic downturns and intergroup hostility is not sufficient and may conceal a more complex relationship in which both relative deprivation and relative gratification are associated with greater levels of intergroup hostility.
References Appelgryn, A. E., & Nieuwoudt, J. M. (1988). Relative deprivation and the ethnic attitudes of Blacks and Afrikaans speaking Whites in South Africa. Journal of Social Psychology, 128, 311–323. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Bettencourt, B. A., Dorr, N., Charlton, K., & Hume, D. L. (2001). Status differences and in-group bias: A meta-analytic examination of the effects of status stability, status legitimacy, and group permeability. Psychological Bulletin, 127, 520 –542. Brewer, M. B., & Brown, R. J. (1998). Intergroup relations. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., pp. 554 –594). Boston: McGraw-Hill. Brook, R. J., & Arnold, G. C. (1985). Applied regression analysis and experimental design. New York: Marcel Dekker. Brown, R. J. (1995). Prejudice: Its social psychology. Oxford, England: Blackwell. Brown, R. J., Condor, S., Mathews, A., Wade, G., & Williams, J. (1986). Explaining intergroup differentiation in an industrial organization. Journal of Occupational Psychology, 59, 273–286. Campbell, D. T. (1965). Ethnocentric and other altruistic motives. In D. Leaven (Ed.), Nebraska Symposium on Motivation (Vol. 13, pp. 183– 311). Lincoln: University of Nebraska Press. Castano, E., Yzerbyt, V., Paladino, M.-P., & Sacchi, S. (2002). I belong, therefore, I exist: Ingroup identification, ingroup entitativity, and ingroup bias. Personality and Social Psychology Bulletin, 28, 135–143. Cialdini, R. B., & Richardson, K. D. (1980). Two indirect tactics of image management: Basking and blasting. Journal of Personality and Social Psychology, 39, 406 – 415.
1043
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., pp. 504 –553). Boston: McGraw-Hill. Dambrun, M., & Guimond, S. (2001). La the´orie de la privation relative et l’hostilite´ envers les Nord-Africains [Relative deprivation theory and hostility towards North Africans]. Revue Internationale de Psychologie Sociale, 14, 57– 89. Dambrun, M., Maisonneuve, C., Duarte, S., & Guimond, S. (2002). Mode´lisation de quelques de´terminants psychosociaux de l’attitude envers l’extreˆme droite [Modeling the social-psychological determinants of support for extreme right-wing movements]. Les Cahiers Internationaux de Psychologie Sociale, 55, 49 – 63. Davies, J. C. (1962). Toward a theory of revolution. American Sociological Review, 27, 5–19. De La Rey, C., & Raju, P. (1996). Group relative deprivation: Cognitive and affective components and protest orientation among Indian South Africans. Journal of Social Psychology, 136, 579 –588. Dollard, J. W., Miller, N. E., Doob, L. W., Mowrer, O. H., & Sears, R. R. (1939). Frustration and aggression. New Haven, CT: Yale University Press. Doosje, B., Spears, R., & Ellemers, N. (2002). Social identity as both cause and effect: The development of group identification in response to anticipated and actual changes in the intergroup status hierarchy. British Journal of Social Psychology, 41, 57–76. Duckitt, J. (2001). A dual-process cognitive-motivational theory of ideology and prejudice. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 33, pp. 41–112). San Diego, CA: Academic Press. Duckitt, J., & Mphuthing, T. (2002). Relative deprivation and intergroup attitudes: South Africa before and after the transition. In I. Walker & H. J. Smith (Eds.), Relative deprivation: Specification, development and integration (Vol. 4, pp. 69 –90). New York. Cambridge University Press. Duckitt, J., Wagner, C., Plessis, I., & Birum, I. (2002). The psychological bases of ideology and prejudice: Testing a dual process model. Journal of Personality and Social Psychology, 83, 75–93. Esses, V. M., Jackson, L. M., & Armstrong, T. L. (1998). Intergroup competition and attitudes toward immigrants and immigration: An instrumental model of group conflict. Journal of Social Issues, 54, 699 – 724. Fein, S., & Spencer, S. J. (1997). Prejudice as self-image maintenance: Affirming the self through derogating others. Journal of Personality and Social Psychology, 73, 31– 44. Fiske, S. T. (1998). Stereotypes, prejudice, and discrimination. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., pp. 357– 411). Boston: McGraw-Hill. Garza, R. T., & Herringer, L. G. (1987). Social identity: A multidimensional approach. Journal of Social Psychology, 127, 299 –308. Grant, P. R., & Brown, R. (1995). From ethnocentrism to collective protest: Responses to relative deprivation and threats to social identity. Social Psychology Quarterly, 58, 195–211. Green, D. P., Glaser, J., & Rich, A. (1998). From lynching to gay bashing: The elusive connection between economic conditions and hate crime. Journal of Personality and Social Psychology, 75, 82–92. Grofman, B. N., & Muller, E. N. (1973). The strange case of relative gratification and potential for political violence: The V-curve hypothesis. American Political Science Review, 57, 514 –539. Guimond, S., & Dambrun, M. (2002). When prosperity breeds intergroup hostility: The effects of relative deprivation and relative gratification on prejudice. Personality and Social Psychology Bulletin, 28, 900 –912. Guimond, S., Dambrun, M., Michinov, N., & Duarte, S. (2003). Does social dominance generate prejudice? Integrating individual and contex-
1044
DAMBRUN, TAYLOR, MCDONALD, CRUSH, AND ME´OT
tual determinants of intergroup cognitions. Journal of Personality and Social Psychology, 84, 697–721. Guimond, S., & Dube´-Simard, L. (1983). Relative deprivation theory and the Que´bec nationalist movement: The cognitive– emotion distinction and the personal– group deprivation issue. Journal of Personality and Social Psychology, 44, 526 –535. Guimond, S., & Palmer, D. L. (1993). Developmental changes in ingroup favouritism among bilingual and unilingual Francophone and Anglophone students. Journal of Language and Social Psychology, 12, 318 – 351. Gurr, T. R. (1970). Why men rebel. Princeton, NJ: Princeton University Press. Hepworth, J. T., & West, S. G. (1988). Lynchings and the economy: A time series reanalysis of Hovland and Sears (1940). Journal of Personality and Social Psychology, 55, 239 –247. Hovland, C., & Sears, R. R. (1940). Minor studies in aggression. VI: Correlation of lynching with economic indices. Journal of Psychology, 9, 301–310. Jackson, L. M., & Esses, V. M. (2000). Effects of perceived economic competition on people’s willingness to help empower immigrants. Group Process & Intergroup Relations, 3, 419 – 435. Jackson, S. E. (1981). Measurement of commitment to role identities. Journal of Personality and Social Psychology, 40, 138 –146. Kawakami, K., & Dion, K. L. (1995). Social identity and affect as determinants of collective action: Toward an integration of relative deprivation and social identity theories. Theory and Psychology, 5, 551–577. Leach, C. W., Snider, N., & Iyer, A. (2002). “Poisoning the consciences of the fortunate”: The experience of relative advantage and support for social equality. In I. Walker & H. J. Smith (Eds.), Relative deprivation: Specification, development, and integration (pp. 136 –163). Cambridge, England: Cambridge University Press. Levin, S., Federico, C. M., Sidanius, J., & Rabinowitz, J. L. (2002). Social dominance orientation and intergroup bias: The legitimation of favoritism for high-status groups. Personality and Social Psychology Bulletin, 28, 144 –157. Levine, R. A., & Campbell, D. T. (1972). Ethnocentrism: Theories of conflict, ethnic attitudes, and group behavior. New York: Wiley. Martin, J. (1981). Relative deprivation: A theory of distributive injustice for an era of shrinking resources. Research in Organizational Behavior, 3, 53–107. Mattes, R., Taylor, D. M., Poore, A., & Richmond, W. (1999). Still waiting for the barbarians: South African attitudes to immigrants & immigration. Southern African Migration Project (Migration Policy Series No. 14). Cape Town, South Africa: Idasa.
Mummendey, A., Kessler, T., Klink, A., & Mielke, R. (1999). Strategies to cope with negative social identity: Predictions by social identity and relative deprivation theory. Journal of Personality and Social Psychology, 76, 229 –245. Perreault, S., & Bourhis, R. Y. (1999). Ethnocentrism, social identification and discrimination. Personality and Social Psychology Bulletin, 25, 91–103. Pettigrew, T. F. (2002). Summing up: Relative deprivation as a key social psychological concept. In I. Walker & H. J. Smith (Eds.), Relative deprivation: Specification, development, and integration (pp. 351–373). Cambridge, England: Cambridge University Press. Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in western Europe. European Journal of Social Psychology, 25, 57–75. Runciman, W. G. (1966). Relative deprivation and social justice. London: Routledge and Kegan Paul. Sachdev, I., & Bourhis, R. (1987). Status differentials and intergroup behaviour. European Journal of Social Psychology, 17, 277–293. Sherif, M. (1966). Group conflict and cooperation: Their social psychology. London: Routledge and Kegan Paul. Sherif, M., Harvey, O. J., White, B. J., Hood, W. R., & Sherif, C. W. (1961). Intergroup conflict and cooperation: The robber’s cave experiment. Norman: University of Oklahoma. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York: Cambridge University Press. Smith, H. J., Spears, R., & Oyen, M. (1994). The influence of personal deprivation and salience of group membership on justice evaluations. Journal of Experimental Social Psychology, 30, 277–299. Stephan, W. G., Ybarra, O., & Bachman, G. (1999). Prejudice toward immigrants. Journal of Applied Social Psychology, 99, 2221–2237. Stouffer, S. A., Suchman, E. A., DeVinney, L. C., Starr, S. A., & Williams, R. M. (1949). The American soldier: Adjustment during army life (Vol. 1). Princeton, NJ: Princeton University Press. Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In S. Worchel & W. Austin (Eds.), The social psychology of intergroup relations (pp. 33– 48). Pacific Grove, CA: Brooks/Cole. Tripathi, R. C., & Strivastava, R. (1981). Relative deprivation and intergroup attitudes. European Journal of Social Psychology, 11, 313–318. Vanneman, R. D., & Pettigrew, T. F. (1972). Race and relative deprivation in the urban United States. Race, 13, 461– 486.
Received September 27, 2004 Revision received March 22, 2006 Accepted March 24, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1045–1065
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1045
Relationship Perceptions and Persistence: Do Fluctuations in Perceived Partner Commitment Undermine Dating Relationships? Ximena B. Arriaga, Jason T. Reed, Wind Goodfriend, and Christopher R. Agnew Purdue University The authors propose specific temporal profiles that reflect certainty versus doubt about where a partner stands with respect to a dating relationship over time. Two multiwave longitudinal studies focused on within-participant changes in perceived partner commitment. Results from multilevel modeling indicate that individuals whose perceptions of partner commitment fluctuate over time were more likely to be in a relationship that eventually ended than were individuals whose perceptions remained relatively steady. For individuals in recently initiated relationships, the association of fluctuation in perceived partner commitment with later breakup was significant regardless of the initial level of perceived partner commitment or the trend, and for all participants, it remained significant when initial level, trend over time, and fluctuation over time of other meaningful variables were controlled. Keywords: uncertainty, commitment, perceptions of partner, relationship stability, longitudinal
Juliet:
However, certainty about one’s own sentiments may not allay concerns about the partner’s sentiments; an individual may feel confident about the partner’s sentiments one week only to be surprised the following week by an unexpected partner behavior that is perceived to reflect less commitment. We suggest that individuals who vacillate in their perceptions of their partner’s level of commitment—perceptions of how inclined the partner is to remain in the relationship—are likely to experience continual doubts over where things stand; they are likely to wonder whether the partner wants the same level of commitment or instead wants more or less commitment. We examined this idea among individuals in relatively newly formed dating relationships (Study 1) and dating relationships of all durations (Study 2, a replication study). Partners in novel relationships may wonder about the uncertain current and future status of their relationship, given that it is precisely things not yet fully known that elicit a search for answers (Sanbonmatsu, Posavac, Vanous, & Ho, 2005). Thus, individuals in newly formed relationships may be particularly interested in, and affected by, their perceptions of the partner’s commitment so as to protect themselves against getting closer than the partner will allow. Several studies indirectly suggested that perceiving a partner as being committed should positively influence the relationship (Drigotas, Rusbult, & Verette, 1999; Holmes & Rempel, 1989; Miller & Rempel, 2004; Murray, Bellavia, Rose, & Griffin, 2003; Murray, Holmes, & Griffin, 2000; Wieselquist, Rusbult, Foster, & Agnew, 1999). If such positive perceptions were to change—if, for example, a person loses faith over time that the partner is committed—these studies implied that this would result in negative outcomes for the relationship. However, not all types of changes over time in perceived partner commitment are the same. We identify specific change markers or temporal profiles with specific implications for a relationship, suggesting that beyond mere increases or decreases over time, temporal patterns marked by fluctuation in perceived partner commitment directly reflect doubts and are particularly detrimental. We present the results of two
[. . .] O gentle Romeo: If thou dost love, pronounce it faithfully: [. . .]
Romeo: Lady, by yonder blessed moon I swear That tips with silver all these fruit-tree tops – Juliet:
O, swear not by the moon, the inconstant moon, That monthly changes in her circled orb, Lest that thy love prove likewise variable. —William Shakespeare, Romeo and Juliet
In this famous balcony scene from Romeo and Juliet, Romeo first professes his love to Juliet. Juliet wants to know that Romeo loves her not just now, but always; that is, she seeks assurance that his love is constant over time. Just like Juliet, people in lasting romantic relationships seek assurance that their own relationships are based on ever present, or temporally stable, factors that keep the partners together over time (Kelley, 1983). More so than in marital relationships, in dating relationships such factors are likely to rely heavily on feelings and perceptions and perhaps less on tangible investments or structural barriers to leaving (Le & Agnew, 2003). Past research has shown that a dating relationship is disrupted when a person’s own feelings vacillate over time and is more likely to persist when a person has relatively stable feelings over time, even when these stable feelings are not particularly positive toward the relationship (Arriaga, 2001). A person with relatively unchanging sentiments is afforded a sense of certainty about where he or she stands with respect to the relationship.
Ximena B. Arriaga, Jason T. Reed, Wind Goodfriend, and Christopher R. Agnew, Department of Psychological Sciences, Purdue University. We thank Niall Bolger for his thoughtful comments regarding this research. We also thank the College of Liberal Arts at Purdue University for supporting this research through an incentive grant. Correspondence concerning this article should be addressed to Ximena B. Arriaga, Department of Psychological Sciences, Purdue University, 703 Third Street, West Lafayette, IN 47907-2081. E-mail:
[email protected] 1045
1046
ARRIAGA, REED, GOODFRIEND, AND AGNEW
longitudinal studies of individuals in dating relationships to examine whether those who vacillate over time in their perceptions of their partner’s commitment are at greater risk for experiencing relationship dissolution.
Relationship Confidence Versus Doubt The Role of Perceived Partner Commitment A growing body of research has suggested that people are motivated to form a strong sense of conviction about their romantic relationships—a strong sense that “the partner really is the ‘right’ person and can be counted on to be caring and responsive across time and situations” (Murray, 1999, p. 23; see also Holmes, 2004). This “quest for conviction” account stipulates that conviction is necessary before allowing oneself to feel close to a partner, lest one become anguished over being tied to a partner with undesirable qualities or hurt by a partner’s uncaring acts (Murray, 2005). Unfortunately, disappointment is inevitable—no partner is always perfect or able to act in ways that promote the relationship. The quest for conviction account suggests that the only way for relationships to survive such disappointing moments is to adopt a sense of closure, namely to remove all doubts about the relationship and embrace the idea that even when there are setbacks or disappointments, these are minor and the relationship remains on course toward a healthy future (i.e., making a “leap of faith,” Rempel, Holmes, & Zanna, 1985; see also Holmes, 2004). In contrast to individuals who follow a quest for conviction, individuals marred by doubts must continually reevaluate where the relationship stands and whether it has a viable future. Doubts about a relationship may arise from any number of sources, such as from (a) feeling uncertain about one’s own feelings toward the relationship or partner (Arriaga, 2001; Murray & Holmes, 1999), (b) feeling that the partner is not concerned with the relationship (Murray et al., 2000; Holmes & Rempel, 1989), or (c) doubtinducing influences beyond the dyad, such as persistent perceived disapproval from others (Agnew, Loving, & Drigotas, 2001; Etcheverry & Agnew, 2004; Lehmiller & Agnew, 2006). In the current article, we focused on the second source of doubt, namely doubts about a partner’s commitment. We suggest that a general inference about a partner’s commitment may be more strongly associated with relationship persistence than perceptions of a partner’s behaviors or even inferences about those behaviors. Why should perceived partner commitment have such a strong effect? There are several reasons to expect that general inferences such as perceptions of partner commitment are particularly important for relationship well-being, more so than inferences about a partner’s behaviors, his or her level of satisfaction, quality of alternatives, investments, or other specific partner inferences. First, research has confirmed that inferences made about a partner’s behavior are just as or even more important than the behaviors themselves (Bradbury & Fincham, 1990). Second, general inferences about how concerned a partner is with the relationship— his or her relationship motives— have profound effects on how rewarding a relationship is experienced to be (Kelley, 1979). Third, general partner inferences form the bases of broad attitudes or orientations toward a relationship (i.e., macromotives, Holmes, 1981) that temper the perceptions of specific partner actions or
even interpretations of specific actions; perceived partner commitment is precisely such a broadly framed inference (Wieselquist et al., 1999; see also the discussion of trust by Holmes & Rempel, 1989). Fourth, predicting a broad outcome (relationship stability) is likely to require a broad-based inference rather than a more specific inference (Ajzen & Fishbein, 1977). Satisfaction, alternatives, and investments all are specific causes of commitment; as such, perceptions of these variables are less likely to predict stability directly than are perceptions of commitment (Rusbult, Olsen, Davis, & Hannon, 2001). There is ample extant research suggesting that perceptions of partner motives have implications for relationship well-being. Perceiving that a partner acts in ways that support the relationship (i.e., perceiving partner accommodation and partner willingness to forego his or her own self-interest for the sake of the relationship) is concurrently associated with trusting the partner (Wieselquist et al., 1999). Conversely, individuals who have doubts about whether their partner will be responsive to their needs make negative inferences about their partner’s behavior and harbor further doubts (Miller & Rempel, 2004). Moreover, individuals who assume their partner sees them in more negative ways than the partner actually sees them (i.e., perceived low regard) report subsequent increases in their own levels of conflict and ambivalence and have partners who report subsequent decreases in satisfaction and trust (Murray et al., 2000). It is important to note that the links between low partner regard and declines in relationship well-being remain significant even after controlling for a partner’s actual level of regard, underscoring the importance of perceptions of the partner (see also Wieselquist et al., 1999). It stands to reason that doubts about a partner’s commitment also have negative repercussions for the relationship. Although these studies suggested positive relationship implications when a partner is perceived to be caring, supportive, and concerned with maintaining the relationship, there are several caveats that limit the relevance of these studies to the current research. First, none of these studies provided direct evidence that perceptions of the partner predict the ultimate indicator of wellbeing, namely whether a relationship lasts. Second, these studies examined variables that only indirectly tap subjective perceptions of partner commitment. For example, Wieselquist et al. (1999) examined reports that the partner had engaged in prorelationship behaviors; Miller and Rempel (2004) examined one’s level of trust. Drigotas et al. (1999) have come closest to examining absolute levels of perceived partner commitment, namely perceived matches versus mismatches in own and partner commitment. We sought to directly examine perceptions of partner commitment per se given our broader focus on doubt and the strong link between subjective commitment and resolving doubts (Brickman, 1987). Most notably, there has been no research that identifies specific patterns in partner perceptions over extended periods of time, extended patterns that would meaningfully reflect sustained certainty versus repeated doubts (Kelley, 1979). Two individuals may exhibit similar levels of perceived partner commitment at two time periods separated by 6 months, but 1 may have vacillated in his or her perceptions of the partner’s commitment whereas the other may have had perceptions that followed a steady path. As we describe below, these two patterns are distinct in ways that are theoretically meaningful and are likely to have different outcomes.
Moreover, when examining perceptions, it becomes critically important to capture prospective changes over time; data that capture prospective changes in perceptions versus retrospective accounts of changes in perceptions differ descriptively and in their predictive value (Karney & Frye, 2002). In short, much can be gained theoretically and in predictive value by capturing changes that unfold over time (Arriaga, 2001; Karney & Bradbury, 1995).
Doubt as Temporal Fluctuations in Perceived Partner Commitment
Perceived Partner Commitment
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
1047
Profile 1 Profile 2 Profile 3 Profile 4
Time
How do people who do not treat their partner’s commitment as an open question look different over time from those who do (Holmes & Rempel, 1989)? Does experiencing sustained certainty versus repeated doubts make a difference in whether a dating relationship lasts? To answer these questions, we advance an analysis of specific within-person temporal patterns, or profiles, in perceived partner commitment. As elaborated in Shoda’s notion of a personality signature (Shoda, 1999; Shoda, Mischel, & Wright, 1994), we suggest that distinct patterns of intraindividual variability (in our analysis, intraindividual changes over time) have coherence and are theoretically meaningful (see also Murray, Holmes, & Collins, 2006). Our major thesis is that measuring each individual’s idiosyncratic pattern of fluctuation in perceived partner commitment is critical (Arriaga, 2001; Campbell, Simpson, Boldry, & Kashy, 2005), by which a person’s ongoing state of relationship doubt is reflected in perceptions that vacillate markedly and sustained certainty is reflected in perceptions that stick to a linear path. This assertion requires identifying temporal patterns with more precision than is afforded by examining only general changes over time, which has presented its own challenges but increasingly is captured by a multiwave trend in a variable (i.e., its slope over time, preferable to averaging between two measurement occasions; Karney & Bradbury, 1995). We identify four specific temporal patterns in perceived partner commitment that capture theoretically relevant experiences of certainty versus doubt. We are not suggesting that every person exactly fits one of these four profiles or that data naturally cluster into these four patterns. Rather, these are four theoretically meaningful profiles, and it is possible to set cutoffs on continuous data so as to approximate each of these four profiles. These four patterns are a combination of whether the general trend is one of increase or decrease over time and whether the pattern fluctuates along or steadily follows a linear path.1 Figure 1 illustrates a hypothetical example of each profile. One temporal profile is reflected in steadily increasing levels of perceived partner commitment (e.g., Profile 1 in Figure 1). These individuals see their partner as predictably committed and capture what theoretically has been described as “exaggerating the case for commitment” (Murray, 1999). Even if the partner acts destructively, the perception of a single destructive act becomes folded into a broader partner perception that is positive and optimistic. Thus, these are individuals who have likely developed strong defenses against potentially negative partner information, defenses that preempt perceiving declines in a partner’s commitment. By deepening their positive view of the partner’s commitment over time, and even enhancing their view in the face of unresolved problems (see Holmes & Rempel, 1989, high trust individuals),
Figure 1. Temporal profiles in individuals’ perceptions of their partner’s commitment level, measured repeatedly over time.
these individuals become immune to major relationship threats and thus are likely to have relatively longer lasting relationships. A second temporal profile is characterized by individuals who generally perceive their partner to be committed, or even exhibit a trend of increasing perceptions of partner commitment, but whose perceptions fluctuate along that course (e.g., Profile 2). Their relationship experience is likely to be different, defined at one moment by optimism and the next moment by relative disappointment. These are individuals who capture what theoretically has been described as having only moderate trust in the partner (Holmes & Rempel, 1989); they acknowledge moments when the partner behaves positively but hesitate in generalizing the positive moment to future moments in the relationship. Their reactivity to specific instances becomes apparent in a temporal profile marked by fluctuation. The third profile concerns those who have steadily low or decreasing perceptions of their partner’s commitment, suggesting a subjective sense of certainty that the partner is decreasingly committed (e.g., Profile 3). They expect little of their partner, yet the steady decline affords some level of certainty and predictability over where their partner stands (Sorrentino, Holmes, Hanna, & Sharp, 1995). They capture what theoretically has been described as individuals who sustain their relationships with routines that may not be satisfying or involve closeness but nonetheless keep partners intertwined (Berscheid, 1983; see Holmes & Rempel, 1989, low trust individuals). As they avoid having to adjust their expectations time and again, remaining in their relationship be1 In addition to these two indicators (whether perceived partner commitment follows a trend of increase or decrease and whether it fluctuates), recent applications of growth curve analysis would suggest also looking at initial level of perceived partner commitment (the intercept). We were less concerned with strictly adhering to growth curve variables (a methodological focus) and more concerned with identifying patterns that have clear theoretical relevance to the experience of doubt. Changes in a variable have been shown to be more theoretically meaningful than the absolute level of that variable (e.g., initial level; Karney & Bradbury, 1995), and our emphasis is on examining increases and decreases that vary in degree of fluctuation, as fluctuation is theoretically relevant to the experience of doubt. We later present regression analyses that take into account level (the intercept), in addition to increase versus decrease (slope) and fluctuation; absolute level did not predict relationship persistence beyond the effect of the linear trend over time or fluctuations over time.
1048
ARRIAGA, REED, GOODFRIEND, AND AGNEW
comes a “path of least resistance” albeit not necessarily a positive path. A fourth temporal profile is reflected in individuals whose perceptions of a decreasingly committed partner fluctuate over time (e.g., Profile 4). In contrast to individuals with steady downward perceptions (Profile 3), these individuals are unable to form stable expectations of what the future holds given their vacillating perceptions. Struggling to know where the partner stands, they remain watchful of their partner’s actions for possible changes thus readjusting their inferences time and again.
Predicting Breakup Status The general trend of change, or slope, in perceived partner commitment is likely to be important in predicting breakup status, in light of seminal research showing that the linear trend in own level of commitment (Rusbult, 1983), the trend in own satisfaction (Karney & Frye, 2002), and the trend in attributions about a partner (Karney & Bradbury, 2000) each are associated with potential or actual relationship outcomes. Similarly, perceptions of partner commitment that decrease over time (e.g., Profiles 3 and 4) are more likely to result in breakup than those that increase (e.g., Profiles 1 and 2). To establish temporal profiles or signature patterns that reflect doubt versus certainty in particular, researchers must move beyond examining increases or decreases only so as to also look at whether the temporal path fluctuates versus steadily follows a straight line (Surra, Hughes, & Jacquet, 1999). Arriaga (2001) found that fluctuations in one’s own level of satisfaction—that is, uncertainty in one’s own feelings—predicted later dissolution, even controlling for overall increases or decreases over time in one’s level of satisfaction. However, one’s own feelings are but one source of doubt and do not address an important additional source of doubt, whether one thinks the partner is committed. We anticipated that the extent of fluctuation over time in perceived partner commitment would affect the odds of breakup such that individuals fitting Profile 2 would be more likely to be in relationships that end relative to those fitting Profile 1, and the same holds for Profile 4 individuals relative to Profile 3 individuals. In short, fluctuation in perceived partner commitment should have a main effect on relationship dissolution independent of the trend of increase or decrease in these perceptions or the overall level of these perceptions (Hypothesis 1). Why should this be so? The lack of partner predictability embedded in fluctuating perceptions means that these individuals are repeatedly adjusting their expectations. Research has shown that individuals plagued by uncertainty (as reflected in fluctuating perceptions of partner commitment) pay special attention to information that violates their expectations (Driscoll, Hamilton, & Sorrentino, 1991) and thus adjust their impressions time and again, making them highly reactive (or sensitive) to each partner interaction (Surra & Hughes, 1997). Relative to low reactivity, such high reactivity is associated with greater relationship distress, less satisfaction, and less closeness (Campbell et al., 2005; Jacobson, Follette, & McDonald, 1982). Our logic suggests that certainty over perceived commitment stands to have an effect that is independent of the general trend in perceived partner commitment (Sorrentino et al., 1995). We are not suggesting that steadily declining perceptions of partner com-
mitment (third profile) are good for a relationship; rather, declining perceptions that fluctuate (fourth profile) are particularly bad for a relationship, worse than steadily declining perceptions (third profile). They share a pattern of decline in perceptions but differ in level of fluctuation in perceptions and thus should differ in odds of breakup; certainty that a partner lacks commitment should predict more persistence than doubt over a partner’s lack of commitment because doubts encourage constant monitoring and redefinition of the status of the relationship whereas certainty retains the status quo. An alternate model would suggest that only the first profile— steadily increasing perceived partner commitment (i.e., a “felt security” profile, see Murray et al., 2006)—is uniquely associated with persistence whereas the other profiles are associated with dissolution. That is, the effect of steady versus fluctuating perceptions may matter more when the perceptions are positive (i.e., Profile 1, but not Profile 2, predicts persistence), and less or not at all when they are negative (i.e., Profiles 3 and 4 predict breakup equally); we explored the possibility that the trend in perceptions would moderate the effect of fluctuation but anticipated that fluctuation would exhibit an independent (i.e., main) effect on relationship dissolution. In keeping with the analysis of the importance of perceived partner commitment relative to other inferences about the partner (e.g., behavioral inferences), we also predicted that fluctuations in perceived partner commitment would be more strongly associated with relationship dissolution than would fluctuations in perceptions of a partner’s behaviors or fluctuations in inferences about partner behaviors (Hypothesis 2).
Possible Correlates and Constraints of Fluctuation in Perceived Partner Commitment Not all variables that fluctuate over time necessarily reflect doubt. There are theoretical reasons to expect some variables to correlate with fluctuation in perceived partner commitment more than others and to expect some variables to constrain or moderate the association of fluctuation in perceived partner commitment with later relationship disruption. We explored possible concomitant, antecedent, and limiting conditions of fluctuation in perceived partner commitment.
Own Level of Satisfaction Because fluctuations in own satisfaction (Arriaga, 2001) and in perceived partner commitment both reflect doubt, they are likely to be positively correlated. Yet, as theoretically distinct and unique sources of doubt stemming from own feelings versus perceptions of the partner, they should each provide independent prediction when examined simultaneously. We predicted that, when examined simultaneously, both should exhibit independent associations with later relationship dissolution (Hypothesis 3). We did not examine perceived partner satisfaction, given that perceiving partner changes in overall commitment is likely to have more pervasive effects than perceiving changes in satisfaction only (Rusbult, 1983). Rather than perceived satisfaction, perceiving committed partner acts more closely captures confidence in a partner’s intentions (Wieselquist et al., 1999).
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
Own Level of Commitment We examined several possible types of association between own level of commitment and perceived level of partner commitment. First, we anticipated that when examined simultaneously, fluctuation in perceived partner commitment and fluctuation in own commitment both should exhibit independent associations (main effects) with later relationship dissolution (Hypothesis 4). Both reflect doubt, and as such, both should influence whether a relationship lasts. Second, we explored whether own commitment might be an antecedent of perceived partner commitment. It is possible that a relatively uncommitted person is inclined to assume a partner is equally uncommitted—that is, one’s own level of commitment may color inferences about a partner’s commitment. Yet, it is also possible that people who perceive a partner to be uncommitted will adjust their own commitment so as to not be disappointed in the future (e.g., “My partner doesn’t want this relationship, so I should not put much into it myself”)—that is, inferences about a partner’s commitment may cause (and thus precede) adjustments in own commitment level. We explored whether (a) initial level of perceived partner commitment correlated with subsequent increases in own commitment and (b) initial level of own commitment correlated with subsequent increases in perceived partner commitment. Third, we explored whether one’s own commitment level at the outset of the study (initial commitment) constrains the extent of doubt one experiences. It is possible that people who initially are relatively uncommitted may not care very much whether the relationship continues, and so they may not heed their perceptions of the partner’s commitment. On the other hand, people who very much want the relationship to continue—those who initially are highly committed—may seek confirmation of the partner’s commitment; they may be more affected by evidence of a stably committed partner versus a partner plagued by doubts. That is, initial commitment level may moderate whether fluctuation in perceived partner commitment predicts relationship dissolution. It is equally possible that uncertainty about a partner’s commitment undermines a relationship regardless of one’s own level of commitment given the pervasive and arguably independent effects of doubt (Holmes, 2004), suggesting no such moderation.
1049
Overview of Studies We have suggested that over time, fluctuating perceptions of partner commitment disrupt a relationship and foretell its ending, a straightforward proposition with rather complex methodological implications. To adequately capture specific temporal patterns in perceptions of the partner, we conducted two longitudinal studies with multiple measurement occasions relatively close in time. For each predictor variable, we estimated each individual’s initial level at the start of the study, the trend of increasing or decreasing perceptions over time, and the extent of variation or fluctuation over time. Variation in a variable has been used in previous research to capture theoretically meaningful patterns (cf. Aronson & Inzlicht, 2004; Arriaga, 2001; Campbell et al., 2005; Crocker, Karpinski, Quinn, & Chase, 2003). Study 1 provided a test of all hypotheses, focusing exclusively on individuals in relatively newly formed relationships (no more than 6 months in duration). Participants completed eight measurement occasions on a weekly basis and a follow-up session 2 months after Time 8 to assess breakup. Because this study was time intensive (it involved weekly sessions), many of the variables were measured with single items, as has been done in other time-intensive studies (e.g., Bolger, Zuckerman, & Kessler, 2000) or studies demonstrating adequate validity of single-item measures (e.g., Robins, Hendin, & Trzesniewski, 2001, with respect to self-esteem; Agnew, Van Lange, Rusbult, & Langston, 1998, as well as Aron, Aron, & Smollan, 1992, with respect to self–partner cognitive overlap). Study 2 served to validate the single-item measures used in Study 1 and assess whether findings generalize to (a) people in relationships of varying durations and (b) profiles based on multiple occasions separated by greater time spans (i.e., nine measurement occasions separated by 4 weeks each rather than by 1 week).
Method Study 1 was conducted with undergraduate students who volunteered to participate as partial fulfillment of a course requirement at a large Midwestern university. Study 2 relied on extant data collected as part of a major study on substance use in a sample of entering freshman students at a large Midwestern university.
Own Attachment Style The temporal profiles described above may have origins in individual differences that are relevant to close relationships, such as attachment style (see Simpson & Rholes, 1998, for an overview of attachment processes). More so than individuals described as secure or avoidant, those who are anxious gauge their perceptions of the partner on the basis of their daily partner interactions and thus may exhibit more volatility in inferring partner motives (Campbell et al., 2005). We anticipated that fluctuation in perceived partner commitment would be positively correlated with anxious attachment tendencies (Hypothesis 5) but not necessarily with avoidant tendencies. We examined the correlations of fluctuation in perceived partner commitment specifically with the absolute level of each insecure attachment variable, rather than changes in insecure attachment variables, given that attachment style is considered to be stable over relatively short periods of time (Fraley, 2002).
Participants In Study 1, of the 130 individuals who attended the Time 1 session, 25 participants were eliminated because they did not meet the criteria for inclusion in this sample. Through a series of checks and assessments, it became clear that 4 participants made up their responses, 4 were in relationships longer than 6 months in duration at the start of the study (i.e., they ignored inclusion criteria provided during recruitment), 2 were dating another participant in the study (they were randomly chosen over their partner to ensure independence of observations), and 15 participants reported at Time 4 that their relationship had ended and thus did not provide sufficient observations to establish a reliable longitudinal pattern. In the absence of guidelines on how many observations are necessary to establish whether a pattern of fluctuation occurs, it seemed reasonable to assume that three data points would not be sufficient (at most, if they fluctuate, one would capture a curvilinear pattern). Participants whose relationships ended after Time 4 but before the follow-up session (n ⫽ 7), or ended by the follow-up (n ⫽ 21), were coded as a breakup.
1050
ARRIAGA, REED, GOODFRIEND, AND AGNEW
Of the original 130 participants, 23 (18%) were lost because of attrition: 9 participants voluntarily dropped the study after Time 1 whereas 14 could not be reached for the follow-up session and thus had missing information regarding their relationship status (persisted vs. ended). These 23 participants did not differ significantly in their Time 1 level of perceived partner commitment from those who were retained in the Study 1 sample, F(1, 104) ⫽ 1.96, ns. The final Study 1 sample of 82 individuals consisted of 44 women and 38 men in relatively newly formed dating relationships (6 months or less), who had usable data from at least four consecutive measurement occasions and who subsequently provided data on their breakup status. Study 2 examined individuals in relationships of all durations. The larger study from which the Study 2 sample was derived involved 912 incoming freshman students who were recruited to provide weekly data on various topics over the course of their freshman year. Initial recruitment took place during the summer prior to starting their freshmen year, when students visited campus as part of an orientation event. Study inclusion criteria included some past experience with cigarette smoking (i.e., at least one puff) prior to the study. Participants were paid for completing each weekly survey, resulting in a subject retention rate of more than 90% over the 35 consecutive weeks of the freshmen year. Every 4 weeks there were questions assessing whether they were in a romantic relationship and tapping various characteristics of their relationships; we examined the sample of individuals who reported being in a relationship (n ⫽ 630) at some point over the course of the study. Of these, 38 (6%) were lost because of attrition (i.e., they left the university or voluntarily stopped participating in the study before it ended). Given that these were freshman students, with many involved in fluid, short-term relationships, many did not meet the criteria for inclusion in this sample: They did not have a relationship that remained intact for 4 consecutive times and thus did not provide sufficient observations to establish a reliable longitudinal pattern (n ⫽ 339). The final Study 2 sample of 253 individuals consisted of 111 women (44%) and 142 men (56%) who had usable data from at least four measurement occasions. Unlike Study 1, there was not a follow-up session to assess breakup; instead, participants who provided data about an intact relationship for 4 consecutive times and whose relationships subsequently ended over the course of the study were included in the sample and coded as a breakup. Study 1 and Study 2 participants were similar, except for the duration of their relationships. Study 1 participants’ relationships were, on average, 3 months in duration at Time 1; Study 2 participants’ relationships were, on average, 16 months in duration when they became included in the current sample. At Time 1, Study 1 participants were 19 years old on average (SD ⫽ 1.36); Study 2 participants were almost exclusively 18 years old given that they were college freshmen. The majority were White (in Study 1, 88% White, 7% Asian American, 4% Latino, and 1% African American; in Study 2, 87% White, 5% Asian American, 2% Latino, 1% African American, and 3% Other). Sixty-six percent (n ⫽ 54) of Study 1 participants continued to be in their relationships at follow-up whereas 34% (n ⫽ 28) were no longer dating their Time 1 partners. Seventy-five percent (n ⫽ 189) of Study 2 participants continued in their relationships over the course of the study whereas 25% (n ⫽ 64) were in relationships that ended. Given that Study 1 participants were exclusively in relatively newly formed relationships, it is not surprising that a higher percentage of these relationships ended.
Procedure Study 1 data were collected over the course of two semesters; the results did not differ between the two semesters (i.e., there were no main or interaction effects for semester designation). Data collection sessions for Time 1 through Time 8 were conducted on a weekly basis in a small classroom; approximately 10 –20 participants took part in each session. At Time 1, the experimenter described the study tasks and obtained written
consent from participants. At each time period thereafter, the experimenter reviewed the activities for the day’s session, assured participants that their responses would remain confidential, and distributed questionnaires. Each session lasted approximately 15 minutes. At Time 8, participants completed an additional one-page questionnaire that probed for dishonest responding. The instructions reiterated that they would receive full credit regardless of their responses to these final probes. At Time 8, participants were also debriefed and thanked for their assistance. Follow-up questions were administered by telephone roughly 2 months after Time 8. For Study 2, participants logged onto the study website (with a preassigned username and a personally selected password) each week and were presented with a set of survey questions. The initial weekly survey was administered at the beginning of the fall semester and the final weekly survey was administered during the final week of the spring semester. Surveys were available every week, including winter and spring breaks, for a total of 35 consecutive weeks. Participants were paid for participation each week, which varied from week to week, but over 87% of participants completed all surveys. A set of questions was presented each week to participants focusing primarily on substance use over the preceding week (e.g., cigarette use). In addition, participants responded to a differing subset of questions each week about other aspects of their lives (relationships, stress, sleep habits, etc.). These additional questions were rotated on a 4-week schedule; thus, relationship variables were administered every 4 weeks for a total of nine times over the course of the 35-week study.
Measures Included in Both Studies Perceptions of the partner’s commitment was measured by a single item in Study 1. Participants were asked, “How committed is your partner to the relationship?” followed by a 9-point response scale ranging from 0 (not at all committed) to 8 (very committed). In Study 2, this variable was measured with four items (␣ at initial time ⫽ .93), adapted from the Investment Model Scale (Rusbult, Martz, & Agnew, 1998): “My partner is committed to maintaining our relationship,” “My partner intends to stay in this relationship,” “My partner feels very attached to our relationship – very strongly linked to me,” “My partner is oriented toward the long-term future of our relationship (for example, imagines being with me several years from now)”; participants indicated their level of agreement with each item on a 9-point response scale ranging from 0 (do not agree at all) to 8 (agree completely). To validate the single item used to tap perceived partner commitment in Study 1, we derived two variables, one based on a single Study 2 item (“My partner is committed to maintaining our relationship”) that is similar to the single Study 1 item and a second one based on the four Study 2 items averaged together. We then correlated these two variables at each time. The average correlation across all times was .95 (range ⫽ .91–.98). Study 2 analyses of perceived partner commitment included the variable based on an average of the four items. Own level of satisfaction was measured by a single item in Study 1—“I feel satisfied with our relationship at the moment”—followed by a 9-point response scale ranging from 0 (do not agree at all) to 8 (agree completely). In Study 2, this variable was measured with three items adapted from the Investment Model Scale (␣ at initial time ⫽ .89): “I feel satisfied with our relationship,” “My relationship is better than others’ relationships,” and “Our relationship makes me very happy”; participants indicated their level of agreement with each item on a 9-point response scale ranging from 0 (do not agree at all) to 8 (agree completely). To validate the single item used to tap own satisfaction level in Study 1, at each time we correlated a variable based on a single Study 2 item (“I feel satisfied with our relationship”) that was almost identical to the single Study 1 item with a variable based on the three Study 2 items averaged together. The average correlation across all times was .95 (range ⫽.91– .97). Study 2 analyses of own satisfaction level included the variable based on an average of the three items.
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT Own level of commitment was measured in Study 1 by a single item from the Investment Model Scale—“I am committed to maintaining my relationship with my partner”—and in Study 2 by four items from this scale (␣ at initial time ⫽ .93): “I am committed to maintaining my relationship with my partner,” “I intend to stay in this relationship,” “I feel very attached to our relationship – very strongly linked to my partner,” and “I am oriented toward the long-term future of my relationship (for example, I imagine being with my partner several years from now).” In all cases, participants indicated their level of agreement with each item on a 9-point response scale ranging from 0 (do not agree at all) to 8 (agree completely). To validate the single item used to tap own commitment level in Study 1, at each time period we correlated a variable based on the Study 2 item that was identical to the single Study 1 item, with a variable based on the four Study 2 items averaged together. The average correlation across all times was .95 (range ⫽ .92–.98). Study 2 analyses of own commitment level included the variable based on an average of the four items. Breakup status was measured in Study 1 with the following question: “Are you still dating the same person that you were when you last participated in this study?” We derived a two-level breakup status variable consisting of the group of participants whose relationships persisted (n ⫽ 54) versus the group whose relationships ended (n ⫽ 28). Similarly in Study 2, at each time when participants were asked about their relationship, they were asked whether they had a romantic partner. If they indicated that they did have a romantic partner, they were asked to provide the first name and first letter of the last name of that person. Subsequently, they were given that person’s name and asked if they were still involved in a romantic relationship with that person. The persisted group was based on participants who indicated they had the same partner throughout the study (n ⫽ 189), and the ended group was based on participants who indicated they no longer had the same partner from the previous time (n ⫽ 64).
Measures Included Only in Study 1 Perceptions of positive partner behavior was measured by using two items (“Based on what has occurred in the past week, how much did your partner do nice things that matter to you?” “How representative are these nice behaviors of the type of person your partner is?”), followed by a 9-point response scale ranging from 0 (very little/not at all representative) to 8 (very much/very representative). These two items were averaged (␣ at initial time ⫽ .86). Perceptions of negative partner behavior was measured by using two items asking participants “Based on what has occurred in the past week, how much did your partner do negative things that matter to you?” and “How representative are these negative behaviors of the type of person your partner is?”, followed by a 9-point response scale ranging from 0 (very little/not at all representative) to 8 (very much/very representative). These two items were averaged (␣ at initial time ⫽ .74). Attachment style was measured for other purposes beyond the current research that demanded using a multiitem scale. We used the 17-item Adult Attachment Questionnaire (Simpson, Rholes, & Phillips, 1996), which consists of a series of statements and a 9-point response scale ranging from 0 (do not agree at all) to 8 (agree completely). Nine items captured an anxious–ambivalent attachment dimension (e.g., “I often worry that my partner(s) don’t really love me”; ␣ at initial time ⫽ .79); these items were averaged such that higher numbers reflect anxious–ambivalent tendencies. Eight items captured an avoidant attachment dimension (e.g., “I don’t like people getting too close to me”; ␣ at initial time ⫽ .77); these items were averaged such that higher numbers reflect avoidant tendencies. For purposes of validating the measure of perceived partner commitment, there was an item in Study 1 only that tapped confidence in a partner’s dependability. Participants were asked, “How confident are you that you can depend on your partner?” followed by a 9-point response scale ranging from 0 (not at all confident) to 8 (very confident). At Time 1,
1051
perceived partner commitment and confidence in a partner’s dependability were highly significantly correlated, r(82) ⫽ .79, p ⬍ .001, providing evidence of convergent validity. Furthermore, at Time 1, correlations of perceived partner commitment with other relationship variables were also significant but lower in magnitude: own level of commitment, r(82) ⫽ .34, p ⫽ .002; own level of satisfaction, r(82) ⫽ .62, p ⬍ .001; and the correlations with individual difference variables were even lower in magnitude, thus providing evidence of divergent validity: anxious–ambivalent attachment style, r(82) ⫽ ⫺.27, p ⫽ .014; avoidant attachment style, r(82) ⫽ ⫺.12, ns.
Results Strategy for Analyzing Change Over Time Analysis of the data proceeded in two stages. First, we used SAS software’s PROC MIXED (SAS Institute, 1992; Singer, 1998) and additional SAS data steps to derive the three within-person change estimates (i.e., initial level, linear trend over time, fluctuation over time) for each perception variable. Second, we used these estimates to predict breakup status in subsequent regression models. We adopted this two-stage strategy to address several challenges described below. One challenge concerned the hierarchical (nested) nature of the data; repeated ratings over time were nested within a participant. That is, each observation corresponded to a particular time for a particular participant, so there were multiple observations for each participant (one for each time). Because observations were clustered by participant, they were likely to violate the assumption of being independent (Gable & Reis, 1999). By adopting a multilevel model approach, PROC MIXED accounted for nesting of observations within participant by treating the intercept and slope as random rather than fixed variables.2 Were repeated ratings clustered (nested) within a person? The intraclass correlation () provides an indicator of clustering; it reflects the proportion of total variance explained by betweenperson variation versus within-person variation (i.e., betweenperson variance divided by the sum of between-person variance and within-person variance; Singer, 1998). In Study 1, the extremely high value ( ⫽ .97) indicated little variance within participants relative to variance between participants (i.e., a particular participant’s responses were homogenous), suggesting a multilevel model approach over standard (ordinary least squares) regression. In Study 2, the intraclass correlation ( ⫽ .24) indicated less within-participant clustering than in Study 1 but a fair amount of clustering nonetheless that could inflate the Type 1 error rate (Kashy & Kenny, 2000). A second challenge was more difficult to overcome and concerned how to examine whether fluctuation over time in one or 2 By treating the intercept and slope as random variables, we can assume we sampled ratings from a participant rather than obtained all possible ratings for each participant (that is, we did not obtain the population of ratings for each participant). Just as samples vary in the number of observations included, participants varied in the number of observations included. It was not essential that all participants have the same number of observations (i.e., the same number of measurement occasions), and participants who missed an occasion could still be included in the analysis. As in hierarchical linear and nonlinear modeling (see Karney & Bradbury, 1997), PROC MIXED adjusts participant-specific estimates for their precision.
1052
ARRIAGA, REED, GOODFRIEND, AND AGNEW
more variables predict later breakup status—for example, whether vacillating versus generally stable levels of (a) perceived partner commitment and (b) own commitment each predicted later breakup when examined simultaneously. We first describe our approach and then explain why we adopted it. Our strategy was to derive growth curve estimates for each participant (cf. Karney & Bradbury, 1995). Growth curve analysis can be thought of as calculating a regression model for each participant, where his or her repeated perception ratings are regressed onto time. Specifically, we used PROC MIXED to test the following (Level 1) model, where the time variable was recoded so that the first time period had a value of zero: variable measured at all times ⫽ intercept ⫹ slope(time)⫹ residual This provided output to construct a new data set with several observations for each participant (one observation at each time). The new data set included each participant’s actual rating at each time and the predicted rating based on the estimated linear slope in his or her ratings from one time to the next.3 For each participant, we derived three change estimates of each predictor variable. One estimate was a participant’s initial perception at the start of the study, on the basis of the estimated intercept (i.e., given the way time was coded, each participant’s predicted level when time equaled zero). A second estimate was a participant’s trend in a given variable over time on the basis of the linear slope in ratings (i.e., the predicted rating at one time minus the predicted rating at the previous time), which reflected whether a participant’s ratings on a variable generally followed a general pattern of increase versus decrease over time. A third variable was the amount of fluctuation over time in a variable for a given participant, on the basis of variation in the participant’s slope. We used the deviation between an actual (observed) score at a given time and the predicted score (based on the Level 1 model) to calculate the standard deviation around the slope line in each participant’s ratings.4 For each predictor, we derived these three variables reflecting a given participant’s change estimates— his or her initial level of a given predictor, the linear trend over time, and fluctuation over time. We used these participant-specific estimates in subsequent standard regression analyses predicting breakup status (persisted vs. ended). We repeated all of the analyses by using logistic regression; this did not alter the patterns of significance. Rather than derive change variables in one step (based on the Level 1 model) and analyze them in a second step (the Level 2 regression model), an alternate strategy might have been an “allin-one” model, in which a single step is used to compare breakup groups in their intercepts, slopes, and of noted importance, fluctuation. Commonly used multilevel model programs, such as HLM and SAS software’s PROC MIXED, typically take this approach to analyze intercepts and slopes as random variables. However, several of our hypotheses required two critical tasks: (a) examining fluctuation in a variable (based on the within-person residual, rij), and (b) assessing the association of breakup status with several Level 1 variables simultaneously—that is, we wanted to see if breakup groups differed in one fluctuation variable (e.g., fluctuation in perceived partner commitment), controlling for another fluctuation variable (e.g., fluctuation in own commitment). None
of the commonly used programs accomplish both tasks in a single step. HLM can be used to model intercepts and slopes for several variables simultaneously in a single step (in the Level 1 model). However, it currently cannot be used to compare ended versus persisted relationships in Level 1 fluctuation in these variables, which was the crucial variable in this research. On the other hand, SAS’s PROC MIXED can be used to examine fluctuation in a single variable, but it is not straightforward how to examine several fluctuation variables simultaneously (i.e., several variables in the Level 1 model).5 In short, we could have used SAS’s PROC MIXED to test an all-in-one model to reflect hypotheses concerning fluctuation in only one perception variable (e.g., Hypothesis 1, the association of breakup status with fluctuation in perceived partner commitment, controlling for the slope and intercept). However, we could not use this program to examine several fluctuation variables simultaneously, as was specified in several hypotheses. Testing these hypotheses required adopting the two-stage approach. It is important to note that in cases for which we could test a hypothesis by using SAS PROC MIXED’s all-in-one approach (i.e., the association of breakup status with fluctuation in perceived partner commitment, controlling for the slope and intercept), we analyzed the data this way as well as by using the two-step approach (deriving estimates in one step and analyzing them in a second step); in no case were the results weakened, and in most cases we obtained greater significance by using the all-in-one approach. For the sake of consistency, we present the results of the two-step approach for all hypothesis tests.
Descriptive Statistics and Correlations Table 1 presents the estimates and standard deviations for all within-person change variables for each sample. The standard deviation for each fluctuation variable was not included because multilevel model analyses do not yield this information (i.e., the variance for the residual term). In both studies, participants initially perceived their partner to be relatively committed (7.19 for Study 1 and 7.43 for Study 2, both on a 0 – 8 scale). In Study 1, perceptions of partner commitment significantly declined over time (⫺.06), whereas the decline was a nonsignificant trend in Study 2 (⫺.01). Overall levels of fluctuation in perceived partner commitment were comparable in the two studies (.47 and .48). 3
A linear model suggests that, on average, individuals’ repeated ratings follow a linear pattern more closely than other patterns (e.g., a curvilinear pattern). Linear patterns have been shown to adequately approximate change over relatively short periods of time in relationship variables (Arriaga, 2001), even when the true model follows another pattern (Rogosa, Brant, & Zimowski, 1982). 4 This was the standard error of the estimate, which in this case was the standard deviation in a linear model with two estimates (an intercept and a slope). Given that the model had two estimates rather than just one estimate (such as a model that calculates only the mean level over time rather than the intercept and slope), we calculated the standard deviation by using n – 2 in the denominator to adjust for two model estimates rather than n – 1. 5 We thank Niall Bolger for showing us a way to use SAS PROC MIXED to examine fluctuation in a single (Level 1) variable, which cannot be achieved with hierarchical linear and nonlinear modeling. For more information on the SAS approach, see the Appendix.
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
1053
Table 1 Estimates and Standard Deviations of Within-Person Change Variables (Study 1 and Study 2) Study 1 (n ⫽ 82) Variable Initial level (intercept) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level Trend over time (slope) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level Fluctuation over time (residual) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level
Study 2 (n ⫽ 253)
Estimate
p⬍
SD
p⬍
Estimate
p⬍
SD
p⬍
7.19 6.84 1.80 6.68 7.03
.001 .001 .001 .001 .001
1.01 1.06 1.63 1.25 1.18
.001 .001 .001 .001 .001
7.43 — — 7.01 7.29
.001 — — .001 .001
0.84 — — 1.04 1.04
.001 — — .001 .001
⫺.06 ⫺.07 .09 ⫺.05 ⫺.08
.010 .003 .002 .049 .001
.15 .15 .18 .18 .14
.001 .005 .002 .001 .001
⫺.01 — — ⫺.01 ⫺.03
.195 — — .488 .005
.10 — — .10 .13
.001 — — .001 .001
0.47 0.87 1.27 0.75 0.55
.001 .001 .001 .001 .001
0.48 — — 0.74 0.58
.001 — — .001 .001
Note. Columns labeled p ⬍ indicate whether the corresponding estimate or standard deviation is significantly different from zero. Results of multilevel models do not yield the standard deviation for fluctuation over time (i.e., the within-person residuals); thus, those estimates are omitted above. Dashes indicate that perceived positive partner behavior and perceived partner negative behavior were not measured in Study 2.
Table 2 presents these estimates by breakup group for each study. In both studies, at the univariate level (not controlling for other variables) individuals in relationships that ended had levels of perceived partner commitment that decreased and fluctuated more over time, levels of satisfaction that were initially lower and fluctuated more over time, and levels of own commitment that
were initially lower, decreased more, and fluctuated more over time, relative to individuals in relationships that persisted. The ended group also had perceptions of positive partner behavior that were initially lower and fluctuated more over time, relative to the persisted group (examined in Study 1 only). Group differences in initial level of perceived partner commitment and the trend in own
Table 2 Estimates of Within-Person Change Variables, for Each Breakup Status Group (Study 1 and Study 2) Study 1 (n ⫽ 82) Persisted (n ⫽ 54) Variable Initial level (intercept) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level Trend over time (slope) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level Fluctuation over time (residual) Perceived partner commitment Perceived positive partner behavior Perceived negative partner behavior Own satisfaction level Own commitment level
Study 2 (n ⫽ 253)
Ended (n ⫽ 28)
Persisted (n ⫽ 189)
Ended (n ⫽ 64)
Estimate
p⬍
Estimate
p⬍
Estimate
p⬍
Estimate
p⬍
7.42a 7.08a 1.68 6.99a 7.34a
.001 .001 .001 .001 .001
6.77b 6.42b 2.03 6.10b 6.45b
.001 .001 .001 .001 .001
7.54 — — 7.26a 7.47a
.001 — — .001 .001
7.23 — — 6.35b 6.85b
.001 — — .001 .001
.00a ⫺.05 .07 ⫺.03 ⫺.04a
.753 .036 .020 .117 .007
⫺.17b ⫺.15 .12 ⫺.10 ⫺.17b
.006 .025 .086 .169 .006
.01a — — .01a ⫺.01a
.123 — — .537 .353
⫺.16b — — ⫺.09b ⫺.16b
.001 — — .038 .002
0.26a 0.67a 1.32 0.60a 0.44a
.001 .001 .001 .001 .001
0.89b 1.27b 1.11 1.08b 0.77b
.001 .001 .001 .001 .001
.35a — — .56a .42a
.001 — — .001 .001
0.97b — — 1.48b 1.20b
.001 — — .001 .001
Note. Columns labeled p ⬍ indicate whether the corresponding estimate or standard deviation is significantly different from zero. For each study, estimates within a row with different subscripts indicate a significant difference between persisted and ended, p ⬍ .05. Dashes indicate that perceived positive partner behavior and perceived partner negative behavior were not measured in Study 2.
ARRIAGA, REED, GOODFRIEND, AND AGNEW
1054
Table 3 Correlations Among Within-Person Change Variables (Study 1) Variable Initial level (intercept) 1. Perceived partner commitment 2. Perceived positive behavior 3. Own satisfaction level 4. Own commitment level Trend over time (slope) 5. Perceived partner commitment 6. Perceived positive behavior 7. Own satisfaction level 8. Own commitment level Fluctuation over time (residual) 9. Perceived partner commitment 10. Perceived positive behavior 11. Own satisfaction level 12. Own commitment level * p ⬍ .05.
1
2
3
4
5
6
7
— .58** .58** .49**
— .76** .65**
— .85**
—
.54** .28* ⫺.11 .38**
.62** .73** .22* .58**
.67** .65** .53** .77**
.59** .48** .36** .82**
— .64** .40** .63**
— .60** .57**
⫺.66** ⫺.37** ⫺.34** ⫺.55**
⫺.58** ⫺.74** ⫺.45** ⫺.47**
⫺.54** ⫺.47** ⫺.55** ⫺.65**
⫺.52** ⫺.30** ⫺.44** ⫺.73**
⫺.55** ⫺.49** ⫺.30** ⫺.40**
⫺.28* ⫺.54** ⫺.28* ⫺.30**
— .56** .04 ⫺.10 ⫺.08 ⫺.15
8
9
10
11
12
— .63** .57** .52**
— .49** .28*
— .55**
—
— ⫺.45** ⫺.29** ⫺.27** ⫺.49**
** p ⬍ .01.
second sample was composed of the remaining participants (n ⫽ 177). The mean relationship duration of participants in the first Study 2 subsample was 3 months, as was the case in Study 1; the mean relationship duration of participants in the second Study 2 subsample was 22 months. Thus, the Study 1 sample and first Study 2 subsample were comparable in the type of participant and sample size, and the only major difference was the time lag between measurement occasions (1 week in Study 1 vs. 4 weeks in Study 2).
satisfaction level were less robust (i.e., significant for one but not both studies). None of the results involving perceived negative partner behavior were significant, so this variable was not examined in further analyses. Tables 3 and 4 display respectively Study 1 and Study 2 correlations among predictor variables. Correlations among variables tapping a particular type of change (i.e., a particular change estimate, such as fluctuation) were generally higher in Study 2 than Study 1. Correlations among the three change estimates for a single predictor variable tended to be comparable in both studies and generally fell into a moderate to high range (the average magnitude was .54).
Testing Hypothesis 1 Do fluctuations in perceived partner commitment uniquely predict breakup? As can be seen in Table 5 (top half for Study 1, bottom half for Study 2: rows for Perceived Partner Commitment, Simple Association column), initial level, linear trend, and fluctuation in perceived partner commitment each were significantly associated with breakup status. In both studies, when examining initial level of perceived partner commitment, the linear trend, and fluctuation in a simultaneous regression, fluctuation had a unique association with breakup status, as did the trend (see the Individual
Comparing Study 1 and Study 2 We assessed support for each hypothesis in Study 1 and Study 2. In cases where results of the two studies differed in ways that were relevant to the hypotheses, we divided the Study 2 sample into two subsamples. The first subsample, comparable to Study 1, included participants who had been dating 6 months or less (n ⫽ 76); the Table 4 Correlations Among Within-Person Change Variables (Study 2) Initial level Variable Initial level (intercept) 1. Perceived partner commitment 2. Own satisfaction level 3. Own commitment level Trend over time (slope) 4. Perceived partner commitment 5. Own satisfaction level 6. Own commitment level Fluctuation over time (residual) 7. Perceived partner commitment 8. Own satisfaction level 9. Own commitment level * p ⬍ .05.
** p ⬍ .01.
1
2
Trend over time 3
4
5
— .75** .67**
— .79**
—
.20** .38** .17**
.35** .48** .31**
.16** .32** .16*
— .67** .64**
— .77**
⫺.67** ⫺.47** ⫺.45**
⫺.59** ⫺.68** ⫺.60**
⫺.46** ⫺.41** ⫺.57**
⫺.50** ⫺.42** ⫺.35**
⫺.54** ⫺.55** ⫺.52**
Fluctuation over time 6
7
8
9
— .67** .64**
— .79**
—
— ⫺.36** ⫺.46** ⫺.51**
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
1055
Table 5 Regression Analyses Predicting Breakup Status (Ended vs. Persisted), Study 1 and Study 2 Simultaneous effects
Variable
Simple association (Pearson r)
Individual estimate ()
Overall model R2
F
df
Study 1 (n ⫽ 82) Perceived partner commitment Initial level Linear trend Fluctuation Perceived positive partner behavior Initial level Linear trend Fluctuation Own satisfaction Initial level Linear trend Fluctuation Own commitment Initial level Linear trend Fluctuation
.37** .46** ⫺.48**
.01 .28* ⫺.32*
.28**
10.24**
3, 78
.37** .31** ⫺.34**
.19 .08 ⫺.16
.14**
4.56**
3, 78
.42** .17 ⫺.33**
.36* ⫺.03 ⫺.13
.19**
6.04**
3, 78
.44** .41** ⫺.29**
.35 .13 .03
.20**
6.57**
3, 78
Study 2 (n ⫽ 253) Perceived partner commitment Initial level Linear trend Fluctuation Own satisfaction Initial level Linear trend Fluctuation Own commitment Initial level Linear trend Fluctuation
.28** .37** ⫺.40**
.09 .24** ⫺.22*
.20**
21.02**
3, 249
.41** .30** ⫺.41**
.24** .08 ⫺.21*
.21**
21.50**
3, 249
.33** .31** ⫺.41**
.17* .16* ⫺.23**
.20**
20.45**
3, 249
Note. Breakup status was coded 0 for ended and 1 for persisted. * p ⬍ .05. ** p ⬍ .01.
estimate column). We performed a second simultaneous regression that also included higher order interactions among initial level, trend, and fluctuation. None of the interactions were significant in Study 1, but there was a significant two-way interaction between the linear trend and fluctuation in Study 2, t(253) ⫽ ⫺3.29, p ⫽ .001. To decompose the interaction, we did a median split on the linear trend. Controlling for initial level, in Study 2 greater fluctuation was highly associated with less persistence among individuals whose perceptions increased over time ( ⫽ ⫺.46, p ⫽ .002) and not associated with breakup status among those whose perceptions decreased over time ( ⫽ ⫺.18, p ⫽ .116). Initial level of perceived partner commitment did not have a unique association with breakup status in any of these analyses. Given the inconsistent finding between Study 1 and Study 2, we repeated the interaction analysis in the two subsamples of Study 2 described above, one with individuals in more recently initiated relationships (comparable with Study 1) and another with the remaining Study 2 participants. When regressing breakup status onto initial level, the linear trend, fluctuation, and higher order interactions, the interaction between linear trend and fluctuation
was not significant in the Study 2 subsample of individuals in newly formed relationships, t(76) ⫽ ⫺1.01, ns, as was the case in Study 1. However, this interaction was significant in the Study 2 subsample of individuals in relatively established relationships (longer than 6 months), t(177) ⫽ ⫺2.29, p ⫽ .023, and decomposition of this interaction revealed a pattern similar to the full Study 2 sample. This suggests that Hypothesis 1 received robust support among individuals in newly formed relationships: Controlling for their initial level and increases or decreases in perceived partner commitment, we found that they were more likely to have relationships that ended to the extent that their levels of perceived partner commitment fluctuated over time; the linear trend in perceived partner commitment also had a unique association with breakup status, but initial perceived partner commitment did not. Support for Hypothesis 1 among individuals in more established relationships was conditional: The unique association of fluctuation in perceived partner commitment with breakup status occurred only among individuals whose perceptions increased over time and not among those whose perceptions decreased over time.
1056
ARRIAGA, REED, GOODFRIEND, AND AGNEW
Do specific temporal profiles have different odds of breakup? The regression results just described can be interpreted in terms of the four specific temporal profiles outlined in the introduction. The moderated effect of fluctuation among individuals in longer lasting relationships suggests that, when one’s overall perception of the partner’s commitment (initial or absolute level) is controlled for, the amount of fluctuation in perceptions is linked to breakup status but only among those individuals whose perceptions are increasing over time (Profiles 1 and 2) and not among those whose perceptions are decreasing over time (Profiles 3 and 4). In contrast, the main effect of fluctuation among individuals in newly formed relationships suggests that, regardless of one’s perception of the partner at the outset of the study or whether one’s perceptions increase or decrease over time, greater fluctuation (Profiles 2 and 4) goes with greater odds of breakup. Tables 6 and 7 provide a concrete illustration of the odds of breakup for each of the four profiles among individuals in newly formed relationships, where there was a main effect for fluctuation (entire sample of Study 1 and subsample of Study 2). We created two groups to reflect the trend in perceived partner commitment— those whose levels increased over time (i.e., positively sloped values; n ⫽ 36 for Study 1 and n ⫽ 47 for the Study 2 subsample) versus those whose levels decreased (i.e., negatively sloped values; n ⫽ 46 for Study 1 and n ⫽ 29 for the Study 2 subsample). We conducted a median split on fluctuation in perceived partner commitment to create two groups—those whose levels fluctuated over time (n ⫽ 41 for Study 1 and n ⫽ 38 for the Study 2 subsample) versus those whose levels were relatively steady (n ⫽ 41 for Study 1 and n ⫽ 38 for the Study 2 subsample). Combining these two categorical variables allowed us to examine frequencies and percentages of ended versus persisted relationships in each of four groups approximating a temporal profile.6 The results are displayed in the top halves of Table 6 (Study 1) and Table 7 (Study 2 subsample). The four groups were defined by actual levels on the trend and fluctuation variable and initial levels were not controlled for; as such, this analysis does not fully correspond to the regression models used to test Hypothesis 1, which controlled for initial level. Two findings stand out from this analysis. First, individuals whose perceptions of their partner’s commitment increased over time were much more likely to have steady than fluctuating perceptions (28 of 36, or 78%, in Study 1; 29 of 47, or 62%, in the Study 2 subsample), and individuals whose perceptions of their partner’s commitment decreased over time were much more likely to have fluctuating than steady perceptions (33 of 46, or 72%, in Study 1; 20 of 29, or 69%, in the Study 2 subsample). This is consistent with the correlations reported in Table 3 (Study 1) and Table 4 (Study 2): Individuals’ trends over time in perceived partner commitment and the extent of fluctuation in their trends were closely aligned: Study 1, r(82) ⫽ ⫺.55, p ⬍ .001; Study 2 full sample, r(253) ⫽ ⫺.50, p ⬍ .001; Study 2 subsample, r(76) ⫽ ⫺.24, p ⫽ .038. Second, when one examines individuals whose perceptions of partner commitment declined over time (i.e., rows labeled Profiles 3 and 4 in Tables 6 and 7), those whose perceptions followed a steady decline (Profile 3) were likely to be in relationships that persisted despite the decline (77% in Study 1 and 78% in the Study 2 subsample). This was not the case for those whose perceptions followed a fluctuating pattern of decline over time (Profile 4)—
roughly half of these individuals were in relationships that ended (58% in Study 1 and 60% in the Study 2 subsample). A similar pattern occurred among individuals whose perceptions followed a pattern of increase over time (Profiles 1 and 2)—a pattern of fluctuation increased odds of breakup. We repeated all of the descriptive analyses by calculating each individual’s mean level of perceived partner commitment over time and doing a median split on this variable rather than using his or her slope over time. The same pattern of results emerged.
Testing Hypothesis 2 Hypothesis 2 suggested that, when examined simultaneously, perceptions of partner commitment should predict later breakup status better than perceptions of specific positive and negative partner behaviors (measured in Study 1 only). We did not examine perceptions of negative behaviors given that breakup groups did not differ on any of the three change estimates of this variable (see Table 2). All three of the change estimates of the perceived positive behavior variable had a significant correlation with breakup status (see Table 5, first column) but none had a unique association (Table 5, second column), possibly because of the high intercorrelations among these three variables, particularly initial level of perceived positive behavior with the linear trend and with fluctuation (see Table 2). To test Hypothesis 2, we compared the predictive value of each set of variables—initial level, trend, and fluctuation for perceived positive partner behavior versus the same three variables for perceived partner commitment. First, when breakup status was regressed onto all six variables, the change in R2 associated with adding the three perceived partner commitment variables as a group (i.e., the change in R2 when comparing a model with all six variables vs. one with the three perceived positive behavior variables only) was significant, F(3, 75) ⫽ 4.82, p ⫽ .004, whereas the change in R2 associated with adding the three perceived positive partner behavior variables as a group (i.e., the change in R2 when comparing a model with all six variables vs. one with the three 6
As stated in the introduction, we do not intend to suggest that individuals perfectly cluster into these four profiles; individuals vary continuously on the variables used to approximate the four profiles. Tables 6 and 7 include the number of participants approximating each profile for Study 1 and the Study 2 subsample. Not all participants had levels that increased or decreased over time. In Study 1, 35% of participants (n ⫽ 29) had slopes that were flat (i.e., they indicated the same response at each time); in Study 2, 36% of participants (n ⫽ 90) had flat slopes. Whereas the actual data of someone who provides the same response over time would suggest a slope of 0, multilevel modeling assigns slope values that are adjusted for the sample (or group) average slope. Categories in Tables 6 and 7 were based on the slope values assigned in a multilevel model analysis, some of which were close to 0 but slightly positive or negative (e.g., ⫺.001). The extent of increase over time for those whose perceptions increased was less than the extent of decrease over time for those whose perceptions decreased (.03 vs. ⫺.12 in Study 1; .02 vs. ⫺.08 in Study 2). When one removes individuals who started at the highest level and remained stable week after week at the highest level (i.e., an analysis removing respondents at the ceiling, as reported in a later section), the extent of increase versus decrease still differs but not as much (.07 vs. ⫺.12 in Study 1; .04 vs. ⫺.08 in Study 2).
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
1057
Table 6 Breakup Rates as a Function of Perceived Partner Commitment Over Time for the Entire Study 1 Sample and for a Subsample Controlling for a Possible Ceiling Effect Breakup group Persisted Perceived partner commitment
%
Ended n
%
n
86 75
24 6
14 25
4 2
77 42
10 14
23 58
3 19
90 50
9 2
10 50
1 2
75 35
15 9
25 65
5 17
Entire sample (n ⫽ 82) Perceptions of partner commitment increase over time (n ⫽ 36) Profile 1: Steady increase (n ⫽ 28) Profile 2: Fluctuating increase (n ⫽ 8) Perceptions of partner commitment decrease over time (n ⫽ 46) Profile 3: Steady decline (n ⫽ 13) Profile 4: Fluctuating decline (n ⫽ 33)
Subsample controlling for possible ceiling effect (n ⫽ 60) Perceptions of partner commitment increase over time (n ⫽ 14) Profile 1: Steady increase (n ⫽ 10) Profile 2: Fluctuating increase (n ⫽ 4) Perceptions of partner commitment decrease over time (n ⫽ 46) Profile 3: Steady decline (n ⫽ 20) Profile 4: Fluctuating decline (n ⫽ 26)
Note. Profiles 1– 4 correspond to the four temporal profiles depicted in Figure 1 and described in the Introduction. For the entire sample, n for the persisted group ⫽ 54 and n for the ended group ⫽ 28. For the subsample, n for the persisted group ⫽ 35 and n for the ended group ⫽ 25.
perceived partner commitment variables only) was not significant, F(3, 75) ⫽ 0.14, ns. Thus, Hypothesis 2 was supported. We sought to determine whether each perceived partner commitment variable predicted breakup status above and beyond the effect of the positive behavior variables. We did not attempt to interpret the effects of specific estimates in a simultaneous regression with six predictors, either in this analysis or subsequent analyses combining perceived partner commitment with own satisfaction or with own commitment, because of high levels of multicollinearity among the predictors (i.e., several with variance inflation factors above 4.0). Instead, we conducted three follow-up analyses, one for each perceived partner commitment variable, controlling for the three positive behavior variables. Controlling for the three perceived positive behavior variables in each of three separate regressions, breakup status was predicted by initial level in perceived partner commitment: overall model, F(4, 77) ⫽ 4.87, p ⫽ .002 (individual estimate for initial level of perceived partner commitment,  ⫽ .30, p ⫽ .027); by the linear trend in perceived partner commitment: overall model, F(4, 77) ⫽ 5.78, p ⬍ .001 ( ⫽ .39, p ⫽ .005); and by fluctuation in perceived partner commitment: overall model, F(4, 77) ⫽ 6.85, p ⬍ .001 ( ⫽ ⫺.46, p ⬍ .001). Thus, each of the three perceived partner commitment variables was a robust predictor of breakup when controlling for perceived positive behaviors.
Testing Hypotheses 3 We examined whether fluctuation in perceptions of the partner’s commitment and fluctuation in one’s own level of satisfaction each had a unique association with subsequent breakup status when examined simultaneously (Hypothesis 3). In regressing breakup status onto both variables simultaneously, the overall model was
significant in Study 1, F(2, 79) ⫽ 12.01, p ⬍ .001, and Study 2, F(2, 250) ⫽ 30.38, p ⬍ .001; fluctuation in perceived partner commitment provided unique prediction in Study 1 ( ⫽ ⫺.42, p ⬍ .001) and Study 2 ( ⫽ ⫺.23, p ⫽ .003), and fluctuation in own satisfaction provided unique prediction in Study 2 ( ⫽ ⫺.26, p ⬍ .001) but not Study 1 ( ⫽ ⫺.08, ns). In the Study 2 subsample that was comparable with the Study 1 sample, each of the two fluctuation variables uniquely predicted breakup status (fluctuation in perceived partner commitment,  ⫽ ⫺.28, p ⫽ .025; fluctuation in own satisfaction,  ⫽ ⫺.25, p ⫽ .043). Thus, Hypothesis 3 was moderately supported, with strong support for fluctuation in perceived partner commitment and partial support (in two of three samples) for fluctuation in own level of satisfaction. We compared the relative predictive value of own satisfaction variables versus perceived partner commitment variables in analyses parallel to those for Hypothesis 2. First, as seen in Table 5 (rows labeled Own Satisfaction in top half for Study 1 and bottom half for Study 2), each of the satisfaction variables was correlated with breakup status except for the linear trend in Study 1 (Simple Association column). Initial level in own satisfaction had a unique effect in Study 1 and Study 2 (Individual Estimate column) and fluctuation had a unique effect in Study 2; the linear trend in own satisfaction did not have a unique effect. Next, in Study 1, when entering the three perceived partner commitment variables as a group, controlling for the three own satisfaction variables, the change in R2 was significant, F(3, 75) ⫽ 3.74, p ⫽ .015, whereas the change in R2 associated with adding the three own satisfaction variables, controlling for the three perceived partner commitment variables, was not significant, F(3, 75) ⫽ 0.41, ns. In Study 2, when entering the three perceived partner commitment variables as
ARRIAGA, REED, GOODFRIEND, AND AGNEW
1058
Table 7 Breakup Rates as a Function of Perceived Partner Commitment Over Time for Study 2 in Newly Formed Relationships and for a Subsample Controlling for a Possible Ceiling Effect Breakup group Persisted Perceived partner commitment
%
Ended n
%
n
93 83
27 15
7 17
2 3
78 40
7 8
22 60
2 12
95 79
19 11
5 21
1 3
60 40
6 6
40 60
4 9
Entire sample (n ⫽ 76) Perceptions of partner commitment increase over time (n ⫽ 47) Profile 1: Steady increase (n ⫽ 29) Profile 2: Fluctuating increase (n ⫽ 18) Perceptions of partner commitment decrease over time (n ⫽ 29) Profile 3: Steady decline (n ⫽ 9) Profile 4: Fluctuating decline (n ⫽ 20)
Subsample controlling for possible ceiling effect (n ⫽ 59) Perceptions of partner commitment increase over time (n ⫽ 34) Profile 1: Steady increase (n ⫽ 20) Profile 2: Fluctuating increase (n ⫽ 14) Perceptions of partner commitment decrease over time (n ⫽ 25) Profile 3: Steady decline (n ⫽ 10) Profile 4: Fluctuating decline (n ⫽ 15)
Note. Profiles 1– 4 correspond to the four temporal profiles depicted in Figure 1 and described in the Introduction. For the entire sample, n for the persisted group ⫽ 57 and n for the ended group ⫽ 19. For the subsample, n for the persisted group ⫽ 42 and n for the ended group ⫽ 17.
a group, controlling for the three own satisfaction variables, the change in R2 was significant, F(3, 246) ⫽ 5.23, p ⫽ .002; the change in R2 associated with adding the three own satisfaction variables, controlling for the three perceived partner commitment variables, was also significant, F(3, 246) ⫽ 5.64, p ⬍ .001. Thus, perceived partner commitment variables were more predictive of breakup status than own satisfaction variables in Study 1, and both sets of variables were predictive in Study 2. We repeated the analysis in the subsample of Study 2 that was comparable to Study 1. The change in R2 associated with adding perceived partner commitment variables was significant, F(3, 69) ⫽ 7.62, p ⬍.001, as was adding own satisfaction variables, F(3, 69) ⫽ 3.45, p ⫽ .021, but the change in R2 was larger from adding the perceived partner commitment variables (.22 to .41) than from adding satisfaction variables (.32 to .41). To determine whether each perceived partner commitment variable might have a unique role in predicting breakup status when controlling for the three own satisfaction variables, we conducted follow-up analyses, one for each perceived partner commitment variable. In three analyses of Study 1 data, controlling for the three own satisfaction variables, breakup status was significantly predicted by the linear trend in perceived partner commitment: overall model, F(4, 77) ⫽ 6.57, p ⬍ .001 (individual estimate  ⫽ .35, p ⫽ .011); and by fluctuation in perceived partner commitment: overall model, F(4, 77) ⫽ 7.19, p ⬍ .001 ( ⫽ ⫺.40, p ⫽ .001); but initial level in perceived partner commitment only approached significant prediction of breakup status: overall model, F(4, 77) ⫽ 5.38, p ⬍ .001 ( ⫽ .27, p ⫽ .091). In three analyses of Study 2 data, controlling for the three own satisfaction variables, breakup status was predicted by the linear trend in perceived partner commitment: overall model, F(4, 248) ⫽ 20.16, p ⬍ .001 ( ⫽
.27, p ⫽ .001); and by fluctuation: overall model, F(4, 248) ⫽ 17.33, p ⬍ .001 ( ⫽ .16, p ⫽ .046), but not by initial level. Thus, when controlling for own satisfaction, the linear trend and fluctuation had robust unique associations with breakup status; initial level did not.
Testing Hypotheses 4 We examined whether fluctuation in perceptions of the partner’s commitment and fluctuation in one’s own level of commitment each had a unique association with subsequent breakup status, when examined simultaneously (Hypothesis 4). In regressing breakup status onto both variables simultaneously, the overall model was significant in Study 1, F(2, 79) ⫽ 11.88, p ⬍ .001, and Study 2, F(2, 250) ⫽ 30.82, p ⬍ .001; fluctuation in perceived partner commitment provided unique prediction in Study 1 ( ⫽ ⫺.45, p ⬍ .001) and Study 2 ( ⫽ ⫺.23, p ⫽ .002), and fluctuation in own commitment provided unique prediction in Study 2 ( ⫽ –.26, p ⬍ .001) but not Study 1 ( ⫽ ⫺.06, ns). The same analysis in the Study 2 subsample that was comparable with the Study 1 sample replicated Study 1: Fluctuation in perceived partner commitment uniquely predicted breakup status ( ⫽ ⫺.29, p ⫽ .023), but fluctuation in own commitment did not ( ⫽ ⫺.21, p ⫽ .103). Thus, Hypothesis 4 was moderately supported, with strong support for fluctuation in perceived partner commitment and partial support for fluctuation in own commitment level (only among individuals in established relationships). We compared the relative predictive value of own commitment variables versus perceived partner commitment variables in analyses parallel to those for Hypotheses 2 and 3. First, as seen in Table 5 (rows labeled Own Commitment in top half for Study 1
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
and bottom half for Study 2), each variable tapping own commitment was correlated with breakup status (Simple association column). None of the own commitment variables had unique effects in Study 1 (the variance inflation factor for initial level of own commitment was over 5.0), but all had unique effects in Study 2 (Individual Estimate column). In Study 1, when entering the three perceived partner commitment variables as a group, controlling for the three own commitment variables, the change in R2 was significant, F(3, 75) ⫽ 4.00, p ⫽ .011, whereas the change in R2 associated with adding the three own commitment variables, controlling for the three perceived partner commitment variables, was not significant, F(3, 75) ⫽ 1.06, ns. In Study 2, when entering the three perceived partner commitment variables as a group, controlling for the three own commitment variables, the change in R2 was significant, F(3, 246) ⫽ 5.21, p ⫽ .002; the change in R2 associated with adding the three own commitment variables, controlling for the three perceived partner commitment variables, was also significant, F(3, 246) ⫽ 4.72, p ⬍ .003. Thus, perceived partner commitment variables were more predictive of breakup status than own commitment variables in Study 1, but both were predictive in Study 2. Given this inconsistency, we repeated the analysis in the Study 2 subsample that was comparable with Study 1; replicating Study 1, the change in R2 was significant in adding the three perceived partner commitment variables, F(3, 69) ⫽ 4.06, p ⫽ .010, but not in adding the three own commitment variables, F(3, 69) ⫽ 0.78, ns. We also conducted follow-up analyses, one for each perceived partner commitment variable, to determine whether each of these three variables predicted breakup status beyond the three own commitment variables. In three analyses of Study 1 data, controlling for the three own commitment variables, breakup status was significantly predicted by the linear trend in perceived partner commitment: overall model, F(4, 77) ⫽ 6.60, p ⬍ .001 (individual estimate  ⫽ .30, p ⫽ .021); and by fluctuation in perceived partner commitment: overall model, F(4, 77) ⫽ 8.06, p ⬍ .001 ( ⫽ ⫺.37, p ⫽ .002); but initial level in perceived partner commitment only approached significance in prediction of breakup status: overall model, F(4, 77) ⫽ 6.10, p ⬍ .001 ( ⫽ .24, p ⫽ .051). In three analyses of Study 2 data, controlling for the three own commitment variables, breakup status was predicted by the linear trend in perceived partner commitment: overall model, F(4, 248) ⫽ 19.57, p ⬍ .001 ( ⫽ .27, p ⬍ .001); and by fluctuation in perceived partner commitment: overall model, F(4, 248) ⫽ 17.61, p ⬍ .001 ( ⫽ ⫺.21, p ⫽ .007), but not initial level. Thus, when controlling for own commitment, breakup status is reliably predicted from the linear trend in perceived partner commitment and fluctuation, but not from initial level. We conducted several exploratory analyses to further examine the link between perceived partner commitment over time and own commitment over time. First, we explored whether own commitment precedes perceptions of partner commitment or vice versa. In Study 1, the correlation of initial level of own commitment with the subsequent linear trend in perceived partner commitment, r(82) ⫽ .59, p ⬍ .001, was higher than the correlation of initial level of perceived partner commitment with the subsequent trend in own commitment, r(82) ⫽ .38, p ⬍ .001. In Study 2, these two correlations were roughly equal: initial own commitment with trend in perceived partner commitment, r(253) ⫽ .16, p ⬍ .001; initial perceived partner commitment with trend in own commit-
1059
ment, r(253) ⫽ .17, p ⬍ .001. The Study 2 subsample of individuals in newly formed relationships revealed a pattern similar to Study 1: initial own commitment with trend in perceived partner commitment, r(76) ⫽ .39, p ⬍ .001; initial perceived partner commitment with trend in own commitment, r(76) ⫽ .28, p ⫽ .015.7 With the current correlational design, we attempted to rule out rather than confirm a possible causal path; neither causal path could be ruled out (all correlations were significant) rendering both causal paths possible, but among individuals in newly formed relationships, there was more support suggesting that own commitment precedes perceptions of the partner’s commitment. An additional analysis aimed to explore whether one’s own commitment level at the outset of the study (initial commitment) constrains the extent of doubt one experiences—that is, whether one’s initial commitment level moderated the link between fluctuation in perceived partner commitment and breakup. It is possible that people who are not very committed at the outset do not pay much attention to their partner’s level of commitment and thus are not affected by the extent that they see their partner as stably committed. On the other hand, people who are initially highly committed may seek similar evidence of commitment from their partner and as such are sensitive to perceived vacillations in their partner’s commitment; for them, vacillating perceptions may predict later relationship outcomes more than for those who are not initially committed. We regressed breakup status onto initial level of own commitment, fluctuation in perceived partner commitment, and the interaction between these two variables. The interaction was not significant in either study: Study 1, t(82) ⫽ 0.23, ns; Study 2, t(253) ⫽ ⫺0.17, ns. One’s own initial level of commitment did not constrain the extent to which fluctuations in perceptions of the partner’s commitment predicted later breakup status.
Testing Hypothesis 5 Was there evidence that experiencing fluctuations in one’s perceptions of partner commitment has its origins in a dispositional tendency to be anxiously attached to partner (measured in Study 1)? Consistent with Hypothesis 5, initial level of anxious attachment and subsequent fluctuation in perceived commitment were positively correlated in Study 1, r(82) ⫽ .39, p ⬍ .001; initial level of avoidant attachment and subsequent fluctuation in perceived commitment were also positively correlated, r(82) ⫽ .29, p ⬍ .001. We regressed fluctuations in perceived partner commitment onto initial anxious attachment style and initial avoidant attachment style simultaneously; initial anxious attachment exhibited a 7 A statistical issue in attempting to interpret these correlations concerns the correlation between initial level and linear trend within each variable. The variable (perceived partner commitment vs. own commitment) with a lower link between initial level and subsequent trend will appear to be a cause rather than an outcome (Rogosa, 1981). There was little concern in Study 1, as the correlation between initial level and the linear trend was higher for own commitment (.82) than for perceived partner commitment (.54), and yet the cross-correlations (initial level of one variable with linear trend of the other variable) suggested own commitment as a cause; the same pattern occurred in the Study 2 subsample. In the full Study 2 sample, the correlation between initial level and linear trend in own commitment (.16) was comparable with the correlation between initial level and linear trend in perceived partner commitment (.20).
1060
ARRIAGA, REED, GOODFRIEND, AND AGNEW
significant unique association with subsequent fluctuation in perceived partner commitment ( ⫽ .33, p ⫽ .002), whereas the association of initial avoidant attachment style only approached significance ( ⫽ .19, p ⫽ .070). We also conducted exploratory analyses with all of the attachment change estimates. None of the other attachment variables were correlated with fluctuation in perceived commitment: anxious attachment trend, r(82) ⫽ ⫺.14, p ⫽ .206; anxious attachment fluctuation, r(82) ⫽ ⫺.11, p ⫽ .343; avoidant attachment trend, r(82) ⫽ .11, p ⫽ .316; avoidant attachment fluctuation, r(82) ⫽ .00, p ⫽ .999. Also, none of the attachment variables were correlated with later breakup status. Together these findings suggest that anxiously attached individuals may be predisposed to greater volatility in their perceptions of partner commitment, but this predisposition does not account for the association between fluctuation in perceptions of partner commitment and later breakup.
Do the Findings Reflect a Measurement Artifact? A general concern with inferring meaning from temporal profiles generated from self-report responses is that the profiles are influenced by the response scales. A specific concern is that when individuals are asked how committed they perceive their partner to be, those who endorse the highest scale value time after time would have both high and steady (unchanging) levels of perceived partner commitment. This would create a ceiling effect in which initial levels would be high, the slope would be zero or flat (neither increasing nor decreasing), and there would be no fluctuation. By virtue of their indicating the highest possible level of perceived partner commitment, it may be level of perceived partner commitment—not stability per se—that is driving the significant association with breakup status. On the other hand, for those who report steady levels, but levels that are lower than the absolute highest level, there would not be a scale ceiling forcing steady levels. We examined whether support for each hypothesis was robust when controlling for a possible ceiling effect by eliminating any individuals who endorsed the highest possible level of perceived partner commitment at all times. This reduced the sample sizes to 60 participants in Study 1 and 168 participants in Study 2. Despite the reduction in statistical power, the pattern of significant findings for each hypothesis test remained the same; there were minor changes in the specific estimate values (e.g., initial levels of perceived partner commitment, own satisfaction, and own commitment were slightly lower for this subsample). The results of two ancillary analyses changed slightly: The simple correlation between fluctuation in own commitment and breakup status became nonsignificant in Study 1, r(60) ⫽ ⫺.17, ns, and the unique effect of fluctuation in own satisfaction in predicting breakup, when we controlled for satisfaction initial level and linear trend, became nonsignificant in Study 2 ( ⫽ ⫺.15, p ⫽ .104). Descriptively, when categorizing individuals in newly formed relationships (see bottom half of Tables 6 and 7) into each of the four temporal profiles, the percentages of ended versus persisted relationships were similar or slightly more compelling. For example, in comparing the full Study 1 sample with the Study 1 subsample with the ceiling removed, the pattern for Profile 1 remained the same—those whose perceptions of their partner’s commitment steadily increased over time were overwhelmingly likely to be in persisting relationships. However, the pattern for
Profile 2 changed; in the full Study 1 sample, the majority of individuals whose perceptions fluctuated as they increased were likely to remain in their relationship, but in the subsample with the ceiling removed, those fitting this profile were equally likely to be in relationships that ended (50%) versus persisted (50%) despite perceiving an increasingly committed partner. Other Study 1 comparisons or comparisons for the Study 2 subsample revealed that removing participants at the ceiling had little effect on the results.
Discussion Summary of Main Findings and Conclusions The main findings suggest new avenues for understanding how certainty versus doubt unfolds over time and how each relates to later outcomes. Individuals whose perceptions of their partner’s commitment fluctuated over time were more likely to be in relationships that ended than individuals whose perceptions were relatively steady. The association of fluctuation in perceived partner commitment with relationship dissolution remained robust when controlling for fluctuation in one’s own level of commitment, fluctuation in one’s own level of satisfaction, and perceptions of the partner’s positive behavior (consistent with Hypotheses 2, 3, and 4). The link between fluctuation in perceived partner commitment and relationship dissolution also remained robust when controlling for initial perceptions of the partner’s commitment among all participants and when controlling for whether these perceptions increased or decreased over time among individuals in relatively newly formed relationships (consistent with Hypothesis 1). Among individuals in relationships of longer duration (i.e., longer than 6 months), greater fluctuation in perception of partner commitment was associated with greater odds of dissolution if those perceptions increased over time but not if they decreased over time (qualified support for Hypothesis 1). These findings suggest different conclusions, depending on how far along a relationship is. Individuals in relatively established relationships whose perceptions of their partner’s commitment decrease over time are not affected by the extent of volatility in their perceptions—things are not good (as inferred from decreasing levels of perceived partner commitment), and perceived ups and downs bear little on having greater odds of breakup. However, those whose perceptions of their partner’s commitment increase over time mirror individuals in relatively novel relationships; they remain hopeful and thus sensitive to ups and downs in their perceptions of the partner. The latter findings suggest some counterintuitive (but hypothesis-consistent) conclusions: Even among individuals who perceived their partner to be increasingly committed, odds of having a relationship that ended increased simply by virtue of having perceptions that fluctuated over time. Also, even among individuals who perceived their partner to be decreasingly committed, odds of having a relationship that lasted increased simply by virtue of having perceptions that were steady over time. Another difference between individuals in relatively newly formed relationships versus those in more established relationships concerned the relative predictive value of perceptions of partner commitment versus own level of commitment or own level of satisfaction. In the sample of individuals in relatively established relationships, all of these variables predicted later relationship
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
persistence versus dissolution. On the other hand, in the two samples of individuals in more recently initiated relationships, perceived partner commitment variables accounted for more variance in later relationship status than did own commitment or own satisfaction. This provides suggestive evidence (albeit not conclusive evidence given the correlational design) that among relatively novel relationships, perceptions of where the partner stands may figure more prominently in the ultimate course of the relationship than a sense of one’s own satisfaction or commitment, whereas in more established relationships concerns about the partner and about oneself are similarly implicated. The current studies also replicated past research focusing on changes in one’s level of satisfaction or one’s level of commitment. Consistent with Rusbult (1983), participants whose own level of commitment increased over time were more likely to be in lasting relationships, and those whose commitment decreased were more likely to be in relationships that ended. However, when controlling for level and fluctuation in commitment, this association remained significant only among individuals in established relationships. Consistent with Arriaga (2001), greater fluctuation in satisfaction was associated with greater odds of relationship dissolution, and this association remained robust in two of three samples (the two Study 2 samples) when controlling for initial level of satisfaction and the linear trend over time. We have suggested that the extent of fluctuation in perceptions of partner commitment reflects varying levels of certainty or doubt over time about where the partner stands with respect to the relationship. Not all individuals may experience the same degree of certainty or doubt over time. Murray et al. (2005) have suggested that each individual may have unique ways of navigating situations of interdependence with a partner, and they have identified meaningful patterns in individual responses. In a related vein, we have suggested that each individual exhibits an idiosyncratic temporal pattern in his or her view of a partner’s commitment, and we identified four patterns in individual responses— or profiles (illustrated in Figure 1)—that reflect distinct and theoretically meaningful experiences of certainty versus doubt.
Origins of Fluctuations in Perceived Partner Commitment What drives some people to perceive their partner as being stably committed and others to perceive wavering levels of partner commitment? There are several possible answers to this question. One answer that we did not examine in the current research is that their partners really are stably versus unstably committed. To an extent, people are accurate in their general perceptions of a partner, but there is mounting evidence that accurate information is processed in a biased way—for example, individuals may perceive objective information about partner faults but then integrate it into a broader impression that downplays the importance of these faults and sustains a positive perception (cf. Murray, 1999). Because we did not have partner data, we could not examine this issue directly. We sought indirect evidence, however, of whether individuals were relatively accurate in their perceptions of partner commitment. We examined reports of individuals whose relationships ended and specifically compared own level of commitment and perceived level of partner commitment among those who initiated the breakup (i.e., leavers, n ⫽ 19 in Study 1 and n ⫽ 29 in Study 2) versus those whose partners initiated the breakup (i.e., aban-
1061
doned, n ⫽ 9 in Study 1 and n ⫽ 31 in Study 2).8 In general, leavers made more distinctions between their own level of commitment and their partner’s level in the ways one might expect; they reported lower initial levels of their own commitment, greater declines in their own commitment, and more fluctuation over time in their own commitment, relative to what they reported for the partner (significantly so for all three variables in Study 2 but only one variable in Study 1). On the other hand, abandoned individuals did not make these distinctions; despite ending the relationship, the partner was not perceived to be less committed initially, less over time, and less stably. This provides very tentative and indirect evidence that individuals are cognizant of where their partner stands relative to them but there is also room for bias, as abandoned individuals were more reluctant or unable to recognize differences— differences that did not work in their favor—than were leavers. We also obtained slightly stronger evidence that one’s own initial level of commitment predicted subsequent changes in perceptions of the partner’s commitment rather than that initial perceptions of partner commitment predicted subsequent changes in one’s own commitment. This, too, indirectly suggests that individuals shape their perceptions of the partner’s commitment in light of their own commitment (a biased process), more than they use what they believe to be accurate perceptions of the partner’s commitment in adjusting their own commitment. It is also possible that individuals vary in their propensity to perceive a partner as stably or not stably committed—that is, the fluctuation pattern may have some origins in individual differences. Individuals who were predisposed to be anxiously attached to the partner were more likely to have perceptions of the partner’s commitment that fluctuated over time (as stated in Hypothesis 5). This is consistent with research by Campbell et al. (2005), who showed that anxiously attached individuals (more than securely attached or avoidant individuals) were highly reactive to daily interactions with the partner, whereby their impressions of the partner were closely tied to daily levels of conflict. Similarly, fluctuation in perceived partner commitment may reflect a predisposition to be highly reactive to partner interactions, stemming from an anxious attachment style. However, the correlation between fluctuation and anxious attachment was moderate rather than high, suggesting that fluctuation in perceived partner commitment stems from more than only an anxious attachment style. It is unlikely that perceiving a partner as being stably or unstably committed reflects individual differences in response sets. If the 8
The main analyses did not differentiate between leavers and abandoned for several reasons. First, the pattern of results was the same with or without the distinction; that is, when compared against people whose relationships persisted, the differences between leavers and abandoned were relatively small and not significant. We collapsed these two groups to keep the results relatively straightforward. Second, the number of individuals in the abandoned group was too small from which to draw firm conclusions. Third, theoretically, there are competing hypotheses of how being more committed than the partner might affect one’s breakup status; it is possible that a less committed partner would be more likely to break up, but it is also possible that the more committed person would preemptively break up so as to avoid being hurt by the partner. Much remains to be learned about the implications of disparities between own and partner commitment—issues that were beyond the current goals.
1062
ARRIAGA, REED, GOODFRIEND, AND AGNEW
extent of fluctuation in perceived partner commitment simply reflected an individual tendency to respond to survey items in unstable ways (cf. Gable & Nezlek, 1998), we would have obtained evidence of an unstable response set across several variables—that is, all fluctuation variables would have been highly correlated. On the contrary, fluctuation in perceived partner commitment was not correlated with all fluctuation variables (e.g., fluctuation in attachment variables), but it was correlated with fluctuation in variables in which the pattern of fluctuation could be said to reflect doubt (e.g., fluctuation in own satisfaction). Thus, the theoretically coherent pattern of correlations was unlikely to be caused by an artifact of using self-report response scales. In a similar vein, it is unlikely that an individual’s pattern of fluctuation was caused by an individual’s deliberate effort to respond a particular way, a self-report bias that can undermine the validity of self-report measures. The change estimates examined in the current research were not themselves self-report variables; they were derived from prospectively observed patterns in each individual’s self-reported data. As such, they are distinct from retrospective recollections of change over time, which are susceptible to self-report bias (McFarland & Ross, 1987). If prospective reports of change over time (as obtained in this research) are similarly susceptible to self-report bias, they should reveal a data pattern similar to retrospective reports of change over time; these two methods when compared directly have yielded distinct data patterns, suggesting that prospective methods are less susceptible to self-report biases (Karney & Frye, 2002). Finally, we explored whether a stable pattern in perceived partner commitment might have its origins in another response bias, namely using the top or ceiling of the scale. Individuals who endorsed the highest possible level of perceived partner commitment time after time would have high and steady levels of perceived partner commitment, making it impossible to determine whether it is level or stability of perceived commitment that is driving the association with later breakup status. When we eliminated individuals who time after time consistently endorsed the highest possible rating of perceived partner commitment, the findings remained the same. Thus, a ceiling effect in ratings of perceived partner commitment did not account for the association between fluctuation in perceived partner commitment and breakup.
Broader Implications Are these findings compelling, or obvious and trivial? From a common sense standpoint, it could seem trivial to report that people who vacillate in their perception of their partner’s commitment are more likely to be in relationships that end than people who believe they know where their partner stands and thus vacillate less. The group whose perceptions vacillate experience more doubt, and all things being equal, it seems self-evident that doubt could hurt a relationship. This finding is less trivial when one considers what might have been the case but was not the case. First, the key variables predicting breakup among novel relationships were not the same as those among more established relationships. Yet, it is important to advance precise knowledge on the causal factors that operate at different relationship stages. When faced with a partner who seems to be decreasingly committed over time (a decreasing trend of perceived commitment), individuals may react differently depending on the stage of their
relationship. As was shown, unstable perceptions of a decreasingly committed partner (i.e., high fluctuation) disrupted relatively novel relationships (negative link between fluctuation in perceptions and persistence), suggesting it is better to end the relationship than persist in a state of uncertainty. Fluctuating perceptions were no different than stable perceptions among individuals in more established dating relationships (no link between fluctuation and persistence)—a decreasing trend of perceived partner commitment was associated with dissolution regardless of how stable the perceptions were. It is conceivable that among marital relationships or other relationships that are difficult to end, fluctuating perceptions of a decreasingly committed partner might be associated with persistence (a positive link between fluctuation and persistence); if one feels the relationship must continue no matter what, but the partner is decreasingly committed, perceiving an occasional commitment increase in an otherwise uncommitted partner may lead to hope that the partner’s commitment will improve. This remains to be examined in future research. Second, and more importantly, common sense would suggest that what should really matter in predicting a breakup is whether one perceives the partner to be committed or not—if a partner is not likely to be there through thick and thin, one would likely leave the relationship rather than remain vulnerable to being left by the partner. The common sense view would stop there. However, this view is overly simplistic given that the amount of fluctuation in perceptions of a partner’s commitment makes a difference as much or more than generally seeing a partner as committed or not committed (Surra et al., 1999). It is not self-evident that people who are increasingly convinced that their partner wants to leave (declining partner commitment) would be likely to stay in the relationship so long as their partner’s lack of commitment is predictable (i.e., steadily declining). Even less obvious is the finding that after taking into account changes in perceived partner commitment (the trend and extent of fluctuation), one’s initial level of commitment simply did not predict the fate of the relationship. Changes one sees in the partner’s level of commitment are more important than simply seeing the partner as being committed or not so committed. Third, common sense might also suggest that perceptions of the partner should matter only to people who are highly committed (high initial level of own commitment). Why should those who are relatively uncommitted care about the partner’s commitment? We did not find this to be the case, as one’s own initial level of commitment did not affect (i.e., moderate) the link between steady perceptions of partner commitment and persistence. The theoretical message of this research resonates with previous research on perceived partner regard (e.g., Murray et al., 2003): When a person has doubts about a partner’s feelings about the relationship— even when one generally believes that the partner is generally committed but doubts the consistency of the partner’s commitment—it becomes difficult to assume that all will turn out well, particularly when the relationship is not yet well established. More generally, perceptions of where the partner stands predict the course of a relationship, particularly in relatively new relationships (6 months or less). We are not suggesting that a partner’s actual level of commitment is irrelevant; instead we are underscoring the importance of perceptions and attributions in directing the course of relationships (Bradbury & Fincham, 1990; Kelley, 1979). Ultimately, the current research advances research on certainty and
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT
doubt in relationships by demonstrating the importance of invariances over time in perceptions of partner motives (see Holmes, 2004). There are also several methodological implications of this research. One is that multiple measurement occasions afford an analysis of specific changes over time better than fewer measurement occasions. Designs that attempt to establish a longitudinal pattern from only two measurement occasions are relegated to averaging (or correlating) reports over the two time periods (Arriaga, 2001; Karney & Bradbury, 1995). Yet as suggested in Figure 1, individuals can exhibit the same initial level of perceived partner commitment, the same mean level over time, and even similar linear trend over time, and yet have vastly different experiences of certainty versus doubt on the basis of the extent of fluctuation in their ratings (e.g., Profile 3 vs. Profile 4). As such, examining fluctuations in meaningful variables not only increases the pragmatic aim of predicting later relationship outcomes but also affords more precision in deriving theoretically meaningful data patterns. A second methodological implication in studies with multiple measurement occasions is that checks must be in place to ensure the results are not attributed to participant response sets. When participants answer different questions in the same way within a given measurement occasion, this will artificially inflate the correlations among variables. As Gable and Nezlek (1998) suggested, it is important to demonstrate divergent validity (as was the case in this research) by showing that an outcome is associated with fluctuation in theoretically relevant variables only and not all variables. Similarly, when participants answer certain questions in the same way across several measurement occasions (e.g., endorsing the highest possible rating every time), this too creates a response set that confounds level and variation over time in a variable. One solution is to eliminate these participants from the analysis, but this might eliminate a true data pattern. In the current research, we repeated the analyses excluding participants at the ceiling of the scale, which eliminated those who truly believed that their partner was consistently and whole-heartedly committed. Some studies have attempted to get around this problem by changing the scale endpoints (e.g., “My partner is more committed to this relationship than anyone in the world will ever be”), but even this strategy has not been successful in eliminating ceiling or floor effects (see Arriaga, 2001, for similarly extreme measures). The best approach might be to use scales with extreme endpoints and perform analyses on a full sample versus a sample excluding participants with possible response sets, to test whether the findings possibly reflect a response artifact. A third methodological issue concerns the lag between measurement occasions. This is less of a concern when examining the linear slope in that it summarizes changes over time, but more of a concern when attempting to capture specific patterns of change (e.g., ups and downs, curvilinear), in which long time lags may miss the relevant changes. In Study 2 (4-week time lag), the effect sizes were larger in magnitude for the linear trend in perceived partner commitment than for fluctuation in perceived partner commitment; in contrast, in Study 1 (1-week time lag) the effect sizes were similar or slightly larger for fluctuation.9 This suggests that 4-week time lags may gloss over dynamic changes in perceptions of the partner whereas 1-week time lags may be more appropriate for this particular type of variable. In the absence of clear guide-
1063
lines, the appropriate time lag to capture specific changes is likely to depend on the construct of interest; constructs that are more susceptible to change may require shorter time lags, and those that are less susceptible may be captured adequately with longer time lags.
Caveats The research reported here has some limitations. The most salient limitation is the focus on individual data patterns. We demonstrated that one can predict later relationship outcomes from examining an individual’s ratings over time. Ultimately, however, developing sound theories of relationship processes will require understanding partner influences that occur in addition to, and independently of, individual processes (Kenny, 2006). Another limitation was that both samples were derived from college students in the Midwestern region of the United States, and one cannot assume these findings generalize to samples of other college students, other young adults, relatively new dating relationships of older adults with longer relationship histories (e.g., those who are divorced), or samples of people from different cultures. Moreover, these are correlational data. Although we have attempted to control for artifacts and discriminate variables that do versus do not account for the link between perceived partner commitment and later breakup status, it is still possible that other variables not measured in this research are causing the observed changes in both perceived partner commitment and later breakup status. Finally, we established that different temporal profiles are predictive of later relationship status but did not provide conclusive evidence on the origins of distinct profiles. We encourage future research that attempts to overcome these limitations.
9
We examined the effect sizes in the subsample of Study 2 that was comparable with Study 1 (participants in relationships of similar duration, similar sample size, differing only in the time lag). When we regressed breakup onto initial level, linear trend, and fluctuation in perceived partner commitment simultaneously, the effect size for linear trend ( ⫽ .41) was larger than the effect size for fluctuation ( ⫽ ⫺.33); initial level was not significant. In analyses controlling for the three satisfaction variables, the effect of adding the linear trend in perceived partner commitment ( ⫽ .44) exceeded the effect of adding fluctuation ( ⫽ ⫺.22); adding initial level was not significant. The same occurred when controlling for the three own commitment variables (linear trend  ⫽ .43, fluctuation  ⫽ ⫺.25, initial level was not significant).
References Agnew, C. R., Loving, T. J., & Drigotas, S. M. (2001). Substituting the forest for the trees: Social networks and the prediction of romantic relationship state and fate. Journal of Personality and Social Psychology, 81, 1042–1057. Agnew, C. R., Van Lange, P. A. M., Rusbult, C. E., & Langston, C. A. (1998). Cognitive interdependence: Commitment and the mental representation of close relationships. Journal of Personality and Social Psychology, 74, 939 –954. Ajzen, I., & Fishbein, M. (1977). Attitude– behavior relations: A theoretical analysis and review of empirical research. Psychological Bulletin, 84, 888 –918. Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self
1064
ARRIAGA, REED, GOODFRIEND, AND AGNEW
scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63, 596 – 612. Aronson, J., & Inzlicht, M. (2004). The ups and downs of attributional ambiguity: Stereotype vulnerability and the academic self-knowledge of African American college students. Psychological Science, 15, 829 – 836. Arriaga, X. B. (2001). The ups and downs of dating: Fluctuations in satisfaction in newly formed romantic relationships. Journal of Personality and Social Psychology, 80, 754 –765. Berscheid, E. (1983). Emotion. In H. H. Kelley et al. (Eds.), Close relationships (pp. 110 –168). New York: Freeman. Bolger, N., Zuckerman, A., & Kessler, R. C. (2000). Invisible support and adjustment to stress. Journal of Personality and Social Psychology, 79, 953–961. Bradbury, T. N., & Fincham, F. D. (1990). Attributions in marriage: Review and critique. Psychological Bulletin, 107, 3–23. Brickman, P. (1987). Commitment, conflict, and caring. Englewood Cliffs, NJ: Prentice-Hall. Campbell, L., Simpson, J. A., Boldry, J., & Kashy, D. A. (2005). Perceptions of conflict and support in romantic relationships: The role of attachment anxiety. Journal of Personality and Social Psychology, 88, 510 –531. Crocker, J., Karpinski, A., Quinn, D. M., & Chase, S. K. (2003). When grades determine self-worth: Consequences of contingent self-worth for male and female engineering and psychology majors. Journal of Personality and Social Psychology, 85, 507–516. Drigotas, S. M., Rusbult, C. E., & Verette, J. (1999). Level of commitment, mutuality of commitment, and couple well-being. Personal Relationships, 6, 389 – 409. Driscoll, D. M., Hamilton, D. L., & Sorrentino, R. M. (1991). Uncertainty orientation and recall of person-descriptive information. Personality and Social Psychology Bulletin, 17, 494 –500. Etcheverry, P. E., & Agnew, C. R. (2004). Subjective norms and the prediction of romantic relationship state and fate. Personal Relationships, 11, 409 – 428. Fraley, R. C. (2002). Attachment stability from infancy to adulthood: Meta-analysis and dynamic modeling of developmental mechanisms. Personality and Social Psychology Review, 6, 123–151. Gable, S. L., & Nezlek, J. B. (1998). Level and instability of day-to-day psychological well-being and risk for depression. Journal of Personality and Social Psychology, 74, 129 –138. Gable, S. L., & Reis, H. T. (1999). Now and then, them and us, this and that: Studying relationships across time, partner, context, and person. Personal Relationships, 6, 415– 432. Holmes, J. G. (1981). The exchange process in close relationships: Microbehavior and macromotives. In M. L. Lerner & S. C. Lerner (Eds.), The justice motive in social behavior (pp. 261–284). New York: Plenum. Holmes, J. G. (2004). The benefits of abstract functional analysis in theory construction: The case of interdependence theory. Personality and Social Psychology Review, 8, 146 –155. Holmes, J. G., & Rempel, J. K. (1989). Trust in close relationships. In C. Hendrick (Ed.), Close relationships: Vol. 10. Review of personality and social psychology (pp. 187–220). Thousand Oaks, CA: Sage. Jacobson, N. S., Follette, W. C., & McDonald, D. W. (1982). Reactivity to positive and negative behavior in distressed and nondistressed married couples. Journal of Consulting and Clinical Psychology, 50, 706 –714. Karney, B. R., & Bradbury, T. N. (1995). Assessing longitudinal change in marriage: An introduction to the analysis of growth curves. Journal of Marriage and the Family, 57, 1091–1108. Karney, B. R., & Bradbury, T. N. (1997). Neuroticism, marital interaction, and the trajectory of marital satisfaction. Journal of Personality and Social Psychology, 72, 1075–1092.
Karney, B. R., & Bradbury, T. N. (2000). Attributions in marriage: State or trait? A growth curve analysis. Journal of Personality and Social Psychology, 78, 295–309. Karney, B. R., & Frye, N. E. (2002). “But we’ve been getting better lately”: Comparing prospective and retrospective views of relationship development. Journal of Personality and Social Psychology, 82, 222–238. Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and groups. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 451– 477). Cambridge, England: Cambridge University Press. Kelley, H. H. (1979). Personal relationships: Their structures and processes. Hillsdale, NJ: Erlbaum. Kelley, H. H. (1983). Love and commitment. In H. H. Kelley et al. (Eds.), Close relationships (pp. 265–314). New York: Freeman. Kenny, D. A. (2006). The partner (and the participant) in social psychology. Invited address presented at the Seventh Annual Meeting of the Society for Personality and Social Psychology, Palm Springs, CA. Le, B., & Agnew, C. R. (2003). Commitment and its theorized determinants: A meta-analysis of the investment model. Personal Relationships, 10, 37–57. Lehmiller, J. J., & Agnew, C. R. (2006). Marginalized relationships: The impact of social disapproval on romantic relationship commitment. Personality and Social Psychology Bulletin, 32, 40 –51. McFarland, C., & Ross, M. (1987). The relation between current impressions and memories of self and dating partners. Personality and Social Psychology Bulletin, 13, 228 –238. Miller, P. J. E., & Rempel, J. K. (2004). Trust and partner-enhancing attributions in close relationships. Personality and Social Psychology Bulletin, 30, 695–705. Murray, S. L. (1999). The quest for conviction: Motivated cognition in romantic relationships. Psychological Inquiry, 10, 23–34. Murray, S. L. (2005). Regulating the risks of closeness: A relationshipspecific sense of felt security. Current Directions in Psychological Science, 14, 74 –78. Murray, S. L., Bellavia, G., Rose, P., & Griffin, D. W. (2003). Once hurt, twice hurtful: How perceived regard regulates daily marital interactions. Journal of Personality and Social Psychology, 84, 126 –147. Murray, S. L., & Holmes, J. G. (1999). The (mental) ties that bind: Cognitive structures that predict relationship resilience. Journal of Personality and Social Psychology, 77, 1228 –1244. Murray, S. L., Holmes, J. G., & Collins, N. L. (2006). Optimizing assurance: The risk regulation system in relationships. Psychological Bulletin, 132, 641– 666. Murray, S. L., Holmes, J. G., & Griffin, D. W. (2000). Self-esteem and the quest for felt security: How perceived regard regulates attachment processes. Journal of Personality and Social Psychology, 78, 478 – 498. Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Trust in close relationships. Journal of Personality and Social Psychology, 49, 95–112. Robins, R. W., Hendin, H. M., & Trzesniewski, K. H. (2001). Measuring global self-esteem: Construct validation of a single-item measure and the Rosenberg Self-Esteem Scale. Personality and Social Psychology Bulletin, 27, 151–161. Rogosa, D. (1981). A critique of cross-lagged correlation. Psychological Bulletin, 88, 245–255. Rogosa, D., Brant, D., & Zimowksi, M. (1982). A growth curve approach to the measurement of change. Psychological Bulletin, 92, 726 –748. Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and deterioration) of satisfaction and commitment in heterosexual involvements. Journal of Personality and Social Psychology, 45, 101–117. Rusbult, C. E., Martz, J. M., & Agnew, C. R. (1998). The Investment Model Scale: Measuring commitment level, satisfaction level, quality of alternatives, and investment size. Personal Relationships, 5, 357–391.
FLUCTUATIONS IN PERCEIVED PARTNER COMMITMENT Rusbult, C. E., Olsen, N., Davis, J. L., & Hannon, P. A. (2001). Commitment and relationship maintenance mechanisms. In J. Harvey & A. Wenzel (Eds.), Close romantic relationships: Maintenance and enhancement (pp. 87–113). Mahwah, NJ: Erlbaum. Sanbonmatsu, D. M., Posavac, S. S., Vanous, S., & Ho, E. A. (2005). Information search in the testing of quantified hypotheses: How “all,” “most,” “some,” “few,” and “none” hypotheses are tested. Personality and Social Psychology Bulletin, 31, 254 –266. SAS Institute. (1992). SAS Technical Report P-229, SAS/STAT Software: Changes and Enhancements. Cary, NC: Author. Shoda, Y. (1999). Behavioral expression of a personality system: Generation and perception of behavioral signatures. In D. Cervone & Y. Shoda (Eds.), The coherence of personality: Social-cognitive bases of consistency, variability, and organization (pp. 155–181). New York: Guildford Press. Shoda, Y., Mischel, W., & Wright, J. C. (1994). Intraindividual stability in the organization and patterning of behavior: Incorporating psychological situations into the idiographic analysis of personality. Journal of Personality and Social Psychology, 67, 674 – 687. Simpson, J. A., & Rholes, W. S. (Eds.). (1998). Attachment theory and close relationships. New York: Guilford Press. Simpson, J. A., Rholes, W. S., & Phillips, D. (1996). Conflict in close
1065
relationships: An attachment perspective. Journal of Personality and Social Psychology, 71, 899 –914. Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics, 24, 323–355. Sorrentino, R. M., Holmes, J. G., Hanna, S. E., & Sharp, A. (1995). Uncertainty orientation and trust in close relationships: Individual differences in cognitive styles. Journal of Personality and Social Psychology, 68, 314 –327. Surra, C. A., & Hughes, D. K. (1997). Commitment processes in accounts of the development of premarital relationships. Journal of Marriage and the Family, 59, 5–21. Surra, C. A., Hughes, D. K., & Jacquet, S. E. (1999). The development of commitment to marriage: A phenomenological approach. In J. M. Adams & W. H. Jones (Eds.), Handbook of interpersonal commitment and relationship stability (pp. 125–148). New York: Kluwer Academic/ Plenum Publishers. Wieselquist, J., Rusbult, C. E., Foster, C. A., & Agnew, C. R. (1999). Commitment, pro-relationship behavior, and trust in close relationships. Journal of Personality and Social Psychology, 77, 942–966.
Appendix Using SAS to Analyze Within-Person Variation in a Single Variable SAS provides a way of analyzing (Level 2) group differences in the magnitude of within-person variation (rij) in a Level 1 variable while controlling for the intercept and slope. We wanted to assess whether the two breakup groups (ended vs. persisted) differed in the amount of withinperson fluctuation over time. This can be achieved by comparing two models, one in which the two breakup groups are constrained to have the same average within-person variation versus one in which the average within-person variation can vary between the two breakup groups (Niall Bolger, personal communication, June 4, 2004). As an example, one would examine an initial model in which the within-person variation in perceived partner commitment is constrained to be equal for individuals in ended and persisted relationships by using the following SAS code, where id is a subject identifier, time is the time period, ppc is perceived partner commitment measured at each time, and breakup is a two-level variable for the ended versus persisted groups: proc mixed method ⫽ ml; class id breakup; model ppc ⫽ time breakup time*breakup; random intercept time/subject ⫽ id type ⫽ un grp ⫽ breakup; run; The following code would test a second model in which the magnitude of within-person variation for one breakup group (ended) could vary from that of the other breakup group (persisted):
proc mixed method ⫽ ml; class id breakup; model ppc ⫽ time breakup time*breakup; random intercept time/subject ⫽ id type ⫽ un grp ⫽ breakup; repeated/subject ⫽ id grp ⫽ breakup; run; The repeated statement elicits an analysis of within-person variance for each participant (subject ⫽ id) and compares the two breakup groups (grp ⫽ breakup) in the extent of within-person variance. Each model would yield a chi-square value; the two chi-square values are compared with a chi-square difference test. A significant reduction in the chi-square value would indicate that the two groups differed in the amount of fluctuation (i.e., within-person variation). Also, in both models, the grp ⫽ breakup command in the random statement allows Level 2 residuals of the random effects (00 and 11 in the Tau or random effects covariance matrix) to vary by breakup group; this translates into a more conservative test of group differences in within-person variation, as tested in the second model. As stated in the text, the limitation of this model comparison approach is that one can compare within-person variation in only one variable, without simultaneously controlling for or analyzing within-person variation in other Level 1 variables.
Received November 22, 2005 Revision received May 11, 2006 Accepted May 14, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1066 –1079
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1066
Perceiving Outgroup Members as Unresponsive: Implications for Approach-Related Emotions, Intentions, and Behavior David A. Butz and E. Ashby Plant Florida State University In 2 studies, the authors investigated the determinants of anger and approach-related intentions and behavior toward outgroup members in interracial interactions. In Study 1, White and Black participants who were led to believe that their interracial interaction partner was not open to an upcoming interaction reported heightened anger and approach-related intentions concerning the interaction, including viewing their partner as hostile, intending to ask sensitive race-relevant questions during the interaction, and planning to blame the partner if the interaction went poorly. Results of Study 2 showed that White participants who received negative feedback about their Black partner’s openness to interracial interactions behaved in a hostile manner toward their interaction partner. The findings are discussed in terms of their implications for the quality of interracial interactions. Keywords: expectations, outgroup members, interracial interactions, intergroup emotions
The increasing racial and ethnic diversity of the United States indicates that people may have greater opportunity for interracial contact (Kent & Mather, 2002). Although such trends are encouraging, they do not ensure that people will engage in more interracial contact or that interracial interactions will be positive. Indeed, interracial interactions provoke anxiety and avoidance for many people (e.g., Gudykunst, 1993; Plant & Devine, 2003; Stephan & Stephan, 1985, 1989). Identifying those who are likely to be anxious and to avoid interracial interactions has proven important to understanding the sources of negativity in intergroup relations and has received considerable empirical attention (e.g., Plant, 2004; Plant & Devine, 2003; Stephan & Stephan, 1985, 1989). Comparatively little research has examined the factors that determine people’s negative approach-related responses to interracial interactions (but see Mackie, Devos, & Smith, 2000). We contend that, in addition to understanding the factors that result in anxiety and the avoidance of interracial interactions, it is important to understand the determinants of anger and hostility in interracial interactions. The present research addresses this gap by experimentally examining whether expectations centered on the openness of outgroup members to interracial interactions determine anger and approach-related behavioral reactions but not anxiety and avoidance-related responses to outgroup members.
Approaching Versus Avoiding Interracial Interactions Emerging theoretical and empirical work indicates that anger and anxiety are distinct emotions that may have differential implications for interactions with outgroup members (e.g., Cottrell & Neuberg, 2005; Mackie, Devos, & Smith, 2000; Smith, 1993). In general, anger and anxiety both occur in response to aversive stimuli but differ in that anger tends to be associated with approach-related tendencies and anxiety tends to be associated with avoidance-related tendencies (e.g., Frijda, Kuipers, & ter Schure, 1989; Harmon-Jones, 2003). Accordingly, anger in intergroup contexts is linked to taking action against outgroup members (Mackie et al., 2000), but anxiety is associated with avoiding intergroup contact (Cottrell & Neuberg, 2005; Plant & Butz, 2006). We propose that these distinct patterns of approach and avoidance-related emotions in intergroup contexts may arise from people’s expectations about the outcomes of interracial interactions. Previous work indicates that people who anticipate negative outcomes in interracial interactions tend to both anticipate and experience more negative affective reactions (e.g., Britt, Boniecki, Vescio, Biernat, & Brown, 1996; Devine, Evett, & VasquezSuson, 1996; Plant & Butz, 2006; Plant & Devine, 2003). In examining the sources of intergroup anxiety and anger, it may, therefore, be important to consider the different reasons that people expect negative outcomes in interracial interactions. For example, some people may expect interracial interactions to be awkward because they expect to convey a negative, racially biased impression in interactions (e.g., Plant & Butz, 2006; Shelton, 2003). We hypothesized that such negative efficacy expectations result in anxiety and the avoidance of interracial interactions. However, people may also expect interracial interactions to be negative because they perceive that outgroup members are not open to interacting with them (Plant & Devine, 2003), which we posit may instead result in anger and hostile behavior. In the current work, we examine the differential implications of these two distinct types of expectations for interracial interactions and test the hypothesis
David A. Butz and E. Ashby Plant, Department of Psychology, Florida State University. This research was supported by a National Science Foundation Graduate Research Fellowship Award given to David A. Butz. We would like to thank Matthew Gailliot, Celeste Doerr, and Jonathan Kunstman for helpful comments on earlier versions of this article. Correspondence concerning this article should be addressed to E. Ashby Plant, Department of Psychology, Florida State University, Tallahassee, FL 32306. E-mail:
[email protected] 1066
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
that negative expectations centered on the perceived openness of outgroup members determine approach-related reactions such as anger, blame, and antisocial behavior relevant to interracial interactions. In this article, we consider the theoretical and empirical basis for these predictions. In seminal work, Bandura (1977, 1982) distinguished between expectations that are centered on people’s ability to perform behaviors to reach a desired outcome (self-efficacy expectancies) and expectations that are centered on the effectiveness of one’s behaviors given a set of external factors (response expectancies, which have also been referred to in the literature as outcome expectancies). Both self-efficacy and response expectancies are important determinants of people’s emotional and behavioral reactions in a given situation, although self-efficacy has received greater empirical attention than response expectancies (e.g., Maddux, Norton, & Leary, 1988). For example, Bandura and colleagues (e.g., Bandura, 1977; Bandura, Adams, Hardy, & Howells, 1980) showed that across many domains, people with negative self-efficacy expectancies tend to exhibit fear-related reactions and avoidant behaviors (e.g., less perseverance on a task). Similarly, when people doubt their ability to convey positive social impressions, anxiety and the avoidance of social interactions may ensue (e.g., Leary & Atherton, 1986; Schlenker & Leary, 1982). In contrast, Bandura’s (1977, 1982) theorizing indicates that when people expect their goals to be blocked because of situational or external events (e.g., negative response expectancies), they may respond with frustration, feelings of futility, efforts to change the situation, or avoidance. In social interactions, the negative response expectancies may result in anger and other-directed blame (e.g., Leary & Atherton, 1986). Further supporting this idea is wide empirical evidence indicating that threats to the self (e.g., Baumeister, Smart, & Boden, 1996) and perceived social rejection (e.g., Twenge, Baumeister, Tice, & Stucke, 2001) tend to result in antisocial behavior. On the basis of this previous work, we proposed that negative response expectancies regarding interracial interactions result in a host of negative approach-related reactions such as anger and other-directed blame, as well as in antisocial behavior. We thought that these distinct types of expectations, which elicit specific patterns of approach and avoidance-related responses in interpersonal contexts, may elicit similar patterns of responses in intergroup contexts. Specifically, if people expect that their efforts to convey a positive impression to outgroup members will be inefficacious, they will experience increased anxiety and heightened avoidant inclinations. In contrast, if people’s negative expectations center on the perceived unresponsiveness of outgroup members, they will experience heightened anger and antisocial behavior directed at outgroup members. Moreover, we argue that added concerns for interracial interactions regarding whether one is being perceived by outgroup members as biased render such types of expectations particularly relevant to interracial encounters and may result in stronger responses to negative expectations. Consistent with the hypothesis that negative expectancies may have stronger implications for interracial than for intraracial interactions, F. E. Frey and Tropp (2006) argued that although people’s perceptions of what others think of them pertain to all types of social interactions, intergroup interactions differ because they include the possibility of being viewed by outgroup members in terms of group membership. Expecting to be viewed on the basis
1067
of one’s racial group can lead people to anticipate race-related rejection or intolerance by outgroup members (see also Krueger, 1996; Mendoza-Denton, Downey, Purdie, Davis, & Pietrzak, 2002; Shelton & Richeson, 2005). Shelton and Richeson (2005) recently found that both White and Black people indicate that they would avoid interracial interactions because of perceived rejection by outgroup members but that outgroup members would likely avoid interracial interactions because they are not interested in or open to such interactions. Further, Vorauer and colleagues (Vorauer & Kumhyr, 2001; Vorauer, Main, & O’Connell, 1998) showed that when people expect outgroup members to view them according to their group membership, they expect be viewed in a negative, stereotypical manner (e.g., “racist,” “snobby”). As a result, they anticipate more negative emotional responses to intergroup interactions, and their behavior may evoke discomfort and negative affect in outgroup members (e.g., Vorauer & Kumhyr, 2001). As a direct test of our hypothesis that response expectancies determine approach-related reactions to interracial versus samerace interactions, we conducted a preliminary study in which 116 White introductory psychology students were led to believe that they would have either an interracial interaction or a same-race interaction. They were then provided with positive, negative, or no response expectancy feedback regarding their partner’s openness to the upcoming interaction, after which they reported their response expectancies and anger regarding the upcoming interaction.1 Results revealed that the response expectancy manipulation had the intended effect for both same-race and interracial interactions such that participants who received the negative response expectancy feedback predicted that their partner would be less open to the interaction than participants who received positive or no response expectancy feedback ( ps ⬍ .01). Consistent with the hypothesis that response expectancies determine approach-related inclinations for interracial interactions in particular, the response expectancy feedback was predictive of participants’ anger for an interracial interaction ( p ⬍ .01) but not for a same-race interaction ( p ⫽ .53). Moreover, among participants who received negative response expectancy feedback, those anticipating an interracial interaction reported relatively more anger than those anticipating a same-race interaction ( p ⬍ .05). The results from this preliminary study provide initial evidence that response expectancies underlie people’s approach-related responses to interracial interactions in particular.
The Current Work In the current work, we examined whether negative expectations about the openness of outgroup members to interracial interactions (i.e., negative response expectancies) resulted in specifically approach-related reactions such as anger, other-focused blame, and hostile behavior. Across two studies using different methodologies, we manipulated participants’ response expectancies for an anticipated interaction by providing them with information regarding their interracial–interaction partner’s openness to the interaction. Both studies examined the implications of the response ex1
More detailed information regarding the methodology and results of the preliminary study are available on request.
BUTZ AND PLANT
1068
pectancy feedback for participants’ responses to the interaction. Participants’ self-reported efficacy was also assessed in the current studies before an anticipated interaction to examine the independent effects of response expectancies and self-efficacy expectancies. We predicted that, consistent with previous findings, selfefficacy expectancies would be specifically associated with avoidant-related emotional reactions (i.e., anxiety) and an increased desire to avoid the interaction (e.g., Plant & Butz, 2006).
Study 1 The primary goal of the first study was to examine the influence of negative response expectancies for participants’ emotional reactions (i.e., anxiety and anger) and their approach- and avoidancerelated intentions for interracial interactions. Participants’ response expectancies regarding an upcoming interracial interaction were manipulated through feedback related to their partner’s openness to the interracial interaction. We predicted that the negative response expectancy feedback would have specifically approachrelated consequences, which is supported by theory suggesting that interaction partners who are not open to interactions represent an external source of potential tension in the interaction (e.g., Leary & Atherton, 1986). People’s emotional and behavioral reactions to unresponsive outgroup members should reflect the externally based nature of their response expectancy—anger and attributions of blame directed outward and toward the perceived source of tension or awkwardness in the interaction. Accordingly, if people’s inclinations toward unresponsive outgroup members are distinctly approach-oriented, they should experience anger as opposed to anxiety, which is a functionally different type of intergroup emotion associated with avoiding or retreating from intergroup contact (e.g., Cottrell & Neuberg, 2005; Mackie et al., 2000). People’s emotional reactions to negative response expectancies may, as a result, draw them into intergroup conflict. To explore participants’ approach-related behavioral intentions for interracial interactions, we asked participants to indicate their interest in approaching racial issues in the upcoming interaction. Previous work has shown that when people possess negative expectancies related to a potential interaction partner, they are more likely to probe the source of the negative expectancy by selecting a greater number of questions that are relevant to the negative expectancy than questions that are irrelevant to the negative expectancy (e.g., Darley, Fleming, Hilton, & Swann, 1988). Participants in the current study were provided with a list of questions, some of which were related to race and diversity, and were instructed to select those questions that they were interested in discussing with their interaction partner. We predicted that if negative response expectancies elicited approach-related reactions intended to probe the source of their negative expectancy, participants in the negative response expectancy feedback condition would select more race-relevant discussion questions than would participants in the other feedback conditions. We also anticipated that negative response expectancies would shape people’s initial impressions of their interaction partner and potentially set the stage for biased information processing. People who expect that outgroup members will not be open to them in the interaction may interpret the demeanor and behavior of outgroup members through a lens tinted by these initially negative expectations (i.e., perceive that their outgroup-member interaction part-
ner is a hostile and angry person). Although prior research has consistently shown that negative expectations bias the impressions formed of others (Hamilton & Sherman, 1996; Hilton & Darley, 1985), we argue that response expectancies in particular influence the dispositional inferences formed of outgroup members. In contrast to the approach-related consequences resulting from negative response expectancies, we anticipated that self-efficacy expectancies would be distinctly associated with heightened anxiety and avoidant intentions regarding the upcoming interaction but not with increased anger and approach-related intentions. To examine these predictions, we assessed participants’ self-reported efficacy and response expectancies before the anticipated interracial interaction. We anticipated that people who possessed negative self-efficacy expectancies would tend to report negative response expectancies and vice versa because both reflect a negative outlook for the interaction. Therefore, we explored the impact of our manipulation on each type of expectancy independent of the other type of expectancy by statistically controlling for the other type of expectancy and thereby accounting for the shared variance between the two expectancies. Similarly, people who are highly anxious about an interaction are also likely to report other negative emotions including anger, and those who are angry are also likely to report anxiety. Our interest focused on the influence of the types of expectancies on the emotional responses independent of each other (i.e., which was the primary emotional response). Therefore, we examined the factors that predict these emotional reactions after controlling for the conceptually distinct emotional reactions (see Devine et al., 1996, Studies 1 and 3). To ascertain whether the theoretical predictions outlined in the current work apply to both majority and minority group members’ concerns regarding interracial interactions, we examined in this study the implications of response expectancies and self-efficacy expectancies for both White and Black participants anticipating an interracial interaction. On the basis of previous work, we predicted that response expectancies would have a similar influence on Black participants’ reactions to interactions with White people as they do for White people’s interactions with Black people. For example, a growing body of research suggests that some Black people perceive White people as not open to interracial interactions and are concerned that Whites will be prejudiced or biased in such interactions (Branscombe, Schmitt, & Harvey, 1999; Mendoza-Denton al., 2002; Monteith & Spicer, 2000; Plant, 2004; Shelton, 2003; Shelton & Richeson, 2005). Such concerns tend to be associated with negative affective reactions to interactions with Whites (e.g., Plant, 2004) and in some cases, hostility toward Whites (e.g., Branscombe et al., 1999). Therefore, we hypothesized that Black people’s expectations about the openness of White people to interracial interactions (i.e., response expectancies) would also predict anger and approach-related intentions. It was less clear, however, whether self-efficacy expectancies would differentially predict anxiety and avoidance for White and Black participants. Shelton (2003) asserted that particular types of self-efficacy concerns (e.g., concerns about appearing prejudiced to one’s interaction partner) are primarily experienced by majoritygroup members in interactions with minority-group members, which suggests that self-efficacy concerns and anxious responses may be particularly salient to White people in their interactions with Black people. However, Bandura’s (1977, 1982) theorizing maintains that more generalized self-efficacy concerns may be
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
relevant to all social encounters in which people perceive that they will not convey a positive impression. Moreover, Hyers and Swim (1998) showed that White and Black perceivers reported similar levels of anxiety regarding interracial interactions.
Method Participants and Design Participants were 111 introductory psychology students (82 women, 29 men). Sixty-four participants identified their ethnicity as White/European American and 47 participants identified their ethnicity as Black/African American. An additional 5 participants distributed across the conditions completed the experiment, but their data were not included in the analyses because they expressed suspicion that they would not be having an interaction. The design of the study was a 2 (race of participant: Black vs. White) ⫻ 3 (response expectancy feedback condition: positive vs. negative vs. no feedback) between-participants factorial design.
Procedure Participants came into the lab and were told that they would be engaging in a same-sex interracial interaction. Black participants were always told that they would be interacting with a White person, and White participants were always told that they would be interacting with a Black person. All participants then completed a questionnaire that assessed their openness to participating in an interracial interaction, which was presented to participants as a brief measure of their preinteraction expectancies. This questionnaire included items such as “How much would you enjoy talking to a person of a different race?” The items were rated on a scale of 1 (not at all) to 7 (very much). Participants in the feedback conditions next took part in a rigged drawing that led them to believe that they had been randomly selected to receive information from their partner via his or her responses on the questionnaire. They were instructed that their partner would not be able to view their questionnaire. After ostensibly retrieving the questionnaire from the interaction partner’s lab room down the hall, the experimenter handed the questionnaire to the participant. To keep experimenters blind to the condition, we kept the questionnaire in a folder, and we had instructed experimenters not to look at the questionnaire. Participants were told to take a minute to read over their partner’s responses. Participants in the control condition were led to believe that they were completing a measure of preinteraction expectancies and were given no expectation that they would receive their partner’s questionnaire or that their partner would view their questionnaire. Participants in the positive response expectancy condition received a questionnaire completed in a manner suggesting that the other person anticipated a positive interaction and would be open to the interaction. The items assessing the estimated quality of an interracial interaction were all circled with the most positive response, implying that the interaction partner would enjoy talking to and working with their partner and expected a very positive interaction. Participants in the negative response expectancy condition received a questionnaire completed in a manner suggesting that the other person anticipated a negative interaction and would not be open to the interaction. The key items were all circled with the second-most negative response, implying that the other person would not enjoy talking to or working with their partner and overall expected a negative interaction. The experimenter then informed the participants that in order to investigate first impressions prior to meeting, participants would be photographed and would receive a photograph of their interaction partner. White participants always received a photograph of a same-sex Black person, whereas Black participants always received a photograph of a same-sex White person. All photographs of the interaction partners had been obtained before the experimental sessions and were posed such that the faces in the photos had neutral, emotionless expressions. Participants were asked
1069
to evaluate their partner on a series of traits, some of which were related to hostility. Next, participants completed a modified version of the Social Interaction Questionnaire (SIQ; Plant & Devine, 2003) that assessed their self-efficacy and response expectancies about the upcoming interaction, desire to avoid the interaction, and tendency to blame the other person for a negative interaction. They also completed a questionnaire asking about their emotional reactions to the upcoming interaction. After the participants completed this packet, the experimenter gave each of them a list of 18 questions, requesting that the participant select questions that he or she would like to ask his or her partner during the interaction. Participants were not limited to the number of questions they could select. Embedded in this list of questions were three questions relevant to race and racial issues (“Describe a time that you had a positive interaction with a person of a different race,” “Describe some advantages of being a member of a minority group at this university,” and “Describe some disadvantages of being a member of a minority group at this university”). Participants were instructed to place a check mark next to those questions that they were interested in asking their partner. While the participant selected questions, the experimenter excused him- or herself under the guise of checking to see if the other person was ready for the interaction. On the experimenter’s return, the participant was informed that the experiment had concluded; the participant then was probed for suspicion, fully debriefed, given credit, and excused.
Measures Expectations about the interaction. As a check of the response expectancy manipulation, participants rated their expectations regarding their partner’s openness to the interaction on a scale from 1 (strongly disagree) to 7 (strongly agree). Six items were reverse-coded when necessary and averaged to form an index of response expectancies such that higher scores indicated more negative response expectancies (␣ ⫽ .89). Response expectancy items included “I think my interaction partner is open to interacting with me” and “I am concerned that my partner will not like me.” To assess self-efficacy regarding the interaction, we had the participants respond to five items including “I feel that I don’t have the skills to have a positive interaction” and “I think that I am capable of having a pleasant interaction” (␣ ⫽ .71). Items were reverse-coded when necessary such that higher scores on the self-efficacy scale indicated more negative selfefficacy expectancies. Photograph ratings. After viewing the photograph of their partner, participants rated it on a series of traits. Of particular interest were the 12 traits related to hostility (e.g., frustrated, hostile, annoyed, friendly), which were reverse-coded when necessary and averaged to create a hostility index with higher numbers indicating greater perceived hostility (␣ ⫽ .91). Emotional reactions. Participants indicated their emotional reactions to the upcoming interaction by responding to a series of emotion descriptors using a 1 (does not apply at all) to 7 (applies very much) scale. To form an index of anxiety, we averaged seven anxiety-related emotions (e.g., anxious, tense, worried; ␣ ⫽ .84) with higher numbers indicating greater anxiety. We formed an anger index by averaging four items that were related to angry emotions (e.g., angry, hostile, agitated, frustrated) and three items that were specific to anger and frustration about the upcoming interaction (e.g., “I am frustrated that I have to participate in this interaction”). Higher numbers indicated greater anger about the interaction (␣ ⫽ .79). Avoidance and blame. To assess participants’ desire to avoid the interaction, we averaged three items on the SIQ to form an index of participants’ desire to avoid the interaction (e.g., “I wish I could avoid having this interaction;” ␣ ⫽ .92). Finally, two items assessed participants’ tendency to blame their partner for tension or awkwardness in the interaction (e.g., “If this interaction doesn’t go well, it will be my partner’s fault” and “I will blame my partner if our interaction is unpleasant;” ␣ ⫽
BUTZ AND PLANT
1070
F(1, 102) ⫽ 8.59, p ⬍ .01, 2 ⫽ .04, the response expectancy feedback was still a highly significant predictor of participants’ self-reported response expectancies (see Table 2). The response expectancy feedback also predicted Black and White participants’ self-reported efficacy expectancies, although the effect size was comparatively smaller, F(2, 103) ⫽ 4.96, p ⬍ .01, 2 ⫽ .09. Participants in the negative response expectancy feedback condition reported more negative self-efficacy expectancies (M ⫽ 3.25, SD ⫽ 1.12) than participants in the positive feedback condition (M ⫽ 2.43, SD ⫽ 1.14; p ⬍ .01) and marginally more negative efficacy expectancies than participants in the no feedback condition (M ⫽ 2.79, SD ⫽ 0.87; p ⫽ .07). Participants’ efficacy expectancies did not differ between the positive and no feedback conditions, p ⫽ .16. However, after accounting for participants’ self-reported response expectancies, the feedback did not predict efficacy expectancies (F ⬍ 1, p ⬎ .90). This indicates that the response expectancy feedback had the intended effect of manipulating participants’ response expectancies but not their efficacy expectancies for the upcoming interaction. Across the analyses, the race of participant did not predict participants’ expectancies (Fs ⬍ 1), nor did race of participant interact with the response expectancy feedback, (Fs ⬍ 1). It is also worth noting that in the no feedback control condition, Black and White participants reported similar levels of response expectancies and efficacy expectancies (Fs ⬍ 1) indicating that in the absence of experimental manipulation, Black and White participants felt similarly about the upcoming interracial interaction.
.81). These two items were averaged such that higher numbers indicate greater other-focused blame. Tendency to approach racial issues. Of the questions that participants could select to ask their partner during the interaction, 15 were unrelated to race or racial issues, whereas 3 specifically targeted potentially sensitive racial issues. Therefore, we tallied the number of the 3 possible racerelevant and 15 possible race-irrelevant questions that participants selected to discuss with their partner.
Results Initial Analyses Intercorrelations between all measures are presented in Table 1. To examine the effect of the response expectancy manipulation on participants’ self-reported efficacy and response expectancies, we conducted 2 (race of participant: Black vs. White) ⫻ 3 (response expectancy feedback condition: positive vs. negative vs. no feedback) analyses of variance (ANOVAs) on response and selfefficacy expectancies. To examine the impact of the feedback on the expectancies after controlling for the shared variance between the types of expectancies, we followed up these initial analyses with parallel analyses of covariance (ANCOVAs) in which we had controlled for the other type of expectancy (e.g., analysis of efficacy controlled for response expectancies). When the analyses revealed a significant effect of response expectancy feedback, Bonferroni tests were conducted to identify significant differences between the conditions. Consistent with the intentions of the manipulation, the response expectancy feedback had a significant effect on response expectancies, F(2, 103) ⫽ 48.73, p ⬍ .001, 2 ⫽ .47, such that participants in the negative response expectancy feedback condition reported more negative response expectancies (M ⫽ 3.91, SD ⫽ 1.12) than participants in the positive feedback condition (M ⫽ 1.60, SD ⫽ .74) and participants in the no feedback condition (M ⫽ 2.32, SD ⫽ 1.09; ps ⬍ .01). Participants in the positive feedback condition also reported more positive response expectancies than participants in the no feedback condition ( p ⬍ .01). In addition, the results of the ANCOVA revealed that after the significant effect of the self-efficacy expectancies was controlled,
Overview of Analyses To isolate the effects of the response expectancy feedback from participants’ self-efficacy expectancies, we categorized participants as high or low in self-efficacy on the basis of a median split of their self-reported efficacy expectancies about the upcoming interaction (Mdn ⫽ 3.00). All dependent measures were then submitted to a 2 (race of participant: Black vs. White) ⫻ 3 (response expectancy feedback condition: positive vs. negative vs.
Table 1 Intercorrelations Between Measures for Study 1 Measures
1
2
3
4
5
6
— .16
—
.06 .09 ⫺.13
⫺.05 .08 .02
7
8
9
— .64** .54**
— .39**
—
Approach-Related 1. 2. 3. 4. 5. 6.
Response expectancies Hostile evaluations Anger Other-focused blame Race-irrelevant questions Race-relevant questions
— .52** .49** .53** ⫺.05 .36**
— .33** .28** ⫺.02 .23*
— .42** .02 .07
— ⫺.13 .16
Avoidance-Related 7. Efficacy expectancies 8. Anxiety 9. Desire to avoid * p ⬍ .05.
** p ⬍ .01.
.40** .39** .44*
.25** .35** .21*
.51** .48** .67**
.19* .21* .36**
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
1071
Table 2 Measures as a Function of Response Expectancy Feedback: Study 1 Feedback Positive Measures
M
None SD
M
Negative SD
M
SD
F
3.79 4.46 2.03 2.81 2.23 1.54
1.02c 0.68b 0.76b 1.66c 1.68a 1.19b
39.71 19.29 3.69 11.67 1.05 4.26
1.30a 1.12a 1.78a
0.09 0.48 2.53
df
p⬍
2
102 96a 96 97 97 97
.001 .001 .03 .001 .35 .02
.36 .23 .05 .19 .02 .07
.92 .62 .09
.00 .01 .04
Approach-Related Response expectancies Hostile evaluations Anger Other-focused blame Race-irrelevant questions Race-relevant questions
1.70 3.12 1.56 1.42 2.78 .86
1.01a 1.05a 0.73a 0.63a 2.07a 0.88a
2.35 3.59 1.70 1.96 2.82 1.11
0.99b 0.95a 0.75a,b 1.43a,b 2.68a 1.02a,b
2, 2, 2, 2, 2, 2,
Avoidance-Related Efficacy expectancies Anxiety Desire to avoid
2.77 3.06 2.10
1.22a 1.09a 1.83a
2.87 2.98 2.32
1.05a 1.03a 1.56a
2.88 3.22 3.19
2, 102 2, 96 2, 97
Note. Expectancy analyses control for the other type of expectancy, and emotion analyses control for other type of emotion. Means with unique subscripts differ at p ⬍ .05 on Bonferroni tests. Means collapse across race of participant. a Different degrees of freedom reflect missing data from one participant.
no feedback) ⫻ 2 (self-efficacy: positive vs. negative) ANOVA.2 Unless explicitly mentioned, there were no significant effects of participant gender. Below, we first present the findings for the approach-related responses and then the findings for the avoidance-related responses.
Approach-Related Responses Hostile evaluations of interaction partner’s photograph. The analysis of participants’ hostile evaluations of their partner’s photograph revealed the predicted effect of the response expectancy feedback (see Table 2).3 Consistent with predictions, participants who received the negative response expectancy feedback viewed their partner as more hostile in the photograph than did participants in the positive and no feedback conditions. Participants’ perceived hostility did not significantly differ between the positive and no feedback conditions. There was also an unexpected Race of Participant ⫻ Efficacy interaction, F(1, 96) ⫽ 14.36, p ⬍ .001, 2 ⫽ .09. White participants perceived their partner as more hostile when they had negative efficacy (M ⫽ 4.11, SD ⫽ 0.74) compared with when they had positive efficacy (M ⫽ 3.34, SD ⫽ 1.04), F(1, 61) ⫽ 11.49, p ⬍ .01, 2 ⫽ .16. However, Black participants perceived their partner as less hostile when they had negative efficacy (M ⫽ 3.35, SD ⫽ 1.19) compared with when they had positive efficacy expectancies (M ⫽ 4.20, SD ⫽ 1.05), F(1, 43) ⫽ 6.45, p ⬍ .02, 2 ⫽ .13. The analysis also revealed an unexpected main effect of participant gender such that women rated their partner more negatively (M ⫽ 3.86, SD ⫽ 1.02) than men did, (M ⫽ 3.39, SD ⫽ 1.11), F(1, 85) ⫽ 11.64, p ⬍ .01, 2 ⫽ .06. Anger. Examination of the participants’ anger revealed that both the response expectancy feedback, F(2, 97) ⫽ 5.50, p ⬍ .01, 2 ⫽ .08, and self-efficacy expectancies, F(1, 97) ⫽ 9.90, p ⬍ .01, 2 ⫽ .08, predicted anger. However, to control for the shared variance between anger and anxiety, we also conducted an analysis with anxiety as a covariate. Once we controlled for anxiety,
self-efficacy expectancies did not predict anger about the interaction, F ⫽ 1.60, p ⫽ .21, whereas the effect of the response expectancy feedback remained significant. Participants who received negative response expectancy feedback were significantly angrier about the interaction than participants in the positive feedback condition. Participants’ level of anger in the no feedback condition fell between the other two conditions and did not significantly differ from either. There were no significant main effects or interactions involving race of participant, Fs ⬍ 2.95, ps ⬎ .08. Including gender in the analysis revealed that men (M ⫽ 2.00, SD ⫽ .97) reported higher levels of anger than women (M ⫽ 1.67, SD ⫽ 0.77), F(1, 85) ⫽ 4.69, p ⬍ .04, 2 ⫽ .03. Other-focused blame. The analysis of other-focused blame revealed only the predicted effect of the response expectancy feedback. Participants in the negative feedback condition indicated that they would be more likely to blame their partner if the interaction were to go poorly than did participants in the positive feedback condition and in the no feedback condition. The positive and no feedback conditions did not significantly differ. Partici2
Because it is never ideal to dichotomize a continuous variable, we also conducted parallel regression analyses with self-efficacy entered as a continuous predictor along with two dummy-coded variables to examine the three levels of the response expectancy feedback. The results were highly consistent with the parallel analyses obtained by the ANOVAs. For example, efficacy expectancies predicted the key avoidance measures including anxiety ( ⫽ .50, p ⬍ .001) and desire to avoid the interaction ( ⫽ .50, p ⬍ .001), whereas the negative response compared with the no feedback response expectancy code predicted key approach-related measures including anger ( ⫽ .20, p ⬍ .05), blame ( ⫽ .27, p ⬍ .02), hostile evaluations, ( ⫽ .37, p ⬍ .001), and marginally predicted race-relevant questions ( ⫽ .22, p ⫽ .06). Because the response expectancy condition had three levels, the analyses were more straightforward using ANOVAs. 3 Differences in degrees of freedom are due to missing data for one participant.
1072
BUTZ AND PLANT
pants’ self-efficacy expectancies were unrelated to their tendency to blame their partner, F ⬍ 1, p ⫽ .87. There were no main effects or interactions involving the race of participant, Fs ⬍ 1.05, ps ⬎ .35. Tendency to approach racial issues. The number of racerelevant and race-irrelevant questions participants selected for their partner were submitted to a 2 (race relevance: race-relevant vs. race-irrelevant) ⫻ 2 (race of participant: Black vs. White) ⫻ 3 (response expectancy feedback condition: positive vs. negative vs. no feedback) ⫻ 2 (self-efficacy: positive vs. negative) mixedmodel ANOVA with race relevance as the within-participants factor. The analysis revealed an effect of race relevance, F(1, 97) ⫽ 39.33, p ⬍ .001, 2 ⫽ .26, such that overall participants asked more race-irrelevant (M ⫽ 2.61, SD ⫽ 2.18) than racerelevant questions (M ⫽ 1.19, SD ⫽ 1.08). However, this main effect was qualified by the predicted Race Relevance ⫻ Feedback Condition interaction, F(2, 97) ⫽ 3.58, p ⬍ .04, 2 ⫽ .05. To examine the nature of this interaction, we analyzed the effect of the feedback condition with separate ANOVAs for race-relevant and race-irrelevant questions. As indicated in Table 2, the ANOVA for race-irrelevant questions indicated that the feedback did not have a significant influence on the number of race-irrelevant questions that participants selected for their partner. In contrast, participants in the negative response expectancy feedback condition asked significantly more race-relevant questions than did participants in the positive feedback condition. The number of race-relevant questions selected in the no feedback condition fell between the positive and negative feedback conditions but did not significantly differ from either, ps ⬎ .25. There were no significant main effects or interactions involving the race of participant, Fs ⬍ 1.11, ps ⬎ .29.
Avoidance-Related Responses Anxiety. The analysis of participants’ anxiety revealed only the significant effect of self-efficacy expectancies, which was consistent with predictions. More negative efficacy expectancies were related to greater anxiety regarding the interaction, F(1, 97) ⫽ 25.83, p ⬍ .001, 2 ⫽ .18. Even after controlling for anger, we found that the significant effect of self-efficacy remained, F(1, 96) ⫽ 16.14, p ⬍ .001, 2 ⫽ .10. Participants with more negative self-efficacy expectancies reported more anxiety about the interaction (M ⫽ 3.58, SD ⫽ 1.04) than participants with positive efficacy expectancies (M ⫽ 2.68, SD ⫽ 1.07). Of importance, both before and after controlling for anger, Fs ⬍ 2.15, p ⬎ .11, we found that the response expectancy feedback was not a significant predictor of anxiety. There were no significant main effects or interactions involving the race of participant, Fs ⬍ 2.91, ps ⬎ .09. Desire to avoid the interaction. The analysis of participants’ desire to avoid the interaction revealed the predicted effect of efficacy expectancies, F(1, 97) ⫽ 19.81, p ⬍ .001, 2 ⫽ .14. Participants with negative efficacy expectancies were more interested in avoiding the interaction (M ⫽ 3.28, SD ⫽ 1.87) than were participants with positive efficacy expectancies (M ⫽ 1.80, SD ⫽ 1.31). The influence of the response expectancy feedback was much weaker and only of marginal significance. These findings indicate that participants’ expectations regarding their ability to convey positive impressions were strongly related to their desire to avoid the interaction, whereas expectations regarding their part-
ner’s openness to the interaction were only weakly related to avoidant tendencies. There were no significant main effects or interactions involving the race of participant, Fs ⬍ 1.42, ps ⬎ .23.
Discussion In the current study, we examined the implications of response expectancy feedback for both White and Black people’s emotional reactions and behavioral intentions regarding an upcoming interracial interaction. We hypothesized that negative response expectancy feedback would result in heightened approach-related emotions such as anger and approach-related intentions such as externalizing blame, perceiving one’s partner as hostile, and selectively focusing on racial issues in the interaction. Consistent with these predictions, both Black and White participants who received negative feedback about their partner’s openness to the interaction reported relatively more anger and other-focused blame than did participants who received positive feedback about their partner’s openness to the interaction. In addition, the participants who received negative feedback rated their partner’s photograph as relatively more hostile and selected more race-relevant questions than did participants who were led to believe that their partner was open to the interaction. The response expectancy feedback was not significantly associated with avoidance-related responses such as anxiety or the desire to avoid the interaction. These results indicate that negative response expectancies evoke specifically approachrelated emotions and intentions for both Black and White participants anticipating interracial interactions. In contrast to the approach-related consequences resulting from the negative response expectancy feedback, negative self-efficacy expectancies were associated with specifically avoidance-related responses such as anxiety and the desire to avoid the upcoming interaction. Further supporting the hypothesis that negative response expectancies and negative self-efficacy expectancies would differentially predict approach- and avoidance-related emotions and intentions, self-efficacy expectancies were not associated with anger or with the measures of approach-related intentions. Finally, the current study showed that majority group members (e.g., White participants) and minority group members (e.g., Black participants) responded similarly when they expected that their partner was not open to interracial interactions. However, although not predicted, White and Black participants’ interpretations of their partner’s hostility differed as a function of their self-efficacy. White participants with negative rather than positive self-efficacy expectancies perceived their partner as more hostile, whereas Black participants with negative rather than positive self-efficacy expectancies perceived their partner as less hostile. Such results suggest that concerns about being perceived as racially biased may be particularly salient to majority group members in interracial interactions (e.g., Shelton, 2003) and negatively influence the perceptions that majority group members form of outgroup members. Together, these results complement Shelton and Richeson’s (2005) work showing that both White and Black participants report that they would like to have more interracial contact but that they believe that outgroup members do not want to have contact with them. Extending these prior findings, the current work shows that in addition to anticipating similar outcomes in interracial interactions, majority and minority group members may, as a result,
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
experience similar approach-related emotions and intentions regarding interracial interactions.
Study 2 The first study provided evidence that response expectancies and self-efficacy expectancies have different implications for White and Black people’s intentions regarding an upcoming interaction. Although findings of this study suggest that people who have negative response expectancies may respond with anger and hostile behavior in interracial interactions, it is important to both replicate the previous findings and consider the behavioral implications of negative response expectancies. Thus, the primary aims of the current study were to replicate the findings of the previous work for response and efficacy expectancies and to examine the implications of the response expectancy feedback for participants’ hostile behavior toward outgroup members. In the second study, we also used a somewhat different methodology than that used in the first study to ensure that the effects generalized across methodologies and to address potential methodological limitations of the first study. In the first study, participants were explicitly told that we were examining the implications of race for interracial interactions, which may have sensitized the participants to the issue of race and created somewhat artificial responses. White participants in the second study instead were told that they would be having an interaction but would first view a videotape of their interaction partner (who was always Black). In the second study, we also used a different manipulation of response expectancies to ensure that the effects were not limited to the manipulation used in Study 1. In a procedure similar to that used by Vorauer, Main, and O’Connell (1998), participants in the current study viewed a videotape of their partner responding to questions regarding his or her career goals and personal attributes. In the final question, the interviewee was asked to comment on social experiences that he or she had encountered at college. In the negative response expectancy condition, the partner remarked that he or she was not very open to interacting with White people because he or she anticipated prejudice in such interactions. In the control condition, the partner remarked that there had not been any particularly positive or negative social experiences at college thus far.
Method Participants and Design Participants were 27 White introductory psychology students (19 women, 8 men). An additional 3 participants (1 in the negative feedback condition) completed the experiment, but their data were not included in the analyses because they expressed suspicion that they would not have an interaction. The design of the study was a 2 (response expectancy feedback condition: negative vs. neutral) between-participants design.
Procedure Participants came into the lab and were told that the purpose of the study was to understand people’s experiences during casual first-time social interactions and the implications of learning information about people before they meet. They were told that later in the session, they would be interacting with a same-sex student.
1073
Participants were then told that with the increasing use of technology, people often meet each other for the first time via computer or videoconferencing. Therefore, the participants were told, the current study was being conducted to examine differences between first meeting someone via a computer interaction or via a more expanded video exchange. All participants were told that they would be in the condition in which they would share information over videotape. That is, they would answer some questions before the interaction that would be videotaped, and their partner would also undergo the same procedure; afterwards, the video of each member of the pair would be shown to the other person. After a short delay ostensibly to check on the progress of the participants’ interaction partner, the experimenter returned with a video in hand. Participants were randomly assigned to view either a video in which the partner indicated that he or she anticipated a negative interaction because of the possibility of prejudice or a control video in which the partner did not mention his or her expectancies. To ensure that experimenters were blind to the experimental condition, the experimenter left the room while the participant viewed the videotape. Next, participants were informed that they would prepare a videotaped interview that would be shown to their partner. Participants were asked two questions related to their personality attributes (e.g., Vorauer, Main, & O’Connell, 1998) and an additional question related to their social experiences at college. After the interview had ended, participants completed a questionnaire packet containing measures of their self-efficacy and response expectancies regarding the upcoming interaction and desire to avoid the interaction. After completing this packet, the experimenter told the participants that for the next part of the session, they would be completing a series of activities and games individually and (eventually) with their partner. During this part of the experiment, participants were provided with the opportunity to behave prosocially or in a hostile manner toward their interaction partner. Participants were told that for the first individual game, they would be completing a computer task in which they could earn money. However, before starting on this task, they were told that they would play a role in their partner’s first activity called Word Builder. Specifically, they were told that their partner would have 10 min to form as many words as possible with a series of nine letters and that their partner would receive $0.25 for each word created. In a procedure similar to that used by D. Frey and Gaertner (1986), the participants selected letters for their partner to use in this task. Of interest was the utility of the letters that participants selected for their partner’s word-building task. Allocation of harder, less useful letters (e.g., Qs, Zs, and Xs) instead of easy, useful letters (Es, Ss, and Ts) would restrict the participants’ partner from forming many words and earning more money. Hence, assigning hard letters constituted hostile behavior toward the interaction partner. After selecting letters for their partner, participants completed a questionnaire packet containing measures of their partner’s likely reaction to the letters that had been assigned to him or her. This measure tapped into participants’ cognizance of their own hostile behavior. If participants assigned their interaction partner harder letters and reported that their partner would be angry or unsatisfied with this allocation, this would indicate that participants were aware of how the letter allocation would likely be received by the interaction partner. On the experimenter’s return, the participants completed a manipulation check and suspicion probe. Finally, the participants were fully debriefed, given credit, and excused.
Materials Response expectancy manipulation. Participants in the negative response expectancy condition viewed a video that contained a same-sex Black actor responding to a series of interview questions. The first two questions were neutral in nature (e.g., Vorauer, Main, & O’Connell, 1998) and were the same for both feedback conditions. The last question was critical to the manipulation because it asked the interviewee to describe his or her social experiences at college. In the negative feedback condition, the confederate responded:
BUTZ AND PLANT
1074
In general, I think I’ve had an easy time of getting to know others . . . . The only negative thing that comes to mind is some difficulties I’ve had in interactions with White people. I guess I’d go as far as to say that my interactions with some White people . . . have been not very pleasant. As a Black person, I’m sometimes worried that White people might be biased or prejudiced toward me. These concerns come up a lot when I find myself talking to or interacting with a White person. All in all, I’d say that because of this possibility, I find myself not very open to having many interactions with White people. In the neutral response expectancy feedback condition, the confederate responded, “In general, I think I’ve had an easy time of getting to know others. I’ve made some really good friends, and we enjoy hanging out. I can’t think of any particularly positive or negative social experience.” Expectancies, emotional reactions, and avoidance. The self-efficacy expectancy scale (␣ ⫽ .71) included the same five items as Study 1 and two additional items (“I expect that it will be difficult to have a conversation with my interaction partner” and “I anticipate that I will be uncomfortable during the interaction”). Responses were reverse-coded where necessary and averaged across these items such that higher scores on these scales indicate more negative self-efficacy expectancies. Consistent with Study 1, participants were instructed to report their emotional reactions to the upcoming interaction using a series of affect descriptors. The index of anxiety (␣ ⫽ .86) included the same items used in Study 1. The index of anger contained the five items: “frustrated,” “hostile,” “angry,” “resentful,” and “bothered” (␣ ⫽ .79). The index of avoidance (␣ ⫽ .75) included two items assessing participants’ desire to avoid the interaction: “If given the option, I would avoid having this interaction,” and “I wish I could avoid having this interaction.” As a check of the response expectancy manipulation, the final questionnaire packet included the item, “Given what you know about your partner so far, how open does he or she appear about the interaction?” This scale was anchored by the endpoints of 1(not at all open) and 7 (very open). To maintain consistency with the other measures, we reverse-coded this item such that higher numbers on this scale indicate less perceived openness to the interaction. Hostile behavior. Participants were told that they would be selecting the letters that their partner would use for the word-building task. To ensure that participants understood the nature of the activity, the experimenter handed them a sheet that contained instructions explaining that their partner would receive $0.25 for each word that was three letters or longer and appeared in the Merriam-Webster Collegiate Dictionary. The sheet also contained a series of nine lines for the participant to write the letters that he or she was assigning to the interaction partner for the task. The nine letters that the participants allocated to their partner were each
assigned a corresponding score ranging from 1 through 10 on the basis of the difficulty assigned to the letter in the game Scrabble. Higher numbers indicate less useful, harder letters (e.g., Qs and Zs). Endpoints on this scale were 9 (assigned all easy letters with a score of 1 each) to 90 (assigned all difficult letters with a score of 10 each). Scrabble scores for each letter were then summed to create an overall index of hostile behavior such that higher scores reflect harder letters and, hence, greater hostile behavior. Hostile intentions. To assess participants’ hostile intentions regarding their choice of letters for their interaction partner, we asked the participants to respond to three items (e.g., “How angry will your partner be at you for assigning these particular letters to him or her?”). Items on the hostile intentions scale were anchored by the endpoints 1(not at all) and 7 (very). These items were reverse-coded when necessary and were averaged to form an index of hostile intentions such that higher scores reflect greater hostile intentions (␣ ⫽ .86).
Results Initial Analyses Intercorrelations between measures are presented in Table 3. As in Study 1, we first examined the effect of the response expectancy feedback on participants’ self-reported response expectancies and self-efficacy expectancies. Consistent with a successful manipulation, participants in the negative response expectancy feedback condition reported more negative response expectancies (M ⫽ 3.42, SD ⫽ 0.77) than did participants in the no feedback condition (M ⫽ 2.11, SD ⫽ 0.43), F(1, 25) ⫽ 30.69, p ⬍ .001, 2 ⫽ .55. This effect remained significant even after controlling for selfefficacy expectancies (see Table 4). The response expectancy feedback did not influence participants’ self-efficacy expectancies regardless of whether we controlled for response expectancies, Fs ⬍ 1.95, ps ⬎ .17.
Overview of Analyses Consistent with the analyses used in Study 1, we categorized participants as high or low in self-efficacy on the basis of a median split (Mdn ⫽ 2.86). All dependent measures were then submitted to a 2 (response expectancy feedback condition: negative vs. neutral feedback) ⫻ 2 (self-efficacy: positive vs. negative) analysis
Table 3 Intercorrelations Between Measures for Study 2 Measures
1
2
3
4
5
6
7
— .75** .34
— .40*
—
Approach-Related 1. 2. 3. 4.
Response expectancies Anger Hostile behavior Hostile intentions
— .33 .39* .48*
— .34 .04
— .23
—
Avoidance-Related 5. Efficacy expectancies 6. Anxiety 7. Desire to avoid * p ⬍ .05.
** p ⬍ .01.
.41* .14 .30
.30 .44* .47*
.23 .06 .28
.34 ⫺.06 .41*
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
1075
Table 4 Measures as a Function of Response Expectancy Feedback: Study 2 Feedback Neutral Measures
M
Negative SD
M
SD
F
df
p⬍
2
.001 .05 .05 .05
.43 .13 .16 .15
.80 .25 .80
.00 .04 .00
Approach-Related Response expectancies Anger Hostile behavior Hostile intentions
2.16 1.35 14.40 2.20
0.60 0.61 3.23 0.87
3.36 1.86 18.17 3.06
0.60 0.62 4.82 0.96
25.75 4.64 4.66 4.72
1, 1, 1, 1,
24 22 23 23
Avoidance-Related Efficacy expectancies Anxiety Desire to avoid
3.02 3.27 2.47
1.10 1.01 1.72
2.89 2.79 2.83
1.15 1.04 0.76
0.07 1.39 0.07
1, 24 1, 22 1, 23
Note. The analyses of the expectancies control for the other type of expectancy, and the analyses of the emotions control for the other type of emotion.
of variance (ANOVA).4 Effects not explicitly mentioned did not reach significance. All analyses were also conducted with participant gender included; however, there were no significant effects of gender, Fs ⬍ 1, ps ⬎ .36, and including gender did not influence the significance of reported effects. Therefore, gender was not included in the reported analyses. Participants’ approachrelated responses are first described followed by their avoidancerelated responses.
Approach-Related Responses Anger. The analysis of participants’ anger paralleled the approach used in Study 1 in which we first conducted an ANOVA with the response expectancy feedback and efficacy expectancies as predictors and then moved to an analysis of covariance that included anxiety as a covariate. The ANOVA revealed a marginal effect of the response expectancy feedback, F(1, 23) ⫽ 3.55, p ⫽ .07, 2 ⫽ .10, and a significant effect of self-efficacy expectancies, F(1, 23) ⫽ 4.54, p ⬍ .05, 2 ⫽ .13. However, after including anxiety as a covariate (i.e., accounting for variance shared between anxiety and anger), there was only the predicted effect of the response expectancy feedback (see Table 4). Efficacy expectancies were no longer significantly associated with participants’ anger about the interaction, F ⬍ 1, p ⬎ .37. Hostile behavior and intentions. The analysis of the difficulty of the letters assigned to participants’ interaction partner revealed the predicted effect of response expectancy feedback. Consistent with the hypothesis that negative response expectancies determine hostile behavior, participants in the negative response expectancy feedback condition assigned their partner more difficult letters than did participants in the neutral response expectancy feedback condition. It was also possible that the allocation of several of the same easy letters (e.g., nine Es) would also limit the partner’s ability to form words. However, inspection of the actual letters assigned to participants revealed that this was not the case (only 3 participants allocated the same letter more than once). Participants’ efficacy expectancies were not associated with the difficulty of letters assigned to the interaction partner, F ⬍ 1.19, p ⬎ .30.
To examine whether participants in the negative response expectancy feedback condition were aware of how their hostile behavior would be received, we also analyzed the extent to which participants anticipated that their partner would be angry with the letter allocation. This analysis revealed an effect of the response expectancy feedback. Participants in the negative response expectancy feedback condition anticipated that their partner would be angrier with the letter allocation than participants in the neutral feedback condition. These findings suggest that the participants who assigned their partner difficult letters were aware that a hostile response from their partner was likely. Efficacy expectancies were not related to the measure of hostile intentions, F ⬍ 1.52, p ⬎ .22.
Avoidance-Related Responses Anxiety. The initial analysis of anxiety that did not include anger as a covariate revealed only the predicted effect of selfefficacy expectancies, F(1, 23) ⫽ 9.69, p ⬍ .01, 2 ⫽ .30, which was also present when anger was included as a covariate, F(1, 22) ⫽ 5.11, p ⬍ .04, 2 ⫽ .14. Participants with negative selfefficacy expectancies reported more anxiety about the interaction (M ⫽ 3.50, SD ⫽ 1.04) than participants with positive self-efficacy expectancies (M ⫽ 2.56, SD ⫽ 1.05). The response expectancy feedback did not predict participants’ anxiety about the upcoming interaction in either analysis, Fs ⬍ 1.46, ps ⬎ .24. 4 Consistent with Study 1, we also conducted regression analyses that paralleled the ANOVAs but included self-efficacy as a continuous predictor. As in Study 1, these results were similar to those obtained by the ANOVAs. For example, the response expectancy code was associated with more anger ( ⫽ .50, p ⬍ .01) and, particularly interesting, with less anxiety ( ⫽ ⫺.34, p ⬍ .02). The feedback was also marginally associated with more antisocial intentions ( ⫽ .37, p ⫽ .06) and antisocial behavior ( ⫽ .41, p ⬍ .05). Self-efficacy expectancies were associated with anxiety after controlling for anger ( ⫽ .74, p ⬍ .001) and, although not statistically significant, were moderately related to participants’ desire to avoid the interaction ( ⫽ .33, p ⫽ .11).
BUTZ AND PLANT
1076
Desire to avoid the interaction. The analysis of participants’ desire to avoid the interaction revealed the predicted effect of self-efficacy expectancies, F(1, 23) ⫽ 7.49, p ⬍ .02, 2 ⫽ .20. Participants who reported negative self-efficacy expectancies were more interested in avoiding the upcoming interracial interaction (M ⫽ 3.35, SD ⫽ 1.45) than participants who reported positive self-efficacy expectancies (M ⫽ 1.96, SD ⫽ 0.99). As anticipated, there was no significant main effect of the response expectancy feedback. However, there was also an unexpected Response Expectancy Feedback ⫻ Efficacy interaction, F(1, 23) ⫽ 5.01, p ⬍ .04, 2 ⫽ .13. Examining the effect of efficacy in each response expectancy condition revealed that, as one would expect, in the neutral response expectancy feedback condition, participants with negative self-efficacy expectancies (M ⫽ 3.83, SD ⫽ 1.91) reported a heightened desire to avoid the interaction compared with participants with positive self-efficacy (M ⫽ 1.56, SD ⫽ 0.77), F(1, 13) ⫽ 10.53, p ⬍ .01, 2 ⫽ .45. However, in the negative response expectancy condition, participants’ desire to avoid the interaction did not differ between those reporting more positive efficacy expectancies (M ⫽ 2.70, SD ⫽ 0.97) and those reporting more negative efficacy (M ⫽ 2.93, SD ⫽ 0.84), F ⬍ 1, p ⬎ .67.
Discussion The primary goals of the present study were to replicate the findings from Study 1 and to examine the implications of response expectancies for White people’s hostile behavior toward outgroup members in interracial interactions. Using a different manipulation of response expectancies, we found that the results closely paralleled the results from the first study. Participants receiving negative response expectancy feedback reported relatively more anger (but not more anxiety) regarding the interaction than participants receiving neutral response expectancy feedback. It is notable that participants receiving the negative response expectancy feedback did not report an increased desire to avoid the interaction, but they did show evidence of more hostile behavior directed toward their interaction partner. Specifically, participants who were led to believe that their partner was not open to the interaction assigned their partner less useful letters on a word-building task, which limited their partner’s opportunity to form words and earn money. Further, participants in the negative response expectancy feedback condition were aware that their partner would likely be angry with the letters that he or she received. Taken together, these results illustrate the strong influence that negative response expectancies have on anger and hostile behavior directed toward outgroup members. An alternative explanation for the findings related to participants’ hostile intentions may be that the negative response expectancy feedback led participants to perceive that their partner was dispositionally angry and that he or she was a “complainer.” If this were the case, even participants with positive intentions or participants who assigned easy letters to their partner would be expected to report that their allocation would be met with anger because of their partner’s negative disposition. Future research should address whether people’s antisocial behavior toward outgroup members stems from their own negative intentions or from their perceptions of negative dispositions in outgroup members. Replicating the results from the first study, participants in the current study with negative self-efficacy expectancies reported
greater anxiety and a greater desire to avoid the interaction but not anger or hostility. Together, the results from the two present studies provide further support for the hypothesis that response expectancies and self-efficacy expectancies differentially predict people’s approach and avoidance-related reactions to interracial interactions.
General Discussion In light of continuing initiatives to promote racial and cultural diversity, it is critical to understand the factors that determine who is likely to avoid interracial interactions, as well as who is likely to respond positively or negatively when such interactions are unavoidable. Much previous work has focused on the determinants of people’s avoidance of interracial interactions (e.g., Britt et al., 1996; Plant & Butz, 2006; Plant & Devine, 2003; Stephan & Stephan, 1985). Replicating previous work (e.g., Plant & Butz, 2006), negative self-efficacy expectancies across the current studies were distinctly associated with avoidance-related reactions, including heightened anxiety and an increased desire to avoid interracial interactions. However, the current work extends previous research by providing insight into the factors that determine approach-related responses to interracial interactions. We proposed that people who expected that outgroup members would not be open to interracial interactions would become angry and blame outgroup members for negative interactions. Further, we anticipated that people with negative response expectancies would report hostile intentions toward outgroup members and, if given the chance, would behave in a relatively more hostile manner than people who possessed positive response expectancies. Across two studies, these predictions were confirmed. In Study 1, White and Black participants who were led to believe that their other-race interaction partner was not open to interracial interactions reported more anger about the interaction (but not more anxiety) than participants in the positive response expectancy feedback condition. Moreover, participants in the negative response expectancy feedback condition exhibited more approachrelated intentions toward their interaction partner. Compared with participants in the positive response expectancy feedback condition, participants in the negative feedback condition evaluated their interaction partner’s appearance in a photograph as relatively more hostile and threatening, (even though they were viewing the exact same photographs), and they attributed blame to their partner for negativity in the interaction. These findings suggest that people who perceive outgroup members as not open to interacting may enter interracial interactions feeling angry at outgroup members and may be more prone to viewing them as hostile. Further, focusing blame on outgroup members for any difficulty in the interaction allows people to absolve themselves of personal responsibility for a failed interaction and may be used to justify a more hostile demeanor during the interactions. Study 2 provided compelling evidence that negative response expectancies result in relatively greater hostile intentions and behavior. Compared with participants who were given neutral expectations about their partner’s openness to the interaction, White participants who were led to believe that their Black interaction partner was not open to the interaction made it harder for their partner to succeed on a meaningful task. Further, these participants were aware that their partner would be angry at their
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS
actions. Thus, it was not by chance or out of carelessness that participants in the negative response expectancy condition doled out harder letters to their partner than did participants in the neutral feedback condition. The results across the studies point to the possibility that people with negative response expectancies could actually elicit negative responses from their outgroup interaction partners, potentially resulting in behavioral confirmation of such negative expectancies (see Devine et al., 1996; Miller & Turnbull, 1986). Consider that in the current studies, people who were led to believe that their partner was not open to interracial interactions responded with relatively more anger, hostile intentions, and hostile behavior than participants in the other feedback conditions. In actual interracial encounters, such negatively intentioned behavior may elicit hostile reactions from outgroup members that seemingly confirm that outgroup members are not open to interracial interactions. People may be unaware that their own hostile behavior may have influenced how outgroup members reacted to them, which may invoke a cyclical pattern of negative emotions and behaviors directed toward and elicited from outgroup members. By examining the perspectives of both majority and minority group members, we showed in Study 1 that many of the findings for response expectancies and self-efficacy expectancies generalized to Black and White participants. When Black and White participants were led to expect that their outgroup interaction partner was not open to interracial interactions, they responded with anger and a range of approach-related intentions. In contrast, when Black and White participants perceived that they would be inefficacious in the interracial interaction, they responded with anxiety and the intention to avoid. These findings are consistent with recent work indicating that although majority and minority group members may not be aware that they anticipate similar outcomes in interracial interactions, their actual expectations and experiences tend to be strikingly similar (e.g., Hyers & Swim, 1998; Shelton & Richeson, 2005). Indeed, we believe that, consistent with the findings of Shelton and Richeson (2005), our findings show that promoting awareness of the similarities among minority and majority group members’ experiences and concerns in interracial interactions may reduce intergroup tension and miscommunication.
Limitations and Future Directions One potential limitation of the studies presented in the current work is that they assessed participants’ reactions prior to an anticipated interaction. In future work, it will be important to examine how response expectancies influence actual interracial interactions and whether participants’ anger and hostility are apparent to their interaction partner. Furthermore, it will be important to examine whether people’s expectancies are important for other types of intergroup interactions, for example, interactions with other racial or ethnic outgroups (e.g., between non-Hispanics and Hispanics) or interactions with different social groups (interactions between heterosexuals and gays or lesbians). With the exception of our preliminary study, the current work has primarily focused on the implications of negative response expectancies for intergroup exchanges. Indeed, we believe that response expectancies may be particularly important for understanding the sources of approach-related negativity in intergroup
1077
interactions because another’s lack of openness may be attributed to race or racial intolerance. As a result, negative response expectancies for ingroup members may generate emotional and behavioral reactions that differ in strength, but not necessarily in form, from negative response expectancies for intergroup interactions (e.g., see F. E. Frey & Tropp, 2006). To examine these ideas, researchers in the future could expand on our preliminary study by including additional measures related to the extent to which participants personalize their partner’s unresponsiveness (i.e., items assessing the extent to which participants feel personally culpable for their partner’s unresponsiveness and experience negative selfdirected affect). One might also examine the attributions that participants make for their partner’s unresponsiveness with a focus on whether participants report more group-based and personalized attributions in the intergroup interaction condition than in the intragroup condition. Another avenue for future research may be to consider whether factors such as the level of prejudice moderate the intensity and nature of participants’ reactions to outgroup members. Compared with participants with a low level of prejudice, participants with a high level may be particularly likely to expect that outgroup members are not open to interacting with them and, concomitantly, to experience approach-related reactions such as anger. On the other hand, concerns about responding without bias in interracial interactions (i.e., self-efficacy) may be more relevant to people with a low compared with a high level of prejudice, which suggests that people with a low level of prejudice may be more prone to anxious or avoidant reactions. These predictions are consistent with the findings of Devine and colleagues (1996), who showed that participants who were highly prejudiced expected to experience high levels of antipathy regarding intergroup interactions, whereas those who had low and moderately low levels of prejudice reported anxious but not antipathy-based reactions. Future work should also examine the role of interracial contact in determining people’s expectations about interactions and emotional reactions to outgroup members. People who have had more positive interracial contact may report more positive self-efficacy and response expectancies for interracial interactions (e.g., Plant & Devine, 2003) and, consistent with findings from a recent metaanalysis, report more positive emotions regarding interactions with outgroup members (Tropp & Pettigrew, 2005). It is also possible that there are other individual difference variables that determine people’s responses to interracial interactions. For example, people who have a strong internal locus of control may be more likely to internalize blame and experience self-directed emotions (e.g., guilt) for a negative interaction, whereas people who have a strong external locus of control may instead externalize blame for a negative interaction and experience other-directed emotions (e.g, anger).
Conclusions We believe that an understanding of the factors that lead to negativity in interracial interactions is critical for efforts to improve the quality of such interactions. The current work showed that people’s approach and avoidance-related tendencies for interracial interactions may be explained by their specific expectations about interracial interactions. Although people who anticipate that outgroup members will not be open to them in interracial interac-
BUTZ AND PLANT
1078
tions may react with emotions and behaviors that bring them into conflict with outgroup members, people who perceive that they will convey awkward, unpolished impressions in interracial interactions may react with emotions and behaviors that pull them away from outgroup members. In its examination of the factors that contribute to intergroup anger, the present work calls for an expansion of previous models of interracial interactions that have focused solely on intergroup anxiety (e.g., Plant & Devine, 2003; Stephan & Stephan, 1985). Our hope is that by identifying the sources of negative reactions to interracial interactions, our current work provides some insight into how these negative reactions can be eliminated, thereby improving the quality of interracial interactions for all people involved. Such efforts may increase the likelihood that instead of avoiding interracial contact or approaching it with hostility, people will approach interracial interactions with the confidence that will result in pleasant interracial interactions.
References Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191–215. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122–147. Bandura, A., Adams, N. E., Hardy, A. B., & Howells, G. N. (1980). Tests of the generality of self-efficacy theory. Cognitive Therapy & Research, 4, 39 – 66. Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5–33. Branscombe, N. R., Schmitt, M. T., & Harvey, R. D. (1999). Perceiving pervasive discrimination among African Americans: Implications for group identification and well-being. Journal of Personality and Social Psychology, 7, 135–149. Britt, T. W., Boniecki, K. A., Vescio, T. K., Biernat, M., & Brown, L. M. (1996). Intergroup anxiety: A person ⫻ situation approach. Personality and Social Psychology Bulletin, 22, 1177–1188. Cottrell, C. A., & Neuberg, S. L. (2005). Different emotional reactions to different groups: A sociofunctional threat-based approach to “prejudice.” Journal of Personality and Social Psychology, 88, 770 –789. Darley, J. M., Fleming, J. H., Hilton, J. L., & Swann, W. B. (1988). Dispelling negative expectancies: The impact of interaction goals and target characteristics on the expectancy confirmation process. Journal of Experimental Social Psychology, 24, 19 –36. Devine, P. G., Evett, S. R., & Vasquez-Suson, K. A. (1996). Exploring the interpersonal dynamics of interracial context. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: The interpersonal context (Vol. 3, pp. 423– 464). New York: Guilford. Frey, D., & Gaertner, S. L. (1986). Helping and the avoidance of inappropriate interracial behavior: A strategy which perpetuates a nonprejudiced self-image. Journal of Personality and Social Psychology, 50, 1083–1090. Frey, F. E., & Tropp, L. R. (2006). Being seen as individuals versus as group members: Extending research on metaperception to intergroup contexts. Personality and Social Psychology Review, 10, 265–280. Frijda, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness. Journal of Personality & Social Psychology, 57, 212–228.
Gudykunst, W. B. (1993). Toward a theory of effective interpersonal and intergroup communication: An anxiety/uncertainty management perspective. In R. L. Wiseman & J. Koester (Eds.), Intercultural communication theory (pp. 8 –58). Thousand Oaks, CA: Sage. Hamilton, D. L., & Sherman, S. J. (1996). Perceiving persons and groups. Psychological Review, 103, 336 –355. Harmon-Jones, E. (2003). Anger and the behavioral approach system. Personality and Individual Differences, 35, 995–1005. Hilton, J. L., & Darley, J. M. (1985). Constructing other persons: A limit on the effect. Journal of Experimental Social Psychology, 21, 1–18. Hyers, L. L., & Swim, J. K. (1998). A comparison of the experiences of dominant and minority group members during an intergroup encounter. Group Processes and Intergroup Relations, 1(2), 143–163. Kent, M. M., & Mather, M. (2002, December). What drives the U.S. population growth? Population Bulletin, 57(4), 1– 44. Krueger, J. (1996). Personal beliefs and cultural stereotypes about racial characteristics. Journal of Personality and Social Psychology, 71, 536 – 548. Leary, M. R., & Atherton, S. C. (1986). Self-efficacy, social anxiety, and inhibition in interpersonal encounters. Journal of Social and Clinical Psychology, 4, 256 –267. Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602– 616. Maddux, J. E., Norton, L. W., & Leary, M. R. (1988). Cognitive components of social anxiety: An investigation of the integration of selfpresentation theory and self-efficacy theory. Journal of Social and Clinical Psychology, 6, 180 –190. Mendoza-Denton, R., Downey, G., Purdie, V. J., Davis, A., & Pietrzak, J. (2002). Sensitivity to status-based rejection: Implications for African American students’ college experience. Journal of Personality and Social Psychology, 83, 896 –918. Miller, D. T., & Turnbull, W. (1986). Expectancies and interpersonal processes. Annual Review of Psychology, 37, 233–256. Monteith, M. J., & Spicer, C. V. (2000). Contents and correlates of Whites’ and Blacks’ racial attitudes. Journal of Experimental Social Psychology, 36, 125–154. Plant, E. A. (2004). Responses to interracial interactions over time. Personality and Social Psychology Bulletin, 30, 1458 –1471. Plant, E. A., & Butz, D. A. (2006). The causes and consequences of an avoidance-focus for interracial interactions. Personality and Social Psychology Bulletin, 32, 833– 846. Plant, E. A., & Devine, P. G. (2003). The antecedents and implications of interracial anxiety. Personality and Social Psychology Bulletin, 29, 790 – 801. Schlenker, B. R., & Leary, M. R. (1982). Social anxiety and self-presentation: A conceptualization and model. Psychological Bulletin, 92, 641– 669. Shelton, J. N. (2003). Interpersonal concerns in social encounters between majority and minority group members. Group Processes and Intergroup Relations, 6, 171–185. Shelton, J. N., & Richeson, J. A. (2005). Intergroup contact and pluralistic ignorance. Journal of Personality and Social Psychology, 88, 91–107. Smith, E. R. (1993). Social identity and social emotions: Toward new conceptualizations of prejudice. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 297–315). San Diego, CA: Academic Press. Stephan, W. G., & Stephan, C. W. (1985). Intergroup anxiety. Journal of Social Issues, 41, 157–175. Stephan, W. G., & Stephan, C. W. (1989). Antecedents of intergroup anxiety in Asian-Americans and Hispanic-Americans. International Journal of Intercultural Relations, 13, 203–219. Tropp, L. R., & Pettigrew, T. F. (2005). Differential relationships between
APPROACH-RELATED RESPONSES TO OUTGROUP MEMBERS intergroup contact and affective and cognitive dimensions of prejudice. Personality and Social Psychology Bulletin, 31, 1145–1158. Twenge, J., M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001). If you can’t join them, beat them: Effects of social exclusion on aggressive behavior. Journal of Personality & Social Psychology, 81, 1058 –1069. Vorauer, J. D., & Kumhyr, S. M. (2001). Is this about you or me? Selfversus other-directed judgments and feelings in response to cross-group interaction. Personality and Social Psychology Bulletin, 27, 706 –719. Vorauer, J. D., Main, K. J., & O’Connell, G. B. (1998). How do individuals
1079
expect to be viewed by members of low status groups? Content and implications of meta-stereotypes. Journal of Personality and Social Psychology, 75, 917–937.
Received August 3, 2005 Revision received May 31, 2006 Accepted June 1, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1080 –1093
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1080
Group Decision Making in Hidden Profile Situations: Dissent as a Facilitator for Decision Quality Stefan Schulz-Hardt
Felix C. Brodbeck
Georg-August-University Goettingen
Aston University, Birmingham
Andreas Mojzisch
Rudolf Kerschreiter and Dieter Frey
Georg-August-University Goettingen
Ludwig-Maximilians-University Munich
The effect of diversity in individual prediscussion preferences on group decision quality was examined in an experiment in which 135 three-person groups worked on a personnel selection case with 4 alternatives. The information distribution among group members constituted a hidden profile (i.e., the correct solution was not identifiable on the basis of the members’ individual information and could be detected only by pooling and integrating the members’ unique information). Whereas groups with homogeneous suboptimal prediscussion preferences (no dissent) hardly ever solved the hidden profile, solution rates were significantly higher in groups with prediscussion dissent, even if none of these individual prediscussion preferences were correct. If dissent came from a proponent of the correct solution, solution rates were even higher than in dissent groups without such a proponent. The magnitude of dissent (i.e., minority dissent or full diversity of individual preferences) did not affect decision quality. The beneficial effect of dissent on group decision quality was mediated primarily by greater discussion intensity and to some extent also by less discussion bias in dissent groups. Keywords: group decision making, information pooling, hidden profile, dissent, minority influence
Some 20 years ago, Garold Stasser and William Titus (1985) published their seminal article on group decision making in situations in which the decision-relevant information is distributed
among members. In their article, they introduced a paradigm that has subsequently been labeled hidden profile (Stasser, 1988). In a hidden profile, part of the information is shared among group members (i.e., all members possess this information prior to discussion), whereas other pieces of information are unshared (i.e., information known to only one member prior to discussion). Furthermore, shared information and unshared information have different decisional implications, and the alternative implied by the unshared information is the correct one given all information available to the group. However, no group member can detect this best solution on the basis of her or his individual information prior to discussion; it can only be found by pooling the unshared information during group discussion. Stasser and Titus (1985) demonstrated that groups predominantly fail to solve hidden profiles, and subsequent studies have shown this failure to be very robust (for overviews, see Brodbeck, Kerschreiter, Mojzisch, & Schulz-Hardt, in press; Wittenbaum, Hollingshead, & Botero, 2004). However, Stasser and Titus (1985) also attempted to demonstrate another effect. Their idea was that groups should be more likely to solve hidden profiles if group members experience dissent about the choice to be made. Whereas in their “unshared/consensus” condition all four group members received individual information that implied Alternative B to be the best choice (given all information, A was best), in the “unshared/conflict” condition two group members received individual information in favor of B, and the other two members received information in favor of C (again, given all information, A was best). The rationale was that conflict between the two emerging preference factions should stimulate the
Stefan Schulz-Hardt and Andreas Mojzisch, Institute of Psychology, Economic and Social Psychology Unit, Georg-August-University Goettingen, Germany; Felix C. Brodbeck, Aston Business School, Work and Organisational Psychology Unit, Aston University, Birmingham, United Kingdom; Rudolf Kerschreiter and Dieter Frey, Department of Psychology, Social and Economic Psychology Unit, Ludwig-MaximiliansUniversity Munich, Germany. Felix Brodbeck and Stefan Schulz-Hardt contributed equally to this article; their authorship order was determined by a coin flip. Parts of the data were presented at the 13th General Meeting of the European Association of Experimental Social Psychology in San Sebastian, Spain, June 2002, and at the 11th European Congress on Work and Organizational Psychology in Lisbon, Portugal, May 2003. The research reported in this article was made possible by grants from the German Scientific Foundation (Deutsche Forschungsgemeinschaft) to Felix Brodbeck, Stefan Schulz-Hardt, and Dieter Frey (Project No. SCHU 1279/1-1). We thank Shanshan Chen, Andrea Csanadi, Marina Deiss, Tanja Deiss, Yasmin Dirkes, Beate Dorsch, Stephanie Flo¨ter, Nadira Faulmu¨ller, Ingrid Mayer, Simone Schickel, Sybille Schuhwerk, Tatjana Schweizer, Eva Traut-Mattausch, Adriana Tzvetkova, and Martin Winkler for their assistance in collecting the data reported in this article. Correspondence concerning this article should be addressed to Stefan Schulz-Hardt, Institute of Psychology, Economic and Social Psychology Unit, Georg-August-University, Gosslerstrasse 14, D-37073 Goettingen, Germany, or to Felix C. Brodbeck, Aston Business School, Work and Organisational Psychology Unit, Aston University, Birmingham B4 7ET, United Kingdom. E-mail:
[email protected] or
[email protected] 1080
DISSENT AND GROUP DECISION QUALITY
exchange of information and, hence, the solution of the hidden profile. Unfortunately, this second suggestion was not confirmed. The solution rate was by no means higher in the conflict condition (M ⫽ 12%) than in the consensus condition (M ⫽ 24%). The absence of any beneficial effects of the conflict manipulation is striking because, in the group decision making literature, dissent among group members’ individual prediscussion preferences is generally viewed as a facilitator for group decision quality (e.g., De Dreu & Beersma, 2001; Dooley & Fryxell, 1999; Simons, Pelled, & Smith, 1999). In this article, we subject the role of dissent in the hidden profile paradigm to a new and methodologically more sound empirical test. We first outline why understanding the role of dissent is important for both theoretical and practical reasons and derive predictions about why and how prediscussion dissent should help groups to solve hidden profiles. We then report an experiment that was designed to test these ideas.
Dissent as a Facilitator for Solving Hidden Profiles— Theoretical and Practical Importance Although it is widely assumed that group decision quality benefits from prediscussion dissent, empirical support for this assumption is weaker than it might seem at first glance. In one type of study, other aspects of diversity in groups (e.g., with regard to personality, functional background, or training) have been investigated, and it has been concluded that predominantly those types of diversity that foster disagreement in the decision-making process are beneficial for decision quality (e.g., Williams & O’Reilly, 1998). A second type of study has shown that artificial dissent, introduced by techniques like devil’s advocacy or dialectical inquiry, raises the quality of group decisions (see Katzenstein, 1996, or Schwenk, 1990, for reviews). However, neither type of study directly addresses genuine dissent in prediscussion preferences, and thus, aspects other than dissent might be responsible for the findings (e.g., structuring of the group decision process by means of a dialectical technique). In a third type of study, it has been shown that exposure to dissenting opinions, especially if they come from a minority, raises creativity as well as quantity and quality of individual problem solutions (e.g., Nemeth, Rogers, & Brown, 2001; Nemeth & Wachtler, 1983). However, being individually exposed to a diverging opinion (e.g., on a piece of paper or by feedback from the experimenter) and discussing diverging opinions in a group might be two different things and lead to different results. Finally, in a fourth type of study, positive correlations between dissent and group decision quality have been shown, but with measurement (particularly of dissent) based on retrospective ratings (e.g., De Dreu & West, 2001; Dooley & Fryxell, 1999). Therefore, it cannot be ruled out that, for example, subjective theories about the determinants of group performance have led to these findings. Thus, what is largely missing are experimental studies in which prediscussion dissent is introduced as an independent variable and group decision (or judgment) quality is subsequently measured as a dependent variable. To our knowledge, only four studies approach this design (Brodbeck, Kerschreiter, Mojzisch, Frey, & Schulz-Hardt, 2002; Hightower & Sayeed, 1996; Sniezek & Henry, 1989; Wanous & Youtz, 1986), all of which resulted in better decisions being associated with higher diversity in predis-
1081
cussion preferences. However, even in these laboratory studies, prediscussion dissent was either not manipulated or, at best, manipulated only indirectly. For example, in their hidden profile studies, Brodbeck et al. (2002) as well as Hightower and Sayeed (1996) manipulated preferences via different information distributions, which makes it possible that these informational differences rather than the amount of dissent accounted for their findings. Furthermore, diversity in opinions appears to be confounded with the quality of the best member’s individual preference. For example, Brodbeck et al. (2002) and Hightower and Sayeed (1996) showed that groups with all members preferring different alternatives solve hidden profiles more often than do homogeneous groups. However, in their homogeneous groups, all preferences were suboptimal, whereas all dissent groups contained one member preferring the correct solution—which should facilitate decision quality independent of dissent.1 In summary, an unequivocal test of the effects of prediscussion dissent on group decision quality requires a situation in which dissent can be manipulated independent of the quality of the best member’s individual solution. In our experiment, we used a special type of hidden profile that allows for this. Besides the theoretical and methodological advantages of the hidden profile paradigm— especially the fact that it allows grouplevel effects on decision quality to be clearly identified—its practical relevance also makes it well suited for such an investigation: In all kinds of political, economic, and societal contexts, important decisions are often made by groups, and one of the reasons for this is that groups possess larger informational resources (e.g., Clark & Stephenson, 1989) and thus are expected to make better decisions than are individuals (e.g., Vroom & Jago, 1988). As has been argued by Brodbeck et al. (in press), the higher costs (with regard to time, money, and effort) of group decision making compared with individual decision making or polls of individual votes can only pay off with regard to decision quality if (most) group members’ preferences prior to discussion are suboptimal and if exchanging information during discussion has the potential to help them find the best solution. Hidden profiles are a prototype of such situations. If groups consistently perform suboptimally in situations in which their use should be beneficial, interventions are called for that enable groups to deal with this particular type of decision problem more successfully. The search for such interventions has hitherto not been very successful (see Stasser & Birchmeier, 2003, for a review). Thus, finding that prediscussion dissent fosters a 1
In the case of the Wanous and Youtz (1986) study, the problem is even somewhat more complicated. Wanous and Youtz used two variants of the moon survival task for their study, which requires a group to rank order 15 items according to their importance for survival. Diversity was assessed as the inverse of the concordance of individual prediscussion rankings. Before regressing solution quality on diversity, Wanous and Youtz entered the best member’s solution quality as a covariate, which seems to rule out the problem mentioned. However, because these survival tasks consist of multiple subtasks (ranking the different items separately, which can give up to 14 subtasks), different “best members” for at least some of the different subtasks should exist. Thus, increasing the diversity of individual solutions increases the likelihood that the group contains a member with a correct (or almost correct) ranking for any specific item, and this effect is not statistically controlled for in the Wanous and Youtz study.
SCHULZ-HARDT ET AL.
1082
group’s ability to solve hidden profiles might open new avenues for such interventions.
Overview of the Present Study and Hypotheses The present study was designed to test whether prediscussion dissent per se is beneficial for the quality of group decisions in hidden profile situations. To this end, dissent had to be operationalized independent of the quality of group members’ prediscussion preferences. This was achieved by constructing a decision case with four alternatives for three-person groups, with three equally attractive suboptimal alternatives and one superior decision alternative (the correct solution). As a consequence, complete dissent in prediscussion preferences (which we label full diversity dissent) could be obtained without one group member necessarily having to prefer the correct alternative. Five different hidden profile conditions were realized: 1.
No dissent (homogeneity, i.e., all group members prefer the same suboptimal alternative prior to discussion).
2.
Pure minority dissent (two members prefer the same and the third member prefers a different suboptimal alternative).
3.
Pure full diversity dissent (all members prefer different suboptimal alternatives).
4.
Minority dissent with proponent (two members prefer the same suboptimal alternative and the third member prefers the best alternative).
5.
Full diversity dissent with proponent (two members prefer different suboptimal alternatives and the third member prefers the best alternative).
The first three conditions (1 vs. 2 and 3) permitted testing the pure effect of prediscussion dissent (labeled pure dissent effect), whereas the comparison of Conditions 4 and 5 with Conditions 2 and 3 tested for the effects of a proponent for the correct solution (the proponent dissent effect). Furthermore, comparing Conditions 3 and 5 with Conditions 2 and 4 allowed us to test whether full diversity dissent, independent of the presence or absence of a proponent for the correct solution, is more beneficial than minority dissent (the magnitude-of-dissent effect). Predictions about beneficial effects of prediscussion dissent on the solution of hidden profiles can be derived if we consider the two group-level processes that hinder groups from solving hidden profiles (cf. Brodbeck et al., in press; Mojzisch & Schulz-Hardt, 2006; Winquist & Larson, 1998)2: The first one is that groups tend to negotiate the final decision on the basis of their members’ prediscussion preferences rather than openly discussing the available information (Gigone & Hastie, 1993). This premature preference negotiation is detrimental for decision quality in hidden profiles because (a) hidden profiles predispose group members to individually prefer suboptimal alternatives prior to discussion, and because of this any prematurely emerging consensus will also be suboptimal, and (b) premature preference negotiation precludes an intensive discussion of the total information available in the group, so that the group fails to exchange sufficient information to detect
the superiority of the best alternative. Premature consensus on the basis of members’ prediscussion preferences should be less likely to occur the more disagreement there is among these preferences. Compared with members of homogeneous groups, members of dissent groups should engage in a more intense debate to argue out the pros and cons of their diverse preferences. In turn, discussion should last longer and more information should be exchanged than in homogeneous groups (Parks & Nelson, 1999). We summarized this by proposing that dissent intensifies discussion. The second hurdle for the solution of hidden profiles is that even if the group engages in information exchange, group discussion is systematically biased against the best solution. Detecting the correct solution in a hidden profile requires the group members to discuss information that is both unshared and inconsistent with their individual preferences. However, shared as well as preference-consistent information is introduced and repeated in group discussions more often than unshared and preference-inconsistent information (e.g., Dennis, 1996; Larson, Foster-Fishman, & Keys, 1994). Both types of discussion bias should be reduced in groups with prediscussion dissent: As we know from minority influence research (e.g., Moscovici, 1980; Nemeth, 1986), minority dissent instigates a self-critical check of one’s own position as well as divergent thinking, which means that the person openly evaluates all available options, including ones that are proposed neither by herself nor by the minority. Thus, if a group consists of a minority and a majority faction, these influence processes should make at least the majority members highly receptive to information that is new to them (unshared information) and that contradicts their individual preferences (Schulz-Hardt, Frey, Lu¨thgens, & Moscovici, 2000; Schulz-Hardt, Jochims, & Frey, 2002). This effect should be even larger if all group members prefer different alternatives because in this case all members are exposed to minority influence (Brodbeck et al., 2002). On the basis of these findings, we predicted that dissent debiases discussion. Both consequences of dissent, namely greater discussion intensity and less discussion bias, should facilitate discussion of the correct alternative because in a hidden profile this alternative is largely supported by unshared information and is inconsistent with the group members’ individual prediscussion preferences. More discussion about the correct alternative should, in turn, make the solution of the hidden profile more likely, because discussing this alternative should increase the likelihood that its superiority is detected. These improvements brought about by prediscussion dissent should occur even if all members prefer suboptimal alternatives. Thus, we predicted a pure dissent effect on decision quality that is mediated by increased discussion about the correct alternative, and this is, in turn, mediated by more discussion intensity and less discussion bias in dissent groups compared with in homogeneous groups. Moreover, we predicted that the likelihood of solving the hidden profile is further enhanced if one of the diverging prediscussion 2
The failure of groups to solve hidden profiles is not exclusively caused by group processes. As recent studies have shown (e.g., Greitemeyer & Schulz-Hardt, 2003), biases in the individual evaluation of information also contribute to this failure (for a detailed outline, see Brodbeck et al., in press). However, because we did not measure information evaluation in our experiment (which is difficult to realize in real-group discussions), we abstain from deriving predictions for the impact of prediscussion dissent on such individual-level processes here.
DISSENT AND GROUP DECISION QUALITY
preferences is in favor of the correct alternative (proponent dissent effect). As the pure dissent effect, this proponent dissent effect on decision quality should also be mediated by discussion about the correct alternative, because dissent groups with a proponent for this alternative should discuss it even more extensively than dissent groups without such a proponent. However, in contrast to the pure dissent effect, this further enhancement in discussion about the correct alternative should not be the consequence of more discussion intensity and less discussion bias—we see no reason why dissent by a supporter of the correct alternative should increase discussion intensity more and decrease discussion bias less than dissent by a supporter of a suboptimal alternative does. Instead, a proponent of the correct alternative should have a direct effect on discussion of the correct alternative, simply because he or she prefers this alternative and keeps discussion about his or her preferred alternative alive. These two different mediational chains are illustrated in Figure 1. Finally, we predicted that both dissent effects should be stronger in groups with full preference diversity than in groups with minority dissent (magnitude-of-dissent effect). Both mediational chains should benefit from full diversity: As outlined, discussion intensity should be highest and discussion bias should be lowest in full diversity groups in which all members act as minorities (Brodbeck et al., 2002), and proponents for a correct solution should have more influence if they do not have to act against a uniform majority.
Method Participants and Design All data were collected from students at Ludwig-MaximiliansUniversity, Munich, Germany. A total of 447 students (317 women, 130 men) with an average age of 23.86 years participated in the experiment, with 3 persons forming a group. They received €9 (⬃$11) for their participation. The experiment is based on a one-way design with six levels, constituted of five hidden profile conditions (homogeneity, pure minority dissent, pure full diversity dissent, minority dissent with proponent, and full diversity dissent with proponent—for a closer description, see the previous section) and a control condition in which group members each received complete information (full information, no hidden profile). In accordance with the previous literature, the latter condition served as a control for the impact of the hidden profile distribution of information.
+
Discussion intensity
+ Discussion about correct alternative
Pure (prediscussion) dissent -
Discussion bias
+
Choice of correct alternative
-
1083
Material The decision case deals with an airline company looking for a new pilot for long-distance flights. The participants play the role of a member of the personnel selection committee of this airline company. They have to choose between four candidates named A, B, C, and D. In the full information set, each of the four candidates is characterized by 10 attributes that are either positive or negative. These 40 attributes had been selected in a pretest in which 100 items were rated by a sample of 112 students. From this item pool, those 40 attributes were chosen that were rated as most unambiguously positive or negative and as being of comparable importance and strength. An example of a positive attribute is “The candidate can concentrate very well over long periods.” An example of a negative attribute is “The candidate is said to be a know-it-all.” In contrast to positive attributes, extremely negative attributes were avoided because it would be implausible that such a candidate would have survived the organizational preselection. The distribution of information about the four candidates is shown in Table 1. We made sure that the four candidates received attributes of similar average strength and importance, so that the number of positive and negative attributes per candidate should decide their ranking (full information about all pretested and selected items as well as all statistics can be obtained from Stefan Schulz-Hardt). Given the full information set, Candidate C was the best choice.3 Whereas this candidate had seven positive and only three negative attributes, all other three candidates (A, B, and D) had four positive and six negative attributes. This ranking was confirmed in a second pretest with 71 students who were given the full candidate information. Of these 71 participants, 62 participants (87%) chose Candidate C. In the hidden profile conditions, each member received a subset of this information. For Candidates A, B, and D, all positive attributes were shared and all negative attributes were unshared. Thus, for each group member, each of the Candidates A, B, and D had four advantages and only two disadvantages prior to discussion. In contrast, all negative attributes and only one positive attribute about Candidate C were shared, with the other six positive attributes unshared. Thus, for each group member, Candidate C had three advantages and three disadvantages prior to discussion. As a consequence, most group members should prefer Candidate A, B, or D prior to discussion. By detecting that Candidate C, who initially seems to be the least attractive one, is in fact really the best candidate, the groups can realize a group-specific surplus in decision quality. However, because the initial difference in positive and negative attributes is not large and because there is always some variation with respect to participants’ idiosyncratic interpretations of the perceived importance and valence of information, at least some participants should prefer Candidate C prior to discussion, which allowed us to form groups with a proponent for the correct choice. These expectations were confirmed in a third pretest with 83 students. Each participant received one of the three individual prediscussion profiles (differing only with regard to the unshared items in the profile). Of these, 21 (25%) chose Candidate A, 27 (33%) chose Candidate B, 26 (31%) chose Candidate D, and only 9 (11%) chose Candidate C. Thus, all pretests confirmed that our decision case material successfully induces a hidden profile.
Procedure We invited 6, 9, or 12 persons to each experimental session in the lab rooms of the Social Psychology Unit at Ludwig-Maximilians-University. If
Dissent by a proponent of the correct choice
+
Discussion about correct alternative
+
Choice of correct alternative
Figure 1. Proposed effects of pure dissent and proponent dissent on decision quality via group discussion.
3 To be precise, we also had a second rotated version with the same information but switched candidate labels, in which Candidate A was the best choice. However, for the sake of simplicity and clarity, methods and results are presented according to the unrotated version. No differences between these two versions occurred in our experimental data.
SCHULZ-HARDT ET AL.
1084
Table 1 Distribution of Information in the Hidden Profile Conditions Candidate Information type and valence Shared information Positive Negative Unshared information Positive Negative Information available to each individual Positive Negative Full information available to the group Positive Negative
A
B
C
D
4 0
4 0
1 3
4 0
0 6
0 6
6 0
0 6
4 2
4 2
3 3
4 2
4 6
4 6
7 3
4 6
1 or 2 persons failed to arrive, the remaining 1 or 2 persons that could not participate in a 3-person group were assigned to a different experiment. The participants were welcomed by the experimenter and briefly informed about the procedure and aims of the experiment. Specifically, it was emphasized that the experiment focuses on the process and quality of group decision making. To investigate this, the participants would first receive and work on individual material about a personnel selection case. Afterward, they would be assigned to groups that should make a common, final decision about which of the four candidates should be hired. The group discussion would be videotaped, with the videos exclusively being used for scholarly purposes. If participants did not agree to being videotaped, they were assigned to a different experiment. The experimenter then handed out a cover letter introducing the decision case and providing some basic information about the airline company and the selection situation. On this sheet, the participants also indicated their sex, age, and main subject at the university. In addition, they were given a code to use on all subsequent questionnaires. The code consisted of a three-digit number as well as a letter indicating whether they would be Group Member X, Y, or Z in the following group discussion. The experimenter then handed out a candidate information sheet and an information evaluation questionnaire. On the candidate information sheet, each of the four candidates A, B, C, and D was characterized by six attributes. Three versions of this information sheet existed, one each for prospective X, Y, and Z members. The three versions did not differ with regard to the number of positive and negative attributes about the candidates (see Table 1), but they did differ with regard to the specific unshared items that this participant received. Taken together, an X, a Y, and a Z profile in combination contained the full candidate information. The participants were then asked to deeply elaborate the information about the candidates and memorize it because later on during the discussion they would not have access to the candidate information sheets. To support this elaboration and fixation phase, we had participants copy the attributes onto the information evaluation questionnaire word by word and rate each attribute with regard to how positive or negative it was for the suitability of the particular candidate. We allocated 15 min for this task. Afterward, the participants had an additional 10 min to learn the information. Finally, they were asked to indicate on a separate questionnaire which of the candidates they individually preferred. All information sheets and questionnaires were then collected by the experimenter. On the basis of the individual preference questionnaires, 3-person groups were assembled by the experimenter. The assignment of participants to groups was conducted as randomly as possible, but with some restrictions. Each group had to consist of 1 X, 1 Y, and 1 Z member. In addition, because a random assignment is least likely to lead to homogeneous preferences (with three alternatives being equally attractive in the begin-
ning), the experimenters were instructed to form homogeneous groups whenever possible. A slightly larger number of homogeneous groups compared with the other conditions was also intended because in the statistical analyses this single condition would be contrasted with the mean of two other conditions (pure minority dissent and pure full diversity dissent). Finally, because we could not influence how many participants would prefer Candidate C, forming groups with proponents (for this best alternative) was also partially a nonrandom process. Each 3-person group was seated at a table in a different room and had a separate experimenter running the group session. This experimenter took care that the 3 members took their places in accordance with the predetermined seating plan for Members X, Y, and Z. The experimenter started the video camera and ensured that the group members read through an “instruction to group discussion” sheet that had previously been handed out. In accordance with previous hidden profile research, these instructions emphasized that only part of the group members’ individual information was identical and that each group member also had some information that was unique. In addition, it was emphasized that on the basis of the full information set held within the group, one of the candidates clearly was the best choice and that it was the group’s task to find out this correct solution (which, again, was the instruction that had been used in most previous hidden profile studies). If the group arrived at the correct choice in the end, each group member would be entered into a raffle and could win 1 of 25 music CD vouchers. A unanimous group decision was required. After the experimenter had ensured that all group members had understood these instructions, the group started its discussion. No time limit was set. However, if a decision had not been reached within 45 min, the experimenter briefly interrupted the discussion and pointed out that it was now time to make the final decision. This occurred in only five groups (with the longest discussion taking 55 min). When the group stated that the final decision had been made, the experimenter handed out a questionnaire on which this decision was noted. Afterward, the 3 members were separated and seated at different tables. Each member was given a recall questionnaire on which they wrote down all attributes about each candidate that they could remember. Thereafter, the experiment was finished. The experimenter thanked the participants, gave them their participation credit, and briefly explained the theoretical background of the experiment. Participants also entered their e-mail address on a list before they were dismissed. When the whole study was completed, participants received a document via e-mail indicating the correct solution, the code numbers of those persons who had won the CD vouchers, and additional information about the study. On average, the whole experimental session took about 100 min.
Dependent Measures The main dependent variable was decision quality, which was dichotomous (choice of the optimal candidate vs. choice of one of the suboptimal candidates) and could be directly derived from the group decision questionnaire. Those dependent measures that were expected to mediate dissent effects on decision quality were derived from the discussion videotapes. Discussion time (as one indicator of discussion intensity) was directly available from the videos. To assess information exchange, two coders trained in coding the discussion content but blind to the experimental hypotheses analyzed the videotapes. Coding was done by noting the number of each item as it was mentioned and marking who the speaker was on a specially designed form. The coders received a written manual with specific instructions and the coding criteria. The criteria defined which deviations from the original wording were tolerable for an item to be counted as a correct mentioning. In addition, for a statement to be counted as a correct mentioning the group member had to link the information to the corresponding candidate explicitly or by context. If one of these criteria was not fulfilled, an item was not coded. If an item had been mentioned by some other group member before or had previously been mentioned by the
DISSENT AND GROUP DECISION QUALITY
no reliable effects of participants’ ages or the groups’ gender composition on the main dependent variables were found. We first report analyses of the decision quality measure to test our central hypotheses regarding the beneficial effects of dissent on decision quality. Then, we report analyses of discussion intensity and discussion bias dependent on dissent to test whether these variables qualify for mediation in accordance with our hypotheses. In both parts of the analyses, overall tests of the complete experimental design are followed by planned comparisons to test for our three dissent effects (pure dissent effect, proponent dissent effect, and magnitude-of-dissent effect). In the final part of the section, mediation analyses are reported for the previously identified possible mediators.
same group member, with at least one other item having been mentioned in between, it was coded as a repetition. One coder coded all 116 discussions that were subsequently entered into data analyses (see the Results section). To estimate coding reliability, a second coder independently coded 22 of these discussions, randomly selected from each condition with largely equal numbers across conditions. The two coders agreed on 87% of coded statements (including both the introduction and the repetition of information). For the 13% of the cases in which the second coder differed from the first coder, no systematic pattern or bias was observed. For the data analyses, the data from the first coder were used. From these data, all dependent variables with regard to discussion content were derived. These include proportion of mentioned information and repetition rates for shared versus unshared information and for each of the four alternatives. Proportions of mentioned information were obtained by counting the number of items per category (e.g., shared information) that were mentioned at least once in the group and dividing it by the total number of items available in that category (e.g., 16 in the case of shared information). Repetition rates were obtained by counting the number of all repetitions per group within a category (e.g., shared information) and dividing it by the number of items from that category that were mentioned at least once. From these separate measures, the discussion intensity and discussion bias variables were computed (see the Results section). From the final individual recall questionnaire, the information gain was derived. An information gain was counted each time a group member correctly recalled an item he or she had not received prior to discussion. Thus, the information gain shows how much unshared information the particular group member learned from other members during discussion.
Decision Quality Overall, 59 of 135 groups (44%) made the correct decision. A chi-square analysis of decision quality (1 ⫽ solved, 0 ⫽ not solved) across the six experimental conditions revealed significant differences between these conditions, 2(5, N ⫽ 135) ⫽ 51.31, p ⬍ .001; the corresponding percentages are given in Figure 2. To clarify these differences, planned comparisons were conducted. First, the full information condition differed significantly from the hidden profile conditions, 2(1, N ⫽ 135) ⫽ 28.48, p ⬍ .001. Whereas all 19 groups (100%) in the full information condition chose the correct candidate, only 40 of the 116 groups (35%) in the hidden profile conditions made the correct choice. This replication of the well-known hidden profile effect confirms that our decision case worked as intended. To test the pure dissent effect, we compared the pure minority dissent condition and the pure full diversity dissent condition with
Results Of the 149 three-person groups in the sample, 14 groups (from all experimental conditions) had to be discarded because of technical problems with the videotapes. For the remaining 135 groups,
Percentage of correct solutions
100%
1085
100%
80% 65% 59%
60%
40% 28%
25%
20% 7% 0% FI
HP Hom.
HP Min(-)
HP Div(-) HP Min(+) HP Div(+)
Figure 2. Percentage of correct group decisions dependent on dissent. FI ⫽ full information, all group members received all information, no hidden profile; HP Hom. ⫽ hidden profile, homogeneous preferences with no proponent of the correct solution; HP Min(-) ⫽ hidden profile with minority–majority distribution of preferences and no proponent of the correct solution; HP Div(-) ⫽ hidden profile with full diversity distribution of preferences and no proponent of the correct solution; HP Min(⫹) ⫽ hidden profile with minority–majority distribution of preferences and one proponent of the correct solution; HP Div(⫹) ⫽ hidden profile with full diversity distribution of preferences and one proponent of the correct solution.
SCHULZ-HARDT ET AL.
1086
Table 2 Means for Discussion Intensity Measures Dependent on Experimental Condition Experimental condition Pure minority dissent (n ⫽ 26)
Homogeneity (n ⫽ 28)
Pure full diversity dissent (n ⫽ 20)
Minority dissent with proponent (n ⫽ 20)
Full diversity dissent with proponent (n ⫽ 22)
Measure
M
SD
M
SD
M
SD
M
SD
M
SD
Average proportion of information introduced Average repetition rate of information Discussion time (min)
.54 1.06 15.18
.22 0.94 13.41
.69 2.39 24.31
.14 1.10 8.01
.73 2.39 27.15
.09 1.05 9.10
.66 2.11 23.45
.14 0.99 10.04
.75 2.23 28.32
.11 0.89 12.36
the homogeneity condition. This comparison was significant, 2(1, N ⫽ 74) ⫽ 4.07, p ⫽ .044. Whereas only 2 of 28 (7%) homogeneous groups made the correct choice, 12 of 46 (26%) pure dissent groups (with either minority or full diversity dissent) solved the hidden profile. The proponent dissent effect was also significant, 2(1, N ⫽ 88) ⫽ 11.48, p ⫽ .001. Compared with the abovementioned 12 of 46 (26%) pure dissent groups, 26 of 42 (62%) dissent groups (either minority or full diversity dissent) with a proponent for the correct alternative were successful in solving the hidden profile. In contrast, no significant magnitude-of-dissent effect occurred, 2(1, N ⫽ 88) ⫽ 0.03, p ⫽ .953. Solution rates were almost identical for minority dissent groups with or without a proponent (20 of 46 were correct; 44%) and for full diversity groups with or without a proponent (18 of 42 were correct; 43%).
Discussion Intensity For all of the following analyses, only the five hidden profile conditions were considered because only in those conditions could mediators for the dissent effects on decision quality be identified (in the full information condition, information exchange is relatively meaningless for the final decision, because all group members have all information from the beginning and, thus, overwhelmingly start the discussion with the correct solution in mind). Discussion intensity was assessed by three indicators, namely average proportion of information mentioned, average repetition rate of information, and discussion time. The average proportion of information mentioned was the unweighted mean of the proportions of mentioned shared and unshared information (similarly for the average repetition rate).4 In one-factorial analyses of variance (ANOVAs), significant effects of the dissent factor emerged for each of the three variables: F(4, 111) ⫽ 7.02, p ⬍ .001, 2 ⫽ .20, for average proportion of information mentioned; F(4, 110) ⫽ 8.13, p ⬍ .001, 2 ⫽ .23, for average repetition rate of information; and F(4, 111) ⫽ 5.75, p ⬍ .001, 2 ⫽ .17, for discussion time.5 The corresponding means and standard deviations are given in Table 2. Planned comparisons revealed that the dissent groups differed significantly from the homogeneous groups on each of these three measures: Groups in the four dissent conditions introduced a higher proportion of information into discussion (M ⫽ 71%) than did homogeneous groups (M ⫽ 54%), t(32.097)6 ⫽ 3.66, p ⫽ .001; dissent groups repeated mentioned information more often (M ⫽ 2.28) than did homogenous groups (M ⫽ 1.06), t(110) ⫽ 5.56, p ⬍
.001; and dissent groups also spent longer in discussion (M ⫽ 25.81 min) than did homogeneous groups (M ⫽ 15.18 min), t(111) ⫽ 4.50, p ⬍ .001. Within the dissent groups, groups with a proponent for the correct solution did not differ from groups without such a proponent on any of the three measures (兩t兩s ⬍ 1.10, ps ⬎ .30). However, full diversity dissent led to somewhat higher discussion intensity than did minority dissent: Groups with full diversity dissent introduced even more information (M ⫽ 74%) than did groups with minority dissent (M ⫽ 67%), t(76.01) ⫽ 2.54, p ⫽ .013. For discussion time, the trend is in the same direction (M ⫽ 27.74 min for full diversity dissent; M ⫽ 23.88 min for minority dissent) but falls short of significance, t(111) ⫽ 1.65, p ⫽ .101. No such trend was observed for the repetition rate of information (M ⫽ 2.25 for minority dissent; M ⫽ 2.31 for full diversity dissent), t(110) ⫽ 0.30, p ⫽ .77.
4 Usually, the proportion of mentioned information is calculated by simply counting how many items are mentioned at least once and dividing this number by the total number of items available. However, because our decision case contains more unshared than shared items (24 vs. 16), this would lead to an interdependence of discussion intensity and discussion bias measures: If discussion bias is reduced by facilitating unshared compared with shared information, this would also increase the proportion of information mentioned (same argument for repetitions). Our measures avoid this problem. 5 As the degrees of freedom indicate, in this as well as several following analyses of information exchange, one or more cases were lost because none of the items in question were exchanged or repeated in these groups. For example, in one of the homogeneous groups, no information was mentioned at all (discussion consisted only of exchanging preferences and subsequently choosing the candidate that was favored by all members), so the repetition rate for this group (as well as all bias measures) could not be calculated. Separate analyses for shared and unshared information confirmed that dissent facilitates the introduction of both shared and unshared information, F(4, 111) ⫽ 5.12, p ⫽ .001, 2 ⫽ .16, and F(4, 111) ⫽ 6.90, p ⬍ .001, 2 ⫽ .20, as well as the repetition of both shared and unshared information, F(4, 110) ⫽ 5.55, p ⬍ .001, 2 ⫽ .17, and F(4, 110) ⫽ 8.19, p ⬍ .001, 2 ⫽ .23. 6 If fractional degrees of freedom occur, this is due to correction for nonhomogeneous variances. Variances were considered to be nonhomogeneous if p ⬍ .10 in the corresponding test for homogeneity of variances.
DISSENT AND GROUP DECISION QUALITY
1087
Table 3 Means for Discussion Bias Measures Dependent on Experimental Condition Experimental condition
Homogeneity (n ⫽ 28)
Pure minority dissent (n ⫽ 26)
Pure full diversity dissent (n ⫽ 20)
Minority dissent with proponent (n ⫽ 20)
Full diversity dissent with proponent (n ⫽ 22)
Measure
M
SD
M
SD
M
SD
M
SD
M
SD
Proportion of shared information introduced Proportion of unshared information introduced Introduction bias in favor of shared information Repetition rate of shared information Repetition rate of unshared information Repetition bias in favor of shared information Repetition rate of preference-consistent information Repetition rate of preference-inconsistent information Repetition bias in favor of preference-consistent information
.67 .42 .61 1.22 0.90 .64 1.18 0.93 .65
.29 .19 .09 1.00 1.00 .21 0.94 1.00 .21
.80 .57 .59 2.61 2.16 .53 2.57 2.19 .54
.14 .15 .05 1.62 0.99 .13 1.36 1.02 .09
.87 .59 .60 2.66 2.13 .55 2.66 2.17 .56
.10 .13 .06 1.44 0.88 .11 1.19 0.98 .08
.76 .56 .58 2.30 1.92 .54 2.30 1.92 .54
.16 .15 .06 1.14 0.90 .07 1.10 1.00 .13
.87 .63 .58 2.50 1.96 .56 2.35 2.11 .53
.12 .12 .04 1.12 0.83 .08 0.96 0.92 .08
Discussion Bias Discussion bias was calculated separately for shared versus unshared and for preference-consistent versus preferenceinconsistent information. Shared versus unshared information. Overall, a higher proportion of shared information (M ⫽ 79%) than unshared information (M ⫽ 55%) was introduced into discussion, F(1, 111) ⫽ 310.90, p ⬍ .001, 2 ⫽ .74. In addition, shared information, once it had been introduced into discussion, was repeated more often (M ⫽ 2.22) than unshared information (M ⫽ 1.78), F(1, 110) ⫽ 21.24, p ⬍ .001, 2 ⫽ .23. In accordance with Stasser, Vaughan, and Stewart (2000), the bias in favor of shared information was calculated by dividing the introduction (repetition) rate of shared information by the sum of the introduction (repetition) rates for shared and unshared information. This bias measure ranges between 0 and 1; a value of .50 indicates that discussion is unbiased. The larger the value, the more the discussion is biased toward shared information. The average introduction bias in favor of shared information was .59, which is significantly different from .50, t(114) ⫽ 15.71, p ⬍ .001. In an overall ANOVA, the effect for the experimental conditions was not significant, F(4, 110) ⫽ 1.10, p ⫽ .360, 2 ⫽ .04. The corresponding means are shown in Table 3. Planned comparisons revealed a marginal difference between homogeneous groups and dissent groups, t(110) ⫽ 1.74, p ⫽ .084: Dissent groups (M ⫽ .59) had a somewhat lower bias toward shared information than did homogeneous groups (M ⫽ .61). The comparisons within the dissent groups (minority dissent vs. full diversity dissent; dissent with proponent vs. dissent without proponent) did not reach significance (兩t兩s ⬍ 1.10, ps ⬎ .28). The average repetition bias was .57; again, this bias significantly differs from .50, t(113) ⫽ 5.11, p ⬍ .001. With regard to this bias, significant overall differences were found in the ANOVA, F(4, 109) ⫽ 2.61, p ⫽ .040, 2 ⫽ .09. The corresponding means are also shown in Table 3. Planned comparisons revealed that homogenous groups had a larger repetition bias (M ⫽ .64) than did groups with
dissent (M ⫽ .54), t(28.52) ⫽ 2.25, p ⫽ .032. Again, no significant differences were found within the dissent groups (兩t兩s ⬍ 0.80, ps ⬎ .42). Preference-consistent versus preference-inconsistent information. Because the preference consistency of information depends on the individual prediscussion preference of the speaker and because in most conditions these preferences are not homogeneous, it is not possible to determine the proportion of discussed preference-consistent and preference-inconsistent information similarly to the case of shared and unshared information. The problem is that the same piece of information that is consistent for one member can be inconsistent for another member. Hence, if one member introduces a piece of information that is preference consistent for him or her, this reduces another member’s possibilities for preference-inconsistent information introductions. After careful investigation, we decided that in a hidden profile situation there is no appropriate way to calculate a preference-consistency bias for the introduction of information that (a) tells us whether there is in fact a bias toward preference-consistent information (i.e., that leads to an unequivocal reference value for an unbiased discussion against which the empirical bias values can be tested) and (b) allows for a fair test between the experimental conditions.7 Hence, we concentrated on information repetitions when analyzing discussion bias toward preference-consistent information. A preference-consistent repetition was counted each time a group member repeated an advantage of his or her preferred candidate or a disadvantage of the nonpreferred candidates. The 7 To give just one illustration, the typical form of preference-consistent information introduction is to mention the advantages of the preferred candidate. Now, if we consider homogeneous groups in our experiment, these are the same four advantages for all three members. Hence, each group member can only mention one third of these advantages on average. In contrast to that, in a full diversity dissent group, each group member can mention all four advantages of her or his preferred candidate. One might try to solve this problem by calculating the introduction bias only on the basis of the unshared items (see also Dennis, 1996). However, in this case, no fair comparison with the proponent dissent groups is possible, because for the proponents all unshared items are preference consistent.
SCHULZ-HARDT ET AL.
1088
amount of preference-consistent repetitions per member was divided by the total amount of mentioned information that was preference consistent for this member, and the resulting values for the 3 members were averaged. Hence, this repetition rate tells us how often, on average, a mentioned piece of information was repeated as preference-consistent information (and vice versa for preference-inconsistent repetitions). Overall, the repetition rate was higher for preference-consistent information (M ⫽ 2.17) than for preference-inconsistent information (M ⫽ 1.83), F(1, 110) ⫽ 27.81, p ⬍ .001, 2 ⫽ .23. To calculate the repetition bias similarly to the case of shared versus unshared information, the repetition rate for preference-consistent information was divided by the sum of the two repetition rates. The average bias was .57, which is significantly different from .50, t(113) ⫽ 5.07, p ⬍ .001. An overall ANOVA of the experimental conditions shows significant differences with regard to the repetition bias, F(1, 109) ⫽ 3.71, p ⫽ .007, 2 ⫽ .12; for the corresponding means, see Table 3. Planned comparisons revealed that dissent groups (M ⫽ .54) had a lower repetition bias than did homogeneous groups (M ⫽ .65), t(28.30) ⫽ 2.61, p ⫽ .014. Within the dissent groups, no significant differences were found (兩t兩s ⬍ 0.66, ps ⬎ .50).
Discussion About Candidate C (Correct Choice) Discussion about the correct choice (introduction and repetition of information about Candidate C) was analyzed in two 5 ⫻ 1 ANOVAs of the experimental design. In the overall analyses, significant effects of the dissent factor emerged for both dependent variables: F(4, 111) ⫽ 10.25, p ⬍ .001, 2 ⫽ .27, for proportion of information introduced about Candidate C, and F(4, 107) ⫽ 5.88, p ⬍ .001, 2 ⫽ .18, for repetition rate of information about Candidate C. The corresponding means and standard deviations are given in Table 4. Planned comparisons revealed that dissent groups without a proponent for Candidate C introduced more information about Candidate C (M ⫽ 58%) than did homogeneous groups (M ⫽ 46%), t(45.47) ⫽ 1.99, p ⫽ .053, and repeated information about Candidate C more often (M ⫽ 1.67) than did homogeneous groups (M ⫽ 1.04), t(107) ⫽ 1.70, p ⫽ .093. Compared with dissent groups without a proponent for Candidate C, dissent groups with such a proponent introduced even more information about that candidate (M ⫽ 77%), t(69.07) ⫽ 4.99, p ⬍ .001, and repeated such information even more often (M ⫽ 2.75), t(107) ⫽ 3.40, p ⫽
.001. No differences were found between minority dissent and full diversity dissent (兩t兩s ⬍ 0.78, ps ⬎ .44).
Mediation Analyses We now report the tests of our proposed mediational chains that were illustrated in Figure 1. In our mediation analyses, we followed the R. M. Baron and Kenny (1986) approach, but by reporting regression coefficients as well as t values, we provide the interested reader with sufficient statistical information to recalculate the analyses according to at least some of the alternative approaches that have recently been reviewed and compared by MacKinnon, Lockwood, Hoffman, West, and Sheets (2002). To maximize comparability across steps, we report beta weights and statistics from linear regressions throughout all of the following analyses, although in the case of decision quality the criterion is dichotomous. However, if binary logistic regression is used for the latter cases, similar results are obtained. Dissent effects on decision quality. We first tested whether the pure dissent effect on decision quality is mediated by discussion about Candidate C. Therefore, the two measures for discussion about Candidate C (introduction and repetition rate) were z transformed and averaged. In cases in which repetition rates could not be calculated (because the group did not mention any information about Candidate C—see also Footnote 5), the means of the corresponding experimental conditions were assigned to these cases (a similar procedure was followed in all subsequent analyses). The internal consistency of this scale was ␣ ⫽ .759. If decision quality is regressed on the pure dissent contrast (homogeneous groups vs. dissent groups without a proponent for Candidate C), this contrast receives a significant weight,  ⫽ .235, t(72) ⫽ 2.05, p ⫽ .044, showing that higher solution rates are found in dissent groups than in homogeneous groups. The same occurs if average discussion about Candidate C is regressed on this contrast,  ⫽ .263, t(72) ⫽ 2.32, p ⫽ .023, showing that dissent groups discussed more about Candidate C than did homogeneous groups. If decision quality is regressed on both discussion about Candidate C and the pure dissent contrast, F(2, 71) ⫽ 29.05, p ⬍ .001, discussion about Candidate C receives a significant weight,  ⫽ .652, t(71) ⫽ 7.14, p ⬍ .001, whereas the weight for the dissent contrast is near zero and no longer significant,  ⫽ .063, t(71) ⫽ 0.69, p ⫽ .492. Hence, the pure dissent effect on decision quality is mediated by the amount of discussion about the correct alternative.
Table 4 Means for Discussion About Candidate C (Correct Choice) and Information Gain Dependent on Experimental Condition Experimental condition
Homogeneity (n ⫽ 28)
Pure minority dissent (n ⫽ 26)
Pure full diversity dissent (n ⫽ 20)
Minority dissent with proponent (n ⫽ 20)
Full diversity dissent with proponent (n ⫽ 22)
Measure
M
SD
M
SD
M
SD
M
SD
M
SD
Proportion of information introduced about Candidate C Repetition rate of information about Candidate C Information gain
.46 1.04 5.21
.26 1.07 4.46
.58 1.72 9.42
.18 1.68 5.07
.58 1.63 9.10
.22 1.67 4.04
.77 2.95 9.10
.13 1.65 5.00
.77 2.55 10.55
.15 1.24 4.40
DISSENT AND GROUP DECISION QUALITY
Regressing decision quality on the proponent dissent contrast (dissent groups with vs. without a proponent for Candidate C) leads to a significant weight for this contrast,  ⫽ .361, t(86) ⫽ 3.59, p ⫽ .001, indicating that more correct solutions were found in dissent groups with a proponent for the correct choice than in dissent groups without such a proponent. A significant weight is also obtained if average discussion about Candidate C is regressed on this contrast,  ⫽ .434, t(86) ⫽ 4.47, p ⬍ .001, showing that dissent groups with a proponent for Candidate C discussed more about this candidate than did dissent groups without such a proponent. Finally, regressing decision quality on both the proponent dissent contrast and discussion about Candidate C, F(2, 85) ⫽ 32.93, p ⬍ .001, leads to a significant weight for discussion about Candidate C,  ⫽ .614, t(85) ⫽ 6.80, p ⬍ .001, whereas the weight for the proponent dissent contrast is reduced by more than two thirds and no longer significant,  ⫽ .095, t(85) ⫽ 1.05, p ⫽ .299. Hence, the proponent dissent effect on decision quality is also mediated by discussion about Candidate C. Dissent effects on discussion of Candidate C. In the next step, we tested whether the increased discussion about Candidate C in dissent groups without a proponent (compared with the discussion in homogeneous groups) is due to more discussion intensity and less discussion bias in these groups (cf. Figure 1). Therefore, two overall indices for discussion intensity and discussion bias were calculated. To avoid a logical dependence between the criterion (discussion about Candidate C) and the mediators, we removed discussion about Candidate C from the information exchange measures constituting intensity and bias. For the discussion intensity index, the average proportion of mentioned information and the average repetition rate of information as well as discussion time were z transformed and averaged. The internal consistency of this scale was ␣ ⫽ .773. For the average discussion bias, first the two sharedness biases (introduction and repetition) were averaged, and the resulting measure was then averaged with the preferenceconsistency bias (giving sharedness and preference consistency an equal weight in this index). The scale had an internal consistency of ␣ ⫽ .784. The correlation between discussion intensity and discussion bias was ⫺.371 ( p ⬍ .001), indicating that more intensive discussions were, on average, less biased. However, because of the moderate size of the correlation, it is empirically justified to test them separately. As already reported, pure dissent groups discussed more about Candidate C than did homogeneous groups,  ⫽ .263, t(72) ⫽ 2.32, p ⫽ .023. With regard to the mediators, pure dissent groups had a higher discussion intensity,  ⫽ .566, t(72) ⫽ 5.82, p ⬍ .001, and a lower discussion bias,  ⫽ ⫺.242, t(72) ⫽ ⫺2.12, p ⫽ .037, than did homogeneous groups. If discussion about Candidate C is regressed on both the pure dissent contrast and discussion intensity, F(2, 71) ⫽ 15.44, p ⬍ .001, the regression weight for discussion intensity is significant,  ⫽ .586, t(71) ⫽ 4.88, p ⬍ .001, whereas the influence of the pure dissent contrast is completely eliminated,  ⫽ ⫺.068, t(71) ⫽ ⫺0.57, p ⫽ .571. Thus, discussion intensity mediates the pure dissent effect on discussion about Candidate C. If the same analysis is conducted with discussion bias instead of discussion intensity, F(2, 71) ⫽ 5.94, p ⫽ .004, the discussion bias also receives a significant regression weight,  ⫽ ⫺.280, t(71) ⫽ ⫺2.50, p ⫽ .016, indicating that less discussion bias is associated with more discussion about Candidate C. The weight for the pure
1089
dissent contrast is reduced and no longer significant,  ⫽ .195, t(71) ⫽ 1.73, p ⫽ .089, indicating a weak partial mediation. However, discussion intensity seems to be the more powerful of these two mediators. If discussion of Candidate C is regressed on both mediators as well as the pure dissent contrast, F(3, 70) ⫽ 10.37, p ⬍ .001, only discussion intensity receives a significant weight,  ⫽ .546, t(70) ⫽ 4.08, p ⬍ .001, whereas the weight for discussion bias is nonsignificant,  ⫽ ⫺.078, t(70) ⫽ ⫺0.68, p ⫽ .497. As before, the influence of the pure dissent contrast is eliminated,  ⫽ ⫺.065, t(70) ⫽ ⫺0.54, p ⫽ .594. As predicted, the proponent dissent effect on discussion about Candidate C is not mediated by discussion intensity or discussion bias: If discussion about Candidate C is regressed on both variables and the proponent dissent contrast, the weight for this contrast ( ⫽ .532, p ⬍ .001) is not reduced when compared with a simple regression ( ⫽ .434, p ⬍ .001). Overall mediation analysis for the pure dissent effect. As a final test for our proposed mediational chain for the pure dissent effect (see Figure 1), all three mediators were entered in a multiple regression analysis as predictors together with the pure dissent contrast with decision quality as the criterion, F(4, 69) ⫽ 15.76, p ⬍ .001. Confirming our predictions, only the proximal mediator, namely discussion about Candidate C, received a significant weight,  ⫽ .746, t(69) ⫽ 7.13, p ⬍ .001, whereas nonsignificant weights were obtained for discussion intensity,  ⫽ ⫺.244, t(69) ⫽ ⫺1.87, p ⫽ .066; discussion bias,  ⫽ ⫺.035, t(69) ⫽ ⫺0.35, p ⫽ .727; and the pure dissent contrast,  ⫽ .168, t(69) ⫽ 1.58, p ⫽ .118. Although the weight for discussion intensity is marginal, it should be noted that the sign has changed, indicating that the facilitative effect of discussion intensity on the solution of hidden profiles completely vanishes if discussion about Candidate C is controlled for.
Additional Findings Information gain. Information gain (i.e., the amount of unshared information per group that the members had learned from each other, as evident from the recall questionnaire) was analyzed in a one-factorial ANOVA of the experimental design. A significant overall effect emerged, F(4, 111) ⫽ 5.04, p ⫽ .001, 2 ⫽ .15; the means are displayed in Table 4. Planned comparisons revealed that dissent groups (M ⫽ 9.54) had a higher information gain than did homogeneous groups (M ⫽ 5.21), t(111) ⫽ 4.31, p ⬍ .001. Within the dissent groups, no significant differences were found (兩t兩s ⬍ 0.57, ps ⬎ .57). An additional mediation analysis shows that this higher information gain in dissent groups is a consequence of their more intense discussion of unshared information: The dissent contrast (dissent groups vs. homogeneous groups) receives a significant weight in predicting both information gain,  ⫽ .378, t(114) ⫽ 4.36, p ⬍ .001, and discussion of unshared information (average of proportion of mentioned unshared information and repetition rate of unshared information, both z transformed),  ⫽ .518, t(114) ⫽ 6.46, p ⬍ .001. If information gain is regressed on both the dissent contrast and discussion of unshared information, only the latter receives a significant weight,  ⫽ .549, t(113) ⫽ 6.27, p ⬍ .001, whereas the effect of the dissent contrast is largely eliminated,  ⫽ .093, t(113) ⫽ 1.07, p ⫽ .289. Majorities versus minorities. In two of the five hidden profile conditions (the minority dissent conditions), majority members
SCHULZ-HARDT ET AL.
1090
could be compared with minority members; in one of these two conditions the minority member was the proponent of the correct solution. Neither with respect to discussion intensity (proportion of information introduced, repetition rate of information) nor with respect to discussion bias (in favor of shared and/or preferenceconsistent information) or information gain did we find any significant differences between minority and majority members ( ps ⬎ .20).
Discussion The goals of the present hidden profile experiment were threefold: (a) to provide an unequivocal test of the effects of prediscussion dissent on group decision quality (pure dissent effect) while controlling for the effects of a proponent for the correct solution within a group (proponent dissent effect); (b) to test for differential effects of minority versus full diversity dissent on group decision making (magnitude-of-dissent effect); and (c) to examine the extent to which discussion intensity, discussion bias, and the amount of information about the correct decision alternative discussed mediate dissent effects on group decision quality. In line with our hypotheses, we found that both minority and full diversity dissent have positive effects on group discussion and group decision quality in hidden profile situations, even if none of the dissenting pre-discussion preferences are in favor of the correct alternative. When a dissent group contains a proponent of the correct solution, the likelihood that the hidden profile is solved is even higher as compared with dissent groups in which all members enter group discussion with a suboptimal preference. Counter to our expectations, the dissent effects on decision quality were not affected by the amount of dissent. Our experiment also demonstrates how decision quality is improved by dissent. As predicted, both the pure dissent effect and the proponent dissent effect were mediated by discussion about the correct candidate: Because dissent groups exchanged more information about the best alternative and repeated it more often, they were more likely to solve the hidden profile. However, the two dissent effects differ with regard to how these increases in discussion about the best alternative are achieved. The proponent dissent effect is directly mediated via discussion about the best candidate. Thus, a proponent who happens to prefer the correct alternative (often termed best member within a group) helps group decision making mainly by keeping discussions about the correct decision alternative alive. In contrast, in diverse groups that are “blind” to the correct alternative because none of their members propose the correct alternative, the increase in discussion about the correct candidate is achieved via two different (although moderately intercorrelated) mechanisms. On the one hand, diversity in prediscussion preferences results in higher levels of discussion intensity, that is, more information is introduced and repeated during group discussion, which also takes more time. As a consequence, the group also introduces and discusses more information about an alternative that initially had not been preferred by any group member— namely the best alternative. On the other hand, groups with prediscussion dissent conduct a less biased discussion than do groups without prediscussion dissent, that is, members of preferencediverse groups focus less on information that is shared and consistent with their initial preferences. It is not surprising that this
debiasing mainly affected the repetition rather than the introduction of information. Dissent should make group members more open to new (unshared) and inconsistent information, but at least in the case of unshared information, this greater openness and receptivity cannot directly affect the introduction of information because before a piece of information is introduced into discussion no group member knows whether it is shared. As a consequence of this greater openness to new and inconsistent information, the group is also more open to discuss an alternative that had not been preferred by any group member before discussion—the best alternative. Mediation analyses revealed discussion intensity to be the more important mediator. Discussion intensity received a stronger regression weight in separate mediation analyses than discussion bias did, and in the common mediation analysis, discussion intensity also mediated on its own whereas discussion bias mediated only in conjunction with discussion intensity (i.e., the variance common to discussion intensity and discussion bias is relevant for the criterion). On the basis of the pattern of results, one might even suspect that the mediation effect of discussion bias is spurious and only discussion intensity matters. However, what strongly speaks against this interpretation is the fact that in several hidden profile studies, pure increases in discussion intensity did not result in significant effects on group decision quality (e.g., Hollingshead, 1996; Mennecke, 1997). The evidence from these studies supports a different interpretation within which a reduction in discussion bias is a necessary ingredient alongside higher discussion intensity for increasing solution rates in a hidden profile. Although the focus of our experiment was on decision quality, prediscussion dissent was also shown to facilitate individual learning of new information not held before discussion (information gain), which is mediated by an increased proportion of unshared information discussed. Such improved knowledge acquisition brought about by dissent may, for example, be helpful in the implementation phase of a group decision by aiding group members to better anticipate consequences of their decision.
Theoretical Implications Pure dissent effect. The results of our experiment provide what is, to the best of our knowledge, the first methodologically sound empirical demonstration that group decision quality benefits from prediscussion dissent independent of the quality of the members’ individual judgments or preferences. In contrast to previous studies (Brodbeck et al., 2002; Hightower & Sayeed, 1996; Sniezek & Henry, 1989; Wanous & Youtz, 1986), the design of our experiment completely rules out the alternative explanation that the dissent effect may be based on the fact that an increase in preference diversity also increases the likelihood that at least one of the members prefers the optimal or a near optimal solution. An important implication of this finding is that 20 years after its publication, the second central idea expressed in Stasser and Titus’ (1985) seminal article has proven to be valid—namely that conflict initiated by prediscussion dissent among group members facilitates the solution of hidden profiles even if no group member initially favors the best alternative. We see at least two plausible explanations for why we were able to show an effect that Stasser and Titus (1985) failed to demonstrate. On the one hand, our method of directly manipulating the
DISSENT AND GROUP DECISION QUALITY
preference distribution in the group may have been stronger and more successful than Stasser and Titus’s method of manipulating dissent indirectly via the information distribution in the group— even more so given that the latter manipulation seems to have failed in most groups (cf. Stasser & Titus, 1985, Table 4, p. 1474). On the other hand, Stasser and Titus aimed to investigate dissent brought about by two conflicting factions, each consisting of two persons with homogeneous preferences, whereas we investigated minority–majority compositions and groups composed of 3 persons who all differed with regard to their individual prediscussion preference. It is possible that “faction dissent” leads to different influence processes (e.g., less minority influence) than the preference distributions that we investigated. Systematically testing for such differences would be an interesting topic for further research. Two other questions also deserve further investigation. First, although full diversity dissent led to somewhat higher discussion intensity than did minority dissent, no differences in decision quality were obtained. However, it should be noted that the groups in our experiment operated in a context that was strongly facilitative for the solution of hidden profiles: For example, on the basis of the full information, the correct alternative was vastly superior to the other alternatives, so that even the somewhat lower discussion intensity instigated by minority dissent (as compared with full diversity dissent) might be sufficient to detect this superiority. Furthermore, prior to discussion, the suboptimal alternatives were equal with regard to the number of advantages and disadvantages and only had a slight advantage over the best candidate. Thus, initial preferences should have been relatively weak, so that even one dissenting opinion might have been sufficient to create doubts about their correctness. Finally, groups operated under strictly cooperative goals, whereas in real groups, members often have vested interests in the “success” of their preferred alternative (Wittenbaum et al., 2004). It is, thus, an interesting question for subsequent studies whether full diversity dissent might be superior to minority dissent in less-than-optimal contexts for choosing the best alternative. The second question concerns the complete absence of differences between minorities and majorities in our experiment. On the basis of the dominant theories about minority and majority influence (e.g., Moscovici, 1980; Nemeth, 1986), we should have expected minority influence to facilitate more discussion intensity, less discussion bias, and more information gain than majority influence. However, research that has shown differences between minority and majority influence has usually been conducted either completely outside the dynamic context of face-to-face discussions or in discussion settings with confederates who acted as minority or majority members (for an exception, see, e.g., Smith, Tindale, & Dugoni, 1996). Hence, findings from the latter context do not necessarily generalize to freely interacting groups (cf. SchulzHardt, Mojzisch, & Vogelgesang, in press), and our results indicate that more research about minority and majority influence in interactive group settings is needed. Proponent dissent effect. At first glance, the proponent dissent effect demonstrated in our experiment seems to be far less noteworthy than the pure dissent effect. Some evidence already exists showing that hidden profiles are more likely to be solved if at least one group member favors the best alternative prior to discussion (McLeod, Baron, Marti, & Yoon, 1997; Sassenberg, Boos, & Klapproth, 2001; Stewart & Stasser, 1998), and this is also what
1091
most social decision schemes (cf. Davis, 1973) would suggest. However, in contrast to some previous studies (McLeod et al., 1997; Stewart & Stasser, 1998), our proponents were not provided with full information about the decision case. Thus, we can rule out alternative explanations for their influence as, for example, higher competence or higher expert status ascribed to these proponents (cf. Wittenbaum, 1998). Furthermore, the solution rates in the proponent conditions are remarkably high (65% and 59%), given that only 1 group member favored the correct solution at the beginning. From a social decision schemes perspective, one would not expect such high solution rates unless the proponent of the correct choice can demonstrate the superiority of his or her preference to the other group members (Laughlin & Ellis, 1986)—and the proponents’ individual information sets obviously did not allow them to do that. This discrepancy can be resolved if we introduce a distinction, namely between individual and collective demonstrability. Whereas individual demonstrability is given if a proponent of the correct choice has enough individual resources to demonstrate its correctness, collective demonstrability is given if the group as a whole can generate sufficient information to show the superiority of the correct choice (cf. Laughlin & Hollingshead, 1995). In our experiment, no individual demonstrability was given, but collective demonstrability could emerge from social interaction and could explain why the proponents of the correct solution were so successful in securing their preferred candidate as the group choice.
Practical Implications Our results clearly demonstrate that group decision quality benefits from prediscussion dissent among group members. Thus, if high decision quality is required, organizations should attempt to design decision-making groups with at least some amount of prediscussion dissent among their members. Although our results do not demonstrate a superiority of full diversity dissent over minority–majority dissent (at least not with regard to decision quality), from a practical point of view one should prefer to realize full diversity dissent for two reasons. The first is that prediscussion dissent is useless if the dissenting opinions are not expressed. In organizations, members of decision-making groups often withhold diverging views (Stanley, 1981), which can be due to formal or informal communication barriers (R. A. Baron & Greenberg, 1989), evaluation apprehension (Gallupe, Bastianutti, & Cooper, 1991), or conformity pressures within the group (Janis, 1982). In a group with a highly diverse preference distribution, it should be more likely that dissent will be expressed than in a group with a large majority and a small minority faction, because less conformity pressure operates and more people can express dissent in the former than in the latter group. The second reason is that the two dissent effects that were separately demonstrated in our study, namely the pure dissent effect and the proponent dissent effect, are to some extent confounded in real-world decision making: The more diverse the preference distribution in the group is, the more likely it is that the group contains at least one member with a preference for the best alternative. Thus, maximizing prediscussion dissent raises the likelihood of capturing not only the pure dissent effect but also the proponent dissent effect. The recommendation to use preference-diverse groups to maximize group decision quality has three caveats. First, nobody
SCHULZ-HARDT ET AL.
1092
knows in advance when a decision situation constitutes a hidden profile, so one cannot design preference-diverse groups exclusively for hidden profiles. Hence, preference-diverse groups could be recommended even more firmly if further studies showed prediscussion dissent to be, at least, not detrimental for decision quality in situations other than hidden profiles. Second, prediscussion dissent is not without costs. Once established, authentic dissent is likely to result in, for example, lower cohesiveness and more conflict (e.g., Jehn, Northcraft, & Neale, 1999; Williams & O’Reilly, 1998), which increases the propensity of disengagement from the task, the group, or both. Hence, one has to consider how such drawbacks can be counteracted— candidates for this may be critical group norms (Postmes, Spears, & Cihangir, 2001) as well as the development of a shared identity valuing diversity (cf. van Knippenberg & Haslam, 2003). Finally, forming preference-diverse groups often may not be possible: For example, teams in organizations often make multiple decisions, and one cannot make sure that prediscussion dissent will be given for all of these decisions. The literature on dialectical decision techniques like devil’s advocacy or dialectical inquiry (e.g., Katzenstein, 1996) seems to indicate that contrived dissent brought about by these techniques could be a good substitute for authentic prediscussion dissent. However, Greitemeyer, SchulzHardt, Brodbeck, and Frey (2006) recently demonstrated that such a dialectical technique, although increasing discussion intensity and facilitating the exchange of unshared information, does not necessarily improve the solution of hidden profiles. Hence, future research is called for to investigate how the beneficial effects of prediscussion dissent, as obtained in our experiment, can be successfully mimicked if group composition is fixed.
References Baron, R. A., & Greenberg, J. (1989). Behavior in organizations: Understanding and managing the human side of work (3rd ed.). Boston, MA: Allyn and Bacon. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research. Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Brodbeck, F. C., Kerschreiter, R., Mojzisch, A., Frey, D., & Schulz-Hardt, S. (2002). The dissemination of critical, unshared information in decision-making groups: The effects of pre-discussion dissent. European Journal of Social Psychology, 32, 35–56. Brodbeck, F. C., Kerschreiter, R., Mojzisch, A., & Schulz-Hardt, S. (in press). Improving group decision making under conditions of distributed knowledge: The information asymmetries model. Academy of Management Review. Clark, N. K., & Stephenson, G. M. (1989). Group remembering. In P. B. Paulus (Ed.), Psychology of group influence (Vol. 2, pp. 357–391). Hillsdale, NJ: Erlbaum. Davis, J. H. (1973). Group decision and social interaction: A theory of social decision schemes. Psychological Review, 80, 97–125. De Dreu, C. K. W., & Beersma, B. (2001). Minority influence in organizations: Its origins and implications for learning and group performance. In C. K. W. De Dreu & N. K. De Vries (Eds.), Group consensus and minority influence: Implications for innovation (pp. 258 –283). Oxford, England: Blackwell. De Dreu, C. K. W., & West, M. A. (2001). Minority dissent and team innovation: The importance of participation in decision making. Journal of Applied Psychology, 86, 1191–1201.
Dennis, A. R. (1996). Information exchange and use in small group decision making. Small Group Research, 27, 532–550. Dooley, R. S., & Fryxell, G. E. (1999). Attaining decision quality and commitment from dissent: The moderating effects of loyalty and competence in strategic decision-making teams. Academy of Management Journal, 42, 389 – 402. Gallupe, R. B., Bastianutti, L. M., & Cooper, W. H. (1991). Unblocking brainstorms. Journal of Applied Psychology, 76, 137–142. Gigone, D., & Hastie, R. (1993). The common knowledge effect: Information sharing and group judgment. Journal of Personality and Social Psychology, 65, 959 –974. Greitemeyer, T., & Schulz-Hardt, S. (2003). Preference-consistent evaluation of information in the hidden profile paradigm: Beyond group-level explanations for the dominance of shared information in group decisions. Journal of Personality and Social Psychology, 84, 322–339. Greitemeyer, T., Schulz-Hardt, S., Brodbeck, F. C., & Frey, D. (2006). Information sampling and group decision making: The effects of an advocacy decision procedure and task experience. Journal of Experimental Psychology: Applied, 12, 31– 42. Hightower, R., & Sayeed, L. (1996). Effects of communication mode and prediscussion information distribution characteristics on information exchange in groups. Information Systems Research, 7, 451– 465. Hollingshead, A. B. (1996). The rank-order effect in group decision making. Organizational Behavior and Human Decision Processes, 68, 181– 193. Janis, I. L. (1982). Groupthink (2nd, rev. ed.). Boston: Houghton Mifflin. Jehn, K. A., Northcraft, G. B., & Neale, M. A. (1999). Why differences make a difference: A field study of diversity, conflict, and performance in work groups. Administrative Science Quarterly, 44, 741–763. Katzenstein, G. (1996). The debate on structured debate: Toward a unified theory. Organizational Behavior and Human Decision Processes, 66, 316 –332. Larson, J. R., Jr., Foster-Fishman, P. G., & Keys, C. B. (1994). Discussion of shared and unshared information in decision-making groups. Journal of Personality and Social Psychology, 67, 446 – 461. Laughlin, P. R., & Ellis, A. L. (1986). Demonstrability and social combination processes on mathematical intellective tasks. Journal of Experimental Social Psychology, 22, 177–189. Laughlin, P. R., & Hollingshead, A. B. (1995). A theory of collective induction. Organizational Behavior and Human Decision Processes, 61, 94 –107. MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. S. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83–104. McLeod, P. L., Baron, R. S., Marti, M. W., & Yoon, K. (1997). The eyes have it: Minority influence in face-to-face and computer-mediated group discussion. Journal of Applied Psychology, 82, 706 –718. Mennecke, B. E. (1997). Using group support systems to discover hidden profiles: An examination of the influence of group size and meeting structures on information sharing and decision quality. International Journal of Human Computer Studies, 47, 387– 405. Mojzisch, A., & Schulz-Hardt, S. (2006). Information sampling in group decision making: Sampling biases and their consequences. In K. Fiedler & P. Juslin (Eds.), Information sampling and adaptive cognition (pp. 299 –325). Cambridge, England: Cambridge University Press. Moscovici, S. (1980). Towards a theory of conversion behavior. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 13, pp. 209 –239). New York: Academic Press. Nemeth, C. J. (1986). Differential contributions of majority and minority influence. Psychological Review, 93, 23–32. Nemeth, C. J., Rogers, J. D., & Brown, K. S. (2001). Devil’s advocate vs. authentic dissent: Stimulating quantity and quality. European Journal of Social Psychology, 31, 707–729. Nemeth, C. J., & Wachtler, J. (1983). Creative problem solving as a result
DISSENT AND GROUP DECISION QUALITY of majority vs. minority influence. European Journal of Social Psychology, 13, 45–55. Parks, C. D., & Nelson, N. L. (1999). Discussion and decision: The interrelationship between initial preference distribution and group discussion content. Organizational Behavior and Human Decision Processes, 80, 87–101. Postmes, T., Spears, R., & Cihangir, S. (2001). Quality of decision making and group norms. Journal of Personality and Social Psychology, 80, 918 –930. Sassenberg, K., Boos, M., & Klapproth, F. (2001). Wissen und Problemlo¨sekompetenz: Der Einflu von Expertise auf den Informationsaustausch in computervermittelter Kommunikation [Knowledge and competence: The influence of expertise on information exchange in computer-mediated communication]. Zeitschrift fu¨r Sozialpsychologie, 32, 45–56. Schulz-Hardt, S., Frey, D., Lu¨thgens, C., & Moscovici, S. (2000). Biased information search in group decision making. Journal of Personality and Social Psychology, 78, 655– 669. Schulz-Hardt, S., Jochims, M., & Frey, D. (2002). Productive conflict in group decision making: Genuine and contrived dissent as strategies to counteract biased information seeking. Organizational Behavior and Human Decision Processes, 88, 563–586. Schulz-Hardt, S., Mojzisch, A., & Vogelgesang, F. (in press). Dissent as a facilitator: Individual and group-level effects on creativity and performance. In C. K. W. De Dreu & M. Gelfand (Eds.), The psychology of conflict and conflict management in organizations. Hillsdale, NJ: Erlbaum. Schwenk, C. R. (1990). Effects of devil’s advocacy and dialectical inquiry on decision making: A meta analysis. Organizational Behavior and Human Decision Processes, 47, 161–176. Simons, T., Pelled, L. H., & Smith, K. A. (1999). Making use of difference: Diversity, debate, and decision comprehensiveness in top management teams. Academy of Management Journal, 42, 662– 673. Smith, C. M., Tindale, S. R., & Dugoni, B. L. (1996). Minority and majority influence in freely interacting groups: Qualitative versus quantitative differences. British Journal of Social Psychology, 35, 137–150. Sniezek, J. A., & Henry, R. A. (1989). Accuracy and confidence in group judgment. Organizational Behavior and Human Decision Processes, 43, 1–28. Stanley, J. D. (1981). Dissent in organizations. Academy of Management Review, 1, 13–19. Stasser, G. (1988). Computer simulation as a research tool: The DISCUSS
1093
model of group decision making. Journal of Experimental Social Psychology, 24, 393– 422. Stasser, G., & Birchmeier, Z. (2003). Group creativity and collective choice. In P. B. Paulus & B. A. Nijstad (Eds.), Group creativity (pp. 85–109). New York: Oxford University Press. Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decision making: Biased information sampling during discussion. Journal of Personality and Social Psychology, 48, 1467–1478. Stasser, G., Vaughan, S. I., & Stewart, D. D. (2000). Pooling of unshared information: The benefits of knowing how access to information is distributed among group members. Organizational Behavior and Human Decision Processes, 83, 102–116. Stewart, D. D., & Stasser, G. (1998). The sampling of critical, unshared information in decision-making groups: The role of an informed minority. European Journal of Social Psychology, 28, 95–113. van Knippenberg, D., & Haslam, S. A. (2003). Realizing the diversity dividend: Exploring the subtle interplay between identity, ideology, and reality. In S. A. Haslam, D. van Knippenberg, M. J. Platow, & N. Ellemers (Eds.), Social identity at work: Developing theory for organizational practices (pp. 66 –77). Philadelphia: Psychology Press. Vroom, V. H., & Jago, A. G. (1988). The new leadership: Managing participation in organizations. Englewood Cliffs, NJ: Prentice-Hall. Wanous, J. P., & Youtz, M. A. (1986). Solution diversity and the quality of group decisions. Academy of Management Journal, 29, 149 –159. Williams, K. Y., & O’Reilly, C. A. III. (1998). Demography and diversity in organizations: A review of 40 years of research. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 20, pp. 77–140). Greenwich, CT: JAI Press. Winquist, J. R., & Larson, J. R. Jr. (1998). Information pooling: When it impacts group decision making. Journal of Personality and Social Psychology, 74, 371–377. Wittenbaum, G. M. (1998). Information sampling in decision-making groups: The impact of members’ task-relevant status. Small Group Research, 29, 57– 84. Wittenbaum, G. M., Hollingshead, A. B., & Botero, I. C. (2004). From cooperative to motivated information sharing in groups: Moving beyond the hidden profile paradigm. Communication Monographs, 71, 286 – 310.
Received May 19, 2005 Revision received April 20, 2006 Accepted May 18, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1094 –1110
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1094
Knowing Your Place: Self-Perceptions of Status in Face-to-Face Groups Cameron Anderson
Sanjay Srivastava
University of California, Berkeley
University of Oregon
Jennifer S. Beer
Sandra E. Spataro
University of California, Davis
Cornell University
Jennifer A. Chatman University of California, Berkeley Status is the prominence, respect, and influence individuals enjoy in the eyes of others. Theories of positive illusions suggest that individuals form overly positive perceptions of their status in face-to-face groups. In contrast, the authors argue that individuals’ perceptions of their status are highly accurate— that is, they closely match the group’s perception of their status— because forming overly positive status self-perceptions can damage individuals’ acceptance in a group. Therefore, the authors further argue that individuals are likely to refrain from status self-enhancement to maintain their belongingness in a group. Support for their hypotheses was found in 2 studies of status in face-to-face groups, using a social relations model approach (D. A. Kenny & L. La Voie, 1984). Individuals showed high accuracy in perceiving their status and even erred on the side of being overly humble. Moreover, enhancement in status self-perceptions was associated with lower levels of social acceptance. Keywords: status, self-perception, self-enhancement, positive illusions, social relations model
Haig’s volatile tenure as secretary of state and the termination of his political career illustrate the primary arguments we put forth in this article about the consequences of overestimating one’s status in face-to-face groups. Status in face-to-face groups is the prominence, respect, and influence individuals enjoy in the eyes of other group members (Anderson, John, Keltner, & Kring, 2001; Berger, Cohen, & Zelditch, 1972; Goldhamer & Shils, 1939). We argue that status self-enhancers—individuals who believe they possess higher status in a group than is actually accorded to them by the group—are disliked and rejected by other group members because they are seen as illegitimately demanding social privileges and trying to usurp control from others. We further argue that to prevent social rejection, individuals tend to avoid engaging in status self-enhancement, instead perceiving their status relatively accurately; in other words, individuals try to avoid Alexander Haig’s mistake of overestimating their own status. Our view contrasts with those offered by the positive illusions perspective, which suggests that individuals tend to form unrealistic, overly positive perceptions of themselves in a wide variety of domains to maintain their self-esteem (Barkow, 1975; Pfeffer & Cialdini, 1998; Taylor & Brown, 1988). We argue, instead, that status considerations offer an important exception to predictions made by the theory of positive illusions. To evaluate these hypotheses, we conducted two laboratory studies of status in naturally interacting face-to-face groups, using a social relations model design (Kenny & La Voie, 1984). In Study 1, we examined the social consequences of status selfenhancement and the accuracy of self-perceptions of status using a longitudinal design. In Study 2, we again examined social consequences and accuracy. In addition, however, we tested whether
“I’m in control here.” Alexander M. Haig uttered these infamous words at a press conference after the shooting of President Ronald Reagan on March 30, 1981. As secretary of state, Haig was actually fourth in the line of succession and did not have executive authority. According to many reports, Haig’s claim was emblematic of his general attitude while U.S. secretary of state. Haig apparently had an overinflated view of his status in the Reagan White House— he felt entitled to more authority than the president over foreign policy issues, treated highly ranked colleagues with little respect, and demanded many high-status privileges, such as a better seat on the president’s airplane, Air Force One (Weisman, 1982). In turn, fellow cabinet members reportedly bristled at Haig’s inflated self-perception. They fought with him repeatedly on policy issues, generally excluded him from daily interactions, and ridiculed him so extensively in the press that one political observer remarked, “The public beating Mr. Haig received at the hands of the White House was virtually unprecedented” (Gelb, 1981, p. 23). Under these circumstances, it is not surprising that Haig resigned after only a year and a half. Cameron Anderson and Jennifer A. Chatman, Walter A. Haas School of Business, University of California, Berkeley; Sanjay Srivastava, Department of Psychology, University of Oregon; Jennifer S. Beer, Department of Psychology, University of California, Davis; Sandra E. Spataro, Johnson Graduate School of Management, Cornell University. Correspondence concerning this article should be addressed to Cameron Anderson, University of California, Walter A. Haas School of Business, 545 Student Services Bldg #1900, Berkeley, CA 94720-1900. E-mail:
[email protected] 1094
SELF-PERCEPTIONS OF STATUS
1095
individuals’ concern for social acceptance kept them from engaging in status self-enhancement and whether status selfenhancement damaged groups’ overall functioning.
with other researchers who have argued that self-enhancement in general has a number of deleterious effects (e.g., John & Robins, 1994; Paulhus, 1998; Robins & Beer, 2001).
The Case for Self-Enhancement in Status Perceptions
The Functions of Status Hierarchies in Face-to-Face Groups
According to the prominent positive illusions perspective, people have a strong desire to view themselves positively, which can lead them to construct distorted, unrealistically positive selfperceptions (e.g., Greenwald, 1980; Taylor & Brown, 1988). Researchers have found that people form overly positive selfperceptions on a variety of dimensions, including their intelligence (Kruger & Dunning, 1999), physical abilities (Dunning, Meyerowitz, & Holzberg, 1989), personality traits (Messick, Bloom, Boldizar, & Samuelson, 1985), and physical attractiveness (Heine & Lehman, 1997). This tendency to self-enhance is thought to stem from the broader motivation to maintain self-esteem (e.g., Baumeister, 1998; Dunning, Leuenberger, & Sherman, 1995), and positive illusions are thought to have a number of personal and interpersonal benefits (Taylor & Brown, 1988; Taylor, Lerner, Sherman, Sage, & McDowell, 2003). An individual’s status in a face-to-face group has a strong impact on his or her self-esteem; that is, the level of respect and admiration individuals achieve in a group strongly shapes how they feel about themselves (Barkow, 1975; Frank, 1985; Heaven, 1986; Leary, Cottrell, & Phillips, 2001; Raskin, Novacek, & Hogan, 1991). Given the importance of status to bolstering selfesteem (Barkow, 1975), one might expect people to be especially likely to form distorted, overly positive perceptions of their status in face-to-face groups. Indeed, a number of theorists have suggested that status selfenhancement is pervasive (Barkow, 1975; Krebs & Denton, 1997; Pfeffer & Cialdini, 1998). For example, Barkow (1975) argued that individuals typically distort status-relevant information to satisfy the imperative for self-esteem. They do so by ignoring some relevant information and emphasizing other information, which allows them to believe they possess higher status than they actually do. Pfeffer and Cialdini (1998) also argued that the drive for a positive self-concept leads individuals to form unrealistically positive perceptions of their influence over others’ behavior. However, empirical studies of accuracy in self-perceptions have generally focused on constructs other than status, such as skills, abilities, or personality traits, leaving unanswered the question of whether self-enhancement biases shape perceptions of status in face-to-face group settings.
Status hierarchies serve a number of important functions for face-to-face groups. One of the primary challenges groups face is the division of influence among members. Groups often experience process inefficiencies because too many group members want to make decisions for the group, give out commands to others, and dominate group discussions, which can create chaos and conflict. Status hierarchies are a primary way groups solve this problem by facilitating an orderly division of influence among group members, using such means as allowing or denying different individuals the rights to perform certain behaviors (Bales, 1950; Berger, Rosenholtz, & Zelditch, 1980). For example, high-status individuals are allowed to control group interactions, make decisions for the group, and give verbal directives to others, whereas low-status individuals are expected to defer to others, speak less in social interactions, and keep their opinions more to themselves (Bales, 1950; Berger et al., 1980; Goffman, 1967; Keltner, Gruenfeld, & Anderson, 2003). Groups also face the problem of self-interest. To succeed as a collective, groups must motivate members to act selflessly or to behave in ways that benefit the group, even when such behavior requires personal investment and sacrifice. Status hierarchies can help groups achieve this by rewarding individuals who contribute to the group’s success (Berger et al., 1972; Blau, 1964; Frank, 1985; Homans, 1951; Thibaut & Kelley, 1959). Face-to-face groups allocate status to group members who are believed to contribute to the group’s goals; individuals believed to make important contributions to the group are typically granted high status, whereas individuals believed to make fewer contributions or even to undermine a group’s success are assigned low status. Valued contributions can take several forms, such as expending effort for the group or providing needed expertise. By rewarding group-oriented behavior, status hierarchies compel individual members to work toward the group’s goals, which facilitates collective success.1
The Case for Accuracy in Status Perceptions In contrast to arguments drawn from the positive illusions perspective, we propose that individuals avoid status selfenhancement because of the severe social costs a person might bear by inflating his or her status. On the basis of functionalist accounts of status (e.g., Thibaut & Kelley, 1959), we propose that when members of face-to-face groups overestimate their status in the group, their behavior provokes conflict and disorganization, which in turn leads the group to dislike and reject that member. Therefore, individuals tend to refrain from engaging in status self-enhancement so that they can maintain their belongingness in groups (Baumeister & Leary, 1995). Our perspective thus concurs
Hypothesis 1: Group Members Who Engage in Status Self-Enhancement Will Be Less Accepted by Other Members Than Those Who Perceive Their Status Accurately On the basis of this functional account of status hierarchies, we hypothesized that status self-enhancers would be less accepted by fellow group members than those who perceived their status accurately. Social acceptance is the degree to which individuals are liked by others and included in the group (Baumeister & Leary, 1 This functional perspective of status hierarchies contrasts with the view that status is allocated through dominance contests (e.g., Lee & Ofshe, 1981; Mazur, 1973). Although the debate between these two perspectives has not been resolved, Ridgeway and colleagues have provided ample evidence that functional considerations play a major role in the statusorganizing processes of task groups (Ridgeway, 1987; Ridgeway & Berger, 1986; Ridgeway & Diekema, 1989).
1096
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
1995; Hogan, 1983). It involves how well individuals get along with others and is conceptually distinct from social status, which involves how well individuals get ahead (Hogan, 1983; Homans, 1951; Leary et al., 2001; Wiggins, 1979). We based our hypothesis on three converging lines of evidence. First, because status hierarchies provide stability and order, face-to-face groups actively protect their status hierarchies by ostracizing and excluding individuals who challenge or subvert their hierarchy (e.g., Ridgeway, 1982; Ridgeway & Berger, 1986). When individuals possess self-perceptions of status in a face-toface group that are higher than the status actually given to them by the group, they are likely to behave according to their own selfperceptions— by speaking frequently in group discussions, asserting their opinions forcefully, and making verbal commands and directives to others (Anderson & Berdahl, 2002; Bugental & Lewis, 1999; Galinsky, Gruenfeld, & Magee, 2003). As a result of such behavior, status self-enhancers would instigate conflict and disorder in a group by refusing to defer to those who actually have higher status, attempting to take charge of the group, and working against those with higher status. They would be viewed by others as challenging and subverting the existing status order and undermining the stability of the group (Homans, 1951; Ridgeway & Berger, 1986). Second, status self-enhancers implicitly claim that they are making larger contributions to the group than others believe they are making. As mentioned above, status is a reward: It is a form of social currency that groups give to members who contribute to the group’s success (Thibaut & Kelley, 1959). When individuals engage in status self-enhancement in a face-to-face group, therefore, they are claiming more of this social payment than the group believes they are entitled to receive. Status self-enhancers would thus be less socially accepted because other group members would perceive them as claiming scarce social rewards that they did not deserve.2 Third, status self-enhancement may be threatening to fellow group members. Groups typically treat status as a zero-sum commodity, affording high status to a few individuals and giving lower status to others (Blau, 1964; Clark, 1990; Frank, 1985). When some individuals have high status, this necessitates that others have low status. Therefore, when individuals claim status for themselves, they take status away from others (Blau, 1964). For example, individuals who give unsolicited advice, interrupt others while speaking, or give verbal directives indicate that they perceive themselves to be superior to others (Clark, 1990). Status self-enhancers’ expressed superiority, viewed as illegitimate by others, is likely to provoke harsh reactions from the group. A few studies have provided suggestive evidence that selfenhanced perceptions of status can damage individuals’ social acceptance. Early observational studies of peer groups found that when low-status group members tried to behave in high-status ways, such as by taking control of the group’s activities, they were ridiculed and ostracized by other group members (Homans, 1951; Roethlisberger & Dickson, 1938; Whyte, 1943). Studies of selfpresentational styles found that individuals who boasted were liked less than individuals with a more self-deprecating style (Jones & Shrauger, 1970; Platt, 1977, cited in Powers & Zuroff, 1986). Research on trait self-enhancement found that individuals who were generally high on self-enhancement were viewed by others as arrogant, hostile, cold, and defensive (Colvin, Block, & Funder, 1995; Paulhus, 1998), suggesting they may have been less socially
accepted. In the current studies, we extended these findings by directly testing the link between self-enhanced perceptions of status and social acceptance in groups.
Social Costs for Status Self-Effacement? Our arguments and prior research suggest substantial costs associated with overestimating one’s own status, but are there social costs associated with underestimation as well? That is, on a continuum from self-enhancement to accuracy to self-effacement, should one expect a linear relationship with social acceptance across the whole continuum or an inverted U-shaped curve, where self-effacers face social rejection just as self-enhancers do? Strictly speaking, individuals who self-efface are violating a status hierarchy, and they might also be failing to perform leadership-related behaviors that the group is expecting them to perform. However, self-effacers are not claiming scarce social resources or threatening others’ high status within the group; in fact, they are making more of both resources and social status available to others. Past research has suggested that status selfeffacers might be more socially accepted than accurate perceivers because they signal a particularly high degree of selflessness or an extreme willingness to put the group’s interests above their own (Ridgeway, 1982). Given these various considerations, we found no strong basis to hypothesize a curvilinear relationship, though we considered it an important enough question to test in the data.
Hypothesis 2: Face-to-Face Group Members Will Perceive Their Own Status Accurately If Hypothesis 1 is correct and status self-enhancement decreases social acceptance, people have an incentive to view their status accurately. We agree that people might desire to engage in status self-enhancement to maintain their self-esteem (e.g., Barkow, 1975; Pfeffer & Cialdini, 1998). However, people also have a fundamental human need to belong and be included in social groups (Baumeister & Leary, 1995; Maslow, 1968). According to Hypothesis 1, status self-enhancement would work against this second motive. Consistent with prior theorizing (e.g., Baumeister & Leary, 1995), we believed that individuals’ need to belong would outweigh their desire for higher self-esteem. Thus, our second hypothesis was that individuals’ self-perceptions of status would be accurate.3 In considering accuracy, we adopted peer-rated status as the criterion against which self-perceptions would be compared. So2
This argument is not meant to imply that status-organizing processes are always fair. Groups allocate status to those they believe contribute to the group. Yet group perceptions can be misguided, as when individuals’ status is based on demographic characteristics such as race or sex rather than on contributions to the group. That notwithstanding, once a status order is established, individuals who believe they possess more status than the group believes they possess will be less accepted. 3 We hypothesized that individuals would accurately perceive their status, rather than underestimate it, because underestimating one’s status would mean forgoing the social benefits that status affords, such as the ability to express one’s opinions. Thus, although the need to belong would keep individuals from engaging in status self-enhancement, we believed the lure of reaping the benefits of status would keep individuals from being overly humble.
SELF-PERCEPTIONS OF STATUS
cial perception researchers often face the dilemma that when an individual’s self-perceptions diverge from peers’ perceptions, it is unclear who has a stronger claim to truth (Robins & John, 1997). However, status differs from other dimensions of social perception in an important way, in that status hierarchies are socially constructed: In a face-to-face group, the group members’ perceptions are the very definition of status. Given this conceptualization of status, peer-rated status is an appropriate criterion for testing questions of accuracy. Moreover, in social perception research, accuracy can mean many things both conceptually and operationally (Cronbach, 1955), and it is important to consider the definition of accuracy on which to focus. One definition provided by Cronbach (1955) is elevation accuracy, which is concerned with whether mean levels of self-perceptions are higher, lower, or about the same as others’ perceptions. This is the sense of accuracy implied by the notion of self-enhancement versus self-effacement or of overestimating versus underestimating one’s own status. If the average level of self-perceived status were higher than the average level of peerrated status, consistent with positive illusions predictions, we could conclude that the average individual self-enhances on status. In contrast, on the basis of our hypothesis that status selfenhancement has social costs, we expected that self-perceived status would be about equal to peer-rated status or perhaps even lower (if individuals were being overly modest in their status self-perceptions). A second definition of accuracy is differential accuracy (Cronbach, 1955), which, in the present context, we defined as whether self-perceived status is positively correlated with peer-rated status.4 Whereas elevation accuracy is based on differences between mean levels of social perceptions, differential accuracy is concerned with the rank-order correspondence between selfperceptions and others’ perceptions. We expected to find evidence for both elevation accuracy and differential accuracy. However, elevation accuracy and differential accuracy are logically distinct, and it would not be necessary for them to give converging results. For example, it is possible that individuals might have little idea where they stand relative to others or within the status hierarchy of their group (poor differential accuracy), but when they estimate their own absolute status level, they might not be biased to be especially high or low (good elevation accuracy). Such a pattern in the data would still be somewhat consistent with Hypothesis 2 insofar as it would suggest that individuals were avoiding the pitfalls of overestimating their own status, but it would also suggest that individuals were not highly accurate in gauging their standing relative to others.
Study 1 For Study 1, we ran a longitudinal study of small groups. Groups of individuals began the study as relative strangers; they met and interacted once a week for 4 weeks. At the end of each meeting, participants rated status and social acceptance in a roundrobin design, wherein each participant in a group rated every other participant in that group and also provided self-ratings. Using Kenny and La Voie’s (1984) social relations model, we were able to derive indices of the extent to which each individual was seen by other group members as having relatively high versus low status within the group and also whether each individual was
1097
relatively more versus less accepted in the group. The design also made it possible to derive indices of self-enhancement bias and to test questions about accuracy of self-perceptions using others’ perceptions as the accuracy criterion.
Main Hypotheses In Study 1, we examined two main hypotheses. First, we examined whether status self-enhancement led to lower levels of social acceptance (Hypothesis 1). The longitudinal design allowed us to test directionality by using lagged-effects analyses (Kenny & Campbell, 1999; West, Biesanz, & Pitts, 2000). If the analyses indicated that status self-enhancement at one point in time predicted social acceptance at a later point in time, this would support our hypothesis that groups are less accepting of individuals who engage in status self-enhancement. We also considered whether the lagged relationship from status self-enhancement to social acceptance was linear or curvilinear. That is, are individuals who self-efface (i.e., who underestimate their own status) met with more, less, or about the same amount of social acceptance as individuals who are fairly accurate or who self-enhance? To address this question, we also tested a cross-lagged model that included a quadratic effect of status self-enhancement on social acceptance. Second, we tested whether self-perceptions of status were accurate (Hypothesis 2). We examined both elevation accuracy and differential accuracy (Cronbach, 1955). Elevation accuracy analyses are unlikely to show exactly zero differences between selfperceptions and peers’ perceptions, and differential accuracy analyses are unlikely to show correlations of 1.0 between selfperceptions and peers’ perceptions. Therefore, to provide a context for interpreting the analyses, we compared the elevation and differential accuracy of self-perceived status with the elevation and differential accuracy of self-perceived social acceptance. Because of the special properties of status hierarchies, Hypothesis 1 applies specifically to status, not to self-enhancement on other social dimensions such as social acceptance. Unlike status, self-enhancement in domains such as social acceptance does not illegitimately claim scarce resources because social acceptance is not a scarce or zero-sum commodity (Blau, 1964). In fact, engaging in acceptance self-enhancement might only communicate an eagerness to be socially integrated and included. Previous research has shown that self-enhancement in many domains, including social acceptance, has either neutral (Srivastava & Beer, 2005) or positive (Taylor et al., 2003) consequences for actual acceptance. Moreover, previous studies have consistently shown that, as predicted by the theory of positive illusions, individuals tend to overestimate themselves in most domains, including social acceptance (Kenny, 1994; Taylor & Brown, 1988). Differential accuracy in many domains of self-perception, including social acceptance, can be substantial (Kenny, 1994); nevertheless, we expected differential accuracy to be even higher for status than for social acceptance. 4 Our present treatment of differential accuracy is simplified somewhat from Cronbach’s (1955) original decomposition, which operationalizes four components of accuracy in patterns of many trait ratings made at once (i.e., profiles) rather than ratings of a single characteristic at a time (in our case, status).
1098
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
Additional Questions In addition to testing these main hypotheses, we also explored some additional issues relevant to our arguments. As part of the cross-lagged analysis, we were also able to conduct an independent test for effects in the reverse direction—to test whether being less accepted by others led to self-enhancement of status. It is possible, for example, that individuals who are less socially accepted might try to convince themselves that they have high status as a way to compensate for being disliked. This was posed as an additional exploratory question, not a rival to Hypothesis 1. In discussing cross-lagged models, Rogosa (1980) has written, “Measures of strength and duration of the reciprocal relationship and of the specific causal effects [italics added] are more informative than the determination of the causal winner” (p. 246). Consistent with this philosophy, we used the cross-lagged model to independently estimate the specific effects in each direction, and conceptually, we treated them as independent questions. Another question concerned the possible effects of selfperceived status on one’s actual status (i.e., status as attributed by peers). Although status self-enhancement might harm one’s social acceptance, it is unknown whether it has consequences for actual status, and it is possible that positive beliefs about one’s own status might act as a self-fulfilling prophecy (Jussim, 1991; Krebs & Denton, 1997). For example, in one study, confederates who boasted about themselves were perceived as less likeable but also as more capable and less dependent on others (Gergen & Wishnov, 1965), suggesting people might dislike self-enhancers but still concede high status to them. Because we assessed groups over time, we were able to test for a lagged effect of self-perceived status on actual status.
Method Participants Participants were undergraduate students attending a West Coast university who participated for course credit; the participants were 19 years old on average (SD ⫽ 0.9 years). The sample for the present study were restricted to those participants who came to all four sessions of the study and thus provided complete data (N ⫽ 152). The sample represented 72% of the total number of participants who attended the first week. Attrition analyses comparing complete-data participants with those who dropped out indicated no significant differences in status, social acceptance, or sex (all absolute rs ⬍ .09). Participants were assigned to one of 28 groups that met once a week for 4 weeks; the great majority of participants were strangers (97% of all possible pairings reported that they did not know one another at all, and fewer than 1% described their relationship as a preexisting friendship). The groups’ size ranged from 4 to 8, with a modal group size of 6. The percentage of women in each group ranged from 25% to 80%; the average percentage of women in each group was 50%.
Procedure Groups interacted for about 20 minutes the first week and for about 40 minutes all subsequent weeks. In Week 1, we used a task that would allow group members to interact on a collaborative project; thus, status differences could emerge right away. Specifically, we used a task called Lost on the Moon, in which the group is told it has crash-landed on the moon and needs to get back to the mother ship using a list of 15 items (Robins & Beer, 2001). In Week 2, we used a task that would facilitate personal
disclosure and the development of interpersonal ties to mimic the same process that occurs in real-world groups. Specifically, group members engaged in an informal exercise where they asked each other a series of questions such as “What was your most embarrassing moment?” (adapted from Aron, Melinat, Aron, Vallone, & Bator, 1997). In Week 3, we used a more competitive task to allow for conflicts to emerge, which is also a part of real-world group experiences. Specifically, groups role-played a university’s alumni committee; the committee awarded prize money to deserving alumni, and each group member was asked to advocate for a specific nominee (adapted from John & Robins, 1994). Finally, in Week 4, we used a fun task to help alleviate any potential tension left over from the competitive task in Week 3 and to provide more variability in the tasks in which groups engaged. Group members played the board game Beyond Balderdash (Gameworks Creations, 1995), in which they tried to guess the correct definitions of various words from a list of potential definitions.5
Measures Self-perceived and actual status. After each of the four group meetings, each participant privately rated the status of every other group member by indicating agreement with the item “This person had a lot of status within the group today” on a scale from 0 (Disagree very strongly) to 10 (Agree very strongly). They also rated their own status with the item “I had a lot of status within the group today,” using the same response scale. We used the software program SOREMO (Kenny, 1995) to implement the social relations model analyses of the round-robin (i.e., peer) status ratings. For each of the four sessions, SOREMO calculated two scores for each participant: a target score, which is an index of how that individual was typically perceived by the others in the group, and a perceiver score, which is an index of how the individual typically perceived others. SOREMO removed group differences, making target and perceiver scores statistically independent of group membership and thus appropriate for conventional least squares procedures that assume independence (see Kenny & La Voie, 1984). Target scores for status showed statistically significant amounts of variance in all 4 weeks (relative variances were .40, .20, .34, and .19, respectively), indicating group members agreed about one another’s status at better than chance levels in all 4 weeks. The extent of agreement on status differed across weeks; pairwise comparisons of consensus between different weeks indicated that all weeks differed from each other, with the exception of Weeks 2 and 4. Status self-enhancement. The social relations model design allowed us to utilize the index of self-enhancement recently developed by Kwan and colleagues (Kwan, John, Kenny, Bond, & Robins, 2004). Selfenhancement has been operationalized in different ways in the literature. Some studies have used a self-insight approach, in which an individual’s self-perceptions are compared with peers’ perceptions of the individual; for example, if an individual believes himself or herself to be more intelligent than others believe him or her to be, this is considered evidence of self-enhancement. Other studies have used a social comparison approach, in which an individual’s self-perceptions are compared with that individual’s perceptions of others; for example, if an individual believes himself or herself to be an above-average driver, this is considered evidence of self-enhancement. The Kwan index, based on the social relations model, represents a conceptual and methodological breakthrough because it integrates both approaches and corrects biases present in each, providing a measure of self-enhancement with fewer confounds (Kwan et al., 2004). To examine the consequences of status self-enhancement, we calculated a self-enhancement index for status based on the technique described by Kwan et al. (2004). In this technique, status self-enhancement is calculated as
5 Srivastava and Beer (2005) reported an investigation of selfperceptions and target ratings of acceptance in this same data set. That investigation did not include any analyses of status.
SELF-PERCEPTIONS OF STATUS SE ⫽ S ⫺ T ⫺ P ⫺ G, where SE is self-enhancement, S is the self-perception, T is the (groupmean-deviated) target score, P is the (group-mean-deviated) perceiver score, and G is the group mean.6 The self-insight approach is represented by the subtraction of target scores (which index how an individual is viewed by others), and the social-comparison approach is represented by the subtraction of perceiver scores (which index how an individual views others). Subtraction of the group mean scales the self-enhancement score so that the zero point indicates an unbiased self-perception; the group subtraction also makes self-enhancement scores statistically independent of group membership. Self-perceived and actual social acceptance. Participants rated the other members of their group on the item “I like this person” on a scale from 0 (Disagree very strongly) to 10 (Agree very strongly). Participants were also asked to rate themselves on the item “I am a likable person” in the context of the group setting, using the same 0 to 10 scale. We used SOREMO (Kenny, 1995) to implement the social relations model analyses of the round-robin acceptance ratings and operationalized social acceptance as the target score (i.e., the group’s collective judgment of how much they liked an individual). Target scores for social acceptance showed lower relative variances than did target scores for status, though in all weeks except the last, the target variances were significant (relative variances were .06, .07, .06, and .02, respectively).
Results and Discussion Hypothesis 1: Group Members Who Engage in Status Self-Enhancement Will Be Less Accepted by Other Members Than Those Who Perceive Their Status Accurately Because our design assessed individuals over time, we analyzed the data using hierarchical linear modeling (Bryk & Raudenbush, 1992) with cross-lagged effects. Each of the individuals in the study had four status self-enhancement scores and four acceptance scores from each of the 4 weeks of the study. At Level 1, we modeled how an individual’s acceptance varied over time as a function of the individual’s prior status self-enhancement (STATUS_SE), controlling for the individual’s prior acceptance (ACCEPTANCE) to account for autocorrelations in acceptance. Thus, we specified the following equation at Level 1 to model within-person effects: ACCEPTANCE(t) ⫽ b0 ⫹ b1 ⫻ ACCEPTANCE(t ⫺ 1) ⫹ b2 ⫻ STATUS_SE(t ⫺ 1). Level 2 of the model aggregated the individual effects to yield a samplewide estimate and t test of each of the coefficients, including the effect of status self-enhancement on acceptance (b2). A random effect on the intercept at Level 2 also accounted for variance in acceptance attributable to individual differences. (Because the social relations model indices remove any group-level dependence in the data, any Level 3 effects—i.e., between-groups effects—would be predetermined to be zero. However, for completeness, we also included a Level 3 random effect on the intercept so that our software would calculate the correct degrees of freedom.) In this model, the key test of the hypothesis is the lagged effect of status self-enhancement on actual acceptance (b2). The results of the analysis, shown at the top of Table 1, indicate that over and
1099
above any influence of prior acceptance, status self-enhancement predicted lower levels of acceptance over time (b2 ⫽ ⫺.020, SE ⫽ .001, p ⬍ .05). That is, over and above the substantial stability in individuals’ acceptance across the 4 weeks, status selfenhancement still predicted subsequent decreases in acceptance. In follow-up analyses, we included sex as a moderator variable; a null result indicated there was no significant difference in the magnitude or direction of the status self-enhancement effect between men and women. Simple within-week correlations were consistent with these lagged effects, showing a consistent negative relationship within week between status self-enhancement and acceptance (average r ⫽ ⫺.13 across weeks). In a model that added a quadratic term for status selfenhancement, we did not find evidence for a curvilinear relation between status self-enhancement and acceptance (b3 ⫽ .000, SE ⫽ .002, p ⫽ .91). Thus, it appears that individuals who status selfenhanced were liked and accepted by fellow group members less than accurate self-perceivers, who in turn were liked and accepted less than self-effacers.
Hypothesis 2: Face-to-Face Group Members Will Perceive Their Own Status Accurately Elevation accuracy. Examining elevation accuracy involved comparing mean levels of status self-perceptions with mean levels of peer perceptions (Cronbach, 1955). If individuals’ selfperceptions of status were not significantly different from peer perceptions, they would be exhibiting elevation accuracy; if selfperceptions were higher, they would be exhibiting selfenhancement. Our primary interest was to compare status selfperceptions with peer-rated (i.e., actual) status; however, to provide context for that analysis, we also included comparisons between self-reported social acceptance and peer-rated social acceptance. Because elevation accuracy concerns the actual means of selfand peer ratings, we did not use the SOREMO target effects, which are mean centered; rather, we analyzed the raw scores for self- and peer ratings. Because the raw scores were dependent on group membership, we conducted the analysis at the group (rather than the individual) level to eliminate any dependence in the data, by averaging self- and peer ratings for each group (Kenny, 1996). In addition to making comparisons between status and social acceptance, we also included factors to account for effects of week and of sex composition of the groups (i.e., operationalized as the proportion of men in the group). Thus, the analysis was a four-way (2 ⫻ 2 ⫻ 4 ⫻ 10) mixed analysis of variance (ANOVA) at the group level with three within-groups factors (self- vs. peer percep6
One can measure self-enhancement by constructing a discrepancy score, in which peers’ perceptions are subtracted from self-perceptions, or by partialing out peers’ perceptions from self-perceptions in the regression analyses. Each method has advantages and disadvantages. The discrepancy-score method can confound the effects of self- and peer perceptions (e.g., self-esteem might correlate with a discrepancy score only because it correlates with self-perceptions, not because it correlates negatively with peer perceptions). However, partialing out peer perceptions can create problems of multicollinearity, leading to unstable estimates. We followed Kwan et al.’s (2004) approach and used discrepancy scores.
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
1100
Table 1 Multilevel Models Predicting Social Acceptance and Status Self-Enhancement in Study 1 Unstandardized coefficient
Parameter DV: Social acceptance Lag-1 acceptance (b1) Lag-1 status self-enhancement (b2) DV: Status self-enhancement Lag-1 status self-enhancement (b1) Lag-1 acceptance (b2)
SE
t test
0.611 ⫺0.020
0.034 17.70** 0.009 ⫺2.36*
0.130 ⫺0.262
0.041 3.16* 0.173 ⫺1.51
Note. N ⫽ 152. DV ⫽ dependent variable. p ⬍ .05. ** p ⬍ .01.
*
tions, status vs. acceptance, and week) and one between-groups factor (the sex composition of the group). Table 2 presents self- and peer ratings for status and acceptance across all 4 weeks; statistics (including standard deviations) are presented at the group level because that is how we conducted the analysis. There was not a significant main effect for self- versus peer perception at the group level, F(1, 18) ⫽ 1.64, ns, indicating that there was not support for a general self-enhancement effect across both status and acceptance. However, consistent with expectations, there was a significant interaction between source of rating (self vs. peer) and type of rating (status vs. acceptance), F(1, 18) ⫽ 39.20, p ⬍ .01. A planned comparison showed that selfratings of acceptance were significantly higher than peers’ social acceptance ratings, F(1, 27) ⫽ 39.07, p ⬍ .01. Self-ratings of status were, however, significantly lower than peers’ status ratings, F(1, 27) ⫽ 18.69, p ⬍ .01. Thus, although people showed a self-enhancement bias in perceiving their acceptance, they showed a bias toward self-effacement in perceiving their status. There was not a significant effect for the groups’ sex composition on perceptions of status or acceptance, nor were there any interactions involving group sex composition. The self-effacement effect in perceiving status was not moderated by week, F(1, 25) ⫽ 1.80, ns. There was, however, a significant moderating effect of week on how much individuals selfenhanced when perceiving their acceptance, F(1, 25) ⫽ 11.19, p ⬍ .01. Specifically, self-rated acceptance exceeded social acceptance
Table 2 Elevation Accuracy: Self- and Peer Ratings of Status and Acceptance Across 4 Weeks in Study 1 Status Week 1 2 3 4
Self-perception 5.06 (0.91) 5.05 (0.72) 5.42 (0.75) 5.36 (1.04)
much more in Week 1 (a difference of 1.39) than in the other three weeks (differences of 0.66, 0.65, and 0.70, respectively). People might have self-enhanced when perceiving their acceptance particularly in Week 1 because it was the beginning of the group’s development, and thus, individuals did not have as much information to inform their judgments; as self-perceptions of acceptance are not constrained in the same way as status, there was more opportunity for self-enhancement tendencies to emerge. Differential accuracy: Correlations between self- and peer ratings. Examining differential accuracy involved correlating selfperceptions of status with peer perceptions (Cronbach, 1955). If individuals’ self-perceptions of status were to correlate highly with peer perceptions, they would be exhibiting differential accuracy. For these analyses, carried out on the individual-level data, we used target scores as the measure of peer perceptions and selfratings as a measure of self-perceptions. Because the self-ratings were potentially dependent on group membership, we followed Kenny, Kashy, and Cooks (2006) recommendation and computed partial correlations between self-ratings and target scores, with group effects partialed out by using 27 dummy variables representing membership in the 28 groups. Shown in Table 3 are the self–peer partial correlations for status and acceptance across the 4 weeks. As shown, self–peer agreement was high for status, indicating that individuals were significantly accurate in perceiving their status. Partial correlations of accuracy were as high as .59 and were, on average, .46 across weeks. It is interesting to note that status accuracy was lowest in Week 2, in which group members engaged in the get-acquainted task. It might have been more difficult for individuals to assess their status in such an informal exercise that had no explicit goal. We tested for sex differences in individuals’ accuracy in perceiving their status using moderated multiple regression analyses (Aiken & West, 1991) with group dummy variables entered as controls and found no significant or substantial differences in any of the 4 weeks. Self-rated acceptance correlated somewhat with peer-rated social acceptance; all self–peer partial correlations were significant and averaged .20 across the 4 weeks. These partial correlations were lower than the self–peer correlations for status perceptions; we tested the differences between them using Raghunathan, Rosenthal, and Rubin’s (1996) method. Self–peer partial correlations were significantly higher for status than for acceptance in Week 1 (Z ⫽ 3.30, p ⬍ .01) and in Week 3 (Z ⫽ 2.28, p ⬍ .01). Table 3 Differential Accuracy in Self-Perceptions of Status and Acceptance in Study 1
Acceptance Peer rating 5.37 (0.52) 5.75 (0.54) 5.98 (0.50) 5.80 (0.74)
Self-perception 7.13 (0.95) 7.39 (0.92) 7.36 (0.89) 7.59 (0.74)
Peer rating 5.72 (0.58) 6.74 (0.58) 6.68 (0.60) 6.90 (0.61)
Note. Between-groups standard deviations are in parentheses. We consider peer ratings of status and acceptance to reflect the person’s actual status and acceptance, respectively. Self-perceptions of status are significantly lower than peer ratings across weeks. Self-perceptions of acceptance are significantly higher than peer ratings across weeks.
Accuracy correlation Week 1 2 3 4
Status **
.59 .34** .48** .42**
Acceptance .16* .28** .16** .21**
Note. Shown are partial correlations between self-perceptions and target effects (or peer ratings), with group membership partialed out through dummy variables. Accuracy partial correlations for status are significantly higher than accuracy partial correlations for acceptance in Weeks 1 and 3. * p ⬍ .05. ** p ⬍ .01.
SELF-PERCEPTIONS OF STATUS
The correlations were not significantly higher in Weeks 2 and 4, however. These self–peer partial correlations may have been lower because there was also lower peer–peer agreement on social acceptance (Kenny, 1994). Differential accuracy: Comparing accuracy with consensus. Another way to create a context to interpret differential accuracy, which can be operationalized within Kenny and La Voie’s (1984) social relations model, is to compare whether self– other covariance (i.e., differential accuracy) is as high as consensus variance (i.e., the agreement among different peers). If self– other agreement is about equal to consensus, this indicates that the average individual knows his or her own status as well as the average peer observer. Following Kenny’s (1994) method, we conducted a two-factor (2 ⫻ 4) ANOVA on the group level, with self–peer covariance versus target variance as the first within-groups factor (which tested individuals’ accuracy against peers’ accuracy) and week as the second within-groups factor. Self–peer covariance was not significantly different from target variance, F(1, 17) ⫽ 0.13, ns. There was no main effect of agreement type, nor was there any interaction effect with week. Thus, this suggests individuals’ perceptions of their status were as good an indicator of their actual status as any other group member’s perception of their status. Although we hesitate to overinterpret this null effect, it is interesting in light of the finding that on most dimensions of social perception, self–peer agreement is typically lower than peer–peer agreement (Kenny, 1994). Thus, on the basis of prior research, one would expect individuals’ accuracy in viewing their status to be lower than peers’ consensus in viewing the individual’s status. The relative accuracy with which people perceived their status provides support for the idea that individuals are uniquely motivated to perceive their status accurately.
Additional Questions Predicting status self-enhancement from acceptance. We found a lagged effect of status self-enhancement on social acceptance. Was the reverse direction also supported by the data? For example, did lower levels of acceptance predict higher levels of status self-enhancement over time? To evaluate this hypothesis, we specified a similar hierarchical linear model, this time with status self-enhancement as the dependent variable and with acceptance as a predictor: STATUS_SE(t) ⫽ b0 ⫹ b1 ⫻ STATUS_SE(t ⫺ 1) ⫹ b2 ⫻ ACCEPTANCE(t ⫺ 1) . The results of this analysis are shown at the bottom of Table 1. There was no significant lagged effect of acceptance on status self-enhancement. Is status self-enhancement a self-fulfilling prophecy? We also explored whether self-perceived status influenced individuals’ actual status (i.e., their peer-rated status). We conducted a similar hierarchical linear model, this time with actual status as the dependent variable and self-perceived status, lagged by 1 week, as the predictor, with lagged actual status as a control. A significant lagged effect of self-perceived status, after controlling for prior levels of actual status, would indicate a self-fulfilling prophecy effect.7 The results of this analysis showed no predictive relation between self-perceived status and actual status (b ⫽ 0.02, SE ⫽ .020, p ⫽ .34). Further, follow-up
1101
analyses indicated this relation did not vary according to participants’ sex. Therefore, the evidence does not indicate that engaging in status self-enhancement helped achieve higher status in the group or that it harmed one’s status in the group.
Summary Supporting our first main hypothesis, individuals who engaged in status self-enhancement were less accepted by the group. Using cross-lagged analyses, we found a directional effect of status self-enhancement on one’s social acceptance but did not find a directional effect of social acceptance on status self-enhancement. Although these findings do not unequivocally establish causation, they do provide evidence consistent with the conclusion that status self-enhancement leads to being less socially accepted. We also found no support for a self-fulfilling effect of self-perceived status on actual status. Thus, these analyses are consistent with the notion that status self-enhancement has only social costs and no apparent social benefits. That we found a linear, rather than curvilinear, effect of status self-enhancement on social acceptance suggests that status selfenhancers were less liked and accepted than were accurate status perceivers (as predicted) but also that accurate perceivers were less liked and accepted than status self-effacers. We hesitate to conclude that status self-effacers are indeed liked more than accurate status perceivers because of the statistical difficulty of testing nonlinear effects (McClelland & Judd, 1993). Tests of curvilinear effects typically have low power, decreasing our chances of detecting such an effect. Therefore, we examined this issue again in Study 2 with a larger number of groups, which provided greater statistical power in a new sample. Supporting our second main hypothesis, participants tended to perceive their status accurately relative to how they perceived their social acceptance. Not only did participants refrain from engaging in status self-enhancement (i.e., they did not overestimate their status) but also they tended to engage in status self-effacement, such that individuals’ self-perceptions of status were consistently lower than their actual status. This suggests individuals might have been so concerned about their acceptance in the group that they were overly humble in perceiving their status. It also might suggest that individuals status self-effaced to further increase their social acceptance. That individuals did engage in self-enhancement when perceiving their acceptance helps rule out a methodological concern— namely, that the accuracy we observed in self-perceptions of status might have been due to our methods or measures. Given that these same individuals in this same setting did engage in selfenhancement when perceiving their acceptance, we can be more assured that our accuracy findings are unique to status. 7
In his reflection-construction model, Jussim (1991) pointed out that the beliefs that produce self-fulfilling prophecies can be based on initially valid information (which produces a special kind of self-fulfilling prophecy that he called the self-sustaining effect) as well as on biased or flawed beliefs. Thus, we used self-perceived status, rather than the self-enhancement index, as the predictor in this analysis.
1102
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
Study 2 We have argued that engaging in status self-enhancement has social repercussions (Hypothesis 1) and that these social repercussions lead people to keep their self-perceptions of status in check because people are concerned for acceptance and inclusion in social groups (Hypothesis 2). For Study 2, we expanded this idea into a third hypothesis—that the more an individual was concerned with social acceptance, the lower that individual’s self-perceptions of status would be relative to his or her actual status (Hypothesis 3). To test this hypothesis, we directly measured individuals’ concern for acceptance in a group. We also examined how individual status self-enhancement affects the group as a whole. As we have argued, when group members view their status in an overly positive way, they are likely to violate peers’ expectations of their behavior and provoke conflict. Therefore, we predicted that groups with members who self-enhanced would show evidence of greater levels of conflict while working on a task together (Hypothesis 4). That is, when group members’ self-perceptions were higher on average than their peer-rated status, the group would exhibit more conflict. Furthermore, along a more exploratory vein, we examined whether average levels of status self-enhancement affected the group’s task performance. If self-enhancement in status perceptions provoked conflict, this less harmonious group environment might hinder the group’s effectiveness and efficiency and thus decrease their performance on tasks. Thus, we had groups perform tasks with easily measurable performance. Examining group-level consequences also allowed us to further examine why status self-enhancement would damage individuals’ acceptance in a group, whereas acceptance self-enhancement would not. We have argued that status self-enhancement decreases social acceptance because it instigates intragroup conflict; however, acceptance self-enhancement does not damage individuals’ actual acceptance because it does not have any effect on the group’s stability and does not provoke intragroup conflict. We tested this assertion directly in Study 2 by examining the relation between self-enhancement in perceptions of acceptance and intragroup conflict. Finally, to better understand the mechanisms underlying accuracy, we used a more multifaceted measure of status in Study 2, which allowed us to differentiate several components of status. One potential alternative explanation for the accuracy we have observed in status perceptions is that status is a relatively clear and unambiguous dimension, precluding individuals from distorting their self-perceptions. That is, ambiguous dimensions can describe a wide variety of behaviors and thus allow people to use selfserving definitions when providing self-evaluations; however, unambiguous dimensions are more objective, are more precisely defined, and thus promote more accurate self-perceptions. Previous studies have found that individuals engage in selfenhancement more on ambiguous dimensions (e.g., moral, sophisticated, and sensible) than on unambiguous dimensions (e.g., intelligent, athletic, and punctual; Allison, Messick, & Goethals, 1989; Dunning et al., 1989). If the unambiguous nature of status were responsible for the accuracy we have observed, we would expect that individuals in Study 2 would be more accurate (for both elevation and differential accuracy) on the less ambiguous aspects of status than on perceiving the more ambiguous aspects and that, in terms of elevation accuracy, individ-
uals would be more self-enhancing on the ambiguous aspects of status than on the unambiguous aspects of status.
Method Participants Participants were 432 undergraduate students (211 men, 221 women) at a West Coast university. They were 20 years old on average (SD ⫽ 2.88 years). Participants were assigned to experimental sessions through a combination of volunteer sign-ups and phone calls inviting them to participate in return for course credit; participants were again assigned to groups in which they did not know other group members. The percentage of men in each group ranged from 0% to 100%; on average, the percentage of men in each group was 50%.
Procedure Participants worked together in 4-person groups on problems randomly selected from previous versions of the Graduate Management Aptitude Test (GMAT), a test used primarily for selection into graduate schools of business. They were given 40 minutes to work together on as many problems as possible. Groups reported their answers on a single form, and experimenters explicitly instructed participants to work together as a group. They were told that the highest performing group experimentwide would receive $100. Their performance would be coded similar to how it would be scored on similar standardized tests, in that they would be rewarded for correct answers and penalized for wrong answers. The total number of items answered correctly on average was 33.64 (SD ⫽ 11.99); the proportion of items answered correctly was .81 (SD ⫽ .10). Following the group task, participants privately completed a questionnaire.
Measures Status. Individuals rated all 4 members of their group, including themselves, on seven status-related items: how much each person received respect from other group members, made valuable contributions, demonstrated high ability, influenced group decisions, led the group’s activities, participated, and contributed to the group overall. Each item was rated on a scale from 1 (Not at all) to 7 (A great deal). Using the software program SOREMO (Kenny, 1995), we computed the target scores for each of the seven status ratings, which is essentially their peer-rated average for those ratings. We then computed an alpha reliability (internal consistency) of the seven peer-rated indices that constituted our status measure. The peer-rated indices had an alpha of .95, indicating high reliability of the status measure. Furthermore, there were significant amounts of peer-rated variance at the p ⬍ .05 level in all seven status indices (M ⫽ .36), indicating that there was high peer consensus as to who was high and who was low in status (Kenny, 1994). Status self-enhancement. For analyses examining the consequences of self-enhancement, we again calculated a self-enhancement index for status based on Kwan et al.’s (2004) technique. We averaged the seven status self-enhancement indices to form an overall score (␣ ⫽ .87). The social relations model calculations remove any group-level dependence in the data. Social acceptance. Individuals rated the other members of their group on five dimensions chosen to be relevant to the context of Study 2: how much they would want to work with the target again, how much the target had earned their trust as a coworker, how similar they felt to the target, how much in common they felt they had with the target, and how different they seemed from the target (reverse-scored). They also rated how much their fellow team members would want to work with them again, how much they had earned their fellow team members’ trust, and so on. Each item was rated on the same scale, from 1 (Not at all) to 7 (A great deal). We used these items to measure social acceptance because the experiment was
SELF-PERCEPTIONS OF STATUS presented to participants as a simulation of work in organizations; thus, for example, rating their desire to work together in the future seemed more appropriate than rating their likeability. We used SOREMO to compute the peer-rated (or target) effects of the five social acceptance ratings. There were significant amounts of variance for only two of the five peer-rated indices: how much they would want to work with the target again and how much the target had earned their trust (M ⫽ 0.06). Thus, we used those two indices to measure social acceptance (␣ ⫽ .81). Again, the social relations model indices removed any grouplevel dependence in the data. Acceptance self-enhancement. We calculated a self-enhancement index for acceptance by computing Kwan et al.’s (2004) index from an average of the acceptance ratings (␣ ⫽ .69). The acceptance selfenhancement index correlated significantly but not highly with the status self-enhancement index, r(432) ⫽ .28, p ⬍ .01. This suggests that individuals who perceived their status in an overly positive way were slightly more likely to perceive their social acceptance in an overly positive way but that these two forms of self-enhancement were also somewhat distinct. Intragroup conflict. Individuals rated on a scale from 1 (Not at all) to 7 (Extremely) how much conflict their group experienced with the item “To what extent did group members disagree over the task solutions during the group task?” (intraclass correlation ⫽ .60, p ⬍ .01). These ratings were aggregated within group (M ⫽ 2.78, SD ⫽ .78). Concern for social acceptance. Individuals rated two items, “How aware were you that you were being evaluated by the other group members?” and “How worried were you about saying/doing the wrong thing?”, on a scale from 1 (Not at all) to 7 (Extremely). These items were combined to form one measure of concern for social acceptance (␣ ⫽ .61). After controlling for group effects, M ⫽ 0.00, SD ⫽ 1.21. Ambiguity of status dimensions. Eight independent judges rated the seven dimensions of status on how ambiguous versus unambiguous they are. Specifically, the judges were told that in a recent study, groups of 4 people worked on GMAT problems together and that following the task, each group member was asked to rate all group members on these seven dimensions. We asked the judges to imagine themselves making these ratings and to gauge the ambiguity or lack of ambiguity of each dimension using a scale from 1 (Very
1103
ambiguous, less objective) to 7 (Very unambiguous, more objective). The raters agreed highly on which dimensions were more or less ambiguous (␣ ⫽ .86). We created self- and peer-rated scores for each individual on the status dimensions rated as more ambiguous (i.e., received respect from other group members, influenced group decisions, and demonstrated high ability; average score ⫽ 4.20) and those rated as less ambiguous (i.e., participated, made valuable contributions, led the group’s activities, and contributed to the group overall; average score ⫽ 5.43).
Results and Discussion Hypothesis 1: Group Members Who Engage in Status Self-Enhancement Will Be Less Accepted by Other Members Than Those Who Perceive Their Status Accurately Although Study 2 did not have a longitudinal structure and, thus, we could not test for lagged effects, we were able to test for crosssectional associations between status self-enhancement and social acceptance target scores. Both of these scores were independent of group effects and thus appropriate for regression analyses. As expected, individuals who engaged in status self-enhancement were less socially accepted ( ⫽ ⫺.25, p ⬍ .01; unstandardized B ⫽ ⫺.24, SE ⫽ .04). This effect is illustrated in Figure 1. Moderated multiple regression analysis indicated no interaction with sex, suggesting that this effect was equally strong for both men and women. Also, similar to Study 1, there was no evidence of a curvilinear relation between status self-enhancement and acceptance. In a model that added a quadratic term for status self-enhancement, we did not find evidence for a curvilinear relation between status self-enhancement and acceptance (b ⫽ .004, SE ⫽ .098, p ⫽ .968). Thus, status self-effacers were liked more than accurate status perceivers, who in turn were liked more than status self-enhancers.
Social acceptance (z-scored)
0.3 0.2 0.1 0 Self - ef f ac ers
Self -enhanc ers
-0.1 -0.2 -0.3
Figure 1. Study 2: The social consequences of status self-enhancement. For illustrative purposes, presented are average levels of acceptance for status self-enhancers, or individuals whose self-perceived status was higher than their actual (or peer-rated) status, and for status self-effacers, or individuals whose self-perceived status was lower than their actual status. No individuals perceived their status with perfect accuracy. We used regression analyses to test the significance of the effect of status self-enhancement on social acceptance.
1104
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
Hypothesis 2: Face-to-Face Group Members Will Perceive Their Own Status Accurately Elevation accuracy. To examine elevation accuracy, we ran a three-factor (2 ⫻ 2 ⫻ 5) ANOVA at the group level, similar to Study 1; the three factors were self- versus peer perceptions, status versus acceptance perceptions, and the sex composition of the group (all male, majority male, balanced, majority female, and all female). This analysis revealed a main effect for self- versus peer perception, F(1, 103) ⫽ 72.32, p ⬍ .01, indicating a general self-enhancement effect across perceptions of status and acceptance. However, consistent with expectations, there was also a significant interaction in that this self-enhancement effect was moderated by status versus acceptance perceptions, F(1, 103) ⫽ 111.25, p ⬍ .01. As illustrated in Table 4, participants overestimated their social acceptance on average, F(1, 103) ⫽ 128.06, p ⬍ .01, consistent with Study 1. However, participants again underestimated their status on average, F(1, 103) ⫽ 3.14, p ⬍ .05. This indicates, again, that individuals self-enhanced in perceiving their acceptance but self-effaced in perceiving their status. The groups’ sex composition was not related to perceptions of status or social acceptance, nor did any interactions involving group sex composition emerge. Differential accuracy: Correlations between self- and peer perceptions. To examine differential accuracy, we followed Kenny et al. (2006) and correlated self-perceived status with actual (peerrated) status, while partialing the group effects with 107 dummy variables to represent membership in the 108 groups. Consistent with Study 1, participants showed high differential accuracy in perceiving their status. The relation between self-perceptions of status and peer-rated status was substantial (partial r ⫽ .626, p ⬍ .01). The self–peer correlation for social acceptance was also significant, but lower (partial r ⫽ .155, p ⬍ .01). We tested whether accuracy was significantly higher for perceptions of status than for perceptions of acceptance by using Raghunathan et al.’s (1996) method. As expected, individuals were more accurate in perceiving their status than their acceptance (Z ⫽ 4.45, p ⬍ .01). We found no sex difference in individuals’ accuracy in perceiving their status or acceptance in moderated multiple regression analyses (Aiken & West, 1991). Differential accuracy: Comparing accuracy and consensus. Also consistent with Study 1, individuals were as accurate in perceiving their own status as peers were in perceiving that individual’s status. That is, self–peer agreement (covariance M ⫽ .65, SD ⫽ .97) was not significantly different from peer–peer agree-
Table 4 Elevation Accuracy: Self- and Peer Ratings of Status and Social Acceptance in Study 2 Dimension
Self-perception
Peer rating
Status Social acceptance
5.30 (0.58) 5.96 (0.57)
5.40 (0.45) 5.01 (0.73)
Note. Standard deviations are in parentheses. We consider peer ratings of status and acceptance to reflect the person’s actual status and acceptance, respectively. Self-perceptions of status are significantly lower than peer ratings. Self-perceptions of acceptance are significantly higher than peer ratings.
ment (target variance M ⫽ .65, SD ⫽ .85), F(1, 107) ⫽ 0.00, ns. This indicates that individuals’ perceptions of their status were as close to their actual status as any other group member’s perception of their status. Was accuracy in self-perceptions of status due to the unambiguous nature of status? Prior research has found that individuals self-enhance more on dimensions that are relatively ambiguous than on those that are less ambiguous. To examine whether this pattern emerged in different dimensions of status, we ran a twofactor (2 ⫻ 2) within-subjects ANOVA at the group level with self- versus peer perceptions and unambiguous versus ambiguous status dimensions as the two within-subjects factors. This analysis revealed a main effect for self- versus peer perception, F(1, 107) ⫽ 7.50, p ⬍ .01, again indicating an overall self-effacement effect in status perceptions. It also revealed a main effect for the ambiguity of the status dimension, F(1, 107) ⫽ 24.31, p ⬍ .01, indicating higher self- and peer ratings for the unambiguous dimensions. However, there was not a significant interaction, in that individuals were equally self-effacing on both the more ambiguous dimensions (self-rating M ⫽ 5.22, SD ⫽ .62, compared with peer rating M ⫽ 5.34, SD ⫽ .48) and the less ambiguous status dimensions (self-rating M ⫽ 5.36, SD ⫽ .59, compared with peer rating M ⫽ 5.45, SD ⫽ .45), F(1, 107) ⫽ 1.92, p ⫽ .17. In fact, the direction of the effect was such that individuals were more self-effacing on the more ambiguous dimensions than on the less ambiguous dimensions (not vice versa). In terms of differential accuracy, we found that accuracy was high for the more ambiguous dimensions of status (partial r ⫽ .535, p ⬍ .01), as well as for the less ambiguous, more concrete dimensions of status (partial r ⫽ .646, p ⬍ .01). Therefore, although the lack of ambiguity in some dimensions of status might have helped boost the accuracy with which individuals perceived their status, individuals were still substantially accurate in viewing the more ambiguous dimensions of status.
Hypothesis 3: Individuals’ Concern for Social Acceptance Keeps Their Self-Perceptions of Status in Check The concern for acceptance in the group was related negatively to status self-enhancement ( ⫽ ⫺.08; unstandardized B ⫽ ⫺.06, SE ⫽ .03), a hypothesized effect that was significant at p ⬍ .05 by a one-tailed test (two-tailed p ⫽ .07). Thus, although individuals on average engaged in status self-effacement, individuals more concerned about being socially accepted were especially likely to do so. This finding supports the idea that the fear of being less socially accepted attenuates self-enhancement tendencies in perceiving one’s status and even leads people to self-efface.
Hypothesis 4: Status Self-Enhancement Instigates Conflict in the Group The more individual group members engaged in status selfenhancement, the higher the level of conflict there was in the group as a whole. On the group level, the correlation between average status self-enhancement levels in groups and intragroup conflict was r(108) ⫽ .23, p ⬍ .01. We illustrate this relation in Figure 2, categorizing groups as self-enhancing if the average self-rating in the group was higher than the average peer rating in the group or as self-effacing if the average self-rating in the group was lower
SELF-PERCEPTIONS OF STATUS
than the average peer rating in the group. As shown, groups whose members self-enhanced on average (n ⫽ 46) engaged in more conflict than did groups whose members self-effaced on average (n ⫽ 60). On an exploratory level, we next examined whether the relation between status self-enhancement and intragroup conflict was mediated by lower levels of interpersonal liking. That is, did groups whose members engage in status self-enhancement experience more conflict because they liked each other less? Mediation analyses did not support this account. When controlling for average levels of liking in groups, the relation between status selfenhancement and group conflict was not significantly reduced, Sobel test t(108) ⫽ ⫺.359, p ⫽ .719. Thus, the effect of status self-enhancement on intragroup conflict was not mediated by lower levels of interpersonal liking. Intragroup conflict was also related to lower group performance, r(82) ⫽ ⫺.31, p ⬍ .01, suggesting the more a group engaged in conflict, the worse it performed. However, there was not a significant direct relation between a group’s average levels of status self-enhancement and its performance; the direct correlation between average levels of status self-enhancement and group performance was r(82) ⫽ ⫺.10, ns. Because groups’ performance was measured as a proportion of correct to incorrect answers, we then looked separately at the raw number of problems answered correctly and the number of problems answered incorrectly. This analysis provided an intriguing picture: Groups whose members exhibited more status self-enhancement answered more problems correctly, r(82) ⫽ .24, p ⬍ .05, but also answered more problems incorrectly, r(82) ⫽ .16, p ⫽ .08, though the latter correlation was marginally significant. Therefore, these findings depict groups
with more status self-enhancers as producing more, but not always accurate, answers. Finally, we found that self-enhancement in perceptions of social acceptance was not related to intragroup conflict, r(108) ⫽ .08, ns. This supports our argument that self-enhancement in perceiving one’s social acceptance does not damage individuals’ actual acceptance because it does not provoke conflict. When group members perceived how much they were trusted and accepted in an overly positive way, this did not seem to disrupt the group’s harmony.
Summary The findings from Study 2 are highly consistent with those from Study 1. Individuals were accurate in perceiving their status and even self-effaced; at the same time, they perceived their social acceptance in an overly positive way. Furthermore, self-enhancers were again less likely to be socially accepted by other members of their group. The findings from Study 2 also extend our previous results. First, individuals who were more concerned about their acceptance in the group were less likely to engage in status self-enhancement. This lends further support to the idea that the need to belong helps keep individuals’ perceptions of their status in check. Second, groups with more status self-enhancers engaged in more conflict, and this conflict, in turn, was related to lower group performance, providing some insight into why self–peer agreement in status perceptions is so important: Group members who agree on their status cohere better as a group, whereas those who disagree instigate more conflict and disruptions. Furthermore, we did not find a
3.5
Intra-Group Conflict
3.25
3
2.75
2.5
2.25
2 Self - ef f ac ing gr oups
1105
Self - enhanc ing gr oups
Figure 2. Study 2: Status self-enhancement and intragroup conflict. For illustrative purposes, presented are intragroup conflict levels for self-enhancing groups, or groups whose members’ self-perceived status was higher on average than their actual (or peer-rated status), and for self-effacing groups, or groups whose members’ self-perceived status was lower on average than their actual status. In no groups were self-perceptions of status exactly equal to peer perceptions. We used regression analyses to test the significance of the effect of status self-enhancement on intragroup conflict.
1106
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
significant relation between acceptance self-enhancement and group conflict; though, again, we hesitate to make too much of a null finding, considered along with our other findings it is consistent with the idea that individuals are freer to engage in acceptance self-enhancement because this form of self-enhancement does not disrupt group functioning. Finally, we found that individuals were accurate and self-effacing on both ambiguous and less ambiguous elements of status, which helps rule out the alternative explanation that the accuracy with which individuals perceive their status is simply due to its unambiguous nature.
General Discussion Accuracy in Self-Perceptions of Status In two studies, one of which included four separate group assessments, we found consistent evidence that people generally did not engage in status self-enhancement but instead perceived their status with considerable accuracy. Self–peer correlations were high, on average .50, across all group assessments. It is notable that self-perceptions of status were accurate even in groups of relative strangers that collaborated for short periods of time—as little as 20 minutes. Furthermore, self-perceptions of status were consistently accurate across group tasks—whether groups members worked collaboratively, competed with each other for scarce resources, or even did not have any clear taskrelated goal (as in the informal discussion groups in Week 2 of Study 1). Men and women were also equally accurate in perceiving their status in that there was not a single sex difference in any of our assessments. Finally, individuals were accurate in perceiving the more ambiguous elements of status (e.g., how respected they were) as well as the less ambiguous elements of status (e.g., how much they participated). Taken together, the current findings suggest that accuracy in self-perceptions of status emerges quickly in various kinds of group settings, among individuals of both sexes, and on diverse components of status. Our studies also provide insight as to why individuals are accurate rather than self-enhancing in perceiving their status, even though status self-enhancement might boost self-esteem (e.g., Barkow, 1975): When group members engaged in status selfenhancement, they were less socially accepted—they were less liked, viewed as less enjoyable to be around, not as preferred as future work partners, and less trusted by others. The concern over maintaining belongingness in groups thus might be a driving factor in keeping individuals from forming overly positive perceptions of their status. In support of this explanation, we found a relation in Study 2 between humility in status perceptions and the concern for social acceptance: the more people cared about their membership in their group, the less likely they were to status self-enhance. Along a more speculative vein, might there be evolutionary origins for the tendency to accurately view one’s status? Throughout human evolutionary history, people have lived in face-to-face groups because social living provides survival and reproduction advantages over living in isolation (e.g., Baumeister and Leary, 1995; Buss & Kenrick, 1998; Cosmides, Tooby, & Barkow, 1992); those who maintained positive inclusion and acceptance in faceto-face groups were more likely to survive and pass on their genes to the next generation. Status hierarchies are thought to have been part of human social groups throughout evolutionary history (e.g.,
Buss, 1999; Wright, 1994). Therefore, humans might have evolved an acute sensitivity to hierarchical dynamics in face-to-face groups, including the tendency to view their status highly accurately— given the costs of inaccuracy, this sensitivity would help maintain inclusion and enjoy better reproductive success. In contrast, those who viewed their status in an overly positive light might have been rejected and ostracized and found it more difficult to survive and reproduce. Of course, it is also important to consider the possibility that participants refrained from engaging in status self-enhancement because they had little motivation to exaggerate their status in the first place. Participants may have cared little about their status because these were temporary groups of strangers, not of their own choosing, that simply fulfilled a course requirement; engaging in status self-enhancement might have had little benefit for their self-esteem. Though this is possible, we do not believe this explanation accounts for our findings for two reasons. First, previous work has shown that even in temporary groups of strangers such as the ones we studied, individuals’ status has a strong effect on their self-esteem (Leary et al., 2001). Thus, people seem to care about their status even in these laboratory-based, ad hoc groups. This is perhaps because participants are among their peers— colleagues from the same university, with whom they are implicitly comparing themselves throughout their entire enrollment. Second, if participants cared little about these groups, we would expect low variance in their level of caring about their status in these groups and would not expect to be able to explain this variance. Yet we did find variance, and more importantly, we found the opposite of what would have been expected by a lack of engagement explanation: the more people cared about their membership in these groups, the less likely they were to engage in status self-enhancement. This suggests that participants who cared little about these laboratory groups were more likely to engage in more status self-enhancement rather than less. It is also important to consider that all of our analyses were correlational in nature, and we could not rule out the possibility that the link between status self-enhancement and decreased social acceptance was driven by variables that we did not assess directly, such as personality traits. For example, rude, hostile people might tend to engage in status self-enhancement, and their generally pugnacious ways might also lead them to be disliked by others. It is important for future research to rule out this possibility using experimental designs or by measuring and controlling for variables, such as personality traits, that might play a role in this process. Finally, our results were obtained using mostly participants from the United States. As these participants came from a relatively individualistic Western culture, our results might change if we examined individuals from Eastern cultures. For example, individuals from Eastern cultures might be even less likely to engage in status self-enhancement because there might be more severe social costs for violating group status hierarchies (Heine, Lehman, Markus, & Kitayama, 1999).
Status Self-Effacement When there were deviations from peer ratings, or biases in self-perceptions of status, self-perceptions of status were not higher on average than peer ratings in any of our assessments but
SELF-PERCEPTIONS OF STATUS
in fact were lower than peer ratings in all assessments. That is, although individuals tended to be accurate in terms of differential accuracy, they also tended to be slightly effacing in terms of elevation accuracy. This indicates that not only did individuals refrain from engaging in status self-enhancement but also they tended to self-efface in perceiving their status. One explanation for this finding is that the need to belong and be included is so strong that people overcompensate, perceptually at least, to ensure that they do not engage in what Ridgeway and Berger (1986) called status violations, for example, talking too much or asserting their opinions too forcefully. Individuals might tend to be so concerned about their belongingness that they err on the side of being overly humble. This notion is similar to an argument posed in politeness research. Many norms of politeness lead people to be overly humble, cautious, and deferential because this helps them avoid interpersonal conflict (Keltner & Anderson, 2000). For example, it is common for individuals who accidentally bump into someone to apologize automatically, regardless of who may be at fault. Another possibility is that there are additional social benefits to status self-effacement that motivate people to be overly humble. In both studies, we found a linear relation between self-perceptions of status and social acceptance in that status self-effacers were liked more than accurate status perceivers, who in turn were liked more than status self-enhancers. Obtaining significant curvilinear effects is notoriously difficult (McClelland & Judd, 1993), and thus, the failure to obtain a curvilinear effect might have been due to a lack of statistical power. However, if status self-effacers are indeed more socially accepted than accurate status perceivers, people might self-efface to further increase their liking and acceptance. With regard to this self-effacement effect, it is worth considering again whether our findings might be unique to the kinds of group contexts we studied. It might be the case that in these short-term groups of strangers, underestimating one’s status does not lead to forgoing any meaningful social benefits. Thus, in these groups, the benefits of being overly humble (i.e., more acceptance and liking) might outweigh the loss of some status-related rewards (e.g., forgoing decision-making control or the ability to express one’s ideas and opinions). In real-world groups in which individuals’ status is associated with more valuable rewards, underestimating one’s status may mean passing up on more meaningful benefits. For example, in organizational groups, individuals who do not take credit for their contributions might receive less monetary compensation and lose out on opportunities to increase their formal authority in the organization (Flynn, 2003). Finally, some individuals might be less likely to perceive their status with such humility. For example, narcissists have a grandiose sense of self and entitlement as well as a preoccupation with success and demands for admiration (see Morf & Rhodewalt, 2001, for a review). Narcissism has been linked to the tendency to self-enhance (Gosling, John, Craik, & Robins, 1998; John & Robins, 1994), which suggests that narcissists might be more likely to view their status in an overly positive, rather than overly humble, way.
Self-Enhancement in Perceptions of Social Acceptance In contrast to the accuracy we observed in self-perceptions of status, people did exhibit self-enhancement biases in perceiving their social acceptance. Individuals tended to form self-enhanced perceptions of how likeable they were, how much others trusted
1107
them, and how much others preferred them as future coworkers. This finding is important in part because it indicates that there is nothing unique about our participants or about the methods we used that led people to be less self-enhancing in general. Rather, there is something unique about self-perceptions of status that led to such high accuracy and humility. We believe individuals view their social acceptance in an overly positive way because doing so does not damage their actual acceptance in the same way as overestimating one’s status; people are freer to form positive illusions of their acceptance to help boost their self-esteem. In support of this argument, we found no effects of acceptance self-enhancement on group conflict; thus, groups are likely less prone to punish acceptance self-enhancers by alienating and ostracizing them. It is possible that acceptance self-enhancement has other negative social consequences we did not observe here. For example, if individuals overestimate their closeness with others, they may overestimate the likelihood of receiving political support when they need it. Such illusions of alliance might have severely detrimental consequences when, in times of political strife, individuals decide to take political risks or to engage in conflict with other individuals under the mistaken belief that they have the support of numerous friends and allies. Overestimating one’s closeness to others might also lead individuals to ask others for favors under the assumption that others will gladly comply. As recent research suggests, asking for favors in this way might elicit enmity because others are likely to comply out of obligation and then later resent the favor request (Flynn & Bohms, 2006). A remaining question is why individuals would keep their selfperceptions of status in check if they can fool themselves into thinking they are socially accepted. In other words, why would individuals care about maintaining social acceptance if they can simply construct an overly positive self-perception of acceptance? It is important to point out that although participants did engage in acceptance selfenhancement, they were also significantly accurate in viewing their social acceptance. Therefore, we believe individuals still feel the sting when they are socially rejected (Leary et al., 2001), even if they do have a somewhat overinflated view of their acceptance.
Why Is Status Self-Enhancement Socially Punished? Why would engaging in status self-enhancement decrease one’s social acceptance in the first place? We have proposed three possible reasons: Status self-enhancement challenges the existing status order and provokes conflict and discord in groups, it is seen as an illegitimate claim for rewards, and it is threatening to others because it signals a sense of superiority. Our group-level findings from Study 2 provide support for at least the first explanation. We found that groups whose members engaged in status selfenhancement experienced more conflict, and in turn, this conflict was related to lower group performance. In essence, these groups seemed to have too many cooks in the kitchen, with too many people trying to make decisions for the group and too few people deferring to others. Groups might therefore punish individuals who engage in status self-enhancement because such self-enhancement provokes intragroup conflict and undermines group progress toward collective goals. This explanation is given further support by the lack of relation between self-enhancement in perceptions of acceptance and intragroup conflict. Group functioning did not
1108
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
seem to be hindered when individuals viewed their acceptance in an overly positive way. Thus, groups might not alienate individuals who view their acceptance in an overly positive way because these individuals do no harm to the group. Future research should also examine the other two possible reasons why status self-enhancement is socially punished. Is status self-enhancement indeed seen as an illegitimate claim for rewards? Is it threatening to others because it signals a sense of superiority? In particular, evidence for this latter notion might provide broader insights as to why self-enhancement in general is sometimes linked with negative social outcomes (Paulhus, 1998) and other times linked with positive social outcomes (Taylor & Brown, 1988). We have argued that status in face-to-face groups is a zero-sum commodity, typically treated as a rank-order variable by groups; therefore, when individuals status self-enhance and signal their higher status, they necessarily signal others’ lower status. On a more general level, this idea implies that individuals incur social costs when they self-enhance on a dimension in which they directly compete with others—that is, when individuals’ expressed superiority necessarily implies others’ inferiority— but not when they self-enhance on noncompetitive dimensions. For example, if an individual on a tennis team is in competition with other members to be the top player on the team and forms an overly positive perception of his or her tennis abilities, this might lead to social rejection and ostracism on the team. If that same individual forms an overly positive perception on an unrelated dimension (e.g., the ability to play chess), however, this might have little to no effect on the individual’s acceptance on the tennis team because the individual is not in competition with the other team members for chess-playing prowess. The issue of why people are punished for engaging in status self-enhancement also prompts questions about who is most likely to reject and ostracize status self-enhancers. We can envision three possible scenarios: First, all group members might reject status self-enhancers. Second, only individuals higher in the status hierarchy might dislike and reject status self-enhancers because it is only those individuals whose authority is being subverted. Third, only those closest in the hierarchy to the status self-enhancers (i.e., those slightly lower or higher in the hierarchy than the status self-enhancers) might dislike and reject them because those are the individuals most directly competing for status with the status self-enhancers. We explored these possibilities and did not find any evidence for the latter two hypotheses. Although our analyses are only suggestive and not definitive, status self-enhancers were not more likely to provoke rejection by those higher than them in the hierarchy or by those closest to them in the hierarchy; status self-enhancement seemed to provoke rejection by the group as a whole. For example, in Study 2, we ranked each group member according to his or her peer-rated status. Status self-enhancement did not lead to less liking by those ranked higher in the status hierarchy (r ⫽ ⫺.12) than those ranked lower in the status hierarchy (r ⫽ ⫺.12; the lack of reliability due to fewer data points likely constrained the magnitude of each of these correlations). Status self-enhancement also did not instigate less liking by those immediately above or below in the status hierarchy (r ⫽ ⫺.13) than those two rankings away in the hierarchy (r ⫽ ⫺.11), or three rankings away in the hierarchy (r ⫽ ⫺.14).
Just as important as understanding why status self-enhancement would be socially punished, it is worth exploring why status self-effacement would be socially rewarded. Why might status self-effacers be liked and accepted more than accurate status perceivers? First, status self-effacers might signal a particularly high degree of selflessness or an extreme willingness to put the group’s interests above their own (Ridgeway, 1982). Status selfeffacers essentially forgo the social rewards they are entitled to in the eyes of the group based on the contributions they make to the collective. Groups might thus be even more accepting of status self-effacers because of the sacrifices they make for the good of the group. Further, status self-effacers might make other group members feel good about themselves because they signal others’ relative superiority. Inasmuch as status is a zero-sum commodity in groups (Bales, 1950), self-effacers’ humble behavior might complement others and boost others’ self-esteem.
Would Status Self-Enhancement Ever Provide Social Benefits? Some theorists have argued that having an overly positive view of one’s status is adaptive because it helps convince others that one has high status (e.g., Krebs & Denton, 1997). Once convinced of an individual’s lofty standing, others might begin treating the individual with more deference and respect, which would increase the individual’s actual status. This notion is similar to the theorized rationale behind conspicuous consumption, wherein individuals purchase highly visible and expensive objects such as exotic cars or large houses (Frank, 1985). These purchases, it is thought, communicate one’s high status to others, with the aim of gaining friends in a higher social bracket or having a wider choice of romantic partners. Although we did not find evidence for a self-fulfilling prophecy in our studies (there were no increases in status over time when individuals engaged in status self-enhancement), we believe there is a way to reconcile Krebs and Denton’s (1997) arguments with our findings. Specifically, we believe that status self-enhancement has social costs primarily within groups that have established clear status hierarchies. However, in situations where individuals are interacting with individuals outside their group or when the status order of a given group is not clear, it is possible that overly positive self-perceptions of status may have social benefits. For example, if an individual falsely conveys to people outside his or her organization that he or she has high status, those people may not react negatively because they are not aware of the individual’s actual place within the organization. In fact, given their lack of information about the individual’s actual place in the organization’s hierarchy, they might believe the individual and thus provide him or her with the deference and respect the individual seeks. Moreover, self-enhancement in status perceptions might benefit individuals in groups with no well-defined hierarchy. In group contexts, overly positive self-perception of status may have the effects that Krebs and Denton (1997) predicted; it might lead to actual increases in status because other group members would show that individual deference and respect. By doing so, they would give the individual opportunities to lead the group, which in turn could lead to actual increases in status (Berger et al., 1972).
SELF-PERCEPTIONS OF STATUS
Conclusion Perhaps the most important implication of our findings is that they present a significant boundary condition to the positive illusions perspective. As previous work has shown, individuals form overly positive perceptions of themselves across variety of dimensions to maintain their self-esteem. Because people’s perceptions of their status lie at the core of their self-esteem, individuals would thus seem particularly likely to engage in status self-enhancement to maintain self-esteem (Barkow, 1975). However, we found that individuals did not engage in status self-enhancement but instead perceived and interpreted information about their status accurately, even if such information might potentially harm their self-esteem. On a broader level, we believe the current findings reflect a tension between the need for self-esteem and the need to belong. Although individuals have a strong need to think positively about their attributes to maintain self-esteem, they have an even stronger need to socially belong (Baumeister & Leary, 1995; Maslow, 1968). Many times, these two forces can work independently from each other; for example, an individual can believe he or she is a better driver than he or she actually is, and such a self-serving bias does not damage the individual’s interpersonal relationships and belongingness. However, possessing overly positive perceptions of one’s status does lead to decreased social acceptance, and thus, the individual’s need for self-esteem in this case is working against the need to socially belong. Consistent with prior theorizing, the need to belong outweighs the individual’s need to inflate self-esteem.
References Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Allison, S. T., Messick, D. M., & Goethals, G. R. (1989). On being better but not smarter than others: The Muhammad Ali effect. Social Cognition, 7, 275–295. Anderson, C., & Berdahl, J. L. (2002). The experience of power: Examining the effects of power on approach and inhibition tendencies. Journal of Personality and Social Psychology, 83, 1362–1377. Anderson, C., John, O. P., Keltner, D., & Kring, A. M. (2001). Who attains social status? Effects of personality traits and physical attractiveness in social groups. Journal of Personality and Social Psychology, 81, 116 –132. Aron, A., Melinat, E., Aron, E. N., Vallone, R. D., & Bator, R. (1997). The experimental generation of interpersonal closeness: A procedure and some preliminary findings. Personality and Social Psychology Bulletin, 23, 363–377. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Chicago: University of Chicago Press. Barkow, J. H. (1975). Prestige and culture: A biosocial interpretation. Current Anthropology, 16, 553–562. Baumeister, R. F. (1998). The self. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 1, pp. 680 –740). Boston: McGraw-Hill. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. Berger, J., Cohen, B. P., & Zelditch, M. (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., Rosenholtz, S. J., & Zelditch, M. (1980). Status organizing processes. Annual Review of Sociology, 6, 479 –508. Blau, P. M. (1964). Exchange and power in social life. New York: Wiley. Bryk, A., & Raudenbush, S. W. (1992). Hierarchical linear models for social and behavioral research: Applications and data analysis methods. Newbury Park, CA: Sage.
1109
Bugental, D. B., & Lewis, J. C. (1999). The paradoxical misuse of power by those who see themselves as powerless: How does it happen? Journal of Social Issues, 55, 51– 64. Buss, D. M. (1999). Human nature and individual differences: The evolution of human personality. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (2nd ed., pp. 31–56). New York: Guilford Press. Buss, D. M., & Kenrick, D. T. (1998). Evolutionary social psychology. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 2, pp. 982–1026). Boston: McGraw-Hill. Clark, C. (1990). Emotions and micropolitics in everyday life: Some patterns and paradoxes of “place.” In T. D. Kemper (Ed.), Research agendas in the sociology of emotions (pp. 305–333). Albany: State University of New York Press. Colvin, C. R., Block, J., & Funder, D. C. (1995). Overly positive selfevaluations and personality: Negative implications for mental health. Journal of Personality and Social Psychology, 68, 1152–1162. Cosmides, L., Tooby, J., & Barkow, J. H. (1992). Introduction: Evolutionary psychology and conceptual integration. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture (pp. 3–18). New York: Oxford University Press. Cronbach, L. J. (1955). Processes affecting scores on “understanding others” and “assumed similarity.” Psychological Bulletin, 52, 177–193. Dunning, D., Leuenberger, A., & Sherman, D. A. (1995). A new look at motivated inference: Are self-serving theories of success a product of motivational forces? Journal of Personality and Social Psychology, 69, 58 – 68. Dunning, D., Meyerowitz, J. A., & Holzberg, A. D. (1989). Ambiguity and self-evaluation: The role of idiosyncratic trait definitions in self-serving assessments of ability. Journal of Personality and Social Psychology, 57, 1082–1090. Flynn, F. J. (2003). How much should I give and how often? The effects of generosity and frequency or favor exchange on social status and productivity. Academy of Management Journal, 46, 539 –553. Flynn, F. J., & Bohms, V. (2006). Just ask: How people underestimate the likelihood of compliance in favor exchange. Manuscript submitted for publication. Frank, R. H. (1985). Choosing the right pond: Human behavior and the quest for status. New York: Oxford University Press. Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). From power to action. Journal of Personality and Social Psychology, 85, 453– 466. Gameworks Creations. (1995). Beyond Balderdash. Pawtucket, RI: Hasbro. Gelb, L. H. (1981, May 31). How Haig is recasting his image. The New York Times, p. 23. Gergen, K. J., & Wishnov, B. (1965). Others’ self-evaluations and interaction anticipation as determinants of self-presentation. Journal of Personality and Social Psychology, 2, 348 –358. Goffman, E. (1967). Interaction ritual. New York: Anchor Books. Goldhamer, H., & Shils, E. A. (1939). Types of power and status. American Journal of Sociology, 45, 171–182. Gosling, S. D., John, O. P., Craik, K. H., & Robins, R. W. (1998). Do people know how they behave? Self-reported act frequencies compared with on-line codings by observers. Journal of Personality and Social Psychology, 74, 1337–1349. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603– 618. Heaven, P. C. (1986). Authoritarianism, directiveness, and self-esteem revisited: A cross-cultural analysis. Personality and Individual Differences, 7, 225–228. Heine, S. J., & Lehman, D. R. (1997). The cultural construction of self-enhancement: An examination of group-serving biases. Journal of Personality and Social Psychology, 72, 1268 –1283. Heine, S. J., Lehman, D. R., Markus, H., & Kitayama, S. (1999). Is there
1110
ANDERSON, SRIVASTAVA, BEER, SPATARO, AND CHATMAN
a universal need for positive self-regard? Psychological Review, 106, 766 –794. Hogan, R. (1983). A socioanalytic theory of personality. In M. Page (Ed.), Nebraska symposium on motivation: Vol. 30. Personality—Current theory and research (pp. 55– 89). Lincoln: University of Nebraska Press. Homans, G. C., (1951). The human group. London: Routledge & Kegan Paul. John, O. P., & Robins, R. W. (1994). Accuracy and bias in self-perception: Individual differences in self-enhancement and the role of narcissism. Journal of Personality and Social Psychology, 66, 206 –219. Jones, S. C., & Shrauger, J. S. (1970). Reputation and self-evaluation as determinants of attractiveness, Sociometry, 33, 276 –286. Jussim, L. (1991). Social perception and social reality: A reflectionconstruction model. Psychological Review, 98, 54 –73. Keltner, D., & Anderson, C. (2000). Saving face for Darwin: The functions and uses of embarrassment. Current Directions in Psychological Science, 9, 187–192. Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110, 265–284. Kenny, D. A. (1994). Interpersonal perception. New York: Guilford Press. Kenny, D. A. (1995). SOREMO Version V.2: A FORTRAN program for the analysis of round-robin data structures. Unpublished manuscript, University of Connecticut. Kenny, D. A. (1996). The design and analysis of social-interaction research. Annual Review of Psychology, 47, 59 – 86. Kenny, D. A., & Campbell, D. T. (1999). On the measurement of stability in over-time data. Journal of Personality, 57, 445– 481. Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York: Guilford Press. Kenny, D. A., & La Voie, L. (1984). The social relations model. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 18, pp. 142–182). Orlando, FL: Academic Press. Krebs, D. L., & Denton, K. (1997). Social illusions and self-deception: The evolution of biases in person perception. In J. A. Simpson & D. T. Kenrick (Eds.), Evolutionary social psychology (pp. 21– 48). Mahwah, NJ: Erlbaum. Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it: How difficulties in recognizing one’s own incompetence lead to inflated selfassessments. Journal of Personality and Social Psychology, 77, 1121–1134. Kwan, V. S., John, O. P., Kenny, D. A., Bond, M. H., & Robins, R. W. (2004). Reconceptualizing individual differences in self-enhancement bias: An interpersonal approach. Psychological Review, 111, 94 –110. Leary, M. R., Cottrell, C. A., & Phillips, M. (2001). Deconfounding the effects of dominance and social acceptance on self-esteem. Journal of Personality and Social Psychology, 81, 898 –909. Lee, M. T., & Ofshe, R. (1981). The impact of behavioral style and status characteristics on social influence: A test of two competing theories. Social Psychology Quarterly, 44, 73– 82. Maslow, A. H. (1968). Toward a psychology of being. New York: Van Nostrand. Mazur, A. (1973). A cross-species comparison of status in small established groups. American Sociological Review, 38, 513–530. McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and moderator effects. Psychological Bulletin, 114, 376 –390. Messick, D. M., Bloom, S., Boldizar, J. P., & Samuelson, C. D. (1985). Why we are fairer than others. Journal of Experimental Social Psychology, 21, 480 –500. Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of narcissism: A dynamic self-regulatory processing model. Psychological Inquiry, 12, 177–196. Paulhus, D. L. (1998). Interpersonal and intrapsychic adaptiveness of trait self-enhancement: A mixed blessing? Journal of Personality and Social Psychology, 74, 1197–1208.
Pfeffer, J., & Cialdini, R. B. (1998). Illusions of influence. In R. M. Kramer & M. A. Neale (Eds.), Power and influence in organizations (pp. 1–20). Thousand Oaks, CA: Sage. Powers, T. A., & Zuroff, D. C. (1986). Interpersonal consequences of overt self-criticism: A comparison with neutral and self-enhancing presentations of self. Journal of Personality and Social Psychology, 54, 1054 –1062. Raghunathan, T. E., Rosenthal, R., & Rubin, D. B. (1996). Comparing correlated but nonoverlapping correlations. Psychological Methods, 1, 178 –183. Raskin, R., Novacek, J., & Hogan, R. (1991). Narcissistic self-esteem management. Journal of Personality and Social Psychology, 60, 911–918. Ridgeway, C. L. (1982). Status in groups: The importance of motivation. American Sociological Review, 47, 76 – 88. Ridgeway, C. L. (1987). Nonverbal behavior, dominance, and the basis of status in task groups. American Sociological Review, 52, 683– 694. Ridgeway, C. L., & Berger, J. (1986). Expectations, legitimation, and dominance behavior in task groups. American Sociological Review, 51, 603– 617. Ridgeway, C. L., & Diekema, D. (1989). Dominance and collective hierarchy formation in male and female task groups. American Sociological Review, 54, 79 –93. Robins, R. W., & Beer, J. S. (2001). Positive illusions about the self: Short-term benefits and long-term costs. Journal of Personality and Social Psychology, 80, 340 –352. Robins, R. W., & John, O. P. (1997). Effects of visual perspective and narcissism on self-perception: Is seeing believing? Psychological Science, 8, 37– 42. Roethlisberger, F. J., & Dickson, W. J. (1938). Management and the worker: An account of a research program conducted by the Western Electric Company, Hawthorne Works, Chicago. Cambridge, MA: Harvard University Press. Rogosa, D. (1980). A critique of cross-lagged correlation. Psychological Bulletin, 88, 245–258. Srivastava, S., & Beer, J. S. (2005). How self-evaluations relate to being liked by others: Integrating sociometer and attachment perspectives. Journal of Personality and Social Psychology, 89, 966 –977. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103, 193–210. Taylor, S. E., Lerner, J. S., Sherman, D. K., Sage, R. M., & McDowell, N. K. (2003). Portrait of the self-enhancer: Well adjusted and well liked or maladjusted and friendless? Journal of Personality and Social Psychology, 84, 165–176. Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: Wiley. Weisman, S. R. (1982, June 26). Haig resigns over foreign policy course, but cites no issues; Reagan names Schultz. The New York Times, p. 1. West, S. G., Biesanz, J. C., & Pitts, S. C. (2000). Causal inference and generalization in field settings: Experimental and quasi-experimental designs. In H. T. Reis & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 40 – 84). New York: Cambridge University Press. Whyte, W. F. (1943). Street corner society: The social structure of an Italian slum. Chicago: Chicago University Press. Wiggins, J. S. (1979). A psychological taxonomy of trait-descriptive terms: The interpersonal domain. Journal of Personality and Social Psychology, 37, 395– 412. Wright, R. (1994). The moral animal: The new science of evolutionary psychology. New York: Pantheon Books.
Received March 1, 2006 Revision received June 11, 2006 Accepted June 17, 2006 䡲
PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES
Emotion Without a Word: Shame and Guilt Among Rara´muri Indians and Rural Javanese Seger M. Breugelmans
Ype H. Poortinga
Tilburg University
Tilburg University and the Catholic University of Leuven
The Rara´muri Indians in Mexico use 1 word for guilt and shame. In this article, the authors show that the Rara´muri nevertheless differentiate between shame and guilt characteristics, similar to cultural populations that use 2 words for these emotions. Emotion-eliciting situations were collected among the Rara´muri and among rural Javanese and were rated on shame and guilt by Dutch and Indonesian students. These ratings were used to select 18 shame-eliciting and guilt-eliciting situations as stimuli. The Rara´muri (N ⫽ 229) and the Javanese (N ⫽ 213) rated the situations on 29 emotion characteristics that previously had been found to differentiate shame from guilt in an international student sample. For most characteristics, a pattern of differentiation similar to that found among the students was found for both the Javanese and the Rara´muri. Keywords: emotion, shame, guilt, cross-cultural, culture
The relationship between emotion words and emotion processes is important for both methodological and theoretical reasons. First, almost all culture-comparative studies use emotion words, either as independent or dependent variables (for exceptions, see Ekman & Friesen, 1971; Mesquita, 2001). As a consequence, differences in the emotion lexicon limit the reach of such studies to only those emotions that can be readily translated. Second, theoretical debates about the extent of cultural variation in emotions are still far from settled. Although most scholars agree that cultures differ in aspects of the emotion process, there is much disagreement on the pervasiveness of these differences. A major reason for disagreement is the prevalence of a dichotomy in emotion psychology between the theoretical positions of universalism and relativism (e.g., Manstead & Fischer, 2002; Matsumoto, 2001). Universalism sees emotions as products of phylogenetic development that have arisen as specialized, adaptive programs in the human species (Ekman, 1992; Tooby & Cosmides, 1990). Relativism sees emotions as social and cultural constructions that are anything but natural (Lutz, 1988; Kitayama & Markus, 1994). Although most researchers take an intermediate position, the dichotomy still has a strong impact on the interpretation of cross-cultural data. Researchers tend to emphasize either similarities as evidence that emotions are universal or observed differences as evidence that emotions are culturally constructed (Ellsworth, 1994), even though empirical studies consistently point to the existence of both similarities and differences (see Mesquita & Frijda, 1992; Mesquita, Frijda, & Scherer, 1997; Scherer & Wallbott, 1994). Discussions are often also about the appropriate methodology to study emotions across cultures (see Breugelmans et al., 2005; Philippot & Rime´, 1997; Scherer, Wallbott, & Summerfield, 1986). The status of cultural differences in the emotion lexicon as an indicator of differences in emotional experiences is an important issue in such discussions (Bedford & Hwang, 2003).
When people think or talk about emotion experiences, they often use different words to distinguish one experience from another. However, when communicating with people from other cultures, they may discover that emotion terms do not translate directly across languages (for a review, see Russell, 1991). People report experiencing emotions that do not translate well into English, like the Ifaluk emotions of fago and song (Lutz, 1988), or the Javanese emotion sungkan (Geertz, 1959), or use a single emotion term for two emotions that are distinct in the English language (e.g., vergu¨enza for both shame and embarrassment in Spanish (Iglesias, 1996). A recurring question in crosscultural research is to what extent such differences in emotion words indicate differences in emotion experiences. This question is the focus of the present article. Seger M. Breugelmans, Faculty of Social and Behavioural Sciences, Tilburg University, Tilburg, The Netherlands; Ype H. Poortinga, Faculty of Social and Behavioural Sciences, Tilburg University and Department of Psychology, Catholic University of Leuven, Leuven, Belgium. We would like to thank Richard Robins as well as Fons van de Vijver and Johnny Fontaine for their valuable comments on previous versions of this article. We would also like to thank Jesu´s Vaca, Benito Martı´nez, Imelda Gameros, Abril Olmos, German Vergara, Alejandra Lambarri, and Roxana Espinoza for their assistance in the study with the Rara´muri. We thank Rosario Valde´z and the Escuela Libre de Psicologı´a, Chihuahua, and the Escuela Nacional de Antropologı´a e Historia, Chihuahua, for their support in our Mexican studies. In addition, we would like to thank Priyo Widiyanto, Johana Hadiyono, and the University of Sanata Dharma in Yogyakarta for their help in our Indonesian and Javanese studies. This article benefited from the stay of Ype H. Poortinga as a visiting scholar in the Department of Psychology at Boston College. Correspondence concerning this article should be addressed to Seger M. Breugelmans, Faculty of Social and Behavioural Sciences, Tilburg University, P.O. Box 90153, Tilburg 5000 LE, The Netherlands. E-mail:
[email protected] Journal of Personality and Social Psychology, 2006, Vol. 91, No. 6, 1111–1122 Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1111
1111
1112
BREUGELMANS AND POORTINGA
Scholars with a relativist orientation (e.g., Lutz, 1988; Shweder & Haidt, 2000; Wierzbicka, 1998) tend to ascribe more importance to cross-cultural differences in the emotion lexicon than do those with a universalist orientation. Few researchers contend that each emotion word necessarily refers to a distinct emotion process (see Sabini & Silver, 2005), but the categorization of emotions is often assumed to influence emotional experience. For example, Barrett (2006) recently proposed a categorization view of emotions, stating that “there is cultural variation in the experience of emotion that is intrinsically driven by cultural differences in emotion categories and concepts” (p. 39). In a similar vein, Wierzbicka (1999) stated that “whether or not two feelings are interpreted as two different instances of, essentially, ‘the same emotion’ or as instances of ‘two different emotions’ depends largely on the language through the prism of which these feelings are interpreted; and that prism depends on culture” (p. 26). Scholars with a universalist orientation tend to assume that emotion processes are cross-culturally similar even if lexicons differ (e.g., Ekman, 1994; Scherer & Wallbott, 1994). However, in order to test this assumption, one should measure emotions by indicators other than emotion words. Facial expressions can only be used for a limited set of basic emotions (Ekman, 1992), leaving most of the social emotions, such as shame and guilt, outside the scope of comparative research. Another approach is to measure emotions not by a single indicator but by a range of characteristics on various emotion components, such as cognitive appraisals (Roseman, Antoniou, & Jose, 1996; Scherer, 1997), body sensations (Breugelmans et al., 2005; Scherer & Wallbott, 1994), and action tendencies (Frijda, Kuipers, & ter Schure, 1989). An approach in which multiple indicators are used to study emotions across cultures has the additional advantage of being less vulnerable to cultural item bias than single-item measurements (Fontaine, Poortinga, Setiadi, & Markam, 2002). Thus, a componential approach should allow for the cross-cultural comparison of emotion processes even if the emotion lexicon differs (Frijda, Markam, Sato, & Wiers, 1995).
Shame and Guilt Among the Rara´muri and the Javanese In a previous study (Breugelmans et al., 2005), we found no adequate translation for the English emotion term guilt in the language of the Rara´muri Indians from northern Mexico, whereas for other emotions (anger, disgust, fear, joy, sadness, shame), corresponding terms could be identified. Shame (riwe´rama) was the emotion word typically reported by various Rara´muri informants in response to guilt-eliciting situations of varying intensity. This finding was corroborated in consultations with Rara´muri bilinguals, anthropologists (Escuela Nacional de Antropologı´a e Historia, Chihuahua, Mexico) and psychologists (Escuela Libre de Psicologı´a, Chihuahua, Mexico) with extensive experience in working with the Rara´muri.1 We also examined the ethnographic literature on the Rara´muri and found some descriptions of events (e.g., confessing a theft; Bennett & Zingg, 1935/1976) and concerns (e.g., “doing right,” not stealing, harming others; Heras Quezada, 2000; Merrill, 1988) related to guilt in a Western context. However, no references to feelings of guilt were found. The absence of a word for the emotion of guilt among the Rara´muri allowed us to test the expectations of universalism and relativism with regard to the effects of emotion lexicon on emo-
tional experiences. In addition, it allowed us to examine the generalizability of emotion processes to a rural, non-Western population. Because the Rara´muri tend to use a word for shame (riwe´rama) in situations in which we would expect to elicit guilt, we contrasted the experience of these two emotions. Given the accumulating body of evidence that shame and guilt represent distinct emotional experiences in Western societies (e.g., Keltner & Buswell, 1996; Tangney & Dearing, 2002; Tangney, Miller, Flicker, & Barlow, 1996), the absence of a word for guilt among the Rara´muri allowed for a rather strong test of the possibility that guilt should be indistinguishable from shame. Previous studies suggest that there can be differences in the intensity and frequency of shame and guilt experiences (e.g., Ha, 1995), their distinctness (e.g., Marsella, Murray, & Golden, 1974; Wallbott & Scherer, 1995), and the type of situations in which they occur (e.g., Bedford & Hwang, 2003; Stipek, 1998). However, other studies suggest that the phenomenological characteristics of shame and guilt are rather similar across cultures (e.g., De Rivera, 1989; Fontaine et al., 2006; Hong & Chiu, 1992; Scherer & Wallbott, 1994). It should be noted that all previous studies were conducted in cultures in which local words for shame and guilt were available, providing only a weak test of the possibility that these emotions are indistinct in some cultures. To narrow the gap between universalist and relativist positions, researchers should specify more precisely the extent to which emotions should be the same across cultures and the extent to which differences are to be anticipated (Berry, Poortinga, Segall, & Dasen, 2002; Poortinga & Soudijn, 2002). Differences can be described at various levels of cross-cultural equivalence (see Fontaine, 2004). Currently the most cited are those described by Van de Vijver and Leung (1997), who have distinguished three levels of equivalence. The most basic level, that of construct equivalence, implies that the same psychological construct is measured across cultures, although not necessarily on the same quantitative scale. Construct equivalence is generally investigated using structural analyses (e.g., factor analysis, multidimensional scaling). In cases in which data satisfy higher levels of equivalence (i.e., metric equivalence and full-score equivalence), quantitative comparisons can be made, such as comparisons of mean levels of intensity (using analysis techniques such as analysis of variance [ANOVA]). For the present article, we treated the possibility that the Rara´muri do not distinguish guilt from shame as a question of 1
We would like to thank Benito Martı´nez, Jesu´s Vaca-Corte´z, Francoise Brouzes, William Merrill, and Margot Heras-Quezada for their kind help on this issue. A Spanish–Rara´muri dictionary (Brambila, 1983) did provide two possible translations of guilt: iyiri and chokira. However, neither of these concepts refers to a feeling of guilt. Iyiri refers to guilt only in the legal sense of being responsible, not of feeling responsible. The meaning of chokira is more complex; it refers to the initial cause or the origin of objects (e.g., the root of a tree) or social events (e.g., the cause of a conflict). A preliminary field study with 30 Rara´muri in the Guachochi area revealed that most did not see these words as emotion words. Iyiri was generally used to describe “objective” states of being responsible, such as burning food or making mistakes. When asked about the emotion that they experienced in such situations, most respondents (⬎75%) answered riwe´rama (or a related term; see Footnote 2), which can be translated as shame (Breugelmans et al., 2005). Other emotions that were mentioned were sadness and fear.
EMOTION WITHOUT A WORD
construct equivalence. This means that we expected the Rara´muri to distinguish characteristics of guilt from those of shame if both emotions are present in their culture. Three methodological issues had to be dealt with to test this expectation. First, reviews (Russell, 1994) and meta-analyses (Van Hemert, Poortinga, & Van de Vijver, 2005) have suggested that more cross-cultural differences are found when non-Western, nonstudent samples are studied. This finding means that failure to replicate the distinction between guilt and shame with the Rara´muri could be due to the absence of an emotion word but also to Western bias in the emotion components. Thus, we included a second non-Western population in the study, namely rural Javanese, who were culturally distant from both Western student samples and the Rara´muri. Javanese culture puts a strong emphasis on shame in the regulation of interpersonal relations, and words for both shame (isin) and guilt (salah) are readily available (Geertz, 1959; Keeler, 1987; Magnis-Suseno, 1997). The inclusion of Javanese could help us to disentangle cross-cultural differences resulting from Western biases and from the Rara´muri’s lack of a word for guilt. The second issue was the selection of indicators for both emotions. A substantial body of literature describes characteristics from various emotion components that should distinguish experiences of shame from experiences of guilt (e.g., Barrett, 1995; Frijda, 1993; Frijda et al., 1989; Gilbert, Pehl, & Allan, 1994; Lewis, 1971; Manstead & Tetlock, 1989; Roseman et al., 1996; Roseman, Wiest, & Swartz, 1994; Scherer & Wallbott, 1994; Tangney et al., 1996; Wicker, Payne, & Morgan, 1983). However, most studies have been conducted with Western samples. Recently, Fontaine et al. (2006) reported strong evidence for construct equivalence with a large set of guilt and shame characteristics among samples in Belgium, Hungary, and Peru. They found that characteristics could be adequately represented in a twodimensional structure, defined by a primary guilt–shame dimension and a secondary intrapersonal–interpersonal dimension. This finding was replicated by Breugelmans et al. (2006) among samples in Belgium, Indonesia, Mexico, and the Netherlands. The characteristics in these studies have proven to be applicable across a range of cultures, so they should also apply to the Rara´muri if a distinction between guilt and shame exists with this group. As a last issue, we adopted two design features to avoid imposition of Western conceptions of shame and guilt on the two rural samples. First, we only used locally gathered situational descriptions as emically derived (Berry et al., 2002) stimuli. Thus, our stimuli should have represented ecologically valid situations. Second, the feasibility of a componential approach depends on whether emotion characteristics can be formulated in a nonethnocentric manner (see Goddard, 1997; Wierzbicka, 1995). Therefore, we adapted a research method used by Fontaine et al. (2002, 2006), in which the structure of emotion characteristics is first examined within cultures and only thereafter compared across cultures. Thus, we first examined the relations among characteristics within both rural samples and then compared this structure with an international student sample derived from Breugelmans et al. (2006). We expected that guilt characteristics would form a distinct cluster from shame characteristics in both rural samples, in a similar manner as was found with the international student sample if both emotions are generalizable across cultures.
1113
To summarize, we analyzed ratings of emotion characteristics indicative of guilt and shame by the Rara´muri and the Javanese in response to locally derived situations, using the ratings of the same characteristics by an international student sample as a comparison standard. We expected to find construct equivalence of students and Javanese and of students and Rara´muri for the various characteristics, even in the absence of an emotion word for guilt among the latter group.
Preparatory Studies Selection of the Stimuli In two preparatory field studies, 170 situation descriptions were solicited among Javanese and Rara´muri. In both groups, descriptions were collected in open-ended interviews by local interviewers. They asked participants to describe a situation in which they had experienced shame (both Rara´muri and Javanese) or guilt (Javanese only). Rara´muri. From 54 Rara´muri (20 women, 34 men), 68 situations involving riwe´rama2 (shame) were collected by three experienced interviewers who spoke Spanish as well as Rara´muri. They recorded verbatim descriptions of the situations. The situations were translated into Spanish by the interviewers and translated into English by two independent Mexican translators. Javanese. From 63 Javanese (31 women, 32 men), 55 situations of isin/lingsem (shame), 39 situations of salah/lepat3 (guilt), and 8 situations in which the participants reported a mixture of both shame and guilt were collected. The data were gathered by four Javanese interviewers who spoke Javanese as their native language and who studied English at Sanata Dharma University in Yogyakarta, Java, Indonesia. Interviewers translated the descriptions directly into English. In the translations, emphasis was placed on rendering the events (i.e., what happened) in the situations as accurately as possible. Specific thoughts and feelings that the participants reported were recorded separately but were not included in the situation descriptions because thoughts and feelings were also the dependent variables in the main studies (Studies 1 and 2). In addition, descriptions were culturally decentered (Van de Vijver & Leung, 1997); names of places or specific animals were replaced by generic terms, and any sentences containing a reference to shame, guilt, or closely related terms were deleted or, if deletion would have disturbed the coherence of the description, were replaced by a neutral substitute (e.g., upset). 2 Several translation variants can be given because Rara´muri is not a written language, and no consensual orthography exists. Words may be pronounced slightly differently in the various Rara´muri variants. Hence, the translations given in the text are only examples of several variants that were recorded, all with a similar root (e.g., riwe or rigue in shame). In the field, interviewers adapted the emotion words to the variant of Rara´muri spoken by the interviewee. 3 Javanese language has different forms, depending on the relative social status of the speakers (Keeler, 1987). The polite equivalent of isin is lingsem, which is used in cases in which the relative status of the speaker is lower than that of the person addressed. According to Koentjaraningrat (1985), both terms signal a position of inferiority in social relationships, but lingsem is a slightly stronger marker of inferiority. The polite equivalent of salah is lepat.
1114
BREUGELMANS AND POORTINGA
Eighty Dutch students (61 women, 18 men, 1 person for whom gender information is missing) from Tilburg University and 74 Indonesian students (52 women, 22 men) from Sanata Dharma University in Yogyakarta rated the decentered situations on the extent to which they would elicit anger, shame, sadness, guilt, and fear, using a 6-point rating scale ranging from 0 (I would not experience this emotion at all) to 5 (I would experience this emotion very strongly). The nontarget emotion terms (anger, sadness, and fear) were included so that situations with a different emotional focus could be identified as such. Procrustes rotation of the unifactorial factor solutions for the ratings of shame and of guilt showed structural equivalence (Tucker’s phi ⬎ .90; Van de Vijver & Leung, 1997) between Indonesian and Dutch raters. Five Javanese situations and two Rara´muri situations (4.18% in total) showed differences between the two sets of raters and were excluded from further analyses. In both groups of raters, difference scores (diff )4 were calculated such that each situation had a single score, indicating a higher shame rating if positive and a higher guilt rating if negative. Intraclass correlation (absolute agreement) of difference scores was very high (ICC ⫽ .88) between Indonesian and Dutch raters, so the average diff across groups was used. The distribution of diff had a median of 0.19 and ranged from ⫺2.05 to 3.14 (absolute values of 0.20 indicate small effects, 0.50 medium effects, and 0.80 large effects; Cohen, 1988). Of the Rara´muri situations, 66% had a positive diff (higher shame than guilt) and 34% had a negative diff (higher guilt than shame). This suggested that situations eliciting Rara´muri riwe´rama encompassed experiences of what would be labeled in English as emotions of guilt as well as shame. In contrast, 82% of the situations that were collected from the Javanese as eliciting shame (isin/lingsem) were also rated higher on this emotion (positive diff), and 74% of the guilt-eliciting (salah/lepat) situations were rated higher on guilt (negative diff). The presence of strong shameeliciting and strong guilt-eliciting situations in both cultures made it possible to use a sample of these situations as emically derived (see Berry et al., 2002) stimuli in the main studies. A set of 18 situations was selected as typically eliciting shame (n ⫽ 6; diff ⬎ 0.80), eliciting guilt (n ⫽ 6; diff ⬍ ⫺0.80), or eliciting both shame and guilt (n ⫽ 6; 兩diff 兩⬍ .20). In this way, we tried to best represent the range of situations that may elicit shame or guilt. Examples of shame situations were stumbling and falling while carrying a bucket of water (Rara´muri) and mispronouncing words during a public speech (Javanese). Examples of shame and guilt situations were inadvertently hitting a visitor with a stone (Rara´muri) and arriving late at a communal task (Javanese). Examples of guilt situations were losing someone else’s cattle because of negligence (Rara´muri) and offending a friend in a discussion (Javanese). All selected situations had average ratings of 3.0 or greater on the target emotions and 2.5 or less on all other emotions. Half the situations in each category originated from the Rara´muri, the other half from the Javanese.
Student Reference Standard Data obtained from Indonesian, Mexican, and combined Flemish–Belgian/Dutch student samples reported in an article by Breugelmans et al. (2006) were reanalyzed for the purpose of creating a reference standard. The original data set contained
ratings of the intensity with which respondents would experience each of 47 emotion characteristics (among which were the respective emotion words for shame and guilt: schaamte and schuld in Belgium and the Netherlands; malu and bersalah in Indonesia; vergu¨enza and culpa in Mexico) in response to various situations, using a 6-point rating scale ranging from 0 (not at all) to 5 (very strongly). Preliminary probing suggested that this number was causing loss of concentration with participants in the two rural samples. Therefore, we conducted a second analysis of the ratings of 27 items (plus the ratings for shame and guilt) that we selected for the studies with the Rara´muri and Javanese. In each sample, a Situations (N ⫽ 15) ⫻ Items (N ⫽ 29) matrix was created, with each cell representing the mean rating of an item in a situation. In each matrix, bivariate correlations were calculated between all items. The ensuing correlation matrices were represented in a two-dimensional space with multidimensional scaling (MDS; Borg & Groenen, 1997) using PROXSCAL in SPSS Version 11.5 (SPSS, Chicago). In each sample, a two-dimensional representation could account for almost all of the dispersion (normalized raw stress ⫽ .01, Tucker’s coefficient of congruence ⬎ .97). These three representations were compared using generalized Procrustes analysis (GPA; Commandeur, 1991, 1996), which can be considered the MDS equivalent of Procrustes rotation for factor solutions (see Van de Vijver & Leung, 1997). GPA yielded a centroid configuration with a good fit (⬎90% of the squared distances explained). The results replicated those of Fontaine et al. (2006), with a first guilt–shame dimension and a second intrapersonal–interpersonal dimension. This centroid was used as a reference standard for the two main studies with the Rara´muri and the Javanese. Characteristics associated with shame were (a) appraisals of being at the center of attention; (b) experiences of the self as confused, powerless and small, and angry with others; (c) bodily sensations of blushing, feeling weak in the limbs, feeling warm, trembling, heart beating faster, and sweating; (d) action tendencies of avoiding the gaze of others, hiding oneself from others, and smiling about what happened; and (e) trying to forget about what happened. Associated with guilt were (a) appraisals of having done damage to someone, being responsible for what happened, experiencing the disapproval of others because of what one has done, harming one’s reputation, having violated a social or moral norm, and deserving to be punished; (b) experiences of oneself as a bad person and anger at oneself; (c) action tendencies of apologizing, changing future behavior, explaining what happened to others, and punishing oneself; and (d) ruminating about what happened.
Study 1: The Javanese Method Participants. The sample consisted of 213 Javanese (107 women, 106 men) with a mean age of 42.14 years (SD ⫽ 15.30). Participants were sampled from several small villages in the central south region of Java, located approximately 40 km from Yogyakarta. In these traditional communities, agriculture is the main source of income. Most participants had had some education in local schools. Participants were arbitrarily divided
2 2 ⫹guilt ⫺ 2* The equation is diff ⫽ (Mshame ⫺ Mguilt)/√(shame r(shame, guilt)*shame*guilt), where ⫽ SD of the rating of the emotion. 4
EMOTION WITHOUT A WORD into six groups for administration of the six versions of the instrument. The number of participants tested per version ranged from 32 to 40. Instrument. The 18 situations were divided over six versions of the instrument, each including one shame-eliciting, one guilt-eliciting, and one shame-plus-guilt-eliciting situation. Three versions were composed of Rara´muri situations, and three were composed of Javanese situations. Each situation description was followed by a list of 29 items (6 appraisals, 6 self-experiences, 7 action tendencies, 6 body sensations, and 2 rumination items, plus the emotion words isin/lingsem [shame] and salah/lepat [guilt]). In interviews, participants were asked to indicate the intensity with which they would experience each of the components for each situation, using a 6-point rating scale that ranged from 0 (not at all) to 5 (very strongly). The scale was visually illustrated by a sheet showing a series of circles of increasing size, representing the different intensities of the responses. A committee of four professional translators who were native speakers of Javanese translated the situations from English into Javanese (see Van de Vijver & Leung, 1997). Translations were made both in formal Javanese and in colloquial Javanese, so that interviewers could adapt their word usage to the form appropriate for an interview. Procedure. Approval of local community leaders was sought prior to data collection. Interviewers asked participants whether they would be willing to cooperate in a study on thoughts and feelings that people could have in various situations. Participants were told that they were going to be presented with some situations that other people had experienced and asked to imagine how they themselves would have felt in each of these situations. The interviewer explained the response scale with the aid of the illustration sheet and several examples. Generally, participants understood the task without problems (⬍2% of the interviews were terminated because of lack of understanding). Women were interviewed by female interviewers and men by male interviewers. Four female and two male Javanese conducted the interviews. They had had previous experience with interviewing and had received specific training for this study. During data collection, interviewers were not aware of the research questions, including the focus of the study on shame and guilt. Interviews typically took between 30 and 50 min to complete.
1115
Results The six versions of the questionnaire were combined in a Situations (N ⫽ 18) ⫻ Items (N ⫽ 29) matrix, with each cell representing the mean rating of an item in a situation, calculated across participants. Bivariate correlations were calculated between all 29 items across the 18 situations. The resulting correlation matrix was analyzed with MDS (PROXSCAL in SPSS Version 11.5). A representation in one dimension accounted for .92 of the dispersion (normalized raw stress ⫽ .08, Tucker’s coefficient of congruence ⫽ .96), and a two-dimensional representation for .98 (normalized raw stress ⫽ .02, Tucker’s coefficient of congruence ⫽ .99). We used GPA (Commandeur, 1996) to compare the Javanese representation with the student reference. The resulting centroid configuration could account for 71% of the squared distances (structure fit ⬍ .90) between Javanese and students. Inspection of fit at the item level showed that in the case of 7 items, the centroid configuration could account for less than 50% of squared distances. GPA on the remaining 22 items accounted for 83% of the squared distances (structure fit ⬎ .90) in the Javanese and student representations. Figure 1 shows the positions of these 22 items in the two-dimensional centroid configuration. For all items, exact positions and percentage of squared distances that were accounted for can be found in Table A1 in the Appendix. The position of the 7 poorly fitting items within this configuration could be determined on the basis of their distances from the 22 centroid items. Exact positions and absolute distances are given in Table A2 in the Appendix. Four of these items mainly differed on the shame– guilt dimension. For example, the item “changing future behavior” was related more to guilt in the students (⫺0.15) but more to shame in the Javanese (0.07). The items “feeling confused,” “feeling powerless and small,” and “deserving punishment” were related more to shame in the students but somewhat more to guilt in the Javanese. Other items showed relatively small
.30 angry with others
Intrapersonal - Interpersonal
.20
explaining
sweating center of attention feeling hot trying to forget
.10 apologizing done damage .00 guilt being responsible moral norm punishing self -.10 others disapprove bad person angry with self ruminating -.20 harms reputation
-.30 -.40
Figure 1.
-.30
-.20
-.10
smiling hiding evading looks blushing shame
.00
.10
Guilt - Shame
.20
.30
.40
Centroid configuration of 22 emotion components for the Javanese and the students.
BREUGELMANS AND POORTINGA
1116
differences (⬍.10) on this dimension and mainly differed on the interpersonal–intrapersonal dimension.
Discussion
people for governmental organizations, and all had received training for this study. Interviewers were not aware of the research questions at the time of the study.
Results
We expected that emotion characteristics associated with shame and guilt in the student reference group would be similarly associated with the two emotions in the Javanese. For a substantial number of items, this expectation was confirmed, suggesting that characteristics of shame and guilt with student samples generalize to an important extent to the non-Western, rural Javanese sample. However, 7 (24%) of the 29 items did not meet criteria for construct equivalence. This lack of equivalence could have several reasons. It may have been caused by bias (e.g., poor translation into Javanese, social desirability, or interviewer effects) but also by culture specificity in reactions. On the basis of the present study, no definite conclusions can be drawn regarding the nature of these differences. For 3 of the 7 items, differences were mainly found on the intrapersonal–interpersonal dimension. However, this dimension contributed relatively little to the total dispersion in the Javanese sample. The other 4 items (i.e., “changing future behavior,” “feeling confused,” “feeling powerless and small,” and “deserving punishment”) were potentially more interesting because they may reflect characteristics for which the association with shame or guilt is not universally present. Hence, these items would merit further attention in a study of culture-specific aspects of shame and guilt on Java. The findings among the Javanese gave an indication of the results that could be expected with the Rara´muri. Given that 24% of the items were found to function differently between the rural Javanese and the students, we expected a similar percentage to differ between the Rara´muri and the students (Study 2), unless the absence of a word for guilt has a clear bearing on how emotional situations are experienced.
Study 2: The Rara´muri Method Participants. Two hundred and twenty-nine Rara´muri (121 women, 108 men) with a mean age of 40.68 years (SD ⫽ 15.33) participated in this study. The Rara´muri resided in small communities located within 30 km of the town of Guachochi in the Mexican state of Chihuahua. Traditionally, separate families live dispersed over the available land, practicing smallscale agriculture of crops such as maize and beans. Contact with Spanishspeaking Mexicans is generally very limited as is access to the Western media. Participants were arbitrarily divided into six groups for administration of the six versions of the instrument. The number of participants per version ranged from 36 to 41. Instrument. The instrument was identical to Study 1 on Java, with the exception that an emotion word for guilt was not included, leaving 28 items. Situations were translated from English into Spanish and were checked by two independent Mexican translators. Translations from Spanish to Rara´muri were done independently by two Rara´muri bilinguals, and any differences were subsequently discussed (see Van de Vijver & Leung, 1997). Variations in words were provided to cover (minor) variations in the Rara´muri language (see also Footnote 3). Procedure. The procedure followed was similar to Study 1. Three Rara´muri women and four Rara´muri men conducted the interviews. Five of the interviewers had had previous experience in interviewing Rara´muri
The six versions of the questionnaire were combined in a Situation (N ⫽ 18) ⫻ Item (N ⫽ 28) matrix, with each cell representing the mean rating of an item in a situation, calculated across participants. Bivariate correlations were calculated between the 28 emotion components across the 18 situations. The resulting correlation matrix was analyzed with MDS (PROXSCAL in SPSS Version 11.5). A representation using one dimension accounted for .81 of the dispersion (normalized raw stress ⫽ .19, Tucker’s coefficient of congruence ⫽ .90), and a two-dimensional representation accounted for .95 of the dispersion (normalized raw stress ⫽ .05, Tucker’s coefficient of congruence ⫽ .98). We compared the Rara´muri representation with the student reference using GPA (Commandeur, 1996). The initial centroid configuration accounted for 61% of the squared distances (structure fit ⬍ .90) between the Rara´muri and the students. Inspection of fit at the item level showed that in the case of 10 items, the centroid configuration could account for less than 50% of squared distances. Subsequent GPA on the 18 well-fitting items accounted for 86% of the squared distances (structure fit ⬎ .90) in the Rara´muri and student representations. Figure 2 shows the positions of these items within the two-dimensional centroid configuration. Again, the same two dimensions, guilt–shame and interpersonal–intrapersonal, were identified in the joint configuration. The position of the item guilt as determined in the student reference group (“student guilt”) has been included in Figure 2. For all 18 items, exact positions and the percentage of squared distances that were accounted for can be found in Table A3 in the Appendix. Of the 10 items with a poorer fit, 5 had a distance of less than .10 on the shame– guilt dimension (i.e., “feeling powerless and small,” “blushing,” “changing future behavior,” “evading looks,” “trying to forget”). Exact positions and absolute distances can be found in Table A4 in the Appendix. Because different outcomes were possible for the Javanese and the Rara´muri, these samples were compared separately with the student reference standard. However, a comparison of all three samples did not notably alter the results. Comparisons between the Rara´muri and Javanese, as well as among the Rara´muri, Javanese, and students, resulted in cross-culturally equivalent configurations (fit ⬎ .90) for 18 items.
Discussion We expected similar patterns of interrelationships among items in the case of both the Rara´muri and the students, not only for shame but also for guilt. These expectations were confirmed for most emotion components (64%). Figure 2 illustrates the distinction between shame characteristics and guilt characteristics that was found irrespective of cultural differences in the emotion lexicon. Of the 28 emotion components, 10 (35%) differed between the students and the Rara´muri, which was more than between the students and the Javanese. Six of these items showed mainly a difference on the intrapersonal–interpersonal dimension; four of
EMOTION WITHOUT A WORD
1117
.30
Intrapersonal - Interpersonal
.20
smiling
apologizing
.10
sweating shame center of attention hiding
“student guilt” done damage .00 bad person moral norm others disapprove -.10 being responsible punishing self harms reputation deserving punishment confused -.20
-.30 -.40
Figure 2.
explaining
-.30
-.20
-.10
feeling hot weak limbs
.00
.10
Guilt - Shame
.20
.30
.40
Centroid configuration of 18 emotion components for the Rara´muri and the students (plus “student guilt”).
these had very similar positions on the guilt–shame dimension. The five items that differed more than .10 on the guilt–shame dimension hold the most promise for further analysis of culturespecificity in shame and guilt characteristics. It may be noted that two items (i.e., feeling powerless and small, and changing future behavior) showed corresponding differences for the Rara´muri and the Javanese, making it less likely that these results can be ascribed to some source of bias. Another noteworthy difference is the association of blushing with guilt components by the Rara´muri because this finding is in line with those of other studies, suggesting that the association of this characteristic with shame may not be universally shared (e.g., Casimir & Schnegg, 2002; Drummond & Lim, 2000).
General Discussion Emotion words are central to most current research of emotion processes, either as stimuli or as dependent variables, but to what extent can differences in emotion processes be inferred from differences in words? We examined this question in a study with the Rara´muri Indians, who do not have a word for the emotion of guilt. In our view, the results in Figure 2 indicate that the Rara´muri did distinguish between two clusters of emotion characteristics of guilt and shame. This finding suggests that differences in the emotion lexicon (see Russell, 1991) cannot be taken as evidence that emotion processes, as identified in terms of associated emotion characteristics, are also different. Approximately one third (36%) of emotion characteristics did not replicate with the Rara´muri, but it appears unlikely that this is caused by the absence of a word for guilt. Twenty-four percent of emotion characteristics did also not replicate with the rural Javanese, who do have emotion words for both guilt and shame. This finding means that the generalizability of guilt and shame characteristics was less clear with the two rural, nonstudent groups when
compared with the international student sample (Breugelmans et al., 2006). Cross-cultural differences in separate characteristics are difficult to interpret unless there is some patterning because differences could also be caused by item bias (Van de Vijver & Leung, 1997). Of most interest for further study are two items for which similar differences were found in Studies 1 and 2. In the two rural groups, the item of feeling powerless and small was associated more strongly with guilt characteristics, and the item of changing future behavior was associated with shame characteristics. This finding suggests that guilt may be related to negative self-affect and that shame may be related to constructive social behavior in non-Western groups, in contrast to what has been argued for these emotions in a Western context (e.g., Tangney & Dearing, 2002). In addition, blushing was not associated strongly with shame in the Rara´muri, a finding that is in line with other evidence that the association between blushing and shame is not so strong as has been assumed on the basis of Western studies (see Casimir & Schnegg, 2002; Drummond & Lim, 2000). The evidence for construct equivalence that we found imposes constraints on the extent to which guilt and shame can be conceived of as different emotional experiences across cultures. Strong relativist views that posit fundamental differences in emotions are less plausible in the light of our data. However, our data do not constrain cross-cultural differences in the intensity or salience of guilt and shame experiences (see Creighton, 1990) because we only addressed construct equivalence. The cross-cultural differences that we found with the rural samples also imply that universality of characteristics of guilt and shame cannot be assumed on the basis of student studies alone. Although patterns of emotion characteristics are very likely to generalize across cultural populations, there can be cross-cultural differences in individual characteristics (see Breugelmans et al., 2005; Fontaine et al., 2002; Matsumoto, Nezlek, & Koopmann, in press).
1118
BREUGELMANS AND POORTINGA
Our findings best fit the notion of modal emotions as described by Mesquita et al. (1997). This theory posits that there are not a fixed number of basic emotions but rather that there may be cross-cultural consistencies in the frequencies with which various emotion characteristics co-occur. In this view, our data are not about whether the Rara´muri do or do not experience a categorical emotion of guilt but rather about whether the emotion characteristics that we associate with a category of guilt co-occur similarly in the Rara´muri, distinct from characteristics of shame. In our view, replication of a guilt–shame distinction with the two rural samples supports a position of psychological universalism with regard to these modal emotions (see Berry et al., 2002; Poortinga & Soudijn, 2002). Psychological universalism posits that basic processes are shared across cultures but that there may be cultural differences in the manifestations of these processes. We sampled emotion characteristics for which plausible evidence of cross-cultural similarity had been found in two recent studies (i.e., Breugelmans et al., 2006; Fontaine et al., 2006). The results with the Javanese and the Rara´muri (Figures 1 and 2) largely replicate the findings of these studies, with a clear first dimension distinguishing guilt-related from shame-related characteristics and a second dimension distinguishing interpersonal from intrapersonal characteristics. It should be noted that the first dimension, which was the main focus of our studies, explained most of the variance in both studies. Many characteristics distinguishing shame and guilt in our studies are compatible with previous results found for these emotions (e.g., Frijda et al., 1989; Gilbert et al., 1994; Roseman et al., 1994; Scherer & Wallbott, 1994; Wicker et al., 1983). However, one important exception concerns the theory put forward by Lewis (1971) and Tangney (1996) that distinguishes shame and guilt in terms of a focus on the global self or on specific behavior. Our findings that guilt—not shame—was associated with negative evaluations of the self (e.g., experiences of the self as a bad person) in both rural samples go against this theory. There can be various explanations for these contrasting findings. First, it is possible that the self– behavior distinction is less relevant for distinguishing shame and guilt in rural, non-Western samples. Another explanation may be that the self– behavior distinction is mainly about the type of attributions that give rise to shame or guilt (see Tracy & Robins, 2004), whereas emotional experiences evoked by a particular situation were the focus of our interviews. As Tracy and Robins (in press) showed, feelings of guilt tend to be elicited by internal, unstable, and controllable attributions (in contrast to internal, stable attributions for shame). However, this does not preclude negative evaluations of oneself (e.g., feeling like a bad person and being angry at or disappointed with oneself) during guilt experiences as assessed in our studies. A third explanation may be that the self– behavior distinction is primarily based on differences among people (i.e., proneness to shame or proneness to guilt), whereas our study analyzed differences across situations. Fontaine et al. (2006) have recently shown that analyses across people yield results that are more compatible with the self– behavior distinction. They also argued that an analysis across situations is most appropriate for studying differences among emotion processes, which was the focus of our studies. There are three possible limitations to our studies. First, we cannot totally exclude effects of cultural diffusion with the Rara´muri and the Javanese. Both groups may have had some exposure
(firsthand or via the media) to Western notions of shame and guilt. However, it appears very unlikely that transfer of such fairly subtle emotion distinctions could have substantially influenced our results. Further, the Rara´muri do have concepts of guilt in a causal or judicial sense, although these do not refer to a feeling or emotion (see Footnote 1). The absence of culture contacts in our studies was not as strong as in the work by Ekman and Friesen (1971) in Papua New Guinea. However, we contend that the present research provides a stronger test for the cross-cultural validity of a distinction between shame and guilt than previous studies have provided with student samples. A second limitation is that the structure of emotion characteristics (Figures 1 and 2) follows from the choice to focus on shame and guilt. In our view, a comparison of these two emotions was the strongest test for the possibility that the Rara´muri did not experience guilt. If we had contrasted guilt with a more distant emotion such as anger or fear, the clustering of characteristics would probably have been different. However, we would argue that this would not be so only for the Rara´muri but also for the other samples. Finally, some of our emotion characteristics, such as smiling, have been argued to be more characteristic of embarrassment than of shame in an American context (Keltner & Buswell, 1996). None of the populations in our studies made a clear linguistic distinction between embarrassment and shame similar to the distinction made in the English language; this circumstance may have led to some confound of shame and embarrassment characteristics. However, most other characteristics in the shame cluster, like a tendency to hide and to evade the looks of others, are also central to shame in the (Western) emotion literature. As a final point, there is no reason to assume that the emotion words in the English language are most representative of the emotion domain (Russell, 1991). Because the exact meaning of emotion words is very difficult to translate across languages (e.g., Goddard, 1997; Wierzbicka, 1999), an approach using multiple characteristics from various emotion components, rather than categorical emotion labels, appears to be better suited for studying the ways in which cultures are similar or different in emotion processes (Mesquita et al., 1997). Our studies suggest how this method could be used to research emotions for which there are no words.
References Barrett, K. C. (1995). A functionalist approach to shame and guilt. In J. P. Tangney & K. W. Fischer (Eds.), Self-conscious emotions: The psychology of shame, guilt, embarrassment, and pride (pp. 25– 63). New York: Guilford. Barrett, K. C. (2006). Solving the emotion paradox: Categorizations and the experience of emotion. Personality and Social Psychology Bulletin, 10, 20 – 46. Bedford, O., & Hwang, K. K. (2003). Guilt and shame in Chinese culture: A cross-cultural framework from the perspective of morality and identity. Journal for the Theory of Social Behaviour, 33, 127–144. Bennett, W. C., & Zingg, R. M. (1976). The Tarahumara, an Indian tribe of northern Mexico. Glorietta, NM: Rio Grande Press. (Original work published 1935) Berry, J. W., Poortinga, Y. H., Segall, M. H., & Dasen, P. R. (2002). Cross-cultural psychology: Research and applications (2nd ed.). Cambridge, England: Cambridge University Press. Borg, I., & Groenen, P. (1997). Modern multidimensional scaling: Theory and applications. New York: Springer.
EMOTION WITHOUT A WORD Brambila, D. S. J. (1983). Diccionario Castellano-Rara´muri (Tarahumar). Mexico City, Mexico: Obra Nacional de la Buena Prensa. Breugelmans, S. M., Poortinga, Y. H., Ambadar, Z., Setiadi, B., Vaca, B., Widiyanto, P., & Philippot, P. (2005). Body sensations associated with emotions in Rara´muri Indians, rural Javanese, and three student samples. Emotion, 5, 166 –174. Breugelmans, S. M., Poortinga, Y. H., Fontaine, J. R. J., Espinoza, R., Luyten, P., & Widiyanto, P. (2006). What’s in a name? A multi-method study of shame and guilt in four countries. Manuscript submitted for publication. Casimir, M. J., & Schnegg, M. (2002). Shame across cultures: The evolution, ontogeny and function of a “moral emotion.” In H. Keller, Y. H. Poortinga, & A. Scho¨lmerich (Eds.), Between culture and biology: Perspectives on ontogenetic development (pp. 270 –302). Cambridge, England: Cambridge University Press. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Commandeur, J. (1991). Matching configurations. Leiden, The Netherlands: DSWO Press. Commandeur, J. (1996). Generalized Procrustes Analysis for (In)complete Configurations (Version 2.5) [Computer software]. Leiden, The Netherlands: University of Leiden. Creighton, M. R. (1990). Revisiting shame and guilt cultures: A forty-year pilgrimage. Ethos, 18, 279 –307. De Rivera, J. (1989). Comparing experiences across cultures: Shame and guilt in American and Japanese. Hiroshima Forum for Psychology, 14, 13–20. Drummond, P. D., & Lim, H. K. (2000). The significance of blushing for fair- and dark-skinned people. Personality and Individual Differences, 29, 1123–1132. Ekman, P. (1992). Are there basic emotions? Psychological Review, 99, 550 –553. Ekman, P. (1994). Strong evidence for universals in facial expressions: A reply to Russell’s mistaken critique. Psychological Bulletin, 115, 268 – 287. Ekman, P., & Friesen, W. V. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17, 124 – 129. Ellsworth, P. C. (1994). Sense, culture, and sensibility. In S. Kitayama & H. R. Markus (Eds.), Emotion and culture: Empirical studies of mutual influence (pp. 23–50). Washington, DC: American Psychological Association. Fontaine, J. R. J. (2004). Equivalence. In K. Kempf-Leonard (Ed.), Encyclopedia of social measurement (Vol. 1, pp. 803– 813). New York: Academic Press. Fontaine, J. R. J., Luyten, P., de Boeck, P., Corveleyn, J., Fernandez, M., Herrera, D., et al. (2006). Untying the Gordian knot of guilt and shame: The structure of guilt and shame reactions based on situation and person variation in Belgium, Hungary, and Peru. Journal of Cross-Cultural Psychology, 37, 273–292. Fontaine, J. R. J., Poortinga, Y. H., Setiadi, B., & Markam, S. S. (2002). Cognitive structure of emotion terms in Indonesia and The Netherlands. Cognition and Emotion, 16, 61– 86. Frijda, N. H. (1993). The place of appraisal in emotion. Cognition and Emotion, 7, 357–387. Frijda, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness. Journal of Personality and Social Psychology, 57, 212–228. Frijda, N. H., Markam, S. S., Sato, K., & Wiers, R. (1995). Emotions and emotion words. In J. A. Russell, A. J. R. Manstead, J. C. Wellenkamp, & J. M. Fernandez-Dols (Eds.), Everyday conceptions of emotions: An introduction to the psychology, anthropology, and linguistics of emotions (pp. 121–143). Dordrecht, The Netherlands: Kluwer Academic.
1119
Geertz, H. (1959). The vocabulary of emotion: A study of Javanese socialization processes. Psychiatry, 22, 225–237. Gilbert, P., Pehl, J., & Allan, S. (1994). The phenomenology of shame and guilt: An empirical investigation. British Journal of Medical Psychology, 67, 23–36. Goddard, C. (1997). Cultural values and “cultural scripts” of Malay (Bahasa Melayu). Journal of Pragmatics, 27, 183–201. Ha, F. I. (1995). Shame in Asian and Western cultures. American Behavioral Scientist, 38, 1114 –1131. Heras Quezada, M. (2000). Identidad y continuidad Rara´muri: Imagenes y discursos de la gente de Banalachi sobre sı´ misma [Rara´muri identity and continuity: Images and discourse of the people of Banalachi about themselves]. Unpublished bachelor’s thesis, Escuela Nacional de Antropologı´a e Historia, Chihuahua, Mexico. Hong, Y-Y., & Chiu, C.-Y. (1992). A study of the comparative structure of guilt and shame in a Chinese society. Journal of Psychology, 126, 171–179. Iglesias, I. (1996). Vergu¨enza ajena [Empathetic embarrassment]. In R. Harre´ and W. G. Parrott (Eds.), The emotions: Social, cultural and biological dimensions (pp. 122–131). London: Sage. Keeler, W. (1987). Javanese shadow plays, Javanese selves. Princeton, NJ: Princeton University Press. Keltner, D., & Buswell, B. N. (1996). Evidence for the distinctness of embarrassment, shame, and guilt: A study of recalled antecedents and facial expressions of emotion. Cognition and Emotion, 10, 155–171. Kitayama, S., & Markus, H. R. (Eds.). (1994). Emotion and culture: Empirical studies of mutual influence. Washington, DC: American Psychological Association. Koentjaraningrat, R. M. (1985). Javanese culture. Singapore: Oxford University Press. Lewis, H. B. (1971). Shame and guilt in neurosis. New York: International Universities Press. Lutz, C. A. (1988). Unnatural emotions: Everyday sentiments on a Micronesian atoll and their challenge to Western theory. Chicago: University of Chicago Press. Magnis-Suseno, F. (1997). Javanese ethics and world-view: The Javanese idea of the good life. Jakarta, Indonesia: Gramedia Pustaka Utama. Manstead, A. S. R., & Fischer, A. H. (2002). Beyond the universalityspecificity dichotomy. Cognition and Emotion, 16, 1–10. Manstead, A. S. R., & Tetlock, P. E. (1989). Cognitive appraisals and emotional experience: Further evidence. Cognition and Emotion, 3, 225–240. Marsella, A. J., Murray, M. D., & Golden, C. (1974). Ethnic variations in the phenomenology of emotions. Journal of Cross-Cultural Psychology, 5, 312–328. Matsumoto, D. (2001). Culture and emotion. Oxford, England: Oxford University Press. Matsumoto, D., Nezlek, J. B., & Koopmann, B. (in press). Evidence for universality in phenomenological emotion response system coherence. Emotion. Merrill, W. L. (1988). Rara´muri souls: Knowledge and social process in northern Mexico. Washington, DC: Smithsonian Institution Press. Mesquita, B. (2001). Emotions in collectivist and individualist contexts. Journal of Personality and Social Psychology, 80, 68 –74. Mesquita, B., & Frijda, N. H. (1992). Cultural variations in the emotions: A review. Psychological Bulletin, 112, 179 –204. Mesquita, B., Frijda, N. H., & Scherer, K. R. (1997). Culture and emotion. In J. W. Berry, P. R. Dasen, & T. S. Saraswathi (Eds.), Handbook of cross-cultural psychology: Vol. 2. Basic processes and human development (2nd ed., pp. 255–297). Boston: Allyn & Bacon. Philippot, P., & Rime´, B. (1997). The perception of body sensations during emotion: A cross-cultural perspective. Polish Psychological Bulletin, 28, 175–188. Poortinga, Y. H., & Soudijn, K. (2002). Behaviour– culture relationships
1120
BREUGELMANS AND POORTINGA
and ontogenetic development. In H. Keller, Y. H. Poortinga, & A. Scho¨lmerich (Eds.), Between culture and biology: Perspectives on ontogenetic development (pp. 320 –340). Cambridge, England: Cambridge University Press. Roseman, I. J., Antoniou, A. A., & Jose, P. E. (1996). Appraisal determinants of emotions: Constructing a more accurate and comprehensive theory. Cognition and Emotion, 10, 241–277. Roseman, I. J., Wiest, C., & Swartz, T. S. (1994). Phenomenology, behaviors, and goals differentiate discrete emotions. Journal of Personality and Social Psychology, 67, 206 –221. Russell, J. A. (1991). Culture and the categorization of emotions. Psychological Bulletin, 110, 426 – 450. Russell, J. A. (1994). Is there universal recognition of emotion from facial expression? A review of cross-cultural studies. Psychological Bulletin, 115, 102–141. Sabini, J., & Silver, M. (2005). Why emotion names and experiences don’t neatly pair. Psychological Inquiry, 16, 1–10. Scherer, K. R. (1997). Profiles of emotion-antecedent appraisal: Testing theoretical predictions across cultures. Cognition and Emotion, 11, 113– 150. Scherer, K. R., & Wallbott, H. G. (1994). Evidence for universality and cultural variation of differential emotion response patterning. Journal of Personality and Social Psychology, 66, 310 –328. Scherer, K. R., Wallbott, H. G., & Summerfield, A. B. (Eds.). (1986). Experiencing emotion: A cross-cultural study. Cambridge, England: Cambridge University Press. Shweder, R. A., & Haidt, J. (2000). The cultural psychology of the emotions: Ancient and new. In M. Lewis & M. Haviland-Jones (Eds.), Handbook of emotions (pp. 397– 414). New York: Guilford. Stipek, D. (1998). Differences between Americans and Chinese in the circumstances evoking pride, shame, and guilt. Journal of CrossCultural Psychology, 29, 616 – 629. Tangney, J. P. (1996). Conceptual and methodological issues in the as-
sessment of shame and guilt. Behavior Research and Therapy, 34, 741–754. Tangney, J. P., & Dearing, R. L. (2002). Shame and guilt. New York: Guilford. Tangney, J. P., Miller, R. S., Flicker, L., & Barlow, D. H. (1996). Are shame, guilt, and embarrassment distinct emotions? Journal of Personality and Social Psychology, 70, 1256 –1269. Tooby, J., & Cosmides, L. (1990). The past explains the present: Emotional adaptations and the structure of ancestral environments. Ethology and Sociobiology, 11, 375– 424. Tracy, J. L., & Robins, R. W. (2004). Putting the self in to self-conscious emotions: A theoretical model. Psychological Inquiry, 15, 103–125. Tracy, J. L. & Robins, R. W. (in press). Appraisal antecedents of shame and guilt: Support for a theoretical model. Personality and Social Psychology Bulletin. Van de Vijver, F. J. R., & Leung, K. (1997). Methods and data-analysis for cross-cultural research. Thousand Oaks, CA: Sage. Van Hemert, D. A., Poortinga, Y. H., & Van de Vijver, F. J. R. (2005). Emotion and culture: A meta-analysis. Manuscript submitted for publication. Wallbott, H. G., & Scherer, K. R. (1995). Cultural determinants in experiencing shame and guilt. In J. P. Tangney & K. W. Fischer (Eds.), Self-conscious emotions: The psychology of shame, guilt, embarrassment, and pride (pp. 465– 487). New York: Guilford. Wicker, F. W., Payne, G. C., & Morgan, R. D. (1983). Participant descriptions of guilt and shame. Motivation and Emotion, 7, 25–39. Wierzbicka, A. (1995). Everyday conceptions of emotion: A semantic perspective. In J. A. Russell, A. J. R. Manstead, J. C. Wellenkamp, & J. M. Fernandez-Dols (Eds.), Everyday conceptions of emotions: An introduction to the psychology, anthropology, and linguistics of emotions (pp. 17– 48). Dordrecht, The Netherlands: Kluwer Academic. Wierzbicka, A. (1998). Angst. Culture and Psychology, 4, 161–188. Wierzbicka, A. (1999). Emotions across languages and cultures: Diversity and universals. Cambridge, England: Cambridge University Press.
EMOTION WITHOUT A WORD
1121
Appendix Item Positions and Percentage of Distances Accounted for in Centeroid Configurations and Culturally Different Item Positions and Absolute Item Differences Among the Javanese, the Rara´muri, and the Student Reference Group Table A1 Item Positions and Percentage of Distances Accounted for by the Centroid Configuration for the Students and the Javanese Position Item
S–G dimension
A–E dimension
Distances accounted for (%)
Apologizing Done damage Explaining Guilt Being responsible Moral norm Bad person Others disapprove Punishing self Angry with self Ruminating Harms reputation Sweating Shame Angry with others Center of attention Feeling hot Blushing Hiding Evading looks Trying to forget Smiling
⫺0.252 ⫺0.212 ⫺0.205 ⫺0.203 ⫺0.190 ⫺0.171 ⫺0.153 ⫺0.144 ⫺0.137 ⫺0.110 ⫺0.083 ⫺0.075 0.078 0.101 0.103 0.112 0.125 0.165 0.225 0.229 0.241 0.335
0.048 0.012 0.173 ⫺0.008 ⫺0.024 ⫺0.021 ⫺0.061 ⫺0.034 ⫺0.019 ⫺0.023 ⫺0.026 ⫺0.070 0.037 ⫺0.091 0.210 0.034 0.022 ⫺0.068 ⫺0.011 ⫺0.050 0.033 0.035
98 98 92 98 98 99 96 89 99 98 61 95 93 83 86 71 79 81 97 94 94 93
Note. G–S ⫽ guilt–shame dimension; A–E ⫽ intrapersonal–interpersonal dimension.
Table A2 Positions of Culturally Different Items of the Javanese and the Students and the Absolute Item Differences per Dimension Students Item
G–S
A–E
Change behavior Confused Powerless and small Deserving punishment Weak limbs Heart beats faster Trembling
⫺0.15 0.09 0.09 0.00 0.05 0.12 0.10
⫺0.03 ⫺0.11 ⫺0.03 ⫺0.08 ⫺0.08 ⫺0.08 ⫺0.06
Note.
Absolute difference
Javanese G–S 0.07 ⫺0.08 ⫺0.05 ⫺0.13 ⫺0.02 0.06 0.06
A–E
G–S
A–E
⫺0.02 0.05 ⫺0.11 0.02 0.12 0.15 0.16
0.21 0.17 0.14 0.13 0.08 0.06 0.04
0.01 0.16 0.08 0.10 0.20 0.23 0.23
G–S ⫽ guilt–shame dimension; A–E ⫽ intrapersonal–interpersonal dimension.
(Appendix continues)
BREUGELMANS AND POORTINGA
1122
Table A3 Item Positions and Percentage of Distances Accounted for by the Centroid Configuration for the Students and the Rara´muri Position Item
S–G dimension
A–E dimension
Distances accounted for (%)
Done damage Others disapprove Being responsible Bad person Apologizing Moral norm Punishing self Harms reputation Deserving punishment Explaining Confused Weak limbs Feeling hot Center of attention Sweating Shame Hiding Smiling
⫺0.223 ⫺0.191 ⫺0.182 ⫺0.166 ⫺0.163 ⫺0.153 ⫺0.103 ⫺0.090 ⫺0.082 ⫺0.061 ⫺0.017 0.101 0.110 0.132 0.144 0.144 0.184 0.270
0.020 ⫺0.042 ⫺0.039 0.004 0.118 ⫺0.002 ⫺0.035 ⫺0.040 ⫺0.102 0.180 ⫺0.139 ⫺0.157 ⫺0.096 ⫺0.019 0.026 ⫺0.010 ⫺0.033 0.154
100 93 95 98 85 98 93 95 73 75 70 81 84 78 70 88 98 93
Note. G–S ⫽ guilt–shame dimension; A–E ⫽ intrapersonal–interpersonal dimension.
Table A4 Positions of Culturally Different Items of the Rara´muri and the Students and the Absolute Item Differences per Dimension Students
Rara´muri
Absolute difference
Item
G–S
A–E
G–S
A–E
G–S
A–E
Powerless and small Blushing Change behavior Evading looks Trying to forget Angry at others Trembling Angry at self Ruminating Heart beats faster
0.08 0.21 ⫺0.14 0.18 0.19 0.09 0.09 ⫺0.11 ⫺0.06 0.10
⫺0.05 ⫺0.04 ⫺0.02 ⫺0.03 ⫺0.04 0.10 ⫺0.08 ⫺0.03 ⫺0.10 ⫺0.09
⫺0.21 ⫺0.07 0.12 ⫺0.01 0.05 ⫺0.01 0.12 ⫺0.08 ⫺0.09 0.11
0.15 ⫺0.27 ⫺0.04 ⫺0.10 0.25 ⫺0.16 0.16 0.22 0.11 0.25
0.29 0.29 0.26 0.18 0.14 0.09 0.03 0.03 0.03 0.01
0.19 0.23 0.02 0.07 0.29 0.26 0.23 0.24 0.21 0.34
Note.
G–S ⫽ guilt–shame dimension; A–E ⫽ intrapersonal–interpersonal dimension.
Received May 23, 2005 Revision received May 17, 2006 Accepted June 1, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1123–1137
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1123
Helping One’s Way to the Top: Self-Monitors Achieve Status by Helping Others and Knowing Who Helps Whom Francis J. Flynn
Ray E. Reagans
Columbia University
Carnegie Mellon
Emily T. Amanatullah and Daniel R. Ames Columbia University The authors argue that high self-monitors may be more sensitive to the status implications of social exchange and more effective in managing their exchange relations to elicit conferrals of status than low self-monitors. In a series of studies, they found that high self-monitors were more accurate in perceiving the status dynamics involved both in a set of fictitious exchange relations and in real relationships involving other members of their social group. Further, high self-monitors elevated their social status among their peers by establishing a reputation as a generous exchange partner. Specifically, they were more likely than low self-monitors to be sought out for help and to refrain from asking others for help. This behavior provides one explanation for why high self-monitors acquire elevated status among their peers—they are more attuned to status dynamics in exchange relations and adapt their behavior in ways that elicit status. Keywords: self-monitoring, exchange relations, helping, social status
high self-monitors better comprehend the networks of relationships around them? And how do high self-monitors attain positions of status and influence among their peers—is it partly driven by their ability to establish a generous reputation as an exchange partner? In this article we address these questions, arguing that high self-monitors do, in fact, better understand the networks around them and that they can elicit conferrals of social status by altering their exchange behavior (e.g., by refraining from seeking help). Taken together, these perceptions and behaviors provide a crucial theoretical link, we suggest, between self-monitoring and social status.
Self-monitoring scholars have called for status to occupy a more prominent role in theory and research on self-monitoring (see Gangestad & Snyder, 2000, p. 547). We heed this call by considering the way in which high self-monitors perceive the status dynamics of exchange relations and alter their exchange behavior in ways that elicit status conferrals from their peers. Previous researchers have suggested that high self-monitors have (a) heightened awareness of their situations—they pay more attention to their social environment—and (b) expressive control—they are more responsive to social and interpersonal cues of situational appropriateness (Snyder, 1987). We draw on these two aspects of self-monitoring behavior—social awareness and expressive control—to explain how high self-monitors perceive the relative status of their own and others’ exchange relations and attain elevated positions of status in social groups. Although previous work alludes to how high self-monitors might function in their exchange relations, no research has directly examined how self-monitoring relates to patterns of exchange across relations or to perceptions of status dynamics within relations. Such evidence could help personality scholars understand how self-monitoring comes to life in everyday interactions: Do
Social Status and Social Exchange Dynamics Sorokin (1927) argued that status can appear in many different forms, including economic, political, informational, and social. In the present research, we have focused our attention on social status, which refers to a position of elevated social standing and interpersonal influence (Bourdieu, 1984). Social status is conferred to people on the basis of their apparent possession of attributes (e.g., competence, generosity) held as ideal by other members of their social group (Wegener, 1992). To the extent that a focal individual possesses a unique value or has provided something of unique value to the group, others are willing to be persuaded by that individual and weigh his or her opinions more heavily in their decision-making (e.g., Anderson, John, Keltner, & Kring, 2001). Exchange behavior, particularly the giving and receiving of help, advice, and social support, can operate as a basic source of social status conferrals. Studies have shown that people tend to be held in higher esteem if others perceive them to be more generous—providing more help and advice to others than they receive in return (e.g., Blau, 1963; Flynn, 2003). Helping behavior can also
Francis J. Flynn, Department of Psychology, Columbia University; Ray E. Reagans, Tepper School of Business, Carnegie Mellon University; Emily T. Amanatullah and Daniel R. Ames, Graduate School of Business, Columbia University. We thank Martin Kilduff for his helpful comments on a draft of this article. Correspondence concerning this article should be addressed to Francis J. Flynn, who is now at the Graduate School of Business, Stanford University, Littlefield 281, 518 Memorial Way, Stanford, CA 94305. E-mail:
[email protected] 1123
1124
FLYNN, REAGANS, AMANATULLAH, AND AMES
act as an important source of interpersonal influence (Jones, 1964; Thibaut & Kelley, 1959). For example, if an individual is having difficulty accomplishing a task, she may enlist the support of others who have received favors from her in the past. Assuming these help recipients have not yet reciprocated, they are obligated to provide help in return (Gouldner, 1960). Thus, status dynamics in exchange relations are partly driven by previous acts of generosity—the more generous you have been in the past, the more status and influence you have over your peers in the future. Beginning with Malinowski’s (1922) and Mauss’s (1925) early work on gift giving and exchange behavior, researchers have recognized the trade-off between help seeking and conferrals of status and influence. Those who assume the role of help seeker tend to occupy a lower status position because they “expose themselves to denial and rejection” and acknowledge their dependence on others (Goffman, 1971, p. 114). If people tend to seek help often, more often than they provide it, they risk ruining their reputation as an exchange partner and undermining their status (e.g., Blau, 1963; Lee, 1997). This implies that we can recognize the status dynamics of an exchange relation by examining the pattern of resource sharing—who tends to give help and who tends to receive it. Consider the following example of two coworkers, Steve and Amy. If Steve and Amy seek (or do not seek) assistance from each other, this would indicate an equivalent-status exchange relation. However, if Steve is willing to request assistance from Amy, but Amy is not inclined to request assistance from Steve, this would indicate a high-status exchange relation for Amy and a low-status exchange relation for Steve.1 Given their concern for maintaining a positive public image, high self-monitors may be more sensitive to the status dynamics of exchange behavior in two ways. First, high self-monitors may be more perceptive in recognizing patterns of exchange relations (i.e., who occupies a position of higher status or which actor is relatively more dependent on the other for assistance). Second, high self-monitors may be motivated to seek conferrals of social status by carefully managing their exchange relations. Specifically, they may attempt to maintain a generous pattern of behavior in which they refrain from requesting help from others but are willing to provide help when others approach them (i.e., leading others to view them as giving more than they receive). We explore these two outcomes—perception and behavior—in the next two sections.
Self-Monitoring and Accuracy in Perceiving Exchange Relations Self-monitoring is characterized by an acuteness of perception, discernment, and understanding of social situations (Gangestad & Snyder, 2000). Whereas most people possess a discriminative facility, or an innate “sensitivity to the subtle clues in the situation” (Mischel & Shoda, 1998, p. 246), a high self-monitor’s discriminative facility may be particularly acute (Snyder, 1974, 1987). High self-monitors attend closely to the behavior of others in their immediate environment. They recognize changes in social dynamics and can diagnose differences in behavioral norms from one situation to the next (see, e.g., Costanzo & Archer, 1989; Funder & Harris, 1986; Hosch, Leippe, Marchioni, & Cooper, 1984). This heightened awareness of social and informational cues can assist high self-monitors in accurately identifying social structures—the
makeup of exchange relations that connect members of their social group. Human beings sometimes have difficulty encoding, representing, and inferring others’ social relationships (e.g., Janicik & Larrick, 2005; Rubin & Zajonc, 1969; Zajonc & Burnstein, 1965; see Kenny, Bond, Mohr, & Hom, 1996, for contrary evidence), but the ability to learn relationship patterns is a critical skill that has been linked to important individual resources, including power and reputation (Kilduff & Krackhardt, 1994; Krackhardt, 1990; Krackhardt & Kilduff, 1999). Researchers have found that high selfmonitors are more aware of the thoughts and feelings of others in their social networks (e.g., Ickes, Stinson, Bissonette, & Garcia, 1990). Such perspicacity should help inform high self-monitors of the exchange relations that exist among members of their social network. That is, high self-monitors should have more accurate representations of others’ cognitive networks, enabling them to answer the question “who is friends with whom in this group?” and, more specifically, “who occupies a position of relatively higher status in these exchange relations?” (i.e., who goes to whom for help and advice?).
Self-Monitoring and Exchange Behavior Aside from having greater awareness of social and informational cues, high self-monitors are motivated to act on these cues in ways that cultivate a favorable public image. High self-monitors are like social pragmatists, attempting to impress others in order to win their approval and respect (Gangestad & Snyder, 2000, p. 531). Previous researchers have found that high self-monitors’ need for social status can affect their decision making as consumers (DeBono, 1987). Whereas high self-monitors react more positively to advertisements for products that are associated with prestige (e.g., a luxury car or a fashionable piece of clothing), low self-monitors focus more on quality and reliability (e.g., DeBono & Snyder, 1989; Snyder & Debono, 1985). This need for a positive public appearance also affects high self-monitors’ decisionmaking in choosing romantic partners—they pursue physically attractive romantic partners (e.g., Snyder & Debono, 1985) to enhance their social standing among their peers (e.g., Sigall & Landy, 1973; Snyder & Debono, 1985). Motivated to maintain a positive public image, high selfmonitors may be particularly sensitive to the status dynamics of dyadic exchange relations—appreciating the negative effect that being indebted to others can have on their reputation. Noting this dynamic, high self-monitors may avoid seeking help from others and instead be inclined to provide help when they are asked for it. This prediction runs counter to findings from previous researchers indicating that high self-monitors are less willing to demonstrate commitment to their exchange partners, particularly their romantic partners (e.g., Snyder & Simpson, 1984). Instead, high selfmonitors may be willing to demonstrate higher levels of commitment to their exchange partners by being generous in their exchange relations (i.e., being the target of requests more often than 1 Some psychologists have noted circumstances in which help seeking can provide a status advantage. For example, Jones (1964) proposed that an individual who requests help from a high-status target may successfully ingratiate themselves to that individual in the short-term, thereby elevating his or her own status in the long-term.
HELPING ONE’S WAY TO THE TOP
requesting help), thereby enhancing others’ impressions of them. Thus, maintaining an asymmetric pattern of exchange behavior, in which people perceive them as more rather than less generous, might serve as a means for high self-monitors to acquire the elevated social status they desire.
Summary of Predictions We made several specific predictions. First, we posited that high self-monitors would be more accurate than other participants in perceiving others’ exchange relations. They would not only recognize whether an exchange relation exists between two people but also recognize which of the two occupied a relatively higher status position. Second, high self-monitors may differ from low self-monitors in their exchange behavior. We predicted that high self-monitors would be viewed as having higher status than other participants, in part because of the way in which they demonstrated more generosity. Being more sensitive to the negative status implications of receiving help, high self-monitors would be less likely than low self-monitors to request help from others. On the other hand, high self-monitors would also be more sensitive to the positive status implications of being sought out for help. Therefore, high self-monitors would cultivate a public image of someone who should be sought out for help. Taken together, these predictions imply a final prediction: Perceived generosity would mediate the relationship between self-monitoring and social status.
Plan of Study We tested our predictions in four studies. In Study 1 we examined the proposed link between an individual’s level of selfmonitoring and his or her need for social status. In Study 2 we investigated the relationship between self-monitoring and accuracy in perceiving exchange relations. Specifically, we measured an individual’s ability to learn an unfamiliar set of exchange relations using an interactive computerized exercise. In Study 3, we gathered data on actual exchange behavior and judgments of social status and examined whether high self-monitors are more likely to elicit conferrals of social status and whether others’ impressions of their exchange behavior mediate this relationship. Finally, in a fourth study, we gathered data on people’s perceptions of emergent exchange relations among members of their social group to determine whether self-monitoring led to greater accuracy in judging interpersonal exchange relations and whether high self-monitors were more likely to occupy a high-status position in these relations.
Study 1 Embedded in our theoretical arguments is a strong assumption that self-monitoring is related to a need for social status. Although several references to this idea exist in the self-monitoring literature, we find little direct empirical evidence of this important theoretical link. We decided to test this assumption—that high self-monitors are motivated by a need for social status— directly.
Method Participants One hundred Columbia University undergraduate students participated in this study in exchange for $5.
1125
Procedure We asked participants to complete a brief questionnaire that included eight items designed to capture the need for social status. Sample items included “being a highly valued member of my social group is important to me” and “I enjoy having influence over other people’s decision making” (see the Appendix for the complete list of these items). Each participant was instructed to rate the extent to which they agreed with each item on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). Their responses to these eight items were averaged to create an overall measure of need for social status (x ⫽ 5.36, SD ⫽ 0.87; ␣ ⫽ .82). The questionnaire also included two measures of self-monitoring. Given the controversy surrounding the reliability of self-monitoring measures (see John, Cheek, & Klohnen, 1996), we felt it was important to replicate our findings using multiple measures. Our first measure, the SelfMonitoring Scale (SMS; Snyder, 1974), consists of 25 self-descriptive statements intended to capture several elements of social adroitness, including concern with situational appropriateness, attention to social cues, and ability to control expressive behavior.2 Each of the items (e.g., “I’m not always the person I appear to be”) was rated using true or false responses. We summed the true responses to all 25 items (some of the items were reverse scored) to create an overall score for self-monitoring (x ⫽ 13.17, SD ⫽ 3.61). Those who are high self-monitors should have high scores on the SMS, and those who are low self-monitors should have low scores. We also measured self-monitoring using a 13-item scale developed and validated by Lennox and Wolfe (1984). Sample items in the Lennox and Wolfe scale include “in social situations, I have the ability to alter my behavior if I feel that something else is called for” and “I am often able to read people’s true emotions correctly through their eyes.” Responses to these 13 items were given using a 4-point scale that ranged from 1 (not like me at all) to 4 (very much describes me). The responses were then averaged to create an overall self-monitoring score (x ⫽ 2.86, SD ⫽ 0.52). The reliability (alpha) coefficient for the entire scale was .84.
Results We had suggested that high self-monitors have an acute need for social status that drives some of their exchange behavior. As expected, the two measures of self-monitoring were highly correlated (r ⫽ .53, p ⬍ .01). Further, the data reveal positive and significant correlations between the participants’ reported need for social status and their self-monitoring scores, both for the SMS (r ⫽ .31, p ⬍ .01) and for the Lennox and Wolfe scale (r ⫽ .25, p ⫽ .01). These preliminary results support our assumption that high self-monitors are motivated by a strong need for social status.
Discussion The results reported here provide some initial evidence of the link between self-monitoring and a need for social status. Participants in our study who rated themselves as high self-monitors also gave higher responses on our measure of need for social status, which included items such as “I want my peers to respect me and hold me in high esteem” and “I am not concerned with my status among my peers” (reverse scored). In the studies that follow, we 2 Following the recommendation offered by John et al. (1996, pp. 772–773), we refrained from using the 18-item revised SMS because it overlapped too closely with extraversion. Given that extraversion can be related to social status (e.g., Anderson et al., 2001), we felt it was important to heed this recommendation.
FLYNN, REAGANS, AMANATULLAH, AND AMES
1126
consider how high self-monitors may show themselves to be more sensitive to status dynamics in exchange relations by demonstrating greater accuracy in judging asymmetric exchange relations and altering their exchange behavior in ways that elicit status conferrals.
Study 2 In our second study, we tested our prediction that high selfmonitors would be more perceptive of status differences in interpersonal relations than low self-monitors (e.g., one actor influences the other actor in the dyad, but that influence is not reciprocal).
Method Participants Seventy undergraduate students at an East Coast university participated in this study. Participants (46 women and 24 men) were recruited via flyers that advertised a study on learning relationships. They were compensated $10 for their participation.
Procedure At the beginning of the session, participants were instructed to fill out a brief questionnaire that included a measure of self-monitoring. They were then asked to complete an exercise that measures how accurately they can learn social relations and the status hierarchy of these relations. The design of this exercise was identical to that originally developed by DeSoto (1960). It focused on a fictitious group of four individuals named Bob, Joe, Mary, and Sally. No other personal information about these four individuals was provided. Participants were told that the objective of the task was to learn the nature of the four targets’ dyadic exchange relations (“who influences whom?”). For many people, this particular exercise can be difficult because the network of exchange relations is incomplete. That is, several of the relationships are asymmetrical, so that one individual has more status than the other (e.g., Sally influences Joe, but Joe does not influence Sally). In each round, the participant was given information about each of the 12 possible relations among the four targets (e.g., “Joe influences Bob”) and asked to indicate whether the information was true or false. After providing a response, the participant was informed whether the response was correct. The first round was a preliminary trial that was intended to give participants a complete set of information about the 12 relationships. Before each subsequent round, the 12 bits of information were shuffled. Rather than have participants administer flash cards and monitor their own performance, as was the case in the original design, we created a computerized version of the exercise. Again, participants were given information about each of the 12 possible dyadic relations and asked to indicate whether the information was true or false. After providing their response to each item, the participant was informed whether it was correct or not and then asked to move on to the next item (the participant had to provide a response before proceeding to the next item). At the end of the round, if the participant answered any of the 12 items incorrectly, the computer automatically shuffled the items and repeated the exercise again. The exercise was completed when the participant correctly answered each of the 12 items for two consecutive rounds (as was the case in the original exercise). Eleven participants were unable or unwilling to complete the exercise. For those who completed the exercise, the average number of rounds they needed to finish was 13.29 (SD ⫽ 7.09).
Measures Accuracy. To assess accuracy, we measured whether the participants answered the items correctly or not. In this case, we consider two dependent variables: (a) the likelihood that a respondent will identify all 12 exchange relations correctly in a given round and (b) the proportion of correct responses given for each round. We did not include responses from the first round because the participant was not yet given information about the nature of each exchange relation. We also removed responses from the last two rounds because the successful completion of these two rounds indicated that the participant had already learned each of the 12 exchange relations. Thus, if the participant required 10 rounds to complete the exercise, we used their responses only from Rounds 2 through 8. Self-monitoring. We assessed the participants’ self-monitoring personality with the SMS. Each of the items (e.g., “I’m not always the person I appear to be”) was rated using true or false responses. We summed the true responses to all 25 items (some of the items were reverse coded) to create an overall score for self-monitoring (x ⫽ 13.83, SD ⫽ 3.96).
Results Our main interest is whether high self-monitors were more accurate in their perceptions of exchange relations. We tested this idea in two ways—first by predicting the number of trials participants needed to complete the task and second by analyzing the accuracy of individual responses on each round (i.e., the proportion of correct responses). We estimated the effect of selfmonitoring on the number of trials needed to complete the task using a hazard model, in which failure in our model is completing the task and the number of rounds is our duration variable. We included the participant’s sex (0 ⫽ male, 1 ⫽ female) and age to control for demographic differences (e.g., Flynn & Ames, 2006). Participants who did not complete the task were treated as censored observations, which contribute to the calculation of our parameter estimates. By including these censored observations, we rule out the possibility that high self-monitors completed the task more quickly but were also more likely to drop out of the study. Given that we focus on participant rounds as the observations in our analysis, we have repeated observations for each participant. This kind of clustering violates the independence assumption and can artificially reduce the size of standard errors. To adjust our standard errors for repeated observations, we included a random effect for each participant in our sample. With these random effects, the results of our duration model indicate that an increase in self-monitoring led to an increase in accuracy ( ⫽ ⫺.89, z ⫽ ⫺2.03, p ⬍ .05). Put differently, high self-monitors required fewer rounds than other participants to complete the task. Thus, high self-monitors appeared to demonstrate greater facility in identifying these fictitious relations, most of which were characterized by status asymmetry. Next, we analyzed the association between self-monitoring and the performance of participants on each round using ordinary least squares regression. Our dependent variable was the proportion of correct responses from the focal round. We included the participant’s sex (0 ⫽ male, 1 ⫽ female) and age to control for demographic differences. In addition to these controls, we included a dummy variable that indicated whether the participant completed the task (0 ⫽ yes, 1 ⫽ no), a measure that represented the number of attempts the participant had previously made, and the proportion of correct responses from earlier rounds. Including the number of earlier rounds and the proportion of correct responses from earlier
HELPING ONE’S WAY TO THE TOP
rounds allowed us to control for the respondent’s ability to learn the relationships. Again, we have repeated observations for each individual, so we introduced a random effect to adjust our standard errors. In this regression analysis, the participant’s self-monitoring score once again had a significant effect on the accuracy measure,  ⫽ .09, t(86) ⫽ 2.68, p ⬍ .01, even when we controlled for the number of attempts made in earlier rounds—that is, even when we controlled for the fact that high self-monitors may learn the overall network more quickly than low self-monitors. Specifically, high self-monitors were more likely to report each exchange relation correctly than were low self-monitors. Taken together with the results from the duration model, these findings suggest that high self-monitors may be better able to perceive status-asymmetric exchange relations, at least in this fictitious social network.
Discussion The results from Study 2 support our prediction that high self-monitors are more accurate in perceiving exchange relations. Their enhanced accuracy reflects not only an ability to detect whether an exchange relation exists but also what the relative status of the actors involved in the relation may be (e.g., Bob influences Mary). These results seem consistent with findings from previous research on self-monitoring and person perception (e.g., Ickes, Stinson, Bissonette, & Garcia, 1990). However, these results are, of course, limited. They do not indicate whether such sensitivity to the status dynamics of exchange relations can also affect high self-monitors’ behavior and their ability to elicit conferrals of social status. In Study 3, we built on the findings from Study 2 by examining the behavior and reputations of high self-monitors in the workplace. In particular, we were interested in whether high selfmonitors acquire more status and influence among their coworkers than do low self-monitors. Further, we examined how selfmonitoring relates to exchange dynamics—are high self-monitors more sensitive to the status implications of exchange and therefore more likely to demonstrate generous exchange behavior (according to their fellow coworkers)? Finally, we tested the idea that patterns of exchange and corresponding impressions of generosity can mediate the relationship between self-monitoring and conferrals of social status. That is, we examined whether high selfmonitors acquire more generous reputations that, in turn, allow them to acquire more status and influence over their colleagues.
1127
As part of a class exercise in an organizational behavior course, participants were required to gather feedback from several former coworkers. Participants identified their respondents and contacted them directly with a standard set of instructions. Respondents were asked to use an anonymous online survey to rate the participant on several dimensions, including generosity and social status. On average, participants gathered 4.32 (SD ⫽ 1.51) responses from work colleagues. We asked raters to clarify how well they knew the ratee using a 4-point scale that ranges from 1 (not well at all) to 4 (extremely well). The average rating for familiarity was 3.15 (SD ⫽ 0.44). Raters were informed that these ratings would remain confidential and would not affect the participant’s course grade. In a separate questionnaire, each participant was also asked to provide self-report ratings of self-monitoring and other personality variables. These measures are described in more detail below.
Measures Social status. According to Anderson, John, Keltner, and Kring (2001) and others (e.g., Bourdieu, 1984; Ridgeway, 1991; Wegener, 1992), highstatus individuals are not only held in higher esteem but also are given greater influence over group decision making and are sought out for their affiliation and support. To capture this notion of interpersonal influence as a key component of social status, we asked respondents to rate the focal participant on five items, including “s/he is able to persuade other people and change their opinions,” “s/he fails to direct and steer meetings in his/her favor” (reverse coded), and “s/he is able to build coalitions to get things done.” Respondents indicated the extent to which each of these statements characterized the target using 7-point scales that range from 1 (never) to 7 (always). The overall reliability (alpha) coefficient for the five-item scale was .83. The average of these responses was used to represent others’ perceptions of each participant’s status (x ⫽ 5.50, SD ⫽ 0.52). Perceived generosity. In addition to rating the participant’s social status, coworkers were asked to provide ratings of the participant’s helpfulness, or generosity. To assess generosity, we used five items: (a) “s/he is willing to help when needed,” (b) “s/he asks for help from others but does not reciprocate in turn (reverse coded),” (c) “s/he is flexible and tries to accommodate others’ needs,” (d) “s/he is not effective at giving helpful/ constructive feedback to others” (reverse coded), and (e) “s/he is unwilling to sacrifice his/her self interest for the good of the team” (reverse coded). Respondents indicate the extent to which each of these statements characterized the target using 7-point scales that range from 1 (never) to 7 (always). The overall reliability (alpha) coefficient for the five-item scale was .70. The average of these responses was used to represent others’ perceptions of the target’s generosity (x ⫽ 5.98, SD ⫽ 0.49). Self-monitoring. We measured self-monitoring using the 13-item scale developed by Lennox and Wolfe. Responses to these 13 items were given using a 6-point scale that ranged from 1 (certainly always false) to 6 (certainly always true; x ⫽ 4.16, SD ⫽ 0.51). The overall reliability (alpha) coefficient for the scale was .80.
Study 3 We tested these ideas by examining exchange dynamics in the workplace, a context in which conferrals of social status are highly valued and earnestly sought.
Method Participants The participants were 306 students enrolled in a 2-year full-time master of business administration (MBA) program at an East Coast university. The sample consisted of 84 women (27%) and 222 men (73%).
Control Variables We have argued that self-monitoring behavior leads to conferrals of social status, but it may be that status leads to self-monitoring behavior (because high-status people feel compelled to maintain their positive public image). Although this possibility seems at odds with research on status and attention focus (e.g., Fiske, 1993; Keltner, Gruenfeld, & Anderson, 2003), we nevertheless thought it was important to consider. To this end, we followed previous research on status conferrals (e.g., Flynn, 2003) by gathering several individual measures of status. In particular, participants were asked to report their sex and race (coded as a dummy variable: 1 ⫽ White; 0 ⫽ non-White), which serve as diffuse status characteristics (e.g., Ridgeway, 1991). To control for access to resources (e.g., Blau, 1963), we
FLYNN, REAGANS, AMANATULLAH, AND AMES
1128
Table 1 Means, Standard Deviations, and Correlations Among Variables in Study 3 Variable 1. 2. 3. 4. 5. 6. 7. 8.
Class White Female Work experience Extraversion Openness Blirtatiousness Overall selfmonitoring 9. Perceived generosity 10. Social status M SD * p ⬍ .05.
1 — ⫺.09 ⫺.13* .13* ⫺.04 .00 ⫺.05
2
4
5
— .00 — ⫺.13* ⫺.06 — .06 .06 ⫺.05 .06 .06 .03 .10 ⫺.12* ⫺.13*
⫺.01 ⫺.03 ⫺.21** .02 ⫺.10 .08 0.58 0.49
3
0.54 0.50
.11 ⫺.02 .13* ⫺.05 ⫺.02 ⫺.02 0.28 0.45
— .20** .43**
6
7
— .13*
—
.26** .18** .28** ⫺.05 ⫺.07 ⫺.08 .12* .09 .07
5.74 3.29
4.98 1.54
5.55 1.13
3.06 0.66
8
9
— .11 — .16** 0.51** 4.16 0.51
5.98 0.49
10
— 5.50 0.52
** p ⬍ .01.
collected two variables. First, as a measure of previous work experience, participants were asked to report the number of years they had worked before enrolling in graduate school. Second, to control for intelligence, we collected each participant’s undergraduate grade point average. To provide evidence of discriminant validity, we attempted to control for several other traits that might explain our findings: extraversion, blirtatiousness, and openness to experience. In the past, measures of selfmonitoring have been shown to overlap significantly with measures of extraversion, which also refer to an outward, or social, disposition (e.g., John et al., 1996). Blirtatiousness, which captures how quickly, frequently, and effusively people respond to their partners in conversation, is also closely linked to the self-monitoring construct, although it is typically associated with low rather than high self-monitoring. Openness to experience has been linked to image enhancement (e.g., Flynn, 2005) as well as social status (e.g., Hogan & Hogan, 1991; Mann, 1959). We measured extraversion and openness using the Ten Item Personality measure (Gosling, Rentfrow, & Swann, 2003), which captures each BigFive dimension with a pair of items (e.g., “is extraverted, enthusiastic”). Participants rated their level of extraversion and openness using a scale that ranges from 1 (disagree strongly) to 7 (agree strongly). We calculated participants’ average responses to these two items (rextraversion ⫽ .61, p ⬍ .01; ropenness ⫽ .46, p ⬍ .01) to compile a score for each construct (extraversion: x ⫽ 4.98, SD ⫽ 1.54; openness: x ⫽ 5.55, SD ⫽ 1.13). We measured blirtatiousness using the BLIRT scale (Swann & Rentfrow, 2001). Participants rated the eight BLIRT items on a scale that ranges from 1 (strongly disagree) to 5 (strongly agree). Sample items include “it often takes me awhile to figure out how to express myself” (reverse scored) and “if I have something to say, I don’t hesitate to say it” (x ⫽ 3.06, SD ⫽ 0.66). The coefficient alpha for this scale is .73.
Results Means, standard deviations, and correlations are reported in Table 1.3 To test our predictions, we conducted regression analyses following the steps outlined by Baron and Kenny (1986). For each regression, we included our entire set of control variables. We predicted that high self-monitors would be better able than low self-monitors to elicit status conferrals from their colleagues. Consistent with this argument, we found a significant positive coefficient for self-monitoring on social status,  ⫽ .15, t(258) ⫽ 2.27, p ⬍ .05. We also argued that high self-monitors can increase
their social status by adapting their behavior in exchange relations, being more rather than less helpful to their peers. That is, a high self-monitor’s elevated status may be partly due to her or his generosity (being the target of help rather than soliciting help). In fact, the effect of self-monitoring on perceived generosity was positive and significant,  ⫽ .14, t(258) ⫽ 2.02, p ⬍ .05. Finally, we proposed that conferrals of social status may be partly driven by perceived generosity, an assumption that has been widely cited but rarely demonstrated (cf. Blau, 1963; Flynn, 2003). In this sample, the link between perceived generosity and social status is positive and significant,  ⫽ .56, t(258) ⫽ 10.17, p ⬍ .01. To test the idea that generosity is a means by which conferrals of social status can be elicited, we conducted a regression in which the measure of status was regressed on the measures of selfmonitoring and generosity simultaneously (e.g., Baron & Kenny, 1986). In a combined model, the predictive power of perceived generosity remained strong,  ⫽ .56, t(257) ⫽ 10.21, p ⬍ .01, whereas the predictive power of self-monitoring dropped more substantially,  ⫽ .08, t(257) ⫽ 1.36, ns, indicating that perceived generosity acted as a mediator between self-monitoring and conferrals of social status. To test the significance of this mediation effect, we calculated a Sobel statistic. The Sobel score was 1.98, which is significant ( p ⬍ .05). A summary of the mediating analyses is depicted in Figure 1.
Discussion Our results suggest that self-monitoring may be a personological determinant of social status. High self-monitors were considered high-status members of their groups, in part because of their exchange behavior. According to our findings, generosity, or at 3
At the request of the school administration, we have not reported the means, standard deviations, and correlations for the participants’ undergraduate grade point average and GMAT scores (from Studies 2 and 3) in Tables 1, 2, and 3. However, it is important to note that the means and standard deviations for these two variables are nearly identical to the means and standard deviations for the entire school population.
HELPING ONE’S WAY TO THE TOP
1129
Perceived Generosity Generosity only: ? = .56; p < .001 ? = .14; p < .05
Combined model: ? = .56; p < .001
Self-monitoring only: ? = .15; p < .05
Self-monitoring Figure 1.
Combined model: ? = .08; p = ns
Social Status
Perceived generosity mediates the relationship between self-monitoring and social status.
least a generous reputation, served as a means by which high self-monitors gained status in the eyes of their peers. Further, the extent to which peers rated participants as being generous mediated the relationship between self-monitoring and social status. In fact, these results remained robust even when we controlled for several measures of demographic- and resource-based status. Although these additional control variables do not completely remove the possibility of reverse causality, they do diminish the likelihood that status led to an increase in self-monitoring in this sample, rather than vice versa. In addition, our hypothesis was supported despite the fact that our generosity measure (the mediator) had a restricted range (mean of almost 6 on a 7-point scale), thereby making the test of our hypothesis more conservative. Taken together, the findings from our first two studies suggest that (a) high self-monitors demonstrate more accuracy in perceiving status dynamics in exchange relations and (b) their generous reputations enable them to increase their social status. Although these results are promising, we also note their limitations. In particular, the relations examined in Study 2 were fabricated, whereas those considered in Study 3 were difficult to compare— participants worked in a wide range of industries, in which specific norms of interpersonal interaction may have affected their behavior. We felt it was important to replicate these findings in a field setting in which group membership is restricted. Further, we wanted to look more carefully at the helping behavior mechanism, specifically the direction of help given and received. Is it the case that high self-monitors are more often sought out for help, more loath to request help from others, or perhaps both? In Study 4, we examined dyadic exchange relations and assessed the likelihood that people would be sought out for help and their inclination to request help. This allows us to determine how high self-monitors are building generous reputations (by providing help to others or by not imposing on them) and whether they tend to occupy high-status positions in these exchange relations. As mentioned earlier, the status dynamics of an exchange relation can be inferred by examining the pattern of resource sharing. For example, if A is willing to request assistance from B but B is not inclined to request assistance from A, this would indicate a highstatus exchange relation for B and a low-status exchange relation for A. High self-monitors, we believed, would be more likely to develop exchange relations in which they occupy high-status positions (as help giver rather than help seeker).
Study 4 Method Participants The participants were 180 MBA students from an East Coast university. During the 1st year of the MBA program, students were required to take courses with the same group of fellow students (referred to as a cluster). Each cluster contained 60 individuals. The sample was 73% men and 27% women; 69% were White, 31% were non-White. This MBA student sample offers several advantages for a study of how high self-monitors may perceive and develop exchange relations. The composition of each cluster is stable over a specific period of time. Further, MBA students have ample opportunity to observe exchange relations among their colleagues because they spend a considerable amount of time with each other (as a cluster), in and out of class. Finally, help from peers is considered a critical resource for students who hope to enhance their academic performance as well as their career prospects.
Procedure Participants were asked to complete a questionnaire that assessed the nature of their exchange relations with and among their “clustermates.” At the time the survey was administered, students had been interacting with other members of their cluster for 8 weeks. The questionnaire was divided into several sections. The first section presented a complete list of students in the cluster. Using this list, participants were asked to indicate “whom would you go to for help or for advice if you had a question or a problem? Such help or advice might include assistance on a course assignment, copies of notes from classes you may have missed, career consultations, or other things.” A small sample of 2nd-year students previously indicated these kinds of helping behaviors were both typical and significant. In the following section of the survey, participants were presented with the same list of clustermates, but in this case they were asked to indicate which members of their cluster might come to them for help or advice. Participants were asked to describe both sides of each dyadic exchange relation (R) because two parties can have different impressions of the same interaction (Krackhardt, 1987; Laumann & Knoke, 1987). Thus, for each pair, we know Riji, Rijj, Rjii, and Rjij. Riji, for example, indicates the exchange relation from person i to person j as reported by person i (i.e., person i’s account of whether he would go to person j for help and advice), whereas Rijj indicates the exchange relation from person i to person j as reported by person j (i.e., person j’s account of whether person i would come to him or her for help and advice). Participants were also asked to describe exchange relations among other members of their cluster. We used a method of data collection pioneered by David Krackhardt (1987) in which a focal participant is asked to describe
FLYNN, REAGANS, AMANATULLAH, AND AMES
1130
Table 2 Means, Standard Deviations, and Correlations Among Study Variables (Accuracy Analysis) in Study 4 Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. M SD
Accuracy Section 1 Section 2 Same team (ij) Same team (iq) Same team (jq) Same race (ij) Same race (iq) Same race (jq) Same sex (ij) Same sex (iq) Same sex (jq) Tie strength (ij) Tie strength (iq) Mutual ties (ij) Mutual ties (iq) Mutual ties (jq) Network size (i) Network size (j) Network size (q) Extraversion (i) Self-monitoring (i)
1
2
3
4
5
6
7
8
9
10
11
12
— ⫺.13** .08** .02** .00 ⫺.19** ⫺.01** ⫺.01** ⫺.05** .00 ⫺.01** ⫺.05** ⫺.05** ⫺.07** ⫺.14** ⫺.15** ⫺.26** ⫺.10** ⫺.18** ⫺.18** ⫺.01** .00
— ⫺.43** ⫺.06** .01** .01** .00 .01** .01** .00 .00 ⫺.01 .07** .08** .32** .30** .32** .31** .34** .30** .08** ⫺.01**
— .05** .01** .01** ⫺.12** ⫺.08** ⫺.08** .02** .00 .00 ⫺.05** ⫺.05** ⫺.22** ⫺.21** ⫺.21** ⫺.26** ⫺.27** ⫺.26** .03** .01
— ⫺.01** ⫺.01** ⫺.04** ⫺.01** .01 ⫺.05** ⫺.01** .01** .26** .00 .06** .00 ⫺.02** .00 ⫺.02** ⫺.02** ⫺.01* .01**
— ⫺.01** .00 ⫺.01** .00 .00 ⫺.04** .00 .00 .30** .01** .09** .00 .02** .00 .01 ⫺.01** ⫺.01**
— .00 .00 ⫺.01** .00 .00 ⫺.05** .00 ⫺.01 .01* .00 .09** .00 .02** .00 .01 .00
— .16** .15** .08** .02** .03** .05** .01* .01** .02** .00 .04** .01** .02** .00 .05**
— .18** .02** .07** .03** .01** .05** .01 .05** .01** .02** .01** .03** ⫺.01** .06**
— .02** .02** .07** .00 .01** .02** .02** .05** .03** .03** .03** ⫺.01* .00
— .16** .17** .00 .00 .03** .01** .00 .01** ⫺.01* .00 .00 .01**
— .15** ⫺.02** .03** .00 .05** .00 .03** ⫺.02** .02** .00 ⫺.01**
— .01** .01** .02** .01** .05** .00 .02** .02** .01** .01**
0.83 0.37
0.32 0.47
0.28 0.45
0.05 0.21
0.05 0.22
0.05 0.22
0.61 0.49
0.59 0.49
0.60 0.49
0.59 0.49
0.60 0.49
0.59 0.49
Note. i represents ego (i.e., the respondent) and j and q represent Alter 1 and Alter 2, respectively. * p ⬍ .05. ** p ⬍ .01.
the exchange relation between each pair of individuals in a group. In Krackhardt’s original studies, small samples were used, which made it feasible for each participant to describe the exchange relations of every other group member. Our large sample size led us to modify this technique by randomly selecting a subset of the group (7 to 9 clustermates) for each participant and asking that participant to describe the exchange relations for each person in the subset. The participants were presented with a customized grid, which included the names of each of their clustermates in the rows and approximately 8 of their randomly selected clustermates in the columns. Each participant was asked to indicate which of their clustermates listed in each of the columns would go to those listed in the rows for help or advice. Participants were also asked to complete Lennox and Wolfe’s 13-item self-monitoring scale, which was used in Study 1 and in Study 3. Items were rated on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree; x ⫽ 3.66, SD ⫽ 0.49). The overall reliability (coefficient alpha) for this scale was .78. In addition to measuring self-monitoring, we collected a measure of extraversion. Although extraversion did not appear to materially affect the results in Study 3, given the fact that previous studies have found a close relationship between these two constructs (e.g., John et al., 1996), we felt it was important to control for extraversion in each of our analyses. We measured extraversion with self-reports of eight items drawn from the Big Five Inventory, which has been found to be both reliable and valid (e.g., John, Donahue, & Kentle, 1991). Each of these items (e.g., “I am talkative”) was rated on a scale ranging from 1 (disagree strongly) to 5 (agree strongly). The items were then averaged to create an overall score for extraversion (x ⫽ 3.44, SD ⫽ 0.81). The coefficient alpha reliability of this scale was .81. The final section of the survey gathered demographic information, including sex and race data. The overall response rate for the questionnaire,
which took about 30 min to complete, was 95%. Means, standard deviations, and correlations among study variables are reported in Tables 2 and 3.
Self-Monitoring and Accuracy in Perceiving Others’ Exchange Relations In our first set of analyses, we focused on the perception of exchange relations among the participants’ (egos) colleagues (alters). Our dependent variable is the accuracy with which the focal individual described the confirmed exchange relations among his clustermates. Each participant (i) was asked to indicate whether individual j went to individual q for help and advice. That person’s response was defined as Rjqi. Rjqi was compared to Rjq. To be rigorous in defining Rjq (see Carley & Krackhardt, 1996; Krackhardt, 1987), we required that both individual j and individual q report an exchange relation between j and q (both alters agreed that an exchange relation exists between them). That is, Rjq ⫽ 1 if and only if Rjqj ⫽ Rjqq ⫽ 1; otherwise Rjq equals zero. Finally, if Rjqi ⫽ Rjq (i.e., the participant correctly perceived the existence or nonexistence of an exchange relation), accuracy equals one. Otherwise, accuracy equals zero. We included several control variables in our equations to rule out alternative explanations. First, we included a measure of network size (NS) because high self-monitors may have larger networks. In addition, people with larger networks may be more accurate in perceiving others’ exchange relations because they have access to more information about other members of their network. We calculated NS as the number of exchange relations the participant had (e.g., Wasserman & Faust, 1994). NSi is the number of relations that involved the focal participant i (x ⫽ 10.21, SD ⫽ 5.49), NSj is the number of relations that involved person j (x ⫽ 10.16, SD ⫽ 5.52), and NSq is the number of relations that involved person q (x ⫽ 10.20, SD ⫽ 5.49).
HELPING ONE’S WAY TO THE TOP
1131
13
14
15
16
17
18
19
— .05** .38** .13** .13** .25** .24** .02** .01** .00
— .13** .38** .13** .25** .02** .24** .02** .00
— .34** .38** .56** .60** .09** .03** ⫺.03**
— .35** .58** .08** .57** .02** .00
— .08** .58** .58** .00 ⫺.03**
— .09** .10** .05** ⫺.01**
— .11** .01* ⫺.03**
10.21 5.49
10.16 5.52
0.17 0.38
0.17 0.38
2.11 2.10
2.11 2.12
2.10 2.13
Participants might have a better sense of the presence or absence of a relationship if they are connected to members of the focal dyad. Therefore, we included a control variable, tie strength (TS), that represents the existence of an exchange relation between the participant and members of the focal dyad. TSij equals one if Rij ⬎ 0 or if Rji ⬎ 0, so TSij equals one if an exchange relation exists between the participant and colleague j. TSiq equals one if Riq ⬎ 0 or if Rqi ⬎ 0, so TSiq equals one if an exchange relation exists between the participant and colleague q. Participants might also have a better sense of exchange relations that involve people with
20
— .01** .00 10.20 5.49
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
Seek advice Give advice Low status High status Section 1 Section 2 Race Sex Extraversion Self-monitoring
M SD * p ⬍ .05.
2
3
4
5
6
— .57** .74** ⫺.01 .26** ⫺.30** .11 .02 ⫺.03 ⫺.09
— .00 .71** .28** ⫺.32** .08 .11 .08 .11
— ⫺.22** .13 ⫺.01 .05 ⫺.06 ⫺.01 ⫺.16*
— .15 ⫺.02 .00 .08 .19* .16*
— ⫺.47** .04 .01 .07 ⫺.02
— ⫺.11 .00 .04 .00
** p ⬍ .01.
7.07 4.54
3.01 2.97
— .20**
— 3.69 0.51
whom they are connected indirectly through mutual third-party connections (MT). Therefore, we control for the number of mutual third-party ties around the focal dyad. For example, MTij is the number of mutual thirdparty ties that include the participant and colleague j. We included demographic predictors to control for the effects of social similarity. Same race (SR) is a dummy variable that equals one if the two focal individuals are of the same racial status. SRij equals one if ego and alter j both occupy a majority racial status or a minority racial status (i.e., both are White or both are non-White). SRiq equals one if ego and alter q
1
7.07 4.80
22
3.42 0.81
Table 3 Means, Standard Deviations, and Correlations Among Study Variables (Analysis of Exchange Patterns) in Study 4 Variable
21
3.01 2.61
0.33 0.47
0.32 0.47
7
8
— .16* — .02 .01 .13 ⫺.03 0.72 0.45
9
10
— .18*
—
0.73 3.44 0.44 0.80
3.69 0.51
FLYNN, REAGANS, AMANATULLAH, AND AMES
1132
Table 4 Summary of Logistic Regression Analyses Predicting Accuracy Overall Model
Predictor variable
Step
(N ⫽ 85,439)
Pseudo R2
STij
STiq
STiq
SRij
SRiq
SRjq
SSij
SSiq
SSjq
TSij
TSiq
1 2 3 4
9688.43** 10586.65** 10606.60** 10610.76**
0.13 0.14 0.14 0.14
0.19** 0.12* 0.12* 0.03*
⫺0.04 ⫺0.03 ⫺0.03 ⫺0.03
⫺1.92** ⫺1.81** ⫺1.81** ⫺1.81**
⫺0.07** ⫺0.08** ⫺0.08** ⫺0.08**
⫺0.03 ⫺0.03 ⫺0.03 ⫺0.03
⫺0.33** ⫺0.30** ⫺0.31** ⫺0.31**
⫺0.03 ⫺0.05* ⫺0.05* ⫺0.05*
⫺0.09** ⫺0.10** ⫺0.10** ⫺0.10**
⫺0.37** ⫺0.34** ⫺0.34** ⫺0.34**
0.05 0.05 0.06
⫺0.05 ⫺0.05 ⫺0.05
2
Note. Individual and section fixed effects were used as controls in each analysis. i represents ego (i.e., the respondent) and j and q represent Alter 1 and Alter 2, respectively; ST ⫽ same team; SR ⫽ same race; SS ⫽ same sex; TS ⫽ tie strength; MT ⫽ mutual ties; NS ⫽ network size; GMAT ⫽ Graduate Management Admission Test Ego; EXTR ⫽ extraversion ego; SM ⫽ self-monitoring ego. * p ⬍ .05. ** p ⬍ .01.
occupy the same racial status. SRjq equals one if alter j and alter q occupy the same racial status. Same sex (SS) equals one if the two focal individuals are the same sex. SSij equals one if ego and alter j are SS. SSiq equals one if ego and alter q are SS. SSjq equals one if alter j and alter q are SS. In this set of data, the participant (i) was asked to describe exchange relations from one colleague (j) to another (q). As a result, we have multiple observations for each i, j, and q. This clustering is a violation of the independence assumption and may artificially decrease the size of our standard errors, which could, in turn, inflate the levels of our significance tests. To control for nonindependence among our observations, we created 179 (N ⫺ 1) individual fixed effects. The predictors for the 3 individuals involved in the focal triad are set equal to one, and the other fixed effects are set equal to zero. One might argue that accuracy increases with self-monitoring because high self-monitors are more intelligent than low self-monitors. To account for the focal participant’s level of intelligence, we included Graduate Management Admission Test (GMAT) scores in our analyses. Finally, we attempted to control for common group membership in a couple ways. First, different norms may have emerged in each of the three clusters that affected the development of exchange relations (although participants were randomly assigned to their respective clusters). To control for this possibility, we included dummy variables for two of the three sections in each of the analyses. Second, in their 1st year of the MBA program, students were assigned to project teams (similar to those described in Study 3), and group assignments composed much of their course work. We attempted to control for common group membership in each of the analyses. Same team (ST) is a dichotomous variable that equals one if the participant and the focal contact are members of the same project team and equals zero otherwise. STij equals one if person i and person j are members of the same team. STiq equals one if person i and person q are members of the same team. Finally, STjq equals one if person j and person q are members of the same team.
Results Accuracy Analyses Parameter estimates from a logistic regression model are reported in Table 4. The predictors are entered in a block format. The final model in the table contains the entire set of predictors, although the fixed effects are not reported. Consistent with our hypothesis, self-monitoring influenced the accuracy with which participants perceived exchange relations among other members of their social network. The coefficient of the self-monitoring variable is positive and significant ( ⫽ .05; p ⬍ .05), which indicates that high self-monitors were more accurate in reporting the exis-
tence (and lack thereof) of exchange relations among other members of their social group.
Self-Monitoring and the Status Dynamics of Exchange Relations In our second set of analyses, we examined whether high selfmonitors were more likely to be sought out for help and whether they were more likely to refrain from requesting help compared with low self-monitors. In addition, we considered whether high self-monitors tended to occupy a relatively higher status position in their exchange relations (i.e., situations in which they do not seek help but are sought out by others for help). Recall that for every interaction, we collected two reports on each side of the exchange relation, Riji and Rijj for the connection from ego to alter as well as Rjii and Rjij for the exchange relation from alter to ego. In this analysis, we again focused on confirmed exchange relations, in which both parties agreed that an exchange relation exists (see Carley & Krackhardt, 1996). To create a measure of ego’s tendency to seek help, we summed all the cases in which Riji ⫽ Rijj ⫽ 1 (the ego seeks help, and the exchange relation is confirmed by the alter). To create a measure of ego’s tendency to be sought out for help, we summed all the cases in which Rjii ⫽ Rjij⫽1 (the ego is sought out for help, and the exchange relation is confirmed by the alter). To test whether high self-monitors were more likely to occupy a position of higher status in their exchange relations, we calculated a measure of the ego’s tendency to occupy a high-status position in an exchange relation by summing the number of cases in which the alter sought help from the ego, but the ego did not seek help from the alter (for cases in which this was confirmed by both parties).4 We also calculated a measure of the ego’s tenden4 There are four possibilities for each exchange relation: the focal actor (a) gives and receives help, (b) does not give and does not receive help, (c) gives help, but does not receive it, or (d) receives help, but does not give it. We examine three variables: first, whether people give help (combining a and c), second, whether people receive help (combining a and d), and third, as a measure of social status, whether people find themselves in exchange relations in which they give help but do not receive it (c). Thus, although all three measures are related, our measure of social status is not the same as the first measure of help giving or the second measure of help receiving.
HELPING ONE’S WAY TO THE TOP
1133
Predictor variable MTij
MTiq
MTjq
NSi
NSj
NSq
GMAT
EXTR
SM
0.06** 0.06** 0.06**
0.01 0.01 0.01
⫺0.16** ⫺0.16** ⫺0.16**
⫺0.06** ⫺0.06** ⫺0.06**
⫺0.07** ⫺0.07** ⫺0.07**
⫺0.05** ⫺0.05** ⫺0.05**
⫺0.03** ⫺0.03**
0.05** 0.04**
0.05*
cyto occupy a low-status position in an exchange relation by summing the number of cases in which the ego sought help from the alter, but the alter did not report seeking help from the ego (and this was confirmed by both parties). Once again, we included the simple demographic controls that were used in our previous analysis. We also included our measures of extraversion, GMAT score, and fixed effects for cluster membership.
Results of Helping Behavior and Status Dynamics Analyses Parameter estimates from an ordinary least squares regression are reported in Table 5. As we predicted, high self-monitors were less likely to seek help and advice from others ( ⫽ ⫺.14, p ⬍ .05). Aside from being less willing to seek help, high self-monitors were also more likely to be sought out for help by other members of their group ( ⫽ .15, p ⬍ .05). It is important to note that these
results are based on confirmed exchange relations rather than on self-report measures. That is, the reports provided jointly by both ego and alter participants indicating whether the ego sought help from the alter and whether the alter sought help from the ego were positively affected by the ego’s level of self-monitoring. In addition to these results indicating the bilateral direction of help seeking and being sought out for help giving, we also considered whether self-monitoring related to the status ordering within dyads. Recall that exchange relations in which actor A seeks assistance from actor B but B does not seek assistance from A are high-status exchange relations for B and low-status exchange relations for A. Indeed, when considering the number of high-status exchange relations as the dependent variable, we found that high self-monitors were more likely than low selfmonitors ( ⫽ .14, p ⬍ .05) to occupy a high-status position in their exchange relations (being sought out for assistance, but
Table 5 Summary of Regression Analyses Predicting Exchange Patterns Overall model Step
F
Adjusted R2
Predictor variable Race
Sex
GMAT
Extraversion
Give advice
Seek advice
Size
Self-monitoring
Seek advice (n ⫽ 177) 1 2
12.46** 11.25**
.34 .34
.07 .07
⫺.05 ⫺.05
⫺.03 ⫺.03
⫺.05 ⫺.06
⫺.14*
.54** .54**
Give advice (n ⫽ 177) 1 2
13.60** 12.18**
.36 .36
⫺.02 ⫺.01
.10 .10
1 2
7.32** 6.72**
.22 .22
.05 .04
⫺.10 ⫺.11
.05 .05
.06 .07
Low status (n ⫽ 177) ⫺.05 ⫺.00 ⫺.05 ⫺.01
.52** .52**
.15*
.48** .48**
⫺.18**
.40** .40**
.14*
High status (n ⫽ 177) 1 2
6.04** 5.42**
.18 .18
⫺.05 ⫺.05
.06 .06
Note. GMAT ⫽ Graduate Management Admissions Test Ego. * p ⬍ .05. ** p ⬍ .01.
.04 .04
.14* .14*
FLYNN, REAGANS, AMANATULLAH, AND AMES
1134
not seeking assistance from others). Further, when considering the number of low-status exchange relations as the dependent variable, we found that high self-monitors were less likely than low self-monitors ( ⫽ ⫺.18, p ⬍ .01) to occupy a low-status position in their exchange relations with others (seeking assistance from others, but not being sought out for assistance).5
Discussion The results for Study 4 confirm our predictions about the relationship between self-monitoring and the perception and dynamics of exchange relations. High self-monitors were more accurate in reporting exchange relations involving other members of their social group. Further, high self-monitors appeared to be more sensitive to the status implications of social exchange, serving as the target of helping requests rather than the requester. This finding was driven both by the tendency to refrain from seeking assistance and by the tendency to be sought out for assistance. These results were robust even when controlling for several alternative explanations, such as the competing influence of demographic similarity, common group membership, network size, intelligence, and extraversion.
General Discussion High self-monitors tend to experience greater levels of success in managing everyday social situations (Gangestad & Snyder, 2000). Laboratory studies have found that high self-monitors are more effective in experimental groups, often emerging as leaders in situations that call for exceptional presentation and communication skills (Garland & Beard, 1979; Zaccaro, Foti, & Kenny, 1991). Beyond the confines of the laboratory, high self-monitors tend to be better performers (Mehra, Kilduff, & Brass, 2001) and to develop more favorable reputations (Kilduff & Krackhardt, 1994), and they are more likely to be promoted to higher ranking positions in organizations (Kilduff & Day, 1994). What may be driving the success of high self-monitors? We propose that one determinant of their success may be a strong need for social status—an elevated position of respect and influence among their peers. In the present research, we have attempted to explain how self-monitoring relates to the perception and development of status dynamics in exchange relations. High self-monitors were better judges of others’ interpersonal exchange relations, outperforming low self-monitors on a task that required them to learn a novel set of relationships—not only who knew whom, but who had relatively higher status (i.e., who influenced whom). We also examined how self-monitoring might affect exchange behavior. As we predicted, high self-monitors were more successful in eliciting conferrals of status, in part because they were judged to be more generous than low self-monitors. Finally, we tested the robustness of these findings by examining exchange relations in a set of social groups in which the boundaries of group membership were clearly defined. We found that high self-monitors were more accurate in perceiving the nature of others’ exchange relations. Again, consistent with our arguments about status, high self-monitors tended to occupy higher status positions in these exchange relations, more frequently being the target of requests for help, and less frequently being the requester.
These findings extend past research on how high self-monitors manage their interpersonal relationships by showing a different side of self-monitoring. For example, Snyder and Simpson (Snyder, 1987; Snyder & Simpson, 1984) found that high self-monitors were more aggressive than low self-monitors in initiating and terminating their relationships, particularly with romantic partners. Rather than settle on a single dating partner, high self-monitors jumped from one relationship to the next. These findings suggest that high self-monitors adopt an uncommitted orientation toward relationships. Our findings tell a different story—we find that high self-monitors are willing to invest resources in their exchange partners by demonstrating generosity, a clear sign of commitment according to exchange theorists (e.g., Thibaut & Kelley, 1959). However, such generosity may not be entirely altruistic. Instead, high self-monitors may be making these investments in their exchange relations with an expectation of a valuable return—a position of elevated status among their peers. This apparent link between self-monitoring and exchange dynamics may help explain previous empirical findings in the selfmonitoring literature. High self-monitors enjoy many benefits in their professional careers, including a faster rate of promotion and more favorable performance evaluations (e.g., Flynn, Chatman, & Spataro, 2001; Kilduff & Day, 1994). At the same time, selfmonitoring has been associated with the development of social exchange, particularly in professional networks (e.g., Mehra et al., 2001). The ability to perceive and manage exchange dynamics may help explain the advantage that high self-monitors hold over low self-monitors. People who have an accurate view of their own and others’ exchange relations, as well as a more favorable reputation, may be in a better position to obtain resources, build support for their ideas, and influence group decisions. This suggests a possible mediating relationship between self-monitoring, exchange relations, and individual outcomes, so self-monitoring may affect individual success by way of its influence on the perception and development of social exchange.
Limitations Although our results were consistent with our predictions, they raise a number of important questions. For example, we were unable to test the lasting impact of self-monitoring on the perception and dynamics of exchange relations because our studies were cross-sectional. It may be that self-monitoring has an initial impact on the perception or management of exchange dynamics but that the effect is fleeting rather than permanent. Are high self-monitors more concerned with making a good first impression or do they maintain a consistent pattern of helping behavior regardless of how 5 It is possible that low self-monitors adopt different relationship strategies from high self-monitors. Whereas high self-monitors seek out highstatus positions in their exchange relations, low self-monitors may seek out equal status (i.e., ego and alter seek one another out for advice or ego and alter do not seek one another out for advice). We tested this idea by examining the association between the number of each kind of equal-status relationship and self-monitoring. In both cases, the self-monitoring score did not have an effect, which indicates that the only link here is link between high self-monitors and their tendency to avoid low-status positions and to attain high-status positions in their exchange relations. We thank an anonymous reviewer for calling this idea to our attention.
HELPING ONE’S WAY TO THE TOP
long they have known their exchange partners? To answer this question, future research might gather longitudinal data on how self-monitoring influences patterns of giving and receiving help over an extended period of time. Some of our measures might also have limitations. In Study 3, we relied on others’ impressions of generosity, but these impressions may have been inflated. Perhaps high self-monitors were not contributing more to their coworkers than they received in return, but they were skillful at managing this reputation. Studies that consider not only reputation but also actual performance are needed to confirm the link between self-monitoring and helping behavior. In addition, in Study 3, the reliability score for our measure of generosity was somewhat low (.70), and our measure of social status was heavily skewed toward interpersonal influence. Future research might try to develop more robust, and perhaps more generalizable, measures of helping behavior and social status. Also, in Study 4, the possibility exists that our results were influenced by order effects. Given that participants first responded to the item “whom would you go to for help or advice?,” they may have been encouraged to think about the status implications of seeking help more than if they had first responded to the opposite item, “who would come to you for help or advice?” Perhaps a diary study in which participants are asked to record each episode of helping behavior would be useful in testing these ideas more rigorously. Finally, the analyses conducted in Study 2 and Study 4 suggest that high self-monitors are more accurate than low self-monitors in perceiving others’ exchange relations. Although this result is intriguing and consistent with our predictions, we also note the magnitude of the accuracy effect in Study 4 is relatively small (5% change in accuracy when moving from a minimal to an extreme level of self-monitoring). The size of this effect is limited, in part, by the size of the groups being studied. The groups examined in Study 4 are considerably larger than those examined in Study 2 (approximately 60 vs. 12), and in these large groups, the overwhelming majority of dyads did not have an exchange relation (the average number of confirmed relationships per student was approximately 10 out of 60). Thus, most people likely (and correctly) assumed that the overwhelming majority of possible dyadic exchange relations do not exist. Not only does this makes our test more conservative, it also limits our ability to judge the extent to which self-monitoring enhances accuracy in the perception of exchange relations. In the future, it would be worthwhile to examine real-world groups that are smaller to test whether the effect of self-monitoring on accuracy in judging exchange relations is substantive.
Future Directions There are several possible directions for this line of research to follow. First, scholars interested in self-monitoring and empathic accuracy might attempt to further delineate the mechanism(s) accounting for some of the present findings. Although we hypothesized that high self-monitors would be more accurate in their perceptions of others’ exchange relations, it remains unclear how they manage to form such accurate impressions. Given a small fictitious group, such as that used in Study 2, high self-monitors may find it easy to retain information about others’ exchange relations, especially when this information is made explicit. In
1135
large groups of 60 people, however, such as those included in Study 4, this information is not explicit or easily retained. Perhaps high self-monitors improve their performance in perceiving others’ exchange relations by effectively organizing the members of their social group into smaller cliques (i.e., they may not know who is connected to whom, but they may be making more informed guesses). Second, we found that helping behavior, or generosity, can lead to conferrals of social status, but this may not be true in all situations. In Study 3 and Study 4, participants were rated by people with whom they shared some interdependence (close coworkers and fellow classmates). Norms of helpfulness may be stronger in these situations because interdependent actors rely on cooperation to achieve mutual success. In contrast, members of a group who are not interdependent with one another, but are instead competing with one another, may not afford social status based on cooperative efforts, at least not to the same degree. If high selfmonitors perform helping behavior to attain status, as we suggest, they may be motivated to provide help only in situations in which cooperative behavior is normative and therefore a determinant of status (e.g., Snyder & Monson, 1975). Identifying the conditions in which high self-monitors are better able to use helping behavior as a means to elicit conferrals of status may be a useful endeavor in future research. Third, we predicted that high self-monitors would be loath to request help and would be sought out more frequently for help. Whereas the former is clearly under the high self-monitor’s control (resisting the urge to request help), the latter is not. How do high self-monitors become seen as a “go-to” person when others are in need? The results from Study 3 indicate that self-monitoring was related to impressions of generosity, but the link between being sought out for help and actually helping is unclear. It may be that high self-monitors receive more requests for help than do low self-monitors but are not necessarily more willing to help when others ask. Future studies that account for the incidence of helping requests are needed to test the possibility that high self-monitors are, in fact, more compliant than low self-monitors when they are presented with requests for help. Finally, we based our predictions on the assumption that requesting help lowers one’s status, whereas providing help increases it. This assumption has often been taken for granted in the literature on helping behavior and social exchange (e.g., Blau, 1963; Homans, 1958; Mauss, 1925). However, there may be situations in which this is not the case. Requesting help from high-status colleagues can be used as a form of ingratiation in which the requester elevates his or her status, at least in the eyes of the target (e.g., Jones, 1964). Given high self-monitors’ need to project a situationally appropriate image, future research might consider ways in which high self-monitors alter their behavior—so that they become help seekers—if the situation provides some advantage for doing so.
Conclusion In summary, these findings represent an important initial step toward explicating the relationship between self-monitoring and social status, especially as it relates to the perception and development of exchange relations. We found that self-monitoring was closely related to the accuracy with which people perceived others’
FLYNN, REAGANS, AMANATULLAH, AND AMES
1136
exchange relations, particularly status-asymmetric relations. Further, consistent with the notion that high self-monitors are more sensitive to status dynamics, we found that high self-monitors developed a more generous image (by providing help and by not requesting it) that, in turn, enabled them to elicit conferrals of status from others. Taken together, the findings suggest that personality traits, particularly self-monitoring, may play a significant role in perceiving and managing the status dynamics of exchange relations. We hope that this work can inspire future research on the connections between self-monitoring, exchange behavior, and social status.
References Anderson, C. P., John, O. J., Keltner, D., & Kring, A. M. (2001). The origins of status within face-to-face social groups. Journal of Personality and Social Psychology, 81, 116 –132. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Blau, P. M. (1963). The dynamics of bureaucracy: A study of interpersonal exchange in two government agencies. Chicago: University of Chicago Press. Bourdieu, P. (1984). Distinction: A social critique of the judgment of taste. Cambridge, MA: Harvard University Press. Carley, K. M., & Krackhardt, D. (1996). Cognitive inconsistencies and non-symmetric friendship. Social Networks, 18, 1–27. Costanzo, M., & Archer, D. (1989). Interpreting the expressive behavior of others: The interpersonal perception task. Journal of Nonverbal Behavior, 13, 225–245. DeBono, K. G. (1987). Investigating the social adjustive and value expressive functions of attitudes: Implications for persuasion processes. Journal of Personality and Social Psychology, 52, 279 –287. DeBono, K. G., & Snyder, M. (1989). Understanding consumer decisionmaking processes: The role of form and function in product evaluation. Journal of Applied Social Psychology, 19, 416 – 424. DeSoto, C. B. (1960). Learning a social structure. Journal of Abnormal and Social Psychology, 60, 417– 421. Fiske, S. (1993). Social cognition and social perception. Annual Review of Psychology, 44, 155–194. Flynn, F. J. (2003). How much should I give and how often? Academy of Management Journal, 46, 539 –553. Flynn, F. J. (2005). Having an open mind: The impact of openness to experience on interracial attitudes and impression formation. Journal of Personality and Social Psychology, 88, 816 – 826. Flynn, F. J., & Ames, D. R. (2006). What’s good for the goose may not be good for the gander: The benefits of self-monitoring for men and women. Journal of Applied Psychology, 91, 272–283. Flynn, F. J., Chatman, J. A., & Spataro, S. A. (2001). Getting to know you: The influence of personality on the impression formation and performance of demographically different people in organizations. Administrative Science Quarterly, 46, 414 – 442. Funder, D. C., & Harris, M. J. (1986). On the several facets of personality assessment: The case of social acuity. Journal of Personality, 54, 528 – 550. Gangestad, S. W., & Snyder, M. (2000). Self-monitoring: Appraisal and reappraisal. Psychological Bulletin, 126, 530 –555. Garland, J., & Beard, J. (1979). Relationship between self-monitoring and leader emergence across two task situations. Journal of Applied Psychology, 64, 72–76. Goffman, E. (1971). Relations in public: Microstudies of the public order. New York: Basic Books.
Gosling, S., Rentfrow, P., & Swann, W. (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37, 504 –528. Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement. American Sociological Review, 25, 161–178. Hogan, R., & Hogan, J. (1991). Personality and status. In D. G. Gilbert & J. J. Connolly (Eds.), Personality, social skills, and psychopathology: An individual differences approach (pp. 137–154). New York: Plenum Press. Homans, G. C. (1958). Social behavior as exchange. American Journal of Sociology, 63, 597– 606. Hosch, H. M., Leippe, M. R., Marchioni, P. M., & Cooper, D. S. (1984). Victimization, self-monitoring, and eyewitness identification. Journal of Applied Psychology, 69, 280 –288. Ickes, W., Stinson, L., Bissonette, V., & Garcia, S. (1990). Naturalistic social cognition: Empathic accuracy in mixed-sex dyads. Journal of Personality and Social Psychology, 59, 730 –742. Janicik, G., & Larrick, R. (2005). Social network schemas and the learning of incomplete networks. Journal of Personality and Social Psychology, 88, 348 –364. John, O. P., Cheek, J. M., & Klohnen, E. C. (1996). On the nature of self-monitoring: Construct explication via Q-sort ratings. Journal of Personality and Social Psychology, 71, 763–776. John, O. P., Donahue, E. M., & Kentle, R. L. (1991). The ‘Big Five’ Inventory–Versions 4a and 54 (Tech. Rep.). Berkeley: University of California, Institute of Personality and Social Research. Jones, E. E. (1964). Ingratiation: A social psychological analysis. New York: Appleton-Century-Crofts. Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110, 265–284. Kenny, D. A., Bond, C. F., Jr., Mohr, C. D., & Horn, E. M. (1996). Do we know how much people like one another? Journal of Personality and Social Psychology, 71, 928 –936. Kilduff, M., & Day, D. V. (1994). Do chameleons get ahead: The effects of self-monitoring on managerial careers. Academy of Management Journal, 37, 1047–1060. Kilduff, M., & Krackhardt, D. (1994). Bringing the individual back in: A structural analysis of the internal market for reputation in organizations. Academy of Management Journal, 37, 87–108. Krackhardt, D. (1987). Cognitive social structures. Social Networks, 9, 109 –134. Krackhardt, D. (1990). Assessing the political landscape: Structure, cognition, and power in organizations. Administrative Science Quarterly, 35, 342–369. Krackhardt, D., & Kilduff, M. (1999). Whether close or far: Social distance effects on perceived balance in friendship networks. Journal of Personality and Social Psychology, 76, 770 –782. Laumann, E., & Knoke, D. (1987). Organization state: Social change in national policy domains. Madison: University of Wisconsin Press. Lee, F. (1997). When the going gets tough, do the tough ask for help? Help seeking and power motivation in organizations. Organizational Behavior and Human Decision Processes, 72, 336 –363. Lennox, R. D., & Wolfe, R. N. (1984). Revision of the Self-Monitoring Scale. Journal of Personality and Social Psychology, 46, 1349 –1364. Malinowski, B. (1922). Argonauts of the Western Pacific: An account of native enterprise and adventure in the archipelagos of Melanesian New Guinea. New York: E. P. Dutton. Mann, R. D. (1959). A review of the relationship between personality and performance in small groups. Psychological Bulletin, 56, 241–270. Mauss, M. (1925). The gift: Forms and functions of exchange in archaic societies (I. Cunnison, Trans.). New York: W. W. Norton. Mehra, A., Kilduff, M., & Brass, D. J. (2001). The social networks of high and low self-monitors: Implications for workplace performance. Administrative Science Quarterly, 46, 121–146.
HELPING ONE’S WAY TO THE TOP Mischel, W., & Shoda, Y. (1998). Reconciling processing dynamics and personality dispositions. Annual Review of Psychology, 49, 229 –258. Ridgeway, C. (1991). The social construction of status value: Gender and other nominal characteristics. Social Forces, 70, 367–386. Rubin, Z., & Zajonc, R. B. (1969). Structural bias and generalization in the learning of social structures. Journal of Personality, 37, 310 –324. Sigall, H., & Landy, D. (1973). Radiating beauty: Effects of having a physically attractive partner on person perception. Journal of Personality and Social Psychology, 28, 218 –224. Snyder, M. (1974). Self-monitoring of expressive behavior. Journal of Personality and Social Psychology, 30, 526 –537. Snyder, M. (1987). Public appearances: Private realities. New York: W. H. Freeman. Snyder, M., & DeBono, K. G. (1985). Appeals to image and claims about quality: Understanding the psychology of advertising. Journal of Personality and Social Psychology, 49, 586 –597. Snyder, M., & Monson, T. C. (1975). Persons, situations, and the control of social behavior. Journal of Personality and Social Psychology, 32, 637– 644.
1137
Snyder, M., & Simpson, J. A. (1984). Self-monitoring and dating relationships. Journal of Personality and Social Psychology, 47, 1281–1291. Sorokin, P. (1927). Social mobility. New York: Harper & Row. Swann, W., & Rentfrow, P. (2001). Blirtatiousness: Cognitive, behavioral, and psychological consequences of rapid responding. Journal of Personality and Social Psychology, 6, 1160 –1175. Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: Wiley. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. New York: Cambridge University Press. Wegener, B. (1992). Concepts and measurement of prestige. Annual Review of Sociology, 18, 253–280. Zaccaro, S. J., Foti, R. J., & Kenny, D. A. (1991). Self-monitoring and trait-based variance in leadership: An investigation of leader flexibility across multiple group situations. Journal of Applied Psychology, 76, 308 –315. Zajonc, R. B., & Burnstein, E. (1965). The learning of balanced and unbalanced social structures. Journal of Personality, 33, 153–163.
Appendix Items Used to Measure Need for Social Status I want my peers to respect me and hold me in high esteem. I am not concerned with my status among my peers. (reverse scored) Being a highly valued member of my social group is important to me. I would like to cultivate the admiration of my peers. I enjoy having influence over other people’s decision making. It would please me to have a position of prestige and social standing.
I don’t care whether others view me with respect and hold me in esteem. (reverse scored) I care about how positively others view me. N ⫽ 100; ␣ ⫽ .82 Received July 25, 2005 Revision received July 14, 2006 Accepted July 17, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1138 –1151
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1138
Higher-Order Factors of the Big Five in a Multi-Informant Sample Colin G. DeYoung Yale University In a large community sample (N ⫽ 490), the Big Five were not orthogonal when modeled as latent variables representing the shared variance of reports from 4 different informants. Additionally, the standard higher-order factor structure was present in latent space: Neuroticism (reversed), Agreeableness, and Conscientiousness formed one factor, labeled Stability, and Extraversion and Openness/Intellect formed a second factor, labeled Plasticity. Comparison of two instruments, the Big Five Inventory and the Mini-Markers, supported the hypotheses that single-adjective rating instruments are likely to yield lower interrater agreement than phrase rating instruments and that lower interrater agreement is associated with weaker correlations among the Big Five and a less coherent higher-order factor structure. In conclusion, an interpretation of the higher-order factors is discussed, including possible neurobiological substrates. Keywords: Big Five, metatraits, stability, plasticity, higher-order factors
variance of Extraversion and Openness/Intellect appears to reflect the ability and tendency to explore and engage flexibly with novelty, in both behavior and cognition (DeYoung et al., 2002; DeYoung, Peterson, & Higgins, 2005). Many questions remain regarding interpretation and explanation of the metatraits. The value of discussing these issues, however, is contingent on the answer to a more basic question: Are the correlations among the Big Five real? Although Big Five scores routinely show intercorrelations, and the higherorder factors have been demonstrated with a variety of instruments and in both self- and observer ratings, several arguments have been made against the substantive reality of these correlations. Costa and McCrae (1992b) have argued that correlations among the Big Five are method artifacts, stemming from the idiosyncrasies of individual instruments. This argument is weakened by demonstrations that the Big Five are correlated even when latent variables are derived from single-informant ratings on multiple instruments (e.g., John & Srivastava, 1999; Yik & Russell, 2001). McCrae and Costa (1999) have also argued that the higher-order factors merely reflect biases in personality assessment, along two evaluative dimensions: Positive Valence (PV) and Negative Valence (NV). Their own prior work counters this assertion, however, in that they found that PV and NV were not associated with biased self-reports of the Big Five (McCrae & Costa, 1995). That the two evaluative dimensions are similar to the metatraits in their associations with the Big Five (McCrae & Costa, 1999), but do not seem to be associated with biased personality ratings, suggests instead that very general evaluations may be based on the metatraits. More recently, Biesanz and West (2004) have argued that correlations among the Big Five are indeed method artifacts, resulting not from the characteristics of individual instruments but from the biases of individual raters. Using confirmatory factor analysis (CFA), these authors found that latent Big Five variables representing the shared variance of self-, peer, and parent reports were uncorrelated, despite the fact that all three
One of the major concerns in personality psychology is the development of a comprehensive model of personality traits, typically conceived as a hierarchy in which correlated lower level traits are grouped together within broader higher level traits. The five-factor model, or Big Five, is a promising candidate (though there is some debate as to whether six- or seven-factor models would be more appropriate; Ashton et al., 2004; Saucier & Goldberg, 2001). The Big Five trait domains— Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness/Intellect— have often been conceived as orthogonal factors and the highest, most general level of the hierarchy of personality traits (Costa & McCrae, 1992a, 1992b; Goldberg, 1993). Investigation of correlations among the Big Five, however, has demonstrated that they are not orthogonal (at least as currently measured; Saucier, 2002) and that they possess a stable higher-order factor solution (DeYoung, Peterson, & Higgins, 2002; Digman, 1997; cf. Markon, Krueger, & Watson, 2005). Emotional Stability (Neuroticism reversed), Agreeableness, and Conscientiousness mark a first factor, whereas Extraversion and Openness/Intellect mark a second. Although Digman (1997, p. 1248) gave the higher-order factors, or metatraits, the “provisional” labels ␣ and , we have suggested that they be labeled Stability and Plasticity (DeYoung et al., 2002). The shared variance of Neuroticism, Agreeableness, and Conscientiousness appears to reflect the individual’s ability and tendency to maintain stability and avoid disruption in emotional, social, and motivational domains, whereas the shared
This study was supported by a grant from the Social Sciences and Humanities Research Council of Canada to Jordan B. Peterson and by an Ontario Graduate Scholarship. Thanks go to Lewis R. Goldberg for his generosity in making these data available and to Lena C. Quilty and William A. Cunningham for statistical advice. Correspondence concerning this article should be addressed to Colin G. DeYoung, Department of Psychology, Yale University, Box, 208205, New Haven, CT, 06520. E-mail:
[email protected] 1138
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
types of rating showed significant intercorrelations among the Big Five when examined separately. Had these findings been reliable, all discussion of higher-order personality factors above the Big Five would be pointless, except inasmuch as one is interested in systematic biases in personality perception. An obvious flaw in Biesanz and West’s argument, however, is that interrater agreement in their sample was quite low. The mean correlation of ratings of the same trait by different informants was only .30 (range ⫽ .18 –.43). By contrast, Costa and McCrae (1992a) reported cross-informant correlations for the Revised NEO Personality Inventory (NEO PI-R) with a mean of .47 (range ⫽ .30 –.67). Given low interrater agreement (i.e., low correlations between different informants’ ratings of the same trait), the correlations between different informants’ ratings of different traits are likely to be affected as well. Poor agreement may reduce not only the magnitude of different-trait, differentinformant correlations but also their systematicity. Reductions either in the consistency of the pattern of correlations or in their magnitude will decrease the likelihood that significant correlations will be evident among the Big Five in latent space.1 The present study involved multi-trait multi-method (MTMM) analyses conducted with a large data set with four informants’ ratings of the Big Five, where each informant was treated as a different method. Results for two instruments, the Big Five Inventory (BFI; John & Srivastava, 1999) and the Mini-Markers (Saucier, 1994), were compared. A first hypothesis was that, in a sample with greater interrater agreement, significant correlations would be evident among latent Big Five traits. A second hypothesis was that, given adequate interrater agreement, the metatraits of Stability and Plasticity would be present as higher-order factors at the latent level. An additional question regarding the metatraits was whether they would be correlated. In his CFAs, Digman (1997) tested only models with the correlation between metatraits fixed at zero. We have found, however, that latent metatraits are fairly strongly correlated; in two samples, the correlations between Stability and Plasticity were .45 and .53 (DeYoung et al., 2002). In reanalysis of Digman’s (1997) data, Mutch (2005) found that when Digman’s CFA procedure was corrected to account for the fact that his data consisted of correlation rather than covariance matrices, a twofactor solution did not fit the data well. This lack of fit might be attributable to the fact that Digman did not allow the metatraits to correlate. The present study therefore examined whether the metatraits were correlated, both in latent traits representing the shared variance across informants and within ratings by single informants. Correlations between the metatraits in data from single informants may reflect a tendency for informants to describe people positively or negatively—as having socially desirable or undesirable qualities—across all trait dimensions, in which case latent metatraits derived from ratings of multiple informants might be less strongly correlated. These latent metatraits would contain only variance agreed upon by all four informants, which should be more reliably linked to the observable patterns of behavior that constitute the substance of the Big Five, rather than to individual informants’ judgments about desirability. Another hypothesis was formulated in response to the question of why interrater agreement was so low in Biesanz and West’s (2004) sample. Whereas many factors could have contributed to poor agreement among informants, one likely possibility presents
1139
itself: The instrument used by Biesanz and West to assess the Big Five was Goldberg’s (1992) measure of 100 trait-descriptive adjectives (TDA). Single adjectives are known to produce less consistent personality ratings than longer items that provide more context, because single words are more subject to idiosyncrasies in the interpretation of their meaning (Goldberg & Kilkowski, 1985; John & Srivastava, 1999). One reason to expect greater interrater agreement in the current study is that BFI items incorporate prototypical adjective markers of the Big Five into short phrases (e.g. “is emotionally stable, not easily upset”), with the explicit intent of increasing the consistency of ratings (John & Srivastava, 1999). Because the current study incorporates a single-adjective rating instrument (the Mini-Markers) as well as the BFI, the hypothesis that use of a single-adjective rating instrument will be associated with lower interrater agreement can be tested directly. Assuming that this hypothesis holds true, a related hypothesis is that lower interrater agreement will be associated with decreased correlations among latent Big Five traits. Although the primary purpose of analyzing ratings from both the BFI and Mini-Markers was to examine the relation of interrater agreement to latent trait correlations across different instruments, the inclusion of multiple instruments also served a second purpose. A model could be fit in which each informant’s Big Five ratings were modeled as latent variables with two indicators, with scores from the BFI and Mini-Markers serving as separate indicators. Method effects could then be assessed for the two instruments and for each informant, simultaneously. This allowed an assessment of what is common across the correlational structures of both instruments in addition to ways in which they differ.
Method Participants Participants were 490 members of the Eugene–Springfield Community Sample, ranging in age from 18 to 80 years (M ⫽ 51.23, SD ⫽ 12.62). Participants were recruited by mail from lists of homeowners and agreed to
1 Biesanz and West (2004) argued against the possibility that low correlations across informants could be responsible for the lack of correlation among the latent Big Five in their multiple-informant analysis, but their reasoning is questionable: They compare their results for multi-informant ratings to their results (in the same sample) for selfratings at multiple time points. Latent Big Five variables derived from self-ratings at three different times showed significant correlations and continued to do so even when correlations among the self-ratings were reduced artificially by increasing the variance of the scores while maintaining the same covariances. In their general discussion, Biesanz and West (2004) stated that this procedure rendered the magnitude of the correlations in the self-report analysis “comparable” (Biesanz and West, 2004, p. 869) to that in the multi-informant analysis. Earlier in their article, however, they noted that their procedure “had the effect of reducing correlations among measures by approximately 50%” (p. 860). The mean different-trait, different-time absolute correlation in their self-report analysis was .25, and half of this (.12) is still more than twice as large as the mean different-trait, different-informant absolute correlation in their multi-informant analysis (.05). I would suggest that .05 is not sufficiently comparable in magnitude to .12 to settle the question of whether interrater agreement is responsible for Biesanz and West’s failure to find significant correlations among latent Big Five traits.
1140
DEYOUNG
complete questionnaires, delivered by mail, for pay. Self-reports and ratings of three additional informants on the BFI were available for 483 participants (283 female, 200 male). Self-reports and three additional informant ratings on the Mini-Markers were available for 487 participants (283 female, 204 male). The sample spanned all levels of educational attainment, with an average of 2 years of post-secondary schooling. Most participants were Caucasian American (97%), with 1% or less (for each category) identifying as Hispanic, Asian American, or Native American or not reporting their ethnicity. Participants were instructed that, in addition to filling out self-report questionnaires, they should distribute additional copies designed for peer ratings to any three people who knew them “very well.” These additional 1,470 informants (550 female, 914 male, 6 with no gender reported) ranged in age from 6 to 94 years (M ⫽ 48.17, SD ⫽ 17.99). (Because of the possibility that children may provide less reliable personality ratings, analyses were repeated with various cutoffs for age, excluding participants with ratings from informants younger than 10, 13, or 17 years. For all three of these cutoffs, results were extremely similar to those obtained in the full sample; hence, only the latter are reported.) Participants described 2.3% of the additional informants as “significant other,” 21.7% as “spouse,” 28.0% as “friend,” 11.4% as “co-worker,” 27.9% as “relative,” 1.2% as “acquaintance,” and 6.3% as “other.” No relationship status was reported for 1.3% of informants. On the whole, these raters felt favorably toward their targets; each responded to a single Likert-scale item asking how much they liked the participant, with possible responses ranging from 1 (like very much) to 6 (greatly dislike). The mean response was 1.21 (SD ⫽ 0.51; range ⫽ 1–5).
Measures Questionnaires were sent to and received from participants by mail. Informants (both self and other) rated participants’ personalities using the Big Five Inventory (BFI; John & Srivastava, 1999) and the MiniMarkers (Saucier, 1994). Both scales are well-validated as measures of the Big Five. The BFI consists of 44 descriptive phrases, with each trait indicated by 8 to 10 items. The Mini-Markers consist of 40 adjectives, with each trait indicated by 8 items. The Mini-Markers were created by taking a subset of the adjectives from the TDA, eliminating many difficult or unusual words; therefore, this measure seems likely to produce more reliable ratings than the TDA; indeed, Saucier (1994) found that the Mini-Markers had a higher mean inter-item correlation within each trait scale than did the TDA. All items were rated for accuracy by informants on a 5-point Likert scale. Trait scores were calculated as the mean item score.2 Each of the three peer ratings for each participant was assigned randomly to one of three groups.
Analyses Following examination of the MTMM correlation matrices (Table 1) to assess inter-rater agreement, the seven MTMM models described below were fitted with confirmatory factor analysis (CFA).3 Each informant was treated as a different method. The best fitting model was retained, and statistical comparisons of differences in fit were made to test the performance of models specifying orthogonality over models allowing correlated traits. 1. Correlated traits, no methods (CTNM): Models five latent trait factors and their correlations but does not assume or model any method effects. 2. Correlated traits, correlated uniquenesses (CTCU; see Figure 1A): Models method effects as correlations among the five uniquenesses for each informant. In CFA, a uniqueness represents the variance in an observed variable not explained by latent variables. No assumptions are made regarding the dimensionality of the method effects. 3. Orthogonal traits, correlated uniquenesses (OTCU): Models method effects identically to the CTCU model but assumes that the five latent trait
factors are uncorrelated. The difference in fit between CTCU and OTCU models provides a statistical test of the orthogonality of the Big Five. 4. Correlated traits, correlated methods (CTCM): Unlike the CTCU model, this model assumes that a single latent factor underlies each method effect. It also allows the latent method factors to be correlated across informants. 5. Correlated traits, orthogonal methods (CTOM; see Figure 1B): Like the CTCM model but assumes that the method factors are uncorrelated. This model is nested under the CTCU model and comparison of these two models permits a test of whether the method effects are unidimensional. 6. Orthogonal traits, correlated methods (OTCM): This model is nested under the CTCM model, being identical to it except for the assumption that the latent traits are uncorrelated. 7. Orthogonal traits, orthogonal methods (OTOM): Assumes uncorrelated latent trait and method factors. The OTOM model is nested under the CTCU, OTCU, CTCM, CTOM, and OTCM models. All models were analyzed with Amos 5.0 (Arbuckle, 2003) with maximum likelihood estimation based on the full covariance matrices. CFAs of higher-order factor structure follow the investigation of trait correlations.
Results Interrater Agreement Correlations between traits as assessed by the BFI and the Mini-Markers within raters (same trait, same informant, different instrument) were quite high (mean r ⫽ .82; range ⫽ .73–.90), indicating that the two scales assess the Big Five very similarly. Nonetheless, as predicted, interrater agreement (correlations for same trait, different informant) was significantly higher for the BFI (mean r ⫽ .41, SD ⫽ .08, range ⫽ .29 –.57) than for the MiniMarkers (mean r ⫽ .36, SD ⫽ .08, range ⫽ .25–.53), F(1, 29) ⫽ 22.87, p ⬍ .001.
Model Fit for Multi-Trait Multi-Method Confirmatory Factor Analyses Table 2 presents fit indices for the CFAs of the seven models described above, for both the BFI and Mini-Markers. In addition to the chi-square test for significant discrepancies between the predicted and observed covariance matrices, the comparative fit index (CFI) and the root mean square error of approximation (RMSEA) are presented. CFI values over .95 are considered to indicate good fit. RMSEA values less than .08 indicate acceptable fit, whereas 2
Factor scores were examined as an alternative to mean item scores. Five factors were extracted from item-level data using principal axis factoring with direct oblimin rotation (delta ⫽ 0). Correlations among the Big Five, in both single-informant and multi-informant analyses, remained very similar with this method, and interrater agreement was unchanged. An orthogonal rotation (varimax) produced factor scores with slightly higher mean interrater agreement (.44 instead of .41 for the BFI; .37 instead of .36 for the Mini-Markers), but at the cost of explaining less variance in the items and artificially preventing any test of the hypothesis in question (i.e., that the Big Five are correlated). An oblique rotation is the appropriate test for whether the underlying factors are correlated in single-informant ratings. 3 Models fit with three item-packets as indicators for each of the Big Five, in order to create latent trait variables for each informant (thereby adding a lower level of latent variables to the model) produced nearly identical results and are not reported because of their additional complexity.
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
1141
Table 1 Multi-Trait Multi-Method Correlation Matrices for the Big Five Inventory and the Mini-Markers Self-report Informant/factor Self Extraversion Agreeableness Conscientiousness Neuroticism Openness/Intellect Peer 1 Extraversion Agreeableness Conscientiousness Neuroticism Openness/Intellect Peer 2 Extraversion Agreeableness Conscientiousness Neuroticism Openness/Intellect Peer 3 Extraversion Agreeableness Conscientiousness Neuroticism Openness/Intellect Big Five Inventory M SD Alpha Mini-Markers M SD Alpha
E
A
C
Peer 1 report N
O
E
A
— .19 .12 ⴚ.04 .19 .53 .00 .15 — .15 ⴚ.35 .09 .12 .32 .21 .24 — ⴚ.17 .08 .08 ⫺.05 ⴚ.16 ⴚ.36 ⴚ.31 — ⴚ.05 ⫺.03 ⫺.15 .25 .06 .09 ⴚ.08 — .01 ⫺.10 .57 .10 .11 ⫺.01 .38 ⫺.01 .00 .03 .35 .01 ⫺.15 ⫺.07 .12 .03 ⫺.04
C
Peer 2 report
N
O
⫺.01 .07 .07 ⫺.01 ⫺.15 .01 .39 ⫺.10 ⫺.05 ⫺.08 .33 ⫺.07 ⫺.10 .10 .35
⫺.07 .12 — .08 .13 ⴚ.08 .15 ⫺.16 ⫺.08 .14 — .26 ⴚ.51 .23 ⫺.12 ⫺.15 .16 .30 — ⴚ.32 .27 .40 .08 ⴚ.15 ⴚ.54 ⴚ.38 — ⴚ.11 ⫺.07 .44 .28 .25 .18 ⴚ.16 —
.56 .01 .02 ⫺.07 .15 .49 .06 ⫺.01 ⫺.05 ⫺.03 .29 ⫺.08 ⫺.15 ⫺.06 .01 .35 .06 ⫺.22 .00 .00 .33 ⫺.08 ⫺.09 .03 .11 .39 ⫺.16 ⫺.03 ⫺.09 ⫺.07 .37 .01 ⫺.01 ⫺.15 ⫺.12 .38 .15 .01 ⫺.09 .02 .48 .10 .02 ⫺.06 .01 .54 .06 .04 .01 .32 .00 .00 .06 .31 .05 ⫺.10 ⫺.04 .10 .03 ⫺.03
⫺.08 .08 .45 .04 .03 ⫺.18 ⫺.10 .02 .36 .07 ⫺.14 ⫺.09 .05 .17 .35 .42 .08 ⫺.00 ⫺.24 ⫺.12 ⫺.05 .52 .07 .03 ⫺.09
E
A
C
N
Peer 3 report O
E
A
C
N
O
.53 .00 ⫺.02 .05 .05 .51 ⫺.04 ⫺.05 .11 ⫺.01 .05 .30 ⫺.04 ⫺.06 .08 .05 .30 .02 ⫺.08 .10 .01 ⫺.07 .34 .03 ⫺.03 ⫺.03 ⫺.06 .33 .00 ⫺.03 .01 ⫺.18 ⫺.01 .27 ⫺.05 ⫺.02 ⫺.20 ⫺.12 .28 ⫺.06 .09 ⫺.02 ⫺.06 .05 .39 .02 ⫺.08 ⫺.06 .08 .38 .47 ⫺.01 .03 .05 .03 .01 .28 .07 ⫺.11 .06 .01 .02 .43 ⫺.07 .02 .03 ⫺.18 ⫺.16 .31 ⫺.02 .09 .04 .06 ⫺.04 .36
.40 .00 ⫺.04 .06 ⫺.03 .04 .30 .11 ⫺.17 .06 .02 .04 .38 ⫺.05 ⫺.02 .02 ⫺.21 ⫺.16 .31 ⫺.08 .07 ⫺.01 .10 .05 .29
.14 — .07 .14 ⴚ.12 .19 .47 ⫺.03 .01 .05 .06 .00 .12 — .23 ⴚ.55 .24 ⫺.04 .25 .08 ⫺.19 .05 .01 .15 .35 — ⴚ.28 .25 ⫺.06 .09 .41 ⫺.13 .03 .04 ⴚ.24 ⴚ.53 ⴚ.41 — ⴚ.11 .08 ⫺.16 ⫺.11 .29 ⫺.02 .41 .35 .22 .19 ⴚ.18 — .05 .03 .08 ⫺.02 .41
⫺.02 .10 .52 .02 .04 ⫺.24 ⫺.03 .01 .34 .13 ⫺.23 .01 .01 .10 .37 .42 .01 ⫺.04 ⫺.22 ⫺.13 ⫺.05 .41 .13 .05 ⫺.02
⫺.02 .17 — .06 .06 .02 .08 ⫺.22 .02 .16 — .32 ⴚ.59 .26 ⫺.14 ⫺.02 .21 .43 — ⴚ.30 .26 .40 ⫺.04 ⴚ.14 ⴚ.59 ⴚ.44 — ⴚ.07 ⫺.04 .44 .28 .23 .24 ⴚ.19 —
3.34 4.07 4.08 2.56 3.69 3.63 4.11 4.19 2.59 3.72 3.68 4.07 4.19 2.64 3.71 3.69 4.11 4.25 2.57 3.72 0.80 0.57 0.60 0.78 0.70 0.78 0.76 0.67 0.91 0.72 0.82 0.74 0.68 0.88 0.71 0.78 0.80 0.69 0.90 0.71 .87 .80 .84 .85 .85 .86 .87 .86 .87 .86 .86 .87 .85 .87 .85 .83 .90 .86 .88 .84 3.46 4.27 4.06 2.37 3.81 3.66 4.26 4.05 2.49 3.90 3.71 4.21 4.08 2.49 3.91 3.70 4.23 4.13 2.44 3.93 0.78 0.54 0.66 0.72 0.65 0.74 0.71 0.77 0.82 0.65 0.76 0.66 0.74 0.81 0.66 0.76 0.72 0.73 0.82 0.64 .85 .81 .86 .81 .83 .81 .83 .85 .82 .82 .81 .84 .85 .82 .82 .81 .87 .85 .82 .78
Note. E ⫽ Extraversion; A ⫽ Agreeableness; C ⫽ Conscientiousness; N ⫽ Neuroticism; O ⫽ Openness/Intellect. N ⫽ 483 for the Big Five Inventory, and N ⫽ 487 for the Mini-Markers. Correlations greater than 兩.08兩 are significant at p ⬍ .05. Correlations within informants and correlations of the same trait between different informants are in boldface. Correlations of different traits between different informants are in plain text. Big Five Inventory correlations are below the diagonal, and Mini-Markers correlations are above the diagonal.
values less than .05 indicate close fit (Kline, 2005). For both the BFI and the Mini-Markers, the CTCU model (Figure 1A) was clearly the best, being the only model with a nonsignificant or nearly nonsignificant chi-square value, which indicates that the covariance matrix predicted by the model does not differ substantially from the observed matrix. (Because the chi-square value is sensitive to sample size, use of a large sample will often cause even good models to differ significantly from the observed data at the traditional significance level of p ⬍ .05; Kline, 2005.) In addition to having the lowest chi-square, the CTCU model also had the highest CFI values and the lowest RMSEAs.
Correlations Among Latent Traits The overall orthogonality of the Big Five was tested by chisquare difference tests comparing the CTCU and OTCU models. If the Big Five were orthogonal, the fit of these two models should not differ significantly. Orthogonality was rejected for both instru2 ments: for the BFI, difference (10, N ⫽ 483) ⫽ 90.37, p ⬍ .00001; 2 for the Mini-Markers, difference (10, N ⫽ 487) ⫽ 65.78, p ⬍ .00001. Thus, even when the effects of specific informants on
ratings were removed, by creating latent variables representing variance shared across all informants, the Big Five remain significantly intercorrelated. Table 3 presents the parameter estimates for the CTCU model for both the BFI and the Mini-Markers. As predicted, some of the correlations among the Big Five were significant. Also as predicted, the correlations were generally weaker for ratings obtained with the Mini-Markers, a single-adjective-rating instrument, than for ratings obtained with the BFI, a phrase-rating instrument. The average absolute correlation among the Big Five for the BFI was .15, whereas for the Mini-Markers a correlation of .11 was obtained. (Both of these were higher than the average absolute correlation of .09 reported by Biesanz & West, 2004). The pattern of correlations, particularly for the BFI, appears consistent with the standard higher-order factor model (DeYoung et al., 2002; Digman, 1997), as Neuroticism, Agreeableness, and Conscientiousness are correlated, and Extraversion and Openness/Intellect are correlated. A formal test of this model is presented following analysis of the correlations among uniquenesses in the CTCU model.
DEYOUNG
1142
N
N S
N P1
A
N P2
N P3
A S
A P1
E
C
A P2
A P3
C S
C P1
C P2
C P3
E S
E P1
O
E P2
E P3
O S
O P1
O P2
O P3
O P2
O P3
A. Correlated Traits, Correlated Uniquenesses (CTCU)
N
N S
N P1
A
N P2
N P3
A S
A P1
Self
C
A P2
A P3
C S
C P1
E
C P2
Peer1
C P3
E S
E P1
Peer2
O
E P2
E P3
O S
O P1
Peer3
B. Correlated Traits, Orthogonal Methods (CTOM) Figure 1. Two multi-trait, multi-method confirmatory factor-analytic models of Big Five traits (N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect) based on ratings from four informants: S ⫽ self-ratings; P ⫽ peer ratings.
Uniquenesses
Table 2 Model Fit Indices for Multi-Trait Multi-Method Confirmatory Factor Analyses of the Big Five Inventory and the Mini-Markers Model Big Five Inventory (N ⫽ 483) CTNM CTCU OTCU CTCM CTOM OTCM OTOM Mini-Markers (N ⫽ 487) CTNM CTCU OTCU CTCM CTOM OTCM OTOM
2
df
CFI
RMSEAa
1577.04** 136.01 226.38** 244.17** 262.35** 312.53** 363.78**
160 120 130 134 140 144 150
.61 .99 .97 .97 .97 .95 .94
.136 (.130–.142) .017 (.000–.029) .039 (.031–.048) .041 (.033–.049) .043 (.035–.050) .049 (.042–.057) .054 (.047–.062)
1194.59** 149.93* 215.71** 294.41** 318.93** 367.19** 399.48**
160 120 130 134 140 144 150
.63 .99 .97 .94 .94 .92 .91
.115 (.109–.122) .023 (.007–.033) .037 (.028–.045) .050 (.042–.057) .051 (.044–.059) .056 (.049–.064) .059 (.052–.065)
Note. CTNM ⫽ correlated traits, no methods; CTCU ⫽ correlated traits, correlated uniquenesses; OTCU ⫽ orthogonal traits, correlated uniquenesses; CTCM ⫽ correlated traits, correlated methods; CTOM ⫽ correlated traits, orthogonal methods; OTCM ⫽ orthogonal traits, correlated methods; OTOM ⫽ orthogonal traits, orthogonal methods; CFI ⫽ comparative fit index; RMSEA ⫽ root mean square error of approximation. a 90% confidence intervals are presented in parentheses. * p ⬍ .05. ** p ⬍ .001.
As seen in Table 3, correlations among uniquenesses were not only larger in magnitude than correlations among the latent Big Five, they were also larger in magnitude than the same-informant, different-trait correlations shown in boldface in Table 1. This finding indicates that after extracting the variance shared among raters, each individual rater’s leftover variance is fairly consistent, which is to say that, relative to the other raters, he or she consistently rated the target as having more desirable or undesirable qualities across all traits. It is hardly surprising that raters’ general impressions of the desirability or undesirability of targets’ personalities should influence their ratings on all trait dimensions. If the correlations among uniquenesses were due exclusively to such a general bias, however, the method effects associated with individual raters would be unidimensional, and the CTOM model (Figure 1B) would fit as well as the CTCU model (Figure 1A). This was not the case, as chi-square difference tests indicated that the CTCU model was significantly better in fit than the CTOM model: for the 2 BFI, difference (20, N ⫽ 483) ⫽ 126.34, p ⬍ .00001; for the 2 Mini-Markers: difference (20, N ⫽ 487) ⫽ 169.00, p ⬍ .00001. Exploratory factor analyses were therefore conducted to examine the dimensional structure of the correlations among uniquenesses. The results of these analyses may be informative regarding the biases present in individual ratings. If the usual two-factor structure were present in the uniquenesses but not in the latent traits, this would suggest that raters’ biases are inaccurate and probably entirely responsible for the higher-order factor solution reported in the past (DeYoung et al., 2002; Digman, 1997). If, however, the same two-factor structure were found both in the uniquenesses and at the trait level, this would suggest that people
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
1143
Table 3 Parameter Estimates for the Correlated Traits Correlated Uniquenesses Model for the Big Five Inventory and the Mini-Markers Shown in Figure 1A Big Five Inventory (N ⫽ 483) Informant/factor
N
A
C
Mini-Markers (N ⫽ 487)
E
O
N
A
C
E
O
.78** .65** .70** .67**
.67** .54** .61** .59**
— .10†
—
Factor loadings Self-report Peer 1 Peer 2 Peer 3
.69** .61** .57** .60**
.63** .60** .54** .52**
.61** .59** .61** .53**
.79** .69** .71** .69**
.78** .61** .62** .67**
.58** .61** .47** .46**
.64** .55** .49** .46**
.64** .62** .62** .56**
Correlations Latent factor Neuroticism Agreeableness Conscientiousness Extraversion Openness/Intellect Uniquenesses Self-report Neuroticism Agreeableness Conscientiousness Extraversion Openness/Intellect Peer 1 report Neuroticism Agreeableness Conscientiousness Extraversion Openness/Intellect Peer 2 report Neuroticism Agreeableness Conscientiousness Extraversion Openness/Intellect Peer 3 report Neuroticism Agreeableness Conscientiousness Extraversion Openness/Intellect
— ⫺.46** ⫺.29** ⫺.06 .02
— .12† .05 ⫺.04
— ⫺.41** ⫺.40** ⫺.25** ⫺.18**
— .41** .24** .17**
— ⫺.56** ⫺.38** ⫺.23** ⫺.29**
— .35** .18** .41**
— ⫺.54** ⫺.48** ⫺.35** ⫺.29**
— .47** .21** .32**
— ⫺.62** ⫺.48** ⫺.24** ⫺.31**
— .48** .24** .39**
— .05 ⫺.16*
— .34** .34**
— .21** .37**
— .26** .39**
— .30** .45**
— .24**
— .30**
— .33**
— .35**
— .35**
—
— ⫺.49** ⫺.22** .09 ⫺.01
— .02 .07 .06
—
— ⫺.31** ⫺.22** ⫺.11† ⫺.09
— .35** .26** .16*
— .23** .26**
— .31**
—
—
— ⫺.52** ⫺.32** ⫺.17** ⫺.16**
— .35** .07 .33**
— .17** .41**
— .18**
—
—
— ⫺.56** ⫺.32** ⫺.27** ⫺.14**
— .32** .13* .29**
— .25** .40**
— .21**
—
—
— ⫺.59** ⫺.30** ⫺.06 ⫺.09†
— .38** .11* .34**
— .16** .36**
— .12*
—
— .00 ⫺.04
Note. N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect. † p ⬍ .10. * p ⬍ .05. ** p ⬍ .01.
have a reasonably accurate (though possibly implicit) expectation regarding how traits covary and also that this expectation leads them to exaggerate trait correlations when they rate their own or others’ personalities. Finally, if the factor structure of the uniquenesses and the latent traits were dissimilar, this would suggest that people have inaccurate expectations about which traits vary together. Maximum likelihood estimation was used for these exploratory factor analyses because it provides a significance test that can be used to evaluate the number of factors necessary to capture the structure of the data. An oblique rotation (direct oblimin, delta ⫽ 0) was used to allow for the possibility of correlated factors. For the BFI, a two-factor solution fit adequately for three of the four sets of uniquenesses, all 2s(1, N ⫽ 483) ⬍ 2.96, p ⬎ .05. The factor structure for self-rating and peer1 rating uniquenesses was
the same as in the standard two-factor solution: Neuroticism (reversed), Agreeableness, and Conscientiousness marked the first factor, and Extraversion and Openness/Intellect marked the second. For the peer3 ratings, Conscientiousness loaded almost equally on both factors. For the peer2 ratings, the two-factor solution was significantly different from the observed data, 2(1, N ⫽ 483) ⫽ 13.46, p ⬍ .001. Three factors were therefore extracted, with principal axis factoring. The first two factors resembled the standard higher-order factors, whereas the third factor was marked primarily by Openness/Intellect. In all four cases, the first two factors were strongly correlated (rs ⬎ .50). For the Mini-Markers, a two-factor solution fit adequately for two of the four sets of uniquenesses, both 2s(1, N ⫽ 487) ⬍ 1.92, p ⬎ .05. The self-rating uniquenesses showed the standard twofactor solution. In the peer3 ratings, Neuroticism and Agreeable-
DEYOUNG
1144
.07 (-.13) Plasticity
Stability -.99**
. 46** (. 48**)
(-1.00**) N
N S
N P1
.30**
A
N P2
N P3
A S
A P1
.59** (.69**)
(.22**) C
A P2
A P3
C S
C P1
.40** (.13)
E
C P2
C P3
E S
E P1
O
E P2
E P3
O S
O P1
O P2
O P3
**p < .01
Figure 2. Higher-order factors of the Big Five traits (N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect), based on ratings from four informants (S ⫽ self-ratings; P ⫽ peer ratings), with parameter estimates for the Big Five Inventory and the Mini-Markers (estimates for the Mini-Markers are in parentheses). See text for indices of fit and Table 3 for parameter estimates for the measurement model.
ness marked the first factor, Conscientiousness and Openness/ Intellect marked the second, and Extraversion loaded weakly on the second factor and not at all on the first. Because the peer1 and peer2 ratings were not adequately described by a two-factor solution, both 2s(1, N ⫽ 487) ⬎ 6.69, p ⬍ .05, three factors were extracted. In both cases, the first factor was marked by Neuroticism (reversed) and Agreeableness, the second by Openness/Intellect and Conscientiousness, and the third by Neuroticism (reversed) and Extraversion. For all four informants, all factors were moderately intercorrelated, with correlations ranging from .26 to .42.
Higher-Order Factors of the Big Five Confirmatory factor analysis was used to test the hypothesis that the Big Five would show the usual higher-order factor structure, in latent space. Figure 2 depicts a hierarchical model with latent Stability and Plasticity variables above the latent Big Five, with parameter estimates for the higher-order factor solution (see Table 3 for parameter estimates for the measurement model).4 For the BFI, this model fit extremely well, 2(125, N ⫽ 483) ⫽ 145.11, p ⫽ .11; CFI ⫽ .99; RMSEA ⫽ .018. Because this model was not nested under the CTCU model, the two could not be compared with the chi-square difference test; however, Akaike’s information criterion (AIC) can be used to compare non-nested models, with lower AIC values indicating better fit (Kline, 2005). The higherorder factor model did fit slightly better: for CTCU, AIC ⫽ 316.01; for the higher-order factors, AIC ⫽ 315.11. The correlation between Stability and Plasticity was nonsignificant, and the fit of the model did not change significantly when the correlation was constrained to zero, 2(126, N ⫽ 483) ⫽ 145.59, p ⫽ .11; CFI ⫽ 2 .99; RMSEA ⫽ .018; difference (1, N ⫽ 483) ⫽ 0.48, p ⫽ .49. The higher-order factor model also fit reasonably well for the Mini-Markers, 2(126, N ⫽ 487) ⫽ 157.43 p ⬍ .05; CFI ⫽ .99; RMSEA ⫽ .023. (The error variance associated with latent Neuroticism was constrained to be non-negative.5) Again, AIC values indicated that this model was preferable to the CTCU model: for CTCU, AIC ⫽ 329.93; for higher-order factors, AIC ⫽ 325.43. Nonetheless, the model was not entirely unproblematic; Openness/
Intellect did not load significantly on Plasticity because of the fact that the correlation between Extraversion and Openness/Intellect was attenuated for the Mini-Markers relative to the BFI (Table 3). As with the BFI, the correlation between Stability and Plasticity was not significant, and the fit of the model did not change significantly when the correlation was constrained to zero, 2(127, N ⫽ 487) ⫽ 159.20, p ⫽ .03; CFI ⫽ .99; RMSEA ⫽ .023; 2 difference (1, N ⫽ 487) ⫽ 1.77, p ⫽ .18. As a test of whether the pattern of correlations among the latent Big Five was significantly multidimensional, the model in Figure 2 was also fitted with the correlation between Stability and Plasticity fixed at unity (1.00). This strategy created a model that is nested under the standard higher-order factor model but is equivalent to a model with only a single higher-order factor marked by all five latent Big Five traits. The model fit well for both instruments: for BFI, 2(126, N ⫽ 483) ⫽ 165.77, p ⫽ .01; CFI ⫽ .99; RMSEA ⫽ .026; for Mini-Markers, 2(127, N ⫽ 487) ⫽ 171.02, p ⬍ .01; CFI ⫽ .98; RMSEA ⫽ .027. However, chi-square difference tests indicated that it did not fit as well as the two-factor model: for BFI, 2 difference (1, N ⫽ 483) ⫽ 20.66, p ⬍ .001; for Mini-Markers, 4 Because a model containing a latent variable with only two indicators is empirically under-identified if that latent variable is not correlated with another latent variable (Kline, 2005), the unstandardized error variances for the latent Extraversion and Openness/Intellect variables were constrained to be equal, to allow identification of the model. 5 This error variance was constrained because without constraint it became negative at some point while the model was fitted. Although negative error variances have sometimes been considered evidence of possible model misspecification, Monte Carlo studies lead to the conclusion that “researchers should not use negative error variance estimates as an indicator of model misspecification” (Chen, Bollen, Paxton, Curran, & Kirby, 2001, p. 501). In the present analyses, the tendency of the error variance for latent Neuroticism to become slightly negative appears to be due to the fact that the value of the loading of Neuroticism on Stability is very near ⫺1.00 (as evidenced by the weight of ⫺.99 seen in the model for the BFI, in which the error variance for Neuroticism did not need to be constrained). If the estimate of a loading goes over |1.00|, the associated error variance will become negative.
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
1145
.07 Plasticity
Stability . 61**
-1.00** N
N S
B F I
M M
.25**
.55**
A
N P1
N P2
B F I
M M
N P3
B F I
A S
M M
BFI
B F I
A P1
M M
E
C
A P2
A P3
C S
C P1
.40**
C P2
C P3
etc…
E S
E P1
O
E P2
E P3
O S
O P1
O P2
O P3
**p < .01
MiniMarkers
Figure 3. Higher-order factor model for multi-trait, multi-informant, multi-instrument confirmatory factor analysis, based on ratings from four informants (S ⫽ self-ratings; P ⫽ peer ratings). For clarity of illustration, the full measurement model is not shown. See text for indices of fit and Table 4 for additional parameter estimates. N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/ Intellect; BFI ⫽ Big Five Inventory; MM ⫽ Mini-Markers.
2 difference (1, N ⫽ 487) ⫽ 13.59, p ⬍ .001. The pattern of correlations among the latent Big Five, therefore, is not unidimensional, providing further support for the two-factor model.
Multi-Trait, Multi-Informant, Multi-Instrument Confirmatory Factor Analysis For the Mini-Markers, correlations among the latent Big Five were reduced relative to the BFI, as predicted, and the standard Plasticity factor marked by both Extraversion and Openness/Intellect was absent. It was therefore of interest to determine whether both metatraits were present if the BFI and Mini-Markers were analyzed simultaneously. Such an analysis would allow some degree of control for method effects specific to each instrument. A model was therefore fitted in which Big Five ratings by each informant were modeled as latent variables with BFI and MiniMarkers scores as separate indicators (Figure 3). Because the CTCU model fit best in the previous analyses, correlated uniquenesses were used at the latent level of individual informant ratings. Two method factors representing variance unique to the BFI and Mini-Markers were included, with each marked by 20 observed variables. (More complex breakdowns of the instrument effects— by trait, for example—were not possible because, in conjunction with the correlated uniquenesses for each informant, they created unidentified models.) The model fit very well, 2(636, N ⫽ 481) ⫽ 1,208.55, p ⬍ .01; CFI ⫽ .96, RMSEA ⫽ .043.6 (Error variance associated with latent Neuroticism was constrained to be non-negative.) Figure 3 displays the factor loadings on the metatraits, and Table 4 displays the loadings of latent traits for each informant on the latent Big Five and loadings of each instrument on each factor for each informant. The loading of Conscientiousness on Stability was low
but significant, and both Extraversion and Openness/Intellect loaded significantly on Plasticity. These results suggest that the absence of the standard Plasticity factor in the model for the Mini-Markers above was due to method variance specific to that instrument. When the Big Five were modeled by the shared variance across both instruments, both metatraits were evident. Notably, there were no significant loadings of observed variables on the two latent variables representing method effects associated with the different instruments. However, when the model was fitted without the instrument effects factors, the fit of the model was significantly worsened: 2(676, N ⫽ 481) ⫽ 2 1467.32, p ⬍ .01; CFI ⫽ .95, RMSEA ⫽ .049; difference (40, N ⫽ 481) ⫽ 258.77, p ⬍ .00001. The instrument effects factors were therefore retained. Notably, several loadings for the Mini-Markers method factor approached significance ( p ⬍ .10), with loadings ranging between .13 and .29; these included loadings for all four ratings of Openness/Intellect, the three peer ratings of Conscientiousness, and two of the peer ratings of Agreeableness. A pattern suggesting stronger method effects for the Mini-Markers than for the BFI is consistent with the differences in higher-order factor structures for the two instruments when analyzed separately.
Correlations Between Stability and Plasticity As seen in the models above, once the variance associated with specific informants was removed, Stability and Plasticity were uncorrelated in latent space. Previous work with self-ratings has found substantial correlations between the metatraits (DeYoung et 6
The full covariance matrix used to fit this model is available from the author on request.
DEYOUNG
1146 Table 4 Latent Big Five Factor Loadings for the Multi-Trait, Multi-Instrument, Multi-Informant Model Shown in Figure 3 Latent factors Informant/measure
N
A
C
E
O
.66 .66 .66 .56
.81 .72 .75 .70
.79 .65 .65 .69
Multi-informant Big Five Self-report Peer 1 Peer 2 Peer 3
.69 .63 .54 .59
.67 .62 .55 .54
Single-informant Big Five Self-report Big Five Inventory Mini-Markers Peer 1 report Big Five Inventory Mini-Markers Peer 2 report Big Five Inventory Mini-Markers Peer 3 report Big Five Inventory Mini-Markers
.83 .90
.96 .81
.96 .87
.97 .93
.98 .83
.86 .90
.97 .85
.90 .91
.97 .91
.95 .79
.85 .89
.97 .86
.95 .88
.96 .93
.95 .83
.86 .88
.97 .88
.98 .86
.99 .90
.99 .78
Note. N ⫽ 481. N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect. All loadings were significant at p ⬍ .01. The loadings in this table correspond to the variables in Figure 3 as follows: multi-informant Big Five loadings are those of the latent (oval) Big Five variables Neuroticism self-report (NS), Neuroticism Peer 1 (NP1), Neuroticism Peer 2 (NP2), and so forth, on the latent Big Five variables Neuroticism (N), Agreeableness (A), Conscientiousness (C), and so forth, whereas single-informant Big Five loadings are those of the observed (rectangular) Big Five variables, labeled BFI (Big Five Inventory) and MM (Mini-Markers), on the latent variables NS, NP1, NP2, and so forth.
al., 2002), but this may have been the result of generally inflated correlations within individual informants’ ratings as a result of a bias toward describing oneself as having uniformly desirable or undesirable traits and/or to more specific biases regarding how traits are assumed to covary. The present study allowed further examination of this issue, by fitting CFA models of the higherorder factors with ratings from single informants (Figure 4) and comparing the resulting correlations between Stability and Plasticity with those obtained in the MTMM analysis above. Significant correlations would suggest that the biases of individual raters are responsible for correlations between the metatraits. As shown in Table 5, Stability and Plasticity were significantly correlated in ratings by all four sets of informants, with either instrument. Table 5 shows the parameter estimates and fit indices obtained for the model depicted in Figure 4 for each set of informants. The model fit very well for the BFI and for self-ratings on the MiniMarkers. The three sets of peer ratings on the Mini-Markers did not show adequate model fit. In these three cases, however, the fit 2 of the model could be significantly improved, all difference s(1, N ⫽ 487) ⬎ 20.90, all ps ⬍ .001, by allowing the uniquenesses for Neuroticism and Openness/Intellect to correlate (dotted line in Figure 4). Fit indices for all three models were then similar to those obtained for the models of peer ratings on the BFI. The correla-
tions between the Neuroticism and Openness/Intellect uniquenesses were all significant (rs ranged from .54 to .75, all ps ⬍ .001). This finding is consistent with our previous findings with the TDA, in which the fit of the higher-order factor model was improved by allowing these two uniquenesses to correlate (DeYoung et al., 2002). One possible reason for this correlation is that in some instruments the positive correlation between Stability and Plasticity may cause the higher-order factor model to overestimate the negative correlation between Neuroticism and Openness/Intellect (thus allowing a positive correlation between their uniquenesses to improve the fit of the model). Another possible reason is that in both the TDA and Mini-Markers, Openness/Intellect and Neuroticism are the only scales that do not have balanced keying (specifically, there are more positively than negatively keyed items). This situation could cause them to covary as a result of acquiescence bias.
Discussion When modeled as latent variables defined by the ratings of four different informants, the Big Five were significantly intercorrelated. Correlations among the Big Five, therefore, cannot be dismissed as artifacts of the biases of individual raters. Because the Big Five were not completely orthogonal in latent space for either the BFI or the Mini-Markers, the question of higher-order factor structure remains relevant. A hierarchical model with latent Stability and Plasticity variables above the latent Big Five fit the data very well for the BFI. The model did not show the standard Plasticity factor for the Mini-Markers, in that Openness/Intellect did not load significantly on it, but this seems likely to be the result of attenuated correlations due to lower interrater agreement. Despite some attenuation, the pattern of correlations among the latent Big Five for the Mini-Markers was similar to that for the BFI, and a model combining the Mini-Markers and the BFI, as indicators of latent Big Five variables for each informant, demonstrated both metatraits. The higher-order factors thus do not appear to be a method artifact specific to the BFI (not surprisingly, given that
Stability
N
A
Plasticity
C
E
O
Figure 4. Higher-order factor model for single-informant ratings of the Big Five traits (N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect). The dotted line represents the correlation between uniquenesses for Neuroticism and Openness/Intellect. Freeing this parameter significantly improved the fit of the model for the three sets of peer ratings on the Mini-Markers.
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
1147
Table 5 Standardized Parameters and Fit Indices for Higher-Order Factor Models of Single-Informant Ratings (Figure 4) Factor loadings Measure/informant Big Five Inventory (N ⫽ 483) Self Peer 1 Peer 2 Peer 3 Mini-Markers (N ⫽ 487) Self Peer 1 Peer 2 Peer 3
2(4)
CFI
RMSEAb
.39** .46** .46** .49**
4.68 11.69* 10.89* 9.61*
.99 .98 .98 .99
.019 (.000–.074) .063 (.023–.107) .060 (.018–.104) .054 (.005–.099)
.41** .50** .45** .41**
10.37* 30.02** 33.86** 35.39**
.95 .90 .90 .91
.057 (.014–.101) .116 (.079–.156) .124 (.087–.164) .127 (.091–.167)
N
A
C
E
O
ra
⫺.64** ⫺.76** ⫺.78** ⫺.76**
.53** .70** .67** .77**
.49** .47** .53** .58**
.76** .46** .57** .46**
.33** .62** .60** .60**
⫺.49** ⫺.69** ⫺.72** ⫺.67**
.70** .79** .75** .87**
.26** .44** .37** .40**
.58** .26** .31** .12
.33** .57** .61** .67*
Note. N ⫽ Neuroticism; A ⫽ Agreeableness; C ⫽ Conscientiousness; E ⫽ Extraversion; O ⫽ Openness/Intellect. CFI ⫽ comparative fit index; RMSEA ⫽ root mean square error of approximation. a Correlation between stability and plasticity. b 90% confidence intervals are presented in parentheses. * p ⬍ .05. ** p ⬍ .001.
almost none of the many previous data sets in which the metatraits were found have used the BFI). The latent metatraits in the multi-informant models were uncorrelated, in contrast to models fit for single-informant ratings, in which the metatraits were fairly strongly correlated, as in previous studies (DeYoung et al., 2002). Thus, whereas the higher-order factor structure does not appear to be an artifact of the biases of individual raters, the correlation between the metatraits may be artifactual. These findings have implications for research utilizing single-informant ratings: Correlation between the metatraits may suppress associations with other variables, when the metatraits predict in opposite directions. We found in a previous study, for example, that Stability predicted conformity positively, whereas Plasticity predicted it negatively; however, the association with Plasticity did not appear in zero-order correlations and only became evident when controlling statistically for Stability (DeYoung et al., 2002). Although in the present study the Big Five were correlated and showed the expected higher-order factor structure in latent space, the magnitude of correlations and loadings on the higher-order factors was generally lower than in ratings by single informants. This suggests that individual informants do inflate the correlations among the Big Five, perhaps in part because of a bias toward rating targets’ personalities as uniformly desirable or undesirable. However, such a general bias cannot be the only factor leading to the inflation of correlations, because statistical comparisons of the CTCU and CTOM models indicated that the method effects associated with specific informants were not unidimensional for either the BFI or the Mini-Markers. Exploratory factor analyses of the correlations among uniquenesses were therefore carried out to investigate their factor structure. The uniquenesses represent variance specific to each informant after shared variance has been removed—in other words, the portion of the rating not agreed upon by all four raters. Correlations among the uniquenesses therefore indicate how individual ratings of the Big Five correlate, above and beyond the actual correlations of the traits in latent space. The correlations among
uniquenesses did not show an entirely consistent factor structure. However, for all four ratings of the BFI and for self-ratings of the Mini-Markers, their factor structure was similar to the standard higher-order factors, in which Neuroticism (reversed), Agreeableness, and Conscientiousness mark the first factor and Extraversion and Openness/Intellect mark the second. It appears, therefore, that the biases associated with individual raters generally conform to the same factor structure that is present in the Big Five at the latent level, but that this is more true for the BFI than for the MiniMarkers, perhaps because ratings on the latter are less consistent, as indicated by lower interrater agreement coefficients. This finding suggests that people’s expectations about which personality traits should vary together are reasonably accurate (in that their individual biases tend to show the same factor structure as the latent traits) but lead them to attribute more covariation than actually exists, producing inflated correlations and higher-order factor loadings, in single-informant ratings. One must also consider the possibility that the uniquenesses contain some genuine variance in addition to bias. The latent Big Five variables represent only the variance that is shared among all informants. It is certainly possible that any particular informant may accurately detect some aspect of the target’s personality that other informants have overlooked. Such disparities are especially likely when comparing self- and other ratings. Individuals may know things about themselves, through introspection, that others do not. Similarly, access to a more objective view of an individual’s behavior may lead others to notice (or report) regularities that the individual does not. This hypothesis is supported by the finding that self- and other ratings yield incremental validity in predicting important criterion variables, such as job performance (Mount, Barrick, & Strauss, 1994). The true magnitudes of the correlations among the Big Five, therefore, probably fall somewhere between those seen in single-informant ratings and those seen in the shared variance of ratings by multiple informants. Correlations in singleinformant ratings are presumably higher than they should be, because of various biases, but correlations among latent variables derived from multiple-informant ratings may be lower than they
1148
DEYOUNG
should be, because of exclusion (from the latent variables) of genuine variance not detected by all informants. The pattern of correlations among the Big Five might be affected by at least three other factors, which could be examined in future research. First, stronger correlations in latent space might be detected with multiple-informant ratings on an instrument prone to even higher levels of interrater agreement than the BFI, such as the NEO PI-R (Costa & McCrae, 1992a). Second, the breadth of assessment within each Big Five domain might also influence correlations, and here again the NEO PI-R would be useful, as it assesses a wider range of content within each domain than either the BFI or the Mini-Markers. Third, the fact that the vast majority of peer raters in this study felt very positively about their targets could have had some effect on the correlational structure of ratings, and it would be informative to attempt a replication in a more evaluatively heterogeneous sample.
Comparison of Instruments The average magnitude of the correlations and the number of significant correlations among latent Big Five variables were significantly greater for the BFI than for the Mini-Markers. This seems likely to be due to the greater interrater agreement associated with the BFI and may explain a previous failure to detect significant correlations among the Big Five in a similar multiinformant analysis (Biesanz & West, 2004). Biesanz and West (2004) found no significant correlations among latent Big Five variables in a data set comprising self-, peer, and parent ratings. However, the interrater agreement obtained in that study was lower than that obtained for either instrument in the present study. Some of these differences in interrater agreement may be due to choice of instrument. Biesanz and West (2004) used a singleadjective rating instrument (the TDA) containing a number of difficult and unfamiliar adjectives (precisely those adjectives that were removed in the construction of the Mini-Markers; Saucier, 1994), which is likely to reduce the consistency of interpretation of items. Of course, other factors related to the participants or the relationships between raters and targets may have contributed to low interrater agreement in their sample. Nonetheless, the present study demonstrated, within one sample, that a single-adjective rating scale (the Mini-Markers) had lower interrater agreement than a scale embedding trait-descriptive adjectives in longer phrases (the BFI) and that interrater agreement was associated with the strength of correlations among the Big Five. At the very least, the results of the current study suggest that one should be attentive to interrater agreement when using multi-informant ratings as indicators of latent variables. At least two strategies could be used in future research, to strengthen the conclusions of this study regarding the different properties of different instruments. First, as mentioned above, it would be ideal to replicate the current findings with another instrument possessing relatively high interrater agreement, such as the NEO PI-R. Second, examining two samples showing different levels of interrater agreement despite using the same instrument would be of interest. Thus far, three instruments have been used to conduct multi-informant analyses of the Big Five, two in the present study and one by Biesanz and West (2004). Across these analyses, interrater agreement has been perfectly correlated with the average absolute correlation among the Big Five at the latent
level, but each analysis has used a different instrument, thereby confounding the effect of interrater agreement with the effect of instrument. In other words, because neither instrument nor level of agreement among raters has been held constant across analyses, one cannot assert with confidence which of these two factors is responsible for differences in the magnitude of correlations. Finally, one should consider two additional factors that might contribute to the finding that correlations among the Big Five as assessed by the Mini-Markers were lower than those as assessed by the BFI. First, the Mini-Markers were intentionally designed to produce relatively weak interscale correlations in single-informant ratings (Saucier, 1994). Second, differences in item content between the two instruments might affect Big Five intercorrelations. One salient example is related to the attenuated correlation of Extraversion and Openness/Intellect seen in the Mini-Markers, which precluded the loading of Openness/Intellect on Plasticity. In the Openness/Intellect scale, the Mini-Markers contain more items emphasizing intellectuality than does the BFI, and the BFI includes items related to curiosity and dislike of routine, whereas the Mini-Markers do not. Curiosity and dislike of routine seem likely to be more strongly related to Extraversion than is intellectuality, and their inclusion might make the BFI’s Openness/Intellect scale a better indicator of Plasticity than the comparable scale of the Mini-Markers.
The Meaning of the Metatraits Having provided evidence that correlations among the Big Five are real and appear to possess the higher-order factor structure first reported by Digman (1997), we now return to questions of interpretation and explanation of the higher-order factors. The present findings suggest that although some of the variance in the metatraits in single-informant ratings is a method artifact stemming from the biases of individual raters, enough of it is genuine that the existence of the metatraits must be taken seriously. At least two additional reasons exist to consider the metatraits important: First, similar higher-order factors have been found in factor analyses combining various measures of normal and abnormal personality traits in conjunction with standard Big Five instruments (Markon et al., 2005). Markon et al.’s analyses indicate that (a) traits considered pathological can be located within the same hierarchy of classification as normal traits and (b) two broad classes of psychopathology (internalizing problems such as depression and anxiety and externalizing problems such as aggression and impulsivity) are associated with low Stability. Second, factors bearing an obvious resemblance to Stability and Plasticity appear in lexical studies when only two factors are extracted (Saucier, 2003; Saucier, Georgiades, Tsaousis, & Goldberg, 2005). These two lexical factors, often labeled Social Propriety and Dynamism, show greater cross-language replicability than do the Big Five (Saucier et al., 2005). A higher-order factoring approach starting with the Big Five thus appears not to be the only method for observing that personality descriptors cluster into two very broad domains. The exact degree to which the Social Propriety and Dynamism factors are similar to the Stability and Plasticity factors is a question for future research. The methods used to discover the two sets of constructs are different enough that one cannot yet judge whether their different labels reflect genuinely different content.
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE
We chose the labels Stability and Plasticity to replace Digman’s (1997) “provisional” (p. 1248) labels, ␣ and , because they seem to be good descriptors of the very general patterns of behavior and experience indicated by the shared variance of Neuroticism (reversed), Agreeableness, and Conscientiousness, on the one hand, and Extraversion and Openness/Intellect, on the other (DeYoung et al., 2002). We have noted elsewhere that this interpretation seems compatible with Digman’s (1997) suggestion that ␣ and  might be associated with socialization and personal growth, respectively. Stability seems likely to make a child easier to socialize (and socialization may encourage Stability), whereas Plasticity seems likely (though not inevitably) to lead to personal growth (DeYoung et al., 2002, 2005). Socialization and personal growth, however, seem more like outcomes than predispositions, whereas “Stability” and “Plasticity” suggest more basic tendencies. We have argued that Stability and Plasticity might be related to two fundamental human concerns (DeYoung et al., 2005): (a) the need to maintain a stable organization of psychosocial function and (b) the need to explore and incorporate novel information into that organization, as the state of the individual changes both internally (developmentally) and externally (environmentally). On this interpretation, some of the variance in the Big Five represents individual differences in emphasis on and competence in meeting these two needs: An absence of Neuroticism reflects emotional stability. Agreeableness reflects the tendency to maintain stability in social relationships (cf. Graziano & Eisenberg, 1997). Conscientiousness appears to reflect motivational stability, the tendency to set goals and work toward them in a reliable and organized manner. Extraversion reflects sensitivity to the possibility of reward (Depue & Collins, 1999; Lucas, Diener, Grob, Suh, & Shao, 2000), producing the tendency to explore the world through action (of course, much of the human world is social, and speech is a form of action). Openness/Intellect reflects the tendency to explore the world perceptually and cognitively (DeYoung et al., 2005). Consistent with our interpretation of Plasticity, both Extraversion and Openness/Intellect are positively related to sensation seeking (Aluja, Garcia, & Garcia, 2003). The present findings suggest that “Stability” is a good label in part because of its similarity to “emotional stability,” the standard label for the negative pole of Neuroticism. In all three multiinformant models that included the metatraits, the loading of Neuroticism on Stability was approximately ⫺1.00, with weaker loadings for Agreeableness and Conscientiousness. Studies using only single-informant ratings have found loadings for Neuroticism to be lower (⬃.60) and similar to loadings for Agreeableness and Conscientiousness (DeYoung et al., 2002; Digman, 1997). The present study reveals that unless the shared variance across informants in this study seriously underestimates the correlation between Agreeableness and Conscientiousness, emotional stability appears to be the primary and dominant component of Stability.7 Nonetheless, Stability is conceptually broader than low Neuroticism because it encompasses those aspects of Agreeableness and Conscientiousness that vary with Neuroticism. The term Plasticity, to denote a broad tendency toward exploration, provides a good complement to Stability, especially with reference to information-processing theory (DeYoung et al., 2002). On the basis of his work with neural network models, Grossberg (1987) used these terms to describe two partially independent subsystems that he argued were necessary for any complex
1149
information-processing system to function well over time in a changing environment: a stability subsystem responsible for maintaining the stability of classification and output and a plasticity subsystem responsible for handling novel information and adjusting categories. The needs met by the functions of these two subsystems seem strongly analogous to the needs described above as the conceptual basis for the traits of Stability and Plasticity. For any interpretation of the metatraits, an important question is how they might be instantiated biologically. Numerous studies have demonstrated that the Big Five show substantial heritability, with at least 40 –50%, and perhaps as much as 80%, of their variance stemming from genetic sources (Bouchard, 1994; Loehlin, 1992; Reimann, Angleitner, & Strelau, 1997). Ample evidence indicates that environmental forces also influence the Big Five over the life span (Roberts, Wood, & Smith, 2005), but environmental forces that affect personality may do so by affecting brain systems, and the question of how traits are instantiated biologically is therefore partially distinct from the question of whether their distal sources are genetic or environmental (DeYoung et al., 2005). Though nonbiological forces may be partially responsible for trait correlations, patterns of covariance among traits are nonetheless useful as clues to the neurobiological underpinnings of personality. The existence of the metatraits suggests that their constituent Big Five traits may share some aspects of their biological substrates. In previous work (DeYoung et al., 2002), we reviewed evidence supporting the hypotheses that Stability reflects individual differences in the functioning of the serotonergic system, which regulates the stability of emotion and behavior (Spoont, 1992; Zald & Depue, 2001) and that Plasticity reflects individual differences in the functioning of the dopaminergic system, which governs exploratory behavior and cognitive flexibility (Ashby, Isen, & Turken, 1999; Braver & Barch, 2002; Depue & Collins, 1999; Panksepp, 1998; Peterson, Smith, & Carson, 2002). Note that this model is not intended to imply that two neurotransmitters might constitute the entire neurobiological substrate of a trait hierarchy based on the Big Five. (We have presented a model of Openness/Intellect, for example, linking it not only to dopaminergic function but also to the functions of the dorsolateral prefrontal cortex; DeYoung et al., 2005). Undoubtedly, many other neurobiological systems are involved in personality (Zuckerman, 2005). What the model does imply, however, is that individual differences in serotonergic and dopaminergic function are likely to be at least partially responsible for the pattern of correlations among the Big Five. Serotonin and dopamine act very broadly in the brain as neuromodulators, and their effects on personality might therefore be expected to be evident at a level of organization as broad as the metatraits. That Neuroticism and Agreeableness showed the strongest correlation among the latent Big Five, for both the BFI and the 7
The importance of emotional stability within Stability suggests a possible link to Tellegen’s higher-order factor Negative Emotionality as does the strong negative correlation between Neuroticism and Agreeableness at the latent level, given the association of Negative Emotionality with aggression (Tellegen & Waller, 1994). Digman (1997) noted this parallel as well. Future studies should empirically compare Stability and Plasticity with higher-order factors from personality models not derived from the lexical tradition (cf. Markon et al., 2005).
DEYOUNG
1150
Mini-Markers (Table 3), deserves particular attention because molecular genetic evidence indicates that covariance between these two traits, in self-reports, is partially mediated by a specific individual difference in the serotonergic system. Jang et al. (2001) found that the correlation between Neuroticism and Agreeableness was genetically based and that variation in the serotonin transporter gene accounted for 10% of the genetic correlation. This kind of genetic investigation, or investigations utilizing pharmacological manipulations, will be necessary to test our biological model of the metatraits directly. For now, the model remains a plausible hypothesis, synthesizing existing findings and suggesting avenues for further research. Notably, Markon et al.’s (2005) finding of association between Stability and both internalizing and externalizing problems is consistent with our biological model, given that both internalizing and externalizing problems are associated with low levels of serotonin (Spoont, 1992). Our model may therefore aid in linking the biological substrates of normal and abnormal variation in personality.
Conclusion The multi-informant analysis presented here provides evidence that correlations among the Big Five, in two commonly used instruments, are not due to raters’ biases. Other instruments seem likely to replicate this pattern, given adequate interrater agreement. One likely exception that should be noted is the orthogonal marker sets developed by Saucier (2002), which yield orthogonal Big Five scores in single-informant ratings. Presumably, these instruments would also yield orthogonal latent traits in a multi-informant analysis like the present one. However, the fact that it is possible to create orthogonal Big Five instruments, through careful item selection, does not necessarily entail that such instruments are desirable. If the Big Five are truly correlated trait domains, then orthogonal marker sets may misrepresent their content. The traditional conception of the Big Five as orthogonal is partly a historical accident resulting from the methods used in their discovery. Had more of the factor analyses that originally validated the Big Five model been performed with oblique rotations that allow correlations among factors, instead of orthogonal rotations that artificially prevent any correlations among factors (at the expense of explaining less variance), we might never have had to debate the reality of correlations among the Big Five. Rather than attempting to eliminate these correlations through orthogonal rotation or techniques of scale construction, one might consider instead whether these correlations, and the higher-order factors they reveal, have substantive meaning. Our model of Stability and Plasticity offers one interpretation of their meaning and provides a hypothesis regarding their biological sources that may aid in the development of neurobiological theories of personality based on the Big Five. Note that this model does not imply that the metatraits should supplant the Big Five as the most important level of trait organization. In the multi-informant models reported here, neither the correlations among the latent Big Five nor most of the higher-order factor loadings were particularly strong. We have noted elsewhere that Extraversion and Openness/ Intellect are probably more different than similar (DeYoung et al., 2005), and the same could be said for Neuroticism, Agreeableness, and Conscientiousness. What is unique to each Big Five trait needs explaining just as much as what is shared. To develop either sort
of explanation, however, one must differentiate what is shared from what is unique, and this can be accomplished only if correlations and higher-order factors are acknowledged and taken into account. The current study suggests strongly that correlations among the Big Five are substantively real and possess a meaningful higher-order structure.
References Aluja, A., Garcia, O., & Garcia, L. F. (2003). Relationships among Extraversion, Openness to Experience, and sensation seeking. Personality and Individual Differences, 35, 671– 680. Arbuckle, J. L. (2003). Amos 5.0 (Build 5138). Chicago: SmallWaters. Ashby, F. G., Isen, A. M., & Turken, A. U. (1999). A neuropsychological theory of positive affect and its influence on cognition. Psychological Review, 106, 529 –550. Ashton, M. C., Lee. K., Perugini, M., Szarota, P., de Vries, R. E., Blas, L. D., Boies, K., & De Raad, B. (2004). A six-factor structure of personality descriptive adjectives: Solutions from psycholexical studies in seven languages. Journal of Personality and Social Psychology, 86, 356 –366. Biesanz, J. C., & West, S. G. (2004). Towards understanding assessments of the Big Five: Multitrait–multimethod analyses of convergent and discriminant validity across measurement occasion and type of observer. Journal of Personality, 72, 845– 876. Bouchard, T. J. (1994). Genes, environment, and personality, Science, 264, 1700 –1701. Braver, T. S., & Barch, D. M. (2002). A theory of cognitive control, aging cognition, and neuromodulation. Neuroscience & Biobehavioral Reviews, 26, 809 – 817. Chen, F., Bollen, K. A., Paxton, P., Curran, P. J., & Kirby, J. B. (2001). Improper solutions in structural equation models. Sociological Methods and Research, 29, 468 –508. Costa, P. T., Jr., & McCrae, R. R. (1992a). Four ways five factors are basic. Personality and Individual Differences, 13, 653– 665. Costa, P. T., Jr., & McCrae, R. R. (1992b). Reply to Eysenck. Personality and Individual Differences, 13, 861– 865. Depue, R. A., & Collins, P. F. (1999). Neurobiology of the structure of personality: Dopamine, facilitation of incentive motivation, and extraversion. Behavioral and Brain Sciences, 22, 491–569. DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2002). Higher-order factors of the Big Five predict conformity: Are there neuroses of health? Personality and Individual Differences, 33, 533–552. DeYoung, C. G., Peterson, J. B., & Higgins, D. M. (2005). Sources of Openness/Intellect: Cognitive and neuropsychological correlates of the fifth factor of personality. Journal of Personality, 73, 825– 858. Digman, J. M. (1997). Higher-order factors of the Big Five. Journal of Personality and Social Psychology, 73, 1246 –1256. Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure. Psychological Assessment, 4, 26 – 42. Goldberg, L. R. (1993). The structure of phenotypic personality traits. American Psychologist, 48, 26 –34. Goldberg, L. R., & Kilkowski, J. M. (1985). The prediction of semantic consistency in self-descriptions: Characteristics of persons and of terms that affect the consistency of responses to synonym and antonym pairs. Journal of Personality and Social Psychology, 48, 82–98. Graziano, W. G., & Eisenberg, N. (1997). Agreeableness: A dimension of personality. In R. Hogan, J. Johnson, & S. Briggs (Eds.), Handbook of personality psychology (pp. 767–793). San Diego, CA: Academic Press. Grossberg, S. (1987). Competitive learning: From interactive activation to adaptive resonance. Cognitive Science, 11, 23– 63. Jang, K. L., Hu, S., Livesley, W. J., Angleitner, A., Reimann, R., Ando, J., Ono, Y., Vernon, P. A., & Hamer, D. J. (2001). Covariance structure of Neuroticism and Agreeableness: A twin and molecular genetic analysis
HIGHER-ORDER FACTORS IN A MULTI-INFORMANT SAMPLE of the role of the serotonin transporter gene. Journal of Personality and Social Psychology, 81, 295–304. John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. In L. A. Pervin & O. P. John (Eds). Handbook of personality: Theory and research (pp. 102–138). New York: Guilford Press. Kline, R. B. (2005). Principles and practice of structural equation modeling. New York: Guilford Press. Loehlin, J. C. (1992). Genes and environment in personality development. Newbury Park, CA: Sage. Lucas, R. E., Diener, E., Grob, A., Suh, E. M., & Shao, L. (2000). Cross-cultural evidence for the fundamental features of extraversion. Journal of Personality and Social Psychology, 79, 452– 468. Markon, K. E., Krueger, R. F., & Watson, D. (2005). Delineating the structure of normal and abnormal personality: An integrative hierarchical approach. Journal of Personality and Social Psychology, 88, 139 – 157. McCrae, R. R., & Costa, P. T. (1995). Positive and negative valence within the five-factor model. Journal of Research in Personality, 29, 443– 460. McCrae, R. R., & Costa, P. T. (1999). A five-factor theory of personality. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 102–138). New York: Guilford Press. Mount, M. K., Barrick, M. R., & Strauss, J. P. (1994). Validity of observer ratings of the Big Five personality factors. Journal of Applied Psychology, 79, 272–280. Mutch, C. (2005). Higher-order factors of the Big Five model of personality: A reanalysis of Digman (1997). Psychological Reports, 96, 167– 177. Panksepp, J. (1998). Affective neuroscience: The foundations of human and animal emotions. New York: Oxford University Press. Peterson, J. B., Smith, K. W., & Carson, S. (2002). Openness and Extraversion are associated with reduced latent inhibition: Replication and commentary. Personality and Individual Differences, 33, 1137–1147. Reimann, R., Angleitner, A., & Strelau, J. (1997). Genetic and environmental influences on personality: A study of twins reared together using the self- and peer report NEO–FFI scales. Journal of Personality, 65, 449 – 476.
1151
Roberts, B. W., Wood, D., & Smith, J. (2005). Evaluating Five Factor theory and social investment perspectives on personality trait development. Journal of Research in Personality, 39, 166 –184. Saucier, G. (1994). Mini-Markers: A brief version of Goldberg’s unipolar Big-Five markers. Journal of Personality Assessment, 63, 506 –516. Saucier, G. (2002). Orthogonal markers of orthogonal factors: The case of the Big Five. Journal of Research in Personality, 36, 1–31. Saucier, G. (2003). Factor structure of English-language personality typenouns. Journal of Personality and Social Psychology, 85, 695–708. Saucier, G., Georgiades, S., Tsaousis, I., & Goldberg, L. R. (2005). The factor structure of Greek personality adjectives. Journal of Personality and Social Psychology, 88, 856 – 875. Saucier, G., & Goldberg, L. R. (2001). Lexical studies of indigenous personality factors: Premises, products, and prospects. Journal of Personality, 69, 847– 879. Spoont, M. R. (1992). Modulatory role of serotonin in neural information processing: Implications for human psychopathology. Psychological Bulletin, 112, 330 –350. Tellegen, A., & Waller, N. (1994). Exploring personality through test construction: Development of the Multidimensional Personality Questionnaire. In S. R. Briggs & J. M. Cheek (Eds.), Personality measures: Development and evaluation (Vol. 1, pp. 133–161). Greenwich, CT: JAI Press. Yik, M. S. M., & Russell, J. A. (2001). Predicting the big two of affect from the Big Five of personality. Journal of Research in Personality, 35, 247–277. Zald, D. H., & Depue, R. A. (2001). Serotonergic functioning correlates with positive and negative affect in psychiatrically healthy males. Personality and Individual Differences, 30, 71– 86. Zuckerman, M. (2005). Psychobiology of personality, second edition. New York: Cambridge University Press.
Received June 14, 2005 Revision received March 21, 2006 Accepted June 16, 2006 䡲
Journal of Personality and Social Psychology 2006, Vol. 91, No. 6, 1152–1165
Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.1152
Love, Work, and Changes in Extraversion and Neuroticism Over Time Christie Napa Scollon
Ed Diener
Texas Christian University
University of Illinois at Urbana–Champaign and The Gallup Organization
The present study examined individual differences in change in extraversion, neuroticism, and work and relationship satisfaction. Of particular interest were the correlations between changes. Data were from the Victorian Quality of Life Panel Study (B. Headey & A. Wearing, 1989, 1992), in which an overall 1,130 individuals participated (ages 16 to 70). Respondents were assessed every 2 years from 1981 to 1989. Four major findings emerged. (a) There were significant individual differences in changes in extraversion and neuroticism. (b) Change was not limited to young adulthood. (c) Development was systematic in that increased work and relationship satisfaction was associated with decreases in neuroticism and increases in extraversion over time; on average, the magnitude of the relation between changes in work and relationship satisfaction and traits was .40. (d) Cross-lagged models indicated traits had a greater influence on role satisfaction; however, marginal support emerged for work satisfaction leading to increased extraversion. Implications of correlated change are discussed. Keywords: traits, personality development, longitudinal, life-span development, well-being
In recent years, several cross-sectional and longitudinal studies have converged on a general picture of personality development in which neuroticism decreases and agreeableness and conscientiousness increase with age (McCrae et al., 1999; Srivastava, John, Gosling, & Potter, 2003). Neuroticism, in particular, decreases with striking consistency with each year of life. In a recent metaanalysis of longitudinal studies of personality, Roberts, Walton, and Viechtbauer (2006) found evidence for change on every dimension of the Big Five, even well into old age. A major goal that remains for personality research is to account for the conditions under which these trait changes occur or are strongest. Are changes in personality traits associated with life experiences? If so, do environmental influences have an impact on trait development only in young adulthood? We sought to address these questions by investigating individual differences in change, or intraindividual change, in extraversion and neuroticism. Specifically, we examined correlated change or the degree to which traits and other variables changed together over time. By focusing on individual differences in change, we hoped to gain a deeper understanding of the relation between important social domains and personality development in adulthood.
Social Roles and Personality Development Dynamic transactional perspectives highlight the codevelopment of the individual and his or her social relationships. For example, Neyer and Asendorpf (2001) found that neuroticism predicted feeling insecure in one’s relationships, but the formation of a romantic partnership also led to decreases in neuroticism over time. More recently, Roberts, Wood, and Smith (2005) have elaborated on transactional views with the social investment model. According to social investment principles, individuals make commitments to important social institutions or roles such as work or marriage. Successful fulfillment of these roles often demands certain behaviors and characteristics, for example, increased emotional stability, agreeableness, and conscientiousness. By committing to and succeeding in these important social roles, over time, the person comes to assume the qualities that the roles promote. Thus, according to a transactional view of development, as role quality increases, individuals should exhibit increases in the corresponding traits that the role promotes.
Important Social Roles: Work and Love As Freud famously noted, the two most important roles in adult life are work and love. Not surprisingly then, a number of tests of transactional development have focused on these two major roles with evidence to suggest that happy work and close relationships may lead to long-term increases in psychological well-being. Roberts, Caspi, and Moffitt (2003), for instance, found that individuals who obtained higher status occupations increased in well-being and agency over time. For women in the 1960s, paid participation in the work force and occupational successes predicted increased agency (Roberts, 1997) and dominance (Roberts, Helson, & Klohnen, 2002) some 20 years later. Subjective aspects of work also predict well-being in that satisfying and engaging employment predicts increases in positive emotion and decreases in negative emotion (Roberts et al., 2003; Roberts & Chapman, 2000).
Christie Napa Scollon, Department of Psychology, Texas Christian University; Ed Diener, Department of Psychology, University of Illinois at Urbana–Champaign and The Gallup Organization, Washington, DC. This article is based in part on a dissertation by Christie Napa Scollon to fulfill the requirements of a doctoral degree at the University of Illinois at Urbana–Champaign. Considerable thanks are due to Bruce Headey and Alexander Wearing for use of their data. We are also indebted to Brent Roberts, C. Y. Chiu, Lawrence Hubert, and Laura King for their insightful comments and guidance. We also thank the Australian Social Science Data Archives. Correspondence concerning this article should be addressed to Christie Napa Scollon, Department of Psychology, Texas Christian University, 2800 South University Drive, TCU Box 298920, Fort Worth, TX 761298920. E-mail:
[email protected] 1152
LOVE, WORK, AND TRAIT CHANGES
By contrast, unemployment leads to long-term decreases in life satisfaction, such that many individuals do not return to their preunemployment levels of well-being, even years after regaining employment (Lucas, Clark, Georgellis, & Diener, 2004). Similar patterns have been noted with regard to close relationships. Widowhood leads to a precipitous decline in life satisfaction that does not return to baseline levels even 7 years after the event (Lucas, Clark, Georgellis, & Diener, 2003). In addition, Robins, Caspi, and Moffitt (2002) found that conflict, abuse, and poor relationship quality predicted increases in negative emotionality over time. Similarly, marital tension predicted increases in femininity, whereas divorce predicted decreases in dominance in an all female sample (Roberts et al., 2002). By contrast, increases in marital satisfaction correlated with increases in well-being and effective functioning and decreases in anxiety over time (Roberts & Chapman, 2000). Likewise, male veterans who married or remarried declined more in neuroticism after 12 years than those who remained single (Mroczek & Spiro, 2003). In some cases, the benefits of marriage extend beyond emotional rewards. Roberts and Bogg (2004) found that time spent married predicted increases in social responsibility, a facet of conscientiousness (see also Robins et al., 2002).
Correlated Change According to transactional views of development, we should expect changes in relationships to correspond to changes in personality. In other words, the two changes should correlate. However, the empirical evidence for correlated change has been somewhat inconsistent. Only one study has found that changes in relationships correlated with changes in personality (Roberts & Chapman, 2000), whereas two other studies (Asendorpf & Wilpers, 1998; Neyer & Asendorpf, 2001) found no relation between the two dynamic constructs. The inability to find significant correlated change could be due to a number of factors other than the lack of a true relation between personality and social relationships. First, studies of correlated changes require a large sample size.1 Second, previous studies have addressed changes at the observed level, in which measurement error can attenuate the correlation between two changes, a point to which we return later. Finally, a focus on relationship variables that provide minimal, if any, indication of the quality of participants’ relationships may not capture the psychological significance of a role or its ongoing functioning. For instance, Asendorpf and Wilpers (1998) had participants record the number of interactions they had, the number of same- and opposite-sex peers, and so on. Roberts et al. (2005) have noted that psychological qualities, such as role satisfaction, are more important determinants of role investment than the mere acquisition of a role. After all, a good relationship might have the power to promote wellbeing, whereas a dysfunctional one might increase ill-being (Roberts, 1997; Robins et al., 2002). Thus, one goal of the present study was to test whether changes in personality traits correlated with changes in role satisfactions when examined at the latent level. Note that although examinations of correlated change are still quite rare in the personality literature, a number of examples on cognitive functioning can be found (e.g., Sliwinski, Hofer, & Hall, 2003).
1153
Implications of Correlated Change Correlated change is essential to understanding development. A central feature of the present study is our focus on change at the individual level. Understanding individual differences in change is essential to understanding life-span development (Hertzog & Nesselroade, 2003). Although experiments are often considered the “gold standard” in psychological research, it is impossible and unethical to experimentally manipulate the long-term variables of theoretical interest to development such as love and work. However, by understanding the relation between changes in different variables, we can uncover important mechanisms that potentially shape development (Hertzog & Nesselroade, 2003; cf. Sliwinski et al., 2003). Our study investigates the relation between two underlying developmental trajectories, which cannot be obtained from simple concurrent correlations. Whereas concurrent correlations address time-specific relations between variables, correlated change provides evidence of personality and social roles enhancing one another over time. Although a handful of past studies have examined individuallevel change in personality (e.g., Helson, Jones, & Kwan, 2002; Jones & Meredith, 1996; Robins, Fraley, Roberts, & Trzesniewski, 2001; Vaidya, Gray, Haig, & Watson, 2002), past research has mainly focused on estimating the number of people who change or on cohort or gender as predictors of change. Although this research has led to important discoveries such as few gender differences in adult development (Helson et al., 2002), the question of what predicts change remains largely underexplored. Our study focuses on the psychological variables associated with change. The study of correlated change also has great potential to inform interventions and programs aimed at self-improvement. For instance, knowing that decreases in neuroticism over time are associated with increased work satisfaction might focus interventions on career counseling. Increases in extraversion are also likely to be salubrious given that extraversion and pleasant affect are consistently and moderately correlated (Lucas & Fujita, 2000). Even if correlated changes are small in magnitude, they may have enormous real-world consequences for an individual’s well-being. In fact, a 1-point difference could be a matter of life or death according to one recent study; Mroczek and Spiro (2005) found that for every half standard deviation increase in neuroticism per decade, the result was a 40% increase in mortality! These results occurred even after they controlled for physical health and age. Clearly, knowing what factors are associated with changes in neuroticism is vital to enhancing physical and psychological well-being.
Extending Previous Research The extant literature on adult personality development leaves unanswered several intriguing questions that are amenable to the present research. 1 Asendorpf and Wilpers (1998) had a sample size of 132 and therefore may have lacked the statistical power necessary to detect an association between the two changes, especially after the investigators controlled for initial status in both variables and made Bonferroni adjustments to avoid Type I error.
1154
SCOLLON AND DIENER
Is Development Limited to Young Adulthood? Many previous studies have examined either young or old participants but not both. Mroczek and Spiro (2003), for example, tracked male veterans initially 43 to 91 years of age. Small, Hertzog, Hultsch, and Dixon (2003) studied men and women initially 55 to 85 years of age. These studies are important because they have demonstrated that change remarkably occurs even well into old age. But given that the stability of personality traits reaches its maximum (at .74) between age 50 and 70 (Roberts & DelVecchio, 2000), a great deal of change in development might be overlooked by not sampling during young adulthood. On the other hand, an exclusive focus on young adults (e.g., participants 18 to 26 years of age in Roberts et al., 2003, and Robins et al., 2002) might fail to detect important changes that occur when one has established a career and a significant long-term partnership. The present study includes individuals as young as 16 and as old as 70 in the first wave of the study. This wide range of participant ages also allowed us to examine whether different age groups have different rates of change. Most studies have focused on the relation between the social world and development with the assumption that the relation remains constant throughout the life course. At the same time, others have argued that personality development ceases or slows down after a particular age (Costa & McCrae, 1994, 2006). Most famous was William James’s (1890/1918) claim that personality becomes “set like plaster” by age 30. Thus, it seems reasonable to test whether the relation between social roles and development differs before and after this important threshold. Although Srivastava et al. (2003) and others have investigated whether indeed personality becomes set by age 30, their study and most others have focused on mean levels or age differences in traits. Unlike past research, the present study addresses this question from the perspective of individual differences in change. Thus, the second goal of this study was to compare individuals under age 30 with those over age 30 to see whether (a) there is more variability in individual change in young adulthood and (b) correlations between changes in personality and changes in social roles are greater in young adulthood.
Latent Growth Modeling (LGM) Lack of correlated changes in previous studies may have been due to measurement error. The present study extends past research by measuring relationship variables and personality variables each on more than two occasions and using LGM to estimate change. Growth models offer more precise estimates of change because they are based on more than two assessments, and they do not require the same number of assessments for all participants in a study; in fact, growth models are tolerant of missing data, thereby allowing researchers to use more of the available data, rather than only complete case data. It is important to note that latent correlated change is unattenuated by measurement error.
Representative Samples Most previous studies have examined a small number of elite individuals who might not be representative of broader society. For example, Roberts (1997), Roberts and Bogg (2004), Roberts and
Chapman (2000), and Roberts et al. (2002) all relied on the Mills Longitudinal sample, which included women who attended Mills College in the 1950s.
Different Types of Change The study of individual differences in change represents a burgeoning area within the field of personality, although ours is not the first to examine this type of change. As early as the 1970s, Baltes and Nesselroade (1979) were pioneers in the study of individual differences in change, and their work continues to influence the theory and techniques behind studying change (see Hertzog & Nesselroade, 2003). In recent years, other investigators have established that individual differences in change exist for several traits (e.g., Jones & Meredith, 1996; Mroczek & Spiro, 2003; Small et al., 2003). Our study extends past findings by identifying predictors of change. Relative to other types of change, however, investigations of individual differences in change are still in the minority. Most research on personality development has focused on group means or test–retest correlations (i.e., rank-order change; see Caspi & Roberts, 1999). Although population statistics can inform us of normative change, any attempts to address predictors of change must ultimately treat change itself as an individual difference. Thus, the question of why change occurs for some individuals and not for others remains vastly underexplored. By examining individual differences in change, we hope to clarify at least part of this process. At the same time, for the sake of comparison across the literature, we also report findings for mean-level and rank-order change.
Implications for a Theory of Traits Early representations of the five-factor theory (FFT; McCrae & Costa, 1990) strongly suggested that traits do not change in adulthood. For example, in 1994, McCrae and Costa stated that “Individual differences in personality traits, which show at least some continuity from early childhood on, are also essentially fixed by age 30” (p. 173). By, 1999, however, McCrae and Costa claimed that “Traits develop throughout childhood and reach mature form in adulthood; thereafter they are stable in cognitively intact individuals” (p. 145). Such a view implied that traits change in childhood and as the result of dementing disorders in adulthood but that traits do not change in normal adults. More recently, these authors (Costa & McCrae, 2006; McCrae, 2002) have conceded that modest trait change after childhood may occur. Unlike the social investment model (Roberts et al., 2005), however, Costa and McCrae (2006) claimed that “changes are more pronounced early in adulthood than either before or after” (p. 26), a view that resonates with James’s plaster hypothesis. Moreover, FFT states that traits are “insulated from the effects of the environment” (McCrae & Costa, 1999, p. 144). Traits are thought to influence characteristic adaptations such as social roles, but social roles clearly do not influence trait development in FFT (see McCrae & Costa, 1999, Figure 5.1). Consequently, the FFT attributes observable trait changes to intrinsic maturation, rather than environmental influences (Costa & McCrae, 2006; McCrae, 2002). The present study examined two important points of scientific contention from the FFT. First, we examined the claim that trait
LOVE, WORK, AND TRAIT CHANGES
changes are more pronounced in early adulthood rather than in later adulthood. As mentioned in the previous section and elaborated on in the analyses section, if development is limited to young adulthood, older adults should exhibit low or zero variability in within-person trait changes (i.e., lower variances around the slopes of extraversion and neuroticism). Unfortunately, our data did not permit an examination of how development differs for young adults compared with children. Second, we addressed whether traits are insulated from the effects of the environment by examining the correlation between changes in traits and changes in role satisfactions. A nonzero correlation between changes in extraversion or neuroticism and changes in work or marital satisfaction would challenge this view of traits. It would be especially convincing if specific traits change more with specific roles, in other words, if changes in personality and social roles do not conform to a pattern of global increases in positivity. Of course, the direction of causality cannot be determined from correlated changes. It is possible for intrinsic maturation to cause changes in both traits and role satisfactions. Thus, a stronger test of causal direction comes from a cross-lagged model in which the influence of traits on changes in role satisfaction can be separated from the influence of roles on trait changes.
Do Satisfying Social Roles Lead to Increased Emotional Stability or Does Increased Emotional Stability Cause People to Enjoy Their Social Roles More? Of studies that have explicitly compared the directionality of paths, Wood and Roberts (2006) found no support for trait effects on roles but significant support for role effects on traits. On the other hand, Neyer and Asendorpf (2001) found evidence for trait effects on social relationships but not vice versa. Thus, the third goal of the present study was to explore the causal relation between roles and changes in personality with a cross-lagged design.
Study Overview We examined mean-level, rank-order, and individual-level changes in extraversion and neuroticism over time. Consistent with theories of transactional development, we focused on work and close relationships as important life domains that relate to increased well-being. We also used high extraversion and low neuroticism as proxies for increased well-being, given the wellreplicated relations between extraversion and pleasant affect (Lucas & Fujita, 2000) and neuroticism and unpleasant affect (Costa & McCrae, 1980). The aims of the present study were three-fold. First, we sought to replicate the finding that despite high rank-order stability, significant individual differences in change in extraversion and neuroticism exist (Jones & Meredith, 1996; Mroczek & Spiro, 2003; Small et al., 2003). In addition, we examined whether trait changes correlate with changes in work and relationship satisfaction. Second, the present study examined whether individual differences in trait changes are more likely to occur among individuals under age 30. If development ceases by age 30, older individuals should show less (or no) individual-level change in traits and a smaller relation (or no relation) between life experiences and development. Third, the present study explored the directionality of changes in traits and social roles. In other words, do satisfying social roles
1155
lead to changes in personality traits or do traits lead to changes in role satisfaction? We examined this final question with a crosslagged path model.
Method Victorian Quality of Life Panel Study The sample consisted of participants in an 8-year longitudinal study in Victoria, Melbourne, Australia. Pioneers in quality of life research, Bruce Headey and Alexander Wearing, in consultation with the Australian government, established the Victorian Quality of Life Panel Study in 1981 and tracked the subjective well-being of Australian residents every 2 years until 1989, for a total of five waves of assessment (see Headey & Wearing, 1989, 1992). The original panel consisted of 941 participants who were selected as a representative sample of Victoria’s population. The sample represented a wide range of socioeconomic statuses and a balance between rural and urban dwellers. In the first wave of the study, participants ranged in age from 16 to 70 (M ⫽ 37.2, SD ⫽ 13.3). In 1983, 189 participants joined the study, replacing participants who dropped out of the study, bringing the total sample size to 1,130 (mean age in 1983 ⫽ 39.9, SD ⫽ 13.3). We excluded 1 person from the analyses because she did not provide age information at any of the assessments. Unfortunately, the panel experienced substantial attrition over the 9 years, leaving available complete data on 33% of participants. Fifty-two percent of participants completed four or more assessments. Sixty-five percent completed three or more assessments, and 74% completed two or more assessments. Headey and Wearing (1992) noted that younger participants and those of lower socioeconomic status were somewhat more likely to drop out of the study. It is important to note that the longitudinal sample did not significantly differ from nonlongitudinal samples on the major variables of interest (see the Results section). Participants responded in interviews in 1981, 1983, and 1985, whereas in 1987 and 1989, respondents completed survey measures. Table 1 shows the number of male and female respondents in each category in the first wave of assessment.2
Measures Extraversion and neuroticism. Participants completed the Eysenck Personality Inventory (Form B; Eysenck & Eysenck, 1968), which consisted of 24 items designed to measure extraversion and 24 items designed to measure neuroticism. Scores on the Neuroticism scale had a theoretical range of 0 to 24. Alphas for the Neuroticism scale ranged from .81 (in 1981 and 1983) to .83 (in 1989). For the Extraversion scale, we omitted 6 items that had low interitem correlations (below .30) and conceptually were closer to impulsivity than extraversion. Internal consistencies for the 18item Extraversion scale ranged from .61 (in 1981) to .68 (in 1989). Extraversion scores had a theoretical range of 0 to 18. Unfortunately, respondents did not complete extraversion and neuroticism measures in 1985. Work satisfaction. Participants responded to questions about their satisfaction with work using a 1 (terrible) to 9 (delighted) scale. Six items measured satisfaction with work, including “How do you feel about the chance you have to use your skills and abilities at work?” and “How do you feel about your job in general?” Alphas ranged from .80 (in 1985) to .86 (in 1989). Relationship satisfaction. Five items assessed satisfaction with one’s romantic relationship. Respondents who were married or living with their romantic partner answered these questions, even if they were not legally married (n ⫽ 653 in 1981; n ⫽ 622 in 1983; n ⫽ 547 in 1985; n ⫽ 490 in
2
Assignment to age groups was based on age in 1981. For individuals who joined in 1983, we estimated age in 1981 as age in 1983 minus 2.
SCOLLON AND DIENER
1156 Table 1 Number of Participants by Gender and Age Age groupa
Male
Female
Under 20 20–29 30–39 40–49 50–59 60⫹ Not reporting age
28 160 134 95 86 35 0
34 178 147 91 98 43 1
62 338 281 186 184 78 1
538
592
1,130
Total
Total
a
Age group assignment was based on participant’s age in 1981. For individuals who joined the study in 1983, age in 1981 was estimated as age in 1983 minus 2.
ences in traits (e.g., Roberts et al., 2006; Srivastava et al., 2003), we included the covariate of age (centered at beginning of the study) in our model. We did this because we were specifically interested in changes over time above and beyond age effects or simple maturation. Two important parameters are associated with the latent slope: mean slope (MS) and variance or deviance around the slope (DS). The mean component is equivalent to a fixed effect in HLM, whereas the variance is a random effect. Therefore, the mean of the slope addresses normative change. If the mean departs from zero, this gives an indication of how the sample as a whole changed. It is possible for the mean to be zero, indicating no normative change, while having a nonzero variance component (i.e., a nonzero random effect). Nonzero variance around the slope indicates significant individual differences in change, so testing whether this component is nonzero determines whether further analyses are possible—a procedure that is also common in analyses using HLM. The intercept is also characterized by a mean intercept (MI) and a variance around the intercept (DI). The mean refers to the average intercept
1987; n ⫽ 361 in 1989). Sample items included “How do you feel about the extent to which your husband/wife understands you?” and “How do you feel about your marriage?” Alphas ranged from .86 (in 1981) to .94 (in 1989).
Analyses LGM. In the present study, we chose to model changes using structural equation models of latent growth curves (see McArdle, 1989, 2005). We elected to use LGM because it takes into account measurement error and uses more of the available data than alternative methods such as withinperson regression (Willett, 1988). With missing data, as is common in longitudinal studies, models cannot be fit to simple covariance matrices. Instead, full information maximum likelihood (FIML) or “direct estimation” procedures must be used to fit models to the raw data. In other words, an algorithm estimates the model using all available data on all cases (Hox, 2000). The structural equation modeling software Amos 5.0 (Arbuckle, 2003) includes this feature. Thus, even participants who provide only one data point can contribute to the modeling of means and variances. Hierarchical linear modeling (HLM) is also capable of this. In contrast, withinperson regression requires an individual to have at least three data points in order to create a meaningful regression for that person. Note that FIML procedures differ from those using only complete case data or data imputation, both of which can lead to biased estimates (Wothke, 2000). Hox (2000) has also shown that in handling missing observations, LGM with FIML is efficient and yields accurate estimates. Additionally, LGM can simultaneously model multiple dynamic variables (e.g., McArdle, 1989), thus allowing for an examination of interrelationships in change or correlations of change components, a major goal of the present study. The basic latent growth model. Figure 1 illustrates a basic linear latent growth model. As is common in structural equation models, circles denote latent variables, and squares denote observed variables. The observed variables, T1 to T5, refer to the repeated measurements taken every 2 years over the course of the study (extraversion in 1981, extraversion in 1983, etc.). Two-headed arrows represent correlations, and single-headed arrows represent regression coefficients or directed paths. A latent slope was modeled with directed paths from the latent variable to the observed variables or measurement occasions. We constrained these paths or factor loadings to equal the number of years that had passed at each assessment since the beginning of the study (e.g., 0, 2, 4, 6, and 8). This is the same as centering at the beginning of the study in HLM and represents linear change over time. Because 1985 assessments of extraversion and neuroticism were not available, we omitted the T3 variable and its associated path (denoted “4” in Figure 1) for these measures. The loadings of the repeated measures on the intercept factor were constrained to unity. In addition, given that previous studies have demonstrated cross-sectional age differ-
Figure 1. Univariate latent growth model. Circles denote latent variables, and squares denote observed variables. The observed variables, T1 to T5, refer to the repeated measurements taken every 2 years over the course of the study. Two-headed arrows represent correlations, and single-headed arrows represent regression coefficients or directed paths. A latent slope was modeled with directed paths from the latent variable to the observed variables or measurement occasions. We constrained these paths or factor loadings to equal the number of years that had passed at each assessment since the beginning of the study (e.g., 0, 2, 4, 6, and 8). Paths from the latent intercept to the observable variables were constrained to unity. Solid lines refer to an intercept-only or no-growth model. We tested a growth model by adding the variables and paths represented by the dashed lines. Du ⫽ unaccounted variation; MI ⫽ mean intercept; DI ⫽ variance around the intercept; MS ⫽ mean slope; DS ⫽ deviance around the slope.
LOVE, WORK, AND TRAIT CHANGES for the group as a whole (fixed effect), whereas the variance component describes individual differences in initial level (random effect). Finally, DU represents unaccounted variation, or error variance. The model constrains error variances to be equal across measurement occasions, rather than “free.” McArdle (2005) noted that using different sized error at different occasions has no substantive or logical basis and has the potential to capitalize on chance in the data. Although we recognized that the possibility of nonlinear growth existed, we elected to examine linear instead of more complex (e.g., quadratic) growth for several reasons. First, with five assessments and no theoretical rationale for nonlinear effects, a linear model seemed reasonable. Second, it is possible to estimate the factor loadings from the latent slope (known as a latent basis model). When we did this, the latent basis models tended not to fit much better than the linear growth models, suggesting that linear growth was a reasonable approximation of the development of these constructs. Third, when we used latent basis models and compared them with linear models, the interrelationships among change components that were of primary interest remained virtually unchanged. Fourth and most important, the meaning of the correlation between two slopes becomes difficult to interpret when the slope functions are nonlinear. Bivariate latent growth model. Figure 2 illustrates a bivariate latent growth model in which two constructs change together over time. For example, this figure might represent a latent slope and intercept for Variable A, extraversion, and a latent slope and intercept for Variable B,
DuA
T1A
1
DuA
DuA
T3A
T2A
1 1
1
1
work satisfaction. The model also includes estimates of the mean slope for each variable, A and B (MSA and MSB), and an estimate of the mean intercept for each variable, A and B (MIA and MIB). DSA and DSB capture deviations or variability around the respective slopes, and DIA and DIB capture deviations around the respective intercepts. DUA and DUB represent error variances, which were constrained to be equal across measurement occasions, but not across the different constructs. The path rSSAB represents the correlation between slopes. For example, this path might represent the relation between change in extraversion and change in work satisfaction. We included other paths such as the relation between slopes and intercepts to control for these associations. Because rate of change is often correlated with initial status, it is important to include these control features in the model. Again, we included age in the model as a covariate to control for cross-sectional age differences. Correlated change. Do extraversion and neuroticism changes accompany work or relationship satisfaction changes over time? A correlation between two slopes (i.e., rSSAB) would suggest that experience shapes personality development to some degree and vice versa. The dotted line in Figure 2 highlights this path. Group differences. Does development differ for younger and older adults? If personality becomes less malleable or “set like plaster” with age, then we should expect less variability in the slopes of extraversion and neuroticism in an older compared with younger sample. In addition, correlated changes should be smaller in magnitude among older adults.
DuA
T4A
2
Intercept A MIA
DuA
4
DuB
T5A
6
8
1
0 rSIAB MSA
DIA
rISA
DuB
T1B
Slope A DSA
1157
DuB
T2B
1 1
1
T3B
1
rIIAB
DIB
DuB
T4B
2
Intercept B MIB
DuB
4
T5B
6
8
0
Slope B rISB
MSB DSB
rSSAB rISAB
Age (centered) Figure 2. Bivariate latent growth model. Circles denote latent variables, and squares denote observed variables. The observed variables, T1 to T5, refer to the repeated measurements taken every 2 years over the course of the study. Two-headed arrows represent correlations, and single-headed arrows represent regression coefficients or directed paths. A latent slope was modeled with directed paths from the latent variable to the observed variables or measurement occasions. We constrained these paths or factor loadings to equal the number of years that had passed at each assessment since the beginning of the study (e.g., 0, 2, 4, 6, and 8). Paths from the latent intercept to the observable variables were constrained to unity. The dashed line represents the correlation between slopes (i.e., correlated change). A and B represent the two variables. Du ⫽ unaccounted variation; MI ⫽ mean intercept; DI ⫽ variance around the intercept; MS ⫽ mean slope; DS ⫽ deviance around the slope.
SCOLLON AND DIENER
1158
That is, satisfying work and relationships should be less correlated with changes in personality with age (i.e., smaller rSSAB). To determine whether developmental patterns differ by age group, we split the sample into two groups: individuals who were under age 30 in 1981 (n ⫽ 400) versus individuals who were age 30 and up in 1981 (n ⫽ 729; see Footnote 2). Of course, because the study spanned a period of 8 years, individuals who were, for example, 24 years old at the first wave of assessment would reach age 32 by the end of the study. According to the plaster hypothesis, fewer or no changes should occur for those individuals who turned 30 before the end of the study, making it more difficult to detect changes in both groups. The decision to split the groups at age 30 at the beginning of the study is, therefore, a conservative test. Presumably, we would detect greater group differences if we split the groups at a younger age. For the under-30 group, on average 143 respondents answered the relationship satisfaction items at each assessment from 1981 to 1989. An average of 200 respondents under age 30 answered the items with regard to work satisfaction each year. There were more responses in the over-30 group for both variables. Approximately 390 adults age 30 and up responded to the relationship satisfaction items at each assessment, and on average 313 responded to the work satisfaction items. We used the multiple-groups feature of Amos 5.0 to simultaneously compare parameters across the two groups. We compared several nested models beginning with the most restrictive model in which all parameters were constrained to be equal across groups. At each subsequent step, we freed a set of parameters (e.g., means, variances, covariance), that is, allowed them to differ between the groups. A significant increase in fit of the model would indicate that the freed parameters differed between the groups, whereas no increase in fit would indicate that the parameters were the same across groups. We were mainly interested in whether the variance around slopes and the correlation between slopes were the same for both groups. Because cross-sectional differences in means (or initial levels) have been documented elsewhere (Roberts et al., 2006; Srivastava et al., 2003), we expected the groups would differ in their mean intercepts. To test whether development differs in young and old adulthood, we compared the variance of the slopes of the two groups. Cross-lagged model. Figure 3 illustrates a cross-lagged model. The diagonal paths marked “t” represent the trait effects on role satisfaction.
e
e at
Trait81
at
Trait83
Descriptive Statistics Table 1 describes the composition of our sample by age and gender at the beginning of the study. Consistent with Helson et al. (2002), we found few gender differences. At the start of the study, women were higher than men in neuroticism (M ⫽ 12.03 vs. 10.40), t(1126) ⫽ 2.33, p ⬍ .05, and lower in extraversion (M ⫽ 10.75 vs. 11.15), t(1127) ⫽ 5.83, p ⬍ .01. Gender did not predict change over time (i.e., slopes); therefore, we omitted gender from further analyses. Table 2 shows the means for all variables at all time periods. In general, the sample decreased in extraversion and neuroticism over time. However, these descriptives obscure age differences because people of all ages participated in each wave. Furthermore, attrition may have influenced the means for later waves of assessment.
e at
Trait85
t
r
e
Results
e
t
Role81
The diagonal paths marked “r” represent role satisfaction effects on personality. To control for the temporal stability of traits and roles over time, we included horizontal paths labeled “at” and “ar”. Because intervals are of equal length, we assume that stability and trait and role effects remain constant across lags. Because observations of traits in 1985 were missing for all participants, we modeled this variable as latent or “phantom” (see McArdle, 1994) and represented it in Figure 3 with the customary circle. We correlated the error terms for variables measured in the same year (e.g., extraversion in 1983 and work satisfaction in 1983). These correlated residuals also represent correlated change, but because they are computed at the observed level (because we only had one indicator of each variable at each wave of assessment), they are not free of measurement error. Furthermore, the correlated residuals do not examine change over the entire study but instead represent change over the shorter time frame (e.g., 2 or 4 years). We examined four models: (a) extraversion and satisfaction, (b) neuroticism and work satisfaction, (c) extraversion and relationship satisfaction, and (d) neuroticism and relationship satisfaction.
Role83
e
Role85
e
Trait89
t
r ar
at
Trait87
t
r ar
e
r ar
Role87
ar
e
Figure 3. Cross-lagged model. The diagram shows traits (neuroticism and extraversion) and roles (relationship and work satisfaction) measured in 1981, 1983, 1985, 1987, and 1989. The circle denotes a latent variable, and squares denote observed variables. The horizontal paths at and ar control for the temporal stability of traits and roles over time. t ⫽ trait effects on role satisfaction; r ⫽ role satisfaction effects on personality; e ⫽ error.
Role89
e
LOVE, WORK, AND TRAIT CHANGES
1159
Table 2 Means, Standard Deviations, and Sample Sizes for the Full Sample and for the Sample Including Only Complete Case Data Extraversion Year
M
SD
Relationship satisfaction
Neuroticism n
M
SD
n
Work satisfaction
M
SD
n
M
SD
n
7.00 7.20 7.16 7.01 6.73
1.5 1.3 1.2 1.3 1.3
653 623 543 490 361
6.68 6.85 6.86 6.83 6.61
1.5 1.2 1.1 1.2 1.4
672 588 500 450 352
1.4 1.3 1.2 1.2 1.3
273 267 272 277 274
6.84 6.80 6.85 6.79 6.56
1.4 1.2 1.2 1.3 1.5
266 262 266 268 275
Full sample 1981 1983 1985 1987 1989
11.00 10.66
2.9 2.9
941 865
11.20 11.31
4.8 4.8
939 862
10.93 10.44
3.1 3.1
649 482
10.26 10.31
4.8 5.0
649 482
Complete case data 1981 1983 1985 1987 1989
10.81 10.54
3.0 3.1
374 374
10.73 10.95
4.8 4.9
372 373
10.77 10.56
3.0 3.0
374 374
10.34 10.32
4.8 5.0
374 374
Mean-Level and Rank-Order Changes in Extraversion and Neuroticism To gain another perspective on the sample as a whole, we created Table 3, which shows cross-sectional age differences in extraversion and neuroticism. We classified participants by age categories in 1981.3 Participants could be classified into one of six categories: under age 20 (n ⫽ 57), age 20 to 29 (n ⫽ 333), age 30 to 39 (n ⫽ 277), age 40 to 49 (n ⫽ 195), age 50 to 59 (n ⫽ 182), and age 60 to 70 (n ⫽ 85). A one-way analysis of variance revealed significant age differences in neuroticism, F(5, 1122) ⫽ 4.35, p ⬍ .01, consistent with previous studies showing that neuroticism decreases with age (e.g., Srivastava et al., 2003). Consistent with previous cross-sectional studies, extraversion also exhibited steady decline with age, F(5, 1123) ⫽ 9.97, p ⬍ .001. It’s interesting to note that post hoc tests (Tukey’s least significant difference) revealed no significant differences among the over-30 groups. Post hoc tests showed that the over-30 groups differed from the under-30 groups, although the age-40 group unexpectedly did not differ significantly from the age-20 group. Thus, at the group level, there did appear to be moderate support for the idea that normative development slows after age 30. Tables 4 and 5 show the stability coefficients for neuroticism and extraversion. The cells below the diagonal of each table report stability coefficients on all available data. The cells above the diagonal show the stability coefficients for complete case data only. Consistent with previous studies (Costa & McCrae, 1988; Robins et al., 2001), both traits exhibited high rank-order consistency even across the 8-year period.
7.07 7.09 7.15 6.96 6.76
variable does not take into account such fine distinctions because we did not have enough data points to support more complex analyses. A regression predicting completeness from the major variables of interest yielded no significant effects. Identical results emerged when we included the 189 individuals who joined the study in 1983 (using their 1983 scores). However, zero-order correlations revealed a slight relation between completeness and neuroticism scores (r ⫽ ⫺.09, p ⬍ .05). We also tested whether attrition was related to slopes and intercepts of and neuroticism. Number of waves completed was unrelated to extraversion. However, number of waves completed was related to the slope of neuroticism ( ⫽ .05, p ⬍ .01) and the initial level of neuroticism ( ⫺.24, p ⬍ .01), suggesting that people who completed more assessments were lower in neuroticism to begin with and exhibited less steep declines in neuroticism over time. The effect is not surprising given that the intercept and slope of growth functions are often related. People who start out lower in neuroticism do not decline as much over time. Thus, we do not claim the participants in the longitudinal sample were randomly selected from the entire sample; however, the effects of attrition were small relative to the correlation between changes that later emerged. Nevertheless, our findings should be considered within this context. Wherever possible we performed analyses on complete case data and on all available data, and few differences in results emerged.
Personality Growth Trajectories
Did the Longitudinal Sample(s) Differ From the Start?
Univariate model. As a baseline, we first fit a no-growth model to the data. This model estimates only the intercepts and is represented by only the solid lines in Figure 1. We then added the slope and its associated components (represented by the dashed
We created a simple variable that reflected the number of waves completed. More complex patterns distinguishing people who completed the first two waves from people who completed, for example, the first and third waves of assessment are possible. Our
3 For the participants who joined the study in 1983, we subtracted 2 from their age in 1983 to calculate their age in 1981. However, in Table 3 we report their 1983 age and 1983 scores.
SCOLLON AND DIENER
1160
Table 3 Cross-Sectional Age Differences in Neuroticism and Extraversion in 1981 Neuroticism Age group Under 20 20–29 30–39 40–49 50–59 60⫹ Total
Extraversion
M
SD
n
M
SD
n
13.0a 11.9a,b 10.7c 11.1b,c 10.8c 10.5c
4.6 4.7 4.8 4.8 4.7 4.5
57 333 277 194 182 85
12.5a 11.5a,b 10.9c 10.4b,c 10.4c 10.1c
2.7 2.9 2.8 2.9 2.7 3.0
57 333 277 195 182 85
11.3
4.8
1,128
10.9
2.9
1,129
Note. Neuroticism: F(5, 1122) ⫽ 4.35, p ⬍ .01. Extraversion: F(5, 1123) ⫽ 9.97, p ⬍ .01. For the 189 participants who entered the study in 1983, their 1983 age and scores were used. Differing subscripts denote significant post hoc differences between groups.
lines) and compared the fit of the models (e.g., ⌬2). We performed these analyses and comparisons on the entire sample, on the under-30 group, and on the over-30 group separately. If normative changes cease to occur after age 30, the no-growth model should fit just as well as the growth model for the older group. However, in all cases, the linear growth model fit better than a no-growth model that only had an intercept, ⌬2s(4) ⬎ 11, ps ⬍ .05.4 We compared the under-30 and over-30 groups to see if they differed in their parameters (e.g., MS). To test whether the groups had different parameters, we compared a model in which both groups were equated on all parameters (most restrictive model) with a model in which the parameter of interest was allowed to vary between the two groups. Thus, each test of each parameter was a 1-df test. Table 6 reports the final parameter estimates from the linear growth models. When the groups had differing parameter estimates, we allowed the parameter estimate to vary across groups in the final model, and two estimates are provided in the table. When the groups had identical parameters, the parameter was constrained to be equal across the two groups in the final model, and only one parameter is reported in the table. As Table 6 shows, the main differences between the two age groups were in mean intercepts (MI). This is no surprise given the cross-sectional differences that other studies have found (Srivastava et al., 2003). A crucial test is whether the variance around the slopes (DS) was nonzero, because it is this parameter that indicates the presence of individual differences in change. All slope variances were nonzero. In only one instance did the groups differ in their variance around slopes, and this was for the variable of relationship satisfaction. Although the younger sample had nonTable 4 Stability Coefficients for Neuroticism Year
1981
1983
1987
1989
1981 1983 1987 1989
— .73 .73 .66
.75 — .75 .68
.75 .76 — .74
.68 .71 .76 —
Note. Values below the diagonal are for all available data. Values above the diagonal were computed on complete case data.
zero variance around the slope of relationship satisfaction (DS ⫽ .01, SE ⫽ .004), the variance was significantly lower than that of the older sample (DS ⫽ .02, SE ⫽ .003). Note that the direction of differences in slope variances directly contradicts the hypothesis that changes slow down with age. Also worth noting were group differences in the relation of age to intercepts. For the under-30 group, the age covariate had a ⫺.16 relation to the intercept of neuroticism and a ⫺.13 relation to the intercept of extraversion, whereas these associations were virtually zero for the over-30 group. The negative covariate indicates that older individuals within that group had a tendency to have lower initial levels of neuroticism and extraversion. Again, this pattern is consistent with cross-sectional differences in traits (Srivastava et al., 2003). Cohort effects. Although not the main focus of our study, we also tested for the presence of cohort effects. We accomplished this by examining whether the path from the age covariate to slope was different from the mean of the slope. If these parameters are not equal, this suggests the presence of cohort effects. Furthermore, we tested this equality constraint for the entire sample, the under-30 group, and the over-30 group. We found the equality constraint for extraversion was not met for the under-30 group, ⌬2(1) ⫽ 4.1, p ⬍ .05, whereas it was met for the over-30 group, ⌬2(1) ⫽ 1.5, ns, and sample as a whole, ⌬2(1) ⫽ 0.9, ns. Thus, there appeared to be cohort effects for extraversion, especially among the younger group. This finding is consistent with Twenge (2001), who noted cohort effects in extraversion. For neuroticism, the equality constraint was met for the under-30 group, ⌬2(1) ⫽ 0.7, ns, whereas it was not met for the over-30 group, ⌬2(1) ⫽ 9.6, p ⬍ .05, and the sample as a whole, ⌬2(1) ⫽ 12.6, p ⬍ .05. Thus, there may have been cohort effects for neuroticism, especially among the older sample, consistent with Twenge (2000), who found cohort differences in neuroticism. Despite the presence of potential cohort effects, however, other studies have noted that trajectories of change across cohorts are similar (Helson et al., 2002). Reliability of change. Variability in responses over time can be due to true change or measurement error. Therefore, it is 4 For chi-square analyses on the entire sample, N ⫽ 1,129; for chi-square analyses on the under-30 group, N ⫽ 400; and for chi-square analyses on the over-30 group, N ⫽ 729.
LOVE, WORK, AND TRAIT CHANGES
1161
important to know how reliably individual differences in change can be measured over the 8-year period. Using parameter estimates from the growth model, we calculated the reliability of change (McArdle, 1986; McArdle, Prescott, Hamagami, & Horn, 1998; McArdle & Woodcock, 1997). These values are similar to alphas for internal consistency of scales administered on a single occasion, but they reflect the precision with which change was measured. In general, the reliabilities of change were quite good, ranging from .63 to .88. Compared with reliabilities that can be obtained from difference scores (see Cronbach & Furby, 1970), the growth models captured change with very high precision. Reliabilities remained virtually unchanged when we excluded the age covariate from the model.
of age group, as work satisfaction increased over time, so did extraversion (r ⫽ .58, p ⬍ .01). Analysis of only complete case data revealed a similar correlation between slopes (r ⫽ .56, p ⬍ .05). Relationship satisfaction and neuroticism. Does increased relationship satisfaction correlate with decreased neuroticism? The restricted model in which correlated change was equated across both groups, 2(101) ⫽ 214.2, yielded an almost identical fit to a less restricted model in which the magnitude of correlated change was allowed to vary across the groups, 2(100) ⫽ 214.1, indicating the two groups did not differ in their degree of correlated change, ⌬2(1) ⫽ 0.1, ns. Increases in relationship satisfaction corresponded to decreases in neuroticism for both samples (r ⫽ ⫺.42, p ⬍ .05). Analysis of only complete case data revealed a consistent, though slightly higher, correlation (r ⫽ ⫺.51, p ⬍ .05). Relationship satisfaction and extraversion. Do changes in relationship satisfaction correspond to changes in extraversion? Again, first we fit a model in which the correlation between slopes was constrained to be the same for both groups, 2(102) ⫽ 226.0, and compared this with a model in which the correlation between slopes was freed, 2(101) ⫽ 223.8. The lack of increased fit in the latter model, ⌬2(1) ⫽ 2.2, ns, indicated correlated change was the same for both age groups. Changes in relationship satisfaction were marginally related to increases in extraversion (r ⫽ .26, p ⬍ .10; r ⫽ .41, p ⬍ .10, for complete case data only).
Correlated Change
Cross-Lagged Model
Work satisfaction and neuroticism. Does neuroticism decrease as work becomes more satisfying? Does this relation vary by age group? Because we already compared the two groups on parameters such as mean intercept in the univariate models, we combined the final univariate models to form bivariate models and allowed the slopes and intercepts of both variables to correlate. All comparisons are based on a 1-df test.5 The correlation between the two slopes answers our first question of whether neuroticism and work satisfaction change together. To determine whether the groups differed in this relation, we compared a model in which the two groups had equal correlations, 2(102) ⫽ 202.0, with a model in which the correlation between the slopes was allowed to vary between the two groups, 2(101) ⫽ 201.9. As indicated by a nonsignificant increase in fit of the latter model, the different age groups did not differ in this correlation, ⌬2(1) ⫽ 0.1, ns. Table 7 shows that the final estimate of the correlation between changes in work satisfaction and neuroticism was ⫺.64 ( p ⬍ .01). As work satisfaction increased, neuroticism decreased for both young and old. We obtained similar results when we performed the same analyses on complete case data only. The correlation between changes in work satisfaction and changes in neuroticism dropped but was still significant at ⫺.43 ( p ⬍ .05). Work satisfaction and extraversion. Does extraversion increase as work becomes more satisfying? Does this relation vary by age group? We first tested a model in which the correlation between the slopes of work satisfaction and extraversion was constrained to be the same for both groups, 2(103) ⫽ 208.0, with a model in which the correlation between slopes was freed between the two groups, 2(102) ⫽ 208.0. The latter model did not fit any better than the former, ⌬2(1) ⫽ 0, ns, indicating that the two groups had the same degree of correlated change. Regardless
We examined four models: (a) extraversion and work satisfaction, (b) neuroticism and work satisfaction, (c) extraversion and relationship satisfaction, and (d) neuroticism and relationship satisfaction. Of primary interest were trait effects versus role effects, as shown in Figure 3. For extraversion and work satisfaction, trait effects emerged as significant ( ⫽ .07, p ⬍ .01) and role effects emerged as marginally significant ( ⫽ .09, p ⫽ .11). Using only complete case data for this model resulted in significant trait ( ⫽ .10, p ⬍ .001) and role effects ( ⫽ .09, p ⬍ .01). For neuroticism and work satisfaction, only trait effects emerged as significant (s ⫽ ⫺.07 and ⫺.09, both ps ⬍ .01, for all data and complete cases only). For extraversion and relationship satisfaction, no paths emerged as significant when we used all the data. However, among complete cases only, role effects emerged as marginally significant ( ⫽ .05, p ⫽ .08), whereas trait effects remained nonsignificant. Finally, for neuroticism and relationship satisfaction, trait effects emerged as marginally significant in analyses on all data ( ⫽ ⫺.04, p ⫽ .08) and significant in analyses on complete case data ( ⫽ ⫺.06, p ⬍ .05).
Table 5 Stability Coefficients for Extraversion Year
1981
1983
1987
1989
1981 1983 1987 1989
— .65 .62 .61
.68 — .68 .65
.65 .70 — .73
.62 .66 .73 —
Note. Values below the diagonal are for all available data. Values above the diagonal were computed on complete case data.
Discussion Does personality change over time? In what contexts do people increase in psychological well-being? Is personality development limited to young adulthood? The present study addressed these 5 The degrees of freedom vary among the models because the two groups differed on some parameters (e.g., mean intercept for neuroticism) but not others (e.g., variance around the slope of extraversion). However, the comparison test is always based on a 1-df test and is the same for all comparisons.
SCOLLON AND DIENER
1162
Table 6 Parameter Estimates and Standard Errors From Linear Growth Models Including Age as a Covariate MI Variable Neuroticism Under 30 Over 30 Extraversion Under 30 Over 30 Work satisfaction Under 30 Over 30 Relationship satisfaction Under 30 Over 30
Est.
DI SE
MS
Est.
11.99a 0.23 16.81b 10.98 0.17 16.46 5.43 11.51d 0.14 10.56 0.10 1.16 6.62f 0.07 6.92 0.06 7.16 0.05 1.30g 1.68
SE
Est.
ⴚ.11 1.46 1.11 0.34 ⫺.02 0.10
DS
⫺.05
SE Est.
CovI
SE
.02 .05
.02
⫺.15 .10 6.10 0.25
.01 .03
.01
⫺.10 .04 2.80 0.11
.01
Est.
DU
SE Est.
.001 .01 .02
0.17 0.13
rIS
SE
Est.
CovS SE
ⴚ.16c .06 ⫺.01 .02
ⴚ.13e .04 ⴚ.03 .01 .003 ⴚ.65 .02 0.82 0.03 .02 .01
.01h .004 ⴚ.58i .02 .02 .003 ⴚ.44 .02
0.57 0.02
.01
.01
Fit
Est.
SE
⫺.01
.002
41.9
25
.02
.00
.001
54.04 26
.03
2
df RMSEA
.001 .001 165.7
41
.05
.00
39
.05
.001 167.8
Note. Values in bold are significant at p ⬍ .05. Age was centered at the mean age for each group in 1981. MI ⫽ mean intercept; D1 ⫽ deviation from the intercept; MS ⫽ mean slope; DS ⫽ deviation from the slope; rIS ⫽ correlation between the slope and the intercept; DU ⫽ unaccounted variation; CovI ⫽ beta from age to intercept; CovS ⫽ beta from age to slope. a Test of group differences: ⌬2(1) ⫽ 13.4, p ⬍ .05. b Test of group differences: ⌬2(1) ⫽ 18.6, p ⬍ .05. c Test of group differences: ⌬2(1) ⫽ 18.8, p ⬍ .05. d Test of group differences: ⌬2(1) ⫽ 33.2, p ⬍ .05. e Test of group differences: ⌬2(1) ⫽ 6.8, p ⬍ .05. f Test of group differences: ⌬2(1) ⫽ 15.9, p ⬍ .05. g Test of group differences: ⌬2(1) ⫽ 8.0, p ⬍ .05. h Test of group differences: ⌬2(1) ⫽ 7.0, p ⬍ .05. i Test of group differences: ⌬2(1) ⫽ 8.0, p ⬍ .05.
questions by examining individual differences in change. Using latent growth modeling, we modeled changes in extraversion, neuroticism, work satisfaction, and relationship satisfaction over time and examined how these constructs change together over an 8-year period. Moreover, we examined whether individual differences in change were different for people under age 30 versus people age 30 and over. Several important findings emerged. First, despite impressive rank-order stability of traits, there were significant within-person changes in extraversion and neuroticism as evidenced by the nonzero variability in the slopes of both neuroticism and extraversion over time (cf. Jones & Meredith, 1996; Mroczek & Spiro, 2003; Small et al., 2003). Furthermore, individuals over age 30 exhibited just as much change as those under 30, thus refuting the idea that personality becomes “set like plaster by age 30” or that development slows down after young adulthood (Costa & McCrae, 2006). In only one case did the two age groups differ in variance around slopes (DS), and this was in variability in change in relationship satisfaction. Although both groups had nonzero variances around the slope of relationship satisfaction, the older group exhibited greater variance in change
Table 7 Correlated Change: Correlations Among Slopes in Bivariate Growth Models for All Available Data and Complete Case Data Neuroticism
Extraversion
Variable
All data
Complete data
All data
Complete data
Work satisfaction Relationship satisfaction
⫺.64** ⫺.42*
⫺.43† ⫺.51*
.58** .26†
.56* .41†
Note. Under-30 and over-30 groups did not differ in their correlations between slopes. † p ⬍ .10. * p ⬍ .05. ** p ⬍ .01.
than the younger group. The direction of group differences directly contradicts the notion that development ceases or slows down with old age. Although it is probably tenuous to conclude that older adults change more than younger ones, development certainly does not cease by age 30. In all likelihood, development at least continues on the same trajectory. Second, individual differences in changes in personality appeared systematic. Consistent with Roberts and Chapman (2000), increased work satisfaction accompanied decreases in neuroticism and increases in extraversion over time. Similarly, increased relationship satisfaction predicted decreases in neuroticism and increases in extraversion as well. Correlations among changes were similar for the under-30 and 30⫹ samples, indicating that transactional influences on development are not limited to young adulthood. Furthermore, the magnitude of the relation between change components was not trivial. In fact, on average, trajectories correlated about .40. In light of Mroczek and Spiro’s (2005) finding that increases in neuroticism are associated with mortality risk, we believe our results have serious implications. Our findings also lend further support to the idea that the social environment shapes personality and vice versa. At the same time, the current results pose a challenge to FFT (e.g., McCrae & Costa, 1990). Contrary to the notion that traits are insular dispositions, the present study suggests that extraversion and neuroticism can and do change over time and that these changes to some extent are related to important social institutions, such as work and romantic relationships, above and beyond the effects of age. Although trait changes found in this study could be interpreted as resulting from intrinsic maturation, we think this is an unlikely explanation. Intrinsic maturation would be more consistent with a general pattern of increasing positivity such that the correlated changes would be similar in magnitude regardless of domain or trait. However, we did not find such a pattern. Instead, work satisfaction had stronger
LOVE, WORK, AND TRAIT CHANGES
relations to trait changes, particularly declines in neuroticism. Relationship satisfaction was associated with decreases in neuroticism but had only a marginal relation to increases in extraversion. Third, in addition to individual differences in change, we also examined mean-level change and rank-order stability and found results comparable to previous research. Mean levels of extraversion and neuroticism declined with age, consistent with Srivastava et al. (2003), McCrae et al. (1999), Roberts et al. (2006), and others. At the same time, the rank-order stability of these traits was remarkably high and consistent with research by Costa and McCrae (1988). Thus, the present study underscores the conceptual and empirical independence of the three types of change (Caspi & Roberts, 1999). Fourth, our findings are consistent with research showing that extraversion and pleasant affect are consistently related. Cunningham (1988a, 1988b) and Lucas (2001) have demonstrated that extraverted behaviors follow pleasant mood induction, whereas Fleeson, Malanos, and Achille (2002) have shown that acting extraverted can lead to increased positive emotion. These past studies, however, focused on the short-term relation between pleasant emotion and extraversion. To our knowledge, ours is the first study to estimate the long-term, dynamic relation between satisfaction (in roles) and extraversion. Fifth, the cross-lagged analyses revealed moderate support for trait effects operating on satisfactions with roles, especially with regard to neuroticism. These findings converge with previous studies (Asendorpf & Wilpers, 1998; Neyer & Asendorpf, 2001) that found mostly trait effects. However, there was also modest support for work satisfaction leading to increased extraversion, a finding that supports the social investment model (Roberts et al., 2005). It may be that work plays a special role in growth in extraversion, but we need more research to fully understand the processes going on here. We caution, however, against drawing strong conclusions about causality from these cross-lagged models or from any observational data for that matter (Freedman, 1987). First, our model is an oversimplification of the development process, essentially treating multiwave data as a series of two-wave “snapshots” (Rogosa, 1980, p. 255). The length of time covered by the lags needs to correspond to the time course of the underlying causal process.6 Lags that are too short or too long may lead to spurious results. Second, there is always the possibility of a third variable affecting change in both of the variables. Only a true experiment can eliminate the possibility of a third variable, although experiments are also not a perfect solution. Experiments raise serious ethical concerns in the study of development, and like cross-lagged models, experiments may also fail to capture the appropriate time course of an underlying process. In all likelihood, developmental processes are too complex to be represented in simple “A causes B” terms. We believe transactional models, which highlight the codevelopment of traits and social relationships, most accurately reflect realworld development, although the cost of such models is that they cannot declare a causal “winner.” Clearly, there are limitations to cross-lagged designs, and we believe our findings make the most sense when considered in conjunction with the growth modeling results.
1163
Limitations and Future Directions The present study stands among only a small handful of largescale longitudinal psychological studies. The large representative sampling allowed for the use of sophisticated models of change and greater generalizability. Nevertheless, several caveats are worth mentioning. First, there is the limitation that the data were collected in the 1980s in Australia. Although Australia does not differ much from the rest of the Western world, especially in terms of well-being (e.g., Diener, Diener, & Diener, 1995), it is possible that results would vary with non-Western samples. In addition, historical context limits the interpretation of these findings. An earlier or a later time period might yield different estimates of means (Twenge, 2000). Indeed, we were unable to rule out the possibility of cohort effects or secular trends in our data as well. Although the interrelationships among change components might differ by historical time periods, the diminished importance of work and close relationships seems unlikely. Second, the heterogeneity in age in our sample allowed us to make intriguing comparisons of young and older adults. The downside of this heterogeneity, however, was a confounding of age and potential cohort effects. It is reassuring, however, that results from our study converged with other studies (Helson et al., 2002; Roberts, 1997). Unlike Roberts (1997), who compared development before age 27 with development after age 27 within individuals, the present study compared individual differences in development before and after age 30 between two groups. Thus, age differences in development in Roberts’s (1997) study invite the possibility of historical differences (e.g., 1960s vs. 1980s), whereas age differences in developmental trajectories in the present study are qualified by potential cohort differences. Coupled with Helson et al.’s (2002) finding that different cohorts tend to change in similar ways, we believe significant individual differences in trait change occur above and beyond cohort effects. Our study adds to the growing literature that finds development is similar before and after age 30. Third, although we found that changes in personality correlated with changes in social roles, the finding is nonetheless a correlation and subject to the limitations of any correlation. Chief among these, of course, is that we cannot infer causality from correlations. However, we believe that correlated changes tell an important story with implications for well-being research and interventions that should not be overlooked because of the lack of firm causal conclusions. For example, the relation between extraversion and pleasant affect is now a well-replicated finding in the subjective well-being literature (see Lucas & Fujita, 2000), and this relation appears to be a bidirectional one. Lucas (2001) demonstrated that people feel more sociable when pleasant affect is high. On the other hand, Fleeson et al. (2002) showed that behaving in an extraverted way, however artificial, leads to an increase in positive affect. Social roles and personality development may share a similar process to extraversion and pleasant affect. Future research should focus on this complex relation. The present study also focused on a limited definition and level of personality. Although most psychologists would probably agree 6
We thank an anonymous reviewer for pointing out this limitation of cross-lagged models and experiments.
SCOLLON AND DIENER
1164
that extraversion and neuroticism are important personality traits, they are not the only ones. In fact, traits alone capture only a facet of the complexities of personality (McAdams, 1996). Thus, evidence that self-reports of extraversion and neuroticism change over time cannot be generalized to other important aspects of personality. Moreover, items on the Eysenck Personality Inventory are behavioral indicators of personality, which can be far removed from molecular or other levels of individual differences; changes occurring at this level do not necessarily reflect changes on other levels. Another limitation of the present study was the use of subjective indicators of role quality (e.g., satisfaction). Future research should examine more specific aspects of work and relationships, including observer ratings. Observer ratings would eliminate the shared variance from self-reports and possibly illuminate specific aspects of roles that mediate the relation between changes in role quality and traits. We would not necessarily expect objective role measures to result in a different pattern of findings, however, because Heller, Watson, and Ilies (2004) found that perceptions of domain satisfaction are often related to objective aspects of the domain.
General Conclusions Four main findings emerged from this study that advance current knowledge of adult development. First, there was evidence of individual differences in change in personality. Second, personality development was systematic and associated with changes in social roles. Third, age 30 did not mark a special time when development ceased or even declined. Instead, people continued to change throughout the life course. Fourth, there was moderate support for trait effects on roles and only mild support for work influencing the trait of extraversion.
References Arbuckle, J. L. (2003). Amos 5.0 [Computer software]. Chicago: Smallwaters. Asendorpf, J. B., & Wilpers, S. (1998). Personality effects on social relationships. Journal of Personality and Social Psychology, 74, 1531– 1544. Baltes, P. B., & Nesselroade, J. R. (1979). History and rationale of longitudinal research. In J. R. Nesselroade & P. B. Baltes (Eds.), Longitudinal research in the study of behavior and development (pp. 1–39). New York: Academic Press. Caspi, A., & Roberts, B. W. (1999). Personality continuity and change across the life course. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 300 –326). New York: Guilford Press. Costa, P. T., Jr., & McCrae, R. R. (1980). Influence of extraversion and neuroticism on subjective well-being: Happy and unhappy people. Journal of Personality and Social Psychology, 38, 668 – 678. Costa, P. T., Jr., & McCrae, R. R. (1988). Personality in adulthood: A six-year longitudinal study of self-reports and spouse ratings on the NEO Personality Inventory. Journal of Personality and Social Psychology, 54, 853– 863. Costa, P. T., Jr., & McCrae, R. R. (1994). Set like plaster? Evidence for the stability of adult personality. In T. F. Heatherton & J. L. Weinberger (Eds.), Can personality change? (pp. 21– 40). Washington, DC: American Psychological Association. Costa, P. T., Jr., & McCrae, R. R. (2006). Age changes in personality and their origins: Comment on Roberts, Walton, and Viechtbauer (2006). Psychological Bulletin, 132, 26 –28.
Cronbach, L. J., & Furby, L. (1970). How should we measure “change”—Or should we? Psychological Bulletin, 74, 8 – 80. Cunningham, M. R. (1988a). Does happiness mean friendliness? Induced mood and heterosexual self-disclosure. Personality and Social Psychology Bulletin, 14, 283–297. Cunningham, M. R. (1988b). What do you do when you’re happy or blue? Mood, expectancies, and behavioral interest. Motivation and Emotion, 12, 309 –331. Diener, E., Diener, M., & Diener, C. (1995). Factors predicting the subjective well-being of nations. Journal of Personality and Social Psychology, 69, 851– 864. Eysenck, H. J., & Eysenck, S. B. G. (1968). Manual for the Eysenck Personality Inventory. San Diego, CA: Educational and Industrial Testing Service. Fleeson, W., Malanos, A. B., & Achille, N. M. (2002). An intraindividual process approach to the relationship between extraversion and positive affect: Is acting extraverted as “good” as being extraverted? Journal of Personality and Social Psychology, 83, 1409 –1422. Freedman, D. A. (1987). As others see us: A case study in path analysis. Journal of Educational Statistics, 12, 101–128. Headey, B., & Wearing, A. (1989). Personality, life events, and subjective well-being: Toward a dynamic equilibrium model. Journal of Personality and Social Psychology, 57, 731–739. Headey, B., & Wearing, A. (1992). Understanding happiness: A theory of subjective well-being. Melbourne, Australia: Longman Cheshire. Heller, D., Watson, D., & Ilies, R. (2004). The role of person versus situation in life satisfaction: A critical examination. Psychological Bulletin, 130, 574 – 600. Helson, R., Jones, C., & Kwan, V. S. Y. (2002). Personality change over 40 years of adulthood: Hierarchical linear modeling analyses of two longitudinal samples. Journal of Personality and Social Psychology, 83, 752–766. Hertzog, C., & Nesselroade, J. R. (2003). Assessing psychological change in adulthood: An overview of methodological issues. Psychology and Aging, 18, 639 – 657. Hox, J. J. (2000). Multilevel analysis of grouped and longitudinal data. In T. D. Little, K. U. Schnabel, & J. Baumert (Eds.), Modeling longitudinal and multilevel data: Practical issues, applied approaches, and specific examples (pp. 15–32). Mahwah, NJ: Erlbaum. James, W. (1918). The principles of psychology. New York: Holt and Company. (Original work published 1890) Jones, C. J., & Meredith, W. (1996). Patterns of personality change across the life span. Psychology and Aging, 11, 57– 65. Lucas, R. E. (2001). Pleasant affect and sociability: Towards a comprehensive model of extraverted feelings and behaviors. Unpublished doctoral dissertation, University of Illinois at Champaign–Urbana. Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2003). Reexamining adaptation and the set point model of happiness: Reactions to changes in marital status. Journal of Personality and Social Psychology, 84, 527–539. Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2004). Unemployment alters the set-point for life satisfaction. Psychological Science, 15, 8 –13. Lucas, R. E., & Fujita, F. (2000). Factors influencing the relation between extraversion and pleasant affect. Journal of Personality and Social Psychology, 79, 1039 –1056. McAdams, D. P. (1996). Personality, modernity, and the storied self: A contemporary framework for studying persons. Psychological Inquiry, 7, 295–321. McArdle, J. J. (1986). Latent variable growth within behavior genetic models. Behavior Genetics, 16, 163–200. McArdle, J. J. (1989). A structural modeling experiment with multiple growth functions. In R. Kanfer, P. L. Ackerman, & R. Cudeck (Eds.), Abilities, motivation, and methodology: The Minnesota Symposium on
LOVE, WORK, AND TRAIT CHANGES Learning and Individual Differences (pp. 71–117). Hillsdale, NJ: Erlbaum. McArdle, J. J. (1994). Structural factor analysis experiments with incomplete data. Multivariate Behavioral Research, 29, 409 – 454. McArdle, J. J. (2005). Latent growth curve analysis using structural equation modeling techniques. In D. M. Teti (Ed.), Handbook of research methods in developmental science (pp. 340 –366). Malden, MA: Blackwell. McArdle, J. J., Prescott, C. A., Hamagami, F., & Horn, J. L. (1998). A contemporary method for developmental-genetic analyses of age changes in intellectual abilities. Developmental Neuropsychology, 14, 69 –114. McArdle, J. J., & Woodcock, R. W. (1997). Expanding test–retest design to include developmental time-lag components. Psychological Methods, 2, 403– 435. McCrae, R. R. (2002). The maturation of personality psychology: Adult personality development and psychological well-being. Journal of Research in Personality, 36, 307–317. McCrae, R. R., & Costa, P. T., Jr. (1990). Personality in adulthood. New York: Guilford Press. McCrae, R. R., & Costa, P. T., Jr. (1994). The stability of personality: Observation and evaluations. Current Directions in Psychological Science, 3, 173–175. McCrae, R. R., & Costa, P. T., Jr. (1999). A five-factor theory of personality. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 139 –153). New York: Guilford Press. McCrae, R. R., Costa, P. T., Jr., Lima, M. P., Simoes, A., Ostendorf, F., Angleitner, A., et al. (1999). Age differences in personality across the adult life span: Parallels in five cultures. Developmental Psychology, 35, 466 – 477. Mroczek, D. K., & Spiro, A. (2003). Modeling intraindividual change in personality traits: Findings from the normative aging study. Journal of Gerontology: Psychological Sciences, 58(B), P153–P165. Mroczek, D. K., & Spiro, A. (2005). Personality change influences mortality in older men. Unpublished manuscript. Neyer, F. J., & Asendorpf, J. B. (2001). Personality–relationship transaction in young adulthood. Journal of Personality and Social Psychology, 81, 1190 –1204. Roberts, B. W. (1997). Plaster or plasticity: Are adult work experiences associated with personality change in women? Journal of Personality, 65, 205–232. Roberts, B. W., & Bogg, T. (2004). A longitudinal study of the relationships between conscientiousness and the social-environmental factors and substance-use behaviors that influence health. Journal of Personality, 72, 325–353. Roberts, B. W., Caspi, A., & Moffitt, T. E. (2003). Work experiences and personality development in young adulthood. Journal of Personality and Social Psychology, 84, 582–593. Roberts, B. W., & Chapman, C. N. (2000). Change in dispositional wellbeing and its relation to role quality: A 30-year longitudinal study. Journal of Research in Personality, 34, 26 – 41. Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency
1165
of personality traits from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25. Roberts, B. W., Helson, R., & Klohnen, E. C. (2002). Personality development and growth in women across 30 years: Three perspectives. Journal of Personality, 70, 79 –102. Roberts, B. W., Walton, K., & Viechtbauer, W. (2006). Patterns of meanlevel change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132, 3–27. Roberts, B. W., Wood, D., & Smith, J. L. (2005). Evaluating five factor theory and social investment perspectives on personality trait development. Journal of Research in Personality, 39, 166 –184. Robins, R. W., Caspi, A., & Moffitt, T. E. (2002). It’s not just who you’re with, it’s who you are: Personality and relationship experiences across multiple relationships. Journal of Personality, 70, 925–964. Robins, R. W., Fraley, R. C., Roberts, B. W., & Trzesniewski, K. H. (2001). A longitudinal study of personality change in adulthood. Journal of Personality, 69, 617– 640. Rogosa, D. (1980). A critique of cross-lagged correlation. Psychological Bulletin, 88, 245–258. Sliwinski, M. J., Hofer, S. M., & Hall, C. (2003). Correlated and coupled cognitive change in older adults with and without preclinical dementia. Psychology and Aging, 18, 672– 683. Small, B. J., Hertzog, C., Hultsch, D. F., & Dixon, R. A. (2003). Stability and change in adult personality over 6 years: Findings from the Victoria Longitudinal Study. Journal of Gerontology: Psychological Sciences, 58(B), P166 –P176. Srivastava, S., John, O. P., Gosling, S. D., & Potter, J. (2003). Development of personality in early and middle adulthood: Set like plaster or persistent change? Journal of Personality and Social Psychology, 84, 1041–1053. Twenge, J. M. (2000). The age of anxiety? Birth cohort change in anxiety and neuroticism, 1952–1993. Journal of Personality and Social Psychology, 79, 1007–1021. Twenge, J. M. (2001). Birth cohort changes in extraversion: A crosstemporal meta-analysis, 1966 –1993. Personality and Individual Differences, 30, 735–748. Vaidya, J. G., Gray, E. K., Haig, J., & Watson, D. (2002). On the temporal stability of personality: Evidence for differential stability and the role of life experiences. Journal of Personality and Social Psychology, 83, 1469 –1484. Willett, J. B. (1988). Questions and answers in the measurement of change. Review of Research in Education, 15, 347– 422. Wood, D., & Roberts, B. W. (2006). Cross-sectional and longitudinal tests of the personality and role identity structural model (PRISM). Journal of Personality, 74, 779 – 809. Wothke, W. (2000). Longitudinal and multi-group modeling with missing data. In T. D. Little, K. U. Schnabel, & J. Baumert (Eds.), Modeling longitudinal and multilevel data: Practical issues, applied approaches, and specific examples (pp. 219 –240). Mahwah, NJ: Erlbaum.
Received April 12, 2005 Revision received April 21, 2006 Accepted May 11, 2006 䡲
Acknowledgment The editors thank the following ad hoc reviewers, who reviewed manuscripts for Volumes 90 and 91 of the Journal of Personality and Social Psychology.
Attitudes and Social Cognition Henk Aarts Andrea E. Abele Dolores Albarracin Daniel Algom Mark D. Alicke Nancy Alvarado Nalini Ambady Daniel Ames David Amodio Susan M. Andersen Craig A. Anderson Hillary Anger Elfenbein Christopher Armitage Jamie Arndt Arthur Aron Lisa G. Aspinwall Ozlem Ayduk Jo-Anne Bachorowski Frank Baeyens Richard P. Bagozzi Paul G. Bain Mark Baldwin Rainer Banse Jamie Barden John A. Bargh Maya Bar-Hillel Tim Barnett John N. Bassili Roy Baumeister Robert B. Bechtel Jennifer Beer Jennifer Berdahl Jesse Bering Leonard Berkowitz Gayle Bessenoff Ann Bettencourt Becky Bigler Hart Blanton Herbert Bless Gerd Bohner Peter Borkenau Jennifer K. Bosson Veronika Brandstaetter Nyla R. Branscombe Miguel Brendl Joel Brockner Jonathon D. Brown Rupert Brown Ryan P. Brown Angela Bryan
Mara Cadinu Larry Cahill Kevin Carlsmith Don E. Carlston Nicholas L. Carnagey Andrea Carnaghi Charles S. Carver Emanuele Castano Luigi Castelli Daniel Cervone Martin Chemers Serena Chen Robert Cialdini Gerald Clore Adam B. Cohen Geoffrey Cohen Tamlin Conner Federica Conrey Andy Conway Joshua Correll Catherine Cottrell Seana Coulson Chris Crandall William B. Crano Jean-Claude Croizet Faye Crosby Amy J. C. Cuddy Anne E. Cunningham William Cunningham John C. Cutting Peter Darke Nilanjana Dasgupta David De Cremer Carsten K. W. de Dreu Jan De Houwer Stephanie Demoulin David DeSteno Brian Detwiler-Bedell Roland Deutsch Siegfried Dewitte Lisa Diamond Mark Dickerson Amanda Diekman Karen M. Douglas Geraldine Downey Muriel Dumont R. I. M. Dunbar Elizabeth W. Dunn
John Eastwood Robert Eisenberger Andy Elliot Nicholas Emler Zachary Estes Leandre Fabrigar Juan Falomir Anne Fausto-Sterling Norman Feather Melissa Ferguson Dan Fessler Klaus Fiedler Andy Field Agneta Fischer Gregory W. Fischer Ayelet Fishbach Alan P. Fiske Monique Fleming Friedrich Foersterling Joseph P. Forgas Francesco Foroni Jens Fo¨rster Shane Frederick Antonio Freitas Ronald S. Friedman Immo Fritsche Susan Fussell Samuel Gaertner Marylene Gagne Adam Galinsky Cindy Gallois Steven Gangestad Leonel Garcia-Marquez Patricia Garcia-Prieto Karen Gasper Bertram Gawronski Nicolas Geeraert Andrew Geers Roger Giner-Sorolla Vittorio Girotto Jack Glaser Jamie L. Goldenberg Peter M. Gollwitzer Gian C. Gonzaga Stephanie Goodwin Ernestine H. Gordjin Sam Gosling Richard Gramzow 1166
Melanie Green Tobias Greitemeyer Dale Griffin James Gross Deborah Gruenfeld Serge Guimond Ana Guinote Geoffrey Haddock Michael Hafner Jonathan Haidt Jamin Halberstadt Judith Hall Eddie Harmon-Jones Christine Harris Monica Harris Martie Haselton Nick Haslam Reid Hastie Curt Haugtvedt Steve Heine Paul Herr Ursula Hess Miles Hewstone Tory Higgins Sara Hodges Wilhelm Hofmann Michael Hogg Rob Holland Thomas Holtgraves Vera Hoorens Kurt Hugenberg Jay Hull Aarti Iyer Jolanda Jetten Remo Job Blair Johnson Carl N. Johnson Joel T. Johnson Lucy Johnston Eva Jonas John T. Jost Lee Jussim Arvid Kappas Minoru Karasawa Andy Karpinski Emiko Kashima Geir Kaufmann
ACKNOWLEDGMENT
Kerry Kawakami Aaron Kay Johannes Keller Markus Kemmelmeier David A. Kenny Nicolas Kervyn Thomas Kessler Tim Ketelaar Jeff Kiesner John Kihlstrom Lisa Kilpatrick Heejung Kim Allan J. Kimmel Shinobu Kitayama Yechiel Klar Joshua Klayman Bill Klein Olivier Klein John P. Kline Eric D. Knowles Sei Jin Ko Vladimir Konecni Sander Koole Ankica Kosic Laura Kray Jon A. Krosnick Joachim I. Krueger Virginia Kwan
Leonard L. Martin Robin Martin David M. Marx Mary Masters David Matsumoto Gail Mauner Iris Mauss Jack McArdle Clark McCauley Gary McClelland Allen McConnell Mike McCullough Craig McGarty Kathleen McGraw Peter McGraw Wendy Berry Mendes Gerold Mikula Dale T. Miller Jason Mitchell Benoit Monin Margo J. Monteith Wesley G. Moons Carey K. Morewedge Gordon Moskowitz Dominique Muller Mark Muraven John D. Murray Jochen Musch
Jessica Lakin Mark Landau Kristin Lane Ellen Langer Rick Larrick Randy Larsen Colin Wayne Leach Mark R. Leary Angela Lee Willy Lens Allison Lenton Howard Leventhal Irwin Levin Sheri R. Levy Jacques Leyens Brian Lickel Debra Lieberman Matthew D. Lieberman Kenneth Livingston Fabio Lorenzi-Cioldi
Steven L. Neuberg Roland Neumann Ian Newby-Clark Sik Hung Ng Paula M. Niedenthal Ara Norenzayan Brian A. Nosek Howard Nusbaum
Joe C. Magee Bertram F. Malle Traci Mann William Marelich Keith Markman Hazel Markus Kerry L. Marsh Carolien Martijn
Gabriele Oettingen Arne Ohman Michael A. Olson Susan Opotow Danny Oppenheimer Stuart Oskamp An Oskarsson Victor Ottati Sabine Otten Jennifer R. Overbeck Brian Parkinson Paul Paulus Keith Payne Francesca Pazzaglia Guido Peeters Cynthia Pickett David Pizarro Jason Plaks Ashby Plant
Beth Pontari Felicia Pratto Deborah A. Prentice Emily Pronin John Pryor Thomas Pyszczynski Diane Quinn Roger Ratcliff Daniel Read Rolf Reber Harry Reis Jason Rentfrow Katherine Reynolds Frederick Rhodewalt Francois Ric Laura S. Richman Courtney Rocheleau Karl Rosengren Michael Ross Klaus Rothermund Paul Rozin Udo Rudolph Janet Ruscher Carey S. Ryan Robert J. Rydell John Sabini Lawrence J. Sanna Julio Santiago Kenneth Savitsky Ulrich Schimmack Brandon Schmeichel Yaacov Schul Christie Scollon Constantine Sedikides Beate Seibt Clive Seligman Todd Shackelford James Y. Shah Phillip Shaver Nicole Shelton James Shepperd David Sherman Michelle Shiota Jim Sidanius Dewitt Siegfried Matthias Siemer Paul J. Silvia Dean K. Simonton Stacey Sinclair Marilyn McKean Skaff Heather J. Smith J. Allegra Smith Pamela K. Smith Richard Sorrentino Russell Spears
v Barbara A. Spellman Adriaan Spruyt Dagmar Stahlberg Charles Stangor Paul C. Stern Jeff Stone Steven Stroessner Art Stukas Eugene Subbotsky Robbie Sutton Janet K. Swim Philip Tetlock Larissa Tiedens Alexander Todorov Zakary Tormala Jessica L. Tracy Jeanne Tsai Tom R. Tyler James Uleman Jeroen Vaes Robert J. Vallerand Leaf Van Boven Stephanie Vance Kees van den Bos Joop Van der Plight Eric van Dijk Ad Van Knippenberg Daan Van Knippenberg Paul van Lange Frank Van Overwalle Leigh Ann Vaughn Bas Verplanken Penny Visser Kathleen Vohs Michaela Waenke Eva Walther Andrew Ward Gregory D. Webster Doug Wedell Thalia Wheatley Ladd Wheeler Katherine White Paul Windschitl Bogdan Wojciszke Wendy Wood Steve Worchel George Wu Natalie Wyer Robert Wyer Vincent Yzerbyt Hanna Zagefka Marcel Zeelenberg Rene Zeelenberg
ACKNOWLEDGMENT
vi
Interpersonal Relations and Group Processes Susanne Abele Linda Albright Mark D. Alicke Nalini Ambady David M. Amodio Cameron Anderson Craig A. Anderson John Archer Jamie Arndt Jo-Anne Bachorowski John A. Bargh Robert S. Baron Manuela Barreto Bruce D. Bartholow Roy F. Baumeister Janet Bavelas Ann M. Beaton Jennifer Beer Jennifer Berdahl Leonard Berkowitz Monica Biernat Rebecca S. Bigler Steven Blader Irene Blair Jennifer G. Boldry Michael H. Bond Jennifer K. Bosson Marcella Boynton Markus Brauer Elisabeth Brauner Jeanne M. Brett Marilynn B. Brewer Joel Brockner Felix C. Brodbeck Ross Buck Roger Buehler Judee K. Burgoon Emanuele Castano Luigi Castelli John Caughlin Chi-Yue Chiu Hoon-Seok Choi Gil Clary Dov Cohen Geoffrey L. Cohen Olivier Corneille Leda Cosmides Catherine Cottrell Phebe Cramer Chris S. Crandall Matthew T. Crawford Jennifer Crocker Michael Cunningham William Cunningham Carolyn E. Cutrona
Mark Davis David De Cremer Michel Desert Lisa Diamond Amanda Diekman Joerg Dietz Kathryn Dindia Geraldine Downey John Duckitt David Dunning Carol Dweck Jennifer L. Eberhardt Nancy Eisenberg Hillary Anger Elfenbein Naomi Ellemers Steve L. Ellyson Nicholas Epley Julie Exline Leandre R. Fabrigar Norman Feather Brooke C. Feeney Frank Fincham Eli J. Finkel Susan T. Fiske Julie Fitness Grainne Fitzsimons Robyn Fivush Donelson R. Forsyth David Funder R. Michael Furr Susan Fussell Faby Gagne Steve Gangestad Ruth Gaunt Bertram Gawronski David Geary Omri Gillath Thomas Gilovich Roger Giner-Sorolla Demis E. Glasford Peter Gollwitzer Karen Gonsalkorale Gian C. Gonzaga Richard Gramzow William Graziano Aiden P. Gregg Dale W. Griffin James Gross Deborah Gruenfeld Ana Guinote Carolyn Hafer Jon Haidt Judith A. Hall
Cheryl L. Harasymchuk Steve Harkins Christine Harris Joshua Hart Steve Heine Ursula Hess Gilad Hirschberger Sara D. Hodges Ying-yi Hong Vera Hoorens Matthew J. Hornsey Pascal Huguet William Ickes Chester Insko Michael Inzlicht Jay W. Jackson Michael Johns Kerri L. Johnson Lee Jussim Satoshi Kanazawa Johan Karremans Harvey J. Keselman Thomas Kessler Shinobu Kitayama Bert Klandermans Yechiel Klar Karl Christoph Klauer Olivier Klein William Klein Chip Knee Sander L. Koole Madoka Kumashiro Larry Kurdek Robert Kurzban Virginia Kwan Jessica Lakin Alan Lambert Richard Larrick Jean-Philippe Laurenceau Loraine F. Lavallee Mark R. Leary Geoffrey J. Leonardelli Kwok Leung Norm Li Allan Lind Robert W. Livingston Penelope Lockwood Brian S. Lowery Richard E. Lucas Geoff MacDonald Diane Mackie Keith Maddox
Joe Magee Robyn Mallett Leonard Martin Robin Martin Michael F. Mascolo David Matsumoto Dan McAdams Allen R. McConnell James K. McNulty Rodolfo (Rudy) Mendoza-Denton Batja Mesquita Dale Miller Norman Miller Daniel C. Molden Benoit Monin R. Matthew Montoya Beth Morling Marian M. Morry Elizabeth Mullen Susan E. Murphy Arie Nadler Janice Nadler Drew Nesdale Steven L. Neuberg Bernard Nijstad Laurie T. O’Brien Daniel O’Leary Mara Olekalns Michael A. Olson Allen M. Omoto Minda Orina Victor Ottati Jennifer Overbeck Michael J. Owren Keith Payne Dan Perlman Katherine W. Phillips Greg Pierce Brad Pinter Ashby Plant Deborah Prentice Tom Pyszczynski Diane M. Quinn Glenn D. Reeder Steve D. Reicher John K. Rempel Janusz Reykowski Kate J. Reynolds Francois Ric Deborah S. Richardson Cecilia Ridgeway Blake Riek
ACKNOWLEDGMENT
Ronald Riggio Richard W. Robins Lee Ross Michael Ross Mark Rubin Laurie A. Rudman Karen D. Rudolph Janet B. Ruscher J. Philippe Rushton Carey S. Ryan Richard M. Ryan Tamar Saguy Barbara Sarason Daan Scheepers Toni Schmader Brandon J. Schmeichel Stefan Schulz-Hardt Marjorie Seaton Jane Sell Robert M. Sellers Guen Semin
John Seta Todd K. Shackelford Sharon Shavitt Ken Sheldon James A. Shepperd David K. Sherman Mark Sibicky Jim Sidanius Eliot R. Smith Peter B. Smith Leanne S. Son Hing Sanjay Srivastava Diederik A. Stapel Richard Street Art Stukas Stefan Stu¨rmer Jerry Suls Seiji Takaku Tania Tam Caroline M. Tancredy Louis G. Tassinary
John M. Tauer Geoff Thomas Francine Tougas Jessica L. Tracy Sisi Tran Paul Trapnell Linda R. Tropp Sarah E. Ullman Jorge Vala Leaf Van Boven Eric Van Dijk Gerben A. Van Kleef Colette Van Laar Eric Vanman Jan Willem van Prooijen Lyn M. Van Swol Mark Van Vugt Maykel Verkuyten Theresa K. Vescio Penny S. Visser
vii Alberto Voci Kathleen Vohs Ulrich Wagner Gregory Walton Duane Wegener Dirk Wentura Nicole E. Werner Tessa V. West Ladd Wheeler Tim Wildschut Kip Williams Christopher Wolsko Steve Wright Gary Yukl Vincent Y. Yzerbyt Stephen Zaccaro Virgil Zeigler-Hill Andreas Zick
Personality Processes and Individual Differences Jennifer L. Aaker William T. Abraham Dominic Abrams Marvin W. Acklin Aaron Ahuvia Ju¨ri Allik Nalini Ambady Cameron P. Anderson Craig Anderson Jennifer Archer Arthur Aron Elizabeth J. Austin Cesar Avila-Rivera Michael Bagby Mark Baldwin Samuel Ball Hans Baumgartner Myriam N. Bechtoldt Jennifer Beer Howard Berenbaum Jeremy Biesanz Gurit E. Birnbaum Bryan Blissmer Charles Bond Peter Borkenau Jennifer K. Bosson Margaret (Peg) Braungart Peter Brecke Marilynn B. Brewer Nathan Brody Kirk Warren Brown Ryan P. Brown
Linda D. Cameron John Campbell W. Keith Campbell Gian Vittorio Caprara Joseph Cesario Cecilia Cheng Sasha Chernyshenko Chi Yue Chiu Incheol Choi Gerald Clore Dov Cohen C. Randall Colvin Luke (Lucian) Conway III Philip J. Corr Catherine Cottrell Kenneth Craik Phebe Cramer Carolyn Cutrona Claudia Dalbert Ken DeBono Mark Dechesne Filip De Fruyt Jan De Houwer Thomas F. Denson Boele De Raad Ed Destaubin Lisa Diamond Sally Dickerson Richard Dienstbier Helga Dittmar M. Brent Donnellan Brian D’Onofrio Geraldine Downey
John Duckitt Lauren Duncan Alice Eagly Collette P. Eccleston Jeffrey R. Edwards Hillary Anger Elfenbein Bruce Ellis Seymour Epstein Leandre Fabrigar Beverley Fehr Allan Fenigstein Pere J. Ferrando Eli J. Finkel Ayelet Fishbach Jens Foerster Patricia Frazier Ron Friedman Frank Fujita Lowell Gaertner Tim W. Gaffney Bertram Gawronski Heidi Gazelle William Gerin Frederick (Rick) Gibbons Betty Glad Karen Glanz Pehr Granqvist William Graziano Melanie C. Green Jeff Greenberg Aiden Gregg
Wendy S. Grolnick James J. Gross Frederick M. E. Grouzet Kathleen C. Gunthert Anne M. Haase Meara Habashi Carolyn Hafer Jon Haidt Michel Hansenne Christine Harris Martie Haselton Nick Haslam Manfred Hassebrauck Adele Hayes Daniel Heller Ravenna Helson Ursula Hess Gilad Hirschberger Geert Hofstede Remus Ilies Linda Jackson Robert Jervis John A. Johnson Wendy Johnson Jeff Joireman Constance Jones Christian Jordan John T. Jost Cheryl Kaiser Todd B. Kashdan
ACKNOWLEDGMENT
viii Yoshi Kashima Aaron Kay Dacher Keltner David Kenny Douglas Kenrick John Kerns Timothy Ketelaar Corey Keyes John Kihlstrom Laura A. King Kris Kirby Shinobu Kitayama Kristen Kling David Klonsky Ariel Knafo Sander Koole Peter Kuppens Virginia S. Kwan Margie E. Lachman Jennifer La Guardia Kevin Lanning Daniel Lapsley Simon Larose Jeff T. Larsen Gary Latham Kimdy Le Angela Y. Lee Kibeom Lee Jennifer Lerner Melvin J. Lerner Chantal Levesque Irwin P. Levin Michael Levine Norman Li Nira Liberman Penelope Lockwood John Loehlin Colin MacLeod Bertram Malle Keith Markman Herbert W. Marsh Grant Marshall Leonard Martin Leslie R. Martin Christina Maslach
David Matsumoto Dan McAdams Edward McAuley Allen McConnell Michael McCullough Sam McFarland Robert E. McGrath Ian McGregor Dean McKay Jeffrey R. Measelle Ivan Mervielde John Michela Elizabeth Midlarsky Roger E. Millsap Beth Morling Mark Muraven Kristin D. Neff Ian R. Newby-Clark Joseph Newman Florrie Fei-Yin Ng Paula Niedenthal Joel Nigg Catherine J. Norris Brian O’Connor Thomas G. O’Connor Shigehiro Oishi Michael Olson Anthony Ong Emda Orr Suzanne Ouellette Clifton M. Oyamot Daniel J. Ozer Crystal Park Lora E. Park Nan-Sook Park James D. A. Parker Del Paulhus Sampo V. Paunonen Ginger L. Pennington Marco Perugini Christopher Peterson Richard E. Petty Alan D. Pickering Gregory Pierce
Paula Pietromonaco Aaron Pincus Jason Plaks Michael Posner Tom Pyszcynski Lisa M. PytlikZillig Karen Quigley Eshkol Rafaeli Glenn D. Reeder Johnmarshall Reeve Steven Reise John K. Rempel Frederick Rhodewalt Kenneth G. Rice Deborah S. Richardson Marsha L. Richins Ronald Riggio John E. Roberts Tom Rodebaugh Glenn Roisman Mary K. Rothbart James A. Russell Lilach Sagiv Jeffrey Sanchez-Burks Kimberly Saudino Steven J. Scher Jeff Schimel Ulrich Schimmack Andreas Schwerdtfeger Christie N. Scollon Paschal Sheeran David Sherman Rebecca L. Shiner Paul Silvia Leonard J. Simms Devendra Singh Luke Smillie Gregory Smith Timothy Smith Annette Stanton Ursula M. Staudinger Joachim Stoeber Peter Suedfeld Eunkook Mark Suh
Cynthia Suveg William B. Swann Janet Swim Jennifer Tackett Romin Tafarodi Maya Tamir Orit Taubman - Ben-Ari J. Kevin Thompson Hulda Thorisdottir Todd M. Thrash Laura Z. Tiedens Eddie Tong Maggie Toplak Harry C. Triandis Jeanne L. Tsai Jo-Ann Tsang Jean Twenge Bert Uchino John Updegraff Tim Urdan Kees van den Bos Alain Van Hiel Simine Vazire Edelyn Verona Bas Verplanken Vivian L. Vignoles Chocklaingam Viveswaran Gifford Weary Bernard Weiner Alexander Weiss Richard F. West Lee Westmaas Keith F. Widaman Kristi Williams Timothy Wilson Connie Wolfe Joanne Wood Wendy Wood Erik Z. Woody Jack C. Wright Robert Wyer John Zelenski Lori A. Zoellner Marvin Zuckerman
ATTITUDES AND SOCIAL COGNITION
Understanding Implicit and Explicit Attitude Change: A Systems of Reasoning Analysis Robert J. Rydell
Allen R. McConnell
University of California, Santa Barbara
Miami University
There is considerable controversy about how to conceptualize implicit and explicit attitudes, reflecting substantial speculation about the mechanisms involved in implicit and explicit attitude formation and change. To investigate this issue, the current work examines the processes by which new attitudes are formed and changed and how these attitudes predict behavior. Five experiments support a systems of reasoning approach to implicit and explicit attitude change. Specifically, explicit attitudes were shaped in a manner consistent with fast-changing processes, were affected by explicit processing goals, and uniquely predicted more deliberate behavioral intentions. Conversely, implicit attitudes reflected an associative system characterized by a slower process of repeated pairings between an attitude object and related evaluations, were unaffected by explicit processing goals, uniquely predicted spontaneous behaviors, and were exclusively affected by associative information about the attitude object that was not available for higher order cognition. Keywords: implicit attitudes, explicit attitudes, attitude change
change through the use of fast-learning, rule-based reasoning, whereas implicit attitudes form and change through the use of slow-learning, associative reasoning (Sloman, 1996). Heretofore, implicit attitude change and explicit attitude change have been studied in relative isolation. Indeed, research on explicit attitude change has been one of the most productive areas of study in social psychology (Eagly & Chaiken, 1993; Petty & Wegener, 1998). Although some researchers have found that implicit attitudes are relatively difficult to change with conventional attitude change manipulations (e.g., Gawronski & Strack, 2004; Gregg, Seibt, & Banaji, 2006; Petty, Tormala, Brin˜ol, & Jarvis, 2006), other research has demonstrated that implicit attitudes can change relatively quickly in response to contextual stimuli or social roles (e.g., Barden, Maddux, Petty, & Brewer, 2004; Dasgupta & Greenwald, 2001; Wittenbrink, Judd, & Park, 2001). But despite these demonstrations, the theory underlying implicit attitude change is relatively underdeveloped (see Devine, 2001; Fazio & Olson, 2003; Wilson, Lindsey, & Schooler, 2000), and experimental paradigms that can systematically examine the concurrent formation and change of implicit and explicit attitudes
The study of attitudes— evaluations of the self, individuals, groups, and other objects— has a long and rich history in social psychology (Eagly & Chaiken, 1993). In recent years, the focus of attitude research has shifted from understanding explicit attitudes (i.e., attitudes that people can report and for which activation can be consciously controlled) to examining implicit attitudes (i.e., attitudes for which people do not initially have conscious access and for which activation cannot be controlled).1 Past research has shown that relying on implicit rather than explicit measures of attitudes can circumvent self-presentational motives (e.g., Dunton & Fazio, 1997) and can often uniquely predict spontaneous behaviors (e.g., McConnell & Leibold, 2001); however, less is known about the processes underlying how implicit and explicit attitudes form and operate. The current work posited that there are important differences between them, especially in how they change. Specifically, we propose that explicit attitudes form and
Robert J. Rydell, Department of Psychology, University of California, Santa Barbara; Allen R. McConnell, Department of Psychology, Miami University. This research was supported by National Institute of Mental Health Grant MH068279 and National Science Foundation Grant BCS 0516931. Portions of this work were submitted by Robert J. Rydell in partial fulfillment of the requirements for a doctoral degree from Miami University. We thank Doris Bergen, Heather Claypool, Dave Hamilton, Kurt Hugenberg, Diane Mackie, Jeff Sherman, and Laura Strain for their extremely helpful comments and guidance in this research. Correspondence concerning this article should be addressed to Robert J. Rydell, Department of Psychology, University of California, Santa Barbara, Santa Barbara, CA 93106-9660. E-mail:
[email protected] 1 Although there is disagreement about the use of the terms implicit attitudes and explicit attitudes in the literature (e.g., Fazio & Olson, 2003), we agree with Strack and Deutsch (2004) who note that “explicit and implicit measures are defined by the cognitive operations that they capture. In this sense, explicit measures tap into people’s knowledge or beliefs, implicit measures tap into their associative structures” (p. 239; see also, Wilson et al., 2000). Because we contrast and compare implicit and explicit measures, we use the terms implicit attitudes and explicit attitudes throughout this article.
Journal of Personality and Social Psychology, 2006, Vol. 91, No. 6, 995–1008 Copyright 2006 by the American Psychological Association 0022-3514/06/$12.00 DOI: 10.1037/0022-3514.91.6.995
995
RYDELL AND MCCONNELL
996
is lacking. The current work seeks to address these issues. Such an advance would not only speak to important theoretical issues (e.g., are the processes underlying implicit and explicit attitude change fundamentally different?), but it has implications for topics ranging from persuasion (e.g., Petty et al., 2006) to intergroup relations (e.g., Jellison, McConnell, & Gabriel, 2004). We now turn to developing a framework that can account for how implicit and explicit attitudes change.
Systems of Reasoning Sloman (1996) proposed a systems of reasoning approach to account for how different cognitive systems affect thinking, language, and behavior (see also, Gawronski & Bodenhausen, 2006; Smith & Decoster, 2000; Strack & Deutsch, 2004). He argued that two independent systems of reasoning use very different processes to operate, learn, and change. The first system of reasoning, the slow-learning system, operates by using paired associations based on similarity and contiguity. In this case, learning is based on the slow accrual of information over time to form and strengthen associations in memory. The second system of reasoning proposed by Sloman, the fastlearning system, relies on logical, verbal, or symbolic representations at a relatively higher order level of cognitive processing. Judgments and behaviors rendered by this system are based on processes requiring at least some degree of conscious control (Strack & Deutsch, 2004). Unlike the slow-learning system, which relies on the accretion of paired associations in memory, the fast-learning system can operate relatively quickly and flexibly to take into account new information that is not associative in nature, but rather, reflects abstractions, language, and logic. In summary, the slow-learning system is characterized by more automatic processes based on the slow accumulation of paired associations in memory, whereas the fast-learning system responds relatively more flexibly and deliberately to abstract information rather than accumulating associations in memory (Smith & DeCoster, 2000). When imported into the attitudes literature, this systems of reasoning approach maps nicely onto implicit attitudes and explicit attitudes. That is, the slow-learning system can shed light on how implicit attitudes form and function because implicit attitudes are posited to follow the basic principles of similarity and paired associations across time (Olson & Fazio, 2001; Smith & DeCoster, 2000; Wilson et al. 2000). On the other hand, the fast-learning system is compatible with explicit attitudes, which can change quickly and often require some degree of cognitive resources in their production and revision (Fazio, 1995; Petty & Wegener, 1998). Indeed, it has been proposed that implicit attitudes and explicit attitudes are the products of different and distinct underlying cognitive processes (Wilson et al., 2000), and accordingly, empirical studies have demonstrated that implicit and explicit attitudes predict different kinds of behavior (spontaneous and nonverbal vs. deliberate and self-presentational, respectively; Dovidio, Kawakami, & Gaertner, 2002; Jellison et al., 2004; McConnell & Leibold, 2001). On the basis of a systems of reasoning account, one would anticipate that implicit and explicit attitudes might be differentially responsive to particular types (nonconscious and associative vs. conscious and verbal, respectively) of attitude– object information. Indeed, Rydell, McConnell, Mackie, and Strain (in press) recently demonstrated that explicit attitudes were formed in response to consciously available information, whereas implicit attitudes
formed in response to the valence of subliminally presented primes when both types of information were available. Specifically, participants were presented with a series of trials in which a target person was preceded by a subliminal prime (either positive or negative in valence), who was described in a sentence as having performed a particular behavior (the valence of which was always opposite of the subliminal prime). After a number of such trials, Rydell et al. found that implicit attitudes toward the person reflected the valence of the subliminal primes, whereas explicit attitudes responded to the valence of the verbally presented behaviors. For example, when presented with negative subliminal primes and positive behavioral sentences, participants reported negative implicit attitudes and positive explicit attitudes toward the same target. Consistent with a systems of reasoning account, the formation of implicit and explicit attitudes were independent of each other, with each reflecting the type of information (associative and nonconscious vs. verbal and conscious) assumed to influence the slow-learning and fast-learning systems, respectively. These findings are difficult to explain by attitude theories that do not assume that people can simultaneously hold different implicit and explicit attitudes about the same object (e.g., Fazio & Olson, 2003; Petty & Wegener, 1998).2 Despite this evidence of independent implicit and explicit attitude formation, the question remains as to what processes underlie the formation and change of implicit and explicit attitudes. In other words, although Rydell et al. (in press) established the independence of implicit and explicit attitudes, they did not evaluate whether slowlearning and fast-learning systems (respectively) account for these outcomes. The current research focuses directly on this issue. It was anticipated that, in general, implicit attitudes would change more slowly than explicit attitudes in response to targetrelevant information because implicit attitudes reflect the slow accrual of paired associations in memory. Although this should be the case when information is presented so that it can be acted on by higher order cognition, there should also be situations (e.g., information is presented outside of conscious awareness) in which attitude-relevant information will impact implicit attitudes but not explicit attitudes, reaffirming their dissociation (Rydell et al., in press). Another consequence of the dissociation between implicit and explicit attitudes should be revealed in the types of behaviors they predict, with implicit attitudes uniquely predicting subtle, less deliberate behavior (e.g., social distance) and explicit attitudes uniquely predicting more thoughtful actions (e.g., Dovidio et al., 2002; McConnell & Leibold, 2001). Earlier, we noted that several studies have shown that reports of implicit attitudes can change relatively quickly. This raises the question of how a systems of reasoning approach would explain abrupt shifts in implicit attitudes in response to positive exemplars (Dasgupta & Greenwald, 2001), contextual features (Wittenbrink et al., 2001), or 2
Instead, these theories assume that people hold an attitude about an object in memory whose expression can be adjusted to accommodate self-presentational concerns, differences in motivation and cognitive resources, societal norms, or persuasive communications. In summary, these models assume that implicit measures reflect an association between an attitude object and its evaluation in memory, whereas explicit measures elucidate more “downstream” consequences of accessing the attitude (Fazio & Olson, 2003).
IMPLICIT AND EXPLICIT ATTITUDE CHANGE
social roles (Barden et al., 2004). In our view (see also, Gawronski & Bodenhausen, 2006; Mitchell, Nosek, & Banaji, 2003; Richeson & Trawalter, 2005), implicit attitudes, just like any other memorial structure, can be affected by priming manipulations that increase the accessibility of a subset of information associated with an attitude object (e.g., increasing the accessibility of positive members of a stigmatized group) or even affect how attitude objects are classified (e.g., encountering nonprototypic group members may temporally impact the categorization of subsequent group members). Thus, situational factors may alter the accessibility of associations related to an attitude object without changing the attitude in an enduring fashion. Instead, these factors may affect which information about an attitude object is activated in memory and may also change the standards used for category membership. Although these effects are important to explore and can speak to the underlying mechanisms of attitudes and attitude activation (especially for well-established attitude objects where many of these outcomes, such as temporarily increasing the accessibility of a subset of group members, are possible), the current work was concerned with understanding how slow-learning and fast-learning processes can account for how implicit and explicit attitudes change. Thus, we had participants learn about a novel attitude object under conditions in which we could manipulate the learning history of the attitude object, allowing us to examine the basic mechanisms through which implicit and explicit attitudes form and change.
Overview of the Current Work Five experiments were conducted to understand whether slowlearning and fast-learning systems could account for implicit and explicit attitudes. To examine a systems of reasoning approach to attitudes, we gave participants information about a novel target person (Bob) in a learning paradigm that initially presented considerable behavioral information about Bob before revealing counterattitudinal behavioral information about him (i.e., behavioral information inconsistent with the valence of the initial information). Afterward, participants reported their implicit and explicit attitudes toward Bob. We sought to understand when and how counterattitudinal information affected implicit and explicit attitudes differently. Experiment 1 examines the conditions under which quick changes in explicit attitudes, but not implicit attitudes, are found. Experiment 2 examines conditions under which implicit attitudes do change in response to counterattitudinal information and how these changes differ from those observed for explicit attitudes in response to the same information. In Experiment 3, we focus on how providing explicit processing goals for forming impressions affects explicit attitudes but not implicit attitudes. Experiment 4 examines how implicit and explicit attitudes toward Bob in the current paradigm uniquely predict different types of behaviors directed toward him. Finally, Experiment 5 uses a modified learning paradigm (incorporating subliminal priming and measuring attitudes at two different times; similar to Rydell et al., in press) to demonstrate conditions under which implicit attitudes, but not explicit attitudes, changed in the face of counterattitudinal information about Bob.
Experiment 1 Experiment 1 was designed to demonstrate that implicit and explicit attitudes reflect different systems of reasoning by present-
997
ing counterattitudinal information to change explicit attitudes (assumed to be governed by the fast-learning system) but not implicit attitudes (presumably governed by the slow-learning system). This experiment was modeled after an elegant study by Kerpelman and Himmelfarb (1971) in which participants were randomly assigned to receive positive reinforcement (i.e., positive behaviors are characteristic of the target and negative behaviors are uncharacteristic of the target) about the behaviors performed by a target person 100%, 80%, or 70% of the time over a block of 100 trials (with the remaining trials providing negative counterattitudinal feedback in the latter two conditions). After this initial learning, participants reported their explicit attitudes toward the target person or they learned in 50 subsequent trials that the target person performed additional behaviors that were of the opposite valence from the majority of the first 100 and then reported their explicit attitudes. Participants in the 100% reinforcement condition displayed a drastic and almost immediate change in their evaluations of the target person in the direction opposite to the originally learned attitude. Thus, to the extent that original explicit attitudes were more extreme because of initially greater consistency in levels of reinforcement, participants showed greater shifts in their explicit attitudes in line with the counterattitudinal information presented. However, because Kerpelman and Himmelfarb (1971) only examined how positive attitudes were changed by negative counterattitudinal information, it is also possible that the processes involved in explicit attitude change in this paradigm are more complex than they acknowledged. Indeed, there is reason to believe that negative counterattitudinal information (i.e., learning negative information following mostly positive initial information) about a target will change attitudes more strongly than positive counterattitudinal information (i.e., learning positive information following mostly negative initial information). Notably, for social judgments involving liking (like those used in the current work), negative information receives greater emphasis and is more crucial in forming impressions (Fiske, 1980; Skowronski & Carlston, 1987). Although these negative asymmetries have been shown for explicit attitudes, it is an open question as to whether they also occur for implicit attitudes. For example, one could argue that implicit attitudes would also be more impacted by negative counterattitudinal information because of its greater attention-grabbing value (e.g., Pratto & John, 1991). However, although negative behaviors are more diagnostic for liking judgments, positive behaviors are more diagnostic for ability judgments (Skowronski & Carlston, 1987). Thus, a simple “negative information is given more weight” explanation seems insufficient. Also, because extracting a trait from behavior may rely on some amount of effortful processing (Bassili & Smith, 1986) and may require verbal processes (Carlston, 1994), it is possible that valence asymmetries are more likely for the fast-learning system than for the slow-learning system. With these latter points in mind, our prediction was that explicit attitudes were more likely to reveal a valence asymmetry (i.e., stronger attitude change following negative counterattitudinal exposure than that following positive counterattitudinal exposure) than implicit attitudes. Thus, we used a learning paradigm similar to that of Kerpelman and Himmelfarb (1971) because it provides a useful way to study how implicit attitudes and explicit attitudes about the same attitude object are formed and changed differently on the basis of the same information. First, participants received a considerable amount of information about Bob, allowing them to form implicit attitudes
998
RYDELL AND MCCONNELL
toward him. Second, the introduction of the counterattitudinal information provided a window in which explicit (which relies on the fast-learning system) but not implicit (which relies on the slow-learning system) attitudes should change in the face of new target-relevant information. Thus, we have an opportunity to assess and understand how this new information affects implicit attitudes and explicit attitudes differently, shedding light on the processes involved in their change. We expected to observe that people would quickly change their explicit attitudes in the face of counterattitudinal information, especially when the initial learning was very consistent (Kerpelman & Himmelfarb, 1971) and when the counterattitudinal information was negative (Fiske, 1980). However, we did not expect implicit attitudes to change as quickly in response to a modest amount of counterattitudinal information nor did we expect to observe a valence asymmetry for implicit attitudes.
Method Participants. A sample of 170 undergraduates at Miami University participated in return for research credit in their introductory psychology courses. They were randomly assigned to a 2 (valence of learned attitude: positive vs. negative) ⫻ 2 (level of reinforcement: 100%, 75%) ⫻ 2 (counterattitudinal condition: control vs. counterattitudinal conditioning) between-subjects factorial. Learning task. The current work used a modified version of the attitude learning paradigm developed by Kerpelman and Himmelfarb (1971). Specifically, participants were presented with a target person’s behaviors that were either relatively positive or negative in valence, and participants judged whether each behavior was characteristic or uncharacteristic of him. As part of a between-subjects manipulation, participants were given different levels of reinforcement in their responses, leading them to form different attitudes toward him. First, participants completed the learning task on a computer, in which they were told that they would be receiving information about a person named Bob. In the initial learning trials, participants read 100 behaviors performed by Bob while a picture of Bob was presented on the computer monitor directly above each behavior.3 After reading each behavior, participants indicated whether they believed that the behavior was characteristic or uncharacteristic of Bob by pressing the C key (characteristic) or the U key (uncharacteristic). After they responded, participants were given feedback about whether the behavior was characteristic of Bob for 5 s. Specifically, feedback consisted of the word correct (in blue text) or incorrect (in red text) positioned in the center of the computer monitor and, at the same time, the behavior was restated “correctly,” on the basis of the assigned reinforcement condition, at the bottom of the computer monitor (e.g., “Helping the neighborhood children is characteristic of Bob.” or “Helping the neighborhood children is uncharacteristic of Bob.”). In the initial 100 learning trials, the feedback given portrayed Bob as positive or as negative in 100% or in 75% of the behaviors (with 25 of the trials in the 75% reinforcement condition being counterattitudinal). The ordering of the behaviors and feedback were randomly determined (in accordance with the experimental condition) for each participant. Following these 100 trials, participants in the control condition received 20 neutral trials (i.e., the behavior performed by Bob was neither positive nor negative; e.g., “Bob waited at the street corner.”). However, participants in the counterattitudinal condition (20 CA) received counterattitudinal feedback about Bob on 20 trials (i.e., the behaviors that were described as characteristic or uncharacteristic of Bob were opposite of the valence presented during the initial learning trials). Finally, participants completed implicit and explicit attitude measures.4 Explicit attitude measure. To assess explicit attitudes, participants judged how likable Bob was on a scale ranging from 1 (very unlikable) to
9 (very likable). In addition, they completed five semantic differential scales, each using a 9-point scale to describe Bob: good– bad, pleasant– mean, agreeable– disagreeable, caring– uncaring, and kind– cruel. Further, participants provided their evaluation of Bob on a feeling thermometer that ranged in temperature from 0o to 100o. The response for each explicit measure was standardized and an overall mean was computed (in all experiments to be reported, ␣s ⬎ .90). Then the standardized scores in the negative valance condition were reverse scored so that greater scores on this measure indicated that explicit attitudes were more extreme in the direction of initial learning. Implicit attitude measure. The Implicit Associations Test (IAT; Greenwald, McGhee, & Schwartz, 1998) was used to assess implicit attitudes toward Bob. The IAT had 26 stimuli: 1 picture of Bob, 5 different pictures of White men who were not Bob, 10 positive adjectives (e.g., wonderful), and 10 negative adjectives (e.g., disgusting). All stimuli were presented in the center of the monitor, and the adjectives were always presented in lowercase letters. This IAT task was a modified version of the task used by McConnell and Leibold (2001), featuring seven blocks with 20 trials per block. Participants were informed that the task involved making category judgments for a variety of stimuli (photos or words) presented on a computer monitor by using one of two responses (the D or K keys on the keyboard). During each block, category label reminders were displayed on the left and right sides of the display (assignment of particular labels to the D and K keys was counterbalanced across participants and produced no effects). Participants were instructed to complete that task quickly while also minimizing errors, and they were told to keep their index fingers on the D and K keys throughout the experiment to minimize delays in responding. There was a 250-ms intertrial interval. In Block 1, participants judged photos of Bob or not Bob and in Block 2 they judged whether the adjectives were “negative” or “positive.” In Blocks 3 and 4 (Combination 1), participants judged whether the stimuli were “Bob or negative” or “not Bob or positive.” In Block 5, participants performed the same judgment task as Block 2 except the assignment of response keys assigned to the two valence categories was reversed. Finally, in Blocks 6 and 7 (Combination 2), participants judged whether the stimuli were “Bob or positive” or “not Bob or negative.” As in past IAT research, half of the participants performed Combination 1 in Blocks 3– 4 and Combination 2 in Blocks 6 –7, whereas the rest performed Combination 2 in Blocks 3– 4 and Combination 1 in Blocks 6 –7 (this counterbalancing manipulation produced no effects). In order to assess implicit attitudes toward Bob, we subtracted the mean response latencies of Combination 2 from the mean response latencies of Combination 1 (regardless of the order they were completed).5 Again, the
3
Photographs of one of 5 different White males were randomly presented as Bob. These 5 White males were judged as equal in attractiveness and the target used did not affect the results in any of the experiments. The positive and the negative behaviors used in the current work were borrowed from those developed by McConnell, Sherman, and Hamilton (1994a). 4 In all of the experiments, half of the participants completed the implicit measure first and the other half completed the explicit measure first. This order variable produced no effects in any of the studies and thus is not discussed further. 5 Following Greenwald et al. (1998), all trials in the critical blocks were retained, responses faster than 300 ms were recoded as 300 ms, and trials slower than 3,000 ms were recoded as 3,000 ms. After any such adjustments were made, each latency was then log transformed to reduce positive skew inherent in response latency data (Fazio, 1990). Alternative scoring techniques for the IAT (e.g., Greenwald, Nosek, & Banaji, 2003) produced the same results in all studies reported. Analyses were performed on the log-transformed values, but means are reported as standardized scores.
IMPLICIT AND EXPLICIT ATTITUDE CHANGE standardized scores in the negative valance condition were reverse scored so that greater scores on this measure indicated that implicit attitudes were more extreme in the direction of initial learning.
Results Explicit attitudes. To examine whether explicit attitudes changed in response to small amounts of counterattitudinal information and were more likely to show attitude change with greater initial reinforcement, a 2 (valence of learned attitude) ⫻ 2 (level of reinforcement) ⫻ 2 (counterattitudinal condition) analysis of variance (ANOVA) was conducted on explicit attitude extremity (see Figure 1). First, there were significant main effects of level of reinforcement, F(1, 162) ⫽ 39.22, p ⬍ .001, and of counterattitudinal condition, F(1, 162) ⫽ 89.90, p ⬍ .001. As one would expect, the main effect of level of reinforcement showed that explicit attitudes were more extreme in the direction of initial learning in the 100% reinforcement condition (M ⫽ 0.94, SD ⫽ 0.62) than in the 75% reinforcement condition (M ⫽ 0.54, SD ⫽ 0.45). Similarly, the main effect of counterattitudinal condition revealed that explicit attitudes were more extreme in the direction of initial learning in the control condition (M ⫽ 1.04, SD ⫽ 0.52) than in the 20 CA condition (M ⫽ 0.44, SD ⫽ 0.47). More important, the anticipated two-way interaction between level of reinforcement and counterattitudinal condition was significant, F(1, 162) ⫽ 19.06, p ⬍ .001. To examine this interaction, the simple effect of counterattitudinal condition was examined for each level of reinforcement. In the 75% reinforcement condition, there was a simple effect of counterattitudinal condition, F(1, 162) ⫽ 13.54, p ⬍ .001, showing that participants in the control condition had more extreme explicit attitudes toward Bob (M ⫽ 0.71, SD ⫽ 0.42); than participants in the 20 CA condition (M ⫽ 0.38, SD ⫽ 0.41). In the 100% reinforcement condition, there was an even stronger effect of counterattitudinal condition, F(1, 162) ⫽ 97.03, p ⬍ .001, indicating that although participants in the control condition had especially extreme explicit attitudes (in the direction of initial conditioning) toward Bob (M ⫽ 1.37, SD ⫽ 0.37), counterattitudinal information led to far less extreme attitudes toward Bob (M ⫽ 0.51, SD ⫽ 0.52). Thus, the interaction reflects the much larger effect of counterattitudinal condition on explicit 2
999
attitude extremity in the 100% reinforcement condition than in the 75% reinforcement condition (replicating Kerpelman & Himmelfarb, 1971). Consistent with negative asymmetries, the two-way interaction between counterattitudinal condition and valence of learned attitude was also significant, F(1, 162) ⫽ 16.64, p ⬍ .001. In the positive learned attitudes condition, those in the control condition had far more extreme explicit attitudes (M ⫽ 1.17, SD ⫽ 0.59) than those in the 20 CA condition (M ⫽ 0.28, SD ⫽ 0.47), F(1, 162) ⫽ 76.70, p ⬍ .001. In the negative learned attitudes condition, this effect was significant but weaker, with those in the control condition having more extreme explicit attitudes (M ⫽ 0.91, SD ⫽ 0.39) than those in the 20 CA condition (M ⫽ 0.61, SD ⫽ 0.42), F(1, 162) ⫽ 8.12, p ⬍ .005. In other words, negative counterattitudinal information had a greater impact on attitude extremity than did positive counterattitudinal information (e.g., Fiske, 1980; Skowronski & Carlston, 1987). No other effects were significant. Implicit attitudes. As with the explicit attitude data, a 2 (valence of learned attitude) ⫻ 2 (level of reinforcement) ⫻ 2 (counterattitudinal condition) ANOVA was conducted on implicit attitude extremity (see Figure 2). In stark contrast to the explicit attitudes, the interaction of reinforcement and counterattitudinal condition and the interaction of valence of learned attitude and counterattitudinal condition were not significant for implicit attitudes (Fs ⬍ 1). In fact, the only effect to obtain for implicit attitudes was an effect showing the that grand mean was significantly different than zero, F(1, 166) ⫽ 55.12, p ⬍ .001 (M ⫽ 0.50, SD ⫽ 0.87). This shows that participants formed implicit attitudes about Bob in accordance with the valence of their initial learning but that subsequent counterattitudinal information had no impact on them. It is important that this effect was not statistically moderated by any of the experimental manipulations, showing no evidence of changes in attitude extremity or negative asymmetries for implicit attitudes.6
Discussion A systems of reasoning conceptualization of attitude change was supported in this experiment because explicit attitudes were changed dramatically by the introduction of counterattitudinal information, whereas implicit attitudes were unaltered by this same information. This suggests that explicit attitudes are the product of a fast-learning system, whereas implicit attitudes reflect a slow-learning system. In this study, participants did form implicit attitudes about Bob, but,
Explicit Attitude Extremity
1.75 1.5 1.25 75% Reinforcement 100% Reinforcement
1 0.75 0.5 0.25 0 Control
20 CA
Positive
Control
20 CA
Negative
Figure 1. Explicit attitude extremity as a function of reinforcement and counterattitudinal condition (20 CA) in Experiment 1. Values for the negative initial learning condition have been reverse scored to reflect attitude extremity.
6 When implicit and explicit attitude measures were simply standardized (i.e., the standardized attitudes in the negative valence of learned attitude condition were not reverse scored) and submitted to a 2 (valence of learned attitude) ⫻ 2 (level of reinforcement) ⫻ 2 (counterattitudinal condition) ⫻ 2 (standardized attitude measure: implicit vs. explicit, a repeated measure) mixed-model factorial ANOVA, the expected four-way interaction was significant, F(1, 156) ⫽ 3.98, p ⬍ .05, reflecting differential responses to counterattitudinal feedback for explicit attitudes and implicit attitudes. In all subsequent experiments, similar omnibus analyses were conducted by using the standardized attitude measure as a within-subjects factor, and the highest order interaction obtained in each experiment (Fs ⬎ 3.88, ps ⬍ .03). These analyses reveal that examining implicit and explicit attitudes separately throughout the article is justified inferentially. In the current work, we present the data as examining attitude extremity by reverse scoring the negative learning condition attitude measures in order to simplify the presentation of how implicit and explicit attitudes are differentially affected by our manipulations.
RYDELL AND MCCONNELL
1000 2 Implicit Attitude Extremity
1.75 1.5 1.25 75% Reinforcement 100% Reinforcement
1 0.75 0.5 0.25 0 Control
20 CA
Positive
Control
20 CA
Negative
Figure 2. Implicit attitude extremity as a function of reinforcement and counterattitudinal condition (20 CA) in Experiment 1. Values for the negative initial learning condition have been reverse scored to reflect attitude extremity.
unlike explicit attitudes, they were unaffected by the introduction of counterattitudinal information. In addition, the current experiment replicated past work on learned attitudes and research on negative asymmetries in impression formation (e.g., Kerpelman & Himmelfarb, 1971; Skowronski & Carlston, 1987), but these effects were observed only for explicit attitudes (and not for implicit attitudes). That is, explicit attitudes showed the greatest change in the face of contradictory information when negative counterattitudinal information followed initially positive feedback and when it came on the heels of consistent feedback in general.
Experiment 2 Although Experiment 1 demonstrates that explicit and implicit attitudes were differentially impacted by counterattitudinal information (with explicit attitudes changing quickly and implicit attitudes remaining unaffected by it), it did not test the systems of reasoning derived supposition that implicit attitudes will change slowly if sufficient counterattitudinal information is encountered. Showing that implicit attitudes do change following substantial counterattitudinal information would provide important support for our systems of reasoning approach. If implicit attitudes are the product of a slow-learning system, they should change when enough counterattitudinal information is encountered. Because there was no significant effect of counterattitudinal information on implicit attitudes in Experiment 1 through the use of just 20 counterattitudinal behaviors (when compared with the control group), Experiment 2 presented some participants with considerably more counterattitudinal information. We expected that explicit attitudes would change quickly in the face of a small amount of counterattitudinal information, whereas implicit attitudes would remain relatively unaffected (replicating Experiment 1). However, we expected that providing participants with a large amount of counterattitudinal information would eventually lead to implicit attitude change as well.
They were randomly assigned to a 2 (valence of learned attitude: positive, negative) ⫻ 2 (level of reinforcement: 100%, 75%) ⫻ 3 (counterattitudinal condition: control, 20 CA, 100 CA) between-subjects factorial. Procedure. All materials, methods, and measures (and scoring of the measures) paralleled those used in Experiment 1, with the exception that, in the current experiment, there are three levels of counterattitudinal learning and all participants received 100 initial learning trials followed by 100 additional trials. The control condition and the 20 CA condition were the same as those used in Experiment 1 (except that the final 100 or 80 descriptions of Bob, respectively, were neutral so that all participants received the same number of trials). In addition, there was another counterattitudinal condition (100 CA) in which participants received 100 trials of counterattitudinal feedback. Thus, in the 100 CA condition, participants had much more information that was inconsistent with the initially learned attitude than in the other two learning conditions, which should lead to implicit attitude change consistent with the valence of the counterattitudinal information.
Results Explicit attitudes. To examine explicit attitude change in response to counterattitudinal information, a 2 (valence of learned attitude) ⫻ 2 (level of reinforcement) ⫻ 3 (counterattitudinal condition) ANOVA was conducted on the explicit attitude extremity score (see Figure 3). First, main effects of level of reinforcement, F(2, 174) ⫽ 36.09, p ⬍ .001, and of counterattitudinal condition, F(2, 174) ⫽ 70.99, p ⬍ .001, were observed. Not surprisingly, explicit attitudes were more extreme in the direction of initial learning in the 100% reinforcement condition (M ⫽ 0.71, SD ⫽ 0.81) than in the 75% reinforcement condition (M ⫽ 0.20, SD ⫽ 0.73). Also, the main effect of counterattitudinal condition showed that explicit attitudes were more extreme in the direction of initial learning in the control condition (M ⫽ 1.09, SD ⫽ 0.64) than in the 20 CA condition (M ⫽ 0.37, SD ⫽ 0.59) and in the 100 CA condition (M ⫽ ⫺0.09, SD ⫽ 0.70), with all means significantly different.7 It is important that the expected interaction of these two effects obtained, F(2, 174) ⫽ 9.24, p ⬍ .001. In the 75% reinforcement condition, there was a simple effect of counterattitudinal condition, F(2, 174) ⫽ 31.90, p ⬍ .001, showing that participants in the control condition had more extreme explicit attitudes toward Bob (M ⫽ 0.69, SD ⫽ 0.54) than those in the 20 CA condition (M ⫽ 0.38, SD ⫽ 0.57) and those in the 100 CA condition, who had a significantly less extreme view of Bob (M ⫽ ⫺0.44 SD ⫽ 0.55) than those in the control condition or in the 20 CA condition. In the 100% reinforcement condition, there was also a simple effect of counterattitudinal condition, F(2, 174) ⫽ 45.55, p ⬍ .001. This effect found that participants in the control condition had relatively extreme explicit attitudes toward Bob (M ⫽ 1.50, SD ⫽ 0.45); however, the presentation of counterattitudinal information led participants to have less extreme attitudes toward Bob, which did not vary between the 20 CA (M ⫽ 0.37, SD ⫽ 0.62) and 100 CA (M ⫽ 0.24, SD ⫽ 0.68) conditions. Revealing the expected negative asymmetries, the two-way interaction between counterattitudinal condition and valence of learned attitude was also significant, F(2, 174) ⫽ 3.94, p ⬍ .03. In the positive learned attitudes condition, those in the control condition had significantly more extreme explicit attitudes (M ⫽ 1.21, SD ⫽ 0.79) than did those in the 20 CA condition (M ⫽ 0.36,
Method Participants. A sample of 186 undergraduates at Miami University participated in return for research credit in their introductory psychology courses.
7
All post hoc tests described as significant differed at the .05 level with Tukey’s honestly significant difference.
IMPLICIT AND EXPLICIT ATTITUDE CHANGE
1.5 1 75% Reinforcement 100% Reinforcement
0.5 0 -0.5 -1 Control 20 CA 100 CA Control 20 CA 100 CA Positive
Negative
Figure 3. Explicit attitude extremity as a function of reinforcement and counterattitudinal conditions (20 CA, 100 CA) in Experiment 2. Values for the negative initial learning condition have been reverse scored.
SD ⫽ 0.45), who had significantly more extreme attitudes than those in the 100 CA condition (M ⫽ ⫺0.25, SD ⫽ 0.64), F(2, 174) ⫽ 42.34, p ⬍ .001. In the negative learned attitudes condition, those in the control condition had significantly more extreme explicit attitudes (M ⫽ 0.96, SD ⫽ 0.40) than did those in the 20 CA (M ⫽ 0.40, SD ⫽ 0.71) and 100 CA (M ⫽ 0.06, SD ⫽ 0.74) conditions. Thus, as in Experiment 1, explicit attitudes were changed more strongly in response to negative counterattitudinal information than to positive counterattitudinal information. Implicit attitudes. Implicit attitude extremity was calculated in the same fashion as in Experiment 1, and it was analyzed in a 2 (valence of learned attitude) ⫻ 2 (level of reinforcement) ⫻ 2 (counterattitudinal condition) ANOVA (see Figure 4). The only effect to obtain was the expected main effect of counterattitudinal condition, F(2, 174) ⫽ 5.02, p ⬍ .01. That is, implicit attitudes in the 100 CA condition (M ⫽ ⫺0.23, SD ⫽ 0.97) were significantly less consistent with the direction of initial learning than were those in the control condition (M ⫽ 0.27, SD ⫽ 0.88) and 20 CA condition (M ⫽ 0.23, SD ⫽ 1.09), which did not differ. As expected, these results show that implicit attitudes did change when sufficient counterattitudinal information (100 CA) was presented. Yet replicating Experiment 1, there was no difference in implicit attitude extremity between the control condition and the 20 CA condition, and once again there was no evidence of negative asymmetry effects (i.e., stronger attitude change when negative information follows initial positive information).
Discussion The results of Experiment 2 show that implicit attitudes change if sufficient counterattitudinal information is encountered. Because we assume that implicit attitudes reflect the totality of the evaluative information associated with an attitude object, a small amount of counterattitudinal information should have little impact in modifying one’s implicit attitudes (i.e., the 20 CA condition). However, once the totality of the counterattitudinal information increased sufficiently (i.e., the 100 CA condition), implicit attitudes did show substantial change. These findings provide evidence that different systems of reasoning are responsible for changing implicit attitudes and explicit attitudes. In response to counterattitudinal information, implicit atti-
tudes changed in line with the slow-learning system, whereas explicit attitudes changed more quickly, consistent with the fast-learning system. Further, we replicated the findings of Experiment 1 for valence asymmetries (e.g., Skowronski & Carlston, 1987) and for greater attitude change following relatively more consistent initial reinforcement (e.g., Kerpelman & Himmelfarb, 1971), but once again, only for explicit attitudes. Although Experiment 2 showed that implicit attitudes were changed by the sufficient presentation of counterattitudinal information, the results of Experiment 2 also show that participants who received less consistent reinforcement (i.e., 75% condition) continued to show explicit attitude change in response to 100 pieces of counterattitudinal information, and those who received consistent reinforcement (i.e., the 100% condition) did not. Why might this occur? We propose that those in the 75% condition may have forestalled judgments of Bob and continued to effortfully process more counterattitudinal information about him. Although perceivers typically form fast on line impressions of individuals (McConnell, Sherman, & Hamilton, 1994b, 1997), they do so because they expect considerable consistency in their behaviors (McConnell, 2001; McConnell et al., 1997). Thus, it is possible that the current 75% level of reinforcement condition provides sufficient inconsistency as to lead perceivers to delay forming their impressions of Bob. In order to test this explanation, Experiment 3 experimentally manipulates the presumed impression formation theory involved to test whether “rushing to judgment” versus “forestalling judgment” could account for the pattern of explicit attitude data observed in the 100% and 75% reinforcement conditions, respectively. If participants are instructed to forestall judgments, then more linear (rather than asymptotic) explicit attitude change should be observed across the conditions (i.e., control, 20 CA, and 100 CA) regardless of level of reinforcement. And similarly, participants who receive rush to judgment instructions should show more asymptotic (than linear) explicit attitude change regardless of level of reinforcement. But more important, Experiment 3 allows us to examine the extent to which explicit processing goals affect implicit and explicit attitudes. If explicit attitudes are the product of a fastlearning system, deliberate processing instructions should affect explicit attitudes but not implicit attitudes. Thus, even though
0.5 0.4 Implicit Attitude Extremity
Explicit Attitude Extremity
2
1001
0.3 0.2 0.1 0 -0.1 -0.2 -0.3 -0.4 -0.5 Control
CA 20 Positive
CA 100
Control
CA 20
CA 100
Negative
Figure 4. Implicit attitude extremity as a function of reinforcement and counterattitudinal conditions (20 CA, 100 CA) in Experiment 2. Values for the negative initial learning condition have been reverse scored.
RYDELL AND MCCONNELL
1002
participants received the exact same information about the attitude objects, holding different processing goals will likely alter higher order cognitive processes and influence explicit attitudes. Implicit attitudes, because they are based on associations and not on higher order logic, should be relatively unaffected by such deliberate processing goals.
bracketed version appearing below) or that first impressions are often incorrect and biased (bracketed version appearing below). Specifically, participants were told the following: When forming your opinion about what type of person Bob is, you should [not] focus on your first impression of what Bob is like. First impressions are almost always [in]correct when forming an impression about a new person, and using your first impressions allows you to avoid [causes you to make] several biases (or errors in thinking) that are caused by thinking too much [not thinking enough] about what a person is like.
Experiment 3 A systems of reasoning account predicts that explicit attitudes will be affected by conscious processing goals, but implicit attitudes should not be affected by such goals. Because people are able to selectively use and effortfully give more emphasis to practical information about an attitude object and to devalue other types of information on the basis of processing goals, explicit attitudes should be especially amenable to change by altering processing goals. Alternatively, because implicit attitudes are based on the accrual of information about an attitude object and are not based on the selective use of information, they should be relatively immune to the effects of conscious goals because they are devoid of the higher order logic necessary to follow the goal. This experiment also allowed us to examine the results for explicit attitudes of Experiment 2 in more detail. We hypothesized that those in the 75% condition were less inclined to rush to judgment to form an early, on line impression of Bob (leading them to process later information and to modify their attitudes accordingly), whereas those in the 100% condition relied on initial counterattitudinal information, modified their impression quickly and then were less impacted by later information. To evaluate this explanation, participants’ processing goals (i.e., to rely on early information vs. to rely on later information) were experimentally manipulated in Experiment 3. If those in the 75% reinforcement condition of Experiment 2 did adopt the goal of forestalling impression formation of Bob, participants explicitly instructed to do so should be more impacted by later information and report relative greater overall attitude change regardless of the actual consistency of initial learning provided (i.e., 75% or 100% reinforcement). Conversely, those instructed to form an early impression should pay less attention to later information and show less overall attitude change, regardless of the consistency of initial learning.
Results Explicit attitudes. The predicted three-way interaction between level of reinforcement by counterattitudinal condition by first impressions was found, F(2, 101) ⫽ 5.37, p ⬍ .005. Thus, level of reinforcement by counterattitudinal condition ANOVAs were conducted for explicit attitudes in the correct first impressions condition and in the incorrect first impressions condition separately. In the correct first impressions condition, the main effects of level of reinforcement and counterattitudinal condition were both significant, F(2, 101) ⫽ 7.96, p ⬍ .005, and F(2, 101) ⫽ 24.20, p ⬍ .001, respectively. The main effect of level of reinforcement showed that explicit attitudes were more positive in the 100% reinforcement condition (M ⫽ 0.49, SD ⫽ 0.67) than in the 75% reinforcement condition (M ⫽ 0.12, SD ⫽ 0.39). The main effect for counterattitudinal condition showed that explicit attitudes were more positive in the control condition (M ⫽ 0.84, SD ⫽ 0.47) than in either the 20 CA (M ⫽ 0.01, SD ⫽ 0.40) or 100 CA conditions (M ⫽ ⫺0.02, SD ⫽ 0.34). It is important that the interaction was also significant, F(2, 101) ⫽ 6.55, p ⬍ .005. As seen in Figure 5, participants in the 75% reinforcement condition showed more positive explicit attitudes toward Bob in the control condition (M ⫽ 0.44, SD ⫽ 0.31) than in the 20 CA condition (M ⫽ ⫺0.12, SD ⫽ 0.33) and 100 CA condition (M ⫽ 0.02, SD ⫽ 0.27), F(2, 101) ⫽ 8.44, p ⬍ .005. Participants in the 20 CA and 100 CA conditions did not differ. Participants in the 100% reinforcement condition showed more positive explicit attitudes toward Bob in the control condition (M ⫽ 1.20, SD ⫽ 0.23) than in the 20 CA condition (M ⫽ 0.15, SD ⫽ 0.45) and 100 CA condition
Method 1.5 Standardized Explicit Attitudes
Participants. A sample of 113 Miami University undergraduates participated in return for research credit and were randomly assigned to a 2 (first impressions: correct, incorrect) ⫻ 2 (level of reinforcement: 100%, 75%) ⫻ 3 (counterattitudinal condition: control, 20 CA, 100 CA) betweensubjects factorial. Procedure. All materials, methods, and measures paralleled those of Experiment 2 with three exceptions. First, only the positive valence condition was used (thus, the initial attitudes were positive and counterattitudinal information, when presented, was negative). Second, we manipulated instructions for the learning task such that participants were told to rely on either initial or later information in forming their impressions. Third, because there was no negative initial learning condition requiring reverse scoring, we discuss our data in terms of more positive attitudes, as opposed to greater learning-consistent attitude extremity, toward Bob. Manipulating the value of first impressions. To manipulate the importance of early versus late information presented about the target, instructions provided before participants learned any information about Bob noted that first impressions are usually correct and rarely lead to errors (non-
1 0.5 0
75% Reinforcement 100% Reinforcement
-0.5 -1 -1.5 -2 Control
20 CA
100 CA
Counterattitudinal Condition
Figure 5. Explicit attitudes as a function of reinforcement and counterattitudinal conditions (20 CA, 100 CA) for the first impressions are the correct condition in Experiment 3.
IMPLICIT AND EXPLICIT ATTITUDE CHANGE
Standardized Explicit Attitude
1.5 1 0.5 0
75% Reinforcement 100% Reinforcement
-0.5 -1 -1.5 -2 Control
20 CA
1.25 Standardized Implicit Attitudes
(M ⫽ ⫺0.04, SD ⫽ 0.39), F(2, 101) ⫽ 34.56, p ⬍ .001. In other words, replicating Experiments 1–2 and Kerpelman and Himmelfarb (1971), the effect of counterattitudinal information was stronger in the 100% reinforcement condition than in the 75% condition. However, when participants were explicitly told to rely on their first impression, any on-going drop in explicit attitudes in the 100 CA condition for those in the 75% reinforcement condition was not evidenced. As expected, adopting a “rely on first impressions” goal eliminated on-going attitude adjustments previously observed in the 100 CA condition for those receiving 75% reinforcement. In the correct first impressions condition, the Level of Reinforcement ⫻ Counterattitudinal Condition interaction was not significant, F(2, 101) ⫽ 2.50, p ⬎ .10. Instead, the main effects of level of reinforcement and of the effect of counterattitudinal condition were both significant, F(2, 101) ⫽ 97.87, p ⬍ .001, and, F(2, 101) ⫽ 22.54, p ⬍ .001, respectively. As seen in Figure 6, participants in the 100% reinforcement condition had more positive explicit attitudes toward Bob (M ⫽ ⫺0.08, SD ⫽ 1.01) than those in the 75% reinforcement condition (M ⫽ ⫺0.56, SD ⫽ 1.06). Also, overall participants showed more positive explicit attitudes toward Bob in the control condition (M ⫽ 0.93, SD ⫽ 0.62) than in the 20 CA condition (M ⫽ ⫺0.13, SD ⫽ 0.66), which were more positive than their attitudes in the 100 CA condition (M ⫽ ⫺1.20, SD ⫽ 0.62). This stair-step pattern across counterattitudinal conditions indicates that, unlike the correct first impressions, explicit attitude change continued across the entire presentation of counterattitudinal information (regardless of level of initial reinforcement) and did not stop at the end of 20 counterattitudinal pieces of information. Implicit attitudes. The three-way interaction for implicit attitudes was not significant (F ⬍ 1). As Figure 7 reveals, the only significant effect to obtain was the predicted main effect of counterattitudinal condition, F(2, 101) ⫽ 19.89 p ⬍ .001. Replicating Experiment 2, implicit attitudes were more positive (i.e., more strongly in the direction of initial learning) in the control condition (M ⫽ 0.39, SD ⫽ 0.89) and in the 20 CA condition (M ⫽ 0.28, SD ⫽ 0.81) than in the 100 CA condition (M ⫽ ⫺0.75, SD ⫽ 0.88). These results for explicit and implicit attitudes again showed that explicit attitudes were more quickly changed than were im-
1003
1 0.75 0.5 0.25 0 -0.25 -0.5 -0.75 -1 -1.25 Control
20 CA
100 CA
First impressions Correct
Control
20 CA
100 CA
First impressions Incorrect
Figure 7. Implicit attitudes as a function of focus and counterattitudinal conditions (20 CA, 100 CA) in Experiment 3.
plicit attitudes when people encountered counterattitudinal information. The amount of explicit attitude change was not different between 20 CA and 100 CA when participants focused on forming first impressions. However, the increase in counterattitudinal information from 20 to 100 behaviors did lead to greater attitude change when participants were instructed not to rely on forming first impressions. Also, these results again showed that implicit attitudes changed more slowly and only when a sufficient amount of counterattitudinal information was encountered. And as expected, explicit impression formation goals did not affect implicit attitudes, although these verbal instructions had considerable impact on explicit attitudes.
Discussion Experiment 3 again found that explicit attitudes were changed by a different system of reasoning than were implicit attitudes. More specifically, explicit attitudes were altered by the introduction of conscious processing goals but implicit attitudes were not. Consistent with Experiment 2, implicit attitudes did change following the presentation of a substantial amount of counterattitudinal information. However, these implicit attitudes were not affected by verbal processing goals. In addition, the experimental manipulation of processing goals explained why differences in explicit attitudes, as a function of level of reinforcement, were found when a large amount of counterattitudinal information was presented in Experiment 2. It appears that less consistent reinforcement led participants to suspend early judgments of the target individual and attend to later information when forming their impression. The preceding experiments provide evidence that implicit and explicit attitudes were formed and changed by slow-learning and fast-learning processes, respectively. However, what implications do these different attitudes have for behavior? Because an important function of attitudes is to predict behavior (Fazio, 1986), in Experiment 4 we sought to understand the relation between learned implicit and explicit attitudes and target-relevant behavior.
100 CA
Counterattitudinal Condition
Figure 6. Explicit attitudes as a function of reinforcement and counterattitudinal conditions (20 CA, 100 CA) for the first impressions are the incorrect condition in Experiment 3.
Experiment 4 Recent work has begun to tease apart when implicit and explicit attitudes guide behavior. In general, this work has found that implicit attitudes predict subtle, spontaneous behavior, whereas explicit atti-
1004
RYDELL AND MCCONNELL
tudes predict more deliberative, intentional behavior (e.g., Jellison et al., 2004; McConnell & Leibold, 2001). We were interested in whether the attitudes toward Bob created in the current experiments could predict behavior in the same manner as past research. Specifically, would explicit attitudes toward Bob only predict deliberate judgments about him but not predict more subtle forms of behavior (i.e., seating distance)? Similarly, would implicit attitudes toward Bob only predict subtle behaviors but not explicit judgments about him? Experiment 4 tests these predictions, anticipating unique predictive value for implicit and explicit attitudes. These findings could be important for at least three additional reasons. First, past research has shown such double dissociations on the basis of measures of group prejudice (e.g., Dovidio et al., 2002; Jellison et al., 2004); however, this would be the first time that such effects have been shown for a different type of attitude object (i.e., a target person). Second, this previous work has shown these outcomes for groups with preexisting attitudes, whereas this would be the first study to demonstrate such dissociation effects on the basis of attitudes engineered in a controlled laboratory setting. For example, it is possible that cultural prescriptions might shape both implicit prejudice and subtle forms of social behavior toward social group members, providing the appearance of an attitude– behavior relation when, in fact, other factors may produce both. By engineering attitudes in the laboratory without any other targetrelevant knowledge, it is far more likely that behavior reflects the influence of attitudes directly. Finally, if we show that implicit attitudes have unique predictive utility for subtle behavior in this study, then the findings would argue against concerns that our implicit measure has poor sensitivity. One might argue that slow changes on our implicit measure may reflect a relatively weak measure (i.e., it is simply less responsive to change than our explicit measures) rather than a slow-learning system. By establishing that our implicit (but not explicit) attitude measure can uniquely predict theoretically derived types of behavior, we could provide evidence inconsistent with a position that our implicit attitude measure is simply a poor measure.
Method Participants. A sample of 29 undergraduates at Miami University participated in return for research credit in their introductory psychology courses. Participants were randomly assigned to receive no counterattitudinal information about Bob (control) or to receive 20 counterattitudinal pieces of information about Bob (20 CA). Procedure. All materials, methods, and measures paralleled Experiment 1, with these exceptions. First, only the positive valence condition was used, and only the 100% reinforcement condition was used. The two experimental conditions (control and 20 CA) were selected to maximize the discrepancy between implicit and explicit attitudes. In Experiment 1 there was a drastic change in explicit attitudes between the control and the 20 CA conditions, however there was no difference in implicit attitudes between them. Additionally, as in Experiment 3, because there was no negative initial learning condition to reverse score, greater standardized measures of attitudes reflected more positive attitudes toward Bob. In addition to the attitude measures, participants completed explicit judgments of desire for social contact with Bob. Specifically, participants rated the extent to which they would want to have Bob as a neighbor, friend, classmate, roommate, and family member, each on 100-point scales (␣ ⫽ .92). Greater scores on this measure indicated that they wanted more social contact with Bob.
After completing the attitude measures and the explicit social contact judgments, participants were told that they would “have a 2-min get acquainted session with Bob.” They were escorted to a different room in which two chairs were set 221 cm apart. One chair had a book bag and a book next to it (where Bob was supposedly sitting), the other chair (for the participant) was on wheels and set against the wall of the room. The experimenter told each participant, “It looks like Bob has stepped out for a moment. Take that seat against the wall and move it so that you can have a face-to-face conversation with Bob.” Participants took the seat and moved it into a position to converse with Bob. Afterward, they were told that they were not going to meet Bob and were then debriefed. The seating distance between the participant’s chair and the chair where Bob had supposedly been sitting served as our measure of subtle, spontaneous behavior.
Results The attitude measures were examined with one-way ANOVAs of counterattitudinal condition. The only effect to obtain was the predicted effect of counterattitudinal condition for explicit attitudes, F(1, 27) ⫽ 12.86, p ⬍ .005. Replicating the findings of Experiment 1, explicit attitudes were more positive in the control condition (M ⫽ 0.48, SD ⫽ 0.80) than in the 20 CA condition (M ⫽ ⫺0.51, SD ⫽ 0.77), F(1, 27) ⫽ 11.57, p ⬍ .005. In contrast, implicit attitude data did not show an effect of counterattitudinal condition (F ⬍ 1). The effect of counterattitudinal condition for social contact judgments was also examined with a one-way ANOVA. This analysis showed, as expected, that people reported wanting more social contact when they were in the control condition (M ⫽ 74.53, SD ⫽ 15.83) than when they were in the 20 CA condition (M ⫽ 61.21, SD ⫽ 17.71), F(1, 27) ⫽ 4.57, p ⬍ .05. Also, there was no effect of counterattitudinal condition on seating distance (F ⬍ 1). Thus, the counterattitudinal condition manipulation affected deliberate behavior (i.e., desire for social contact) but not the subtle behavior (i.e., seating distance). To examine the main hypotheses, the correlation between explicit attitudes, implicit attitudes, deliberate behavior (i.e., desire for social contact), and subtle behavior (i.e., seating distance) were calculated. As expected, more positive explicit attitudes were related to greater desire for social contact (r ⫽ .71, p ⬍ .001) but were unrelated to seating distance (r ⫽ .04, ns). It is important that more positive implicit attitudes were unrelated to desire for social contact (r ⫽ ⫺.03, ns) but were significantly related to closer seating distance (r ⫽ ⫺.41, p ⬍ .03). Moreover, two multiple regressions analyses were conducted in which explicit and implicit attitudes served to predict desire for social contact (first analysis) and seating distance (second analysis). As predicted, explicit attitudes ( ⫽ 0.70, p ⬍ .001) but not implicit attitudes ( ⫽ ⫺0.01, ns) predicted desire for social contact. On the other hand, implicit attitudes ( ⫽ ⫺0.41, p ⬍ .04) but not explicit attitudes ( ⫽ 0.02, ns) predicted seating distance. Thus, explicit attitudes uniquely predicted deliberate judgments and implicit attitudes uniquely predicted subtle, spontaneous behaviors.
Discussion Experiment 4 showed that the differential formation and change of implicit and explicit attitudes demonstrated in Experiments 1–3 have important implications for predicting behavior toward an attitude object, which in turn, reflect different systems of reason-
IMPLICIT AND EXPLICIT ATTITUDE CHANGE
Experiment 5 Although Experiments 1– 4 establish that implicit and explicit attitudes change at different rates in response to the same information, we more seriously consider whether the observed differential rate of implicit and explicit attitude change reflects a less sensitive implicit attitude measure. Thus, Experiment 5 sought to provide additional evidence showing that implicit, but not explicit, attitudes can be changed to further discredit this alternative account. Specifically, Experiment 5 sought to demonstrate that implicit, but not explicit, attitudes would be affected by nonconscious associations with the attitude object. In this study, two types of information about Bob that should be differentially attended to by a slow-learning system of reasoning (based on associations in memory) and a fast-learning system (based on higher order cognition and logic) were presented. Verbal behavioral information was presented, but unlike the previous experiments, all of this behavioral information was relatively neutral (which should produce relatively neutral explicit attitudes toward Bob). In addition, a valenced prime was presented subliminally before the presentation of Bob’s face (which should shape implicit attitudes toward Bob). A systems of reasoning explanation predicts that implicit attitudes should reflect the valence of the subliminal primes, whereas the fast-learning system should not be affected by the primes when other information is available. This pattern of results would provide compelling evidence that implicit attitudes are changed by associative processes and that the implicit measures in the current work are indeed sensitive.
Method Participants. A sample of 50 undergraduates at Miami University participated in return for research credit in their introductory psychology courses. Participants were randomly assigned to one of two conditions in which they were provided with all neutral behavioral information and were exposed to either positive primes first and then to negative primes later or to negative primes first and then to positive primes later. Procedure. The materials and methods in this experiment were based on those used in Experiments 1– 4 but differed in some key respects. Most notably, a subliminal prime was presented prior to Bob’s picture and the
behavioral information about Bob. Specifically, following a fixation point appearing in the center of the computer monitor for 200 ms, a positive or negative word was presented in the center of the computer monitor for 25 ms (serving as a prime). Next, participants saw a screen with only a picture of Bob for 250 ms, and finally, with the picture remaining on the screen, they were given neutral behavioral information that they judged to be characteristic or uncharacteristic of Bob. The valence of the primes that participants saw was varied systematically to be either unambiguously negative (e.g., death) or positive (e.g., love; Fazio, Sanbanmatsu, Powell, & Kardes, 1986). During the first 100 trials, half of the participants received 10 negative primes 10 times each, and the other half received 10 positive primes 10 times each. During the second 100 trials, the valence of the prime presented was switched such that those who had initially seen the positive primes now saw the negative primes and those who had seen the negative primes now saw the positive primes. Thus, overall all participants saw the same 20 primes (10 positive and 10 negative) 10 times each. In another change from the previous experiments, participants’ implicit and explicit attitudes were assessed at two different times during the session: after the first 100 trials (Time 1) and after the second set of 100 trials (Time 2). However, the attitude measures were identical at both times of assessment and paralleled those of Experiments 1– 4 (in addition, these measures were counterbalanced at both times and this manipulation produced no effects). Finally, participants were given a recognition task for the positive and negative primes after the second assessment of attitudes. They were told that words were presented before Bob’s picture and that we were interested in their ability to detect them. To assess whether participants recognized the words that were flashed on the monitor, we gave them a list of 40 words presented alphabetically (20 actual words, 10 positive and 10 negative, and 20 filler words, 10 positive and 10 negative) from which they chose 20 that they believed could have been presented during the session.
Results The attitude measures were examined separately with 2 (prime order: negative prime first, positive prime first) ⫻ 2 (Time 1, Time 2) mixed-model ANOVAs. As expected, the two-way interaction between condition and time was not found for explicit attitudes, F(1, 48) ⫽ 1.66, ns (see Figure 8). Indeed, no effects were statistically significant for explicit attitudes. As Figure 9 reveals, a very different picture emerged for implicit attitudes. The predicted two-way interaction between condition and time was the only significant effect found for implicit attitudes, F(1, 48) ⫽ 10.02, p ⬍ .005. Thus, simple effects analyses of time were conducted for implicit attitudes in each of the two between-subjects conditions.
1 Standardized Explicit Attitudes
ing. As in the previous experiments, explicit attitudes were affected by the introduction of a small amount of counterattitudinal information, and in the current study, these attitudes uniquely predicted deliberate judgments of the target, whereas implicit attitudes did not. Conversely, implicit attitudes were unaffected by the presentation of a small amount of counterattitudinal information, and these implicit attitudes uniquely predicted spontaneous behaviors (i.e., seating distance) that explicit attitudes did not predict. Moreover, the current study provided clear evidence that the implicit measure is sensitive (i.e., it uniquely predicted subtle behavior), and it showed this double dissociation pattern of predicting behavior for the first time for an individual target and for an attitude object for which there were no preexisting beliefs. Indeed, coupling these results with those of Rydell et al. (in press), in which the same measures of implicit and explicit attitudes were opposite in valence when the valence of the subliminal prime and the behavioral information were inconsistent, strongly argues that the current findings (i.e., slow implicit attitude change) are not due to lack of sensitivity in the implicit measure.
1005
0.75 0.5 0.25 Time 1 Time 2
0 -0.25 -0.5 -0.75 -1 Positive Prime First
Negative Prime First
Figure 8. Explicit attitudes as a function of condition and time in Experiment 5.
RYDELL AND MCCONNELL
1006
Standardized Implicit Attitudes
1 0.75 0.5 0.25 Time 1 Time 2
0 -0.25 -0.5 -0.75 -1 Positive Prime First
Negative Prime First
Figure 9. Implicit attitudes as a function of condition and time in Experiment 5.
In the negative prime first condition, implicit attitudes were more negative at Time 1 than at Time 2, F(1, 48) ⫽ 6.94, p ⬍ .03 (Time 1: M ⫽ ⫺0.85, SD ⫽ 1.13; Time 2: M ⫽ 0.44, SD ⫽ 0.86). In the positive prime first condition, implicit attitudes were more positive at Time 1 than at Time 2, F(1, 48) ⫽ 7.16, p ⬍ .02 (Time 1: M ⫽ 0.85, SD ⫽ 0.67; Time 2: M ⫽ ⫺0.44, SD ⫽ 0.63). To ensure that participants did not recognize the words presented before Bob’s picture in the learning task (i.e., to establish that the primes were subliminal), we assessed the mean accuracy for their identifying which 20 of the 40 words they thought were presented. As intended, participants were no better than chance (M ⫽ 0.51, SD ⫽ 0.07) at recognizing the primes, t(49)⫽1.01, ns, even though each prime was presented 10 times, indicating that the presentation of the primes was indeed subliminal.
Discussion Experiment 5 demonstrated that implicit, but not explicit, attitudes were sensitive to the subliminal presentation of the priming words. This is consistent with a slow-learning system that is based on the slow accrual of associations encountered across time and a fast-learning system that is based on mentally manipulating conscious information. Moreover, the current study provides additional evidence, supplementing that of Experiment 4 and Rydell et al. (in press), that the results involving implicit attitudes in Experiments 1–3 were not due to an insensitive IAT, but instead, reflect associations with Bob in memory.
General Discussion This research indicates that implicit and explicit attitudes change because they are governed by different systems of reasoning. Explicit attitudes changed more quickly in response to new information and were responsive to deliberate processing goals, consistent with a quick-learning, rule-based system of reasoning. Alternatively, implicit attitudes changed much more slowly and were unaffected by processing goals, consistent with a slowlearning, associative system of reasoning. An old paradigm was modified in the current work that allows perceivers to form an initial attitude about an attitude object and that also allows for the presentation of counterattitudinal information
about that same attitude object. Consistent with distinct systems of reasoning (Sloman, 1996; Smith & DeCoster, 2000) and dual-attitude approaches (Wilson et al., 2000), these experiments show that explicit attitudes quickly changed following only a small amount of counterattitudinal information, whereas implicit attitudes about the same attitude object did not change in response to the same counterattitudinal information. Thus, counterattitudinal information, like that used in many persuasion paradigms (Petty & Wegener, 1998), may not erase the initial (implicit) attitude (see Petty et al., 2006; Wilson et al., 2000). However, consistent with a slow-learning system of reasoning, Experiments 2 and 3 show that implicit attitudes were changed when sufficient counterattitudinal information was presented. In addition, this work shows that people can hold different implicit and explicit attitudes about the same attitude object at the exact same time based on how the information they encountered impacts different systems of reasoning. The results from Experiments 2 and 3 were clear in elucidating that implicit attitudes change by a slow-learning, associative system of reasoning. Specifically, by presenting substantial amounts of counterattitudinal information, implicit attitudes changed to reflect the accrual of copious amounts of counterattitudinal information. On the other hand, explicit attitudes showed a different pattern. That is, explicit attitude change was best explained by a fast-learning, rule-based system of reasoning. Experiment 1 shows that explicit attitudes changed following a relatively small amount of counterattitudinal information. Consistent with past research, Experiments 1 and 2 also show the greatest amount of attitude change following greater consistency of initial reinforcement (e.g., Kerpelman & Himmelfarb, 1971) and when negative counterattitudinal information followed positive information (e.g., Fiske, 1980). Interestingly, implicit attitudes did not reveal these results, indicating that classic attitude asymmetry effects may be more likely for explicit than implicit attitudes. Moreover, the manipulation of reinforcement level also produced an interesting pattern with explicit attitude change. In Experiment 2, people who had less consistent initial learning showed continued attitude change as more counterattitudinal information was presented. Experiment 3 experimentally established that this pattern resulted from participants not rushing to form strong on line impressions following less consistent feedback (cf., McConnell et al., 1994b). More important, this experiment showed that manipulating impression formation goals (a deliberate process) changed explicit attitudes yet did not change implicit attitudes, further supporting a systems of reasoning explanation of implicit and explicit attitude change. Experiment 4 shows that implicit and explicit attitudes developed in the laboratory predicted different types of behaviors. Consistent with past research (e.g., McConnell & Leibold, 2001), implicit attitudes predicted subtle, spontaneous behaviors toward Bob (i.e., seating distance) but not deliberative judgments toward him (i.e., desire for social contact). Conversely, explicit attitudes predicted deliberate judgments about Bob but not more subtle, spontaneous behaviors. This experiment shows that attitudes created and changed by different systems of reasoning have important implications for when attitudes correspond to behavior. Implicit attitudes only predicted spontaneous behaviors, whereas explicit attitudes only predicted deliberate target-relevant judgments; this double dissociation further supports a systems of reasoning account.
IMPLICIT AND EXPLICIT ATTITUDE CHANGE
Finally, Experiment 5 shows that, consistent with a systems of reasoning prediction, implicit (but not explicit) attitudes were changed by counterattitudinal information that was associated with the attitude object subliminally. Explicit attitudes, on the other hand, reflected the neutral information that was consciously available about Bob. Whereas Experiments 1– 4 showed that explicit attitudes changed more quickly than implicit attitudes, Experiment 5 revealed that implicit attitudes would change even when explicit attitudes did not because of the type of information available to each system of reasoning. However, the results of Experiment 5 raise other important questions. First, why did we observe only implicit attitude change when other research (e.g., Murphy & Zajonc, 1993) has shown that explicit attitudes change in response to valenced primes? We believe that providing participants with a series of neutral behaviors occupied the fast-changing, verbal-based system with valence irrelevant information, which in turn, led to the expression of relatively neutral attitudes. Indeed, past subliminal priming research has not presented supraliminal behaviors in tandem that might engage the fast-verbal system. Another alternative is that the presentation of many neutral behaviors may have led participants to conclude that they should ignore or discount any affect generated by the subliminal primes, leading them to not use these feelings in their explicit judgments (Yzerbyt, Schandron, Leyens, & Rocher, 1994). However, it is unclear whether people would make such an attribution for explicit attitudes when the supraliminal information is neutral. This is not to argue that people do not use meta-informational cues such as “social judgability” (Yzerbyt et al., 1994), but it is not established whether people provided with neutral behaviors feel “unentitled” to render evaluations of a target. If people did feel unentitled, one would expect low-variance judgments around the midpoint. But given the relatively large variability in the current data, it seems more likely that people viewed the behaviors with some idiosyncratic degree of positivity and negativity rather than circling the midpoint because they felt they could not render a judgment. Certainly, future research should address this interesting possibility more directly. Also related to this issue are the data of Rydell et al. (in press), who used the same methods as the current Experiment 5 but presented supraliminal behaviors that were always of the opposite valence to the subliminal primes (e.g., positive subliminal primes with negative supraliminal behaviors). As predicted by a systems of reasoning perspective, explicit attitudes toward Bob mirrored the valence of the supraliminal information, and implicit attitudes toward Bob reflected the valence of the subliminal information. In this work, a social judgabililty alternative seems untenable because the verbally available information about the target should seem coherent. Yet at the same time, explicit attitudes toward the target person were radically different than the implicit attitudes toward the target person, indicating that feelings from the subliminal presentations did not “spill over” on their explicit attitudes in a substantial manner. More broadly, the current work has important implications for existing models of attitudes and persuasion. Indeed, it provides some of the clearest support for two of the dual attitudes model’s most important suppositions: People can hold different implicit attitudes and explicit attitudes about an attitude object at the same time, and implicit attitudes are not changed at the same rate as explicit attitudes (Wilson et al., 2000). This last point is extremely
1007
important for understanding how the current research relates to other research on attitude change, specifically research on the equation likelihood model (ELM) of persuasion (Petty & Wegener, 1998). The ELM predicts that once an attitude is changed, usually by the presentation of compelling arguments or by a peripheral cue to persuasion (e.g., attractiveness), the original attitude no longer exists. Although this model is extremely powerful in predicting explicit attitude change and deliberate behavior toward an attitude object, it may not account for how implicit attitudes change (e.g., Petty et al., 2006). In addition, it does not account for the possibility that implicit attitudes and explicit attitudes predict different types of behavior (e.g., Dovidio et al., 2002; Jellison et al., 2004; McConnell & Leibold, 2001). Thus, models such as the ELM should consider how implicit attitudes change and how they guide behavior in order to provide a fuller account of persuasion. Also, the motivation and opportunity to deliberate (MODE) model argues that differences between implicit and explicit attitudes are evidence that people differ in the extent to which they have the motivation and ability to modify the initial automatic activation of an attitude in memory (e.g., Fazio, 1995). Although this account is undoubtedly true in many circumstances, it may not capture the relation between implicit and explicit attitudes in all situations, especially those involving novel attitude objects and in situations in which attitude accessibility is low. In addition, this account of attitudes has difficulty explaining how implicit and explicit attitudes can differ at the exact same point in time devoid of some motivation to modify the expression of the explicit attitude (e.g., evaluating Bob carries far less social desirability concerns than expressing racial attitudes). However, a systems of reasoning approach predicts this outcome and fits the data obtained in the current work nicely. In summary, the current work shows that implicit attitudes and explicit attitudes form and change on the basis of different processes that support a systems of reasoning approach to attitude change. Understanding attitude change, and more specifically the different processes underlying implicit and explicit attitude change, is extremely important for advancing theoretical conceptualizations of attitude formation and attitude change. The interplay between implicit and explicit attitudes is extremely important for diverse areas of social psychology (e.g., attitude formation, persuasion, prejudice, attitude– behavior correspondence), and the current research begins to disentangle the differences in how implicit attitudes and explicit attitudes form and respond to social information.
References Barden, J., Maddux, W. W., Petty, R. E., & Brewer, M. B. (2004). Contextual moderation of racial bias: The impact of social roles on controlled and automatically activated attitudes. Journal of Personality and Social Psychology, 87, 5–22. Bassili, J. N., & Smith, M. C. (1986). On the spontaneity of trait attribution: Converging evidence for the role of cognitive strategy. Journal of Personality and Social Psychology, 50, 239 –245. Carlston, D. (1994). Associated systems theory: A systematic approach to cognitive representations of persons. In R. S. Wyer, Jr., (Ed.), Advances in social cognition (Vol. 7, pp. 1–78). Hillside, NJ: Erlbaum. Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic attitudes: Combating automatic prejudice with images of admired and disliked individuals. Journal of Personality and Social Psychology, 81, 800 – 814.
1008
RYDELL AND MCCONNELL
Devine, P. G. (2001). Implicit prejudice and stereotyping: How automatic are they? Journal of Personality and Social Psychology, 81, 757–759. Dovidio, J. F., Kawakami, K., & Gaernter, S. L. (2002). Implicit and explicit prejudice and interracial interaction. Journal of Personality and Social Psychology, 82, 62– 68. Dunton, B. C., & Fazio, R. H. (1997). An individual difference measure of motivation to control prejudiced reactions. Personality and Social Psychology Bulletin, 23, 316 –326. Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Orlando, FL: Harcourt, Brace & Jovanovich. Fazio, R. H. (1990). A practical guide to the use of response latency in social psychological research. In C. Hendrick & M. S. Clark (Eds.), Research methods in personality and social psychology (pp. 74 –97). Thousand Oaks, CA: Sage. Fazio, R. H. (1995). Attitudes as object-evaluation associations: Determinants, consequences, and correlates of attitude accessibility. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 247–282). Mahwah, NJ: Erlbaum. Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognition research: Their meaning and uses. Annual Review of Psychology, 54, 297–327. Fazio, R. H., Sonbanmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229 –238. Fiske, S. T. (1980). Attention and weight in person perception: The impact of negative and extreme behavior. Journal of Personality and Social Psychology, 38, 889 –906. Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692–731. Gawronski, B., & Strack, F. (2004). On the propositional nature of cognitive consistency: Dissonance changes explicit but not implicit attitudes. Journal of Experimental Social Psychology, 40, 535–542. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464 –1480. Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197–216. Gregg, A. P., Seibt, B., & Banaji, M. H. (2006). Easier done than undone: Asymmetries in the malleability of implicit preferences. Journal of Personality and Social Psychology, 90, 1–20. Jellison, W. A., McConnell, A. R., & Gabriel, S. (2004). Implicit and explicit measures of sexual orientation attitudes: In-group preferences and related behaviors and beliefs among gay and straight men. Personality and Social Psychology Bulletin, 30, 629 – 642. Kerpelman, J. A., & Himmelfarb, S. (1971). Partial reinforcement effects in attitude acquisition and counterconditioning. Journal of Personality and Social Psychology, 19, 301–305. McConnell, A. R. (2001). Implicit theories: Consequences for social judgments of individuals. Journal of Experimental Social Psychology, 37, 215–227. McConnell, A. R., & Leibold, J. M. (2001). Relations among the Implicit Association Test, discriminatory behavior, and explicit measures of racial attitudes. Journal of Experimental Social Psychology, 37, 435– 442. McConnell, A. R., Sherman, S. J., & Hamilton, D. L. (1994a). Illusory correlations in the perception of groups: An extension of the
distinctiveness-based account. Journal of Personality and Social Psychology, 67, 414 – 429. McConnell, A. R., Sherman, S. J., & Hamilton, D. L. (1994b). On-line and memory-based aspects of individual and group target judgments. Journal of Personality and Social Psychology, 67, 173–185. McConnell, A. R., Sherman, S. J., & Hamilton, D. L. (1997). Target entitativity: Implications for information processing about individual and group targets. Journal of Personality and Social Psychology, 72, 750 –762. Mitchell, J. P., Nosek, B. A., & Banaji, M. R. (2003). Contextual variations in implicit evaluations. Journal of Experimental Psychology: General, 123, 455– 469. Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness: Affective priming with optimal and suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64, 723–739. Olson, M. A., & Fazio, R. H. (2001). Implicit attitude formation through classical conditioning. Psychological Science, 12, 413– 417. Petty, R. E., Tormala, Z. L., Brin˜ol, P., & Jarvis. W. B. G. (2006). Implicit ambivalence from attitude change: An exploration of the PAST model. Journal of Personality and Social Psychology, 90, 21– 41. Petty, R. E., & Wegener, D. T. (1998). Attitude change: Multiple roles for persuasion variables. In D. T. Gilbert, S. T., Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 1, pp. 323–390). New York: McGraw-Hill. Pratto, F., & John, O. P. (1991). Automatic vigilance: The attentiongrabbing power of negative social information. Journal of Personality and Social Psychology, 61, 380 –391. Richeson, J. A., & Trawalter, S. (2005). On the categorization of admired and disliked exemplars of admired and disliked racial groups. Journal of Personality and Social Psychology, 89, 517–530. Rydell, R. J., McConnell, A. R., Mackie, D. M., & Strain, L. M. (in press). Of two minds: Forming and changing valence inconsistent implicit and explicit attitudes. Psychological Science. Skowronski, J. J., & Carlston, D. E. (1987). Social judgment and social memory: The role of cue diagnosticity in negativity, positivity, and extremity biases. Journal of Personality and Social Psychology, 52, 689 – 699. Sloman, S. A. (1996). The empirical case for two systems of reasoning. Psychological Bulletin, 119, 3–22. Smith, E. R., & DeCoster, J. (2000). Dual-process models in social and cognitive psychology: Conceptual integration and links to underlying memory systems. Personality and Social Psychology Review, 4, 108 – 131. Strack, F., & Deutsch, R. (2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review, 8, 220 –247. Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual attitudes. Psychological Review, 107, 101–126. Wittenbrink, B. Judd, C. M., & Park, B. (2001). Spontaneous prejudice in context: Variability in automatically activated attitudes. Journal of Personality and Social Psychology, 81, 815– 827. Yzerbyt, V. Y., Schandron, G., Leyens, J. P., & Rocher, S. (1994). Social judgability: The impact of meta-informational cues on the use of stereotypes. Journal of Personality and Social Psychology, 66, 48 –55.
Received June 17, 2005 Revision received February 27, 2006 Accepted March 12, 2006 䡲
SENSATION SEEKING AND RISKY BEHAVIOR Marvin Zuckerman isky behavior can be an expression of a normal, genetically influenced personality trait, sensation seeking. Its expression in risky behaviors such as extreme and risky sports, vocations, substance abuse, unsafe sex, and crime, among others, is the topic of this fascinating and accessible book. In Sensation Seeking and Risky Behavior, Marvin Zuckerman offers a comprehensive view of the role of sensation seeking in a wide range of behaviors, from risky driving and sports through substance use, sex, and crime or other antisocial behaviors. How the personality trait sensation seeking relates to these risky behaviors is described and explained in terms of genetics, biology, attitudes, and expectancies. Insights into prevention and treatment of maladaptive forms of sensation seeking, like substance abuse and unsafe sexual activity, based on the published research, are offered. The author of this engagingly written book is one of the foremost experts in this important area of behavior. 2007. 320 pages. Hardcover.
R
CONTENTS: Preface ■ Chapter 1. Sensation Seeking ■ Chapter 2. Risk ■ Chapter 3. Risky Driving, Sports, and Vocations ■ Chapter 4. Substance Use and Abuse: Smoking, Drinking, and Drugs ■ Chapter 5. Sex ■ Chapter 6. Crime, Antisocial Behavior, and Delinquency ■ Chapter 7. Prevention and Treatment of Unhealthy Risk Taking Behavior ■ References ■ Author ■ Index ■ Subject Index
List: $59.95 • APA Member/Affiliate: $49.95 • ISBN 1-59147-738-7 • Item # 4317124 • ISBN-13: 978-1-59147-738-9
A L S O
AV A I L A B L E
APA Books Ordering Information
800-374-2721 www.apa.org/books A 2005 CHOICE Outstanding Academic Title PATHOLOGICAL GAMBLING: Etiology, Comorbidity, and Treatment
Nancy M. Petry 2005 • 417 pages • Hardcover • List: $59.95 • APA Member/Affiliate: $49.95 ISBN 1-59147-173-7 • Item # 4316046 • ISBN-13: 978-1-59147-173-8
EMOTION REGULATION IN COUPLES AND FAMILIES Pathways to Dysfunction and Health
Edited by Douglas K. Snyder, Jeffry A. Simpson, and Jan N. Hughes 2006 • 336 pages • Hardcover • List: $59.95 • APA Member/Affiliate: $49.95 ISBN 1-59147-394-2 • Item # 4318032 • ISBN-13: 978-1-59147-394-7
SMOKING CESSATION with Bonnie J. Spring List: $99.95 • APA Member/Affiliate: $69.95 (VHS) 1-55798-966-4 • Item # 4310588 • ISBN-13: 978-1-55798-966-6 All videos are approximately 100 minutes, unless otherwise noted. Videos developed and hosted by Jon Carlson, PsyD, EdD
In Washington, DC, call: 202-336-5510 TDD/TTY: 202-336-6123 • Fax: 202-336-5502 In Europe, Africa, or the Middle East, call: 44-207-240-0856 AD0494
Maximize the Value of Your Membership Order Silver, Gold, or Platinum at members.apa.org /access PsycINFO® | PsycARTICLES® | PsycBOOKS® | PsycEXTRA® | PsycCRITIQUES™ Available to APA Members, Fellows, Associates, and Affiliates at deeply discounted rates, the Silver, Gold, and Platinum packages for access to APA’s renowned scientific databases are one of your most valuable membership benefits. With Platinum, you’ll get convenient access at your own desktop to all APA electronic products—including resources your institution may not offer—plus access rights for one additional person (research associate or spouse). Or choose Gold or Silver to match your specific needs and budget. Order today and start enhancing your research or practice with these invaluable collections of basic, applied, and clinical literature.
(A $999 Value)
ONLY $99
ONLY $149
ONLY $299
APA MEMBER & AFFILIATE RATES:
APA MEMBER & AFFILIATE RATES:
APA MEMBER & AFFILIATE RATES:
Members/Fellows/Associates: $99
Members/Fellows/Associates: $149
Members/Fellows/Associates: $299
Joint APA/CPA* Members: $89
Joint APA/CPA* Members: $139
Joint APA/CPA* Members: $289
Student Affiliates: $89
Student Affiliates: $139
Student Affiliates: $249
International Affiliates: $149
International Affiliates: $299
International Affiliates: $499
Teacher Affiliates: $89
Teacher Affiliates: $199
Teacher Affiliates: $399
Includes:
Includes:
Includes:
The last three years of PsycINFO journal abstracts (2004–present) and PsycARTICLES full text (2004-present)
• The complete (1800s–present) PsycINFO database • The full PsycARTICLES database (1894–present) • PsycBOOKS • PsycEXTRA
PLATINUM
PLATINUM
GOLD
GOLD (A $499 Value)
SILVER
SILVER (A $399 Value)
• Everything in Gold • Access rights for spouse or research associate • Access to all other APA electronic products, including: - PsycCRITIQUES - Graduate Study Online - APA Membership & Affiliate Directory - APA-Style Helper - And more!
See www.apa.org/databases/members/ for full package descriptions. *Canadian Psychological Association Access runs from January 1, 2007, through December 31, 2007.
BEST SELLERS from the American Psychological Association
Executive Wisdom Coaching and the Emergence of Virtuous Leaders
Richard R. Kilburg May 2006. 416 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-402-7 • Item # 4316075
Neural Networks in Organizational Research Applying Pattern Recognition to the Analysis of Organizational Behavior
David Scarborough and Mark John Somers 2006. 330 pages. Hardcover. List: $39.95 APA Member/Affiliate:$34.95 ISBN 1-59147-353-5 • Item # 4312006
The Aging Workforce Realities, Myths, and Implications for Organizations
Jerry W. Hedge, Walter C. Borman, and Steven E. Lammlein 2005. 248 pages. Hardcover. List: $79.95 APA Members/Affiliate: $49.95 ISBN 1-59147-319-5 • Item # 4316062
Creating High-Tech Teams Practical Guidance on Work Performance and Technology
Edited by Clint A. Bowers, Eduardo Salas, and Florian Jentsch 2006. 313 pages. Hardcover. List: $69.95 APA Members/Affiliate: $54.95 ISBN 1-59147-274-1 • Item # 4318025
Mental Health Care for Urban Indians Clinical Insights from Native Practitioners
Edited by Tawa M. Witko March 2006. 272 pages. Hardcover. List: $69.95 APA Member/Affiliate $54.95 ISBN 1-59147-359-4 • Item # 4317098
Psychology in the Service of National Security Edited by A. David Mangelsdorff May 2006. 320 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-355-1 • Item # 4316069
Portraits of Pioneers in Psychology, Volume VI Edited by Donald A. Dewsbury, Ludy T. Benjamin Jr., and Michael Wertheimer
Criminal Profiling Developing an Effective Science and Practice
Scotia J. Hicks and Bruce D. Sales March 2006. 288 pages. Hardcover. List: $69.95 APA Member/Affiliate $54.95 ISBN 1-59147-392-6 Item # 4316073
Emotions and Culpability
Culturally Responsive Cognitive–Behavioral Therapy Assessment, Practice, and Supervision
Edited by Pamela A. Hays and Gayle Y. Iwamasa April 2006. 336 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-360-8 • Item # 4317099
How the Law Is at Odds With Psychology, Jurors, and Itself
Interpersonal Psychotherapy for Depressed Older Adults
Norman J. Finkel and W. Gerrod Parrott June 2006. 304 pages. Hardcover. List: $69.95 APA Member/Affiliate $49.95 ISBN 1-59147-416-7 • Item # 4316078
Gregory A. Hinrichsen and Kathleen F. Clougherty
Strengthening Research Methodology Psychological Measurement and Evaluation
Edited by Richard R. Bootzin and Patrick E. McKnight 2006. 344 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-324-1 • Item # 4318026
Helping Others Help Children: Clinical Supervision of Child Psychotherapy
Edited by T. Kerby Neill May 2006. 232 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-404-3 • Item # 4317106
Fear and Learning: From Basic Processes to Clinical Implications
Edited by Michelle G. Craske, Dirk Hermans and Debora Vansteenwegen
March 2006. 248 pages. Hardcover. List: $59.95 APA Member/Affiliate $49.95 ISBN 1-59147-361-6 • Item # 4317102
Understanding Depression in Women Carolyn M. Mazure and Gwendolyn Puryear Keita April 2006. 200 pages. Hardcover. List: $49.95 APA Member/Affiliate $39.95 ISBN 1-59147-361-6 • Item # 4317102
Personality-Guided Therapy for Depression Neil R. Bockian June2006. 328 pages. Hardcover. List: $69.95 APA Member/Affiliate $49.95 ISBN 1-59147-410-8 • Item # 4317109
Chronic Health-Related Disorders in Children Collaborative Medical and Psychoeducational Interventions
Edited by LeAdelle Phelps
June 2006. 312 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-414-0 • Item # 4318034
March 2006. 312 pages. Hardcover. List: $59.95 APA Member/Affiliate $49.95 ISBN 1-59147-408-6 • Item # 4317108
Psychosocial Issues Near the End of Life
Emotion Regulation in Couples and Families
A Resource for Professional Care Providers
Pathways to Dysfunction and Health
Edited by James L. Werth, Jr. and Dean Blevins
Edited by Douglas K. Snyder, Jeffry A. Simpson, and Jan N. Hughes
2006. 264 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-236-9 • Item # 4317073
May 2006. 336 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-394-2 • Item # 4318032
Psychology Licensure and Certification
A Casebook of Psychotherapy Integration
What Students Need to Know
Edited by George Stricker and Jerry Gold
Edited by Thomas J. Vaughn
April 2006. 418 pages. Hardcover. List: $59.95 APA Member/Affiliate: $49.95 ISBN 1-59147-405-1 • Item # 4317107
List: $79.95 APA Members/Affiliate: $49.95 ISBN1-59147-417-5 Item # 4317112
2006. 264 pages. Paperback. List: $49.95 APA Member/Affiliate: $39.95 ISBN 1-59147-357-8 • Item # 4317097
The Pain Survival Guide
Cognition and Suicide
The Ethical Practice of Psychology in Organizations, Second Edition
How to Reclaim Your Life
Theory, Research, and Therapy
Edited by Rodney L. Lowman
Dennis W. Turk and Frits Winter
Edited by Thomas E. Ellis
2006. 280 pages. Paperback. List: $19.95 APA Members/Affiliate: $19.95 ISBN 1-59147-049-8 • Item # 4441007
2006. 400 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-357-8 • Item # 4317097
January 2006. 272 pages. Paperback. List: $39.95 APA Member/Affiliate: $34.95 ISBN 1-59147-353-5 • Item # 4312006
To O r d e r : 8 0 0 - 374 -2721 • w w w. a p a .or g / b o o ks
AD0458
Preventing Youth Violence in a Multicultural Society NANCY A. GUERRA AND EMILIE PHILLIPS SMITH Preventing Youth Violence in a Multicultural Society highlights the importance of creating culturally compatible interventions to stop violence among the youngest members of diverse populations. Chapters explore how ethnicity and culture can increase or decrease risk for violence among youth depending on contextual factors such as a disadvantaged upbringing, exposure to trauma, and acculturation status. Authors focus on the interaction between environmental conditions and the individual risk factors that foster youth violence. They begin by examining risk factors common to all groups of youth, such as feeling alienated from mainstream culture and searching for self-identity, and then focus on risk, resilience, and distinguishing factors among particular racial and ethnic groups, including Latino, African American, Asian American, Pacific Islander, American Indian, and White youth. The authors recommend intervenCONTENTS: Chapter tions tailored to each group as well as advice on how to 1. Ethnicity, Violence, and the Ecology of incorporate cultural competence into more general youth Development • violence prevention programs. The social-ecological Chapter 2. Ethnic approach taken in this volume emphasizes the learned nature Identity, Social Group Membership, and of aggression and violence, and many of the recommended Youth Violence • Chapter 3. Youth Violence, Immigration, and interventions involve changing the context in which violence Acculturation • Section II. Youth Violence and Prevention in Specific Ethnic Groups – Chapter 4. Youth Violence Prevention Among Latino is taught, therefore truly encouraging long-term violence Youth • Chapter 5. Youth Violence Prevention Among Asian American prevention. and Pacific Islander Youth • Chapter 6. Understanding American Indian This practical, empirically supported book serves as an Youth Violence and Prevention • Chapter 7. Preventing Youth Violence Among African American Youth: The Socio-Cultural Context of Risk important resource to all mental health practitioners working and Protective Factors • Chapter 8. Youth Violence Prevention Among in the field of youth violence. 2006. 304 pages. Hardcover. White Youth • Section III. Developing Culturally-Competent Youth
ISBN 1-59147-327-6 • Item # 4316064 List: $69.95 • APA Member/Affiliate: $49.95
Violence Prevention Programs and Strategies – Chapter 9. CulturallySensitive Interventions to Prevent Youth Violence • Chapter 10. What is Cultural Competence and How Can It Be Incorporated Into Preventive Interventions? • Chapter 11. Preventing Youth Violence in a Multicultural Society: Future Directions
ALSO AVAILABLE JUVENILE DELIQUENCY Understanding the Origins of Individual Differences VERNON L. QUINSEY, TRACEY A. SKILLING, MARTIN L. LALUMIERE AND WENDY M. CRAIG
2003. 240 pages. Hardcover. List: $49.95 APA Member/Affiliate: $39.95 ISBN 1-59147-048-X Item # 4316016 PREVENTING VIOLENCE Research and Evidence-Based Intervention Strategies EDITED BY JOHN R. LUTZKER
2005. 320 pages. Hardcover. List: $69.95 APA Member/Affiliate: $49.95 ISBN 1-59147-342-X Item # 4316067 TREATING CHRONIC JUVENILE OFFENDERS Advances Made Through the Oregon Multidimensional Treatment Foster Care Model PATRICIA CHAMBERLAIN
2003. 186 pages. Hardcover. List: $39.95 APA Member/Affiliate: $34.95 ISBN 1-55798-996-6 Item # 4317007
APA Books Ordering Information
800-374-2721
www.apa.org/books In Washington, DC, call: 202-336-5510 TDD/TTY: 202-336-6123 • Fax: 202-336-5502 In Europe, Africa, or the Middle East, call: 44-207-240-0856
AMERICAN PSYCHOLOGICAL ASSOCIATION AD0402