Attitudes, Behavior, and Social Context: The Role of Norms and Group Membership

cover

title: author: publisher: isbn10 | asin: print isbn13: ebook isbn13: language: subject publication date: lcc: ddc: subject:

next page >

Attitudes, Behavior, and Social Context : The Role of Norms and Group Membership Applied Social Research Terry, Deborah J. Lawrence Erlbaum Associates, Inc. 0805825657 9780805825657 9780585179742 English Interpersonal relations, Social groups--Psychological aspects, Attitude change, Social influence, Persuasion (Psychology) 2000 HM132.B48 1998eb 158.2 Interpersonal relations, Social groups--Psychological aspects, Attitude change, Social influence, Persuasion (Psychology)

cover

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_ii

next page > Page ii

Applied Social Research Marlene Turner and Anthony R. Pratkanis, Series Editors Larwood/Gattiker • Impact Analysis Murrell/Crosby/Ely • Mentoring Dilemmas Sedikides/Schopler/Insko • Intergroup Cognition and Intergroup Behavior Turner • Groups at Work Tyler/Kramer/John • The Psychology of the Social Self Wosinski/Cialdini/Feykowski/Barett • The Practice of Influence in Multiple Cultures

< previous page

page_ii

next page >

< previous page

page_iii

next page > Page iii

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership Edited by Deborah J. Terry Michael A. Hogg University of Queensland, Australia

< previous page

page_iii

next page >

< previous page

page_iv

next page > Page iv

Copyright © 2000 by Lawrence Erlbaum Associates, Inc. All rights reserved. No part of this book may be reproduced in any form, by photostat, microfilm, retrieval system, or any other means, without prior written permission of the publisher. Lawrence Erlbaum Associates, Inc., Publishers 10 Industrial Avenue Mahwah, NJ 07430 Cover design by Kathryn Houghtaling Lacey Library of Congress Cataloging-in-Publication Data Attitudes, behavior, and social context: the role of norms and group membership /edited by Deborah J. Terry and Michael A. Hogg. p. cm. (Applied social research) Includes bibliographical references and index. ISBN 0-8058-2565-7 (cloth: alk. paper) ISBN 0-8058-2566-5 (pbk.: alk. paper) 1. Interpersonal relations. 2. Social groupsPsychological aspects. 3. Attitude change. 4. Social influence. 5. Persuasion (Psychology). I. Terry, Deborah J., 1959- . II. Hogg, Michael A., 1957- . III. Series. HM132.B48 1998 158.2dc21 98-38294 CIP Books published by Lawrence Erlbaum Associates are printed on acid-free paper, and their bindings are chosen for strength and durability. Printed in the United States of America 10 9 8 7 6 5 4 3 2 1

< previous page

page_iv

next page >

< previous page

page_v

next page > Page v

ACKNOWLEDGMENTS As with any major task, there are numerous people to whom we are indebted for their assistance in the preparation of this edited volume. Foremost, we thank Joel Cooper for the encouragement, support, and advice that he provided as a visitor to the Centre for Research on Group Processes, School of Psychology, University of Queensland, during the time that we initially thought about putting together an edited volume on social contextual influences and attitudinal phenomena. We also thank Judi Amsel and Nadine Simms, at Lawrence Erlbaum Associates, for their advice and editorial support throughout the project, and the editors of the Applied Social Research Series, Anthony Pratkanis and Marlene Turner, for their support of the project. We thank Stuart Kinner, Jackie Wellen, and Leda Blackwood for the many hours of editorial assistance that they provided when we were preparing the manuscript for publication. We thank the faculty and graduate members of the social identity and attitudes research group at the University of QueenslandJulie Duck, Katy White, Carol Masel, Jackie Wellen, Blake McKimmieall of whom have provided an ongoing intellectual environment for this project, and the Australian Research Council, which funded our research on group membership, norms, and attitude-behavior relations. Finally, we thank our families (Martin, Nicholas, Harry, Sophie, and Bee, Jessica, and James) for their continued support and tolerance. Without their support, the project would have been much more difficult to complete in a timely manner. DEBORAH J. TERRY MICHAEL A. HOGG

< previous page

page_v

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

next page >

page_vii

Page vii CONTENTS 1 Social Contextual Influences on Attitude-Behavior Correspondence, Attitude Change, and Persuasion Michael A. Hogg And Deborah J. Terry

1

Part I. Attitude-Behavior Relations 2 The Role of Moral Norm in the Attitude-Behavior Relation Antony S. R. Manstead

11

3 Subjective Expected Utility-Based Attitude-Behavior Models: The Utility of Self-Identity Paul Sparks

31

4 A Theory of Attitudes, Subjective Norms, and Private Versus Collective Self-Concepts David Trafimow

47

5 Attitude-Behavior Relations: Social Identity and Group Membership Deborah J. Terry, Michael A. Hogg, And Katherine M. White

67

6 Pluralistic Ignorance and Inconsistency Between Private Attitudes and Public Behaviors Dale T Miller, Benoît Monin, Deborah A. Prentice

95

7 Where Does the Behavior Come From in Attitude-Behavior Relations? Toward a Connectionist Model of Behavior Generation 115 Yoshihisa Kashima And Virginia Lewis Part II. Attitude Change and Persuasion 8 The Impact of Group Membership on Persuasion: Revisiting "Who Says What to Whom with What Effect?" Diane M. Mackie And Sarah Queller

< previous page

page_vii

135

next page >

< previous page

next page >

page_viii

Page viii 9 Group Norms, Prototypicality, and Persuasion Daan Van Knippenberg

157

10 Identity and Persuasion: An Elaboration Likelihood Approach Monique A. Fleming And Richard E. Petty

171

11 Altercasting as an Influence Tactic Anthony Pratkanis

201

12 Cognitive Dissonance and the Social Group Joel Cooper And Jeff Stone

227

13 Social Influence and Identity Conflict Juan Manuel Falomir, Gabriel Mugny, And Juan Antonio Pérez

245

14 The Perceived Impact of Persuasive Messages on "Us" and "Them" Julie M. Duck, Michael A. Hogg, And Deborah J. Terry

265

References

293

Author Index

331

Subject Index

341

< previous page

page_viii

next page >

< previous page

page_1

next page > Page 1

1 Social Contextual Influences on Attitude-Behavior Correspondence, Attitude Change, and Persuasion Michael A. Hogg Deborah J. Terry University of Queensland The study of attitudes lies at the core of social psychology. Thomas and Znaniecki (1918) and Watson (1930) went so far as to define the whole of social psychology as the scientific study of attitudes, and, with only slightly less imperialism, Allport (1935) called attitudes social psychology's most indispensable concept. Attitudes are important in social psychology for at least three reasons. First, if social psychology is essentially cognitive (Fiske & Taylor, 1991; Markus & Zajonc, 1985), then attitudes are the apotheosis of social cognition, because they are unobservable cognitive constructs that are socially learned, socially changed, and socially expressed. Second, attitudes are practically important to social psychologists, because social psychologists use expressed attitudes, in the form of questionnaire responses, as the data base for most theories of social behavior. Finally, attitudes are politically important for social psychology, as they provide a potent entrée to the real-world application of social psychology. In his discussion of the attitude concept in 20th-century social psychology, McGuire (1986) identified three distinct phases of attitude research: a focus in the 1920s and 1930s on static issues of measurement and relation to behavior, a focus in the 1950s and 1960s on dynamics of individual attitude change, and a move in the 1980s and 1990s toward understanding the structure and function of attitude systems. Attitudes have weathered the storm of

< previous page

page_1

next page >

< previous page

page_2

next page > Page 2

behaviorism and undoubtedly will weather the contemporary storm of poststructuralism and discourse analysis. There has certainly been a recent strengthening of interest among mainstream social psychologists in attitude phenomena, as can be seen from recent books such as Eagly and Chaiken (1993); Perloff (1993); Krosnick and Petty (1995); and Pratkanis, Breckler, and Greenwald (1989). This renewed interest has many vehicles, but foremost, we feel, are issues relating to cognitive processing of attitudinal information and group membership and norms. People's attitudes are developed and expressed as behaviors in a context that is social; it contains other people who are actually present or who are invisibly present in the social norms that define social groups to which we do or do not belong. Cognitive processing, group membership and norms may have begun to make an impact in the study of attitudinal phenomena due to advances made in the study of social cognition (e.g., Devine, Hamilton, & Ostrom, 1994; Fiske & Taylor, 1991), social identity and group membership (e.g., Hogg & Abrams, 1988; Robinson, 1996), and group norms (e.g., Turner, 1991), as well as the interrelations of social cognition and group processes (e.g., Abrams & Hogg, 1999; Nye & Bower, 1996). In many ways, this reinvigoration of attitudes mirrors a similar revitalization of social psychological research on groups (see Moreland, Hogg, & Hains, 1994). This book brings together leading attitude researchers from around the world to present their ideas about how the social context in the form of group norms and group membership may influence attitudesspecifically attitude-behavior relations, attitude change, and persuasion. The first half of the book focuses on attitude-behavior relations, and the second half on attitude change and persuasion. The Chapters The book opens with chapter 2, by Anthony S. R. Manstead on moral norms, and directly addresses the commonsense notion that if something is right, people ought to do it. Moral norms seem to be particularly influential, which explains relatively more variance in attitude-behavior relations than do other norms that do not have a component of moral imperative or moral obligation. Manstead reviews research from a wide range of attitudinal and behavioral contexts, including sexual behavior, business ethics, and alcohol consumption. When people really do subscribe to moral norms (as opposed to merely paying public lip service to such norms), behavior seems to be strongly determined by such norms. The question of how norms acquire moral force becomes very interesting both theoretically and practically. Presumably, if a norm can be infused with moral force then it should more strongly determine behavior. That is, if someone can be persuaded that

< previous page

page_2

next page >

< previous page

page_3

next page > Page 3

something is morally right, rather than merely normative, for example, to purchase Product A rather than Product B, then sales of Product A will skyrocket. Manstead discusses developmental theories, in particular the work of Hoffman (1983), Kohlberg (1984), and Vygotsky (1978), to show how the child develops a sense of right and wrong. This fundamental sense instills moral force in certain norms and thus produces moral norms. Another possibility is that norms acquire the equivalent of moral force if they define the selfin other words, if they define a social group that contributes significantly to the self-concept (cf. Terry, Hogg, & White, chapter 5, this volume). Paul Sparks refocuses our attention from the lofty heights of morality to the more basic issue of food (chap. 3). Although food is of course a necessity of life, it also defines who we are. One of the most obvious differences among cultures and economic groups in cultures is what is eaten and how it is eaten. In multicultural or multiethnic societies, cuisine is often the most salient symbol of ethnic or cultural diversity. The study, then, of food attitudes and behavior provides a powerful empirical focus for understanding the relation among attitudes, norms, behavior, and identity. Sparks' chapter adopts this focus. He argues that people whose self-identity or self-concept defines them as the sort of people who perform a particular behavior are more likely to perform that behavior. If someone believes that he or she is the sort of person who eats noodles with chopsticks, then he or she is very likely to eat noodles with chopsticks. In keeping with self-identity perspectives, Sparks does not distinguish between self-identities that may be more or less grounded in group membership. Sparks suggests that one reason that self-identity improves prediction is that specific attitudes may be overly context-bound and often be in conflict with other attitudes or with various values, whereas attitudes and values are more likely to be consonant and enduring within a self-identity. As in Manstead's chapter, attitudes are considered better predictors of behavior if they are ''energized'' by evaluative factors (morality in Manstead's treatment, social identity in Terry et al.'s chapter, and values in Sparks' analysis). David Trafimow organizes his wide-ranging chapter (chap. 4) around five major propositions relating behavior to attitudes. One strand of his argument is that research suggesting that behavior is more under attitudinal than normative control may be attributable to the fact that some people are under attitudinal control across a wide range of behaviors, whereas others are under normative control across a wide range of behaviors. Furthermore, research has tended to use "attitudinal control" people because the research has been conducted in individualistic Western cultures. Trafimow documents that the two parts of the self-conceptthe private and the collective selfare distinct and cognitively largely independent. Attitudinal control is

< previous page

page_3

next page >

< previous page

page_4

next page > Page 4

more closely related to private self, normative control is more closely related to the collective self, and culture can influence whether the private or collective self is more accessible. This analysis has the wider implication that norms will have a greater effect on behavior if the norms are tied into conditions under which collective self is salient. Deborah J. Terry, Michael A. Hogg and Katherine M. White's chapter (chap. 5) follows directly from Trafimow by exploring the role of collective self, but here, collective self is conceptualized as group membership and strictly in terms of social identity and self-categorization theories. When people categorize themselves in terms of a contextually salient ingroup, there is a process of depersonalization that cognitively, behaviorally, and affectively assimilates self to the ingroup prototype. That is, it produces ingroup normative behavior regarding attitudes, feelings, and behaviors. The clear prediction is that attitude-behavior consistency should increase under conditions in which people identify strongly with a group for which the attitudes and the behaviors are prototypical or normative. Terry and her associates describe their program of research, which provides evidence in support of this analysis. Some of these studies also investigate the role of attitude accessibility and mode of behavioral decision making in a social identity analysis of attitudebehavior consistency. Dale T Miller, Benoît Monin, and Deborah A. Prentice turn the traditional issue of attitude-behavior relationship on its head (chap. 6). Instead of asking how behavior reflects underlying attitudes, they ask how people infer underlying attitudes from overt behavior, given that behavior often merely reflects compliance with assumed norms and is therefore uninformative about attitudes. One reason that there is so often a mismatch between attitudes and behavior is, paradoxically, that people are actually conforming to a normative misperception based on pluralistic ignorance. For example, people who oppose excessive alcohol consumption may nevertheless engage in heavy drinking because they falsely believe that most other relevant people are in favor of heavy drinking. If this is true, then dispelling pluralistic ignorance may change behavior. Miller and his associates report data to support this analysis. Pluralistic ignorance may develop over time, because even though private attitudes may shift, people often base their normative beliefs on other people's public behavior. Gradually, attitudes shift out of synchrony with normative beliefs and with behavior. The notion of pluralistic ignorance suggests that behavior is more closely tied to perceived group norms, even if they are completely inaccurate, than to individuals' private attitudes, an idea that seems consistent with the social identity analysis of attitude-behavior relations provided by Terry and her associates. The final chapter on attitude-behavior relations (chap. 7) is by Yoshihisa Kashima and Virginia Lewis. Kashima and Lewis strike right at the heart of

< previous page

page_4

next page >

< previous page

page_5

next page > Page 5

the question of what energizes specific behavioral responses. They describe a novel application of connectionist modeling to help explain the generation of behavioral responses. The tensor product model outlines the mechanism that translates cognitive representations of the self, the attitude object, and the context into representations of behavior. In accord with the central theme of this book, Kashima and Lewis also discuss the role that the social context may play in activating social norms that, in turn, generate behavioral responses. In particular, they propose that self-categorization as an ingroup member generates behavior that is congruent with ingroup norms because the categorization serves as a cue to cognitive representations of the prototypical behavior for group members. Although the tensor product model has not been tested empirically, Kashima and Lewis, outline some of the model's interesting implications for the interpretation of existing findings in the attitude-behavior field. The second half of the book focuses on attitude change and persuasion. Chapter 8, by Diane M. Mackie and Sarah Queller, carefully documents a loss of interest among social psychologists in the impact of group membership on attitude change and persuasion. This is attributed to the separation of normative influence from informational influence and the marginalization of the former as dealing with behavioral compliance and the latter as dealing with true internalized cognitive change. Social cognition's emphasis on asocial cognitive processes rather than more social processes also has contributed to this shift in emphasis. Recent research, however, on systematic versus heuristic and on central versus peripheral route processing of persuasive information, in conjunction with social categorization perspectives on group membership, may endow group influence with a new respectability. Mackie and Queller describe how persuasive messages from an ingroup may engage effortful, systematic, central route processing that results in true internalized attitude change. This is particularly likely to occur when the persuasive message deals with matters that play key roles in defining ingroup membership. Like Mackie, Daan van Knippenberg (chap. 9) discusses social psychology's long tradition of distinguishing between two distinct forms of social influence that produce attitudinal or behavioral change. On the one hand, people may change their behavior or expressed attitudes superficially as a consequence of relatively heuristic or peripheral information processing and to comply with others who have some form of social or material power over them. On the other hand, they may substantively change their behavior or attitudes as a consequence of relatively systematic or central information processing and to gain a more valid and accurate understanding of the stimulus domain. Building on social identity and self-categorization theory, van Knippenberg suggests that messages originating in a salient ingroup and

< previous page

page_5

next page >

< previous page

page_6

next page > Page 6

that are highly prototypical of that group are particularly persuasive in an enduring and substantive sense. People process such information very carefully and deliberately because it is highly self-relevant. In chapter 10, Monique A. Fleming and Richard E. Petty describe how the elaboration likelihood model (ELM) of persuasion can explain the different routes by which common or disjunctive identity can change attitudes. Identity is defined very broadly to include all attributes that constitute the self-conceptindividual and group attributes that are possessed as well as those that are aspired to. When people are motivated and have the opportunity to elaborate arguments (i.e., central route processingelaboration likelihood is high), identity may be persuasive because it encourages one to discredit outgroup and accept ingroup arguments; it biases processing. In a group context, the process is akin to informational influence. When motivation or opportunity to elaborate arguments is low (i.e., peripheral route processingelaboration likelihood is low), identity is persuasive largely because it is easy to simply agree with sources with whom one shares identity. In a group context the process is akin to normative influence. Fleming and Petty review research and describe a series of studies that generally support this analysis of the way that identity may influence attitudes under low and high elaboration likelihood. However, there seem to be many paths to persuasion via identity, because the relationship is complex. In chapter 11, Anthony Pratkanis and Melissa Gliner take a different approach to the impact of the social context on persuasion by focusing on the use of altercasting, in which the influence attempt is successful because the persuader places the recipient into a role that make the incumbent more likely to accept the persuader's position. Pratkanis and Gliner draw on research from a range of different areas of social psychological research to discuss the effectiveness of altercasting. For example, they review research showing that heightening the salience of a role or identity makes people more responsive to role- or identity-related persuasion, as does either attributing a new role to message recipients or asking them to engage in a role-playing exercise. They go on to discuss an altercasting theory of source credibilityone of the aspects of a persuasive attempt most commonly examinedand discuss the type of predictions that this theory makes in comparison to theories of source credibility that focus solely on the source of the message rather than the nature of the role relationship between the message source and the target audience. Pratkanis and Gliner conclude their chapter with a consideration of the type of role sets commonly used in influence attempt, and a discussion of the basis for the reasons that altercasting is such an effective influence attempt. Mass persuasion in the context of antismoking campaigns is the vehicle for Juan Manuel Falomir, Gabriel Mugny, and Juan Pérez's analysis of the

< previous page

page_6

next page >

< previous page

page_7

next page > Page 7

ways beliefs translate into behavior (chap. 12). Many people believe smoking is bad for them, but only a subset of these people stop smoking or do not take up smoking. Falomir and his associates argue that smoking is a behavior that relates strongly to self-concept and social identity, so attempts to change that behavior can represent a threat to identity. Because smokers are generally a stigmatized social group and antismoking messages generally come from the higher status, dominant, majority group, smokers can feel their identity is being threatened. A common reaction to such threat is defensive reinforcement of their "smoker" identity and a concomitant resistance to persuasion and behavioral change. By viewing persuasion, attitude change, and attitude-behavior correspondence in an intergroup context, Falomir and associates introduce a novel and promising new approach. Cognitive dissonance theory demonstrates that people change their attitudes to reduce cognitive inconsistency, largely between their attitudes and their freely chosen, publicly verifiable, overt behavior. Joel Cooper and Jeff Stone (chap. 13) review cognitive dissonance theory and its subsequent refinements and elaborations, with a particular emphasis on the role of responsibility for aversive consequences of behavior. Dissonance is most likely to arise when people freely and knowingly engage in attitude-inconsistent behavior that has aversive consequences. Cooper and Stone's chapter mainly is concerned with the role of groups and group membership in dissonance processes and attitude change. Social support can be viewed as a consonant cognition that reduces the magnitude of dissonance and thus maintains attitudes in the light of behavior that has unwanted consequences. Being in a group also may allow diffusion of responsibility for unwanted consequences and thus reduce dissonance and guard against attitude change. A third way that group membership may protect attitudes is when attitudes define membership in a salient and important self-referent ingroup (i.e., social identity). Under these circumstances, the negative feelings associated with dissonance can be attributed to and vented on a salient outgroup in the form of accentuated outgroup affect and hostility. Thus dissonance is reduced while attitudes remain intact. Alternatively, people may engage in motivated forgetting (i.e., repression) of attitude discrepant behavior, if the attitude is importantly self-definitional (i.e., a part of social identity and thus the selfconcept). Cooper and Stone describe research that supports these analyses. Most of the attitude change and persuasion chapters in this book focus on the process of persuasionhow individuals are persuaded to change their attitudes or behaviors. Julie M. Duck, Michael A. Hogg and Deborah J. Terry (chap. 14) introduce a change in perspective by focusing on people's perceptions of being influenced, specifically the third-person effect, in which people generally believe that others (third parties"them") are more

< previous page

page_7

next page >

< previous page

page_8

next page > Page 8

persuaded than self ("me") and similar others ("you"). One important influence on perceptions of influence is whether the other person is an ingroup or an outgroup member. When being influenced is undesirable (e.g., commercial advertising) outgroup members are thought to be more influenced than self and ingroup members, but when being influenced is desirable (e.g., public health campaigns) the opposite perceptions prevail (cf., Pettigrew's, 1979, ultimate attribution error). Duck and her associates take their intergroup or social identity analysis farther by also examining how perceptions of being influenced are affected by the perceived nature of intergroup relations regarding power, status, legitimacy, stability, permeability, and so forth. Themes and Prospects The chapters in this book are certainly diverse and wide-ranging in their approach to attitudes, behavior, and social context. They deal with many topics, including connectionism, morality, the self-concept, identity, roles, cognitive dissonance, mass communication, information processing, social influence, intergroup relations, and, of course, norms and groups. There are, however, some recurring themes that identify shared concerns. In the attitude-behavior context, a number of authors believe that attitudes need to be energized for them to translate into attitude-consistent behavior. Attitudes can be energized most readily, perhaps, if the attitudes are attached to social values, moral imperatives, or identity and the self-concept. The focus on social values and moral imperatives has clear practical relevance. Much attitude-behavior research is concerned with getting people to do the right thing in the areas of business ethics, sexual behavior, and public and private health-related behavior. The focus on self-concept repeatedly distinguishes between the more personal and idiosyncratic self and the more widely shared collective or group membership related self, the latter being more closely linked to norms. Another important theme is that in social psychology there has been a historic separation between, on the one hand, relatively deliberate and systematic information processing and, on the other, relatively superficial and heuristic information processing. Because the former has been associated with attitudes as internal cognitive or individual constructs and the latter with norms as external social or group constructs, it has been difficult to see how attitudes are affected by norms or groups. Many of the chapters in this book show that this separation is breaking down. Groups and norms can and do affect people via deliberate and systematic information processing. A final theme is that attitudinal phenomenaattitude acquisition, change, and enactmentoccur in intergroup contexts. The social information that influences our attitudes and our decisions to behave in accordance

< previous page

page_8

next page >

< previous page

page_9

next page > Page 9

with our attitudes comes from other people, either directly or indirectly, and these people are readily sorted into those who are like us (ingroupers) or not like us (outgroupers). Sociocognitively generic and sociohistorically specific processes associated with group membership and intergroup relations then influence attitudinal phenomena. From these general themes, sketching out a general perspective on the relation among attitudes, behavior, norms, and group membership may be possible. Attitudes are preeminently social. We acquire them through, and they are held in place or modified by, direct or indirect social interaction. Through behavior we learn about other people's attitudes, others learn about our attitudes, and we can make public or conceal our own attitudes. Social context is critical for attitudinal phenomena. To varying degrees, attitudes tell us what sort of people we are and what sort of people others are. The process of finding out about people often involves trying to discover what they really thinkwhat their attitudes really are. Attitudes can be important markers ofeven the defining attributes ofidentity. In many, perhaps most, cases, attitudes are shared and attitudinal discontinuities among people provide the contours of social groups. In this way, attitudes can be the content of social normsthe stereotypical attributes, even the criterial attributes, of social groups. People in different groups may hold different attitudes, and indeed this may come about as the result of a process of accentuation of intergroup differences. From this analysis, it is not unreasonable to suppose that some attitudes are critical to who we are, in the sense of where we are located in the complex structure of social groups and categories that comprises the wider social context of human existence. These sorts of attitudes are ones that we think about carefully (we prefer to process information relating to them in a systematic and careful manner); they are ones that we are generally more likely to enact to socially affirm who we are; they are ones that are more highly responsive to social norms that validate or disconfirm the identity-relevance of the attitude (we may consolidate these attitudes in the face of outgroup norms or persuasion attempts, or change them if they are disconfirmed by ingroup norms). In diverse ways, the following chapters touch on and elaborate some of these themes and ideas about how attitudes, behavior, norms, and group membership are dynamically interrelated.

< previous page

page_9

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_11

next page > Page 11

2 The Role of Moral Norm in the Attitude-Behavior Relation Antony S. R. Manstead University of Amsterdam In this chapter, I focus on the construct of moral norm in the context of the theories of reasoned action and planned behavior. Moral norm is argued to be distinct from the standard constructs included in the theories of reasoned action and planned behavior, and its inclusion among predictors of behavioral intentions can lead to significant and, in certain cases, substantial increments in the amount of explained variance in intentions. I begin with an analysis of the concept of moral norm, distinguishing it from constructs already included in the theory of reasoned action and the theory of planned behavior, and relating it to anticipated affective consequences of engaging in a behavior. The chapter proceeds with a review of the empirical evidence relevant to the role of moral norm in attitude-behavior relations. In the penultimate section of the chapter, I consider theoretical models of the development of moral norms, and the chapter closes with a summary of what is known about this construct and what still needs to be accomplished by way of theory and research. The Concept of Moral Norm As most readers of this book are well aware, standard models of the attitude-behavior relation, such as the theory of reasoned action (TRA; Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975) regard intention as the most proximal psychological determinant of behavior (assuming that the behavior is largely volitional in nature). Intention, in turn, is jointly determined by atti-

< previous page

page_11

next page >

< previous page

page_12

next page > Page 12

tude toward behavior and subjective norm. The precursors of these two constructs are also specified, in the form of behavioral beliefs and outcome evaluations in the case of attitudes to behavior, and normative beliefs and motivations to comply in the case of subjective norms. The global picture created by this influential model is of a rational human being who makes decisions about how to act that are based on cost-benefit analyses of what would ensue from enacting the behavior in question. Other factors that influence behavior are thought to exercise that influence through the constructs that are included in the model. One such construct is the individual's conviction that acting in a certain way is right or wrong. This conviction that some forms of behavior are inherently right or wrong, regardless of their personal or social consequences, is what is meant by the term moral norm. In fact, Fishbein's (1967b) original statement of what later evolved into the TRA did include the construct of personal normative beliefs, in addition to attitude toward behavior and social normative beliefs. This extra component was later dropped on the grounds that "it seemed to be little more than an alternative way of measuring behavioral intentions" (Ajzen & Fishbein, 1980, p. 257). The reasons Ajzen and Fishbein were led to believe that these two variables measure the same underlying construct are clear: If one believes strongly that using corporal punishment is "wrong", one is very likely to state that one does not intend to use this form of punishment. At the same time, it is not difficult to imagine situations in which an individual has a clear sense of what is morally "right" or ''wrong" but nevertheless forms an intention that conflicts with this moral norm, because the personal or social consequences of behaving consistently with that norm are felt to outweigh the disadvantages of acting inconsistently with the norm. Indeed, this is what is referred to colloquially as a moral dilemma, or a situation in which moral convictions are placed in competition with personal or social gains and losses. So even though personal convictions about the moral rightness or wrongness of a course of action might coincide completely with the person's intention to engage (or not to engage) in that course of action, this need not be the case. To the extent that moral norm and intention do not always tap the same construct, it is worth considering further what might be gained from including such a construct in the TRA. A further issue of potential redundancy arises in connection with subjective norm. This construct is supposed to capture the individual's perception that important others in his or her social environment wish or expect him or her to behave in a certain way. Perceived social pressure to perform (or not to perform) the target behavior would be another way of describing this construct. Moral norm also refers to an expectation that one will behave in a certain way, and this expectation is of course social in origin, insofar as it is presumably acquired in the course of socialization. How, then, do subjective

< previous page

page_12

next page >

< previous page

page_13

next page > Page 13

norm and moral norm differ from one another? Although the two constructs may be closely related under certain circumstances (e.g., where an individual has been raised in a family that adheres strictly to some religious code, and this code becomes internalized), it is also conceivable that individuals entertain moral convictions that do not tally with the perceived expectations of important others in their social milieu. Thus, a teenager may become a vegetarian having developed a conviction that killing animals for consumption is morally unacceptable, despite the carnivorous practices of the rest of his or her family. The moral conviction underlying the decision to become a vegetarian is presumably initially acquired in the course of interaction with the social environment (e.g., friends and peers) but then becomes independent of the immediate expectations and influence of others, virtually regardless of whom those others might be. Thus, although moral norms may have their origins in social or group norms, such norms become internalized and autonomous, exercising influence over the individual's thoughts, feelings, and behaviors independently of the immediate social context. It is clear, then, that a personally held moral norm concerning eating meat results in a rather clear evaluation of that behavior: It is wrong. This raises a third issue of potential overlap with other constructs in the TRA: To what extent is moral norm distinct from attitude toward behavior? After all, if attitude toward behavior reflects the individual's general feeling of favorableness or unfavorableness regarding the performance of the behavior in question, should this not also include the moral conviction that performing the behavior is inherently right or wrong? In this case, much depends, I would argue, on the way in which attitude toward behavior is operationalized. It is standard practice in research using the TRA to distinguish between direct and indirect measures of the constructs in the model. Direct measures are ones that invite the respondent to rate the target behavior on semantic differential scales. Ajzen and Fishbein (1980) provided examples of such scales, including rewarding-punishing, unpleasant-pleasant, harmful-beneficial, foolish-wise, and good-bad. Although all of these are clearly evaluative in nature, only good-bad, I suggest, clearly has the potential to capture the moral concerns of, say, a committed vegetarian. This person might enjoy the taste of meat and find it punishing and unpleasant to avoid eating meat; he or she might even suspect that not eating meat is harmful to her health, and that it is foolish, but may nevertheless be a vegetarian on the basis of purely moral considerations. This distinction between moral norm and attitude toward behavior becomes even clearer if we consider how the latter construct is assessed using indirect measures. These are based on the perceived likelihood that engaging in the target behavior will bring about certain outcomes (i.e., behavioral

< previous page

page_13

next page >

< previous page

page_14

next page > Page 14

beliefs), weighted by the desirability or undesirability of these outcomes in the eyes of the individual (i.e., outcome evaluations). Clearly, beliefs concerning the outcomes of a given behavior can vary from individual to individual; Ajzen and Fishbein overcame this problem by recommending that one should first identify the so-called modally salient beliefs in the research population. This is achieved by conducting pilot research using a sample drawn from the same population, and asking them to state the advantages and the disadvantages of performing the target behavior. This form of question almost inevitably leads respondents to focus on the instrumental consequences of their actions rather than their moral implications (see Manstead & Parker, 1995)although it should be added that instrumental consequences do not necessarily preclude moral implications. In summary, attitudes toward behavior as directly measured may partly capture morally based evaluations of the behavior, to the extent that participants are invited to rate the behavior as good or bad. However, many of the other evaluative scales that are typically used for such direct assessment of attitude toward behavior tend to have a more instrumental or appetitive character, making it possible or even likely that the overall index of attitude reflects the desirability of performing the behavior because doing so will be pleasant or profitable as opposed to morally right. Attitude toward behavior as indirectly measured is even more likely to fail to capture this morally based evaluation of a behavior. One can, in terms of these measures, have a positive attitude to performing the behavior, although simultaneously believe it to be morally wrong. A final point in connection with the concept of moral norm is its relation to the anticipated affective outcomes of a behavior. If moral norm can be said to ''reflect internalized moral rules" (Eagly & Chaiken, 1993, p. 177), it seems reasonable to assume that individuals are unlikely to contemplate the infringement of these rules in a dispassionate way. Rather, they may anticipate experiencing negative feelings such as regret, shame, or guilt after having broken these rules; it is also possible that they anticipate experiencing positive feelings such as contentment and pride after behaving consistently with these rules. The role played by anticipated postbehavioral feelings in decision making was first discussed by Janis and Mann (1977). Subsequent theory (Bell, 1982; Loomes & Sugden, 1982) and research (e.g., Baron, 1992; Beattie, Baron, Hershey, & Spranca, 1994; Simonson, 1992; Zeelenberg, Beattie, van der Pligt, & de Vries, 1996) has confirmed the importance of the role played by anticipated feelings in decision making. Recent research using the theory of planned behavior (TPB) as a theoretical framework has begun to include measures of anticipated affect in general, and anticipated regret in particular, to establish whether they help to account for additional variance in the explanation of intentions. Because the

< previous page

page_14

next page >

< previous page

page_15

next page > Page 15

prospect of infringing moral norms is likely to give rise to anticipated affect, research including measures of anticipated affect is also reviewed. This is not to argue that it is not worth distinguishing between moral norms and anticipated affect: It is clearly possible for affective reactions to be anticipated in the absence of moral considerations. Moral Norms and the Attitude-Behavior Relationship: Research Evidence Altruistic Behaviors Research on moral norms in the context of the TRA or TPB can be traced back to Schwartz and Tessler (1972). These investigators measured personal normative beliefs, attitudes, and subjective norms in the context of a study of intentions with respect to organ donation. Personal normative beliefs were operationalized as sense of moral obligation, and measured by asking respondents "If a [close relative] of yours . . . needed a [kidney] transplant and you were a suitable donor, would you feel a moral obligation to donate a [kidney] ?" (p. 230). Other behaviors assessed were donating a kidney to a stranger; arranging to donate one's heart to (a) a close relative or (b) a stranger after one dies; and donating bone marrow to (a) a close relative or (b) a stranger. For all six behaviors, it was found that the inclusion of the moral obligation measure in the regression equation significantly improved the prediction of behavioral intentions beyond what was explained by attitudes and subjective norms. Indeed, moral obligation accounted for considerably more variance in intentions than did the other two constructs: When entered alone, moral obligation explained between 39% and 53% of variance in intentions; when attitude toward behavior and subjective norm were added, the increment in explained variance ranged from 6% to 10%. Pomazal and Jaccard (1976) conducted a study of blood donation, assessing behavior as well as intentions. Drawing on Schwartz and Tessler's findings, they measured moral norm as well as attitude to behavior and subjective norm. Moral norm was assessed by one item: "I personally feel I have a moral obligation to donate blood at the upcoming drive" (p. 320). They found that donation behavior was substantially related to intention (r = .59) and that adding the moral norm measure to an equation in which intention was regressed on attitude toward behavior and subjective norm improved prediction by 7% (from 41% to 48%). Similar findings were reported by Zuckerman and Reis (1978): Blood donating behavior was again quite substantially related to intentions (r = .45), and whereas intentions were reasonably well predicted by attitudes and subjective norms (r = .500), adding moral norms to the equation resulted in a significant increase in predictive utility (r = .545).

< previous page

page_15

next page >

< previous page

page_16

next page > Page 16

Harrison (1995) used the TRA and TPB to study volunteer motivation; as manifested by willingness to volunteer for a nonprofit organization for the benefit of others. The specific context of Harrison's research was a winter shelter for homeless men, and the behavior involved was overnight work in the shelter, cooking, washing, cleaning, counseling, and so on. In addition to standard measures of TRA and TPB constructs, moral obligation was assessed by means of two items, one of which was "I feel amoral obligation to work at the shelter the next time I'm scheduled," the response scale ranging from extremely weak to extremely strong. Three different samples of volunteers were studied, one of them at two different timepoints. Correlations between attendance at the shelter and intention ranged from .57 to .73, depending on the sample. When regressing intention on the predictors, adding the moral obligation measure after measures of attitude to behavior, subjective norm, and perceived behavioral control (which together explained between 25% and 46% of the variance) consistently resulted in significant increments in explained variance (ranging from 6% to 26%, depending on sample). Indeed, of the various predictors of volunteer intention, moral obligation consistently had the highest regression weight. Illegal, Antisocial, or Dishonest Behaviors Gorsuch and Ortberg's (1983) study included a measure of what they called perceived moral obligation. This construct was assessed, along with measures of the standard TRA constructs, in the context of two hypothetical situations which, in the view of the authors, posed moral conflicts. One situation involved being sent a tax refund for $500 in error, and having to decide whether or not to return it; the other involved being offered a job promotion that would entail working on Sundays and therefore being unable to attend church (the participants were persons attending adult Sunday school classes at a Baptist church). Moral obligation was assessed using the item "I have a moral obligation to [return the $500]." Here, too, it was found that the moral norm measure made an independent contribution to the explanation of behavioral intentions. In the tax situation, moral obligation added 20% to the explained variance in intentions, compared to the 9% already accounted for by attitude toward behavior and subjective norm; in the job promotion situation, the comparable figures were 7% and 29%, respectively. Beck and Ajzen's (1991) study used the theory of planned behavior to predict the commission of dishonest actions: cheating on a test, shoplifting, and lying in order to get out of an assignment. The authors reasoned that moral concerns are likely to be salient with respect to these behaviors, and that a measure of moral obligation may therefore add to the predictive utility of the TPB model. Moral obligation was measured by three items: "I

< previous page

page_16

next page >

< previous page

page_17

next page > Page 17

would not feel guilty if I did x"; "Doing x goes against my moral principles"; and "It would be morally wrong for me to do x"; where "x'' was one of the three behaviors under study. It was found that a scale consisting of these three items added significantly to the prediction of behavioral intentions for all three behaviors, above what could be accounted for by the three TPB constructs, although the increments in explained variance were not large in absolute terms (ranging from 3% to 7%). Beck and Ajzen also found that their moral obligation measure added significantly (above what could be explained by intentions and perceived behavioral control) to the prediction of (self-reported) behavior, measured later, although this was only true of lying and (to a lesser extent) cheating. In short, these findings go beyond what previous researchers had reported, in that a moral norm measure was found (a) to predict intentions independently of TPB (rather than TRA) constructs, and (b) to predict later behavior independently of intentions and perceived behavioral control. Parker, Manstead, and Stradling (1995) examined intentions to commit driving violations, using an extended version of the TPB. All three behaviors included in this study concerned lane violations: cutting sharply across a lane of traffic to leave a motorway (i.e., freeway); weaving from lane to lane in slow-moving traffic to make faster progress; and overtaking a car on the inside lane (an offense in the United Kingdom and many other countries). Respondents read scenarios depicting the commission of each violation, and were asked to respond to questions designed to tap each of the TPB constructs, plus two additional variables: moral norm, and anticipated regret. Moral norm was assessed by means of a single item "It would be quite wrong for me to . . . ," followed by the violation in question. Anticipated regret was measured using two items: "Having [committed the violation] would make me feel sorry," and "Having [committed the violation] would make me feel good" (the latter being reverse scored). The standard TPB predictors accounted for between 34% and 37% of the variance in intentions; adding moral norm and anticipated regret increased this by 10% and 15%. Moral norm had a higher beta weight than the other predictors in all three final regression equations, followed in each case by anticipated regret. Consistent with the argument that moral norm and anticipated regret are related constructs, these two variables were significantly correlated with one another, but the fact that they made independent contributions to the explanation of intentions shows that it is nevertheless worth assessing both constructs. In a follow-up study, Manstead, Parker, and Stradling (1997) used an extended version of the TPB to study two other types of driving violation: flashing headlights at the vehicle ahead to get it to move faster or get out of the way; and driving through traffic lights that are in the process of turning

< previous page

page_17

next page >

< previous page

page_18

next page > Page 18

red, although one has sufficient time to stop safely. In addition to measures of the standard TPB constructs, there were two moral norm items (e.g., "It would be quite wrong for me to flash my headlights like that"), and two items assessing anticipated affect (e.g., "If I flashed my headlights like that I would feel bad afterwards"). Measures of attitudes, subjective norm, and perceived behavioral control accounted for around 51% of the variance in intentions to commit the two violations; moral norm and anticipated affect explained a further 3% to 5%. Employee Behaviors Early research in this behavioral domain using a measure of moral norm was not promising. Hom, Katerberg, and Hulin (1979) and Hom and Hulin (1981) studied the enlistment decisions of national guardsmen. Although they found that the TRA provided a better prediction of employee turnover than did measures of organizational commitment and job satisfaction, they also found that adding a measure of moral obligation to the TRA predictors did not result in any significant improvement in predictive utility. By contrast, a series of studies reported by Prestholdt, Lane, and Mathews (1987; Lane, Prestholdt, & Mathews, 1988, 1990) show that prediction of employee turnover in another profession, nursing, is improved by adding a measure of moral obligation to the TRA model. Specifically, in the Prestholdt et al. (1987) and Lane et al. (1988) studies, it was found that nurses' intentions to remain in or resign from their present job were predicted by attitude, subjective norm, and moral obligation, all three factors being significant predictors. For example, in the Prestholdt et al. (1987) study, moral obligation was assessed by calculating the difference scores between responses to the items "I have a moral obligation to remain on the staff of this hospital" and "I have a moral obligation to resign from the staff of this hospital." Although the entry of this predictor to the regression equation only added a further 1% to the 67% of variance in intentions explained by the other two factors, the beta weight for moral obligation was significant. In the Lane et al. (1990) study, the investigators examined the moderating role of marital status and educational background in determining which of the two normative factors in the model (i.e., subjective norm and moral obligation) was more influential on nurses' intentions. Being married, it was reasoned, would decrease the strength of moral obligation to stay in a job, and perhaps increase the role of subjective norm in forming an intention. Whether a nurse received a nonbaccalaureate (2-year) or baccalaureate (4-year) degree should also have an impact on sense of moral obligation, the authors argued, because the focus of the former type of program is on prepar-

< previous page

page_18

next page >

< previous page

page_19

next page > Page 19

ing technical nurses to perform a caregiver role and the emphasis is on loyalty and commitment, whereas the focus of the latter type of program is training professional nurses, and the emphasis is on personal development and professional skills. In short, moral obligation should play a stronger role among nonbaccalaureate than among baccalaureate nurses. Consistent with the authors' reasoning, moral obligation was found to be a significant independent predictor of intention to remain in or to resign from current job among unmarried (but not among married) and among nonbaccalaureate (but not among baccalaureate) nurses. This suggests that the role played by moral norms in forming behavioral intentions varies as a function of personal factors, as well as type of behavior. Business Ethics Randall and Gibson (1991) used the TPB to study ethical decision making by nurses. The participants were presented with four scenarios depicting instances of inadequate patient care and were asked if they would report the health professionals responsible for the situation to the appropriate authorities. Intention to report was the outcome variable to be predicted by measures of the remaining TPB constructs, together with moral obligation. The latter construct was measured using one item: "I believe that I have a moral obligation to report the [health professional] to my supervisor." Attitude, subjective norm, and perceived behavioral control together accounted for 61% of the variance in intentions; adding moral obligation to the regression equation explained a further 2%, a significant increment. Kurland (1996) examined the issue of sales agents receiving incentives in the form of commission, and how the availability of this commission influences the intentions of agents with respect to their clients. The participants in her research were drawn from a national group of (U.S.) financial service industry professionals. These participants were asked to complete a number of measures relating to one ethical decision, namely whether or not to disclose to a client, prior to recommending a product, information about the quality of the product and about the agent's commission on selling that product. Measures of the standard TPB constructs were taken with respect to this decision; in addition, participants were asked to respond to six items measuring perceived moral obligation, examples being "I believe I have a moral obligation to my clients to disclose this information," and "It is my responsibility as a professional to describe this information to my clients." The standard TPB predictors accounted for 46% of the variance in intentions to disclose the information; adding perceived moral obligation explained a further 12%, and this construct had the highest standardized regression coefficient.

< previous page

page_19

next page >

< previous page

page_20

next page > Page 20

Sexual Behavior Boyd and Wandersman (1991) used the TRA to study condom use in an undergraduate population. In addition to measuring the standard TRA constructs, they included a single-item measure of moral norm, described as "reflecting the subject's sense of moral responsibility to use a condom during the next 3 months" (p. 1819). Attitude to behavior and subjective norm together accounted for 38% of the variance in intention. The authors note that adding the moral norm measure to the equation increased explained variance significantly, although it is unclear exactly by how much. Richard, van der Pligt, and de Vries (1995) investigated the role of anticipated affective reactions to safe and unsafe sexual behaviors, in the context of the theory of planned behavior. The participants were 584 teenagers who had experience of sexual intercourse with more than one partner. They were asked to imagine being in each of three scenarios in which they and a new partner wanted to have sexual intercourse. As well as assessing attitudes to behavior, subjective norms, self-efficacy (comparable with perceived behavioral control), and behavioral expectations (comparable with intentions), the researchers assessed anticipated affective reactions by asking respondents to indicate how they would feel after (a) having sex without intercourse, (b) having intercourse using a condom, and (c) having intercourse without using a condom. Ratings were made on scales labeled worried-not worried, regret-no regret, and tense-relaxed. Two sets of theoretical models were analyzed using LISREL. One concerned the prediction of expectations regarding refraining from intercourse; the other concerned expectations regarding condom use. In both cases, the addition of anticipated affective reactions to a model equivalent to the TPB resulted in a significantly better fit. Furthermore, it was established that a model in which anticipated affective reactions were distinguished from attitudes to behavior provided a better fit with the data than did a model in which these two constructs were represented by a single latent variable. This study therefore established that it is worthwhile to make the distinction between anticipated affective reactions and attitudes to behavior, and also that once this distinction is made, adding the anticipated affective reactions construct to a model equivalent to the basic TPB results in a significant improvement in the amount of explained variance in behavioral expectations. Richard, van der Pligt, and de Vries (1996) replicated these findings using an older sample of 451 students who were asked similar types of questions in relation to a scenario in which they and a new or casual partner wanted to have sex. The model including anticipated affective reactions explained 65% of the variance in behavioral expectations regarding condom use in such a situation, and when the same respondents were questioned a month later it was found that, among those

< previous page

page_20

next page >

< previous page

page_21

next page > Page 21

who had had casual sex in the intervening month, expectations explained 35% of the variance in self-reported behavior. Eating and Drinking Behavior DeCourville and Zanna (1993a, 1993b) used an extended version of the theory of planned behavior to study alcohol consumption. In their 1993a study they examined intention to get drunk as a function of drinking status ("problem" drinker vs. "nonproblem" drinker, determined on the basis of how much alcohol was reportedly consumed in a typical drinking session). Theory of planned behavior constructs were assessed, supplemented by measures of affective beliefs about getting drunk and perceived immorality of getting drunk. The latter construct was assessed by a single item: "It is wrong for me to get drunk." When data from problem and nonproblem drinkers were analyzed together, it was found that attitude, affective beliefs and subjective norm together accounted for 59% of the variance in intentions to get drunk, perceived control added a further 3%, and perceived immorality a further (significant) 1%. However, when drinking status was added after these predictors, together with the interactions between drinking status and the other variables, it was found that several of these additional predictors explained significant amounts of variance in intentions. When data from the two groups of drinkers were analyzed separately, it was found that perceived immorality of drinking added significantly to the prediction of nonproblem drinkers' intentions to get drunk, explaining a further 5% of variance beyond the 43% already accounted for by attitude, affective beliefs, subjective norm, and perceived behavioral control; however, the intentions of the problem drinkers were predicted only by attitude, affective beliefs, and subjective norm: Neither perceived behavioral control nor perceived immorality played a role. The reported frequency of getting drunk was predominantly explained by intention, regardless of drinking status, although intention accounted for 39% of behavior in nonproblem drinkers and only 11% in problem drinkers. It would seem, then, that the perceived immorality of getting drunk served a deterrent role among the nonproblem drinkers, influencing the frequency with which they got drunk via intentions. DeCourville and Zanna's (1993b) study adopted a similar theoretical and methodological approach to examining the prediction of drinking intentions (to get drunk), and drinking behavior (frequency of getting drunk) in a larger sample of respondents (partly overlapping with those who participated in the 1993a study). Data were analyzed by means of structural equation modeling (EQS). Once again it was found that perceived morality (this time in combination with affective beliefs) added significantly to the

< previous page

page_21

next page >

< previous page

page_22

next page > Page 22

amount of variance in intentions that could be explained by the standard TPB constructs (from 46% to 51%). Raats, Shepherd, and Sparks (1995) examined the capacity of TPB constructs to predict intentions to drink milk products differing in fat content (whole milk, semiskim milk, and skim milk). As well as attitude, subjective norm, and perceived behavioral control, they assessed "perceived obligation for family's health." The latter was measured using a 1-item measure, "I feel obliged to use [product] for my family's health." Whether this captures the concept of moral obligation is somewhat open to question, but this predictor did add significantly to the prediction of intentions in relation to two of the three products (whole milk: from 29% to 31%; semiskim milk: from 36% to 37%). A final study involving intentions relating to food is the one reported by Sparks, Shepherd, and Frewer (1995). These authors focused on expectations of eating food produced using gene technology during the next 15 years, and expectations about supporting the use of gene technology in food production. In addition to measures of attitude, perceived attitudes of others (arguably a form of subjective norm), and perceived behavioral control, the authors assessed "perceived ethical obligation" using a 1-item measure: "I feel that I have an ethical obligation to avoid eating food produced by gene technology [to support the use of gene technology in food production]." Perceived ethical obligation proved to be an only marginally significant independent predictor of expectations when other predictors were also in the equation (.05 < p < .10). The Development of Moral Norms The literature just reviewed clearly establishes that moral norms have the capacity to increase the predictive utility of the TRA and TPB. This is consistent with the arguments developed earlier concerning the conceptual independence of the moral norm construct from other constructs that are included in these models of the attitude-behavior relation. Moral norms, it would appear, have a psychological status that is (at least partly) independent of attitudes to behavior, subjective norms, and intentions. This raises the interesting question of how moral norms develop. To address this question I draw on arguments and evidence presented mainly by developmental psychologists. As Gibbs (1991) noted, moral development is an issue that is tackled by two major contemporary theories: Kohlberg's (1984) cognitive-developmental theory, and Hoffman's (1983) moral socialization theory. A key difference between these theories is the extent to which the individual is regarded as internalizing moral norms and values. Kohlberg's (1984) posi-

< previous page

page_22

next page >

< previous page

page_23

next page > Page 23

tion does acknowledge "a trend toward an increasingly internal orientation to norms" but goes on to state quite clearly that "this development cannot be defined as a direct internalization of external cultural norms" (p. 90). Instead, argued Kohlberg, what changes is the nature of moral judgment, which develops from superficial to profound. This development parallels cognitive development, and Kohlberg described it as reflecting the child's maturing ability to make more profound moral judgments. The dynamic force driving this change, then, comes from within the developing individual, which is why Kohlberg resisted the notion of simple internalization of external norms. What changes within the individual as a function of cognitive maturity is the capacity to engage in what Flavell (1985) referred to as "decentration." This is the ability to focus on attributes of the situation or stimulus other than the most salient one or two, and thereby to grasp the "underlying" (as opposed to superficial) reality. Gibbs (1991) extended Flavell's analysis by suggesting that "decentration generates logical and moral prescriptions of equality and reciprocity, in social-cognitive as well as cognitive development. In the social realm, as the child's attention becomes more extensively and equally distributed in social situations, prescriptions of social equality or impartiality tend to emerge. Such an ideal is central to moral development" (pp. 91-92). Thus, the motor for moral development is the child's maturing ability to regard a social situation in ways that go beyond the most obvious (e.g., what would benefit himself or herself) to include the less salient (e.g., how his or her own actions conceivably could harm others). Hoffman's (1983) theory, by contrast, is concerned explicitly with the process of internalizing initially external norms. Thus a norm that is originally external becomes accepted, or internalized, and is then perceived as having emanated from oneself. Especially interesting is Hoffman's (1983) claim that a moral orientation "characterized by independence of external sanctions and by high guilt is associated with the use of . . . inductions, which are discipline techniques that point up the effects of the child's behavior on others" (p. 246). Inductions make moral internalization necessary, because they focus on the consequences the child's behavior has for another and thereby allow the source of this information (i.e., the caregiver or parent) to remain relatively nonsalient. To the extent that the induction information is more salient than the source of the information, conditions are right for self-attribution and internalization. Inductive discipline also capitalizes on a "motivational resource that exists in the child from an early age, namely, the child's capacity for empathy, defined as a vicarious affective response to others" (Hoffman, 1983, p. 252). The combination of the child's empathic distress and awareness of being the agent causing that distress elicits feelings of guilt. As Gibbs (1991) observed:

< previous page

page_23

next page >

< previous page

page_24

next page > Page 24

The prosocial cognition embodied in the inductive teaching gains motivational properties through classical conditioning with empathic affect and its derivative, empathy-based guilt. In the subsequent internalized morality, as empathy and anticipatory guilt are activated in situations of high moral conflict or temptation, the inductiontaught prosocial cognition is endowed with sufficient motive-power to overcome (at least usually) egoistic motives or aggressive impulses. The result is moral conduct. (pp. 94-95) This internalization of moral norms via inductions can be contrasted with a more external moral orientation which in Hoffman's view arises from a different pattern of socialization. For example, power-assertive discipline, especially when administered harshly and arbitrarily, tends to enhance an external orientation in which the main focus is on fear of detection and punishment. Thus, the chances are that even if moral norms are prescribed by the parent, the only norm that is internalized is that one should try to avoid being caught by those with power or authority and that social conflicts should be resolved by using force. Although Hoffman's theoretical analysis lends itself more readily to an understanding of how moral norms come to be internalized, his approach and that of Kohlberg are not as diametrically opposed as the foregoing description might suggest. Indeed, the main thrust of Gibbs' (1991) argument is that these two approaches should be regarded as compatible and complementary rather than as competing. Granted, Kohlberg tends to emphasize the way in which the individual cognitively constructs morality, although Hoffman's analysis focuses on the internalization of social norms, but both theorists are ultimately concerned with how the individual develops an internal moral orientation. The ability to decenter, and, thereby to regard a situation from another person's perspective, is clearly compatible with the capacity to empathize with another's plighteven if the former is primarily a cognitively driven process and the latter an affectively driven one. Gibbs nevertheless concluded that Hoffman's theory provides the better account of the development of that aspect of morality that involves self-attribution (e.g., of prosocial concern) and sanctionsindependent feelings (e.g., of guilt). This type of internal morality comes closest, I suggest, to the concept of moral norm as implicitly theorized and explicitly operationalized in attitude-behavior research. What shapes the intentions of the individual who espouses a moral norm in a given behavioral domain is probably the prosocial concern that it is right to act in a certain way and the anticipation of sanctions-independent feelings of regret or guilt that would arise if one were to breach the norm in question. A slightly different but nevertheless quite compatible approach to understanding the process by which moral norms come to be internalized is represented by Vygotsky's (1978, 1981) sociocultural theory (cf. Buzzelli, 1991).

< previous page

page_24

next page >

< previous page

page_25

next page > Page 25

Vygotsky saw language, and especially private or ''inner" speech, as playing a central role in the development of selfregulation. It is argued that children's talk about their actions as they perform them serves to guide their behavior. Furthermore, the language that children use reflects the language they have heard used in previous dialogues with adults. With specific reference to moral development, then, Vygotsky's theory suggests that a child's moral understanding is importantly influenced by "dialogue with adults who interpret and frame rules within a moral context that reflects their own unique perspectives. While children do construct their own moral norms and use them to guide their behavior in moral situations, this process is a social one influenced by the words adults use in their dialogue with children" (Buzzelli, 1991, p.83). Buzzelli suggested that different adult reactions to a specific incident, such as taking another child's toy, serve to frame that action as "stealing and inherently wrong," as "against the school rules," or as upsetting for the child whose toy has been taken. Some types of framing do, it is argued, help the adult and child to arrive at a shared meaning of the action, and this shared meaning later serves as the basis for the child's construction of a moral norm governing that behavior. Thus, a frame that simply emphasizes that a rule has been broken may help the child to remember the rule, but it does little to help the child to construct a moral norm. A frame that helps the child to arrive at a shared meaning of the event, by contrast, should enable the child to retell the rule later. Such a frame is more likely to arise, Buzzelli suggested, if the adult begins by asking where the child got the toy, whether the child found it, whether it belongs to another child, what it means to take a toy belonging to another child, and what the child thinks the other child feels. The idea here is that the adult's speech builds on the child's response to each question and thereby has a better chance of creating a shared meaning of the situation. Such shared meanings provide a basis for the construction of moral norms that are later used by the child to regulate his or her own behavior. Thus, a Vygotskyan approach to understanding the internalization of moral norms can be seen as complementary to Hoffman's analysis, but it places much more emphasis on the role of language, initially in the from of dialogue between caregiver and child, and later in the form of private speech that regulates own behavior. The crucial factor for Vygotsky is the creation of meanings shared between children and adults: Without such intersubjectivity, the child fails to redefine his or her actions from the adult's perspective and thereby fails to construct a moral norm with self-regulatory force. All three perspectives on the development of moral norms share the notion that individuals differ in the extent to which they have an internalized moral orientation, and that these individual differences have clear developmental antecedents. They differ in their analysis of what these antecedents

< previous page

page_25

next page >

< previous page

page_26

next page > Page 26

are. As we have seen, Hoffman and Vygotsky attach more importance than does Kohlberg to interactions between caregiver and child as processes by which a child arrives at an internal moral orientation. This emphasis on social processes is consistent with evidence from one of the very few published studies in which the TRA has been used to analyze moral behavior (Vallerand, Deshaies, Cuerrier, Pelletier, & Mongeau, 1992). Participants in Vallerand et al.'s study were athletes who answered questions about two hypothetical sports situations raising moral issues: whether or not to criticize an official for making a bad call that cost the athlete the event, and whether to inform an official of an undeserved outcome of one's own, if doing so would cost one the event. One finding to emerge from this study was that the overall fit of the model to the observed data was significantly improved by adding noncausal paths relating normative beliefs, supposedly a component of the subjective norm factor, to behavioral beliefs and outcome evaluations, the two components of attitude toward the behavior. In discussing this relationship, the authors suggested that it can be understood by positing the existence of a type of belief that underlies both behavioral and normative beliefs. This common belief, they suggested, might be "one's beliefs about what should or ought to be done" (p. 106)that is, one's moral norm. A second, related finding was that the overall fit of the model to the observed data was significantly improved by adding a causal path from normative beliefs to attitudes toward the behavior. Adding this path improved the explained variance by no less than 35%. Thus the perceived expectations of others appeared to play a key role in shaping individuals' attitudes to the two moral behaviors under investigation. Vallerand and his colleagues were led to conclude that the social or normative component of the TRA has an important bearing on individuals' intentions to engage or not to engage in these behaviors: Normative beliefs, or the beliefs regarding what is perceived as appropriate in the eyes of important others, appear to loom large in the learning of moral beliefs, evaluations, attitudes, and subjective norms, and probably for good reasons. How do we learn that moral behaviors are to be valued (or devalued) if not through the influence of important others, such as parents, teachers, and peers? (Vallerand et al., 1992, p. 108) Conclusions Where a measure of moral norm has been added to measures of the standard TRA and TPB constructs, there is rather consistent evidence that it adds significantly to the prediction of behavioral intentions. In only three of the studies reviewed above was there less than compelling support for the independent predictive utility of moral norm, and in these cases it is reasonably easy to generate post hoc explanations for fact that moral norm played

< previous page

page_26

next page >

< previous page

page_27

next page > Page 27

a weak or nonexistent role. In the studies reported by Hom et al. (1979) and Hom and Hulin (1981), the target behavior concerned the enlistment decisions of national guardsmen. Whether or not a national guardsman enlists is hardly the sort of issue to create conflict between what the individual himself wishes (or what he sees important others as wishing) and one or more cherished moral norms. This stands in contrast to Lane et al.'s (1990) finding that nonbaccalaureate nurses' decisions to stay or leave their hospital were significantly better predicted if a measure of moral obligation was added to the standard predictors: The training of these nurses emphasized loyalty and commitment, thereby rendering the moral conflict more salient. A third study yielding only marginally significant evidence in favor of the additional predictive utility of the moral norm construct was the study of intentions to avoid eating food produced by gene technology, reported by Sparks et al. (1995). In this case it is possible that some respondents were simply unaware of what is involved in producing food via genetic manipulations and therefore failed to construe this behavior as one that engaged a moral norm. An alternative explanation would center on a possible lack of conflict between perceived moral obligation and attitudes in this domain: Perhaps respondents' attitudes to eating food developed by gene technology are shaped in part by moral considerations, thereby limiting the extent to which a separate measure of moral obligation can account for additional variance. Although the additional variance accounted for by moral norm in the remaining studies was not always large in absolute terms, such results need to be interpreted in the light of the following considerations: (a) Moral norm and associated constructs have often been assessed using single-item measures, thereby increasing the likelihood that these measures contain error, and as a result reducing the chance that they will be related to other measures; and (b) the typical way in which the role of moral norm has been evaluated in these studies is by entering measures of this construct after the predictors specified by the TRA or TPB have been entered into the regression equation. Although this is an excellent procedure for testing whether the moral norm construct offers added explanatory value beyond that provided by the TRA or TPB, it runs the risk of underestimating the predictive utility of this construct. It is not unusual for the beta weight associated with moral norm to be the highest in the final regression equation, showing that although the addition of this variable may not account for a great deal of extra variance, it is in fact often a more powerful predictor of intentions than the standard TRA or TPB constructs. The research evidence reviewed in this chapter leaves several questions unanswered. Most obvious is the absence of any clear basis for identifying the conditions under which moral norms add predictive or explanatory

< previous page

page_27

next page >

< previous page

page_28

next page > Page 28

value to the standard TRA or TPB constructs. To be consistent with the general spirit of these models, one needs to look for behavior-specific factors as determinants of moral beliefs: What is it about a given behavior that elicits more or fewer moral considerations? Although it is clear that moral norms are unlikely to add predictive value unless the behavior in question raises moral issues, there is no a priori way to determine when these issues will be raised. Furthermore, the fact that engaging in a given behavior touches on moral values is unlikely to be a sufficient criterion for determining when the addition of moral norm will enhance the predictive power of the basic TRA or TPB model. Much will depend on the extent to which the moral norm differs from attitude to behavior and subjective norm. Obviously, if moral norm and either attitude to behavior or subjective norm are entirely congruent, adding moral norm is very unlikely to enhance the prediction of intention. However, when the assessment of moral norm (or anticipated affective reactions) captures a perspective on the behavior in question that is independent of (and perhaps in conflict with) attitude toward behavior and subjective norm, it has at least the potential to account for variance in intention that remains unexplained by the other two factors. Prime candidates for behavioral domains in which moral norms may diverge from attitudes or subjective norms are those in which the satisfaction of individual or social goals runs counter to moral imperatives. The individual goals might involve the avoidance of discomfort or threat, the attainment of financial reward or personal advancement, or the satisfaction of physical appetites. The social goals involved might be obedience to an authority figure or simply fulfilling another person's immediate expectations or wishes. When meeting these goals conflicts with personally held moral imperatives, the potential for a measure of moral norm to add to the predictive utility of the basic TRA or TPB model will, I suggest, be greatest. Empirical evidence is needed to test these propositions. Adding to the predictive utility of the TRA or TPB is an important but modest goal. Of greater interest to most researchers is enhancing the explanatory power of the TRA or TPB. Showing that a moral norm measure explains significantly more variance in intentions than can be accounted for by attitudes and subjective norms does not help us to understand the underlying processes unless we have a clear conception of what moral norms entail and how they are formed. Hence, attention in this chapter is paid to the development of moral norms. As we saw, there are several theoretical accounts of the process by which individuals come to develop an internal moral orientation. Although Kohlberg's (1984) theory emphasizes the role of individual cognitive development more than do the explanations offered by Hoffman (1983) and Vygotsky (1978, 1981), one aspect of cognitive development that is vital to the development of a moral orientation is

< previous page

page_28

next page >

< previous page

page_29

next page > Page 29

decentering. In social terms, this entails the ability to regard a situation from another person's perspective. This provides a point of contact with Hoffman's approach, in which empathic response to another's suffering plays an important role. A style of discipline that capitalizes on this empathy is one that emphasizes the impact of the child's behavior on others. Here is a means by which originally external moral norms come to be internalized. Vygotsky's approach also treats moral norms as internalized guides to behavior but places greater emphasis on the construction of these norms via the use of dialogue between caregivers and children. If pitched at an appropriate level, this dialogue establishes a shared meaning of the situation, and the child will draw on this shared meaning later in regulating his or her own behavior in a comparable situation. Returning to the TRA and TPB, it is not difficult to see how these models may fail to take proper account of the influence of moral norms on the formation of behavioral intentions. The two streams of influence on intentions represented in the TRA are either personal (attitude to behavior) or social (subjective norm), and the additional factor included in the TPB (i.e., perceived behavioral control) is a mix of personal (e.g., competence) and situational (e.g., task difficulty) factors. None of these constructs adequately captures the peculiar blend of the social and the personal that results in moral norm. This is not to say that moral norm cannot be assessed by means of a relatively simple adaptation of the usual ''direct" measure of attitude to behavior. Recall that respondents are typically asked to rate performance of the target behavior on a number of semantic differential scales, examples being good-bad, harmfulbeneficial, pleasant-unpleasant. Where researchers have any grounds for thinking that intentions might be influenced by moral concerns, they would be well advised to add two semantic differential scales to this set: right-wrong, and moralimmoral. Furthermore, ratings made on these scales should not simply be summed or averaged with ratings made on the remaining scales. If the arguments and evidence presented are correct, there should be some behavioral domains in which ratings of the morality of performing a behavior will be to some degree independent of other evaluations of the behavior. Thus, it would be worthwhile to check whether a two-factor model provides a better fit with the data than a one-factor model. This better statistical fit may well be associated with superior prediction of intentions. In conclusion, there are good conceptual and theoretical grounds for arguing that moral norm represents a potential influence on behavioral intention that is independent of attitude to behavior, subjective norm, and perceived behavioral control. Individual differences in cognitive development and socialization histories are likely to result in individual differences in moral norm, at least for certain types of behavior. Especially where indi-

< previous page

page_29

next page >

< previous page

page_30

next page > Page 30

vidual and social rewards conflict with moral norms, it is likely that a separate measure of moral norm will enhance not only the predictive utility of the theories of reasoned action and planned behavior, but also their capacity to provide an explanation for and an understanding of variations in intention and behavior. Acknowledgments I thank Icek Ajzen, Nick Emler, Paul Sparks, and the editors for helpful comments on a previous draft of this chapter.

< previous page

page_30

next page >

< previous page

page_31

next page > Page 31

3 Subjective Expected Utility-Based Attitude-Behavior Models: The Utility of Self-Identity Paul Sparks Institute of Food Research Introduction This chapter focuses on the potential role of self-identity in the structure of the theory of reasoned action (TRA; Ajzen & Fishbein, 1980) and the theory of planned behavior (TPB; Ajzen, 1991), perhaps the best known examples of subjective expected utility-based (SEU) attitude-behavior models. Although there is some evidence for self-identity effects in applications of these models, strong theoretical rationales for such effects have not yet been forthcoming. However, this chapter is intended to identify and address a number of potential lines of investigation that may assist in understanding a potential role for self-identity in attitude-behavior models of this kind. The argument is made that selfidentity represents a potentially useful candidate to extend the TPB, with a view to Ajzen's (1991) frequently noted suggestion that "The theory of planned behavior is, in principle, open to the inclusion of additional predictors if it can be shown that they capture a significant proportion of the variance in intention or behavior after the theory's current variables have been taken into account" (p. 199). Because self-identity and attitudes are central to social psychological theorizing and research, it is appropriate to promote the examination of potential points of contact between these concepts.

< previous page

page_31

next page >

< previous page

page_32

next page > Page 32

The TRA proposes that behavior results from intentions and that intentions, in turn, arise from people's attitudes ("a general feeling of favorableness or unfavorableness," Ajzen & Fishbein, 1980, p. 54) and their subjective norm (the perception that important others think that the person should or should not carry out the behavior). For behaviors that are under "incomplete volitional control" (Ajzen, 1991, p. 181), the TPB adds a measure of perceived behavioral control, alongside attitudes and subjective norm, as a predictor of intentions. Perceived behavioral control is also envisaged as sometimes having a direct predictive effect on behavior in conditions when people's perceptions of control match the amount of actual control that they are able to exercise. The antecedents of attitudes are suggested as residing in a combination of the beliefs that people have about the attributes of a behavior (behavioral beliefs) and the values that people ascribe to those attributes (outcome evaluations) In structurally similar ways, the antecedents of subjective norm are proposed as being represented by people's normative beliefs and their motivation to comply, and perceived behavioral control is said to be determined by control beliefs and power. The TRA has been addressed in numerous empirical applications (see e.g., Eagly & Chaiken, 1993); the TPB has also been very successful since its introduction, in the sense that a number of studies have indicated independent predictive effects for perceived behavioral control on people's intentions and behavior (see e.g., Ajzen, 1991; Conner & Sparks, 1996; Manstead, 1996). This chapter focuses on the value or desirability of incorporating, or attempting to incorporate, self-identity within the framework of these models: On the one hand, an assessment is made of the utility of the inclusion of self-identity in such models; on the other hand, and as a part of this, passing mention is made about the value of self-identity for people themselves as a guide to social action. After a rudimentary discussion of the basic concepts central to the theme of this chapter, the empirical evidence for the role of self-identity in the structure of the TRA and TPB is outlined. Following this, some speculative assessment is made of the reasons why one might expect independent predictive effects of measures of self-identity in the structure of these models: Here the discussion encroaches on behavioral decision theory, functional approaches to attitudes, and perspectives on ambivalence. None of the empirical work mentioned is described in any detail (full accounts are available elsewhere). Instead, the bulk of the chapter is given over to a general discussion of some potential directions for future research. The preoccupation of this chapter may consequently be viewed as rather parochial, concerned as it is with one orientation to self-identity in the structure of a particular kind of theory of attitude-behavior relations. However, the implications of

< previous page

page_32

next page >

< previous page

page_33

next page > Page 33

what is discussed extend beyond this rather narrow concern and bear on the ways in which the concepts of attitude and identity are used in social psychological theory and research. Subjective Expected Utility At the outset, a working definition of the concept of utility is appropriate. Here (following Baron, 1988), utility is synonymous with subjective value and desirability: "The concept of utility respects the variety of human goals. It represents whatever people want to achieve. Some people do not want "pleasure" as much as they want other things (such as virtue, productive work, enlightenment, respect, or loveeven when these are painful things to have). The utility of an outcome is also different from the amount of money we would pay to achieve it" (p. 287). Such a definition has its roots in utilitarianism in moral philosophy. This utilitarian doctrine has many different formulations, perhaps the most famous being the dictum of John Stuart Mill (1863, reprinted in Ryan, 1987, p. 278) that "actions are right in proportion as they tend to promote happiness, wrong as they tend to produce the reverse of happiness.". There have been various criticisms of the apparent focus on hedonic factors, and Baron's careful interpretation of the utility concept reflects these kinds of concern. Baron's is a broader definition than that which is often applied in vernacular discourse and in psychological theorizing: The latter often tends to treat utilitarian (or instrumental) concerns as synonymous with concerns about tangible, material outcomes (or, at least, contrasts them with more "symbolic" concerns). This contrast is not made here. Subjective expected utility theory is initially based on the assumption that people seek to "maximize the product of utility and subjective probability" (Edwards, 1967, p. 67). It is essentially a theory of decision making in contexts or situations where several choice options may be available. The focus in the theory on subjective experience can be seen as a development from earlier models that were concerned with objective probabilities and objective values. These models encountered serious problems: hence the development of models that incorporated the perspective of the decision maker. However, it soon became apparent that SEU was not a good descriptive model of how people actually go about making decisions (see Schoemaker, 1982; see e.g., Baron, 1988, van der Pligt, 1996, for accounts of decision making from a psychological perspective). With this in mind, it is interesting to note that Eagly (1992) described the popularity of expectancy-value models in applied social psychology as "unprecedented" (p. 695). The TRA is explicitly based on SEU, because it proposes that attitudes arise as a function of the summed products of behavioral beliefs (cf. subjec-

< previous page

page_33

next page >

< previous page

page_34

next page > Page 34

tive probabilities) and outcome evaluations: The way that attitudes are seen to be a function of behavioral beliefs and outcome evaluations is "essentially equivalent" (Ajzen & Fishbein, 1972, p. 2) to the structure of the relations between SEU, subjective probability, and subjective value. The question as to whether (or when) people actually go through an elaborate weighing up of costs and benefits in making decisions is a moot point (e.g., Eagly & Chaiken, 1993). However, most researchers using the TRA have not really been concerned with the descriptive accuracy of the SEUbased combination of behavioral beliefs and outcome evaluations; rather they have been more concerned with its predictive usefulness. I return to this point later. Self-Identity In sociological theorizing, identity has been notoriously ambiguous (e.g., Gleason, 1983); a philosopher is likely to interpret personal identity in quite a different way than is a social psychologist; the notion of social identity may be used by some sociologists (e.g., Goffman, 1959) in a way quite at odds with the way the concept is used in, for example, social identity theory (SIT; see chapter 5, this volume by Terry, Hogg, & White). It is consequently important to recognize that self and identity are used in quite different ways, both within and between disciplines. William James' (1890/1983) seminal chapter on the self has been described as the most widely cited reference in the self literature in psychology (Scheibe, 1985). James distinguished the Material Self, the Social Self, the Spiritual Self and the Pure Ego. The first three of these are subsumed under what he termed the Empirical Self. The Material Self is represented essentially by the material body, although James also included relations and possessions in this domain; the Social Self is described as "the recognition which he gets from his mates" (p. 281); the Spiritual Self refers to "a man's inner or subjective being" (p. 283); finally, the Pure Ego is described as that "pure principle of personal identity" (p. 314). The notion of the Social Self has been of particular interest to social psychologists, because it combines a concern with how people's social behavior varies not only as a function of different social roles but also as a function of the kind of social others with whom a person is interacting. It is also of interest because it is represented by images of the individual that are carried by other individuals: Thus, although the Social Self is clearly related to an individual's social behavior, the image of the individual in others' minds provides the essence of the concept. These ideas about social dimensions to the self have been developed in various ways: from Cooley's (1902/1964) ideas about the looking-glass self where our self-idea consists in part of "the imagination of our appearance to

< previous page

page_34

next page >

< previous page

page_35

next page > Page 35

the other person'' (see Shrauger & Schoenemann, 1979, p. 152), through to concerns with self-fulfilling prophecies (Merton, 1968), self-presentation (Goffman, 1959), and the relation between authority and identity (Holzner & Robertson, 1980). The widespread idea that a person's self-identity or self-concept influences their experiences, cognitions and behaviors (e.g., Gergen, 1971; Rosenberg, 1981; Turner, 1982) suggests a further important social dimension to the concept: "Society not only defines but creates psychological reality. The individual realizes himself in societythat is, he recognizes his identity in socially defined terms and those definitions become reality as he lives in society" (Berger, 1966, p. 108). In this chapter, self-identity may be interpreted as synonymous with self-perception or self-concept. Essentially, it refers to the relatively enduring characteristics that people ascribe to themselves, which take the form of (or incorporate) socially given linguistic categorizations. Under this interpretation, social identity (as interpreted in SIT) is subsumed as a particular kind of self-identity. The SIT distinction between personal and social identity is not considered to be a crucial distinction for the themes developed here, although this distinction may prove to have important ramifications for attitude-behavior relations in other ways. Social influences are likely to be involved in the constructions of different forms of self-identity (whether, e.g., group membership, social role, or personality characteristic identifications): However, the usual SIT distinction between personal and social identities relates to their content rather than to the formative influences and processes involved in their construction (although these are likely to vary for different forms of self-identity). Self-Identity In Attitude-Behavior Models Research in the late 1980s gave a hint that self-identity might have an important role to play in attitude-behavior relations. The claim that self-identity will contribute to the prediction of behavioral intentions independently of attitudes is made, and empirically illustrated, in the work of Biddle, Bank, and Slavings (1987) and Chamg, Piliavin, and Callero (1988). Evidence for the effect of self-identity on behavior independent of behavioral intentions is provided by Granberg and Holmberg (1990). All these studies are cited in a recent discussion of the role of self-identity in the structure of the TRA (Eagly & Chaiken, 1993). Biddle et al. (1987), in a study of college retention decisions, suggested that self-referent identity labeling has an effect on behavior that is independent of the effect of individual preferences. For these authors, self-referent identity labeling is a generative force behind behavior, as are attitudes or

< previous page

page_35

next page >

< previous page

page_36

next page > Page 36

preferences and norms. Norms here refer to "prescriptive or proscriptive standards" (p. 325). Biddle et al. (1987) suggested that self-identity and other cognitions (such as attitudes or preferences) may not always be consistent, because the following line of thought is quite conceivable: "I would enjoy doing A, and I think I should do A (or B), but I am the type of person more oriented to doing C" (p. 326). In the empirical research that they report, Biddle et al. found that students' ratings of the sort of person they thought they were (e.g., ''someone who seeks a specific career that requires college") contributed to intentions independently of the contribution of preferences. Charng et al. (1988) suggested that repeated behaviors influence a person's self-concept, which then becomes important to that person. They suggested that carrying out such behaviors then "conveys meaning over and above the positive or negative attitudes we may hold toward performing the behavior itself" (p. 304). They went on to propose that role identity ("a set of characteristics or expectations that simultaneously is defined by a social position in the community and becomes a dimension of an actor's self" p. 304) as a blood donor may influence intentions to donate blood independently of attitudes toward blood donation. In their empirical study, they demonstrated that regressions of intentions to donate blood on attitudes and role identity revealed an independent effect for each of the two predictor variables. Moreover, these authors were able to show that the effects of role identity as a blood donor increases with increments in the frequency of blood donation. Following Charng et al. (1988), Granberg and Holmberg (1990), in a study of voting intentions, found that self-identity and prior behavior had independent effects on behavioral intention and on voting behavior. No measures of attitude or subjective norm were taken in their study. Biddle et al. (1987) proposed that different "intellectual traditions" of self-concept and attitude perspectives in the literature may be partly responsible for the lack of integration of the two orientations. Although there is probably some truth in this, Sparks and Shepherd (1992) thought it likely that many attitude researchers aligned to the theoretical position of Fishbein and Ajzen would consider that a person's self-identity would be reflected in that person's beliefs, values, and attitudes and that the inclusion of self-identity as a model component should not represent a theoretical or empirical advance. Sparks and Shepherd (1992) reported some initial skepticism about the applicability of the findings of Biddle et al. (1987) and of Charng et al. (1988) to the TRA and TPB. Although they considered that a person's self-identity is conceptually distinct from his or her evaluative attitudes, and that there is likely to be a bidirectional causal link between a person's

< previous page

page_36

next page >

< previous page

page_37

next page > Page 37

self-identity and his or her attitudinal evaluations, they felt that there is unlikely to be a causal link from a person's selfidentity to behavioral intentions that is independent of the effect of those attitudinal evaluations. That is, they suggested that if people identify themselves as blood donors, as in the study of Charng et al. (1988), then this would be reflected in favorable attitudes towards blood donation, so long as the behavior were voluntary and not coerced or otherwise associated with control problems. Likewise when Biddle et al. (1987) provide the example of the person who thinks "I would enjoy doing A, and I think I should do A (or B), but I am the type of person more oriented to doing C", they understood the person in the example to be, on balance, favorably disposed to doing C, despite any reservations they may have and the lack of enjoyment involved in doing C. Sparks and Shepherd also argued that if a person's self-identity, per se, is important to him or her (as suggested by Biddle et al., 1987) then the creation, affirmation, or bolstering of self-identity through the performance of behavior could be validly as an outcome of the behavior (albeit possibly difficult to articulate and less salient in relation to other more tangible outcomes; cf. Eagly & Chaiken, 1993). Although considering that people's self-identities would find expression in their attitudes, Sparks and Shepherd (1992) noted that the attitude measures used by Charng et al. (1988) and by Biddle et al. (1987) were not the standard measures that Ajzen and Fishbein suggested (cf. Ajzen & Fishbein, 1980). That is, these studies were not specifically structured within the framework of the TRA or the TPB. In fact, although Sparks and Shepherd (1992) took a skeptical view of the potential for an independent role of self-identity in the prediction of behavioral intentions and behavior within the framework of the TPB, their subsequent findings demonstrated independent effects of a measure of selfidentity (as "green" consumers) on intentions to consume organically produced vegetables. Self-Identity and Food A research domain in which self-identity has been examined within the TPB is that of food choice (Guthrie & Sparks, 1997; Sparks & Shepherd, 1992; Sparks, Shepherd, Wieringa, & Zimmermanns, 1995). Although this can be seen as an archetypal "consumer" activity in which self-identity effects might be less expected than in behavioral domains in which group or role identities may be more salient, food choice research attracts many discussions relevant to the current theme. Sociologists and anthropologists have written much on the concept of food and identity. Bourdieu (1984), for example, noted the ways that taste and consumption patterns reflect class identities; Fischler (1988) described the ways that food consumption

< previous page

page_37

next page >

< previous page

page_38

next page > Page 38

involves not only physical absorption of the food into the body but also the ways that the symbolic integration of food characteristics might affect identity (e.g., strength from blood ingestion). Experimental evidence also suggests that people make judgments about others based on the food that they eat (Guthrie & Sparks, 1996; Stein & Nemeroff, 1995); the aphorisms "you are what you eat" and "tell me what you eat and I will tell you what you are" are often noted approvingly in this research area. Conceived more broadly, consumption, like food, can function to establish and bolster a sense of identity and to express differences between social groups (Bourdieu, 1984; Fischler, 1988): "Commodities are not just objects of economic exchange; they are goods to think with, goods to speak with'' (Fiske, 1989, cited in Bocock, 1993, p. 96). The focus of the current chapter is not so much a discussion of whether identity issues are relevant to food choice but is more disciplinarily limited than this: Specifically, it addresses whether a consideration of identity might add to the predictive abilities of models of attitude-behavior relationships such as the TPB. Since the Sparks and Shepherd (1992) research, a number of studies in the domain of food choice behavior have supported further the proposal that selfidentity might provide a useful addition to the structure of the TPB model. Sparks, Shepherd, Wieringa, and Zimmermanns (1995) found an independent effect for a measure of self-identity (as someone concerned with the health consequences of food consumption) on people's expectations of reducing their consumption of a number of different foods contributing to high levels of fat consumption. Dennison and Shepherd (1995), in a study of the food choices of more than 600 adolescents, found that a measure of health conscious self-identity had independent predictive effects on intentions to eat fruit and to eat chips (french fries). Armitage and Conner (1997) showed that self-identification as a healthy eater contributed an independent predictive effect in multiple regressions of undergraduate students' intentions to eat a low-fat diet. Guthrie and Sparks (1997) found that self-identification as a health-conscious consumer had independent predictive effects on respondents' intentions to eat a diet low in fat. Sparks and Guthrie (1998), in a recent cross-national study of attitudes toward consuming diets that are low in animal fats, found independent predictive effects for a measure of identification as a health conscious consumer. Effects of self-identity within the structure of the TRA and TPB also have been reported in other behavioral domains. For example, Theodorakis (1994), in a study of exercise behavior, found predictive effects for a measure of self ("role") identity on female participants' intentions and behavior with respect to participation in a program of exercise activity.

< previous page

page_38

next page >

< previous page

page_39

next page > Page 39

Although there has been some recent accumulation of evidence for self-identity effects within the TRA and TPB, in most of these studies, little in the way of theoretical advance has been made to explain the effects that have been reported. Problems with self-identity measures have also been noted. In the Sparks, Shepherd, Wieringa, and Zimmermanns (1995) study, for example, the authors raised the concern that a measure of self-identity as someone who is health-conscious may merely make certain healthrelated consequences salient: That is, the measure of self-identity may simply make one of the outcomes of a behavior that could potentially influence attitudes to be cognitively available. This essentially echoes the suggestion of Eagly and Chaiken (1993) that self-identity could be subsumed under attitudes if it were not for the fact that independent effects on intentions have been demonstrated: They add that self-identity may not be ''especially salient when respondents rate behaviors on the evaluative scales used to assess attitude toward the act" (p. 178). The nagging concern also remains that measures of identity serve as measures of past behavior (with people possibly inferring their self-identities from an examination of their past behavior), although independent predictive effects of self-identity have persisted even in the presence of measures of past behavior (Sparks & Shepherd, 1992). Fishbein's (1997) suggestion that some measures of "self-image" may essentially constitute measures of behavioral intention also requires examination. Although studies evidencing independent predictive effects for measures of self-identity within the TPB are accumulating rapidly, theoretical developments by which one might understand these effects and examine them more thoroughly and critically have been slow in appearing. However, Kelly and Breinlinger (1995), in a recent study of women's participation in actions to promote women's issues, suggested that the compensatory weighing-up of pros and cons that characterizes the SEU underpinnings of the Sbb.oe component (i.e., the summed product of behavioral beliefs and outcome evaluations) of the TRA and TPB are less characteristic for those people with strong identifications with a social group for whom action is motivated by the affirmation and bolstering of social identities. Their data were congruent with this suggestion, with relations between attitude (measured as (Sbb.oe) and intentions significantly stronger for those with weak identification as someone involved in the promotion of women's issues than for those with strong identifications. This suggestion is similar to that made by Abelson (1982) who suggested that "symbolic attitudes," which are characterized by a great deal of emotional and self-identity involvement, may sometimes be associated with "surprising insensitivity to rational-instrumental considerations" (p. 140). Abelson noted also the moral dimension of such attitudes and the "desultory attention'' that they have

< previous page

page_39

next page >

< previous page

page_40

next page > Page 40

received in the literature. Likewise, in a discussion of the role of rules and principles in decision making, Prelec and Herrnstein (1991) talked about how the use of rules "disengage the cost-benefit calculus" (p. 323). Terry, Hogg, and White (1997) found that a measure of the importance of household recycling to the self-concept made an independent contribution to intentions to participate in household recycling during the subsequent 2 weeks. Their research on self-identity, social identification, and group norms is discussed in chapter 5. Sparks and Shepherd (1992) indicated that many forms of self-identity may involve a moral component; moral norms (variously defined) have also been shown to have independent predictive effects in applications of the TRA (see, e.g., Gorsuch & Ortberg, 1983; Zuckerman & Reis, 1978) and the TPB (Beck & Ajzen, 1991; Sparks, Shepherd & Frewer, 1995; Parker, Manstead, & Stradling, 1995; Raats, Shepherd, & Sparks, 1995; see Manstead, chap. 2, this volume). Accounting for Self-Identity Effects One might understand self-identity as having a predictive effect on intentions and behavior, independent of the effect of attitudes and other components of the TPB, if self-identity encapsulates people's interests or goals that are distinct from those that are expressed by their attitudes. This issue relates to a fundamental problem with the notion of attitudes: Although there may be greater or lesser agreement that attitudes are essentially evaluative in nature (e.g., Zanna & Rempel, 1988), it is difficult to know what sorts of considerations people take into account when they express those evaluations/attitudes. Consider the following scenario as a crude example. Person X may express a positive attitude toward recycling all their household waste; Person Y may express a negative attitude toward the same issue. Person X may say that he or she is in favor of recycling household waste (for environmental reasons) but is unlikely to do so because it is difficult to transport the waste to the required recycling center. Person Y may share the same position regarding the environment but may say he or she has a negative attitude toward recycling household waste because it is difficult to transport the waste to the required recycling center. We are left in the position that, at least in some sense, we might want to say that Person X and Person Y value (or "evaluate") recycling to a similar degree, even if they interpret our question, or choose to answer our question, differently. We invariably both do not know what considerations or criteria people are taking into account when they express their attitudes, nor do we have a way of asking questions in ways that ensure that people respond with something akin

< previous page

page_40

next page >

< previous page

page_41

next page > Page 41

to "all-out judgments" (Davidson, 1980). To complicate matters further, attitudes, like preferences (Shafir, 1993), often are constructed during their elicitation. Conflicting Values Not only are we likely to be unclear about the considerations that people incorporate into their expressions of attitude but also the notion of a plurality of values hints at the potential for conflict between those values. Part of the potential problem here is that people may experience and express conflicting evaluations of an "attitude object" rather than one overall evaluation: People may remain essentially ambivalent towards some of the objects of their attitudes (see, e.g., Sparks, James, Conner, Shepherd, & Povey, 1997; Thompson, Zanna, & Griffin, 1996). Such a position is rather at odds with the central tenets of SEU-based models of attitudes, such as the TPB, which would suggest that people arrive at their attitudes through an analytic, compensatory, weighing-up of positive and negative features of an attitude "object." However, ambivalence is entirely consonant with the position of various authors across various disciplines who have discussed inconsistencies in peoples values, preferences, interests, and wants, and so on (e.g., Harsanyi, 1977; March, 1978; Schelling, 1992). Once we acknowledge the possibility for a degree of conflict or inconsistency within people's evaluations of attitude objects, then we can understand readily that there may be a conflict between the values that are represented by people's self-identities (or their expressions of their self-identity) and the values that are represented within their attitudes (or their expressions of attitude). However, to date, this proposition has not been empirically assessed: Rather, what we have is a collection of empirical studies that point to important independent predictive effects but without a clear theoretical structure in which to locate those effects. To the extent that a particular form of self-identity embodies values discrepant from those that a person may hold and express in their attitudes, then we might expect independent predictive effects on intentions and behavior. As a further speculative departure, one could imagine that long-term interests are sometimes captured within people's self-identities in a way in which they are not captured in people's attitudes toward specific behaviors, insofar as identities incorporate rules for action. Attitudes toward specific behaviors may be more likely to capture relatively transitory, ephemeral, or context-bound motivations relating to shorter-term goals. Other authors have written about people's conflicting preferences and how some preferences may come to dominate under certain contextual conditions (e.g., Ainslie, 1992).

< previous page

page_41

next page >

< previous page

page_42

next page > Page 42

Although not supported by much experimental evidence to date, Ainslie (1992) suggested that people may adopt "precommitting devices" with which they cope with conflicting desires: either to influence their future motives or to place "physical limitations on future behavior" (p. 126). Schelling (1992) discussed similar proposals. One of Ainslie's precommitting devices is "personal rules'' that serves to categorize certain actions as proscribed. These can be interpreted as similar to the kinds of action rules described by Abelson (e.g., 1982). These considerations would suggest that two issues are important in understanding the role of self-identity in attitudebehavior models. First, people may have conflicting values and conflicting ("mixed") attitudes; second, a subset of those values may be expressed in people's self-identities, such that the values expressed in people's attitudes may conflict with those expressed in their self-identities. To reiterate; it may not be clear what values are expressed in people's attitudes. Sociological Ambivalence Not only will there be a potential divergence of interests or values between people's attitudes and their self-identities, but we can also conceive of ambivalent attitudes and ambivalence in relation to different facets of self-identities. That is, there may well be conflicting evaluations within the structure of people's self-identities themselves. In an essay examining how "ambivalence comes to be built into the structure of social statuses and roles" (p. 5), Merton (1976) described how "sociological" ambivalence can result from conflicts both within and between different social positions. Sociological ambivalence is described as "incompatible normative expectations of attitudes, beliefs, and behavior assigned to a status (i.e., a social position) or to a set of statuses in a society" (p. 6). He noted, for example, both the affective detachment and compassion that is required of physicians and the conflicting demands on women in their occupational and family roles. Social roles are also core concerns of "identity theory" (Stryker, 1986), which proposes that commitment to an identity (i.e., role identity) influences the salience of that identity, which in turn influences behavior. Commitment is defined as "the costs to the person in the form of relationships foregone were she/he no longer to have a given identity and play a role based on that identity in a social network" (p. 90), and salience refers to the probability of an identity "coming into play within or across situations" (p. 90). Although it would be interesting to know how something similar to these proposals might be seen to be applicable to forms of identity other than role identity, it would be necessary to provide a clearer account of the psychological processes that form the counterparts of salience and commitment as they are

< previous page

page_42

next page >

< previous page

page_43

next page > Page 43

defined here (cf. Hogg, Terry, & White, 1995; see chapter 5, this volume by Terry, Hogg, & White). It would also be interesting to assess the contextual influences and other cues that differentially affect the salience of certain identities (and the values that they represent) and the accessibility of values represented by people's attitudes. One can therefore envisage potential conflicts between certain forms of social and personal identity, or between people's roles as citizens and their roles as consumers (e.g., Sagoff, 1988), among moral codes, social roles, and personal tastes (March, 1978), and between personal and moral preferences (Harsanyi, 1977), between pleasure and moral values (Etzioni, 1988). In short, there is plenty of scope for people's self-identities to be involved in their conflicts of value or interest, and a variety of theoretical perspectives that bear upon these issues. The Limited Descriptive Value of Seu Within behavioral decision theory and other perspectives that are concerned with the choices that people make and how they make them, there is much criticism of the descriptive validity of SEU approaches. In the first place, the calculations required to fulfill the SEU proposals are thought to be too complex (Schoemaker, 1982); on the other hand, a concern over the fundamental incommensurability of people's values is often voiced. Within image theory (e.g., Beach, 1990; Mitchell & Beach, 1990), it is suggested that important components of decision-making processes are the different "images" that a person may use to evaluate choice options. Images may represent a person's principles, goals, or plans. Decision options may then match or not match these images and be adopted, rejected, or considered further, depending on circumstances. The details of image theory (see Beach, 1990) are not necessary here: What is of interest for this discussion is that image theory, as a descriptive theory of decision making, incorporates both the suggestion that people will evaluate choice options against their principles (e.g., values and ethical positions) as well as against their goals, (which may include "the agenda for what he or she hopes to become" (Mitchell & Beach, 1990, p. 9). It is thus entirely compatible with the idea that a person's self-identity will direct their choices (cf. Markus, 1977, on self-schemata). Functional Perspectives in Attitude Theory Functional perspectives on attitudes also reference self-identity. As Shavitt (1989) noted, perspectives that propose that attitudes serve many different functions have been frequently cited but infrequently examined empirically (see also Herek, 1986; Snyder & Debono, 1987). One of attitude's pur-

< previous page

page_43

next page >

< previous page

page_44

next page > Page 44

ported functionsthe utilitarian functionconcerns developing favorable attitudes toward those attitude objects that promote a person's needs or unfavorable attitudes toward attitude objects that contribute punishments. Katz describes this function as "a recognition [sic] of the fact that people strive to maximize the rewards in their external environment and to minimize the penalties" (Katz, 1960, p. 170). This has, at least superficially, much similarity with the attitude depicted in the TRA and TPB arising out of beliefs usually about consequences or outcomes of some kind, weighted by the evaluation of those potential outcomes. On the other hand, Katz's value-expressive function, which provides the person with "satisfaction from expressing attitudes appropriate to his personal values and to his concept of himself" (p. 170), seems to incorporate an acknowledgment of the role of self-identity. In fact, Shavitt (1990) used the term socialidentity function in much the same way that Katz used the notion of value-expressive function. Shavitt (1989) suggested that different attitude objects elicit different attitude functions: For example, coffee usually serves a utilitarian function, whereas cars may serve a utilitarian function, through providing transport, as well as an identity function through expressing status and identity. Snyder and Debono (1987) made a number of interesting suggestions about the ways attitudes may function for different people: For example, attitudes for people with lower self-monitoring scores may serve a value-expressive function to a greater extent than they will for people with higher self-monitoring scores. Additionally, however, one can also easily focus on the idea that attitudes may serve different functions for the same person and that where a conflict exists between different functions (e.g., between utilitarian and social-identity functions) the expression of attitude reflects, or is dominated by, only one of these functions. Summary and Conclusions In this chapter, the structure of the concept of utility is outlined. The ways it is used within the structure of the TRA and the ways the notion of self-identity may be useful for researchers in their attempts to predict people's intentions and behavior are also outlined. As a part of this discussion, it has been suggested that self-identity may also have utility for people in their everyday behavioras a guide to action, or as a form of commitment in the face of competing interests or wishes that they may have. Subjective expected utility theory has been judged a poor descriptor of the ways people make decisions or construct their attitudes. People are unlikely to arrive consistently at stable overall summary evaluations in the way that this type of model would suggest. Rather, people, may have conflicting values, attitudes, wishes, and preferences. They use heuristic devices,

< previous page

page_44

next page >

< previous page

page_45

next page > Page 45

scripts, personal rules, principles, and so on to the neglect of the decisional balance sheet. Such devices can be construed as normative to the extent that they involve prescriptions about what behavioral choices should be made (Etzioni, 1988). Particular contextual factors may make some values more salient than others. Attitudes may be constructed at their point of elicitation rather than being accessible in memory. People's self-identities (whether personal or social) may embody a particular subset of values or interests, separate from those that people express in their attitudes. Moral values may often be incorporated into expressions of self-identity. Part of a pervasive problem that presents itself in a consideration of the issues under discussion here has to do with how little we know about what is being expressed or taken into account when people offer their attitudes. The integration of self-identity into the framework of the TPB offers the opportunity to examine the social, moral and emotional dimensions of people's attitudes and behavior in greater detail. This recognition of "conflicting" attitudes has long been apparent in psychological theorizing. Allport (1935), for example, discussed how public and private spheres of activity might produce contradictory attitudes, both "sincerely held" (p. 824). Both attitudes and self-identity attract a voluminous research literature, and it would be all too easy to overlook useful developments. The various points of contact between self-identity and attitudes are addressed only fleetingly here. Some areas of research, such as the self-presentational or reputation management aspects of selfidentity, are not addressed at all. Even the issue of the influence of self-identity on attitudes is mentioned only briefly. This is a potentially important role for self-identity within the TPB, possibly in the same way that moral values may impact on attitudes (Raats et al., 1995; Sparks, Shepherd, & Frewer, 1995). Certainly, it has been suggested that behaviors that are important to a particular social identity may lead to stronger attitudes toward those behaviors among people who subscribe to that identity than among people who do not (Boninger, Krosnick, & Berent, 1995). Self-identity may have utility for attitude-behavior researchers if it helps in the prediction and understanding of social behavior. However, clear, interesting, and useful hypotheses about the relation among self-identity, people's attitudes, and their social behavior have been scarce. Having said this, interest in both attitudes and self-identity is so widespread that it is likely that interest in this area will persist for a long time. It is to be hoped that an integration of these perspectives will lead to an enriched account of attitudes, their antecedents, causal networks, and consequences. New, more social directions to attitude research that relate attitudes more explicitly to the social contexts in which they are constructed and expressed, and to

< previous page

page_45

next page >

< previous page

page_46

next page > Page 46

other social variables of one kind or other, would regain some social dimensions for attitude research. They would also offer a more complete and relevant framework within which people's motives, attitudes and social behavior may be investigated.

< previous page

page_46

next page >

< previous page

page_47

next page > Page 47

4 A Theory of Attitudes, Subjective Norms, and Private Versus Collective Self-Concepts David Trafimow New Mexico State University One of the most important tasks for personality and social psychologists is to predict behavior. Consequently, several theories have been proposed for this purpose. One theory that has performed well over the last few decades is Fishbein's theory of reasoned action (TRA; Ajzen & Fishbein, 1980; Fishbein, 1967b; 1980; Fishbein & Ajzen, 1975). According to this theory, the proximal cause of behavior is intention to behave, which, in turn, is caused by attitude and subjective norm. An attitude is the target person's opinion about whether the behavior is positive or negative, and a subjective norm is the target person's opinion about what most others who are important to him or her think he or she should do. Attitudes and subjective norms are determined by beliefs about the consequences of the behavior and beliefs about the opinions of specific important others, respectively. Not surprisingly, recent investigators have added other variables to the model. Some of these are affect (Triandis, 1980), perceived behavioral control (Ajzen, 1988), moral values (Gorsuch & Ortberg, 1983), previous behavior (Bentler & Speckart, 1981; Fredricks & Dossett, 1983), behavioral norms (Grube, Morgan, & McGree, 1986), and habit (Triandis, 1980). Although each of these variables has been shown to increase our understanding of some behaviors, attitudes and subjective norms have remained central components of the theories, and much of my own research has been concerned with them. This chapter summarizes that work by placing it in a

< previous page

page_47

next page >

< previous page

page_48

next page > Page 48

context of some propositions that specify the relations between attitudes, subjective norms, and private and collective self-concepts (see Table 4.1). The propositions are presented in an order that, like the TRA (Fishbein, 1980), traces behavior back to other variables. The first proposition takes into account recent research suggesting that behavioral intentions are not the only proximal determinant of behavior. The second proposition addresses whether attitudes and subjective norms, the presumed determinants of intentional behavior, are really separable from each other. The third proposition suggests that the distinction between attitudes and subjective norms can be applied to people as well as behaviors. Furthermore, five subpropositions suggest variables that influence the extent to which people or behaviors are controlled by attitudes and subjective norms. These include the private self, the collective self, and culture. Finally, the fourth and fifth propositions propose distinctions between different types of beliefs that TABLE 4.1 A Set of Propositions Proposition 1: Behavior can result from unintentional as well as intentional processes. Proposition 2: Attitudes and subjective norms are different constructs. Proposition 3: People differ in the degree to which they are under attitudinal or normative control. Subproposition 3A: The relative accessibility of the private and collective selves is a determinant of whether people or behaviors are under attitudinal or normative control. Subproposition 3B: Culture is a determinant of whether people or behaviors are under attitudinal or normative control. Subproposition 3C: Individual differences in a variety of variables affect the relative importance of attitudes and subjective norms. Subproposition 3D: A variety of sociocontextual variables can affect the relative importance of attitudes and subjective norms. Subproposition 3E: Habit is a moderator of the attitude-intention and subjective norm-intention relationships. Proposition 4: Affective and cognitive influences on behavior are fundamentally different. Proposition 5: Beliefs about behavior-specific referents can be more important determinants of behavior than generalized normative concepts.

< previous page

page_48

next page >

< previous page

page_49

next page > Page 49

are presumed to underlie the formation of attitudes and subjective norms, respectively. The fourth proposition asserts, in contradiction to the TRA, that cognitive beliefs should be distinguished from affective ones; and the fifth proposition suggests that beliefs pertaining to behavior-specific referents should be distinguished from those pertaining to general referents. Proposition 1: Behavior Can Result From Unintentional, As Well As Intentional, Processes Inspired largely by the work of Fishbein (e.g., 1967b, 1980), a great deal of literature indicates that behaviors are largely determined by intentions to perform them (see Sheppard, Hartwick, & Warshaw, 1988, for a review). However, other research suggests that behavior also can result from less intentional processes such as previous behavior (Bentler & Speckart, 1981; Fredricks & Dossett, 1983), habit (Triandis, 1980), behavioral norms (Grube et al., 1986), and perceived behavioral control (Ajzen, 1988), which have all been shown to affect behavior in ways not easily accounted for solely by an intentional process. Furthermore, there is also evidence that behaviors can be directly affected by priming. For example, Bargh, Chen, and Burrows (1996) primed the concept of the elderly, and unobtrusively measured the speed with which the participants walked down a hallway after leaving the laboratory. Consistent with expectations, these participants walked more slowly than did control participants who had not been primed. Proposition 1 implies some interesting possibilities. For example, it might be possible to induce people to perform behaviors that they otherwise would intend not to perform given the ''activation" of their intentions. Alternatively, the priming of intentions could increase the extent to which they affect behavior. Finally, general rules might be developed by which researchers could predict, in an a priori fashion, the circumstances under which specific types of behaviors are under intentional control or not (e.g., Bargh, 1990; Fazio, 1990a). Proposition 2: Attitudes and Subjective Norms Are Different Constructs Given that the vast majority of literature (see Ajzen, 1988; Ajzen & Fishbein, 1980, for reviews) in the area has dealt with intentional rather than unintentional processes, knowing the variables that affect intentions is important. Fishbein (1980) argued that the two most important determinants are attitudes and subjective norms. To reiterate, attitudes are global judgments about the behavior (e.g., the behavior is positive or negative) and subjective norms are judgments about what others who are important to

< previous page

page_49

next page >

< previous page

page_50

next page > Page 50

the person think he or she should do. Interestingly, despite repeated demonstrations that attitudes and subjective norms predict intentions, the fact that they are also correlated with each other has led some researchers (e.g., Miniard & Cohen, 1981) to argue that they are not really different constructs. Indeed, causal modeling of these variables has been shown to result in crossover effects (Grube et al., 1986; Oliver & Bearden, 1985; Shimp & Kavas, 1984; Vallerand, Deshaies, Cuerrier, Pelletier, & Mongeau, 1992). Of course, this could just mean that attitudes (or whatever causes attitudes) and subjective norms (or whatever causes subjective norms) affect each other. Relatedly, whether the arrows connecting attitudes and subjective norms that result from causal modeling really represent causation or something else is a matter of debate. Further, Miniard and Cohen (1981; also see Liska, 1984) pointed out that beliefs about consequences (behavioral beliefs) and beliefs about the opinions of specific others (normative beliefs) that are assumed by Fishbein to determine attitudes and subjective norms, respectively, are not very different. For instance, a behavioral belief that "My mother will disagree with me if I eat unhealthy food" seems similar to the normative belief that "My mother thinks I should not eat unhealthy food." In contrast to the critiques by Miniard and Cohen (1981) and Liska (1984), much recent research strongly supports the distinction between attitudes and subjective norms. For example, Trafimow and Fishbein (1994a) argued that some behaviors tend to be more under attitudinal control (AC) and others tend to be more under normative control (NC). Behaviors can be classified as being under AC or NC by using a standard multiple regression paradigm by which intention is predicted on the basis of attitude and subjective norm. If attitude acquires a large beta weight and subjective norm acquires a small one, then the behavior is under AC; if the reverse is true, then it is under NC. To test this notion experimentally, Trafimow and Fishbein manipulated attitudes toward a behavior that had previously been shown to have a large attitudinal or normative beta weight, and measured intentions to perform that behavior. According to Trafimow and Fishbein's argument, manipulating attitudes should affect intentions to perform behaviors with large attitude beta weights but not behaviors with small attitude beta weights. This hypothesis was confirmed in all three of their experiments. Furthermore, they performed another set of experiments (Trafimow & Fishbein, 1994b) in which subjective norms were manipulated, and analogous results were obtained (normative manipulations affected intentions to perform behaviors with large normative beta weights but not behaviors with small normative beta weights). Thus, the distinction between AC and NC behaviors was well supported, and, by implication, the distinction between attitudes and subjective norms was also supported.

< previous page

page_50

next page >

< previous page

page_51

next page > Page 51

Trafimow and Fishbein (1995) also obtained support for the distinction between attitudes and subjective norms at the belief level. They reasoned that if people compare behavioral beliefs (beliefs about consequences) to each other to form an attitude, and normative beliefs (beliefs about the opinions of specific others) to each other to form a subjective norm, then the way these various beliefs get encoded into memory should reflect this process. In particular, interbelief associative linkages should become forged among behavioral beliefs and among normative beliefs but not between behavioral and normative beliefs. Consequently, if people are asked to retrieve their beliefs, the retrieval of a behavioral belief should elicit the retrieval of another behavioral belief while the retrieval of a normative belief should elicit the retrieval of another normative belief. Thus, people's recall protocols should be clustered by belief type. Finally, because attitudes and subjective norms are presumed to be functional for intention formation, such clustering should be evident for beliefs that had been used to form an intention but not for beliefs that had been used for other (control) purposes. Results from three experiments strongly supported this reasoning. In summary, the distinction between attitudes and subjective norms has received support (Trafimow & Fishbein, 1994a, 1994b, 1995; see Trafimow, 1998, for a review). Further support is described in the following pages. Proposition 3: People Differ in the Degree to Which They Are Under Attitudinal or Normative Control Farley, Lehmann, and Ryan (1981) and Fishbein and Ajzen (1975) found that although most multiple regression studies resulted in larger attitude beta weights than normative ones, the normative ones were often statistically significant. Why should subjective norms contribute slightly, but significantly, to the prediction of intentions for a large set of behaviors? The most obvious reason, particularly in light of the research by Trafimow and Fishbein (1994a, 1994b) just described, is that most behaviors are primarily under AC and slightly under normative control for most people. However, an alternative hypothesis is that most people are under AC across a large number of behaviors (AC people), but a minority of people are under normative control (NC people). If there are some NC people in a particular sample, then it would not be surprising that their presence would be reflected in a significant normative beta weight. Another explanation is that attitude measures are more reliable than subjective norm measures, and therefore they correlate better with intention measures. In contradiction to this possibility, however, Trafimow and Finlay (1996) found that the test-retest reliabilities of the two types of measures were not discernibly different.

< previous page

page_51

next page >

< previous page

page_52

next page > Page 52

Trafimow and Finlay (1996) argued that three criteria should be fulfilled to convincingly demonstrate that there are AC and NC people. First, attitudes or subjective norms should be shown to predict behavioral intentions across a wide range of behaviors in a within-participants design. Second, there must be a subset of people for whom the withinparticipants prediction of intentions is better from subjective norms than from attitudes. Finally, the relation between subjective norms and intentions should be demonstrated to vary predictably with some other individual difference variable. Consistent with this reasoning, Trafimow and Finlay measured attitudes, subjective norms, and intentions to perform 30 behaviors, and conducted traditional between-participants multiple regression analyses on each behavior across all of the participants, as well as nontraditional within-participants analyses on each participant across all of the behaviors. The between-participants analyses replicated previous literature; Twenty-nine of the 30 behaviors were more under AC than NC. The within-participants analyses indicated that for a minority of the participants (21%), the withinparticipants subjective norm-intention correlation was greater than the within-participants attitude-intention correlation. These 21% were designated as NC people (as opposed to the other 79% who were designated as AC people). More interestingly, however, when the NC people were deleted from the sample, and the between-participants analyses were again conducted, the median unique variance in intentions accounted for by subjective norms above and beyond that accounted for by attitudes was .00! Finally, the tendency for people to be under normative control was correlated with an outside variablea collective self-measurewhich is discussed in the following section. This research has been replicated in the domains of health behaviors (Finlay, Trafimow, & Jones, 1997) and sports behaviors (Finlay, Trafimow, & Villarreal, (1999). The description of the above research may have unintentionally implied that it is only people that are under AC or NC and that the behaviors do not matter. In contradiction to this suggestion, however (but consistent with Trafimow & Fishbein, 1994a, 1994b), Trafimow and Finlay performed an additional study (1999) and demonstrated that a substantial number of behaviors were under AC (greater attitude beta weight than subjective norm beta weight) even when only normatively controlled participants were analyzed. Given that there appear to be AC and NC people and behaviors, what are some of the determinants? Subpropositions 3A through 3E deal with this question.

< previous page

page_52

next page >

< previous page

page_53

next page > Page 53

Subproposition 3A: The Relative Accessibility of the Private and Collective Selves Is a Determinant of Whether People or Behaviors are Under AC or NC. Several researchers have suggested that there are different aspects of the self, such as private and collective (Triandis, 1989), personal and social (Hogg & Abrams, 1988; Hogg, Terry, & White, 1995), independent and interdependent (Markus & Kitayama, 1991), and others. The private self includes thoughts about one's own traits or states (e.g., ''I am tall"), whereas the collective self includes thoughts about group membership (e.g., "I am Chinese"). I refer to these different kinds of thoughts as private and collective self-cognitions, respectively. Triandis (1989) reviewed literature indicating that different aspects of the self are emphasized in different cultures. For example, when people from individualistic or collectivist cultures complete statements beginning with "I am," more private selfcognitions tend to be written by individualists than by collectivists, and the reverse is true concerning collective selfcognitions (Triandis, McCusker, & Hui, 1990). Trafimow, Triandis, and Goto (1991) argued that two alternative theories could account for these findings. First, there may be only one location in memory where self-cognitions are stored, but culture affects the relative number of private and collective self-cognitions stored there. Second, people may have a "private self" where private self-cognitions are stored and a ''collective self" where collective self-cognitions are stored, and culture affects the relative accessibility of these locations. Two kinds of findings support the two-location theory over the one-location theory. First, Trafimow et al. (1991) showed that the private self and collective self could be independently primed. Specifically, they found that more private self-cognitions were elicited after the private self was primed, but that more collective self-cognitions were elicited after the collective self was primed. If the one-location theory were true, and private and collective selfcognitions were linked to the same self-concept (i.e., stored in the same location in memory), then the priming manipulation should not have had this effect. Second, they performed a conditional probability analysis on the selfcognitions obtained from the participants. To understand the predictions, consider a basket containing red and blue marbles (analogous to the one-location theory). Suppose a person, without looking, reaches into the basket and selects a red marble. The next marble selected has an approximately equal probability of being either red or blue. In contrast, suppose there is a basket containing red marbles and another basket containing blue marbles. In this case, the selection of a red marble indicates that the person is selecting from the basket of red marbles, and so the next marble selected is likely to be red. So, if we consider self-cognitions, the

< previous page

page_53

next page >

< previous page

page_54

next page > Page 54

two-location theory predicts that the probability of retrieving a private self-cognition is greater if the previously retrieved item is a private than collective self-cognition, but the probability of retrieving a collective self-cognition is greater if the previously retrieved item is a collective than private self-cognition. In contrast, if these self-cognitions are stored in the same location in memory as the one-location theory asserts, then retrieval should be random. In fact, both the priming data and conditional probability data strongly supported the two-location theory and contradicted the onelocation theory (see Tables 4.2 and 4.3). Furthermore, these findings were obtained with both Americans (presumably individualists) and Chinese (presumably collectivists) and replicated by Trafimow, Silverman, Fan, and Law (1997) in an experiment conducted in Hong Kong. Trafimow and Smith (1997) also replicated these findings with a sample of Native Americans (collectivists). Finally, Singelis (1994) and Bontempo (1993), submitted "independent" and "interdependent" items to factor analyses, and, consistent with the two-location theory, they demonstrated that a twofactor solution is superior to a one-factor solution, and that the factors are uncorrelated with each other. Given that people have a private and collective self, it seems reasonable to speculate that there might be a connection between these selves and whether people or behaviors are under AC or NC. Trafimow and Finlay (1996) used a scale developed by Singelis (1994) as a measure of the collective self and found that scores on this scale correlated with the tendency for TABLE 4.2 Proportion of Private and Collective Self-Cognitions Reported as a Function of Whether the Private or Collective Self was Primed (Experiments 1 and 2, Trafimow et al., 1991)a Prime Private Self-Cognitions Collective Self-Cognitions Experiment 1 .77 .12 Private .65 .29 Collective Experiment 2 .80 .09 Private .58 .22 Collective aBecause some items were coded as neither private nor collective selfcognitions, proportions did not sum to 1.

< previous page

page_54

next page >

< previous page

page_55

next page > Page 55

TABLE 4.3 Conditional Probabilities p(P\P) p(P\C) p(C\P) p(C\C) Experiment 1 .74 .47 .16 .42 Experiment 2 .73 .51 .17 .46 Note. p(P\P) = probability of reporting a private self-cognition given that the previously reported item was a private self-cognition; p(P\C) = probability of reporting a private self-cognition given that the previously reported item was a collective self-cognition; p(C\P) = probability of reporting a collective selfcognition given that the previously reported item was a private self-cognition; p(C\C) = probability of reporting a collective self-cognition given that the previously reported item was a collective self-cognition (from Trafimow et al., 1991, Experiments 1 and 2). Because some items were coded as neither private nor collective self-cognitions, p(P\P) + p(C\P) and p(P\C) + p(C\C) do not sum to 1. people to be under NC. They also suggested that a more direct test could be conducted by priming the private and collective selves and measuring attitudes, subjective norms, and intentions to perform a behavior. If the relative accessibility of the private and collective selves is a determinant of the relative importance of attitudes and subjective norms in influencing behavior, then a behavior should be under AC after the private self is primed but under NC after the collective self is primed. Ybarra and Trafimow (1997) performed this test and found that intentions to use condoms were primarily under AC subsequent to a private self-prime but were primarily under NC subsequent to a collective self-prime (i.e., the regression weights associated with attitudes and subjective norms flip-flopped depending on the prime; see Table 4.4). Subproposition 3B: Culture Is a Determinant of Whether People or Behaviors are Under AC or NC. Given that culture affects the relative accessibility of the private and collective selves (Trafimow et al., 1991; Trafimow et al., 1997), and that the relative accessibility of the private and collective selves determines whether people or behaviors are under AC or NC (Subproposition 3A), it follows that culture should be one determinant of whether people or behaviors are under AC or NC. Less clear, however, is whether culture affects AC or NC independently of its effect on the accessibility of the private and collective selves. It is also not clear whether culture affects the number of people who are under AC or NC (Hypothesis A); the number of behaviors that are under AC or NC (Hypothesis B); or both (Hypothesis C).

< previous page

page_55

next page >

< previous page

page_56

next page > Page 56

TABLE 4.4 The Prediction of Intentions to Use a Condom from Attitudes and Subjective Norms as a Function of Priming (Condition (Experiments 1, 2, and 3, Ybarra & Trafimow, 1998). Priming Condition Attitude Beta Weight Normative Beta Weight Experiment 1 .54** .23 Private Self .29 .53*** Collective Self Experiment 2 .65*** .28 Private Self .14 .49** Collective Self Experiment 3 .43* .13 Private Self .17 .66** Collective Self *p < .05. **p < .01. ***p < .001. There are reasons to believe each of these three hypotheses. For example, in at least some collectivist cultures (or in some families in individualistic cultures), where gaining rewards or avoiding punishments is contingent on knowing what is expected and complying with those expectations, children may develop chronically accessible cognitive structures (e.g., the collective self) that function in this regard. These cognitive structures, precisely because they are chronically accessible, are likely to become activated when a variety of behavioral decisions are formed and thus will affect those decisions. To the extent that the chronic accessibility of a self-structure is considered an individual difference variable, culture could be said to affect large numbers of people (Hypothesis A). However, culture may affect behavioral decisions in a completely different way. Even in individualistic cultures, there are some behaviors for which subjective norms are generally important (Stasson & Fishbein, 1990). For example, if a person is deciding whether to use a condom during sex, the sexual partner's opinion is clearly important (particularly if failure to consider

< previous page

page_56

next page >

< previous page

page_57

next page > Page 57

the partner's opinion means not having sex. Fishbein, Middlestadt, & Trafimow, 1993; Fishbein et al., 1993; Fishbein et al., 1995). On the other hand, the opinions of important others are generally considered to be relatively unimportant compared to one's own attitude in making career decisions in individualistic cultures. This differs from collectivist cultures, where the opinions of one's family (particularly the parents) are of paramount importance. More generally, as Triandis (1990, 1994) documented, the opinions of family members seem to be important for more behaviors in collectivist than individualistic cultures (Hypothesis B). Subproposition 3C: Individual Differences in a Variety of Variables Affect the Relative Importance of Attitudes and Subjective Norms. Several individual differences variables have been shown to affect the relative importance people attach to attitudes and subjective norms. For example, Miller and Grush (1986) found that people who were both high in private self-consciousness and low in self-monitoring displayed high attitude-behavior correspondence, but people with other combinations of these traits displayed high subjective-norm behavior correspondence. Arie, Durand, and Bearden (1979) performed a study using U.S. employees at a large southeastern university as participants. They found that opinion leaders' intentions to patronize credit unions were under attitudinal control, but nonleaders' intentions were under normative control. Finally, Bagozzi, Baumgartner, and Yi (1992) found a greater attitude-intention than subjective norm-intention correlation for action-oriented participants, but the reverse was true for state-oriented participants. (According to Bagozzi et al., action-oriented people are less subject to behavioral "inertia" than are stateoriented people. They further suggested that whereas action-oriented people decide what they want to do, and then do it, state-oriented people go along with what other people do. These assumptions are consistent with the described finding.) Trafimow (1994) proposed a hypothesis that also bears on this issue. He argued that the weight people give to subjective norms should depend, at least in part, on how confident they are of the accuracy of their normative perceptions. To test this hypothesis, he measured attitudes, subjective norms, and intentions to use a condom (Experiment 2). Consistent with the hypothesis, participants who were not confident in the accuracy of their normative perceptions failed to exhibit a significant subjective norm-intention correlation, but confident participants exhibited a correlation of .94. In contrast, the attitude-intention correlation was not affected by normative confidence (see Table 4.5). Furthermore, these results were replicated in a second study (Trafimow, 1999b). Interestingly, possibly

< previous page

page_57

next page >

< previous page

page_58

next page > Page 58

TABLE 4.5 The Prediction of Intentions to Use a Condom from Attitudes and Perceptions of Normative Pressure Depending on Level of Confidence (Trafimow, 1994) Level of Confidence R Attitude Normative Attitude Normative Beta Weight Beta Weight Correlation Correlation Low Confidence .67* 2.68** -.26 .61 -.09 Extreme Confidence .97***.12 .88*** .71*** .94** *p < .05. ** p < .01. *** p < .001. because people are generally confident that they know what their attitudes are, an analogous effect of attitudinal confidence on the attitude-intention correlation was not obtained. Subproposition 3D: A Variety of Sociocontextual Variables Can Affect the Relative Importance of Attitudes and Subjective Norms. Distinguishing between individual differences versus sociocontextual variables is often difficult. Certainly, one could reasonably interpret some of the data just presented as sociocontextual in nature and some of the following data just as dealing with individual differences. Consequently, my interpretation of what kinds of data pertain to individual differences or sociocontextual variables is subject to judgment. With this caveat in mind, let us now examine the data. I argued that confidence in one's normative perceptions is an individual difference variable that affects the importance of subjective norms in predicting intentions. However, this can also be thought of as a sociocontextual variable. For example, Trafimow (1994, Experiment 1) found that an experimental manipulation of confidence in normative perceptions affected the weight people gave to subjective norms. Terry and Hogg (1996) also performed research showing that, under some conditions, subjective norms strongly predict intentions. Based on social identity theory (SIT; Hogg & Abrams, 1988; Tajfel & Turner, 1979), these researchers argued that the extent to which people identify with their

< previous page

page_58

next page >

< previous page

page_59

next page > Page 59

ingroups should affect the weight placed on ingroup norms when making a behavioral decision. Two studies confirmed their hypothesis. In Study 1, they found that perceived norms strongly influenced intentions to exercise but only for participants who identified with their ingroup. In Study 2, females' intentions to engage in sun-protective behavior were highly affected by perceived norms but only for those who highly identified themselves with their ingroup. (Friends and peers at a university composed the ingroup in both studies; see chap. 5, this volume by Terry, Hogg, and White for an extension of this line of research.) Fazio (1990a) also argued for the importance of sociocontextual variables. According to his MODE (motivation and opportunity determinants) model, people sometimes perform behaviors on the basis of a spontaneous process and sometimes on the basis of a deliberative one. Attitudes can affect behaviors in both cases, depending on the extent to which other variables (e.g., norms) are seen as important. Because other variables are less likely to be considered when the process is spontaneous than when it is deliberative, the attitude-behavior relation should be stronger (if the attitude is accessible) and the norm-behavior relation should be weaker when the process is spontaneous rather than deliberative. Research by Sanbonmatsu and Fazio (1990) and Schuette and Fazio (1995) supports this conception. For example, Sanbonmatsu and Fazio found that people's decisions about where to buy a camera were highly influenced by general attitudes toward the stores when a spontaneous process was used, but their decisions were more influenced by specific information about the camera departments of the stores when a deliberative process was used. Subproposition 3E: Habit Is a Moderator of the Attitude-Intention and Subjective Norm-Intention Relations. Triandis (1980) argued that being in the habit of performing a behavior can affect actual performance of the behavior in a way not mediated by intentions. Trafimow (1999a) extended this argument by proposing that habit can play a role in intentional processes as well. More specifically, Trafimow argued that when people are in the habit of performing a behavior, they can "intend" to do so in the future without putting much effort into the formation of the intention. Thus, there is no reason to expect attitudes and subjective norms to correlate with such intentions. Only when people are not in the habit of performing a behavior is there reason to devote cognitive resources to the decision. In this case, as Trafimow and Fishbein (1995) suggested, it is worthwhile to use attitudes and subjective norms for making the decision. Trafimow (1999a) tested this notion by performing a study in which people's habitual behavior, attitudes, subjective norms, and intentions pertain-

< previous page

page_59

next page >

< previous page

page_60

next page > Page 60

ing to condom use were measured. The findings were consistent with the above reasoning. When a subsample of people who habitually used condoms was analyzed, attitudes and subjective norms were not good predictors of intentions (r = .18 and r = .10, p = .1, in both cases). In contrast, when people who did not habitually use condoms were analyzed, attitudes and subjective norms were excellent predictors of intentions (r = .81 and r = .61, p < .01, in both cases). Proposition 4: Affective and Cognitive Influences on Behavior Are Fundamentally Different Subpropositions 3A through 3E deal with variables that influence the relative importance of attitudes and subjective norms in determining intentions or behaviors. But what variables affect the formation of attitudes and subjective norms in the first place? One variable that has received much attention, and that is the focus here, is beliefs about the consequences of performing behaviors (Fishbein, 1980). To put it simply, beliefs that performing a behavior will result in positive (and valued) consequences lead to more positive attitudes than do beliefs that performing a behavior will result in negative consequences. Although this suggestion may seem obvious, it has been the source of much controversy. Specifically, several researchers (e.g., Breckler & Wiggins, 1989) have suggested that there is a difference between affect ("feeling") and cognition ("knowing"), and different beliefs may reflect one or the other of these processes. Furthermore, attitudes formed on the basis of affective or cognitive beliefs may differ from each other in a variety of ways (e.g., Millar & Tesser, 1986; Pfister & Böhm, 1992). It is interesting to note, however, that even before this research was conducted, Fishbein (1980) argued against the distinction between affect and cognition whereas Triandis (1980) made it a central part of his theory. Perhaps because affect and cognition seem different, the majority of research efforts in the area have attempted to support rather than disconfirm the distinction. One strategy has been to show that affect and cognition predict global attitudes better than either component alone (Abelson, Kinder, Peters, & Fiske, 1982; Breckler & Wiggins, 1989; Pfister & Böhm, 1992). Another strategy has been to have people respond to "affective" or "cognitive" words, and then use factor-analytic techniques to group the terms into the two categories. Showing that affective words load together and that cognitive words load together is taken as evidence that attitudes do, in fact, have affective and cognitive components (Abelson et al., 1982; Breckler, 1984; Breckler & Wiggins, 1989; Crites, Fabrigar, & Petty, 1994). Sometimes multiple measures of affect and cognition are included (e.g., Breckler, 1984) and a multitrait-multimethod type of paradigm (Campbell & Fiske, 1959) is used to support the distinction. A third strategy has been to manipulate whether peo-

< previous page

page_60

next page >

< previous page

page_61

next page > Page 61

ple focus on affect or cognition (e.g., by presenting affective or cognitive information) when forming their attitudes; people who focus on affect show greater affect-attitude (or affect-behavior) consistency whereas those who focus on cognition show greater cognition-attitude consistency (Millar & Tesser, 1986). Unfortunately, as Eagly, Mladinic, and Otto (1994) pointed out, these studies have several methodological limitations that make it difficult to draw a solid conclusion (these include alternative explanations, psychometric issues, and concerns about whether "real" affect and cognition were tested). Ironically, Crites et al. (1994) subsequently criticized the Eagly et al. research also on the basis of methodological grounds. More generally, for reasons too complicated to address here, Trafimow and Sheeran (1998) argued that all previous research has failed to definitively test the distinction. To test the affect-cognition distinction in a different way, Trafimow and Sheeran made the following proposal. On the basis of differences between affective and cognitive beliefs enumerated by Pfister & Böhm (1992), they suggested that affective beliefs are more similar to other affective beliefs than to cognitive ones and that cognitive beliefs are more similar to one another than to affective ones. Thus, people are more likely to compare affective beliefs to other affective beliefs and to compare cognitive beliefs to other cognitive beliefs than to compare affective to cognitive beliefs. Consequently, interbelief associations should become forged between affective beliefs and other affective beliefs, and between cognitive beliefs and other cognitive beliefs, but not between affective and cognitive beliefs. A further assumption is that it is functional for people to make the distinctionthe process of forming an intention causes these comparisons to take place. These assumptions imply several predictions. For example, if people write down their beliefs about a behavior for which they had previously formed an intention, then the retrieval of a particular belief should trigger the retrieval of beliefs that are associated with it. Because affective beliefs are assumed to be associated with other affective beliefs, and cognitive beliefs are associated with other cognitive beliefs, affective beliefs should elicit the retrieval of other affective beliefs, and cognitive beliefs should elicit the retrieval of other cognitive beliefs; but affective beliefs should not elicit the retrieval of cognitive beliefs, nor should cognitive beliefs trigger the retrieval of affective beliefs. Thus, the lists of retrieved beliefs should be clustered by belief type. Three experiments (Trafimow & Sheeran, 1998) confirmed not only this prediction but also the prediction that such clustering should not be obtained when participants had not previously formed an intention, but had processed beliefs for other reasons. Furthermore, two other experiments, using traditional discriminant validity approaches similar to research described earlier (e.g., Abelson et al., 1982; Crites et al., 1994), also supported the distinction.

< previous page

page_61

next page >

< previous page

page_62

next page > Page 62

The distinction between affect and cognition also has been supported at the physiological level. Johnston and OliverRodriguez (1997) showed that the late positive component of event-related potentials (which they took to be an indicator of affect) was correlated with males' attractiveness ratings of females' faces but not with their attractiveness ratings of other males' faces. If, as these researchers assume, attractive female faces stimulate an affective reaction in males but attractive male faces do not, then these results are consistent with the affect-cognition distinction. Proposition 5: Beliefs About Behavior-Specific Referents Can Be More Important Determinants of Behavior Than Generalized Normative Concepts Proposition 5 is based on a distinction made by Chung (1985) between general and behavior-specific referents. General referents (e.g., mother, father, spouse) are people who are considered to be important irrespective of the behavior under consideration, whereas behavior-specific referents are people who are considered to be important only with respect to specific behaviors and behavioral domains. For example, police may become a salient referent if one is considering speeding, the Pope may become salient if one is Catholic and considering having an abortion, and so on. Trafimow and Fishbein (1994b) tested Chung's distinction in a set of three experiments. In the first experiment, they manipulated a belief about whether a date believed that the participant should or should not perform a particular behavior (going to a Korean restaurant) with the date or alone. Not surprisingly, participants exhibited greater intentions to perform the behavior with the date when the date was thought to feel positively rather than negatively toward the behavior. Furthermore, this effect was significantly attenuated in the "alone" condition (evidently, the date's opinion was not considered as important in a situation where the date was not directly involved). More interestingly, in Experiment 2, when the behavior-specific referent (i.e., the date) was changed to a general normative concept (i.e., "most others who are important to me"), the manipulation had the same effect regardless of whether the behavior was to be performed with the date or alone. Trafimow and Fishbein suggested two possible explanations for their findings. First, manipulating the general normative concept could have made normative considerations so salient that whether the behavior was performed with a date or alone was unimportant. Second, they suggested that when the general normative concept was used in Experiment 2, thoughts about the behavior-specific referent were not accessed. Because the general normative concept was equally important regardless of whether or not the behavior was performed with a date or alone, the effect of the manipulation was approximately the same in both conditions. It follows from

< previous page

page_62

next page >

< previous page

page_63

next page > Page 63

this reasoning that priming the behavior-specific referent when the general normative concept is manipulated should result in findings similar to those obtained in Experiment 1. Consistent with this latter explanation, manipulating the general normative concept in conjunction with a prime of the behavior-specific referent resulted in a much greater effect on intentions to perform the behavior with a date than alone (Trafimow & Fishbein, 1994b, Experiment 3). One implication of Proposition 5 concerns the extent to which behaviors are under AC or NC. Research cited earlier (e.g., Farley et al., 1981; Trafimow & Finlay, 1996; Trafimow & Fishbein, 1995) indicates that the vast majority of behaviors are more under AC than NC. One possible explanation is that behavior-specific referents are often not considered in these studies. Had they been considered, or had participants been reminded to consider behavior-specific referents when responding to items measuring subjective norms, then it seems reasonable to suppose that the findings might have been more supportive of the importance of subjective norms. Of course, there are other explanations for these findings. As we suggested earlier, these findings could be due to using primarily AC people as participants (Trafimow & Finlay, 1996), performing experiments in individualistic cultures (Triandis, 1980, 1990, 1994), and not taking group identity issues into account (Terry & Hogg, 1996). Where the Theory Is Incomplete Because this chapter is mainly concerned with attitudes, subjective norms, and the self (with a strong cultural ''flavor''), various other important variables have not been considered. For example, two currently "fashionable" variables that are worthy of consideration are perceived behavioral control (Ajzen, 1988), which Ajzen (1991) has recently conceptualized as the perceived ease or difficulty of performing the behavior, and perceived self-efficacy, which Bandura (1996) conceptualized as "a judgment of one's capability to accomplish a certain level of performance" (p. 391). These concepts have sometimes been confounded with each other. For example, perceived behavioral control has sometimes been measured by semantic differential scales such as easy-difficult and under my control-not under my control (Ajzen, 1988). The latter scale is arguably a measure of self-efficacy rather than perceived behavioral control as Ajzen (1991) defined it. In fact, Trafimow and Trafimow (1998) found that although intentions of patients experiencing back pain to perform recommended exercises were predicted by a combination of these two scales (r = .68), the scales were uncorrelated with one another, suggesting that they may measure different concepts (see Terry & O'Leary, 1995, for a similar argument). Clearly, it would be desir-

< previous page

page_63

next page >

< previous page

page_64

next page > Page 64

able to have a better understanding of the differences between various control variables, and how they affect behavior. I mentioned earlier that a wide variety of variables, in addition to attitude and subjective norm, have been shown to explain unique variance in intention and behavior. Clearly, one approach would be to expand the theory to account for all of these variables. Alternatively, one could argue that it is not elegant to keep expanding the theory (with more propositions) to include all of the variables that account for unique variance in at least some intentions and behaviors. Instead, perhaps it is desirable to obtain a better understanding of the variables that are the basis for this theory. For example, given the important effects of the self and culture on behavior (Subpropositions 3a and 3b), it seems worthwhile to investigate how the private and collective selves develop (and when), how culture affects their relative accessibility, why and how they affect the way people use attitudinal and normative information, and whether selfconcepts or culture affect behavior independently of attitudes, subjective norms, or other mediating variables. In addition to more propositions about "self" variables, there is reason to suspect that future theories will include propositions about how attitudes and subjective norms affect each other (e.g., see Terry & Hogg, 1996, for a recent demonstration of the effects of subjective norms on attitudes; also see Grube et al., 1986; Oliver & Bearden, 1985; Shimp & Kavas, 1984; Vallerand et al., 1992, for crossover effects). Although the processes by which attitudes and subjective norms affect one another have not yet been explicated, as Terry and Hogg suggested, social-identity-selfcategorization considerations are likely to play a roleattitudes are likely to be developed that are consistent rather than inconsistent with ingroup norms. Furthermore, consistent with recent findings obtained by Hogg and Hains (1996), attitudes may be affected by depersonal attraction for ingroup members based on group membership (as opposed to personal liking based on idiosyncratic characteristics of the individual). Clearly, more research is needed to test these speculations. There are also gaps in the theory regarding the beliefs that are presumed to underlie the formation of attitudes and subjective norms. It would be nice to know (a) the extent to which beliefs can affect behavior directly, not mediated by attitudes or subjective norms; (b) the role variables not considered here (e.g., classical conditioning) play in determining attitudes and subjective norms; (c) the extent to which crossover effects of subjective norms on attitudes are due to the beliefs underlying subjective norms, rather than subjective norms per se (and vice versa for beliefs underlying attitudes); (d) the extent to which causality goes in the reverse direction from that which is examined in the theory (e.g., how do attitudes, subjective norms, intentions, and behavior affect beliefs?); and (e) whether distinctions between belief

< previous page

page_64

next page >

< previous page

page_65

next page > Page 65

types can be made other than those discussed in Propositions 4 and 5. Preliminary evidence for this last issue was provided by a recent set of studies by Trafimow and Duran (1998) suggesting a distinction between behavioral and control beliefs and by Duran and Trafimow (in press) suggesting that beliefs about positive and negative consequences are stored separately in memory. Although Proposition 1 makes clear that nonintentional processes can affect behavior, most of the theory was clearly concerned with intentional processes. Like Fishbein's (1980) TRA, attitudes or subjective norms were assumed to play a large role in the formation of behavioral intentions (e.g., Proposition 2). Unlike the TRA, however, the present theory explicitly assumes that people, as well as behaviors, can be under attitudinal or normative control (Proposition 3). Furthermore, this theory enumerates several categories of variables that determine whether particular behaviors or people are likely to be under attitudinal or normative control (i.e., Subpropositions 3A-3E). Finally, various distinctions are made among different types of beliefs that are presumed to have causal effects on attitudes, subjective norms, intentions, and behaviors (e.g., Propositions 4 and 5). There is at least one facet of beliefs, however, that has not been addressednamely, the factors that determine them. A possible explanation, alluded to earlier, is that there is reverse causation; attitudes, subjective norms, intentions, or behaviors affect beliefs. Certainly, there must be some truth to this, but it is important to be careful about engaging in circular reasoning. Specifically, for example, it seems inadequate to assume that beliefs determine attitudes and that attitudes determine beliefs. Consequently, some other variables are needed to gain a more complete picture. One place to look for these variables is in Subpropositions 3A through 3E; the relative accessibility of the private and collective selves, culture, and various other individual difference or sociocontextual variables may determine the beliefs underlying attitudes and subjective norms, as well as their relative importance. Future research efforts might profitably be directed toward investigating this issue. Acknowledgments I thank Krystina Finlay, Larry Gregory, Michael Hogg, Victor Johnston, Laura Madson, Paschal Sheeran, Walter Stephan, Deborah Terry, and Sabine Trafimow for their insightful comments.

< previous page

page_65

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_67

next page > Page 67

5 Attitude-Behavior Relations: Social Identity and Group Membership Deborah J. Terry Michael A. Hogg Katherine M. White University of Queensland Despite popular opinion to the contrary, early scientific evidence consistently pointed to a lack of support for the view that people's actions are guided by their attitudes (Wicker, 1969). As a consequence, it has been argued that there is no simple attitude-behavior relation, and that to predict behavior accurately it is necessary to take into account other variables. The theory of reasoned action (TRA; Fishbein & Ajzen, 1975), and its recent extension, the theory of planned behavior (TPB; Ajzen, 1987, 1991), have been influential in this respect. A central component of these theories is the assumption that, in addition to attitudes, social norms have an important influence on behavioral outcomes. However, empirical research has failed to find strong support for the role of norms in predicting behaviora pattern of results that has led some commentators to conclude that social factors are not important in this context (e.g., Ajzen, 1991). Nevertheless, on the basis of social identity theory (SIT; Hogg & Abrams, 1988; Tajfel & Turner, 1979; Turner, 1982) and self-categorization theory (SCT; Turner, 1985; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), a strong theoretical case can be made for the view that to understand attitude-behavior consistency it is necessary to take into account the wider social context of group membershipsattitudes themselves can be regarded as social products to the extent that they are likely to be influenced by social norms and

< previous page

page_67

next page >

< previous page

page_68

next page > Page 68

expectations. Furthermore, norms of behaviorally relevant social groups and categories are likely to influence people's willingness to engage in attitudinally consistent behaviorto do so may help to validate an important social identification. In this chapter, we begin by reviewing previous perspectives and research on the role of social norms in attitudebehavior relations. We then discuss limitations of this previous research and describe how the impact of social norms on attitude-behavior relations can be reconceptualized on the basis of SIT and self-categorization theory. We conclude the chapter by discussing, in some depth, the results of a number of recent studiesconducted in both field and laboratory settingsthat support this perspective; that is, that attitude-behavior consistency is strengthened if a person's attitudes are supported by a congruent ingroup norm. Research testing a number of corollaries to this basic prediction is also reviewed. This research has provided some evidence that (a) the impact of group norms on attitude-behavior consistency depends on the extent to which the group membership is a salient basis for self-conception, (b) the effects of attitude accessibility and ingroup norms on attitude-behavior consistency are independent, and (c) ingroup norms influence attitude-behavior relations under spontaneous and deliberative decision making (cf. Fazio, 1990a). Background Theories of Reasoned Action and Planned Behavior To date, most research into the role of norms in attitude-behavior relations has been conducted from the perspective of the TRA (Fishbein & Ajzen, 1975) or the TPB (Ajzen, 1991; see Terry, Gallois, & McCamish, 1993, for an overview of these theories). According to both theories, behavioral intentions are the most proximal determinant of behavior. Intentions, in turn, are proposed to be functions of people's attitude toward the behavior and the subjective norm, which refers to people's perception of social pressure from significant others to perform the behavior. Attitudes and norms are proposed to have additive effects on intentions, with the relative predictive strength of each varying across behaviors and populations. The TRA and TPB propose further that attitudes are a function of people's beliefs about the consequences of the behavior, weighted by the importance or value placed on these consequences, whereas the subjective norm is proposed to be determined by the perceived expectations of specific individuals and groupsidentified in pilot work as being relevant to the behavior and population under considerationweighted by people's motivation to comply with the expectations of these referents. In the TPB, Ajzen (1987, 1991) proposed that perceived behavioral control would emerge as an additional

< previous page

page_68

next page >

< previous page

page_69

next page > Page 69

predictor of intentions and actual behaviorthe latter effect being evident only in relation to behaviors that cannot be performed at will. By acknowledging the role that social influence is likely to play in predicting behavior, the theories of reasoned action and planned behavior are noteworthymost attitude-behavior models have either neglected the possible effects of norms (e.g., the health belief model; Rosenstock, 1974) or regard the influence of such variables as secondary to the role of attitudes (Fazio, 1986). However, tests of the TRA and TPB have found only weak support for the role of norms. In a meta-analysis of 37 tests (26 studies) of the TRA, Farley, Lehmann, and Ryan (1981) found that the average regression weight for attitude (in the prediction of intention) dominated the average regression weight for subjective norm by a factor of 1.5. This pattern of results was consistent across type of sample (student or nonstudent), design of study (experimental or correlational), and discipline (social psychology or marketing). In a similar vein, Ajzen (1991) reported that the norm-intention link was nonsignificant in more than half of the 19 extant tests of the TPB. On the basis of this evidence, Ajzen (1991) concluded that there is a consistent lack of evidence linking norms to intentions, arguing that in light of evidence linking attitude and perceived behavioral control to intentions, personal factors are more influential in the prediction of behavioral outcomes than social factors. From our perspective, this conclusion can be questioned given that there are limitations in Fishbein and Ajzen's (1975) treatment of the role of social influence in attitude-behavior relations. Rather than playing down the role of norms and group influence, we suggest that a reconceptualization of the role of norms in attitude-behavior relations in line with recent social psychological models of group influence, specifically SIT (Hogg & Abrams, 1988; Tajfel & Turner, 1979; Turner, 1982) and self-categorization theory (Turner, 1985; Turner et al., 1987), may help to preserve a central and important role for norms in this context. Prior to outlining the proposed reconceptualization of the role of norms in attitude-behavior relations, we discuss the limitations in Fishbein and Ajzen's (1975) treatment of the role of social influence in attitude-behavior relations. Limitations of the Normative Component of TRA and TPB The normative component of TRA and TPB may be problematic because Fishbein and Ajzen (1975) did not conceptualize norms in line with the wider social psychological literature on social influencethat is, as the accepted or implied rules that specify how group members should behave (Brown, 1988). Instead, they conceptualized the subjective norm as the extent of pressure perceived from others to perform the target behavior. However, such pressures may be latent and may not be perceived for behaviors

< previous page

page_69

next page >

< previous page

page_70

next page > Page 70

that do not influence other people, or for behaviors that are not directly associated with harmful outcomes. Thus, norms may have stronger effects if they are defined more in line with wider social psychological definitions of norms. This broader notion of norms has been assessed with measures of the perceived behavior (Grube, Morgan, & McGree, 1986) and attitudes of others (White, Terry, & Hogg, 1994). Measured in this way, norms have been found to predict intentions, but the effects are typically still not strong (see e.g., White et al., 1994). The fact that norms, even when assessed to reflect the manner in which they are defined in the wider social psychological literature, do not have strong effects on behavioral intentions may reflect the fact that there are more fundamental problems with the way in which norms are conceptualized in the TRA and TPB. First, although Fishbein and Ajzen (1975) acknowledged that the attitudinal and normative components of the theories are likely to be correlateda pattern of results that has been commonly reported in the literature (e.g., Shepherd & O'Keefe, 1984)they assumed that the variables are cognitively independent, to the extent that they are based on different belief structures (Liska, 1984; Miniard & Cohen, 1981; Ryan, 1978). People's attitudes are seen to be functions of their beliefs about the likely costs and benefits of performing a behavior, whereas subjective norms are based on people's beliefs concerning the extent to which particular others want them to perform the behavior. However, evidence that the belief-based predictors of attitudes can affect subjective norms or vice versa suggests that the variables are not psychologically distinct. Recent research using structural equation modeling has indicated that such crossover effects are most likely to occur from normative beliefs to attitudes (Oliver & Bearden, 1985; Vallerand, Deshaies, Cuerrier, Pelletier, & Mongeau, 1992). In other words, people's attitudes are influenced not only by the perceived costs and benefits of performing the behavior, but also by their perceptions that significant others would want them to perform the behavior. This finding may reflect the fact that the consequences of performing a behavior often include pleasing others, so "pleasing others" is a benefit of performing the behavior, thus meaning that it is a factor that influences people's attitudes (Fishbein & Ajzen, 1975; Liska, 1984). Although this is probably true, the normative belief-attitude crossover effect could reflect a more fundamental effect, namely, that people's attitudes are influenced by significant others (Hovland, Janis, & Kelley, 1953; Vallerand et al., 1992). The assumption made by Fishbein and Ajzen (1975) that there is effect independencethat is, that the effects of attitudes and norms are additive or independent of one anotheris related to the issue of the psychological independence of the two constructs. People's intentions to perform a behavior are proposed to be influenced by both their attitude toward the behavior

< previous page

page_70

next page >

< previous page

page_71

next page > Page 71

and by the extent to which they perceive that significant others want them to perform the behavior. However, the effects of attitudes are seen to be independent of the effects of social influence. In other words, the strength of attitudinal influence is not considered to be influenced by, or dependent on, social influence. Theoretically, a more interesting proposal is the contingent consistency hypothesis (see Acock & DeFleur, 1972; Liska, 1984), which proposes that people are more likely to engage in behavior that is consistent with their attitudes if the normative climate supports the attitude. Several studies have provided some support for this hypothesis (e.g., Grube et al., 1986; Grube & Morgan, 1990); however, other studies have failed to find that norms moderate the effects of attitudes on intentions (e.g., Bagozzi & Schnedlitz, 1985), or they have found that the presence of contingent consistency effects varies as a function of some third variable such as extent of experience with the target behavior (Andrews & Kandel, 1979). Nevertheless, evidence of contingent consistency effects suggests further that attitudes and norms may be less independent than proposed by the theories of TRA and TPBthe perceived views of significant others may influence people's willingness to express their attitudes behaviorally. The problems with the normative component of the TRA and TPB may be attributable to the way in which social influence is viewed by these models and by the wider field of attitude-behavior relations. This literature has largely ignored recent developments in basic social psychological understanding of social influence. Thus, it is not surprising that it is difficult to predict when, and in what way, norms should influence intentions or behavior. Fishbein and Ajzen's (1975) conceptualization of subjective norms implies a social influence process whereby people bring their behavior into line with the behavioral expectations of important others. Thus, public compliance based on a need for social approval and acceptance is probably the underlying social influence process (Miniard & Cohen, 1981; Ryan, 1982), in other words, subjective norms influence behavior through normative influence (Deutsch & Gerard, 1955; Kelley, 1952). Traditionally, social psychology has distinguished this type of social influence from informational influence, which occurs when people privately accept and internalize information from others because the information provides a basis for correct perceptions, attitudes, and beliefs. Recently it has been argued (e.g., Abrams & Hogg, 1990; Hogg & Turner, 1987b; Turner, 1991) that such a dual-process model of social influence is problematic, in that it may conceptually compartmentalize facets of social norms that need to be conceptualized in terms of a single process. We would argue that the notion of norms in the TRA and TPB should be reconceptualized to go beyond the focus on normative influencethat is, away from the view of norms as external prescriptions that are linked to behavior to the extent that one believes that one's behavior is

< previous page

page_71

next page >

< previous page

page_72

next page > Page 72

observable by valued others. One way in which this might be done is in terms of SIT and self-categorization theory, which propose a single process model of social influence. This perspective has the potential to predict when norms should influence behavior and also to articulate the more complex role that norms may play in attitude-behavior relations. SIT and SCT and Attitude-Behavior Relations Social identity theory (e.g., Hogg & Abrams, 1988; Tajfel & Turner, 1979; Turner, 1982) is a general theory of group processes and intergroup relations, that distinguishes group phenomena from interpersonal phenomena. When people define and evaluate themselves in terms of a self-inclusive social category (e.g., a gender, class, or team) two processes come into play: (a) categorization, which perceptually accentuates differences between ingroup and outgroup and similarities among ingroupers (including self) on stereotypical dimensions; and (b) self-enhancement, which, because the self-concept is defined in terms of group membership, seeks behaviorally and perceptually to favor things ingroup over things outgroup (cf. ethnocentrism). This theory was recently extended by Turner and his colleagues (Turner et al., 1987) to focus more specifically on the role of the categorization processself-categorization theory. A single-process model of social influence in groupsreferent informational influencehas been derived from these two theories (Abrams & Hogg, 1990; Turner, 1991). This process is seen as being underpinned by the cognitive process of self-categorization. When social identity is salient, people construct a context-specific group norm from available, and usually shared, social comparative information. This norm is subjectively represented as a group prototype that describes and prescribes beliefs, attitudes, feelings, and behaviors that optimally minimize ingroup differences and maximize intergroup differences (the principle of metacontrast). The process of self-categorization also means that there is an assimilation of self to the prototype and thus the self is transformed: Self-perception, beliefs, attitudes, feelings, and behaviors are now defined in terms of the group prototype. In this way, group membership causes people to think, feel, behave, and define themselves in terms of group norms rather than unique properties of the self. People are influenced by norms not for social approval, nor because others have told one to, nor because others are watching but because norms prescribe the context-specific attitudes and behaviors appropriate for group members. Norms can be enacted in private; a particular group membership just needs to be the contextual basis for self-definition. Thus, norms are inextricable properties of groups that influence people through self-categorization.

< previous page

page_72

next page >

< previous page

page_73

next page > Page 73

On the basis of a social identity and self-categorization perspective, three interrelated factors might explain the lack of stronger evidence for the proposed norm-intention link in the TRA and TPB. First, the subjective norm is conceptualized as the extent of perceived social pressure from all important others to perform the behavior, a variable that is assessed by asking participants to indicate how much they think important others would want them to perform the target behavior (see Ajzen & Fishbein, 1980). Thus, social pressure is conceptualized as additive across all referents and reference groups that the participant defines as important to them. From a social identity and self-categorization perspective, however, norms are linked to specific groups, and have their effect because that specific group is relevant to the behavioral context. Thus, researchers should focus on the norms of the reference group that is most relevant in the context under consideration. Second, individual differences in the subjective salience of a specific group membership are not specifically taken into account in the TRA and TPB. Although people's motivations to comply are proposed to influence the subjective norm, this variable is not seen to have a proximal impact on intentions. Moreover, it can only be assumed that greater motivation to comply with the expectations of a reference group is associated with strong group identification. The social identity and self-categorization approach explicitly proposes that behavioral outcomes are influenced by reference group norms but only for those people for whom group membership is a salient basis for self-definition. Third, tests of the TRA and TPB tend to focus on the role of salient others rather than on salient reference groups. As such, the theory essentially focuses on how one individual influences another (interpersonal influence) rather than on how the group influences the individual through internalized social norms (group influence). It is possible that stronger and more predictable effects of norms are yielded under conditions that favor group influence. In addition to these factors that may be responsible for the evidence of weak effects of norms in tests of the TRA and TPB, the evidence that attitudes and norms may not be psychologically independent is not inconsistent with the social identity and self-categorization perspective. In fact, from this perspective, there is no reason to believe that the constructs should be independent (see also Hovland et al., 1953). Attitudes can be personal, idiosyncratic, and unrelated to norms, but they also can be widely shared and normative. These shared normative attitudes are tied to specific group memberships and are part of group prototypes. Having categorized and defined themselves as members of a distinct social category, people form the context-specific stereotypic norms of the group, and these norms are internalized as their attitudes toward an issue. Thus, a social identity and self-categorization perspective would predict that psychological group

< previous page

page_73

next page >

< previous page

page_74

next page > Page 74

membership should influence attitudes. Furthermore, this perspective would be in line with the contingent consistency hypothesis, which, as described earlier, predicts that people are more likely to behave in accordance with their attitudes if the normative climate supports the attitudeexposure to an attitudinally congruent group norm should strengthen attitude-behavior consistency because it validates the attitudinally congruent behavior as appropriate for group members. We would go further and suggest that not only must the normative climate support the attitude but the group membership that defines the norm should be a salient basis for self-conception. Early research provided some evidence for the relevance of group norms to the attitude-behavior relation. Frideres, Warner, and Albrecht (1971) examined the role of three social constraint variables on attitude-behavior consistency. In addition to manipulating disclosure and social participation (both concerning the visibility of the behavior to others), Frideres et al. included a measure of attitudinal congruence, which was defined as the extent to which the attitude of the participant was similar to those of other participants. The results of the study showed that attitude-behavior consistency greatly increased under conditions where others present took an attitudinal position that was congruent with the attitude of the participant, whereas the consistency decreased when others took an incongruent position. In a similar vein, Schofield (1975) found that in a study examining the consistency between young women's attitudes and their behavior toward sex education, attitude-behavior consistency increased when participants were given information that the peer group was performing the behavior (as well as when disclosure of the act was public). In research investigating high school students' marijuana use, Skinner and Cattarello (1989) found that group norms became more important predictors of behavior than did attitudes, as behavioral commitment (from initiation to continuation of a behavior) increased. Consistent with the premises of SIT and SCT, Skinner and Catarello argued that for adolescent drug-using behaviors, an important influence on behavior is whether adolescents are closely associated with a peer culture that reinforces the behavior and defines it as normative, an influence that strengthens as the behavior becomes more habitual (and probably as contact with the reference group increases). Empirical Research on Group Norms and Attitude-Behavior Consistency These studies support the importance of incorporating reference group norms into the attitude-behavior domain. Although they support our

< previous page

page_74

next page >

< previous page

page_75

next page > Page 75

reconceptualization of the role of norms in attitude-behavior relations, a stronger test of the social identity and selfcategorization perspective would be to link the normative information more clearly to a relevant ingroup and to take into account the salience of the group membership (either in terms of perceived group identification or as a manipulated variable)the effects of norm congruency on attitude-behavior consistency should be strongest for people for whom group membership is a salient basis for self-conception. In the remainder of this chapter, we describe the results of our recent research that focuses on the role that salient ingroup norms play in attitude-behavior consistency. First, we overview the results of two studies examining the effects of group norms and strength of group identification on behavioral intentions and self-reported behavior. Second, we describe the results of a number of experiments designed to examinein a more controlled settingthe effects of group norms, salience of the group membership, and attitude accessibility on attitude-behavior consistency. Third, we consider the effects of group norms and group salience as a function of the mode of behavioral decision making. We conclude the chapter with a discussion of possible directions for future research in the area. Group Norms, Identification, and the TPB In two tests of the TPB, Terry and Hogg (1996) reported some support for the proposed reconceptualization of norms in attitude-behavior relations. In a study of exercise behavior, the perceived norms of a behaviorally relevant reference group (friends and peers at university) influenced behavioral intentions but only for subjects who identified strongly with the group. Also consistent with SIT and SCT was the finding that the relation between perceived behavioral control and intention to engage in regular exercise was stronger for participants who did not identify strongly with the reference group than for the high identifiers. This pattern of results is not inconsistent with a social identity and selfcategorization perspective, which would predict that when one's identity as a unique person is salient, personal beliefs and feelings (such as whether the behavior can be performed at ease, viz., perceived behavioral control) are likely to form the most cognitively accessible basis for behavioral choice (see Fazio, 1990a). In a second study of females' reported sun-protective behavior, there was, once again, evidence that the effects of group norms were evident only for participants who identified strongly with the behaviorally relevant group. Furthermore, the effect of attitude, rather than perceived behavioral control, was stronger for the low identifiers than for the high identifiers. This result is likely to be specific to appearance-related behaviors (such as sun-protective behaviors for females), which have few constraints that may interfere with their performance but are vulnerable to influence from per-

< previous page

page_75

next page >

< previous page

page_76

next page > Page 76

sonal preferences and opinions (i.e., attitudes). Additional results of the second study revealed that the perceived group norm was a stronger predictor than attitudes of the intentions of high identifiers, whereas the reverse pattern of results was observed for the low identifiers, a pattern of results that is consistent with SCT. When social identity is salient, depersonalization occurs, such that a person's feelings and actions should be guided more by group prototypes and norms than by personal factors. An opposite pattern is expected when personal identity is salientshifts toward this end of continuum mean that personal factors should have the strongest impact on feelings and behaviors. In a subsequent study of intention to engage in household recycling and reported recycling behavior (Terry, Hogg, & White, in press), the results of the previous research were essentially replicated. The perceived norm of a behaviorally relevant reference group (friends and peers in the community) was related to behavioral intention but only for participants who identified strongly with the group, whereas the relation between perceived behavioral control (a personal factor) and intention was strongest for low identifiers. Although providing some support for the proposed reconceptualization of norms in attitude-behavior relations, the results of the studies reported by Terry and Hogg (1996) and Terry et al. (in press) failed to support the prediction derived from the contingent consistency hypothesis, namely that for high identifiers, attitudes would be more likely to influence intentions when supported by a congruent ingroup norm than by an incongruent norm. The notion that as a consequence of group identification, behavior will be defined in terms of the group prototype is central to the selfcategorization perspective on social influence. Thus it follows that attitude-behavior consistency should be most marked for high identifiers whose attitudes are congruent with the group norm. The research conducted by Terry and Hogg and Terry et al. may have failed to provide support for this hypothesis for two reasons. First, the group membership considered in the researchfriends and peerswas not as strongly linked to any of the behaviors as it might have been. Second, the research did not focus on a clearly defined group categorization; thus participants may have had difficulty responding to items concerning perceived group norms. Moreover, the likely complexities of the relations among perceived ingroup norms, levels of group identification, and attitudes may be difficult to detect in field studies. Group Norms, Group Salience, and Attitude Accessibility To complement the field research just reported, we conducted a number of experiments designed to examine the effects of ingroup norms and salience

< previous page

page_76

next page >

< previous page

page_77

next page > Page 77

of group membership on attitude-behavior relations. The central prediction tested in the research was that the attitudinal congruency of the ingroup norm would influence attitude-behavior consistency, such that consistency would be enhanced when the relevant ingroup norm was attitudinally congruent but only for people for whom the group membership was a salient basis for self-conception. If the group membership is not a salient basis for self-conception, then group norms should have no impact on behavioral outcomes. On the basis that high levels of group identification indicate that the group membership is likely to be a salient basis for self-conception (in an enduring sense at least), this prediction can be examined using measures of identification. However, a stronger test of the moderating influence of the contextual salience of group membership is to manipulate group salience experimentally, such that the group is a latent category for some participantshence, in that context, is not likely to be a salient basis for self-conceptionand an important and salient contextual basis for self-conception for others. In the first set of experimental studies examining the effects of group norms on attitude-behavior consistency, we also examined the role of attitude accessibility. According to Fazio and colleagues (e.g., Fazio, 1986, 1989; Fazio, Powell, & Herr, 1983), the accessibility of attitudes determines whether a person's attitude influences the way in which he or she behaves. Accessibility of an attitude is determined by the salience of the attitude in memory. Specifically, if a given attitude is called to mind (or memory) in a specific situation, then it is proposed that through the process of selective perception, the perception of the attitude object is influenced by the activated attitude. This process of selective perception should, in turn, influence the cognitive definition of the event and, as a result, mean that the behavioral response is attitudinally congruent. Fazio (1986) argued that some attitudes are more chronically accessible than others, for instance, when people are confident in their attitude or they have a vested interest in the target issue. He has also demonstrated that attitude accessibility can be experimentally manipulated by asking participants to repeatedly express the relevant attitude (e.g., Downing, Judd, & Brauer, 1992; Powell & Fazio, 1984). Response latencies provide a measure of the accessibility of attitudesthe faster people can give their response to an attitude question, the more accessible the attitude is (e.g., Downing et al., 1992; Fazio, 1990b). Regardless of whether an attitude is chronically more accessible or whether it has been manipulated experimentally to be more accessible, Fazio and colleagues have demonstrated that the more accessible the attitude is, the greater the likelihood that it will influence behavior (e.g., Fazio, Powell, & Williams, 1989; Fazio & Williams, 1986; Houston & Fazio, 1989; see also Bassili, 1995).

< previous page

page_77

next page >

< previous page

page_78

next page > Page 78

By making group norms salientparticularly norms that are prototypical of the groupgroup-relevant attitudes may become cognitively accessible. In other words, by informing individuals of the group norms, relevant attitudes may become accessible and, in this way, may influence attitude-behavior consistency. A social identity and selfcategorization approach assumes that the impact of activating a group prototype goes beyond making an attitude accessibleit makes an attitude normative, providing the prescriptive power for a cognition to be reflected in action. Thus, activating a group prototype should influence behavior beyond simply making an attitude accessible for two reasons. First, there is a strong motivational basis to behave consistently with a group prototype as a consequence of people's need for both positive self-evaluation (self-esteem) and the reduction of uncertainty (see Hogg, 1996). Second, even if an attitude is made accessible, it is a relatively disembodied construct. A group prototype is a broader construct that embodies how one should behave as well as how one should think. A group prototype can be viewed as a networked schema with associated nodes representing the feelings, thoughts, and actions of a typical group member. Even if accessible attitudes do trigger group-relevant schemas, this process is one step removed from the immediate triggering of a specific group prototype. Thus, it was expected that (a) both attitude accessibility and ingroup norms would influence attitude-behavior consistency; and (b) the effects of attitude accessibility on attitude-behavior relations would be independent of the effects due to ingroup norms. In other words, it was proposed that the effects of ingroup normative support on attitude-behavior consistency should not result simply from making an attitude more cognitively accessible. The first experiment (White, Hogg, & Terry, in press, Experiment 1) focused on university students' attitudes towards comprehensive examsingroup salience (low vs. high) and the attitudinal congruency of the ingroup norms (congruent vs. incongruent) were manipulated. Attitude accessibility was assessed in terms of attitude confidence. Participants (N = 160) completed measures assessing their attitude toward the focal issue (and a number of other nonfocal issues) and their confidence in the attitude, were exposed to ingroup normative information under low or high levels of ingroup salience, and then completed measures assessing their behavior. Salience was manipulated via a self-description procedure developed by Hogg and his colleagues (e.g., Hogg & Hains, 1996; Hogg & Hardie, 1991). Low-salience participants were asked to describe themselves as an individual person and to list attributes that made them unique as an individual, whereas high-salience participants were asked to describe themselves as University of Queensland (UQ) students and to list their similarities with other such students. To manipulate normative support, participants were

< previous page

page_78

next page >

< previous page

page_79

next page > Page 79

exposed to bargraphs showing the percentage of opposition to three issues (including the target issue) from students at UQ (ingroup) and from students at a rival university (outgroup)the graphs indicated either that the ingroup was strongly opposed to comprehensive exams and the rival campus supported them or vice versa. On completion of the experiment, participants were categorized as either having received attitude-congruent ingroup support (if their attitudes and the normative information concurred) or attitude-incongruent normative support. There were three behavioral measures: willingness to engage in behaviors in accord with initial attitude (e.g., participate or not participate in an on-campus rally opposing comprehensive university exams), willingness to sign a relevant petition, and signing a form letter against comprehensive exams (responses to both the latter measures were coded in terms of whether they were attitude-congruent or attitude-incongruent). As expected, participants exposed to an attitudinally-congruent rather than an incongruent ingroup norm behaved more consistently with their initial attitude on the letter-signing and petition-signing measures of behavior; however, contrary to predictions, these effects did not vary as a function of ingroup salience. There was some evidence that on the measure of willingness or not to sign the petition, attitude confidence was associated positively with attitude-behavior correspondenceas expected, this effect was independent of the effect due to norm congruency. To address the limitations of the first experiment, a second experiment was conducted in which attitude accessibility was manipulated and the salience manipulation was strengthened (see White, Hogg, & Terry, in press, Experiment 2). The target issue was ''separate bicycle lanes on roads'' and was linked to gender identity. Under conditions of high or low gender salience, participants (N = 167) were exposed to attitude-congruent or attitude-incongruent ingroup normative support (contrasted with an opposing outgroup norm). Attitude accessibility was manipulated using the repeated expression method, which has been used successfully, as verified with response latencies, to strengthen the association between attitude and attitude object in memory (e.g., Downing et al., 1992; Powell & Fazio, 1984). There were a number of different behavioral measures. Three of these were essentially intention measures: Participants indicated their willingness to engage in attitudinally congruent behaviors (e.g., whether or not to attend a seminar organized by a group supporting separate bicycle lanes) and their willingness to donate both money and time to implement a local student union scheme to introduce a trial system of separate bicycle lanes around campus and the surrounding residential areas. Two other measures constituted more behavioral responses. Participants responded to a ballot ostensibly conducted by the state government concerning a proposal to have separate

< previous page

page_79

next page >

< previous page

page_80

next page > Page 80

bicycle lanes. They were also asked whether, if approached by a relevant organization, they would be prepared to sign a petition supporting separate bicycle lanes. As in Experiment 1, all responses to the behavioral measures were recoded in terms of whether the response was attitudinally congruent. There was some support for the prediction that the effects of ingroup norm congruency would be most marked when the group membership was a salient basis for self-definition. As shown in Fig. 5.1, participants exposed to normcongruent information were more likely to engage in attitudinally congruent behavior (in relation to willingness to donate time) under conditions of high rather low group salience. When exposed to norm-incongruent information, participant attitude-behavior consistency was most marked when the group membership was not made salient, which is consistent with the prediction that ingroup normative information has the least impact on attitude-behavior consistency when the group membership is not a salient basis for self-conception. On both the binary behavioral measures (the state transport ballot and willingness to sign a petition), there was evidence that participants were more likely to behave in accordance with their initial attitudes when (a) they received ingroup normative support and (b) when they were in the high attitude accessibility condition. Taken together, the results of the two experiments provide strong support for the expectation that participants would be more likely to engage in attitudinally congruent behavior when they received normative support for their original attitude from a relevant reference ingroup. In the second experiment, where the manipulation of group salience was more successful than in the first experiment, we also found some evidence that, as expected, the effect of ingroup normative support was stronger under high than low group salience conditions. In both experiments, there was also support for Fazio's (see e.g., Fazio, 1986) prediction that greater attitude accessibility is associated with stronger attitude-behavior consistency. The fact that in both studies accessibility did not interact with group salience or normative support accords with the expectation that social identity and cognitive accessibility processes have independent effects on attitude-behavior consistency. People do appear to behave in accordance with their attitudes if those attitudes are cognitively accessible, but quite independently they also bring their behavior in line with their attitudes (whether they are accessible or not) when there is normative support for their attitudes from a contextually salient ingroup. Although providing some support for the predicted effects of ingroup norms, the first two studies were limited in that an attitudinally incongruent ingroup norm was always accompanied by an attitudinally incongruent outgroup norm. Thus, whether the effects of ingroup norms were a consequence of the fact that an explicit intergroup contrast was drawn between

< previous page

page_80

next page >

< previous page

page_81

next page > Page 81

FIG. 5.1. Effect of norm congruency and salience on attitude-behavior consistency on measure of hours willing to donate to a committee. ingroup norms and the norms of a relevant outgroup was unclear. To examine this possibility, an additional experiment (White, Terry, & Hogg, 1999) was conducted in which ingroup norm congruency (congruent vs. incongruent) and outgroup norm congruency (congruent vs. incongruent) were

< previous page

page_81

next page >

< previous page

page_82

next page > Page 82

independently manipulated. Ingroup salience was held constant at high levels, and attitude accessibility was measured in terms of attitude confidence. The results of the study indicated that information about ingroup norms (rather than outgroup norms) was sufficient to improve attitude-behavior consistency. These results provide evidence that ingroups do exert more influence than outgroups (see Wilder, 1990). They also question the necessity of contrasting outgroup information for ingroup norms to influence attitude-behavior consistency. There may be a similarity between the effects of ingroup norms on attitude-behavior consistency and group polarization (e.g., Turner, 1985; Wetherell, 1987) where there does not appear to be a need to have an explicit outgroup for ingroup polarization to occur (see Hogg & Abrams, 1988). Group Norms, Group Salience, and Mode of Behavioral Decision Making In a second body of experimental research examining the effects of ingroup norms on attitude-behavior consistency, we manipulated the mode of behavioral decision making in addition to the attitudinal-congruence of the ingroup norm. In a similar vein to dual process models of attitude change (i.e., elaboration likelihood model, Petty & Cacioppo, 1986a, and heuristic-systematic model, Chaiken, Liberman, & Eagly, 1989), Fazio (1990a) distinguished two different processes through which attitudes can influence behaviora spontaneous, theory-driven process and a deliberative, data-driven process. In his motivation and opportunity as determinants of mode of behavioral decision-making (MODE) model, Fazio hypothesized that under conditions that favor peripheral route processing of persuasive messages (i.e., low motivation and low ability; see Petty & Cacioppo, 1986a), links between attitudes and behavior arise as a consequence of spontaneous processing, rather than as a consequence of deliberative or reasoned processing. Specifically, when participants are making behavioral decisions via the spontaneous route to the extent to which they are cognitively accessible, attitudes guide behavior through the mediating process of selective perception of the situation (Fazio, 1990a; see also Jamieson & Zanna, 1989). Under deliberative processing, behavioral decisions are made in accord with the mechanism that underpins the TRA and TPBnamely, as a consequence of effortful consideration of the available information, including the consequences of performing the behavior, the perceived expectations of others, and perceptions of behavioral control. Previous research provides some support for the general predictions of the MODE model (Sanbonmatsu & Fazio, 1990; Schuette & Fazio, 1995). Although Fazio (1990a) acknowledged that norms may influence behavioral decision making under spontaneous processing, systematic consider-

< previous page

page_82

next page >

< previous page

page_83

next page > Page 83

ation of a number of different factors, including norms, in the formation of a behavioral decision is restricted to the realm of deliberative processing. Thus, according to the MODE model (Fazio, 1990a), the effects of norms should be more marked under deliberative than spontaneous behavioral decision making. In contrast, on the basis of SIT and SCT, we would predict that, irrespective of mode of behavioral decision making, ingroup norms should influence attitudebehavior consistency. Under both processing routes, assimilation of self to the group prototype means that group norms become a central and important guide to behavioral responses. However, because people for whom the group membership is a salient basis for self-conception are motivated to consider self-relevant group norms carefully (see Mackie, Worth, & Asuncion, 1990; also Forgas & Fiedler, 1996), the effect of norms on such people should be more marked under deliberative than under spontaneous decision-making conditions. Thus, we would expect that the moderating impact of group salience (or group identification) on the effects of ingroup norms would be most marked under conditions of deliberative decision making. In a preliminary study of the effects of ingroup norms and mode of behavioral decision making on attitude-behavior consistency, we conducted a study of career choice in psychology (Terry, Hogg, & McKimmie, 1998, Experiment 1). After indicating their preferred career choice (attitude), participants received one of three sets of normative information: norm-consistent (ingroup members supported their career preference), norm-inconsistent (ingroup members supported another career preference), or no-norm (no clear norm for ingroup members). Participants were then told to make a behavioral decision either spontaneously or deliberatively. On the basis of SIT, it was predicted that participants exposed to an attitudinally-congruent ingroup norm (norm-consistent information) would be more likely to behave in accordance with their attitudes than participants in a no-norm condition (norm neither attitudinally-congruent nor incongruent) who, in turn, would be more likely to behave in accord with their initial attitude than participants exposed to an attitudinally-incongruent norm (norm-inconsistent condition). This effect was expected to strongest for high group identifiers, who made their behavioral decision under deliberative rather than spontaneous decision-making conditions. The main dependent variable was whether participants chose to attend an information session that was consistent with their initial preferred career choiceinformation sessions were ostensibly to be conducted, as part of the experimental session, on each of the three main career choices in psychology (pursuing a career as a clinical psychologist, an organizational psychologist, and a sports psychologist). As shown in Fig. 5.2, results revealed that, as expected, more participants indicated that they would attend the informa-

< previous page

page_83

next page >

< previous page

page_84

next page > Page 84

tion session that was in accord with their initial attitudes in the norm-consistent condition than in the norm-inconsistent condition. There was only weak evidence that participants in the norm-inconsistent condition were less likely to behave in accord with their initial attitude than participants in the no-norm condition. There was a significant normative condition by mode of behavioral decision-making interaction on whether participants' behavioral choice was consistent with their initial attitudenormative condition influenced behavioral choice more strongly for participants in the deliberative condition than for those in the spontaneous condition. Under deliberative decision making, fewer participants behaved in accordance with their initial attitudes in the norminconsistent than in either the norm-consistent or the no-norm conditions (see Fig. 5.3). In the spontaneous condition, participants in the norm-consistent condition were more likely to behave in accord with their attitudes than those in the no-norm condition but not the norm-inconsistent condition. These results reflected the fact that the proportion of participants in the no-norm condition who behaved in accordance with their initial attitude varied as a function of mode of decision making: In comparison to the no-norm participants who made their deci-

FIG. 5.2. Percentage of participants in each normative condition who attended the information session that accorded with their initial attitudes.

< previous page

page_84

next page >

< previous page

page_85

next page > Page 85

FIG. 5.3. Percentage of participants in each normative condition who attended the information session that accorded with their initial attitudes. sion deliberatively, fewer participants in the spontaneous, no-norm condition behaved in accordance with their initial attitude. Strength of identification as a psychology student failed to moderate the effects of normative condition on behavioral choice. However, on a measure of willingness to engage in activities related to finding out more about their preferred career, there was also a significant 3 (normative condition) x 2 (mode of behavioral decision making) x 2 (identification) interaction. Further analysis of this interaction revealed that there was a significant Normative Condition x Mode of Decision Making interaction for participants who did not identify strongly as psychology students, but not for the high identifiers. Low identifiers who received norm-inconsistent information were more willing to engage in actions to find out about their preferred career choice when they made their decision to attend a particular information session under deliberative, rather than spontaneous conditions, (see Fig. 5.4), whereas there were no significant differences in the effects of

< previous page

page_85

next page >

< previous page

page_86

next page > Page 86

norm-consistent and norm-consistent information as a function of mode of behavioral decision making. As predicted, attitude-behavior consistency was influenced by the extent to which the group norm was attitudinally congruent. On the main dependent variablewhether participants chose to attend an information session that accorded with their initial attitudeattitude-behavior consistency was reduced in the norm-inconsistent condition as compared to the norm-consistent and no-norm conditions. The significant normative information by mode of decision making interaction indicated that the impact of normative information was stronger for participants in the deliberative decision making condition than for those in the spontaneous condition, a pattern of results that could be taken as support for Fazio's (1990a) MODE model. However, this conclusion is difficult to justify, given that, irrespective of the mode of behavioral decision making, attitude-behavior consistency was stronger when participants were exposed to an attitudinally

FIG. 5.4. Average willingness to engage in activities in accord with initial attitude or low identifiers in each experimental condition.

< previous page

page_86

next page >

< previous page

page_87

next page > Page 87

congruent ingroup norm (compared to the norm-inconsistent condition under deliberative decision making and compared to the no-norm condition under spontaneous decision making). Under both types of decision making, ingroup norms clearly affected attitude-behavior consistency. Contrary to predictions, the impact of ingroup normative information was similar for low and high identifiers. These results probably reflect the fact that to increase the relevance of the ingroup normative information, the group membership was made salient for all participants. There was, however, evidence that exposure to an attitudinallyincongruent ingroup norm and mode of decision making interacted in a theoretically interpretable manner for low identifiers. Specifically, low identifiers in the norm-inconsistent, deliberative decision making condition were more willing to engage in attitude-consistent behaviors than low identifiers in the norm-inconsistent, spontaneous condition. It is possible that because these people did not identify strongly with the normative reference group, and they had time to think about the oppositional relation between the group norm and their own attitudes, they might have viewed the group as a psychological outgroup, or at least a "non-ingroup." Under these circumstances they might have striven to differentiate themselves from the noninclusive group by being motivated (possibly in pursuit of positive self-esteem) to behave in a way that violates group norms (Crocker, Blaine, & Luhtanen, 1993; Hogg & Abrams, 1990; Tajfel & Turner, 1979). These effects were presumably most marked in the deliberative decision making conditions because of the extensive processing that occurs under these circumstances. A second study was designed to test further the proposed effects of ingroup norms and mode of behavioral decision making on attitude-behavior consistency (Terry et al., 1998, Experiment 2). This study extended the first study by manipulating rather than assessing levels of group salience and using an ability- rather than a memory-based manipulation of mode of behavioral decision making. In an effort to determine the generalizability of the results from the first study, Experiment 2 manipulated the independent variables using a different paradigm to Experiment 1a mock jury paradigm, a paradigm that has been employed widely in a range of different areas of social psychology (e.g., Davis, 1986; Davis, Stasson, Ono, & Zimmerman, 1988; Kaplan & Miller, 1978; MacCoun & Kerr, 1988; Stasser & Davis, 1977). Student participants were divided into jury groups under high or low group salience conditions, and indicated their initial attitudes to a number of crimes, including the focal crime of computer hacking. Then, they were exposed to ingroup normative information, which was congruent or incongruent with their initial attitude. They then read case notes relating to an alleged occurrence of computer hacking, and they were required to make

< previous page

page_87

next page >

< previous page

page_88

next page > Page 88

individual decisions about guilt and punishment (behavior) under restricted or liberal time constraints (manipulation of spontaneous vs. deliberate mode of behavioral decision making). The principal dependent variable was the extent of attitude-behavior inconsistency. On the absolute measure of attitude-behavior inconsistency, there was a significant main effect for norm congruency. As predicted, individuals in the norm-incongruent condition displayed greater attitude-behavior inconsistency than did participants in the norm-congruent condition. Further analyses revealed that in the norm-incongruent condition, attitude-behavior inconsistency, for the most part, reflected movement toward the pole represented by the group norm. Of the participants whose behavioral recommendation was inconsistent with their initial attitude, 80% moved in the direction of the group norm. To assess the relative importance of norms and attitudes in the prediction of the behavioral recommendation, multiple regression analyses were conducted with behavior as the criterion. Participants' perceived level of appropriate punishment and perceived ingroup norm were predictors. Analyses were performed separately for high and low salience participants; however, because there were no effects of mode of decision making in the previous analyses, participants in the two conditions were considered together. In line with the theoretical basis for this research, norms were predictive of behavior only when group membership was highly salient (see Table 5.1). Because there was evidence that the manipulation of group salience did not have a strong effect on perceived identification with the group, a second set of analyses was performed where identification (rather than the manipulated salience variable) was used as an independent variable. These analyses revealed that for participants who did not identify strongly with the group TABLE 5.1 Prediction of Behavior from Attitudes and Norms for High and Low Group Salience Participants Group Salience High (n = 105) Low (n = 97) Variable b b Attitude .38** .39** Norm .28** .19 R2 = .23** R2 = .18**

< previous page

page_88

next page >

< previous page

page_89

next page > Page 89

(i.e., for when it was not a salient basis for self-definition), norm congruency did not have a significant effect on attitude-behavior inconsistency (see Fig. 5.5). In contrast, participants who identified strongly with the group were influenced by the group norm. Those who received an incongruent group norm displayed greater attitude-behavior inconsistency compared to those who received a congruent norm. Among the norm-incongruent participants, there was further evidence that movement towards the group norm was greatest for the high identifiers. Furthermore, norms were predictive of behavior only when individuals strongly identified with the group. In support of the central hypothesis, the results of the second study indicated that the extent of attitude-behavior consistency was influenced by the attitudinal congruence of normative information. On the measure of attitudebehavior inconsistency, there was evidence that an incongruent group norm induced greater behavioral deviance from a previously expressed attitude than did exposure to a congruent group norm; moreover, further analyses revealed that the attitude-behavior consistency in the norm incongruent condition, for the most part, reflected movement toward

FIG. 5.5. Perceived identification by norm congruency interaction on absolute attitude-behavior inconsistency.

< previous page

page_89

next page >

< previous page

page_90

next page > Page 90

the pole represented by the group norm. As expected, there was also some evidence that the influence of norms on the behavioral recommendation tended to be greatest when the group was a salient basis for self-definition, particularly when assessed in terms of strength of group identification. However, there was no evidence that the latter effects were most marked under deliberative decision making conditions. In this second body of experimental research, we found additional evidence that exposure to attitudinally-congruent ingroup norms does improve attitude-behavior consistency. In Experiment 1, there was no evidence for the prediction that the effects of norm congruency would be most marked for high identifiers, a pattern of results that was not surprising, given that fact that the group membership was made salient for all participants. Nevertheless, the fact that an attitudinally-incongruent ingroup norm increased attitude-behavior consistency for low identifiers, at least under deliberative decision making conditions, is consistent with the social identity approach. Experiment 2, which experimentally manipulated levels of group salience, provided stronger support for the prediction that ingroup norms would have most impact when the group membership was a salient basis for self-definition. In accord with the findings reported by Terry and Hogg (1996), there was weak evidence that ingroup norms influenced behavior only for participants in the high salience condition. When strength of group identification, which reflects the psychological importance of the group membership as a basis for self-conception, was considered, there was clear evidence that the impact of the normative information was stronger for high than for low identifiers. This evidence was apparent when three types of data were considered: the absolute amount of attitude-behavior consistency, the direction of movement in the norm incongruent condition, and the prediction of behavior from the ingroup norm. In both experiments, there was no evidence that the effects of normseven when the group membership was a salient basis for self-conceptionvaried as a function of mode of behavioral decision making. This pattern of results is contrary to the MODE model (Fazio, 1990a), which suggests that information, such as ingroup norms, should receive more careful consideration and hence should have more impact on behavior under deliberative than under spontaneous decision making conditions. It is unlikely that the present results are specific to the manipulation of mode of behavioral decision making, given that the same results were obtained across two different types of manipulation of decision makinga motivationally- and an ability-based manipulation. Thus, it appears that, irrespective of the conditions under which behavioral decisions are made, ingroup norms do impact on attitude-behavior consistency.

< previous page

page_90

next page >

< previous page

page_91

next page > Page 91

Conclusions and Future Prospects Taken together, the results of the research reviewed in this chapter support the proposed reconceptualization of the role of norms in attitude-behavior relations along the lines suggested by SIT and SCT. The results of the research provide clear evidence for the central prediction of the research, namely that the attitudinal-congruence of an ingroup norm influences attitude-behavior consistency. In each of the experimental studies reviewed, participants were more likely to behave in accordance with their attitude when exposed to an attitudinally-congruent ingroup norm than when exposed to an incongruent norm. The evidence of a main effect of norm congruency on attitude-behavior consistency is in line with SIT and SCTexposure to an attitudinally-congruent group norm should strengthen attitude-behavior consistency because it validates the attitudinally-congruent behavior as appropriate for group members. This result also accords with the contingent consistency hypothesis and suggests that the lack of clear support for the prediction that attitudebehavior relations are stronger in the presence of a supportive normative climate (e.g., Andrews & Kandel, 1979; Bagozzi, & Schnedlitz, 1985; cf. Grube & Morgan, 1990; Rabow, Neuman, & Hernandez, 1987) is a consequence of the fact that previous tests of the hypothesis have not focused on the norms of relevant ingroups. The second pattern of results that supports our theoretical perspective is the evidence that the extent to which the group membership that defines the norm is a salient basis for self-conception moderates the influence of ingroup norms. In the field studies conducted in relation to two different target behaviors, there was clear evidence that ingroup norms did influence behavioral intentions, but only for respondents who identified strongly with the group membership. In our experimental research, which has focused on the key dependent variable of attitude-behavior consistency, we have also found some evidence for the moderating role of group salience. In the research examining the combined effects of norm congruency and attitude accessibility, attitude-behavior consistency was strongest for participants who received an attitudinally-congruent ingroup norm under conditions of high group salience (White, Hogg, & Terry, in press, Experiment 2). In the studies focusing on mode of behavioral decision making rather than attitude accessibility, Experiment 2, which experimentally manipulated levels of group salience, provided some support for the prediction that ingroup norms would have most impact when the group membership was a salient basis for self-definition. In accord with the findings reported by Terry and Hogg (1996), there was weak evidence that ingroup norms influenced behavior only for participants in the high salience condition. When strength of group identification, which reflects the psychological importance of the

< previous page

page_91

next page >

< previous page

page_92

next page > Page 92

group membership as a basis for self-conception, was considered, there was clear evidence the impact of the normative information on attitude-behavior consistency, and behavioral choice was stronger for high than for low identifiers. The fact that stronger effects were observed for strength of identification than for group salience is entirely consistent with SIT and SCT. According to this perspective, the critical variable for understanding group processes is the psychological importance of the group membership for self-conceptiona variable that is reflected in levels of group identification. In addition to the consistent findings observed for ingroup norms, the experimental research reported in this chapter demonstrates, quite clearly, that the effects of norm congruency are independent of attitude accessibility and are not dependent on mode of behavioral decision making. These results, at a general level, accord with a social identity and self-categorization perspective, although we would expect that ingroup norms would be most influential under conditions that engender deliberative consideration of the normative information. The attitude change literature (see chap. 10, in this volume, by Fleming and Petty, chap. 8, this volume, by Mackie and chap. 9, this volume, by Queller, and van Knippenberg) provides some support for the latter prediction, which suggests that the ability to deliberatively process ingroup information is particularly relevant when group members are exposed to persuasive ingroup messages. The research presented here establishes a clear role for ingroup normative support in attitude-behavior relations. Future research needs to examine in more depth the role that contextual salience of the group membership plays as a moderator of the influence of ingroup norms of attitude-behavior consistency. On the basis of the research reviewed in this chapter, such research clearly should seek to strengthen experimental manipulations of salience. An alternative methodology would be to vary whether the group membership is a self-inclusive social category. On the basis of a social identity and self-categorization perspective, the moderating effect of group norms on attitude-behavior consistency should be evident only if the norms emanate from a self-inclusive membership group. If the group membership is not a basis for self-conceptionif it is not a self-inclusive groupthen group norms should have no impact on behavioral outcomes (see Mackie et al., 1990, for evidence supporting this proposal in the persuasion context; see also Mackie and Queller, chap. 8, this volume). To test the utility of the social identity and self-categorization perspective on norms and attitude-behavior consistency, future research should also test an additional prediction that can be derived from this perspective, namely, that the impact of ingroup norms on attitude-behavior consistency should be most marked on issues that are centrally relevant to social categorization or, in other words, the extent to which the norm is a defining and distinguishing

< previous page

page_92

next page >

< previous page

page_93

next page > Page 93

feature of the group. As Mackie et al. (1990) noted, persuasive messages on group-relevant issues should increase people's capacity, as well as their motivation, to process the message. Similarly, in the context of attitude-behavior relations, whether or not the attitudinal object is group-relevant should influence the extent to which the group norm is processed and, in turn, the extent to which it is regarded as an appropriate guide for action in that context. More broadly, future research on ingroup norms and attitude-behavior consistency should explore in more depth the cognitive mechanisms underlying effects of such norms. A potential model, based on theoretical considerations with some support lent from the research presented here, is that: (a) a context-specific cue is triggered that primes a specific social category; (b) the typical attitudes, norms, feelings, and behaviors associated with being a group member are made available to the individual; (c) the availability of the group prototype results in selective perception and definition of a situation as long as activation of the ingroup makes sense of the situation; and (d) attitudes and behaviors are consistent because they are normative for the salient ingroup. Attitudes and behaviors are inconsistent when the group is not salient in the behavioral context. Thus, an attitude becomes accessible and affects on behavioral outcomes because a group prototype is made cognitively available. Given that the group prototype encompasses attitudes, feelings, and behaviors of a valued group membership, it provides the impetus for behavioral action. Ingroup members adhere to group norms to reduce the uncertainty of the interpretation of social stimuli as well as to satisfy their need for positive self-evaluation (self-esteem; see Hogg, 1996). In conclusion, the results presented here are important in that they reflect clear support for the prediction that norm congruency influences the strength of attitude-behavior consistency. Such results should serve to maintain an important and central role for social influence in attitude-behavior relationsa role that has been questioned by recent commentators (e.g., Ajzen, 1991)along the lines suggested by SIT and SCT As such, they open the way for further research to take a more fine-grained look at the role that group membership and group norms play in this context.

< previous page

page_93

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_95

next page > Page 95

6 Pluralistic Ignorance and Inconsistency Between Private Attitudes and Public Behaviors Dale T Miller Benoît Monin Deborah A. Prentice Princeton University Psychologists have a longstanding interest in the utility of attitudes for predicting behavior, an interest that has been both frustrated and maintained by the ever accumulating evidence that their utility is mediocre at best. The many findings of attitude-behavior inconsistency have been viewed as discouraging for at least two reasons: First, most claims for the validity of attitudes as a social construct have rested on their ability to predict behavior, and second, practitioners have hoped to be able to use persuasion techniques to modify deleterious behaviors, a strategy that requires not just attitude-behavior consistency but also a causal link between the two. As a consequence, researchers have been motivated to find attitude-behavior consistency wherever it might live. They have found a number of successful approaches, including refining the measurement of attitudes (Fishbein & Ajzen, 1975), refining the measurement of behaviors (Fishbein & Ajzen, 1974, 1975), changing the definition of consistency (Campbell, 1963), and identifying individual characteristics and social circumstances that promote consistency (Fazio & Zanna, 1981; Sherman & Fazio, 1983). These efforts have led to the development of models for predicting behavior from attitudes, many of which are described elsewhere in this volume.

< previous page

page_95

next page >

< previous page

page_96

next page > Page 96

In this chapter, we take a slightly different angle on the question of attitude-behavior consistency. In particular, our interest is not in predicting behavior from attitudes but rather in the reverse; we seek to understand the ways in which people use and misuse the data of observable behavior to draw inferences about other people's private attitudes. This task is familiar to ordinary social perceivers who typically do not enjoy the kind of access to people's private thoughts and feelings that research psychologists have. Instead, they must infer these private views from public statements and actions (Miller & Prentice, 1996). Because they have access to behaviors but not to attitudes, social perceivers are interested in a slightly different question than are research psychologists. Whereas psychologists tend to ask, ''To what extent and under what circumstances will this person's attitudes be expressed in his or her actions?'', social perceivers more often ask, "What can I learn about this person's attitudes from the way he or she acts?" Our focus on the prediction of private attitudes from public behaviors takes us away from definitional and methodological solutions to the quandary posed by attitude-behavior inconsistency, for these solutions are more useful for enhancing behavioral prediction and modification than for diagnosing private thoughts and feelings. Indeed, the fact that people exhibit more attitude-behavior consistency when researchers measure attitudes toward specific behaviors, use an aggregated behavioral index, or assess consistency in light of situational thresholds is of little interest to most social perceivers. For them, consistency is not a question; it is an assumption. They expect to find some degree of correspondence between an individual's thoughts, words, and deeds, and this expectation, combined with its lack of validity in most social situations, lead them to social inferences that are seriously in error. We analyze this phenomenon, known as pluralistic ignorance, in the second part of this chapter. First, we consider the nature of public behavior and its implications for attitude-behavior consistency. Public Behavior Our analysis of public behavior draws on the insights of two strong theoretical traditions in social psychology, one focusing on self-presentation and the other on social identity. In the self-presentation tradition, we trace our ideas to Goffman (1959), who likened public behavior to a theatrical performance, in which the individual plays a part on a social stage with others who simultaneously serve as co-actors and as audience to his or her performance. This metaphor seems to us to capture very well the self-consciousness of behaving in the presence of others. Public behavior is enacted onstage with an eye to the impression that behavior creates in the minds of the audience. Because most public actors wish to be seen as good and appro-

< previous page

page_96

next page >

< previous page

page_97

next page > Page 97

priate members of their groups and societies, their behavior is guided by socially shared definitions of how people like them are supposed to act. These shared definitions, or social norms, vary in their generality: Some apply to nearly all social interactions (e.g., norms of politeness, turn-taking, or reciprocity); some are specific to particular situations (e.g., norms of orderliness in queues and crowds, and of exchange in commercial transactions); and some are specific to particular groups or types of people (e.g., norms of delinquency among gang members, of alcohol consumption among college students, of strictness among prison guards). By behaving in accordance with social norms, actors are able to communicate both their definition of the situation and their character or role within it. As to the content of social norms, both self-presentation and social identity theories contribute insights. Goffman (1959), for example, pointed to the ways in which participants in an interaction behave to protect and defend the impressions that they and others project. Public behavior is part of a delicate negotiation in which participants converge on a joint definition of the situation and their respective roles in it. This negotiation defines the norms pertaining to that situation. Social identity and self-categorization theories trace norms to the prototypes of the groups with which individuals identify (Turner, 1991; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). According to this theoretical perspective, the behavior that is normative for group members is that which highlights what they all have in common and at the same time most distinguishes them from members of other groups (i.e., that which maximizes the metacontrast ratio; see Turner et al., 1987). Norms thus serve to maintain the integrity and boundaries of social categories. It is clear from this discussion why social norms often produce public behavior that is inconsistent with individuals' private attitudes. Norms are properties of situations and of groups, not of individuals. They develop through processes that have only an indirect and partial connection to the characteristics and views of those who are influenced by them. Therefore, it is not surprising that the normative behavior that is exhibited in public settings is frequently counterattitudinal for some or even most of the people who are enacting it. As an illustration of the power of social norms to shape public behavior in a direction contrary to private attitudes, consider the classic demonstration of attitude-behavior inconsistency by LaPiere (1934). Over a 2-year period, LaPiere and a young married Chinese couple visited 184 restaurants and 67 hotels, auto camps, and tourist camps. Sometimes all three of them went into the establishment, sometimes the Chinese couple went in first. The couple was refused service only once, at an autocamp. Clearly, the actual practices of the businesses showed no evidence of anti-Chinese prejudice. Neither did the interpersonal treatment the couple received. LaPiere coded

< previous page

page_97

next page >

< previous page

page_98

next page > Page 98

the reception accorded the couple as hesitant, normal, or better than normal and found that the couple received treatment that was normal or better than normal more than 90% of the time. Of course, LaPiere found a very different picture when he sent a questionnaire to these and other establishments, asking (among other questions), "Will you accept members of the Chinese race as guests in your establishment?" Of the 256 establishments that returned the questionnaire (128 of which he and the couple visited previously), only 2 declared that they would accept Chinese guests unconditionally, and 18 said they might, depending on the circumstances. Thus, whereas more than 99% of the first sample accepted Chinese guests, more than 93% of the second sample (which included members of the first sample as well as new individuals) declared in writing that they would not do so. Why did the desk clerks observed by LaPiere admit a Chinese couple to their establishments when they said they would not? Based on LaPiere's observations, it appears that behavior was directed by general norms pertaining to social interactionto what Goffman (1959) called the etiquette of social life. According to Goffman, there exists a powerful norm not to violate the self-definition of others. In Goffman's words: "When an individual projects a definition of the situation and then makes an implicit or explicit claim to be a person of a particular kind, he automatically exerts a moral demand on others, obliging them to value and treat him in the manner that persons of this kind have a right to expect" (p. 185). From LaPiere's account, this dynamic was responsible for the fact that his respondents offered accommodation to members of a group toward which they held negative attitudes. LaPiere pointed to the important role played by the "air of self-confidence" that the couple exuded as they made their request. This projected self-image and the normative obligation it incurred proved much more powerful than the desk clerk's attitudes. In LaPiere's words, ''A supercilious desk clerk could not refuse his master's hospitality to people who appeared to take their request as a perfectly normal and conventional thing" (p. 232). In effect, although refusing accommodation to a Chinese couple might not have constituted a social impropriety, refusing accommodation to a couple (Chinese or otherwise) who clearly defined themselves as deserving of it did. It is instructive here to consider an alternative reading of LaPiere's results. Campbell (1963) argued that the attitudebehavior mismatch observed in LaPiere's and many similar studies do not provide demonstrations of inconsistency, but instead are artifacts of situational thresholds. Because the threshold ("hurdle," as he calls it) for actually refusing accommodation is higher than for expressing a generalized intention to refuse, we should not take the behavioral frequency data as evidence of inconsistency. The mea-

< previous page

page_98

next page >

< previous page

page_99

next page > Page 99

sure is too insensitive. Given the difference in thresholds, the only way respondents could be inconsistent, in Campbell's view, would be to say they would accept Chinese guests and then not accept them, an attitude-behavior pattern that LaPiere did not observe. Raden (1977) translated this position into a systematic model. According to Raden, if one uses discrete measures of attitudes (+ or -) and behavior (+ or -), there are four possible relationships between a person's attitudes and behaviors: The person could possess an attitude and act on it (+ +), the person could possess an attitude but not act on it (+-), the person could act as if he/she possessed an attitude although he/she does not (-+), or the person could neither possess an attitude nor act upon it (- -). Raden argues that because hurdles are lower for attitude expression than for behavior display, only the "-+ " cell, not the " +-" cell, should be considered to reflect attitude-behavior inconsistency. When he re-examined the extant research on the attitude-behavior link using this revised index of inconsistency, Raden found very high levels of attitude-behavior consistency (generally greater than .95). In our view, this attempt to redefine attitude-behavior consistency by introducing the concept of a hurdle or threshold is misguided. True, the threshold for acting on one's attitude in the face of a counterattitudinal norm is quite high, but norm-congruent behavior in these circumstances is still counterattitudinal. Nor is it useful to characterize behavioral measures such as that used by LaPiere as insensitive. The measure he used was, in fact, quite sensitive, although to social norms rather than to respondents' attitudes. Such a measure may be of little use to researchers interested in dispositional prediction, but it would be highly prized by those interested in social influence. Role of Norms in Theories of Attitude-Behavior Relations We are by no means the first to argue for the importance of taking social norms into account in analyses of attitudebehavior consistency. Perhaps the best-known formulation of the role of norms vis-à-vis attitudes in predicting behavior was provided by Fishbein and Ajzen (1975) in their theory of reasoned action. They proposed that attitudes and norms exert independent and additive effects on behavioral intentions, which, in turn, influence overt behavior. The effective norm for a given individual is assessed by weighting that individual's perceptions of the expectations of reference others by his or her motivation to comply with their expectations. Thus, subjective norms, like attitudes, vary across individuals, as a function of whose expectations matter and what those expectations are. In Ajzen's (1991) more recent theory of planned behavior, the assessment of norms

< previous page

page_99

next page >

< previous page

page_100

next page > Page 100

and their role in the prediction of behavior has remained largely the same as in the earlier theory. The theories of reasoned action and planned behavior have proven quite successful at predicting behavior from behavioral intentions and behavioral intentions from attitudes, but their formulation of the role of subjective norms in predicting behavioral intentions has received little empirical support (see Ajzen, 1991 for a review). By way of explanation, its authors (Ajzen, 1991; Fishbein & Ajzen, 1975) argued that this apparent weakness in their theories reflects an unintentional bias in the particular behaviors and populations they have studied; had they chosen different types of behaviors or different populations of participants, they might well have seen a stronger influence of norms on intentions. Others have seen the weakness as more endemic. For example, Terry and Hogg (1996) argued for a reconceptualization of norms and their relation to attitudes in light of social identity and self-categorization theories. In particular, they proposed that subjective norms should predict behavioral intentions only for individuals who identify strongly with the reference others in question and that, for these individuals, norms should relate to attitudes as well. In two studies, they found empirical support for these emendations. Although Terry and Hogg's proposed revisions of the theories of reasoned action and planned behavior seem to us quite sensible, the theories remain extremely limited in their abilities to account for normative influence. This limitation stems, in large part, from the cognitive and individualistic focuses of the theories. Both conceive of attitudes and norms within an expectancy-value framework, measure them at the level of individuals, and predict that they will affect intentions to behave, but not behavior itself. We might expect this kind of model to provide strong prediction of behaviors that are rational, planned, and enacted singly. But what of public behaviors, which are often symbolic, spontaneous, and negotiated between interacting parties? In the public domain, we would expect an entirely different set of relations between attitudes, norms, and behaviors, with important implications for the nature and magnitude of attitude-behavior consistency. Norms and Behavioral Uniformity To illustrate the influence of norms on public behavior, we return to the LaPiere (1934) study described earlier. Most analyses of that study highlight the remarkable degree of attitude-behavior inconsistency it demonstrates. But equally striking is the uniformity of participants' actions: Ninety-nine percent of those approached accepted the Chinese couple into their establishment. Clearly, the norms for social interaction dictated the same response in all cases. Moreover, these norms appear to have influenced

< previous page

page_100

next page >

< previous page

page_101

next page > Page 101

behavior directly, rather than influenced behavioral intentions: Ninety-three percent of an overlapping and comparable sample declared in writing their intentions to keep Chinese people out. Admittedly, this question about behavioral intentions was phrased quite generally. But would the proprietors have responded differentlywould they have intuited the effects of normative pressure on their behaviorif they were asked about their behavioral intentions in that particular situation? We expect not. And what if LaPiere had asked them about the expectations of important reference others, others with whom they strongly identified? Would the views of those others have reflected the norms that ultimately influenced their behavior? Again, we expect not. In short, we believe that LaPiere's findings illustrate the ways in which norms actually influence public behavior. In public contexts, norms reduce the variance in behavior across individuals and affect behavior directly, rather than through behavioral intentions. Indeed, the social influence literature contains numerous examples of laboratory and real-world situations that function similarly. Whether individuals are confronted with a confident Chinese couple asking for accommodation, an unyielding researcher in a learning experiment (Milgram, 1974), a crowd of unresponsive bystanders to an emergency (Latané & Darley, 1970), or a group of peers evaluating hypothetical decision dilemmas (Myers, 1982), their actions tend to become more similar to one anothers', more norm-congruent, and more independent of their private attitudes and perceptions. In addition, dissonance researchers routinely engineer this type of behavior in the high choice condition of the induced compliance paradigm (Festinger & Carlsmith, 1959). The reason people agree to write an essay advocating a position contrary to their attitudes under conditions of high choice (low justification) is the same reason that LaPiere's respondents acted contrary to their attitudesto do otherwise would violate another's definition of the situation. To induce participants to write a counterattitudinal essay under these conditions, the experimenter must project the impression that he or she considers this request reasonable, thereby defining the refusal to do so as a social impropriety. The homogenizing effect of norms on public behavior has long been recognized by personality researchers interested in dispositional prediction. For example, Mischel (1977) noted that powerful situations, in which everyone construes events the same way, holds uniform expectancies regarding appropriate behavior, and is motivated to learn and enact that behavior, tend to induce behavioral uniformity. In a similar vein, Price and Bouffard (1974) found that situations that are high in constraint (i.e., those in which few behaviors are appropriate) are seen as potentially embarrassing, as requiring careful monitoring of one's behavior, and as demanding certain behaviors

< previous page

page_101

next page >

< previous page

page_102

next page > Page 102

rather than others (see also Snyder & Ickes, 1985, for a discussion of the effects of situational constraint). These are precisely the properties of the public situations that we have been describing. Private Attitude-Public Behavior Consistency What are the implications of this analysis for consistency between public behavior and private attitudes? In this connection, there are at least two types of consistency in which we might be interested: correlational consistency (are people ordered in the same way on both the attitude and behavior measures?) and literal consistency (do people do what they say they will do? See Schuman & Johnson, 1976, for a discussion of this distinction). Correlational consistency is indexed by the magnitude of the correlation coefficient relating a measure of attitude and a measure of behavior in a populationthe higher the correlation, the greater the consistency. Literal consistency is indexed by the difference between the level of behavior implied by an individual's or group's attitudes and that exhibited by their overt behaviorthe smaller the difference, the greater the consistency. These two forms of consistency are largely independent of one another: A particular sample of attitudes and behaviors can be high on both types of consistency (e.g., when attitudes are the primary or sole influence on behavior), high on correlational consistency and low on literal consistency (e.g., when attitudes are one of a number of properties of individuals that influence behavior), low on correlational consistency and high on literal consistency (e.g., when normative pressures produce attitude-consistent behavior), or low on both types of consistency (e.g., when normative pressures produce attitude-inconsistent behavior). LaPiere's study, for example, falls into this last category: His data showed no correlation between attitudes and behaviors and showed a sizeable gap between the implications of what establishment owners said and what they did. As an aside, it is interesting to note the extent to which a correlational definition of consistency has come to dominate research on the attitude-behavior relation. We conducted an informal review of all the articles published since 1974 in the Journal of Personality and Social Psychology (JPSP) in which the phrase attitude-behavior appeared either in the title or in the abstract. After omitting 1 review article and 5 that included only a brief mention of the consistency issue in their discussions, we were left with 26 articles that used consistency as their main dependent measure. Of these, 16 used simple correlations, 7 used more elaborate covariance-based approaches like regression or structural equation modelling (still, for our purposes, a correlational measure), and 3 included some other conceptualization of the attitudebehavior relation. Thus, nearly 90% of the JPSP studies that examined attitude-behavior consistency over the last 23 years defined it in

< previous page

page_102

next page >

< previous page

page_103

next page > Page 103

correlational terms, often failing to provide measures of central tendency and dispersion for the attitude and behavior distributions in their reports. We would expect norms to have a very predictable effect on correlational consistency between private attitudes and public behaviors: By restricting variation in behavior, normative pressure should reduce the attitude-behavior correlation (see Nunnally, 1978). Note that this effect is not well captured by most existing theories, which attempt to explain behavior as a function of a linear combination of dispositional and situational forces (e.g., Ajzen, 1991; Fishbein & Ajzen, 1975). The effect of norms on literal consistency is less easy to predict. In some cases, when norms and attitudes prescribe the same behavior, normative influence has no effect on attitude-behavior consistency. For example, consider what LaPiere would have found had he examined attitudes and behaviors toward a well-to-do White couple. In this case, the respect and accommodation that norms of social interaction required would have coincided with respondents' private attitudes. Although their behavior would still have been driven by normative pressure, it would have been authentic. Therefore, no literal inconsistency between attitudes and behaviors would have been observed (and, we might note, the study would have generated no interest in the psychological community). However, in other cases, when norms and attitudes prescribe different behaviors, normative influence reduces literal consistency. It also produces pluralistic ignorance, a social psychological phenomenon to which we now turn attention. Pluralistic Ignorance Pluralistic ignorance is a form of erroneous social inference that is at once both a cause and a consequence of literal inconsistency between private attitudes and public behaviors. The term pluralistic ignorance was coined by Floyd Allport (1924) to describe the situation in which virtually all members of a group privately reject group norms yet believe that virtually all other group members accept them (see Miller & McFarland, 1991; Miller & Prentice, 1994; Prentice & Miller, 1996). Allport introduced this concept to account for an especially puzzling case of literal attitudebehavior inconsistencywidespread public conformity to social norms in the absence of widespread private support. By way of explanation, he argued that people do not act on attitudes unless they believe those attitudes are shared. Thus, in the extreme case, when everyone believes that everyone else holds an attitude that, in fact, no one holds, the result is a complete attitude-behavior disjunction. The first two social psychology doctoral dissertations completed in America, both conducted under the supervision of Allport, demonstrated

< previous page

page_103

next page >

< previous page

page_104

next page > Page 104

the potential of pluralistic ignorance to produce attitude-behavior inconsistency. In the first, Katz (Katz & Allport, 1928) reported a number of instances of inconsistency in the private attitudes and public reactions of Syracuse University students. Katz found that the majority of fraternity members on campus claimed to have no personal objections to admitting students from certain groups to their fraternities and yet consistently voted to exclude these same students. As in the contemporaneous LaPiere study, then, participants' private attitudes and their public behaviors diverged, although in this case, the nature of the mismatch was the opposite to that found by LaPiere. Here, participants attitudes reflected less, not more, prejudice than did their actions. There was also a methodological difference between the two studies: Katz, unlike LaPiere, asked his participants about their peers' attitudes as well as their own. Their responses to these questions suggested an explanation for the particular attitude-behavior divergence Katz had observed. Participants expressed the belief that their fraternity brothers, unlike themselves, did hold negative attitudes toward certain groups (after all, why else would they have voted to exclude them?). Furthermore, participants indicated that it was because of the widespread negative attitudes of their peers that they themselves voted to exclude members of these groups from their fraternity. Their (mis)perception of the consensual attitude became the norm that guided their behavior. The second, and better known of the early demonstrations of pluralistic ignorance was provided by Schanck's (1932) dissertation on the social attitudes of the residents of the community of "Elm Hollow." The Methodist Church in Elm Hollow was very influential, and Schanck found that the community residents were nearly unanimous in stating their public support for the church's prohibitions against card playing, drinking alcohol, and smoking. Yet Schanck reported that, over his extended stay in the community, he himself had played cards, drunk hard cider, and smoked with many, if not most, residents (although always in the privacy of their homes). Consistent with his experience, Schanck's survey of people's private attitudes revealed little private support for the public norms. What his survey also revealed, however, was that residents assumed that their fellow community members actually believed in and closely followed the dictates of the church. Once again, then, we find both an attitude-behavior mismatch (people reporting private attitudes that were more liberal than their public actions) and a possible account for this mismatch (people acting consistently not with their own attitude, but with their mistaken perception of the consensual attitude of their peers). Our own program of research on college students' private and public expressions of comfort with campus drinking practices has provided us with

< previous page

page_104

next page >

< previous page

page_105

next page > Page 105

the opportunity to investigate the phenomenon of pluralistic ignorance more systematically. Our initial investigations turned up strong evidence of pluralistic ignorance in student's attitudes toward alcohol use on campus (Prentice & Miller, 1993). Students indicated that their own comfort with students' drinking habits was less than that of the average student, of their friends, and of their same-gender peers. Moreover, this self-other difference was not driven by concerns with self-enhancement or self-presentation, for students reacted to their estimates of peer opinion as if they actually believed them. They tended to adjust their own attitudes to be closer to their estimates of peer attitudes over time and to experience alienation to the extent that they perceived their own views as deviant. Having established that pluralistic ignorance existed in the attitudes of students, we next conducted a laboratory study (Prentice & Miller, 1996) to determine if this misperception of peer attitudes was driven by inauthentic public behavior in alcohol-related situations. We brought groups of female Princeton undergraduates into the laboratory to participate in group discussions of campus issues. They were first asked to indicate on a private questionnaire both their own comfort with the alcohol drinking habits of students at Princeton and the comfort of the average student of their gender. They were next asked to discuss with one another a series of general questions about the role of alcohol in campus life. Following the discussion, each student completed a second questionnaire in which she rated the comfort with alcohol of each of the other members of her discussion group, how comfortable the other members believed her to be, and how similar her opinions were to theirs. Students' prediscussion ratings of their own comfort and the comfort of the average female student replicated the selfother discrepancy found in Prentice and Miller (1993), with students expressing less comfort than they attributed to the average student of their gender. More interesting to us was the relation of these prediscussion comfort ratings to the comfort they expressed in the group discussion. Participants' ratings of how other group members would describe them confirmed our expectation that they were expressing more comfort publicly than they felt privately. Following the discussions, students indicated that they expected the other group members to see them as significantly more comfortable with campus drinking habits than their own prediscussion ratings suggested. Indeed, their ratings of the comfort they conveyed in the discussions were closer to (and statistically indistinguishable from) the comfort they attributed to the average female student than to their own private views. Our final study in this series focused on the effects of dispelling pluralistic ignorance. If, as we have been arguing, pluralistic ignorance is a cause of literal inconsistency between private attitudes and public behaviors, then dis-

< previous page

page_105

next page >

< previous page

page_106

next page > Page 106

pelling pluralistic ignorance should release the pressure to conform to others' views and render behavior more attitudecongruent. A quasi-experimental test of this proposition was provided by America's experience with the prohibition of alcohol during the early part of this century. Although strongly advocated and enforced by various constituencies in America, prohibition never had majority support. But because people were reluctant to express their widely shared antiprohibition sentiment, it did seem to have public support (Robinson, 1932). This all changed with the advent of public opinion polls, however, as these polls revealed the depth of private antiprohibition sentiment and thereby dispelled a national case of pluralistic ignorance. Prohibition, with its lack of majority support exposed, soon thereafter ''collapsed like a punctured balloon" (Katz & Schanck, 1938, p. 175). We conducted a more controlled test of the effects of dispelling pluralistic ignorance in the context of alcohol use on campus (Schroeder & Prentice, 1997). Entering Princeton students were randomly assigned to participate in one of two types of hour-long discussion sessions about alcohol use during their first week on campus. The two types of sessions were identical except for the content of a 20-minute discussion segment. In the norm-focused condition, the discussion centered on pluralistic ignorance and its implications. Students were presented with the data showing pluralistic ignorance regarding alcohol use on campus and were encouraged to talk about why these misperceptions might have developed. They were also asked to reflect on how misperceiving the norm for drinking might affect social life on campus. In the individual-focused condition, the discussion, following many existing programs designed to change drinking behavior, centered on how individual students can make responsible decisions about alcohol consumption. Students were encouraged to reflect on the types of situations in which they might encounter alcohol at the university, to explore their options in those situations, and to consider the personal and social consequences of various courses of action. Four to 6 months after the discussions, students in both conditions completed self-report measures of their alcohol consumption. Did dispelling pluralistic ignorance reduce students' drinking behavior to a level more consistent with their attitudes? Apparently it did: Students in the norm-focused condition reported consuming almost 40% fewer drinks each week than did students in the individual-focused condition. From their responses to additional measures collected during the initial and follow-up sessions, it appears that dispelling pluralistic ignorance produced a change in drinking behavior by making students aware of the variability of their peers' attitudes. In other words, it was not the perception that students, on average, were comfortable with heavy drinking, but rather the perception

< previous page

page_106

next page >

< previous page

page_107

next page > Page 107

that everybody was comfortable with heavy drinking that was responsible for students' attitude-inconsistent behavior. Sources of Attitude-Inconsistent Public Behavior In the foregoing discussion, we emphasized the ways in which pluralistic ignorance can create literal inconsistency between private attitudes and public behaviors, as individuals act according to their mistaken perceptions of what everybody else truly believes. But these mistaken perceptions on which they base their behaviors are themselves produced by observing the attitude-inconsistent behavior of their peers. What initiates this chain of events? How does the cycle of erroneous inferences and inauthentic behaviors get started? In our discussion of the LaPiere (1934) study, we emphasized the role that social etiquette plays in producing attitudeinconsistent behavior. There, the proprietors were driven to accept the Chinese couple into their establishments because to refuse them would have violated the couple's self-presentation-it would have challenged their face. A similar dynamic may account for a number of cases of pluralistic ignorance, notably those described by Latané and Darley (1970) and Miller and McFarland (1987). It may also play a role in the drinking situation that we have been describing. However, there are many other circumstances under which individuals behave in counterattitudinal ways. Here, we consider the role of temporal factors and social identities in producing attitude-behavior disjunctions. Temporal Decoupling of Attitudes and Behaviors Attitudes and behaviors often follow quite different courses over time. Factors that change attitudes can produce little or no change in behavior; conversely, a rapid change in behavior can be followed by a much slower adjustment in attitudes. These differing temporal patterns often lead to the gradual or sudden development of attitude-behavior mismatches that are then sustained by pluralistic ignorance. Indeed, many of the documented cases of pluralistic ignorance may have arisen as attitudes and behaviors that were once in sync diverged over time. Here, we consider two possible means by which attitudes and behaviors can follow different temporal patterns and thereby produce pluralistic ignorance. Shifting Attitudes and Stable Behaviors: Conservative Lags. The most common case of attitude-behavior divergence occurs when attitude change is not accompanied by behavior change, thus leaving social

< previous page

page_107

next page >

< previous page

page_108

next page > Page 108

practices in place long after they have lost private support. Why might people continue to behave as they always have, despite a change in their attitudes? One possibility is that they fail to recognize that their attitude change has been shared by others (O'Gorman, 1986). The evolution of the attitudes of Whites toward the integration of Blacks in the United States provides a case in point. The civil rights movement, along with other social forces, produced a shift in the private attitudes of White Americans toward integration long before there was any corresponding shift in social norms and public behavior (Breed & Ktsanes, 1961; Fields & Schuman, 1976; O'Gorman, 1975; O'Gorman & Garry, 1976). One reason for this so-called conservative lag was that Whites believed that the changes in their own thinking were not shared by other Whites. Mistakenly assuming that the majority of their White peers continued to support segregation, they acquiesced to a status quo they no longer privately believed in (Fields & Schuman, 1976). A similar analysis may also explain a result that Moscovici and his colleagues (Moscovici & Lage, 1976; Moscovici, Lage, & Naffrechoux, 1969; Moscovici & Neve, 1972) found in their investigations of minority influence. These researchers have shown that even when the majority of a group's members have been influenced by a persistent minority (as assessed by private measures of attitude change), their public behavior tends to remain unchanged. One possible reason for this reluctance to yield publicly is that the converted majority members may mistakenly think that the other majority members remain unmoved. We believe that many of the cases of pluralistic ignorance that we have described can be traced, at least in part, to a conservative lag. It seems quite likely, for instance, that Syracuse fraternity members once held prejudiced attitudes toward members of outgroups, that residents of Elm Hollow once privately supported the dictates of the Methodist Church, and that Princeton students were once comfortable with excessive drinking practices. Whatever factors were responsible for changes in these attitudes, their effects were never publicly acknowledged. As a result, public behavior went on as it always had, with pluralistic ignorance there to sustain it. Stable Attitudes and Shifting Behaviors: Liberal Leaps. Although the tendency of behavior change to lag behind attitude change is a common feature of social life, occasionally the reverse can happen-behavioral change can speed ahead without a corresponding change in attitudes. Revolutions are one such circumstance. For example, Alexis deTocqueville proposed that the private attitudes of the French toward the Church in the middle of the 18th century did not undergo nearly as abrupt a shift after the revolution as did the public support of the Church. De Tocqueville's ac-

< previous page

page_108

next page >

< previous page

page_109

next page > Page 109

count of this provides an eloquent description of pluralistic ignorance at work: ''Those who retained their belief in the doctrines of the Church because of being alone in their allegiance and, dreading isolation more than error, professed to share the sentiments of the majority. So what was in reality the opinion of only a part . . . of the nation came to be regarded as the will of all and for this reason seemed irresistible, even to those who had given it this false appearance" (1955, p. 155). A similar claim was made by Klassen, Williams, and Levitt (1989) in their analysis of the so-called sexual revolution in the United States during the 1960s and 1970s. According to these researchers, people's attitudes during this period did not change nearly as much or as quickly as their public pronouncements and the rhetoric of the times would suggest. As occurred in the French Revolution, however, the doctrine of the new regime tended to go publicly unchallenged because of the mistaken assumption that it enjoyed the private support of the majority. Pluralistic ignorance, thus, seems to have accompanied (and possibly even aided and abetted) the sexual revolution. Social Identities A final reason that individuals might act in ways that are inconsistent with their private attitudes is because they are conforming to the ideals, norms, or values of their social identities (Abrams & Hogg, 1990; Hogg & Abrams, 1988; Turner, 1991; Turner et al., 1987). As Terry and Hogg (1996) stated: "When social identity is salient, depersonalization occurs, such that a person's feelings and actions are guided more by group prototypes and norms than by personal factors" (p. 790). Because a group prototype tends to embody those beliefs, attitudes, and behaviors that minimize ingroup differences and maximize intergroup differences (Tajfel, 1978, 1981; Turner et al., 1987), intergroup contexts are especially likely to produce discrepancies between public behavior and private attitudes. Highly structured social institutions, such as prisons or schools, are one context in which groups with different social identities commonly interact with one another. Studies of social life in such institutions frequently reveal that members of different roles (e.g., prisoners and guards, students and teachers) believe that their peers are more hostile to occupants of other roles than they themselves are (Akers, 1977; Benaquisto & Freed, 1996; Kauffman, 1981; Packard & Willower, 1972; Wheeler, 1961). For example, Klofas and Toch (1982) found that although only 37% of guards agreed with the position that "the best way to deal with inmates is to be firm," almost 70% believed that their fellow guards would agree with this statement. In explaining the tendency of role occupants to underestimate the most common of their private attitudes, researchers frequently point to the

< previous page

page_109

next page >

< previous page

page_110

next page > Page 110

disjunction between private attitudes and public behaviors that occurs in these contexts. As Toch and Klofas (1984) observed, the pressure to defend the ingroup's values results in prisoners' and guards' onstage behaviors being less sympathetic to the outgroup than their private views would dictate. Similarly, Benasquito and Freed (1996) noted that the views inmates expressed to their peers were less consistent with their private view than with "expectations of the code characterizing the inmate social system" (p. 504). Social identities, of course, do not become salient only in formally structured intergroup contexts such as social institutions. Individuals are conscious of, and guided by, their social identities in virtually all intergroup contexts. An illustration of this fact and of the way it can lead to pluralistic ignorance is provided by the group conflict research of Ross and his colleagues (Robinson, Keltner, Ward, & Ross, 1995; Ross & Ward, 1996). For example, in one of their studies, the focal groups were college students who acknowledged having strong pro-choice versus pro-life views relevant to the ongoing abortion rights debates. These students were presented with a series of questions about abortion (e.g., what kind of abortion scenarios and considerations are common vs. uncommon; what positive consequences and what negative consequences would be likely to follow from a tightening of abortion restrictions, etc.) and were asked to give both their own responses to these questions and the responses they believed the average member of their group and the average member of the opposing group would give. The results indicated that the "pro-choice" and "pro-life" groups did differ from one another but not nearly as much as they assumed they did. Partisans tended to view not only members of the opposition but also the members of their own side as extremist and unduly influenced by ideology and bias. They perceived their own more moderate views as atypical vis-à-vis their group. The ingroup pluralistic ignorance and outgroup stereotyping that Robinson et al. (1995) observed may share a common source: The fact that partisans' public behavior reflected the sharply defined norms and positions of their social identity rather than their less-extreme positions of their private attitudes. Indeed, as the students told Robinson et al., "they rarely acknowledged to others the degree of ambivalence in their political beliefsnot in talking to their ideological allies and not in talking to their ideological adversaries (lest their concessions be exploited or misunderstood)'' (Ross & Ward, 1996, p. 123). Groups need not even have an explicit source of conflict, such as the legality of abortion, for there to be pressures on their members to differentiate themselves publicly from one another. The emergence of gender norms is a case in point. By the time children enter elementary school, they know what toys and activities are considered appropriate for each gender (Huston, 1983), the trait adjectives that typically describe girls and those that de-

< previous page

page_110

next page >

< previous page

page_111

next page > Page 111

scribe boys (Williams, Bennett, & Best, 1975), and the occupations that are traditionally held by men and those that are held by women (O'Keefe & Hyde, 1983). Moreover, they act on their knowledge of norms governing genderappropriate behavior by, for example, preferring toys and activities considered appropriate for their gender and actively avoiding those considered appropriate for the opposite gender (Ruble, Balaban, & Cooper, 1981). There may be little dispute that boys and girls have different social identities, but do these identities induce boys and girls to comply publicly with norms or ideals that they do not privately embrace? It appears that they do. Bacon and Lerner (1975) surveyed second-, fourth-, and sixth-grade girls about their mother's occupational status, their beliefs about women's employment opportunities, and their own occupational aspirations. They found that daughters of working mothers held quite egalitarian beliefs about the kinds of jobs men and women can do, and this egalitarianism increased with age. Yet, the girls uniformly indicated that they themselves would still choose traditional female occupations. One of our own studies (Prentice, Girgus, & Miller, 1997) pursued the link between public conformity to prototypical gender norms and pluralistic ignorance among elementary school children. In this study, we asked fourth- and fifthgrade boys and girls to indicate their own beliefs about whether boys, girls, or both boys and girls like each of 30 toys and activities. Next, we asked them to indicate how they thought "other kids" would answer the same questions. Consistent with the prediction that children's beliefs would be characterized by pluralistic ignorance, we found that they expressed more egalitarian views when responding for themselves than for other children. Finally, the mistaken impression that everyone supports the prototypical norms of a social category can arise even in the absence of widespread public conformity to those norms. It is often sufficient that those who do conform (even if the minority) receive disproportionate prominence. Indeed, the nature of group dynamics is such that those who embody a group's norms or ideals are given (or take) the spotlight and thus can provide a skewed view of the group. As an example, consider again the finding that prison inmates systematically underestimated the similarity of their attitudes to those of their peers, each assuming that his own position was more sympathetic toward the prison administration (Akers, 1977; Benaquisto & Freed, 1996; Kauffman, 1981; Toch & Klofas, 1984; Wheeler, 1961). Wheeler attributed this finding not to widespread public opposition to the prison authorities but to the fact that prison dynamics enabled a vocal minority of anti-administration prisoners to have a disproportionate degree of visibility within the prison. In his words, "much of the strength of the inmate culture may reside in the ability of anti-staff oriented inmates to attain positions of

< previous page

page_111

next page >

< previous page

page_112

next page > Page 112

high visibility within the inmate system, thereby generating and reinforcing the image of a culture marked in conflict with the values of the administration." (p. 291). In a similar vein, Klofas and Toch (1982) found that those guards and prisoners with the most hardline positions (subculture custodians, in their terms) were inclined to adopt self-appointed roles as spokesmen for their respective constituencies. Packard and Willower (1972) also acknowledged that the pluralistic ignorance they observed among school teachers may have been triggered by witnessing custodial behavior in places of high visibility within the school. The tendency of those who embody the norms of a group to find themselves in positions of high influence has been observed in less formal situations as well. For example, Newcomb, (1943) in his classic study of social life at Bennington College, noted that those students who espoused most vocally Bennington's tradition of liberalism tended to be popular and prominent in positions of leadership. Korte's (1972) study of social life at Vassar tied the greater prominence of cultural conformists to pluralistic ignorance more directly. Specifically, he attributed the finding that Vassar students perceived themselves to be less socially and politically liberal than the majority of their peers to the fact that those students who embodied the liberal values of the institutional norm tended to be particularly prominent and conspicuous on campus. The mistaken perception of personal deviance among the majority may arise initially because of the high visibility of those minority members who best embody the prototypical group norms. It may be perpetuated, however, by the widespread public conformity it induces. The illusion of personal deviance that begins with the misinterpretation of the representativeness of a vocal minority is thus perpetuated by a misinterpretation of the conformity of a formerly silent majority. Summary and Conclusions In recent years, students of the relation between attitudes and behaviors have contributed significantly to our understanding of the nature and conditions of attitude-behavior consistency. In the process, they have made a number of choices regarding the kinds of situations and behaviors they would study and the kind of consistency they would try to predict. They have chosen to focus on correlational consistency between attitudes and reasoned or planned actions, a strategy that seems to us quite sensible if the goal is to improve models of behavioral prediction. At the same time, this strategy has left a wide range of behaviors and many interesting questions about attitude-behavior consistency unexplored.

< previous page

page_112

next page >

< previous page

page_113

next page > Page 113

Our purpose in this chapter has been to examine the relation of attitudes to a very different type of behaviorthe behavior that is enacted in public settings. Public behavior is rarely reasoned or planned; instead it is spontaneous and negotiated. It is intended for public, rather than for private consumption and therefore is enacted always with an eye to how it will be viewed by others. As a result, public behavior is heavily influenced by social norms that dictate what is appropriate or inappropriate. These norms derive from the demands of social etiquette, the identities associated with social groups, or even just the behaviors enacted by peers. Their effect is to reduce variation in behavior across individuals and thereby to render private attitudes useless as behavioral predictors. Given the many forces that pull public behaviors away from private attitudes, it might seem puzzling, or even perverse, that we have chosen to focus on the relation between the two. But, in fact, our choice has been dictated not by a desire to assess or increase the consistency of these thoughts and deeds, but instead by an interest in the psychological and social consequences of their disjunction. These consequences turn out to be considerable, for social perceivers are naive consistency theorists who assume that public actions offer a valid reflection of private views. When, as is often the case, they do not, the result is pluralistic ignorance, a self-perpetuating cycle of erroneous inference and social misunderstanding. We have analyzed a number of cases of pluralistic ignorance in an attempt to illustrate how it functions as both a cause and a consequence of literal inconsistency between private attitudes and public behaviors. Despite our focus on attitude-behavior inconsistency, the implications of our analysis for efforts to improve behavioral prediction and modification are quite heartening. Cases of pluralistic ignorance provide a fundamental insight into the nature of social life: Its private and public components are largely disjointed. Private experiences frequently occur without public acknowledgment, and public actions often belie private views. As a result, attempts to change public behaviors by changing private attitudes will not be effective unless some effort is also made to bridge the boundary between the public and the private. Only when individuals' private conversions are brought into public awarenessonly when they realize that their new attitudes are shared by their peerswill they feel at liberty to act on them.

< previous page

page_113

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_115

next page > Page 115

7 Where Does the Behavior Come From in Attitude-Behavior Relations? Toward a Connectionist Model of Behavior Generation Yoshihisa Kashima Virginia Lewis La Trobe University The relation between attitudes and behavior has been a central question in nearly three decades of attitude research. Triggered by Wicker's (1969) charge that empirical evidence for an attitude-behavior relation is weak at best, a flurry of research followed. Perhaps now social psychology can lay claim to some solutions to the ways attitudes guide behaviors (Eagly, 1992). With Wicker's pessimistic condemnation long forgotten, the attitude construct is still thriving as a central concept in social psychology, as Allport (1935) crowned it more than half a century ago. Yet, there is a neglected issue in the current literature on attitude-behavior relationswhere does the behavior come from? How do people generate behaviors when attitudes guide their enactment in some fashion? We explicate this question and describe a solution drawing on the emerging literature on connectionist approaches to social psychology. Where Does the Behavior Come From? The major approaches to the question of attitude-behavior relations, one based on the theory of reasoned action (TRA; Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975) and the other seeking to delineate conditions un-

< previous page

page_115

next page >

< previous page

page_116

next page > Page 116

der which attitudes guide behavior (Sherman & Fazio, 1983), failed to shed light on the question of where the behavior comes from in any attitude-behavior relation. The reasoned action approach takes a decision theoretical perspective on attitude-behavior relations. The conceptual core of this approach is the claim that an intention is the central determinant of a behavior; through this construct one's attitude toward performing the behavior (attitude toward behavior) indirectly influences the behavior, although other variables contribute to the prediction of behavior, such as past behavior (Bentler & Speckart, 1979) and behavioral control (Ajzen, 1991). Theoretically, only with regard to some particular behavior (or its cognitive representation) can a behavioral intention be formed. Therefore, the particular behavioral options must be generated first. Fishbein and Ajzen (1975) apparently considered this issue and suggested that potential behavioral responses may be generated in a particular behavioral situation. Out of the generated responses, one response may be chosen through comparisons of attitudes toward performing those possibilities (Davidson & Morrison, 1983; Jaccard, 1981). Still, the theory does not extend far enough to consider the process of generating behavioral options. Methodologically too, the generation of behavioral options does not arise as a question in the research strategy typically adopted in this tradition. Research often begins with the identification of a specific behavior that researchers wish to predict. The behavior under examination is fixed by design. Furthermore, the research design itself may increase the probability that the appropriate behavioral option is generated at the right time. Research participants are typically asked to state their intentions to perform a specific behavior and are examined according to this and to other variables that are deemed relevant within the researchers' theoretical framework. This questioning of behavioral intentions itself may make the behavioral options more memorable and hence increase the probability of generating the behavioral options later on. In this connection, it is interesting to point out that Sherman (1980) and Greenwald, Carnot, Beach, and Young (1987) showed that the very act of asking people whether they intend to act in a given manner increased the probability of that action. The second approach seeks to find variables that moderate attitude-behavior relations. Attitude accessibility (Fazio, 1986) emerged as a central moderator variable in this approach (Eagly, 1992). Fazio endorsed Fishbein and Ajzen's (1975) definition of attitudes as abstract evaluations of objects. According to Fazio's (1986) early formulation, attitude guides behavior when the encounter with an attitude object activates the attitude, which then affects the immediate perception of the attitude object. That is, if one's attitude is positive, positive aspects of the object are selectively perceived, and if one's attitude is negative, negative aspects are attended to. Thus, the

< previous page

page_116

next page >

< previous page

page_117

next page > Page 117

activated attitude "colors" the perception of the attitude object. Fazio contended that when normative constraints do not prevent people from behaviorally expressing their attitudes, the perception of the attitude object dictates the definition of the event. When the event is defined primarily positively due to the activated attitude and subsequent selective perception, approach behavior ensues; when the event is defined negatively, avoidance behavior ensues. Research strategies used by Fazio and his colleagues to test this approach often constrain the situation to such an extent that the generation of behavioral options is not an issue. For instance, Fazio and his colleagues (1983) tested his attitude-accessibility model using puzzles. Participants were given the opportunity to play with the puzzles directly or told to learn indirectly about them. The direct interaction experience, relative to the indirect experience condition, increased attitude-accessibility as measured by reaction time and improved the correlation between attitudes and actual choice of puzzles when the participants were given a chance to play freely with them. In this paradigm, Fazio's discussion of approach and avoidance behaviors makes good sense. The experimental condition was such that there was no need to generate behavioral options. Options were fixed as a design constraint. Most empirical tests of Fazio's model have been conducted under similar conditions (e.g., Doll & Ajzen, 1992). Both major theoretical approaches, thus, are silent on the question of how behavioral options are generated. Consequently, so is Fazio's (1990; Sanbonmatsu & Fazio, 1990a) more recent attempt at integrating the two approaches into a coherent framework. According to the motivation opportunity determinants (MODE) model, the behavior intention and the attitude-accessibility approaches describe two different modes of attitude-behavior relations. The latter describes an effortless, automatic process in which automatically activated attitudes guide behaviors. The behavioral intention approach characterizes an effortful, controlled process that occurs only when the actor has enough cognitive resources and is sufficiently motivated to deliberate on costs and benefits of performing a behavior. This theory attempts to specify conditions under which the attitude activation or the intention formation process drives attitudebehavior relations. Yet, again, the question of how behaviors are generated falls outside the scope of the theory. The generation of behavioral options should be a significant part of a theory of attitude-behavior relations. Consider the following example. In Australia, at the start of the acquired immuno-deficiency syndrome (AIDS) epidemic, a public service advertisement was aired on television. The advertisement depicted AIDS as the Grim Reaper, who was playing a bowling game with 10 humans as pins on a bowling alley. The humans included men,

< previous page

page_117

next page >

< previous page

page_118

next page > Page 118

women, children, and elderly of all colors and shapes. The message was that AIDS is a deadly disease that kills indiscriminately. This campaign, and similar others focusing on drunk driving and seat belt use, are designed to produce attitudes appropriate for public safety and welfare. Yet, these campaigns often fail to tell the public what to do. For instance, the Grim Reaper advertisement might have created negative attitudes toward AIDS successfully, but it did not clearly inform viewers about how to deal with AIDS. To frame the issue in our terminology, the advertisement created attitudes, but it failed to let the viewers know which behaviors to generate to prevent AIDS. With an increasing amount of economic resources spent on public campaigns of all sorts, there may be more cases in which public attitudes are created and shared, and yet public behaviors do not necessarily follow because many people do not know how to behave in accordance with their attitudes. Recently, Posavac, Sanbonmatsu, and Fazio (1997) reported a series of studies that underline the importance of the generation of choice alternatives in attitude-behavior relations. A theory of attitude-behavior relations whose scope includes the generation of behavioral options would enhance the effectiveness of public attitude campaigns aimed at behavior changes. If a unified theory is premature, at least a theory of behavior generation may need to be constructed. Toward a Theory of Behavior Generation We propose a preliminary model of behavior generation based on a connectionist theory. We first describe the basic assumptions embodied in the model, and then review the theoretical underpinnings of the model by drawing on the literatures on attitudes, social norms, and self-categorization theory. Finally, we review some empirical studies that bear on the present model's predictions. A Connectionist Model of Behavior Generation The model describes the process by which mental representations of behaviors are acquired and retrieved within the framework of a connectionist theory called the tensor product model (TPM). Connectionism is an approach to cognitive modeling that was inspired by the brain structure. According to its tenet, multiple processing units, which singly possess a limited information-processing capacity, can collectively perform some surprising cognitive tasks (Hinton & Anderson, 1981; Rumelhart, McClelland, & PDP Research Group, 1986). Recently, some social psychologists (Kashima &

< previous page

page_118

next page >

< previous page

page_119

next page > Page 119

Kerekes, 1994; Kunda & Thagard, 1996; Read & Marcus-Newall, 1993; Shultz & Lepper, 1996) began to use this framework to theorize about traditional social psychological phenomena (for general reviews, see Read, Vanman, & Miller, 1997; Smith, 1996). We use the TPM (Humphreys, Bain, & Pike, 1989; Kashima, Woolcock, & King, 1998; Pike, 1984; Smolensky, 1990) for this purposeit is well suited to model cognitive processes associated with context effects in a variety of cognitive tasks ranging from memory and category learning to natural language processing. Major advantages of the model are that it permits the modeling of the learning of configural associations among various aspects of a concept in a context and that it can be used to integrate a variety of theoretical insights and assumptions within a coherent framework. We briefly describe the model here, but a more detailed treatment appeared in Kashima et al. (in press). Let us assume that a perceiver represents social events as a configuration of a focal event and the context in which it has taken place. The focal event representation consists of three aspects: actor, behavior, and object. In other words, we assume that the cognitive representation of a social event consists of a configural association among the three aspects of the focal event and the context, that is, who did what to whom (or what) where (or when). It is important to note that this scheme treats the representations of the self, other individuals, and groups all as actors with varying degrees of similarity. The TPM specifies the mechanism by which such configural representations of contextualized events are learned. More specifically, we assume that the aspects and context are represented by separate sets of simple processing units. In our case, we assume that four separate sets of a large number of processing units represent the three aspects and the context: actor, behavior, object, and context (see Fig. 7.1). Each unit is assumed to take an activation value ranging from negative infinity to positive infinity. In each set, the units can take on different levels of activation, thus forming a pattern. One fundamental assumption is that the pattern of activation, rather than the magnitude, signifies a meaningful notion. We describe general principles and the most significant implications of the model here and provide mathematical details in the appendix. The model assumes that a social event is encoded as a configuration of actor-behavior-object-context. Each encoded event is stored in the form of a change in connection strengths among the processing units. The TPM is so called because a mathematical entity called a tensor is used to represent the connection strengths in the network. The model predicts that mental representations of behaviors generated at a given time depend on the social situation defined by the configuration of actor, object, and context that demands action.

< previous page

page_119

next page >

< previous page

page_120

next page > Page 120

FIG. 7.1 A tensor product network with separate sets of units for actor, behavior, object, and context. a[i], b[j], o[k], and x[l] represent the levels of activation of the corresponding units in the actor, behavior, object, and context sets. M[i,j,k,l] stand for the connection strength among those units. To express this more precisely, let us characterize the retrieved behavior at time t, b'(t), as a function of the previously remembered behavior at time t, b(t). where s(t,t) is the similarity between the actor-object-context configuration at time t and the actor-object-context configuration of time t. Now note that

< previous page

page_120

next page >

< previous page

page_121

next page > Page 121

where sa(t,t), s0(t,t), and sx(t,t) indicate the similarities of the current actor, object, and context representations with the actor, object, and context representations at time t. This model implies that behavior representations emerge as possible action alternatives. The model (Equations 1 and 2 combined) is generally consistent with Smith and Zárate's (1992) exemplar theory of social judgment (see Kashima et al., in press). The model predicts that, other things being equal, generated behaviors in a given situation are likely to be similar to behaviors that have been learned in similar situations in the past. Behavior representations associated with similar past situations are most prominent, or most accessible, according to this model. Two points need to be emphasized. First, generated behaviors are mental representations of behaviors, rather than motor behaviors per se. Second, the situation in this model is defined as a configuration of actor, object, and context, and the similarity of the current situation with a past situation depends on the current and past configural representations of the actor-object-context. We discuss these issues in turn. Mental Representations of Behaviors Behavior in this context is not a sequence of observable motor movements of a human body but rather its mental representation. Most social behaviors are not simply a sequence of motor behaviors. They are typically meaningful actions, that can be identified by some commonsense description. We call those meaningful representations of motor behaviors behavioral options or mental representations of behaviors. Although the concept of a mental representation of behavior does not figure prominently in attitude-behavior research, the idea often is assumed implicitly. The theory of reasoned action and its variants, for instance, assume that people can imagine performing some behaviors, describe those behaviors by symbolic means (i.e., language), and report their intentions to perform those behaviors. Triandis (1980) explicitly defined those culturally meaningful behaviors as acts. Within a larger context, however, psychologists have examined the structure of cognitive representations of behaviors in terms of such constructs as scripts (Abelson, 1981; Schank & Abelson, 1977). Norman (1981) explicitly theorized the process of action control by a mental representation of behavior. According to him, a cognitive representation of behavior is a hierarchical organization of schemas. A schema in his theory is a sensorimotor knowledge structure equipped with a triggering mechanism, that activates the structure when an appropriate condition is satisfied. At the highest level, a parent schema such as ''driving home'' may be present, and at a lower level, there may be child schemas such as "avoiding obsta-

< previous page

page_121

next page >

< previous page

page_122

next page > Page 122

cles," "maintaining speed," "braking properly," "following the correct route,'' and so on. A child schema may then act as a parent schema for even lower levels of behaviors. Norman called the organized structure an activation trigger schema (ATS) system. According to Norman (1981), for the performance of skilled action such as "driving home," people activate by intention the top level parent schema, and lower level schemas are also activated to control appropriate bodily movements in a coordinated fashion, in accordance with previous action, environment, and perceptual input. An implication of the hierarchical organization of a mental representation of behavior is that a given action sequence can be described in more than one way (as noted by philosophers of action long before; e.g., Danto, 1963). So, the act of "moving a finger while standing in front of a door" may be described as "pushing a doorbell," which in turn may be described as "seeing if someone is home." Wegner, Vallacher, and their colleagues (Vallacher & Wegner, 1987; Wegner & Vallacher, 1986) proposed action identification theory, which can predict the level at which an actor chooses to describe or identify his or her behavior within a hierarchy of action descriptions. The theory postulates three principles. First, the actor's description of action corresponds to the behavior that the actor intends to perform. This level appears to correspond to the highest level of action schema that the actor intentionally activates within Norman's (1981) ATS system. Second, actors prefer to identify behavior at a higher level if both higher and lower level descriptions are available. That is, if the act of "pushing a door bell" can be understood as either "moving a finger" (lower level) or "seeing if someone is home" (higher level), actors prefer to think of it in the latter terms. Third, if a current level of action identification cannot maintain the course of action (that is, if it is too hard to do), actors tend to move down to a lower level of action identification. In other words, if the man who tries to push the doorbell realizes he cannot do so, he will move to the lower level of "moving a finger'' to perform his behavior. Optimally, the actor should form an intention to perform a behavior at the level that enables him or her to complete the action most effectively. When induced to form an intention at a suboptimal level, the resultant action should be less effective. This implication was tested by Vallacher, Wegner, and Somoza (1989). Participants in this experiment were told to deliver a prepared speech advocating a tougher university policy on cheating. The audience was said to be a group of students who agreed with the advocated position (easy action) or a group who opposed a tougher policy (difficult action). The level of action identification was manipulated by instructing the participants to be mindful of the manner of their speech such as articulation and enunciation (low level) or to try to be persuasive (high

< previous page

page_122

next page >

< previous page

page_123

next page > Page 123

level). According to the theory, action identification at a lower level is more effective for a difficult action, whereas action identification at a higher level is more effective for an easy action. The participants' communication effectiveness, which was measured by such objective and subjective variables as speech fluency and satisfaction with the speech, was found to be consistent with the prediction. Generally, a cognitive representation of a behavior appears to be a useful concept in theorizing about the psychological control of social action. Put simply, people mentally represent a behavior, and the representation that is used to control one's action makes a difference in observed social behavior, as Vallacher et al.'s (1989) study clearly showed. These theories specify which mental representation is used to control action once a type of action to be performed is generated and determined. The TPM attempts to describe the process by which mental representations of behaviors are generated. Actor, Object, and Context The tensor product model predicts that behavior representations are generated as a function of the current representation of the actor, the object to which a behavior is directed, and the context in which the behavior is enacted. These assumptions embody theoretical insights of the research areas of attitudes, societal norms, and intergroup relationships. Object Representation and Attitudes. According to the tensor product model, when an object in the current situation is encoded, it acts as part of the current configuration of actor-object-context. Equations 1 and 2 suggest that behavior representations associated with similar actor-object-context configurations in the past should be generated in response to the object. This implication of the model is closely related to tripartite models of attitudes (e.g., Katz & Stotland, 1959; for recent explorations, see Breckler, 1984, and Eagly & Chaiken, 1993). According to this general class of models, an attitude toward an object has an internal structure whose facets are represented by three partscognition, affect, and conation. Conation is often loosely understood as motivation or behavior. In one of the clearest renditions of this basic idea, Greenwald (1968) suggested that these components may derive from different learning processes. Affect, cognition, and behavior are said to result from classical conditioning, information learning, and operant conditioning, respectively. Tripartite models imply that behavioral responses may be produced when attitudes are activated in response to those objects with which the attitudes are concerned. To put it differently, the activation of the attitude is a process mediating the recogni-

< previous page

page_123

next page >

< previous page

page_124

next page > Page 124

tion of the attitude object and the subsequent generation of behaviors toward that object. The object in a social context then triggers behaviors via attitudes. Context Representation and Societal Norms. The TPM implies that the current context can also trigger the generation of behavior representations. Encoding the current context should retrieve behavior representations associated with the past representations of similar contexts. This feature of the model embodies the notion of societal norms. Societal norms imply the sense of "ought" that prescribes or proscribes behavior, and the boundedness of its applicability to some context. Take the following two definitions for example: "Any standard or rule that states what human beings should or should not think, say, or do under given circumstances" (Blake & Davis, 1964, p. 456). "A statement made by a number of members of a group, not necessarily by all of them, that the members ought to behave in a certain way in certain circumstances" (Homans, 1961, p. 46). Both definitions make reference to ''should or should not" or "ought.'' They also specify that norms dictate behaviors under certain circumstances that are meaningfully defined within a society (e.g., cocktail party, football match). For instance, it is a societal norm to "talk sociably with people at a cocktail party," or to "shout and cheer at a football match." Societal norms, therefore, are distinguished from attitudes in terms of their presumed injunctive coercive power (rather than evaluation) and their context dependence (rather than object dependence). The norm concept can shed some light on the process of behavior generation. Once a norm is activated, it should then activate a cognitive representation of the behavior with which the norm is concerned. In the process of behavior generation, then, the activation of a societal norm may act as a mediating process just like the activation of an attitude can. However, what activates societal norms is not an object in a context. Rather, it is the context itself. Consistent with this contention, Cantor, Mischel, and Schwartz (1982) found that a variety of social contexts (e.g., cocktail party) activated the cognitive representation of behaviors appropriate for the contexts (e.g., talking). A brief review of the literature gives some support to the idea that societal norms may contribute to the generation of behavior representations. Like attitudes, the concept of societal norm has a checkered history despite its centrality to the social sciences. Early functionalists (e.g., Durkheim, 1982; Radcliffe-Brown, 1952) introduced norms as a central concept in explaining the maintenance of social order, and, like Parsons (1949), many took for granted the idea that internalized societal norms motivate people to conform to social standards. Yet, this assumption was challenged on both theo-

< previous page

page_124

next page >

< previous page

page_125

next page > Page 125

retical and empirical grounds. On the one hand, as charged by Wrong (1961), the conception of the person as unproblematically driven by norms overemphasizes the social nature of humans. On the other hand, empirical research often finds that prescriptive or proscriptive norms are not predictive of social action (e.g., Cancian, 1975). That is, people do not always do what they are supposed to do. A widespread suspicion ensued regarding the conceptual utility of societal norms (e.g., Garfinkel, 1967; Gibbs, 1981). Nonetheless, recent work in social psychology shows that societal norms sometimes predict behaviors. Let us start with the TRA. According to TRA, behavioral intention not only is influenced by attitude toward the act but also is influenced by subjective norms. Subjective norms are defined as significant others' injunctions that the actor perform or not perform the behavior. In addition, two other types of norms have been proposed. One is the personal norm, in which the actor makes an injunction that he or she perform the behavior. The other is the behavioral norm (Grube, Morgan, & McGree, 1986), which is what significant others do, rather than what they say the actor should do. Although subjective norms are often said to contribute little to the prediction of intentions, as Kashima and Kashima (1988) and Trafimow and Finlay (1996) showed, the importance of subjective norm in intention formation varies across individuals. In addition, when three types of norms (behavioral, personal, and subjective norms) are combined, their contribution to behavioral intentions is substantial (see Kashima & Gallois, 1993, for a review). This implies that societal norms of various kinds are at least cognitively associated with behavior representations as measured by behavior intentions. Cialdini and his colleagues' focus theory (Cialdini, Kallgren, & Reno, 1991; Cialdini, Reno, & Kallgren, 1990) adds another dimension to this line of reasoning. In a series of elegant experiments, Cialdini and his colleagues showed that the descriptive norm, "people do not litter in this context," when made salient, decreased the probability of littering in that context. In addition, the injunctive norm of "one should not litter" can influence behavior when it is made accessible, for instance, by increasing the salience of related concepts such as "one should recycle." Presumably, then, an activated norm can generate a behavior representation of recycling, and then, affect one's intention. All in all, the literature supports the proposition that societal norms contribute to the generation of behaviors. Actor representation and self-categorization. The TPM also embodies the insight of self-categorization theory. According to TPM, when the actor, who is the self, is represented as a member of a social category, this current actor representation can act as a cue to retrieve behavior representations that are associated with the representations of the social category.

< previous page

page_125

next page >

< previous page

page_126

next page > Page 126

This implication is consistent with self-categorization theory. Turner, Hogg, and their colleagues (Hogg & Turner, 1987a; Turner, 1991; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) argued that self-concepts are conceptualized as categorizations of the self into one class in contrast to another class. The self-categories are thought to differ in their levels of inclusiveness, so that they may go from the most inclusive (i.e., human being) to the least inclusive (i.e., individual person). When the social context is such that the self is categorized into a group in contrast to other groups, this leads to a depersonalization of the self, resulting in the adoption of the group norm as one's own. The group norm in self-categorization theory is the prototypical behavior of the group to which the self is categorized. The relevance of this theory to behavior generation is clear. When one categorizes oneself into a group, the group's prototypical behavior is generated as one's own behavioral option. There are two points to note. First, what triggers the behavior generation process is the salience of one's group membership, rather than the object to which, or the context in which, a behavior is to be executed. Clearly, the context influences the type of self-categorization; however, theoretical emphasis is placed on self-categorization, rather than on the context that affects it. Second, a behavior that is prototypical of the group is generated as a behavioral option. Yet, the prototypicality of a behavior of one's ingroup depends on the outgroups to which the ingroup is contrasted. Therefore, different ingroup norms may emerge as a function of which outgroup is salient in the context. Implications of the Behavior Generation Model The TPM of behavior generation is yet to be fully tested. However, the model sheds new light on some of the existing empirical findings on attitude-behavior relations, which go beyond its existing theories. One straightforward implication of the TPM is an effect of past behaviors on the generations of behavior options (Bentler & Speckart, 1979; for some recent examples, see Kashima, Gallois, & McCamish, 1993). It is easy to see that the past behaviors that the self performed are likely to be generated again if the current self-object-context representation is similar to the past self-object-context representations that are associated with those past behaviors. Bentler and Speckart's (1979) findings that past behaviors predicted future behaviors are easily accommodated in the TPM framework. Yet, the current theory does not predict that all past behaviors are predictive of actual behavior. It is noteworthy that behavior options that are similar to past behaviors are generated but not necessarily acted out. This is because there may be a decision making process in which the generated behavior option may be considered and chosen to be executed or not. If the generated

< previous page

page_126

next page >

< previous page

page_127

next page > Page 127

behavior option is not selected for action, past behaviors would not predict behaviors. Second, the TPM sheds light on the effects of "direct experience" on attitude-behavior relations. Fazio and his colleagues (Fazio et al., 1983) showed that people who expressed their liking of a puzzle on the basis of their direct experience of playing with the puzzle tended to behave toward the puzzles in the manner congruent with their expressed evaluations. However, the relation between the expressed evaluations and the behaviors tended to be weak when people did not have a direct experience with the puzzles. Although Fazio and his colleagues explained this finding in terms of attitude accessibility, Doll and Ajzen's (1992) recent findings cast some doubt on this interpretation. In their experiment, participants either did or did not play with computer games before expressing their evaluation and intention to play with the games. They measured both the accessibility of the participants' evaluations and the stability of their expressed intentions, and examined the extent to which these variables mediated the attitude-behavior consistency. Doll and Ajzen showed that stability of intentions, not attitude accessibility, mediated the attitude-behavior consistency. From the TPM perspective, Doll and Ajzen's (1992) finding can be interpreted as follows. When people actually perform behaviors in relation to some object in a particular context (direct experience condition), this leads to the formation of a configural representation of the self performing behaviors toward objects in the context. In the indirect experience condition, it is not the self who performs the behaviors but someone else. Then this indirect experience is represented as a nonself performing behaviors toward the objects in the context. The representations in the direct and indirect experience conditions differ at least with respect to the actor representation and are likely to differ in the representations of all the other aspects as well. Later on, when the participants perform behaviors freely, they are likely to use as a cue the self-object-context configuration to generate behaviors. This cue and the representation of the experience is likely more similar in the direct experience condition than in the indirect experience condition. Let us assume that the expressed intentions reflect the generated behaviors. It follows then that the intentions are likely to be stable in the direct experience condition because the same cue and experience representations are used in both occasions. As well, for the same reason, the same behavior representation used for the report of the intentions is likely to be generated when the participants freely performed their behaviors. Third, the TPM sheds light on Terry and Hogg's (1996) recent finding about the effects of group norms on behavior. In two studies, they examined the effect of group identification and perceived group norm (how much the group members would agree that performing a certain behavior is good) on

< previous page

page_127

next page >

< previous page

page_128

next page > Page 128

the intention to perform the behavior. The behaviors of interest were exercising three times a week and engaging in sun-protective behaviors every time participants stayed out in the sun for more than ten minutes. Terry and Hogg predicted that people intend to perform a given behavior when the group norm was in favor of it and when they identified with the group. They found evidence for this interaction effect in both studies. Their finding can also be explained from the TPM perspective. When the object and context are specified, the generated behaviors depend on (a) the similarity of the current self-representation to the representations of actors stored in the memory system and (b) the type of behaviors performed by these actors. To put it differently, other things being equal, if the person knows that a group of actors performs a certain type of behaviors and that the person's current selfrepresentation is similar to those of the group of actors, then, the behaviors that are typically performed by the group of the actors are likely to be generated. The Terry-Hogg (1996) findings are in keeping with the theoretical prediction of the TPM if we can assume that the behavior that is generated receives a high level of behavior intention and the group identification is in this case related to the similarity between the current self-representation and the representations of the group and its members. Summary The TPM of behavior generation embodies the insights derived from the theories of attitudes, norms, and selfcategorization. This connectionist model specifies the mechanism by which the mental representations of social events are stored and retrieved to aid the generation of the mental representations of behaviors. The model assumes that a social event is encoded as a configural representation of the actor-behavior-object-context. These configural representations of events about one's own, other individuals' and groups' behaviors are stored in a connectionist network by superimposing new representations on top of old ones. When action is called for, the connectionist memory system is accessed by the current representation of the self, the object toward which a behavior is directed, and the context in which the behavior is performed. This retrieval operation generates behavior representations that are associated with the past social events whose actor-object-context configuration most closely matches that of the current configuration. We showed that the TPM can shed new light on some of the findings of attitude-behavior relations that go beyond the existing theories of attitude-behavior relations.

< previous page

page_128

next page >

< previous page

page_129

next page > Page 129

Conclusions Social psychological inquiries into attitude-behavior relations have come a long way. The pessimism of the early 1970s gave way to a decision theoretical reformulation of the question by the TRA and its variations and a process-oriented theory of attitude accessibility. Yet, neither formulation asked the question of where the behavior comes from in attitude-behavior relations. The behavior generation is an issue fundamental to any attempts at understanding behaviors. For an intention to perform a behavior to be formed, that behavior must come to one's mind. For an attitude to be used to decide whether a behavior is to be performed, that behavior must be generated to be selected. The behavior generation is an issue conceptually prior to the decision making. It should come as no surprise that Doob (1947, p. 136) referenced the behavior generation process in his behaviorist approach to attitudes. In fact, his definition of attitude is almost what we have called behavior representations: "An attitude is an implicit response which is both (a) anticipatory and (b) mediating in reference to patterns of overt responses [emphasis added], which is evoked (a) by a variety of stimulus patterns (b) as a result of previous learning or of gradients of generalization and discrimination, which is itself cue- and drive-producing, and which is considered socially significant in the individual's society." Yet, our formulation is different from Doob's in that we used a connectionist model of mental processes in modeling the behavior generation. In Doob's formulation, it is unclear how responses are generated, whether serially one by one or in parallel processes. By contrast, we postulated parallel processes in which a composite behavior representation is generated. Clearly, the current formulation requires further specification and development. Most of all, the theory needs to be tested empirically. Perhaps our formulation should be taken as a call to pay closer attention to action. The inquiry into attitude-behavior relations is one of very few areas in contemporary social psychology, if any, in which the prediction of action is an explicit aim. A strongly action-oriented approach to the issue may be needed. Miller, Galanter, and Pribram (1960) in their Plans and the Structure of Behavior, criticized Tolman's cognitive map approach to learning (p. 9): It is so transparently clear to [cognitive theorists] that if a hungry rat knows where to find food . . . he will go there and eat. What more is there to explain? The answer, of course, is that a great deal is left to be explained. The gap from knowledge to action looks smaller than the gap from stimulus to actionyet the gap is still there, still indefinitely large. Tolman . . . leaps over that gap when he infers the rat's cognitive organization from its behavior. But that leaves still outstanding the question of the rat's ability to leap it.

< previous page

page_129

next page >

< previous page

page_130

next page > Page 130

Their criticism may be as applicable now as four decades ago. Appendix This appendix provides some mathematical details for the discussion of the tensor product model. Notations Each unit in the network described in Fig. 7.1 is assumed to take an activation value ranging from negative infinity to positive infinity. Within each set, the units can take on different levels of activation, thus forming a pattern. One fundamental assumption is that the pattern of activation, rather than the magnitude, signifies a meaningful notion. Note that a pattern of activation values can be mathematically represented by a vector, assuming that each set has N units and the activation level of the ith unit in each set is designated as the ith element in the vector. We use the following convention in this chapter. A pattern in the set for actor, behavior, object, and context is respectively designated as a, b, o, and x. Note that the ith element in each vector is designated as a[i]. For simplicity, we assume that the dot product of each vector with itself is unity. Note that the dot product of a vector, a, with another vector, a', is defined as [a.a'] = Sa[i]a'[i]. It is assumed that the units within the same set are not connected with one another, but the units among different sets are fully connected. The strength of connection at time t among the ith unit in the actor set, the jth unit in the behavior set, the kth unit in the object set, and the lth unit in the context set is designated as M(t) [i,j,k,l]. Let us define a mathematical entity called a tensor, which can be thought of as a multidimensional array. The tensor, M(t), is then defined as a four dimensional array whose element at the [i,j,k,l] coordinate is M(t) [i,j,k,l] Note that this is a straightforward generalization of a matrix. A matrix is a two dimensional array whose element can be specified by the coordinate [i,j]. Note that M(t) can be used to describe the state of connection strengths of all the connection points at time t. Encoding and Learning Process When a social event is encoded, its configural representation is formed and remembered. Suppose that the aspects of actor, behavior, and object, and the context of the event are represented by the patterns of activation, a(t), b(t), o(t), and x(t), with the activation levels, a(t)[i], b(t)[j], o(t)[k], and x(t)[l], respectively, for the ith unit in the actor set, the jth unit in the behavior set, the kth unit in the object set, and the lth unit in the context set. These activations then spread from every unit, and come to the connection

< previous page

page_130

next page >

< previous page

page_131

next page > Page 131

point of these units. For instance, the activations, a(t)[i], b(t)[j], o(t)[k], and x(t)[l], spread to the connection point for the ith unit in the actor set, the jth unit in the behavior set, the kth unit in the object set, and the lth unit in the context set. According to the TPM, the strength of connection among these units increases by the product of these activations. In other words, M(t+ 1)[i,j,k,l] = M(t) [i,j,k,l] + a(t)[i]b(t)[j]o(t)[k]x(t)[l]. We can simplify this learning equation by using the tensor notation. Let us define a tensor product, E(t) = a(t)b(t)o(t)x(t), as a four dimensional array, whose element at the [i,j,k,l] position is a(t)[i]b(t)[j]o(t)[k]x(t)[l]. Then, the learning equation is: Equation Al can be interpreted as follows. When a social event is encoded at time t, its configural representation, E(t), is formed and stored in memory, as if it is superimposed on top of the existing memory representation, M(t). The memory representation is automatically updated to include this event memory. The resultant memory representation is M(t + 1). Therefore, the memory representation at time t + 1 can be rewritten as follows:

Note that s is an integer from 1 to t, and that E(t) = a(t)b(t)o(t)x(t). In this model, time is treated as discrete. Modeling Behavior Generation Humphreys et al. (1989) defined a cognitive operation of retrieval, in which a cue is used to access the memory representation, and to retrieve whatever is associated with the cue. We assume that the process of behavior generation is basically the retrieval of the cognitive representations of behaviors using a variety of cues, such as the object, context, and self-categorization. We assume that a cue is represented as a configuration of actor, object, and context representations. Note that the aspect of behavior is missing. The cue representation may then be the actor, object, or context representations, s'(t), o'(t), and x'(t), or any combinations of these. In this case, the person who is generating a behavior is the self, and therefore, a self-representation is used as the actor part of the cue. Nonetheless, we do not assume that the self-representation remains constant. Rather, its activation pattern could vary across contexts though self representations are likely

< previous page

page_131

next page >

< previous page

page_132

next page > Page 132

to be similar to each other within certain bounds. When only one or two aspects are specified, for the other unspecified aspects, a "white noise" activation pattern is activated. The "white noise" activation pattern is represented by r whose element is all 1/ÖN (Humphreys et al., 1989), where N is the number of processing units in each set. Suppose that the retrieval cue contains a self-categorization at time t, s'(t), an object representation, o'(t), and a context representation, x'(t). These patterns are assumed to be activated in the actor, object, and context set of processing units. Then, the activations of s'(t)[i], o'(t)[j], and x'(t)[k] spread to their connection points. Take one of these connection points, the point connecting the ith unit in the actor set, the jth unit in the behavior set, the kth unit in the object set, and the lth unit in the context set. Recall that M(t + 1) [i,j,k,l] is the strength of this connection at time t + 1. The activations spreading to this connection point are all multiplied and then further multiplied by the connection strength, and that the amount of activation specified by this product spreads to the jth unit in the behavior set. Note that this takes place for all i, k, and l in parallel. The resultant activation in the jth unit in the behavior set, b'[j] is described by the following equation: The retrieved pattern is designated as b'(t) whose jth element is b'[j]. The retrieved behavior, b'(t), can be expressed as a function of the previously remembered behavior at time t, b(t). where s (t,t) is the similarity between the actor-object-context configuration at time t and the actor-object-context configuration of time t. Equation A4 is equivalent to Equation 1 in the text. Now note that where [_._] indicates the dot product as described before. This is equivalent to Equation 2 in the text. A proof is given here.

< previous page

page_132

next page >

< previous page

page_133

next page > Page 133

Let b(0) be the vector whose jth element is b(0) [j]. Also recall s(t,t) = [a(t).s'(t)] [o(t).o'(t)] [x'(t).x(t)] (Equation A5). By substituting Equation 5 into Equation A6 above, we obtain Equation A4 as follows:

< previous page

page_133

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_135

next page > Page 135

8 The Impact of Group Membership on Persuasion: Revisiting ''Who Says What to Whom With What Effect?'' Diane M. Mackie Sarah Queller University of California, Santa Barbara "Who says what to whom with what effect?" For decades of researchers, this question defined how persuasion was studied. Successful influence (effect) could be understood as the end product of an interaction of source variables (who), message variables (what), and audience variables (whom). Although many of the experiments prompted by this approach documented the effectiveness of each kind of variable, these experiments were perhaps most significant in demonstrating the importance of interactions between source and message variables on the one hand and between audience and message variables on the other. For example, high anxiety recipients are more persuaded by low fear appeals whereas low anxiety recipients are more persuaded by high fear appeals (Goldstein, 1959; Wheatley & Oshikawa, 1970). A credible source loses some perceived expertise if he or she presents weak and specious arguments (Petty, Cacioppo, & Goldman, 1981). Messages that are consistent with the recipient's opinions are more persuasive if they are one-sided, whereas counterattitudinal messages are more effective if they are two-sided (Lumsdaine & Janis, 1953). It is curious that the focus on interactions did not extend to interactions between source and audience. The main source features studied, credibility and attractiveness, tended to be treated as properties inherent in the source, even though the impotence of the audience in conferring those properties

< previous page

page_135

next page >

< previous page

page_136

next page > Page 136

was implicit in researchers' verification of them by pretesting with the relevant recipient population. Similarly, the focus on intelligence, gender, or personality traits asserted the effects of such audience variables as fixed and unchanging (Andersen & Clevenger, 1963). One clear exception was the study of similarity, by definition an interactive effect of source and audience. Unfortunately, its treatment was relatively unsophisticated, with the proliferation of lists of dimensions on which source and audience might be similar (Huston & Levinger, 1978; Simons, Berkowitz, & Moyer, 1970) first precluding any theoretical consideration of why or how it might be effective and then resulting in arguments that it operated "only" through expertise and liking (Berscheid, 1985; Byrne, 1969; Wagner, 1984; Woodside & Davenport, 1974). It is no wonder, then, that similarity was first seen as promising increased persuasive effectiveness, only to later be seen as having inconclusive effectssometimes helping and sometimes hurting the persuasive mission (e.g., Brock, 1965; Goethals & Nelson, 1973; Infante, 1978; King & Sereno, 1973; Leavitt & Kaigler-Evans, 1975; Mills & Kimble, 1973). Especially surprising in this limited and confusing literature is the lack of contact with theoretical concepts that were central to the very beginnings of the study of social influence, in particular that of shared group membership between source and audience. Overlooking the concept of reference groups, which encompasses both psychologically meaningful membership groups and groups in which membership is desired, seems particularly puzzling: Because of their psychological significance for the individual, reference groups have the potential to exert influence on the individual's beliefs and behavior. This shared group membership as a basis of influence was the explicit focus of early research but as more and more attention was paid to effective message variables, the role of social factors in persuasion was first inadvertently overlooked and then, as we discuss later, judged to be of less and less importance. In this chapter we return to the idea that shared group membership between source and recipient has an important impact on persuasion, and we explore some of the mechanisms by which such shared membership increases persuasive effectiveness. We use the general term group-mediated persuasion for situations in which a source tries to influence a recipient with whom group membership is shared. For our purposes it is the recipient's subjective perception of belonging to the same group as the source that is important, and category membership is defined as the subjective perception of the self as a member of a group (Tajfel & Turner, 1986; Turner, 1982). Our argument is that such shared membership triggers a variety of mechanisms that increase the persuasive effectiveness of appeals from ingroup members; thus our focus is on the processing impact of group membership. First, how-

< previous page

page_136

next page >

< previous page

page_137

next page > Page 137

ever, we review the importance of group membership in early conceptions of influence and attitude change, with their assumptions that effective persuasion required social relationships between source and audience. We then suggest some reasons why group-mediated change, and the social relations that made it possible, may have fallen from research favor. Primary among these is the idea that group-mediated persuasion does not represent the same kind of "true," "real," or ''valid" attitude change that careful analysis and acceptance of persuasive arguments alone brings about. That is, group-mediated persuasion is assumed to reflect a different process from that of argument-mediated persuasion. Fortunately, the recent development of paradigms and measures that assess the processes underlying effective persuasion allows evaluation of those assumptions. Thus, in the third part of the chapter, we describe a program of research designed to ascertain the processing strategies triggered by receiving a persuasive communication from an ingroup as opposed to an outgroup source. Finally, we draw some conclusions about the importance of interactions between the "who'' and the "whom" of persuasion and point to further research that needs to be done on this important topic. Early Perspectives on Group-Mediated Persuasion The early study of social influence focused on global questions such as, "How does the knowledge of other people's opinions affect our own opinions?" In his now classic studies on conformity, Sherif (1936) investigated this question with an innovative paradigm. He exposed participants to a small speck of light that was actually stationary but that appeared to move because of a perceptual illusion called the autokinetic effect. He then asked participants to report how much the spot of light moved. Compared to participants tested alone, participants who heard other people's responses tended to converge on a single response to describe the distance the light moved. This convergence persisted even after the participants were separated and tested in isolation once again. Thus, it appears that the expression of others' opinions can have a lasting influence on an individual, at least in a situation where physical reality is difficult to discern. Asch (1951, 1952, 1956) studied this question by presenting lines of varying lengths to a group of experimental participants and asking them to verbally identify which line was the same length as a target line. All but one of the participants were experimental confederates and, on designated trials, these confederates all gave the same, incorrect response. Under these conditions, the experimental participants went along with the group's incorrect responses on approximately one third of the trials. Because participants

< previous page

page_137

next page >

< previous page

page_138

next page > Page 138

were able to select the correct response in 99% of the trials when tested under isolated conditions, it seems that people can also be influenced by the opinions of others to change their responses from what they perceive to be observably true in their physical environment. Both of these studies exemplify an emphasis on the relationship between a group and a target of influence that is almost absent in current research on persuasion. As one might expect, these groundbreaking studies spawned further interest in the question of why an individual might go along with the group. Of particular relevance is Kelley's (1952) concept of reference groups. A reference group is a social group that is psychologically important to an individual and that, consequently, influences the individual's attitudes, beliefs, and behaviors. According to Kelley, reference groups can fulfill either a comparative function that helps the individual to evaluate his or her attributes and environment, or a normative function through which the individual aligns his or her own values and beliefs with those of the reference group. In Sherif's (1936) autokinetic experiment, the confederates acted as a reference group that served a comparative function and helped to provide definition to a task that was physically ambiguous. In the Asch (1956) experiment, some participants indicated that the unanimous responses of the confederates made them question their own perceptions, thus describing a comparative influence. But the majority of participants indicated that they did not want to "stand out" or be ridiculed by the group, thus describing a normative influence. Festinger (1950a, 1954) took a similar approach in describing two mechanisms that would lead to conformity with others' opinions. When physical reality is ambiguous, Festinger (like Kelley, 1952) suggested that the individual relies on appropriate reference groups to define a social reality. That is, the individual uses the reference group as a gauge for the appropriateness of his or her own responses and adjusts his or her responses to align them with those of the reference group. He further argued that people have a need to evaluate their abilities and opinions, and they can best do this by comparing to similar others. Thus, they have an attraction for groups of similar others, and they have some motivation for maintaining the homogeneity of these groups so that the groups can continue to be useful in defining social reality. The other mechanism Festinger proposed as leading to conformity is that of group locomotion. Group locomotion provides a pressure toward conformity when the achievement of a group goal can be facilitated by homogeneity within the group. Thus, to the extent that a group agrees on a goal and this goal is thought to be more attainable by providing a unified or homogeneous front, conformity within the group will increase. Because the pressure to attain the group goal is a defining feature of group locomotion, targets of influence may "change" their expressed beliefs and behaviors to go along

< previous page

page_138

next page >

< previous page

page_139

next page > Page 139

with the group without truly accepting and internalizing those beliefs. Group locomotion, then, offers a first example of how group influence has been relegated to a secondary status, in that it produces only a superficial change in beliefs that only comes about as a result of a desire to go along with the group. Other models propose similar mechanisms for social influence that depict group-based influence not as true persuasion, but as transitory compliance. For example, Deutsch and Gerard (1955) described two types of social influence. Normative influence is conformity to social pressure that is mediated by the expectation of punishment or reward by a group that has that kind of power over the individual. Of course, because certain rewards and punishments such as acceptance and rejection are only meaningful in a group about which the recipient of influence cares and with whom he or she identifies, reference groups can exert influence that nonreference groups are not capable of wielding. Informational influence, on the other hand, is motivated by a desire for accuracy and is accomplished when uncertainty leads to reliance on others for cues to objective reality. As in Festinger's (1950a, 1954) concept of social reality, similar sources may exert more informational influence than dissimilar sources, providing another route by which reference groups or same-category sources may have more influence than other sources. A more recent depiction of group-mediated influence as being less "real" than other forms of influence is embodied in Moscovici's (1980) minority influence work. Moscovici argued that influence exerted by a majority is likely to engender compliance based on potential reward and punishment of the individual by the majority. Thus, majority influence is much like Deutsch and Gerard's (1955) normative influence. In contrast, Moscovici argues that minority influence leads to persistent, internalized attitude change. Once again, the type of influence that is most associated with a group source is seen as producing less meaningful change than other forms of influence. A significant departure from the typical description of group-mediated influence as being less meaningful and less stable than other forms of influence is provided by Kelman (1958, 1961). He includes compliance as one form of influence that engenders transitory public acceptance and that is analogous to Deutsch and Gerard's (1955) normative influence. However, he uses two concepts to account for what Deutsch and Gerard call informational influence. Kelman's identification is a privately accepted change that arises from a liking for the source and a desire to identify with him or her. This type of influence is informational in nature, but is situation specific and transitory in that it typically does not outlast the term of the relationship. The other type of informational influence involves the thoughtful and

< previous page

page_139

next page >

< previous page

page_140

next page > Page 140

thorough integration of new material into a broad framework of cognitive associations and Kelman calls this type of influence internalization. Kelman conceptualizes internalization as a form of influence that is meaningful and enduring. In summary, Sherif's (1936) initial work on conformity spawned decades of investigation and theory regarding the mechanisms by which social others influence an individual's beliefs and behaviors. As has been discussed, the role of meaningful reference groups is central to this work. There has been a general tendency, however, to dichotomize accuracy-based influence and group-mediated influence: Influence based on accuracy was typically depicted as "true" change that is internalized by the individual and that endures over time, whereas group-mediated influence was seen as more superficial and transitory. Yet, as we have seen, even accuracy goals can be group-mediated in situations where the physical reality is ambiguous. In addition, Kelman's (1958) group-mediated identification should be of interest in persuasion research even if we restrict our studies to persuasion that clearly alters the belief system of the individual, regardless of whether it is enduring. Given such knowledge of the potential role of group-mediated influence, why is it that group affiliations between source and recipient have received so little attention in recent research on persuasion and attitude change? The next section addresses this question. Why Has a Group Focus Disappeared from Current Research on Persuasion? What might account for the dwindling interest in the impact of group membership on attitude change and persuasion? As social psychology turned more toward elucidating the cognitive mediation of various phenomena, interest was inevitably drawn away from some of the more social antecedents of those phenomena. In addition, however, interest in the social mediation of persuasion diminished because change brought about by groups was linked more and more to compliant or apparent change rather than the "true" or "real" change thought to be triggered by careful processing of persuasive arguments. Theories that considered much of group-mediated change to be mere compliance (Asch, 1951; Deutsch & Gerard, 1955; Festinger, 1953; Moscovici, 1980) may have had particular appeal and impact in North America, for example, because they fit well with cultural evaluation of the rugged individual as superior to the usually negative influences of the group. Some theories did allow for the possibility that group influence resulted in private rather than merely public change. Even in these cases, however, such influence was often accorded a status that was conceptually distinct

< previous page

page_140

next page >

< previous page

page_141

next page > Page 141

from that of influence based on careful and thoughtful consideration of presented information. For example, although lasting social influence can occur as a result of convergence toward a socially defined reality, this influence may be a result of simplistic recalibration to a group norm rather than careful evaluation of relevant information. Kelman's (1958, 1961) views illustrate this distinct status accorded group influence. As described earlier, Kelman distinguished between internalization and identification. Whereas internalization involves the incorporation of the advocated position into one's belief system in a way that integrates the new position with preexisting knowledge and beliefs, identification occurs because of a role-defining relationship between source and recipient. Identification, which of course is a form of group-mediated change, involves little effort and thought, and this distinguishes it from the more "real" change brought about by internalization. Thus, group-induced persuasion was traditionally seen either as resulting in context-specific public compliance, or in superficial or transitory attitude change. Group-mediated change was seen as the result of processes that were quite distinct from those thought to underlie attitude change brought about by careful consideration, evaluation, and acceptance of argumentation. Processing Perspectives on Group-Mediated Persuasion It is somewhat ironic that the rise of cognitive approaches to the study of persuasion has been matched by a decline in interest in group-mediated influence, because the cognitive approach offers new means of assessing the conceptual status of group-mediated attitude change. The cognitive approach has identified two well-established and distinct processes that produce attitude change (Chaiken, 1980, 1987; Petty & Cacioppo, 1981a, 1986a). What Kelman (1958, 1961) termed internalization, for example, is now commonly referred to as systematic (Chaiken, 1987) or central route (Petty & Cacioppo, 1986) processing. Such processing involves the active and intensive evaluation and integration of information, requires considerable motivation and cognitive capacity, and results in resistant and persistent attitude change. The presence of such processing is typically diagnosed from different reactions to manipulations of the quality of the arguments that comprise the persuasive message. Recipients systematically processing an appeal are likely to be more persuaded by strong and valid arguments than by weak and specious arguments. This differential persuasion is mediated by whether cognitive elaborations about the appeal are favorable or unfavorable: Strong arguments stimulate the generation of more favorable elaborations, whereas weak arguments stimulate the generation of more de-

< previous page

page_141

next page >

< previous page

page_142

next page > Page 142

rogatory elaborations. Thus, the presence of systematic processing should be reflected in a strong relationship between the extent of favorable elaboration and the extent of attitude change. Systematic processing can also be assessed via measures of generalization and longevity of attitude change and via the concomitantly reduced impact of factors other than message content on persuasion. Systematic processing can be contrasted with heuristic (Chaiken, 1987) or peripheral route processing (Petty & Cacioppo, 1986a), in which superficial aspects of the message, messenger, or message context are used as cues for adoption or rejection of the advocated position. For example, an attractive source or a lengthy appeal may be sufficient to result in message acceptance, whereas an unattractive source or a brief appeal may lead to message rejection. The use of such cues can replace extensive processing of message content as a means of assessing the advocated position if motivation or capacity is low. Within the framework of Petty and Cacioppo's (1981 a, 1986a) elaboration likelihood model (ELM), source characteristics such as shared group membership were initially described as operating via the peripheral route and thus as producing change because of mechanisms other than processing of available topic-relevant information. Because peripheral route processing involves mechanisms other than processing of topic-relevant information, the implication is that peripheral route processing involves less concern for the true validity of the advocated position than central route processing. Thus, Petty and Cacioppo's model suggests that group-mediated attitude change is tainted with a lack of objectivity. More recently, certain source characteristics, such as expertise, have also been described as informational items and as factors increasing message processing (Hass, 1981; Puckett, Petty, Cacioppo, & Fisher, 1983; see also Mills & Harvey, 1972). When source characteristics such as expertise and attractiveness have been studied in the context of Chaiken's heuristic systematic model (HSM), they have been shown to operate as heuristic cues (Chaiken, 1980, 1987; Pallak, 1983). Although heuristic cues are quick and easy to use, they are viewed as veridical indicators of message validity. Thus, heuristic processing does not imply the same lack of objectivity that peripheral route processing does. Category membership may operate as a heuristic cue either because of the attractiveness of the source to recipients (Kelman, 1958, 1961; Pallak, 1983) or because participants can rely on an ingroup member's views as appropriately defining reality (Chaiken, 1980; Festinger, 1950; Kelman, 1958, 1961). Although the impact of shared category membership has not been featured in either the HSM or the ELM, these approaches echo earlier views that groupmediated change may be rooted in less content-focused and less

< previous page

page_142

next page >

< previous page

page_143

next page > Page 143

extensive information processing than the attitude change brought about by the internalization of informational influence. There are, however, equally plausible motivational and cognitive reasons to believe that exposure to a communication from or about the ingroup could invoke intensive processing. First, a communication from a same-category source might encourage careful and thoughtful processing for the very reason that it is seen as reflecting, defining, and informing about social reality for people similar to the recipient (Festinger, 1950; Turner, 1982, 1987). Turner argued, for example, that the extent to which we think our views accurately reflect reality is a function of the extent to which similar others agree with those views. Thus, if accuracy is a goal, the message recipient should be particularly motivated to process information reflecting the views of ingroup members. Another reason that ingroup sources might invoke intensive message processing is that the ability to process category-relevant messages might be increased if category-relevant information is activated by the recognition of shared group membership (for examples of this effect in particular knowledge domains see Bargh & Thein, 1985; Markus, 1977). From this perspective, ingroup membership may act as a source characteristic that both motivates the recipient for systematic processing and helps increase the capacity available for systematic processing. Finally, exposure to ingroup views that differ from one's own views violates the expectation that our opinions are widely held (Heider, 1958; Ross, Greene, & House, 1977), especially by similar others (Allen & Wilder, 1977, 1978, 1979; Turner, 1982). Violation of such expectancies is likely to initiate active processing of relevant information (Hastie, 1984; Petty & Cacioppo, 1979a, 1979b; White & Carlston, 1983). For any or all of these reasons, same-category sources could bring about attitude change that is based on extensive processing of presented information as well as, or instead of, triggering reliance on group identity as a heuristic cue. Given the possibility that persuasive appeals by ingroup sources might lead to extensive processing of arguments rather than to compliance or superficial processing, and given the tools to assess whether or not extensive processing has occurred, we set out to investigate how source membership might moderate the processing of those appeals. Our purpose in the first experiment (Mackie, Gastardo-Conaco, & Skelly, 1992) was to gauge the impact of receiving an ingroup or outgroup message on the use of different processing strategies by creating a situation in which participants could react to ingroup or outgroup messages with either heuristic or systematic processing. Male and female University of California, Santa Barbara (UCSB) undergraduates were first asked their opinions on a variety of issues including the legalization of euthanasia and handgun possession. Responses indicated that they were generally in favor of legalization of euthanasia and

< previous page

page_143

next page >

< previous page

page_144

next page > Page 144

opposed to generalized handgun possession. In a purportedly unrelated task, participants were told that they would be asked their views on several issues as part of an ongoing intercollegiate survey of student opinions. They were further told that they would first see responses from another participant that would help them acclimatize to the task. All participants were first provided with information about the source in the form of a code number and the source's school affiliation. Half the participants learned the source was a student at UCSB (ingroup) and the other half learned the source was a student at University of Manitoba, Canada (outgroup). All participants then received two other kinds of information from the source. In the position statement, the source expressed support for one of the key issues, revealing his or her position on the matter. In the argument statement, the source presented three arguments supporting his or her position and three arguments opposing it. Arguments were presented one at a time on a computer screen and participants pressed the spacebar after reading each argument to initiate presentation of the next item. The balance of arguments was either strong (three strong supporting arguments plus three weak opposing arguments) or weak (three weak supporting arguments plus three strong opposing arguments). The order in which the three supportive and three opposing arguments were presented was counterbalanced. The arguments made it clear that the source had considered both sides of the issue, whereas the position statement clearly stated which side of the issue the source had found more compelling. Because of differences in participants' initial reactions to the two issues, the source's position on the euthanasia issue was proattitudinal, whereas his position on the handgun issue was counterattitudinal. Once the source's presentation was complete, participants gave their own views on the issue, and this was the primary measure of the source's persuasive effectiveness. Participants read the position and argument statements about one issue and completed dependent measures relevant to it before reading the student's views on the other issue and responding to dependent measures with regard to it. The order in which the gun control and the euthanasia issues were presented was counterbalanced. There were three key manipulations. First, half of the participants learned that the source was an ingroup member, whereas the other half believed the source to be an outgroup member. All participants knew before message presentation whether the source was an ingroup or an outgroup member. Second, half of the participants saw the statement expressing the source's position before the arguments were presented to them, whereas the other half saw the source's position after argument presentation. Consequently, in some conditions a source heuristic could be used to assess validity without message processing, whereas in others it could not. Third, half of

< previous page

page_144

next page >

< previous page

page_145

next page > Page 145

the participants read arguments that, on balance, strongly supported the advocated position, whereas the other half read arguments that weakly supported the advocated position. When the source heuristic was available, we would be able to see if and how it was used. When using the source as a heuristic cue was not possible, we would be able to see how participants dealt with communications from ingroup and outgroup sources. What effect did these manipulations have on the strategy with which participants processed the messages? The primary measure of interest was attitude change toward the positions advocated on the two issues. Analysis indicated that the extent to which participants were persuaded depended on the identity of the source, when the source's position became known, and the quality of the arguments supporting the source's position. To assess the different processing strategies that this interaction might reflect, we looked separately at attitude change that occurred when the source's position was known before argument presentation and when the source's position was not known until after argument presentation. When the source's position was announced before arguments were presented, participants showed a tendency to accept the position advocated by an ingroup member compared to an outgroup member regardless of argument quality (see top panel of Fig. 8.1). Further analysis revealed that this difference was significant when the issue was counterattitudinal but not when the issue was proattitudinal. Such unqualified acceptance of the ingroup position thus appeared to reflect heuristic processing, but evidence for use of group status as a simple cue could be demonstrated only when the issue was counterattitudinal. When the source's position statement was delayed until after argument presentation, the quality of the arguments influenced attitude change but only when the source was an ingroup member (see bottom panel of Fig. 8.1). Participants were significantly more accepting of the ingroup member's opinion when it was accompanied by strong arguments than when it was supported by weak ones. Argument quality had no impact on reactions to an outgroup member's views. These results suggest that the communication received from an ingroup member was systematically processed when the use of source status as a heuristic cue was not possible. The time participants spent reading each argument before moving on to the next also supported these conclusions. When participants knew the source's position on the issue beforehand, ingroup arguments were processed more rapidly than outgroup arguments, as can be seen in Fig. 8.2. When announcement of the source's position was delayed, however, ingroup arguments were processed for a much longer time than were outgroup arguments. These data were thus consistent with the idea that the

< previous page

page_145

next page >

< previous page

page_146

next page > Page 146

FIG. 8.1 Attitude change toward the advocated position as a function of source group membership, argument strength, and the agreement of the advocated position with the participants' initial attitudes. Source group membership always was revealed prior to reading the source's position on the issue and the persuasive arguments. The top panel shows attitude change for participants (cont'd on next page)

< previous page

page_146

next page >

< previous page

page_147

next page > Page 147

(FIG. 8.1. cont'd from previous page) who read the source's position on the issue before reading the persuasive arguments. The bottom panel shows attitude change for participants who read the source's position on the issue after reading the persuasive arguments. The proattitudinal argument advocated the legalization of euthanasia, and the counterattitudinal argument advocated fewer restrictions on gun ownership. ingroup messages were processed heuristically when a group cue was available but processed systematically when it was not. Thus, messages attributed to an ingroup source were most persuasive when the message was strong and when heuristic acceptance of the message was discouraged by delaying the source identification until after the message was delivered. Ingroup messages were least persuasive when the message content was weak but the source heuristic was unavailable because the source was not identified until after the message was delivered. Under conditions where the group membership heuristic was unavailable, then, the findings are indicative of systematic processing. When the ingroup source

FIG. 8.2 The average amount of time spent reading persuasive arguments (in seconds) as a function of when the source's position on the issue was revealed and the source's group membership.

< previous page

page_147

next page >

< previous page

page_148

next page > Page 148

was identified prior to the persuasive arguments, levels of persuasion were intermediate, regardless of message quality. This latter finding is suggestive of a moderate heuristic acceptance of the ingroup source's appeal. Although evidence for both heuristic and systematic processing of ingroup messages was found, the outgroup messages produced no attitude change in any condition. In addition, reading times for outgroup messages were fast, suggesting the absence of systematic processing. Thus, it appears that messages ascribed to outgroup sources were not subjected to careful processing. Our findings thus gave some credence to Kelman's (1958, 1961) and others' suggestions that group-mediated persuasion may be different from persuasion that occurs because of the careful processing of arguments. Both attitude change and processing indices suggested that ingroup status can operate as a heuristic cue. When use of the cue was possible, acceptance of the ingroup messages appeared to occur with little content-focused processing. Although this change was significant only when the message was counterattitudinal, it is possible that diagnosing change on the proattitudinal issue may have been made difficult by ceiling or assimilation effects. Results from the condition in which participants were unaware of the source's position beforehand, however, revealed considerable investment in content-focused processing of the ingroup message. Participants in this condition spent a considerable amount of time processing presented arguments, presumably to ascertain the ingroup source's position. As a result, this extra processing meant that persuasion was dictated largely by argument quality. Although recipients were persuaded by an ingroup position backed by strong arguments, they rejected an ingroup position backed by weak ones. In summary, it appeared from this experiment that private acceptance of ingroup messages is routinely heuristic, unless the ingroup position is difficult to discern, in which case systematic processing might occur. We were aware, however, that the issues about which participants had received persuasive appeals were quite unrelated to group membership. A position on euthanasia or gun control was not definitional of being a UCSB student in any sense, nor was there any particular relevance or involvement for UCSB students as a group in these issues. Thus, it is reasonable to think that processing motivation regarding these issues was somewhat reduced. As reductions in processing motivation are often associated with increases in reliance on heuristic cues, it seems possible that our findings that ingroup sources produced heuristic change reflected issue nonrelevance rather than a default means of dealing with ingroup messages. Assessing this possibility required monitoring processing reactions to messages from ingroup and outgroup sources on an issue that was relevant to group membership. Just such conditions held in an experiment con-

< previous page

page_148

next page >

< previous page

page_149

next page > Page 149

ducted earlier in our laboratory (Mackie, Worth, & Asuncion, 1990). In this experiment, UCSB students read a message on the topic of standardized testing. We chose this issue for use with our predominantly freshman participants because we believed that the issue would be particularly relevant to their membership as UCSB students. Scholastic Achievement Tests (SAT) scores recently had played a large role in determining their acceptance to UCSB and, consequently, in determining their membership as UCSB students. Having ascertained participants' initial attitudes on the issue of continued use of the SAT as a criterion for college admissions at an earlier session, we presented them with a persuasive communication that advocated abolishing the use of SAT scores. This was done under the guise of asking participants to read a speech purportedly delivered by a delegate at an intercollegiate conference on educational policy. After reading the speech, participants answered some questions about the speech and the delegate, ostensibly for a study on ''people's perceptions of delegates who are representing their constituents' views at conferences.'' Embedded in this questionnaire were several items intended to reassess participants' attitudes toward the SAT exam. Two manipulations were important in this study. First, for half of the participants the experimenter identified the source as a UCSB student representing the views of UCSB students (the ingroup), whereas for the other half, the experimenter identified the source as a University of New Hampshire (UNH) student said to be representing UNH students (the outgroup). Second, half the participants in each condition read a message comprised of arguments pretested to be strong and valid, whereas the rest of the participants saw messages comprised of weak and specious arguments. In all cases, the first sentence in the message clearly stated the source's position that the use of SAT scores should be discontinued. If receiving a message from an ingroup member initiated heuristic processing, we expected that both strong and weak messages from the ingroup would result in significant attitude change. If, on the other hand, receiving a message from an ingroup source triggered systematic processing, we expected to find that participants would be persuaded by a strong message from the ingroup source but not by a weak one. Receiving a message from an outgroup member was predicted to be generally less persuasive than receiving a message from an ingroup member, because an outgroup source would act as a heuristic cue indicating the message was not valid or should be ignored. Attitude change toward the position advocated in the message was calculated by comparing each participant's attitude before and after exposure to the message, with positive scores indicating movement toward the position advocated in the message. Analysis yielded several results that provided insight into participants' processing strategies. First, participants

< previous page

page_149

next page >

< previous page

page_150

next page > Page 150

exposed to a message from the ingroup source showed significant change toward the advocated position, whereas those exposed to the UNH source moved slightly, but nonsignificantly, in the direction opposite to the one advocated in the message. Thus overall, ingroup sources were more persuasive than outgroup sources. Second, tests of the simple effect of argument quality for the ingroup source condition revealed that the ingroup's persuasive superiority derived largely from the condition in which the message was strong rather than when it was weak (see Fig. 8.3), a pattern consistent with increased content processing of ingroup messages. This pattern of differentiation between strong and weak messages was not present when the source was a member of another group: The outgroup source was largely ineffective regardless of message quality. Three other measures also supported these conclusions. First, participants recalled a slightly greater number of arguments from ingroup rather than outgroup messages. Second, the valence of elaborations made about both message content and source expertise differed quite dramatically de-

FIG. 8.3 Attitude change toward the advocated position as a function of source group membership and argument strength.

< previous page

page_150

next page >

< previous page

page_151

next page > Page 151

pending on the source of the message. When messages came from an ingroup source, strong messages received more favorable than unfavorable responses, whereas weak messages received more unfavorable than favorable responses. These differences were not significant when the message was attributed to the other group source. A similar pattern was observed for the number of favorable and unfavorable comments participants made regarding the message source. Finally, the favorability of the message elaborations proved a significant predictor of attitude change when participants read ingroup messages but not when participants read outgroup messages. These results are all consistent with the idea that messages describing an ingroup member's opinions received more scrutiny than messages attributed to outgroup sources. As we suspected, then, the relevance of the issue at hand to the salient group membership influences how persuasive appeals are processed. The results of these two studies combined suggested that when messages are not relevant to the ingroup, the ingroup source acts as a heuristic cue prompting acceptance unqualified by careful processing of message content. When the issue at hand is relevant and group defining, however, ingroup messages may be processed routinely in a systematic fashion. Our confidence in these conclusions, however, was dampened by the fact that we inferred them from comparisons across studies. To better assess the impact of issue relevance on the processing of ingroup messages, we manipulated the relevance of the issue to the ingroup as well as source status and message quality in a single experiment (Mackie et al., 1990). To do so, we chose two issues that participants saw as being equally important but of differential relevance to a salient ingroup. Participants' initial attitudes on these issues were measured during a pretesting session. One week later, participants listened to two 2-minute speeches, one on each issue, ostensibly delivered by either an ingroup (UCSB) or an outgroup (UNH) delegate at a student-initiated environmental conference. Participants watched a slide of the delegate as they listened to his or her speech. The source's group identification was made obvious by the T-shirt he or she wore and also was mentioned by the experimenter. In addition to the manipulation of source status, the relevance of the issue to ingroup membership was important. All participants heard one speech discussing an issue that pretesting had confirmed as relevant to the ingroup (oil drilling off the southwestern coast of the United States) and one of much less relevance to the ingroup (acid rain problems in the northeastern United States). Finally, messages were comprised of either strong or weak arguments so that the presence or absence of systematic processing could be diagnosed. After listening to the first speech, participants reported their own attitude on the is-

< previous page

page_151

next page >

< previous page

page_152

next page > Page 152

sue; they then heard a second message from a second delegate on the other issue and completed the dependent measures again. Analysis of the extent of change toward the position advocated in the message revealed the combined impact of source status, issue relevance, and argument quality. When participants received a message on an ingroup relevant issue (oil drilling) from an ingroup source, they accepted the advocated position more when the message was strong than when it was weak. This pattern of attitude change was thus consistent with the occurrence of content-focused processing of the ingroup relevant message from the ingroup source. When the issue was not ingroup-relevant, however, participants showed moderate and significant amounts of change toward the position advocated by the ingroup source, regardless of message quality. This acceptance of the ingroup position without careful processing of its content suggested that source status was acting primarily as a heuristic cue. In contrast, the manipulations had little impact in the outgroup source condition. Other-group members were much less persuasive than ingroup members and attitude change was not significantly greater than zero in any condition. Conclusions and Future Directions What can be concluded from the results of this program of research? First, they indicate that receiving a message from an ingroup source can persuade via a number of different mechanisms, depending on the circumstances. The mere presence of the ingroup source can act as a persuasive cue, and heuristic acceptance of the advocated position is possible as long as that position is made clear. It may be that this is the default mode of processing whenever the message topic is not one of central importance to group identity (as in some of our experiments), or when group identity is particularly salient (a factor not manipulated in our studies). Thus, our results provide evidence of a processing mechanism similar to what Kelman (1958, 1961) may have intended by identification. Despite its apparently heuristic quality, however, such group-mediated persuasion may not have all of the qualities typically associated with heuristic attitude change. Attitude change produced via heuristic or peripheral routes typically is thought of as relatively short-lived and vulnerable to counterpersuasion. However, to the extent that exposure to ingroup sources is frequent, activation of the process may produce long-lasting change. Alternatively, messages from ingroup sources, even on issues not originally ingroup relevant, may become attached to an ingroup or self-concept and become both long-lasting and resistant through this mechanism. These possibilities suggest that the nature of attitude change produced by different kinds of heuristic cues might be quite different, and this idea warrants further investigation.

< previous page

page_152

next page >

< previous page

page_153

next page > Page 153

Second, the results of the current studies make clear that group-mediated change can be produced via increased message processing and not merely as a result of reliance on persuasion cues. When the issue under consideration was relevant to ingroup membership, or when the ingroup position on an issue was not clear, messages from ingroup sources were subject to considerable scrutiny. This scrutiny produced significant interactions with message quality compared to conditions where the message was heuristically processed. Note that this message scrutiny meant that ingroup messages were not routinely acceptedelaboration of weak messages led to less persuasion than scrutiny of strong messages. Although this finding suggests that ingroup messages were being scrutinized fairly objectively, a closer look at the data suggests that this might not be completely true. Compared to the same message delivered by an outgroup source, strong ingroup messages were more persuasive, but weak ingroup messages were not less compelling (in fact, they were as effective as strong other-group messages). Because systematic processing of weak arguments typically leads to less attitude change than that produced by heuristic processing, the pattern described here hints of a bias toward acceptance of ingroup relative to other-group messages. That is, recipients may be predisposed toward acceptance of ingroup messages. In the case of weak messages, further processing reduces the extent of this acceptance, but the "head start" they were given makes them relatively effective. Chaiken (1987) suggested that persuasion cues may have an initial impact on advocacy acceptance that is modified only if further, more deliberative, processing of message content occurs. Our results are consistent with this suggestion: Message processing produced differences in reaction to messages of different argument strength, but the predisposition to agree with the ingroup displaced this differentiation toward acceptance of the message. The third conclusion that can be drawn from our data is that the relevance of the issue to ingroup identity has an important effect on which process or processes occur. At this point, we know very little about why this is true. One possibility focuses on what is activated in the persuasion context. Reception of a message from or about an ingroup presumably activates one's mental representation of the ingroup (Turner, 1982). Among other things, this structure might well contain typical issues and opinions that are of concern to the target group (Andersen & Klatzky, 1987). The activation of an ingroup structure might thus be accompanied by activation of knowledge about these group-relevant issues, as well as knowledge about their importance and relevance to the ingroup (and thus to the self). Because increases in knowledge on a topic (Bobrow & Norman, 1975; Wood, Kallgren, & Priesler, 1985) and increases in the relevance of a topic (Harkness, DeBono, & Borgida, 1985; Petty & Cacioppo, 1979b, 1984) have been found to in-

< previous page

page_153

next page >

< previous page

page_154

next page > Page 154

crease extensive processing, activation of the ingroup structure might increase both the capacity and the motivation to process. In support of these ideas, persuasive advertisements that describe products as having group-relevant features have been found to be particularly effective (Worth, Smith, & Mackie, 1992). Future research should be directed at understanding the nature of mental representations that are activated when group-relevant and group-irrelevant issues are raised by ingroup and outgroup sources. The differential impact of relevant and nonrelevant issues offers one means of explaining some of the apparent inconsistencies in the literature on similarity and persuasion. As noted, this literature suggests that source similarity both is and is not a boon to persuasion. One reason for these mixed findings may be a lack of attention to the dimension of relevancein a study that manipulates source similarity in terms of political group membership, a message concerned with welfare spending will have a different effect than does an appeal regarding the quality of a particular romantic comedy. As Turner (1982) pointed out, Asch's original idea was that people expect to agree with and are persuaded by consensus from those who share the attributes relevant to making a particular judgment. This idea has been supported in the conformity literature. Boyanowsky and Allen (1973) found a Black supporter was effective in decreasing conformity expressed by prejudiced White study participants in an Asch-like paradigm but only when the judgment was a physical measure or an general opinion. When the response involved the expression of a personal opinion, prejudiced White participants conformed to the majority despite the presence of a Black supporter. Presumably, a Black supporter did not share attributes relevant to the prejudiced White participants' personal opinions and did not provide an appropriate social comparison for prejudiced Whites' evaluations of their own personal opinions. Our results speak much less to the processes that guide the rejection of outgroup messages. Because participants basically ignored outgroup messages in our studies, little diagnosis of underlying mechanisms is possible. It is worth noting that the term outgroup may be somewhat of a misnomer in the studies described here. In fact, the source might be better described as a "nongroup member." Situations in which the identity of ingroup and outgroup are so entwined that the presence of an outgroup source activates ingroup identity may lead to outgroup sources producing similar processing outcomes as those found in the present studies for ingroup sources. Although prior research has documented the tendency to disavow and move away from the position advocated by an outgroup source, this need not result from heuristic rejection of the message, nor need it be the only outcome. An alternative outcome might be that outgroup sources activate ingroup

< previous page

page_154

next page >

< previous page

page_155

next page > Page 155

identity and, consequently, trigger increased processing. In some circumstances, for example, when the message was strong, this increased processing could result in acceptance of an appeal delivered by an outgroup member. At a more general level, this program of research highlights the important impact of who says what to whom. The identity, shared or otherwise, of the individuals in source and audience roles can have considerable impact on both the processing and eventual effectiveness of persuasive appeals. Such results remind us that even when message variables appear to determine persuasion, the cognitive processes that mediate their effectiveness often presuppose, assume, and depend on an appropriate social context. The role of assumed consensus or social agreement, for example, seems to be a particularly important concept that should receive more attention in persuasion research. As Kelley (1952) made clear, consensus information is a powerful determinant of informational validity, leading as it does to external attributions. In addition to the impact that explicit consensus has in conformity and majority influence situations, consensus may have much wider effects because it is implicitly assumed. Phenomena such as false consensus and pluralistic ignorance no doubt owe their persuasive capacity to perceived consensus. Additionally, the persuasive power of statistical and expert evidence also may reside in their appearing to embody an appropriate consensus. Thus, social processes are just as essential to conferring validity on information and producing persuasion as are the more popularly investigated cognitive processes. With the means of assessing the processing pathways by which distal factors make their persuasive contribution now in hand, the time is ripe for a return to concern with social psychology's early focus on the social roots of persuasion and an integration of the persuasion and social influence literatures is now in order. Acknowledgments This chapter was written with the partial support of Grant SBR 9507628 from the National Science Foundation to Diane Mackie and a National Science Foundation Graduate Fellowship to Sarah Queller.

< previous page

page_155

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_157

next page > Page 157

9 Group Norms, Prototypicality, and Persuasion Daan van Knippenberg University of Amsterdam Early research on how group norms affect group members' attitudes and behaviors shows that group members construct or adopt a shared judgmental or behavioral frame of reference that guides their judgments, attitudes, and behavior (see Asch, 1956; Festinger, Schachter, & Back, 1950; Newcomb, 1943; Sherif, 1936; for classic examples). In the study by Sherif (1936), for instance, groups of participants were exposed to a highly ambiguous stimulus, a stationary point of light that appeared to move due to the absence of reference points in a totally darkened room. When requested to make a series of judgments of the magnitude of the movement of the light, group members rapidly developed a shared judgmental norm about this magnitude. This norm guided group members' judgments, resulting in norm-congruent responses. Asch (1956) reported a more extreme example of the influence exerted by shared judgmental standards. Participants in Asch's studies were requested to judge which of three lines was similar to a comparison line. When confronted with the unanimous but obviously erroneous judgments of their fellow group members (actually confederates), they tended to conform to this group normative response and express the same erroneous judgment. Perhaps as a result of the salience of such extreme examples of norm-induced influence as the yielding to unambiguously wrong judgments in the Asch studies, norm-induced, or group-induced, influence traditionally has been contrasted with information-based influence, implying a contrast between on the one hand socially dependent conformity and on

< previous page

page_157

next page >

< previous page

page_158

next page > Page 158

the other hand "true" influence that is socially independent and based on objective and valid information (e.g., Deutsch & Gerard, 1955). This dichotomy is, however, not unchallenged. Other researchers have proposed a social basis of informational influence, arguing that informational influence depends on social consensus and that norm-induced influence is based on the informational value of group norms (e.g., Abrams & Hogg, 1990; Hogg & Turner, 1987; Turner, 1982, 1985, 1991). Although this conceptualization rejects the view of norm-induced influence as essentially "informationless," even this latter position does not preclude norm-induced influence from reflecting uncritical acceptance of the information represented by group norms. In contrast, "true" influence is based on elaborate scrutiny of the information presented. Indeed, the two dominant models of persuasion, the elaboration likelihood model (ELM; Petty & Cacioppo, 1981, 1986a, 1986b) and the heuristic systematic model (HSM; Chaiken, 1980; Chaiken, Liberman, & Eagly, 1989) focus on this contrast between influence based on careful scrutiny of information and influence based on uncritical acceptance of a position and suggest that norm-induced persuasion is based on uncritical acceptance rather than elaborate scrutiny of group normative positions (Chaiken, Wood, & Eagly, 1996; Petty, Priester, & Wegener, 1994). This chapter questions this conception of norm-induced influence as a process of uncritical yielding, arguing instead that norm-induced influence may be based on the systematic processing of norm-representing communications. To do so, the chapter builds on the analysis of social influence provided by self-categorization theory (e.g., Abrams & Hogg, 1990; Hogg & Turner, 1987; Turner, 1985, 1991) and on the framework provided by the ELM and the HSM. It also hopes to make a step toward integrating the study of persuasion processes, which typically is concerned with the information processing underlying persuasion but not with the social contexts in which persuasion processes may take place, and the study of social influence, which typically is concerned with the social basis of influence but not with the information processing underlying influence. First, however, the distinction between socially dependent normative influence and socially independent informational influence and the self-categorizational approach to social influence is discussed briefly (for more extensive discussions of these approaches, see Abrams & Hogg, 1990; Hogg & Turner, 1987; Moscovici, 1976; Turner, 1985, 1991; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; Turner & Oakes, 1989). Socially Dependent Versus Socially Independent Influence Traditionally, social influence researchers have distinguished between socially dependent influence and socially independent influence. A straightforward statement of this position is found in Deutsch and Gerard's (1955)

< previous page

page_158

next page >

< previous page

page_159

next page > Page 159

distinction between normative and informational influence. Normative influence instigates conformity to others' expectations to gain approval or acceptance or to avoid disapproval or rejection. Normative influence requires that the target of influence depend on the source of influence in that the influencing agent needs to possess the ability to administer social rewards or punishments in case the target of influence does not yield. Normative influence thus is based on (implicit) coercion and, hence, is associated with compliance more than with true internalized acceptance. Because of its coercive nature, normative influence is socially dependent in that it is contingent on the influencing agent's characteristics (the ability to reward and punish), and the influencing agent's surveillance (needed to reward or punish depending on the target's behavior). Informational influence, on the other hand, is based on the validity of the information provided by the influencing agent or on the persuasiveness of the influencing agent's argumentation, independent of who or what the source is. As a consequence, informational influence is socially independent in the sense that the source of the information is irrelevant (and need not, in fact, be known), and surveillance by the influencing agent is unnecessary to reinforce the influence. Where Deutsch and Gerard (1955) distinguished one single form of socially dependent influence, other researchers have identified subtypes of socially dependent influence, for instance by distinguishing influence based on (a) the ability of the influencing agent to reward or punish, (b) the influencing agent's stature as an expert, (c) the influencing agent's entitlement to influence because of his or her position in a hierarchical structure, and (d) the target's identification with the influencing agent (e.g., French & Raven, 1959; Kelman, 1958, 1961). The latter three forms of influence are socially dependent in that they are contingent on the influencing agent's characteristics, but they do not require surveillance. They thus acknowledge a form of socially dependent influence that is not associated with coercive group pressures. Still, these socially dependent forms of influence are contrasted with information-based socially independent influence, implying that socially dependent influence is essentially a "mindless" influence process independent of the information provided. This perspective suggests that we are influenced either by the nature of the source of influence, irrespective of the information provided, or by the information provided, irrespective of the source. Even so, this perspective recognizes that others' opinions may be informative about social reality, in other words, that others' opinions may have informational value (e.g., Festinger, 1950; Kelley, 1952). However, informational influence based on others' opinions is still conceptualized as source-independent: As far as the informational value of others' opinions is concerned, others are no more than a means of getting access to "objective" (i.e., source-independent) information about physical reality. Thus, the di-

< previous page

page_159

next page >

< previous page

page_160

next page > Page 160

chotomy between a normative and informational form of influence is maintained. Challenging the Dual Process View of Social Influence In essence, the models just discussed are dual process models of social influence (cf. Turner, 1991), in the sense that they contrast socially dependent influence as an uncritical, mindless conformity process with information-based influence as a socially independent, ''objective,'' and "true" influence process. Group norm-induced influence, by definition socially dependent, thus would seem to fall in the category of information-independent "mindless" influence. This implies that norm-induced influence is based on coercion, or at best on uncritical, unthinking acceptance, and has little to do with the objectivity and validity of information. But is there no informational value to group norms? One could rightfully argue that participants in Sherif's (1936) study, faced with no other information than the apparent movement of a point of light and the judgments made by their fellow group members, made good use of the information available when they aligned their judgments with the shared judgmental standard that developed during the group task. In other words, group norms may represent information relevant to the group's members. This view of group norms as representing group-relevant information lies at the core of the more recent selfcategorizational analysis of social influence (e.g., Abrams & Hogg, 1990; Hogg & Turner, 1987; Turner, 1982, 1985, 1991; Turner et al., 1987) that proposes that all influence is socially dependent in that the perceived validity and objectivity of information is contingent on social or cultural consensus (see also Moscovici, 1976; Moscovici & Faucheux, 1972). That is, validity and objectivity should be seen not as attributes of the physical world but rather as products of social agreement: "Objective truths" are perceived as such only because there is social consensus concerning their objectivity and truthfulness. Viewed from this perspective, the distinction between normative and informational influence is meaningless. If the validity and objectivity of information is based on social consensus, the degree to which information is normative (i.e., represents social consensus) determines its validity and objectivity, that is, its informational value. Thus, normative influence is based on the informational value of norms (i.e., is informational), and informational influence is derived from the normativeness of the information (i.e., is normative). In other words, all influence is proposed to be simultaneously informational and normative. Two other aspects of the self-categorizational approach to social influence are essential to the analysis of norm-induced influence. First, social

< previous page

page_160

next page >

< previous page

page_161

next page > Page 161

identification is proposed to mediate norm-induced influence: Group norms exert influence only to the extent that the target of influence identifies with the group. That is, ingroup norms induce normative influence (Abrams, Wetherell, Cochrane, Hogg, & Turner, 1990; Hogg & Turner, 1987) because shared group membership provides the necessary basis for a shared social reality and thus for influence based on norms that represent that shared reality. Second, group norms themselves are unobservable abstract concepts: Group members do not "take a look at the group norm" and then act accordingly. Rather, group members are affected by group norms through social interaction, by observing the behavior of other group members, or by being confronted with the beliefs, opinions, and attitudes of their fellow group members. That is, the actual influence exerted by group norms is mediated by behaviors, opinions, and arguments that function as norm-representing instances from which group normative positions may be derived (cf. Hogg & Turner, 1987). However, not all attitudes, beliefs, and arguments group members express are equally group normative. As a consequence, in the study of norm-induced influence distinguishing more group-normative from less group-normative communications would seem necessary. The degree to which communications (verbal or nonverbal, with communicative intent or not) are group normative, or norm-representative, may be referred to as the prototypicality of the communication (cf. Turner et al., 1987). Prototypicality refers to the extent to which a category instance is representative of both within-group similarities and between-group differences (Turner, 1985; Turner et al., 1987; cf. Rosch, 1978). The more a group member, a statement, or an attitude represents what members of a group have in common and what differentiates the group from other groups, the more group prototypical the group member, statement, or attitude is (note that, because of the role of intergroup differences, the most prototypical position, the group norm, is not necessarily the group mean). In the self-categorization approach to social influence, norm-induced influence is conceptualized as conformity to the most prototypical group position (e.g., Hogg & Turner, 1987; McGarty, Turner, Hogg, David, & Wetherell, 1992; Turner, 1985, 1991; Turner et al., 1987). Thus, from this perspective, the study of norm-induced influence essentially should concern itself with the relation between the prototypicality of communications and influence. Prototypicality and Routes to Persuasion Once we recognize that norm-induced influence may be based on the informational value of group norms, we may raise the question of how this information is processed to engender persuasion. Is norm-induced influence based on uncritical acceptance of the information communicated by the

< previous page

page_161

next page >

< previous page

page_162

next page > Page 162

group norm, as most approaches to normative influence would imply, or is it based on careful consideration of the position implied by the group norm, as the proposition that norm-induced influence is informational might suggest? Attention to the information processing underlying norm-induced influence would seem all the more relevant if we take into consideration that group norms are not represented by single statements of position but rather are more likely to be communicated by behaviors, argumentations, attitude statements, and other intragroup communications that leave room for more elaborate processing than just acceptance or rejection of a position. Both the recognition that some form of communication is necessary to elicit the influence induced by group norms and the contrast between uncritical acceptance and critical scrutiny implied by the question asked earlier turn our attention toward models of persuasion to address the information processing underlying norm-induced influence. For the last decade and a half, research on influence induced by persuasive communication predominantly has focused on determinants and consequences of the way persuasive messages are processed (see, e.g., Chaiken et al., 1996; Eagly & Chaiken, 1993; Petty et al., 1994, for reviews). Central to this research effort is the distinction between persuasion based on the careful processing of a message and persuasion that is relatively independent of the actual message contents, and rather based on the use of persuasion cues (e.g., Chaiken, 1980; Chaiken et al., 1989; Petty & Cacioppo, 198 lb, 1986a, 1986b). This distinction is derived from two models of the determinants of the way persuasive communication is processed and the relationship between processing mode and the extent to which different message and nonmessage factors may affect attitudes, Petty and Cacioppo's (1981b, 1986a, 1986b) ELM and Chaiken's (1980; Chaiken et al., 1989) HSM. Both models distinguish between an effortful systematic (HSM) or central (ELM) processing mode and a relatively effortless heuristic (HSM) or peripheral (ELM) processing mode. Systematic or central processing comprises careful scrutiny of the message contents. Persuasion in this processing mode is contingent on the perceived validity of the information and the strength of the arguments presented in the message. If systematic processing results in attitude change, this change tends to be relatively enduring and predictive of behavior. Heuristic or peripheral processing, on the other hand, implies a lazy, quick-and-dirty processing strategy. Persuasion in this processing mode is contingent on noncontent persuasion cues such as characteristics of the source (e.g., attractiveness, expertise), the length of the message, or the number of arguments in the message rather than on the actual arguments or information presented. Using these cues, for instance, to apply simple decision rules (heuristics) such as "expert sources can be trusted" or "length implies strength," the recipient may decide in a quick-and-dirty fashion

< previous page

page_162

next page >

< previous page

page_163

next page > Page 163

whether or not to accept the position advocated in the message. In contrast to attitude change elicited by systematic or central processing, attitude change resulting from heuristic or peripheral processing tends to be relatively short-lived. Recipients of persuasive communication only engage in the more effortful systematic or central processing mode if they are both motivated and able to do so. Motivation to process has been shown to be affected by such factors as the personal relevance of the communication (e.g., Petty, Cacioppo, & Goldman, 1981), accountability (e.g., Chaiken, 1980), and the need for cognition (e.g., Cacioppo, Petty, Kao, & Rodriguez, 1986), whereas ability has been found to depend on the presence or absence of distracting circumstances and message repetition (e.g., Cacioppo & Petty, 1985), among other factors (see Eagly & Chaiken, 1993; Petty & Cacioppo, 1986a, for comprehensive reviews of the determinants of the routes to persuasion and the consequences of persuasion through the different processing modes). Persuasion through systematic or central processing is contingent on the contents of the message alone and independent of noncontent cues, whereas persuasion in the heuristic or peripheral mode is independent of message contents. Although persuasion in the heuristic or peripheral mode need not be socially dependent (e.g., the number of arguments in the message or message length may be used as heuristic cues), influence instigated by social factors such as source characteristics is located in the content-independent processing mode both within the ELM and within the HSM. Thus, the systematic or central versus heuristic or peripheral processing dichotomy to some extent echoes the distinction between socially dependent influence and informational influence. Although both models are best known for the distinction between the two processing modes just discussed, both models actually allow for a mixed processing mode in which noncontent cues affect systematic processing, potentially allowing for norm-induced influence based on systematic processing of persuasive communications. Specifically, the ELM proposes that noncontent factors (e.g., forewarning of persuasive intent; Petty & Cacioppo, 1979a) may elicit a bias to respond in a favorable or unfavorable fashion when centrally processing the message, rendering persuasion contingent on both the quality of the message and on a predisposition to respond in a certain way. Without reference to response predispositions, the HSM allows both for the use of heuristic cues to determine the validity of the message (especially when the validity of the message is not self-evident; Chaiken & Maheswaran, 1994) and for the independent cooccurrence of systematic and heuristic processing. Although both models thus potentially allow for the mixed processing implied by the notion of socially dependent informational influence, neither model predicts norm-induced processing.

< previous page

page_163

next page >

< previous page

page_164

next page > Page 164

There is, however, both reason to predict norm-motivated information processing and empirical evidence supporting that prediction. Prototypicality as Moderator of Systematic Processing As argued earlier, the study of norm-induced influence essentially is concerned with the relation between the ingroup prototypicality of communications and persuasion. Consequently, questions regarding the information processing underlying norm-induced influence refer to the relation between message prototypicality and message processing. Because ingroup prototypical messages represent ingroup consensus, and ingroup consensus renders information valid and persuasive (cf. Hogg & Turner, 1987; Turner et al., 1987), we may presume that for subjectively valid information, group members are predominantly dependent on ingroup prototypical messages. Because of this informational value of ingroup prototypical communications, we may predict that prototypical messages (i.e., valid and informative messages) instigate more systematic processing than nonprototypical (i.e., less valid and informative) messages. In addition, because prototypical messages represent group consensual positions, they will be perceived to be more valid than nonprototypical messages; and, hence, they will be more persuasive than nonprototypical messages. Both predictions were put to the test in a series of experiments by van Knippenberg and Wilke (1992; van Knippenberg, Lossie, & Wilke, 1994). These studies adopted the research framework provided by the ELM and the HSM (because the ELM is most explicit about this framework, the ELM's terminology is adopted here). The ELM proposes that central processing in contrast to peripheral processing encompasses the generation of issue-relevant thoughts in response to the message (cognitive responses) that mediate persuasion: If responses are predominantly favorable (i.e., positive, agreeing reactions to the message), attitude change may occur. If responses are predominantly unfavorable, the message will have no effect, or it may elicit movement away from the position advocated in the message. As a consequence, indications of recipients' processing mode may be found in the cognitive responses elicited by the message, in the relation between cognitive responses and postexposure attitudes, and in effects of experimental manipulations that may be assumed to affect cognitive response under conditions of high elaboration. Specifically, the number of issue-relevant thoughts reported on a thought-listing task should be indicative of message elaboration, because more issue-relevant thoughts imply more thinking about the message. Second, because attitude change is mediated only by the favorability of issue-relevant cognitive responses in the central route, strong

< previous page

page_164

next page >

< previous page

page_165

next page > Page 165

relations between the favorability of issue-relevant responses and postexposure attitudes should reflect central processing. Third, in the central mode, strong messages are more likely to elicit favorable cognitive responses and attitude change than weak messages are, whereas in the peripheral mode attitudes and responses should be independent of message contents. Hence, effects of experimental manipulations of message quality are informative about message elaboration. Finally, more elaborate processing may result in better recall of message arguments, although argument recall is a more questionable indicator of message elaboration (i.e., recall may be relatively independent of elaborateness of processing; cf. Petty & Cacioppo, 1986b). Guided by these notions, van Knippenberg and Wilke (1992, Experiment 1) conducted an experiment in which participants were exposed to two persuasive messages, one opposing and one supporting the introduction of university entrance exams. The qualities of the pro and the con message were varied independently (i.e., by presenting arguments that were pretested to be either strong and persuasive or weak and nonpersuasive). Both messages allegedly were derived from an opinion survey of undergraduates students from participants' own major (i.e., they originated from an ingroup source). Ingroup prototypicality of the message was manipulated by introducing either the pro or the con message as "representative" of undergraduate (ingroup) attitudes to the issue in a recent attitude survey, and labeling the other message "not representative." After participants had read both the prototypical and the nonprototypical message (order of presentation was counterbalanced), attitudes toward the introduction of entrance exams and perceived quality of the message were assessed. Cognitive responses were measured by means of a thought-listing task that required participants to write down the thoughts they had while reading the message (cf. Petty & Cacioppo, 1986b). The number of issue-relevant thoughts listed was determined, and listed thoughts were categorized as being either in agreement or disagreement with the message position (or neutral toward the position advocated in the message). On the basis of the latter, the overall favorability of responses to the message was determined by computing the difference between the number of positive and the number of negative thoughts. Finally, argument recall was assessed by asking participants to write down everything they could remember of the contents of the message. As predicted, results showed that the prototypical message elicited more systematic processing. First, the favorability of cognitive responses to the prototypical message was predictive of postexposure attitudes (but only when the message was counterattitudinal), whereas favorability of responses to the nonprototypical message was not. Second, argument recall was better for the prototypical message than for the nonprototypical mes-

< previous page

page_165

next page >

< previous page

page_166

next page > Page 166

sage. In addition, prototypicality appeared to engender what would be called biased processing in the ELM. The ratings of message quality were higher for the prototypical message than for the nonprototypical message, and when also proattitudinal, the prototypical message elicited more positive cognitive responses than did the nonprototypical message. This would seem to corroborate the proposition that ingroup consensus renders messages persuasive. Finally, recipients' attitudes were more favorable toward entrance exams when the pro message was introduced as ingroup prototypical than when the con message was said to be ingroup prototypical, showing that prototypical messages are associated with greater influence. These results support the notion that prototypical messages are both more systematically processed and more influential than nonprototypical messages. They do not, however, explicitly address the mediating role that social identification is assumed to play in inducing norm-based influence. Only ingroup norms are likely to elicit norm-induced influence (e.g., Abrams et al., 1990; Mackie, 1986). It follows that the degree to which a message represents any given group's consensus does not elicit systematic processing, but ingroup prototypicality does. To test this hypothesis, van Knippenberg and Wilke (1992) conducted a second study in which participants were exposed to either a strong or a weak message that either supported or opposed the introduction of entrance exams. This message was introduced as either originating from ingroup (undergraduates) or outgroup (ministry employees, a nonmembership group of approximately equal status). Independent of message position, the source group was announced either as supporting or as opposing entrance exams, thus rendering the message either prototypical or nonprototypical, depending on the message position-source group norm congruency. Results indicated that ingroup messages were more systematically processed, especially when they were representative of ingroup norms. First, strong messages elicited more favorable cognitive responses than did weak messages, but only when the message originated from ingroup, suggesting, overall, more elaborate processing of ingroup messages than of outgroup messages. Second, the favorability of cognitive responses was predictive of posttest attitudes only after participants were exposed to a prototypical ingroup message, indicating that mainly norm-representing ingroup messages elicited systematic processing. In addition, ingroup messages elicited more favorable issue-relevant cognitive responses than did outgroup messages, illustrating the social dependency of persuasiveness. Finally, norm-induced influence mediated by social identification was evident in the finding that recipients tended to align their attitudes with the ingroup norm but not with the outgroup norm.

< previous page

page_166

next page >

< previous page

page_167

next page > Page 167

The van Knippenberg and Wilke (1992) experiments arguably employed a rather blunt operationalization of prototypicality, explicitly introducing the prototypical message as representing source group attitudes and explicitly (Experiment 2) or implicitly (Experiment 1) announcing the normative position of the source group on the message topic. In a follow-up study, van Knippenberg et al. (1994) addressed the issue of prototypicality in a more subtle way, focusing on group member prototypicality. Prototypical group members, as compared with nonprototypical members, are similar to other ingroup members and relatively different from relevant outgroups, thus representing both what group members have in common and what distinguishes the group from other groups (Hogg, 1993; McGarty et al., 1992; Turner et al., 1987; see also Hogg & Hardie, 1991, 1992; Hogg, Cooper-Shaw, & Holzworth, 1993). As a consequence, group members might be expected to be motivated to systematically process the message from a prototypical group member as a source of information about ingroup normative attitudes. Nonprototypical members, on the other hand, are less representative of the group, and, therefore, persuasive messages attributed to such group members should be less likely to engender systematic processing. This hypothesis, which is a generalization of the hypothesis formulated earlier, was tested in an experiment in which the prototypicality of the source was varied. In addition, van Knippenberg et al. (1994) argued that, although prototypicality may render systematic processing of a persuasive message more likely, under conditions of low elaboration likelihood prototypicality also might function as a heuristic cue affecting attitudes independent of message contents. To investigate this potential double function of prototypicality, a condition was introduced that was expected to render heuristic processing relatively more likely. Specifically, in half the conditions, the source's position on the message issue was announced before message presentation. It was expected that prior knowledge about the position the source would advocate in the message would render the message itself less interesting as a source of information and would hence decrease the likelihood of systematic processing and, at the same time, increase the likelihood that prototypicality information would be used as a heuristic cue. Under conditions of low elaboration likelihood, the prototypical source was predicted to elicit more attitude change than the nonprototypical source, unmediated by cognitive responding to the message contents. Participants received a strong or a weak message that either supported or opposed the introduction of entrance exams. The source of the message was introduced as a fellow student. As a manipulation of source prototypicality, the source's attitudes toward other student issues were described, suggesting

< previous page

page_167

next page >

< previous page

page_168

next page > Page 168

that the source held attitudes that were either typical or atypical of the student population (which was determined in a pretest). In half the conditions, the source's attitude toward the message topic was announced before message presentation. In the other conditions, no mention of the source's opinion was made. Results supported the predictions. The message from the prototypical source elicited more systematic processing but only when the position advocated in the message was not known beforehand. First, when no prior information about the source's position was provided, the message from the prototypical source elicited more issue-relevant cognitive responses. Second, favorability of cognitive responses was a better predictor of attitudes in the prototypical source, when no prior information about the source's opinion condition than in the other conditions. Moreover, generalizing van Knippenberg and Wilke's (1992) demonstration of socially dependent persuasiveness, the quality of the prototypical source's message was rated higher than the quality of the message from the nonprototypical source. Finally, recipients' posttest attitudes were more influenced by the position advocated by the prototypical source than by the position advocated by the nonprototypical source (independent of the foreknowledge manipulation). The results of these studies show that prototypicality simultaneously may motivate systematic processing and elicit a tendency to respond favorably to the message. Moreover, they demonstrate that, if circumstances render heuristic processing more likely (e.g., because the source's position is known beforehand), prototypicality information may function as a heuristic cue affecting the willingness to accept the position advocated in the message. Implications for the Study of Social Influence and Persuasive Communication Although the finding that prototypical messages elicit more attitude change than nonprototypical messages might be considered to be a replication or generalization of earlier observations of norm-induced influence (although group pressures to conform were arguably absent in the van Knippenberg & Wilke studies), the observation that prototypical messages are also more likely to elicit systematic processing is incongruent with earlier suggestions that norm-induced influence isat bestbased on uncritical acceptance of group normative position. Rather, it suggests a form of informational influence that is based on the systematic processing and persuasiveness of prototypical communications. Obviously, this suggestion is more in line with a conceptualization of social influence as rooted in a social reality defined by ingroup consensus than with a conceptualization of influ-

< previous page

page_168

next page >

< previous page

page_169

next page > Page 169

ence as based on either group pressures or source-independent information (cf. Abrams & Hogg, 1990; Hogg & Turner, 1987; Turner, 1982, 1985). In addition, viewed from the perspective of the ELM and the HSM, the observation that norm-induced influence may be the result of systematic processing suggests that "normative influence" may be internalized, true influence that is relatively enduring and predictive of behavior (cf. Chaiken et al., 1989; Petty & Cacioppo, 1986a) rather than the shortlived form of influence it has been implied to be. Moreover, the fact that ingroup prototypicality affects both the motivation to process and the perceived validity of a communication also implies that the motivation to process a message and the persuasiveness of a message may be more source-dependent than has previously been recognized in the field of persuasive communication. Both observations suggest a reconsideration of the role of source variables in persuasion. This is corroborated by the fact that prototypicality may be linked to the source of a message (i.e., sources may be more or less ingroup prototypical; cf. van Knippenberg et al., 1994). More generally, if we view prototypicality as the degree of "ingroupness" of a category instance, we may predict that the differences found in both the processing and the persuasiveness of messages from prototypical and nonprototypical ingroup sources by van Knippenberg et al. (1994) generalize to messages originating from one's own group as compared with messages from other groups. Ingroup messages should elicit more systematic processing and should be more persuasive than messages from other groups are, because ingroup messages are more ingroup prototypical than are messages from nonmembership groups. Indeed, results of van Knippenberg and Wilke's (1992) second experiment showed that, overall, ingroup messages tended to elicit more systematic processing and were more influential than outgroup messages. Mackie, Worth, and Asuncion (1990), Mackie, Gastardo-Conaco, and Skelly (1992), McGarty, Haslam, Hutchinson, and Turner (1994), and van Knippenberg and Wilke (1991), all obtained similar results and reported evidence of both more systematic processing and a greater impact on attitudes of ingroup compared to outgroup messages. These findings corroborate the line of reasoning presented here and demonstrate that the prototypicality concept may explain both within-group and between-group differences in social influence. Probably one of the major contributions of the self-categorizational approach to social influence is the focus on the intergroup context in which social influence takes place is (e.g., Hogg & Turner, 1987; Hogg, Turner, & Davidson, 1990; Turner et al., 1987). It not only follows from this approach that norm-induced influence is mediated by social identificationthat is, we tend to be affected only by ingroup norms (cf. Abrams et al., 1990; Mackie, 1986)but it also follows that group norms themselves may vary as

< previous page

page_169

next page >

< previous page

page_170

next page > Page 170

a function of the intergroup context. A straightforward demonstration of this notion may be found in a recent study by van Knippenberg and van Knippenberg (1996). In this study, participants were allegedly assigned to a group that would be required to make the decision either to take a bet offered by the experimenter or not. Bogus feedback about the group suggested that the group was either more likely or less likely to make risky decisions than a salient outgroup. Although this information did not suggest differences in absolute riskiness of the ingroup between conditions, the relatively higher or lower tendency toward risky decision making was sufficient to make group members see either the risky alternative (taking the bet) or the more cautious alternative (not taking the bet) as group normative (see also Hogg et al., 1990). That is, what is prototypical (group normative) in one situation need not be so in another. This notion of context-dependent prototypicality (see also Haslam, Oakes, McGarty, Turner, & Onorato, 1995) implies that, for instance, group members that are influential in one situation may not be so in another situation in which, because of contextual differences, they may be less prototypical. Similarly, contextual differences may affect the relative importance of group member prototypicality in determining influence. A recent study by Hains, Hogg, and Duck (1997), for instance, demonstrated that group prototypicality is predictive of leader effectiveness under conditions of high group salience but not under conditions of low group salience (see also Fielding & Hogg, 1997; Hogg, 1996). The argument presented in this chapter suggests that such contextual differences in either the degree of group prototypicality or the relative importance of prototypicality very well might affect the motivation to process and the persuasiveness of ingroup communications. This hypothesis concerning context-dependent motivation to process persuasive communications will, however, have to await future research.

< previous page

page_170

next page >

< previous page

page_171

next page > Page 171

10 Identity and Persuasion: An Elaboration Likelihood Approach Monique A. Fleming Richard E. Petty The Ohio State University Imagine a woman who encounters an advertisement for a new car that states it is endorsed by other women. How will the woman respond to the advertisement? Is she more likely to be persuaded to buy the car than she would be if the advertisement stated that the car is endorsed by men? How likely is her resulting attitude to last over time (e.g., long enough to get to a car dealer)? Finally, do the answers to these questions depend on the extent to which this woman likes being a member of the group ''women'' and feels that it is an important part of her self-concept? Although these are seemingly simple questions, their answers can be quite complex. Recent studies have suggested that an individual's characteristics such as group memberships (e.g., belonging to the group "women") affect behavior to the extent that those characteristics are salient and have been adopted as identities (i.e., they form an important part of an individual's self-concept). For example, a survey of Hispanic consumers found that those who strongly identified with their ethnic group reported being more likely to buy brands advertised with Hispanic spokespersons than did those who weakly identified with their ethnic group (Deshpande, Hoyer, & Donthu, 1986). In the political realm, Plutzer and Zipp (1996; Zipp & Plutzer, 1985) found that women were particularly likely to vote for women political candidates when they identified with "women's issues" and were registered as Independents

< previous page

page_171

next page >

< previous page

page_172

next page > Page 172

(and thus did not have a salient competing party identity).1 Plutzer and Zipp (1996) also found that this groupmotivated voting occurred particularly in favor of women candidates who were easily identified as "feminists." An important question raised by such findings is how identification with aspects of oneselfsuch as one's groupaffects behavior. One plausible mechanism for the impact of identity on behavior is through persuasion. Specifically, identity would influence persuasion processes, and the attitudes resulting from persuasion would then guide behavior. For example, an Hispanic individual's identification with the group "Hispanic" might help guide the attitude he or she forms about advertised products, and those attitudes would guide his or her consumptive behavior. The major goal of this chapter is to outline an integrative approach to understanding the role of identity in persuasion and subsequent behavior. A number of studies have examined the impact of making an aspect of one's identity temporarily salient (e.g., when students are told that an important change is being considered for their university, their student identity is presumably made salient; Petty, Cacioppo, & Haugtvedt, 1992). However, this chapter focuses on theories and research illuminating the question of the impact of relatively enduring aspects of the self-concept (e.g., group memberships, personal traits, or characteristics) on persuasion and behavior. Conceptualization and Assessment of Identity In this chapter, we use the term identity to refer to an individual's adoption of a particular group or groups (e.g., "I'm a Democrat") and/or trait or traits (e.g., "I'm honest") as part of her or his self-concept. Importantly, identities are not necessarily ascribed to individuals by others, as traits (e.g., "You are tardy'') or group memberships (''You are White") can be, but instead are the subjective perceptions of individuals, willfully chosen as part of who they define themselves to be. More specifically, an identity can be either a group identity or a personal identity. Our use of the term group identity incorporates ideas from social identity theory (SIT) and reference group theory about identifying with a group. Social identity theory defines a social identity as an existing group membership that becomes part of the self-concept through categorization of the self as a member of the group (Tajfel & Turner, 1986). In SIT, existing group memberships are based on social similarities and differences between people that society recognizes as defining membership in certain social categories in contrast to others. Thus, SIT does not allow for the possibility of 1Although identification with issues relevant to one's group is not the same as identification with one's group itself, it is plausible that the two constructs are highly related.

< previous page

page_172

next page >

< previous page

page_173

next page > Page 173

identifying with and incorporating into the self-concept a group to which one does not belong but to which one nevertheless aspires to belong. In contrast, reference group theory explicitly posits that one can use groups to which one does not belong (i.e., reference groups) to define the self (e.g., Siegel & Siegel, 1957). Our definition of group identity thus includes any groups that form part of the self-concept, whether the person is actually a member of the group or aspires to be a member. A personal identity as conceptualized here is a trait or individual difference that is incorporated into the self-concept. Our conceptualization of personal identity, like that of group identity, also includes both traits that the person already possesses along with traits that the person aspires to possess as long as those traits or groups are part of the selfconcept. Membership in a group or possession of a trait that is unimportant to a person's self-definition is not an identity. Of course, the group memberships and traits (whether real or desired) in one's self-system can vary in the extent to which one identifies with them. For example, people might be a member of Groups A and B (or possess Traits A and B) and willfully consider these traits and groups to be at least nominally part of the self-concept, but some groups (or traits) can be higher in identity than others if the person cares more about these groups or traits (cf. Markus, 1977). We combine the concepts of group and personal identity in our consideration of identity effects in persuasion because the existing empirical evidence suggests that the two concepts, though traditionally examined separately within the social psychological literature, might often be indistinguishable in their impact on persuasion. In a recent review, Deaux (1996) suggested that social (i.e., group) and personal identities do not differ in content. She argued that social categories can be based on one defining characteristic, such as being female or being a professor, and that what it means to be a member of a social category requires reference to traits, characteristics, and behaviors. Likewise, personal traits, at least implicitly, have meaning because they refer to a group. That is, they draw a similarity between the self and others who share the trait (e.g., "intellectual") and a difference between the self and others who do not share the trait. Empirical evidence also suggests that group and personal identities are not separate constructs in individuals' selfconcepts. For example, Reid and Deaux (1996) found that when responding to the task "Who am I?", individuals clustered their responses on the basis of conceptual meaning (e.g., "athleticism," comprising both group and trait terms) rather than on the basis of personal versus group identities. Based on the prevailing evidence, we make the simplifying assumption that the mechanisms through which group identities affect persuasion processes do not differ from those through which personal identities affect persuasion processes.

< previous page

page_173

next page >

< previous page

page_174

next page > Page 174

Identity can be made relevant in persuasion situations in various ways. The majority of previous empirical research relevant to identity effects has operationalized identity as a match between the group membership of a particular message source and the message recipient (e.g., the source and recipient are both female), or a match between the framing of the message arguments and an aspect of message recipients' self-concept (e.g., the message makes an appeal to some dimension that the recipient values such as an image appeal to a high self-monitor). However, it is important to note that identity can be operationalized in ways other than matching source or message features to the person's selfconcept. For example, the persuasion context itself can make identity salient (e.g., simply receiving the message in the presence of ingroup or outgroup members). Overview of Theories Explaining the Effects of Identity on Persuasion Previous approaches to studying the role of identity in attitude change have tended to focus on specifying just one process by which identity determines attitudes, with different theorists focusing on different processes. The assumption that variables tend to have one outcome on persuasion (e.g., appealing to one's identity is "good" for persuasion), and there is one process by which this outcome is achieved, is a common scenario for variables in the persuasion literature. Regarding identity effects, there appear to have been two bursts of theoretical activity addressing this question, the first in the 1950s and early 1960s, and the second in the 1980s and 1990s. One of the earliest relevant theories is Festinger's (1950) theory of social reality testing, which posited that individuals are influenced by similar others (e.g., ingroup members, or those who share one's traits) in forming judgments for which physical reality is low (e.g., attitudes) because people gain subjective validity that their opinions are "correct" to the extent that similar others agree with them. To the extent that similar others disagree, there is pressure to change attitudes to agree with those others. Festinger further argued that the impact of similar others on attitudes would increase, through this same process of gaining subjective validity, as the message recipients' attractiveness to (i.e., identification with) the others increased. In examining this hypothesis, Festinger and his colleagues (e.g., Back, 1951) found that the more cohesive a group was (i.e., the more attractive the group was to its members) the more its members influenced each others' attitudes about an ambiguous stimulus. Importantly, the theory of social reality testing posited that individuals' identifications have an impact on attitudes through one processthe process of social reality testing, gaining confidence in the subjective validity of one's judgments through agreement with similar others.

< previous page

page_174

next page >

< previous page

page_175

next page > Page 175

Alternatively, Deutsch and Gerard (1955) hypothesized that the process of influence associated with identity (particularly group identity) is that of normative social influence. That is, they suggested that ingroups exert normative social influence to a greater extent than do outgroups. Normative social influence is conceptualized as "an influence to conform with the positive expectations of another," where the fulfillment of positive expectations leads to positive feelings of solidarity, and unfulfillment leads to alienation (Deutsch & Gerard, 1955, p. 629). Replicating and extending the findings of Asch's line-judging conformity experiments (e.g., Asch, 1951, 1952), they found that participants conformed to the incorrect judgments of others (confederates) more than twice as often when they believed they and the others were part of a group than when they and the others were merely individuals participating in the experiment (Deutsch & Gerard, 1955). Thus, Deutsch and Gerard hypothesized that the unique process through which group memberships affect attitude change is through normative social influence, a process that could be assumed to be more likely the more an individual wants to feel solidarity with the group (i.e., the greater their identification with the group). One of the most well-known theories of identification is Kelman's (1958, 1961) three-process model. Kelman suggested that individuals are persuaded by expert sources because people accept the information on its merits (internalization), but information from attractive sources (whether attractive groups or individuals) is accepted because people want to establish or maintain a self-defining relationship with the source. Kelman argued that in adopting an attitude through this identification mechanism, the recipient is not so concerned with the content of the adopted attitude (as in internalization) but is satisfied by the act of agreement itself. Kelman's theory suggests that identification is more likely the more attractive the group or individual source is to the message recipient. Finally, Sherif and Hovland's (1961) social judgment theory of persuasion also suggested a process through which identity affects attitude change, although this was not their primary intent. Specifically, in much of their early research on the effects of ego-involvement, they operationalized involvement in terms of a person's group memberships. For example, for the issue of alcohol prohibition, highly ego-involved participants were members of the Women's Christian Temperance Union and Salvation Army (who held extremely pro-prohibition attitudes) and informal antiprohibition groups, whereas uninvolved participants were college students (who held moderately antiprohibition attitudes). Sherif and Hovland suggested that members of highly ego-involved groups would be more resistant to attempts to change their attitudes on group-relevant issues than members of groups for which the issue was not involving, because, among other things,

< previous page

page_175

next page >

< previous page

page_176

next page > Page 176

ego-involved individuals have larger latitudes of rejection than uninvolved individuals and thus are more likely to view discrepant messages as unacceptable. Although a number of studies have indicated that high ego-involvement (suggesting high group identity) can produce resistance to persuasion (at least from sources not affiliated with the group), the mechanism behind this effect is in some dispute (see Petty et al., 1992, for additional discussion). In the past several years, a second set of theories has been proposed to explain the process through which identity affects persuasion. The first of these modem approaches, which is most similar to traditional approaches, is the theory of referent informational influence (e.g., Turner, 1982, 1991; Turner & Oakes, 1989). A modified version of Festinger's (1950) theory of social reality testing, the theory of referent informational influence rests on the idea that for all judgments, regardless of their manifestation in physical reality, the subjective validity of (i.e., confidence in) one's views is a function of the extent to which similar others agree with those views. As a result of self-categorization as a member of certain groups and not others, ingroup members are perceived as similar to the self, and outgroup members as dissimilar. Once self-categorization occurs, the theory suggests the following: (a) we expect to agree with those who are functionally identical to ourselves (i.e., ingroup members); (b) when we do not agree with similar others, subjective uncertainty arises; and (c) we are influenced by the views of similar others because these views provide information about the social validity of our views, and we perceive that shared responses of ingroup members are objectively demanded (i.e., that they reflect objective reality). Shared responses refer to the prototypicality of the ingroup member's view.The more prototypical of the group (in Kelley's, 1967, attributional terms, the more consensual the view across members), the more influential the view. Thus, Turner and colleagues argued that the views of ingroup members are persuasive and lead to true internalized attitude change because they are taken as an indicator of reality for all judgments (including attitudes, beliefs, and the nature of the external physical world). On the other hand, outgroup members, with whom we do not expect to agree, are less likely to produce attitude change, and thus must resort to coercion to change behavior. Hogg and Abrams (1993) also suggested that people are motivated to reduce subjective uncertainty, and that uncertainty can be reduced by agreement with others who are relevant and similar to the self (e.g., ingroup members, those who share individual traits). Empirical tests have not supported the prediction by the theory of referent informational influence that increased subjective validity (i.e., confidence) will result from agreement with similar others on all judgments. Specifically, Goethals and Nelson (1973) predicted from Kelley's (1967)

< previous page

page_176

next page >

< previous page

page_177

next page > Page 177

theory of attribution that on judgments of verifiable beliefs or facts (i.e., "a potentially verifiable assertion about the attributes of an entity," p. 118), an individual's confidence that his or her judgment reflects a quality of the entity itself, rather than her or his idiosyncratic biasing characteristics, would increase more when a dissimilar other (one who has a different perspective) agrees than when a similar other (who shares the same biasing characteristic) agrees. Consistent with this prediction, they found that agreement with a dissimilar other (i.e., one who was shown to differ from participants in judging others) on a matter of a verifiable belief (i.e., which of two applicants would be more academically successful at the participant's university) led to greater confidence in one's judgment than did agreement with a similar other (i.e., one who judged others similarly; see also, Goethals, 1972). In contrast, Goethals and Nelson (1973) found that agreement with a similar other when making a judgment of an unverifiable belief or "value" (i.e., which of two applicants they personally liked better) led to greater confidence in judgment than did agreement with a dissimilar other (and to less attitude change, Berscheid, 1966), because individuals want to agree with those who share their perspectives when making judgments of preference. Thus, it appears that Festinger's (1950) theory of social reality testing has received better empirical support than the theory of referent informational influence, although more research is needed using groups for which identity is higher than in the existing research. Conover (1988) proposed the cognitive-affective model, an alternative contemporary theory outlining how identity affects political attitudes. This theory suggests that high identification with the ingroup, along with other individual and cultural factors, leads individuals to expend more energy on information relevant to the ingroup and to perceive and evaluate the ingroup in a positively biased manner (i.e., impacts on "political thinking"), leading to more favorable political attitudes, particularly on issues made chronically or temporarily relevant to the ingroup. Finally, van Knippenberg and Wilke (1992) proposed a similar information-processing theory regarding the process through which social identity affects persuasion. Specifically, they suggested that individuals are more interested in information expressing ingroup attitudes than information not about their group. In addition, individuals are postulated to perceive arguments that are more prototypical of the ingroup as stronger and more valid, and thus more persuasive, than nonprototypical arguments. Thus, persuasive messages that represent ingroup norms (i.e., are more prototypical of the ingroup) elicit greater favorable thought than do nonprototypical messages, and these lead to greater attitude change. Although similar to Conover's (1988) theory, this proposal suggests that individuals are more motivated to think favorably about information about the ingroup only

< previous page

page_177

next page >

< previous page

page_178

next page > Page 178

when it represents the "prototypical" views of the ingroup, rather than any information about the ingroup. To summarize thus far, many single-process theories have been proposed to explain how identity has an impact on persuasion. These theories can be organized into those that propose either a relatively nonthoughtful process (e.g., high identification with the message source leading to greater persuasion because agreement leads to satisfaction; Kelman, 1958, 1961) or a relatively thoughtful process (e.g., high identification with the ingroup motivating message recipients to engage in favorably biased thinking about information relevant to the ingroup; Conover, 1988) through which identity affects persuasion. The major assumption underlying our approach is that identity can affect attitudes in multiple ways (i.e., through multiple processes) under different specifiable conditions. This assumption is derived from the elaboration likelihood model (ELM) of persuasion (Petty & Cacioppo, 198la, 1986), which proposes that any variable in a persuasion context, including identity, can affect attitudes in four ways: (a) by acting as a simple cue, (b) by determining the extent of information processing, (c) by biasing the type or valence of thoughts that come to mind in response to information, and (d) by acting as an additional argument or piece of information. Diversity of Initial Findings on the Effects of Identity The notion that one's identity can have an important impact on the attitudes one holds can be traced back to the classic Bennington College study, in which students' political attitudes were found to become progressively more similar to the political leanings of fellow classmates (liberal) than their parents (conservative) from their freshman to senior year (Newcomb, 1943). Since then, many investigators found that persuasive messages lead to greater attitude change when they are presented by a source who shares the message recipients' group membership than when they are presented by a source who does not share this membership (e.g., McGarty, Haslam, Hutchinson, & Turner, 1994; Siegel & Siegel, 1957; Wilder, 1990) and when the message content matches rather than does not match an aspect of the message recipients' identity (e.g., DeBono & Packer, 1991; Snyder & DeBono, 1985; see Snyder & DeBono, 1989, for a review). In addition, it appears that the effects of identity on persuasion are increased under conditions in which it is more likely that the relevant identity is being activated for message recipients. For example, the persuasiveness of ingroup sources has been shown to increase as the salience of message recipients' ingroup membership increases (e.g., by having participants explicitly categorize

< previous page

page_178

next page >

< previous page

page_179

next page > Page 179

themselves as a member of the ingroup or as a nonmember of the outgroup before being exposed to a persuasive message, McGarty et al., 1994). However, under certain conditions, messages appealing to an identity the message recipient does not hold or presented by an outgroup member have been found to be equally, and sometimes more, persuasive than messages associated with the ingroup. For example, Freedman (1967) and White and Harkins (1994, Experiment 2) found that a source who belonged to the message recipients' own ethnic group (i.e., White) could be less persuasive than a source of a different ethnic group (African American or Hispanic). Wilder (1990) found that outgroup sources who were individuated (by giving recipients personal information about the outgroup sources such as their names, hometowns, and majors) were as persuasive as ingroup sources. In addition, Petty and Wegener (1998b) found that messages that matched recipients' identity were sometimes less persuasive than messages that did not match recipients' identity. Importantly, the conditions under which these different effects are likely to occur and the processes underlying them are only beginning to attract conceptual and empirical investigation, and cannot easily be explained by previous single-process theories. The complex and conflicting findings observed for identity present a similar situation to that which has existed for many other variables in the persuasion literature. For example, although much literature shows that increasing source credibility increases persuasion, some studies have shown that increasing source credibility can decrease or have no effect at all on persuasion (see Petty & Wegener, 1998a, for review). A complete understanding of the role of identity in persuasion requires an overall conceptual framework that can account for the diverse effects of identity and specify the processes by which these effects are obtained. The Elaboration Likelihood Model of Persuasion The ELM provides a theoretical framework for understanding and integrating the many seemingly conflicting findings in the persuasion literature. The ELM specifies four roles through which source, message, context, and other variables have an impact on attitude change (Petty & Cacioppo, 1986). The ELM rests on the assumption that attitude change can occur as a result of two relatively different routes to persuasion: an effortful route and a route that requires less cognitive engagement. That is, the ELM holds that sometimes people engage in a relatively effortful examination of all of the information relevant to judging the validity of the position advocated, but sometimes they do not. When people are thinking carefully about (i.e., elaborating) the issue-relevant information

< previous page

page_179

next page >

< previous page

page_180

next page > Page 180

presented in the persuasive message, they are said to be following the central route to attitude change. When attitude change is a result of relatively low cognitive effort, people are said to be following the peripheral route. According to the ELM, when people are highly motivated and able to think about the issue-relevant information presented, variables such as identity can have an impact on persuasion in two ways. First, identity can serve as a persuasive argument that provides issue-relevant information when people think about it. For example, the identity of a source could serve as an issue-relevant argument for a hangout location. That is, if ingroup members advocate a certain hangout location, their endorsement alone could serve as an issue-relevant argument because a central feature of a hang out place is that others central to one's identity (e.g., members of one's important groups) spend time and are welcome there. When people are engaged in effortful thought, making identity features salient also might bias the ongoing elaboration. For example, making identity salient might motivate people to selectively find the flaws in the arguments presented by sources who do not share their group membership (or contained in a message not framed toward an aspect of their self-concept). However, these people might selectively find the strengths in the arguments presented by sources who share their group membership (or contained in a message framed toward an aspect of their self-concept). Individuals engage in elaborative processing of persuasive information when they are motivated and able to do so. Thus, for example, individuals have been found to elaborate when a message topic is temporarily made personally relevant to them (e.g., Petty & Cacioppo, 1979b), and when there are no distractions present in the persuasion context (e.g., Petty, Wells, & Brock, 1976). If an individual is not motivated or is unable to attend to information in a persuasive message, then persuasion can still take place through minimal central route processing (e.g., processing only the first argument in a persuasive message) or by other qualitatively different processes that often depend on the presence of relatively simple cues in the persuasion context. Attitudes can be changed through this peripheral route through mere identification with a source without consideration of the merits of what the source has to say (Kelman, 1958, 1961), simple associations (e.g., as in classical conditioning; Staats & Staats, 1958; arm flexion-extension; Cacioppo, Priester, & Berntson, 1993), or the use of simple decision rules stored in memory (e.g., heuristics such as, "If the attitude object is associated with an aspect of my identity, it must be good;" Chaiken, 1987). Thus, when the elaboration likelihood is low in a specific context (e.g., when the personal relevance of the issue is low, and there are many distractions), identity, to the extent that it has any impact at all, would likely influence attitudes by serving as a peripheral cue. For example, under low elaboration

< previous page

page_180

next page >

< previous page

page_181

next page > Page 181

conditions, individuals might agree with a source of the same group membership more than a source of a different group membership, even when source group membership is not relevant to the issue (e.g., in an advertisement for an unfamiliar dessert). This might occur because individuals want to identify with (i.e., maintain a satisfying relationship with) the source through agreement and not because the merits of what the source has said were considered. Thus far, three roles that variables can have in persuasion processes have been outlined. Variables can serve (1) as persuasive arguments, (2) to bias thinking or attention to arguments, or (3) as a persuasive cue. According to the ELM, a fourth possible role exists. Variables can serve to determine the amount of elaboration. That is, they can serve to motivate or enable an individual to engage in thinking about a persuasive message. Put differently, many variables are capable of moderating the route to persuasion, either central or peripheral. Some variables, such as personal relevance (Petty & Cacioppo, 1979b), and need for cognition (Cacioppo & Petty, 1982), help to determine a person's level of motivation to elaborate. Other variables, such as message repetition (Cacioppo & Petty, 1989), and distraction (Petty et al., 1976), determine message recipients' ability to elaborate. Identity might serve to determine message recipients' motivation to elaborate by determining the personal relevance of the message, or it might determine their ability to elaborate by determining whether information relevant to evaluating the message (e.g., knowledge about identityrelevant issues) is called to mind. One of the most important features of the ELM is that any one variable can serve in multiple roles in persuasion. That is, a variable can serve as a peripheral cue in one context, as an argument in another, to motivate thinking in a third, and to bias thinking in a fourth. For example, source attractiveness has been found to: (1) serve as a peripheral cue when it was irrelevant to evaluating the merits of the attitude object (a typewriter), and the motivation to elaborate was low (Haugtvedt, Petty, Cacioppo, & Steidley, 1988); (2) motivate thinking about a persuasive message when the elaboration likelihood was not constrained to be high or low (Puckett, Petty, Cacioppo, & Fisher, 1983); and (3) serve as an argument when it was relevant to evaluating the merits of the attitude object (a new shampoo) and the elaboration likelihood was high (Petty & Cacioppo, 1980). Identity would serve in these multiple roles in similar situations. That is, identity would be most likely to serve as a peripheral cue when the elaboration likelihood was low, serve as an argument (if relevant to judging the attitude object) or bias processing when the elaboration likelihood was high, and determine the extent of processing when other factors have not already constrained the recipient's processing to be high or low (see Petty & Wegener, 1998).

< previous page

page_181

next page >

< previous page

page_182

next page > Page 182

Use of Argument Quality and Thought Listings in Uncovering Persuasion Processes A manipulation of the quality of arguments contained in a persuasive message and a measure of thoughts generated during message exposure are two convergent methods of determining the route, central or peripheral, through which persuasion has taken place, and, in turn, the role a variable has served in the persuasion process (see Petty & Cacioppo, 1986, for a description of the method by which messages differing in argument quality are generated, and additional procedures to assess the extent of elaboration). Individuals engaged in elaborative processing have been found to be more persuaded and to report more positive thoughts in response to strong than weak messages, whereas those relying on peripheral cues are less influenced in their attitudes and thoughts by message quality (Petty, Cacioppo, & Goldman, 1981) If the variable of interest is serving as a peripheral cue, then its effect on attitudes should occur regardless of argument quality. That is, a main effect of the variable on attitudes should be found, unmoderated by argument quality. In addition, the variable should have a direct effect on attitudes, unmediated by recipients' thoughts. If the variable serves to determine the extent of elaboration, then one should see a greater differentiation of both attitudes and thoughts in response to strong and weak arguments at the level of the variable eliciting greater thinking than at the level of the variable eliciting less thinking (i.e., a two-way interaction between the variable and argument quality should be obtained). If a variable is serving to bias elaboration, a main effect of a variable on attitudes and thoughts should occur, and the variable's effect on attitudes should be mediated by thoughts.2 Consequences Associated With Different Attitude Change Processes Finally, the process through which persuasion occurs (i.e., through the central or peripheral route) has been shown to have several consequences for the resulting attitudes. Specifically, attitudes formed or changed through the central route are relatively strong in that they exhibit greater persistence over 2As noted previously, a final possibility is that a variable can also serve as an argument when it is relevant to judging the merits of the attitude judgment. Whether or not a variable is serving as an argument can be determined through manipulating elaboration likelihood (e.g., high and low personal relevance of the topic), and the relevance of the variable to the topic. If a variable is serving as an argument, it should have an impact on persuasion under high elaboration conditions (e.g., high personal relevance) when it is relevant but not when it is irrelevant to the topic of the persuasive message. The possibility that identity can serve as an argument under high elaboration conditions when it is relevant to the attitude judgment has not yet been investigated and thus is not discussed further here.

< previous page

page_182

next page >

< previous page

page_183

next page > Page 183

time (e.g., Haugtvedt & Petty, 1992), resistance to counterpersuasive information (e.g., Haugtvedt & Wegener, 1994), and attitude-behavior correspondence (e.g., Cacioppo, Petty, Kao, & Rodriguez, 1986), than attitudes resulting from peripheral route processes, which induce relatively weak attitudes (see Petty, Haugtvedt, & Smith, 1995, for a review). Thus, the process through which persuasion occurs and the role a variable serves in persuasion are important. To return to our original voting example, attitudes impacted upon by identity are more likely to guide voting behavior at the polls when they have been formed through the central route than through the peripheral route. Identity Processes in the ELM Identity Under Conditions of Low Elaboration Likelihood If an individual has relatively low motivation and ability to process a persuasive message, then identity, to the extent that it has any effect at all, could serve as a peripheral cue. Some early investigators of the effect of shared characteristics (i.e., similarity) on persuasion assumed that all identity-induced attitude change occurred through simple associations. That is, in ELM terminology, some prior theories postulated that shared characteristics always served the role of a peripheral cue in persuasion. In addition, these theories explicitly proposed that identity-induced persuasion did not lead to long-lasting attitude change but was context specific (e.g., lasted only as long as the similar source was present, Deutsch & Gerard, 1955; Kelman, 1958, 1961). This proposal, of course, is consistent with the ELMpostulated consequences of peripheral change. Following the framework provided by the ELM, we propose that if identity is serving as a cue, it should lead to more favorable attitudes (i.e., serve as a positive cue) when the identity implications of the persuasive situation match that of the message recipient (e.g., the source, message, topic, or context share or are relevant to the identity of the message recipient) than when the identity implications of the persuasive situation are not salient. In the same vein, identity should lead to less favorable attitudes (i.e., serve as a negative cue) when the identity implications of the message explicitly mismatch that of the message recipient (e.g., the source, message, topic, or context share or are relevant to the identity of some group with whom the individual disidentifies) than when the identity implications of the persuasive message are not salient. Two recent investigations support the view that shared group membership with a source can serve as a peripheral cue in persuasion when the elaboration likelihood is relatively low. In one set of conditions of an experiment

< previous page

page_183

next page >

< previous page

page_184

next page > Page 184

using a topic that was irrelevant to the ingroup (acid rain in the northeastern United States for ingroup University of California at Santa Barbara [UCSB] students), Mackie, Worth, and Asuncion (1990) found that individuals tended to be more persuaded by an ingroup (a fellow UCSB student) than an outgroup (a University of New Hampshire [UNH] student) source when the source's issue position was stated before the message, regardless of the quality of the arguments contained in the communication. Results for participants' thoughts also were consistent with the interpretation that the persuasion observed was peripheral (i.e., nonthoughtful), as under these same conditions, participants' thought listings did not reliably predict their attitudes in response to the message by either the ingroup or outgroup source. Thus, it appears that an ingroup source served as a positive peripheral cue when the persuasive message was about a topic that was irrelevant to the ingroup, and the source's issue position was stated before the message. This finding was replicated in one set of conditions in a second investigation (Mackie, Gastardo-Conaco, & Skelly, 1992). This study found a marginal main effect for source group membership on attitudes, suggesting that the ingroup source tended to be more persuasive than the outgroup source when the source's issue positions were presented before persuasive messages about topics which were irrelevant to both the ingroup and outgroup (handgun control and euthanasia for ingroup UCSB and outgroup University of Manitoba students). It is plausible to suggest that the elaboration likelihood under these conditions was low. Specifically, a topic that was irrelevant to the ingroup was presumably of low relevance to message recipients. In addition, message recipients did not need to scrutinize the message to discern the source's position since the position was clearly stated before the arguments were presented. A third investigation provides support for the notion that personal identity also can serve as a peripheral cue in persuasion, when the topic is identity-irrelevant and the source issue-position is stated before the message. DeBono (1987) presented recipients with a message containing arguments on both sides of an issue but told recipients that one position (i.e., the mentally ill should be institutionalized) was consistent with an aspect of their self-concept (i.e., value or image-driven, for low and high self-monitors, respectively). Following exposure to the message, recipients expressed more agreement with the side they had been told was consistent with their identity than the other side of the issue. In addition, this study found that message-relevant thoughts were equally predictive of attitudes regardless of identity match, and these attitude results were replicated in a second study when no messages were presented to process (i.e., recipients simply were told which side was associated with their identity), suggesting that the message frame was capable of changing attitudes through a peripheral process.

< previous page

page_184

next page >

< previous page

page_185

next page > Page 185

These findings that individuals use information associated with an aspect of the self-concept as a guide in forming their own attitudes about identity-irrelevant issues rules out a possible explanation for why a match between persuasion setting and recipient characteristic can serve as a positive peripheral cue. Specifically, it does not appear that a match between persuasion setting and recipient characteristic serves as a cue because matched messages are necessarily seen to be more informative about the issue than mismatched messages. In fact, the Mackie et al. (1990) investigation suggests that an ingroup source (i.e., matched message) can serve as a cue even when the outgroup (i.e., mismatched message) might be expected to be more knowledgeable about the topic (i.e., acid rain in the northeast when the outgroup source was a UNH student). This research also suggests the conditions under which a match between persuasion setting and recipient characteristic can serve as a peripheral cue (i.e., on identity-irrelevant topics, when the issue position is stated before the message). However, although these findings suggest why a match between persuasion setting and recipient characteristic does not serve as a peripheral cue (i.e., it is not because the matched message is seen as more informative), they do not provide direct evidence as to why a match between persuasion setting and recipient characteristic does serve as a peripheral cue. We conducted an experiment to investigate whether identity with the group rather than group membership per se is important in producing cue effects for ingroups. Specifically, if identification is the mechanism for why a match between a persuasion feature (such as an ingroup source) and a recipient characteristic affects persuasion as a peripheral cue, then a group membership match per se should not necessarily lead to greater persuasion under low elaboration conditions. Rather, an ingroup source should be relatively more effective only for individuals who strongly identify with that particular ingroupthat is, those who have chosen the group as an identity (for a similar view, see Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). If shared group membership alone is sufficient for the cue effect, then strength of identification should not matter. To provide preliminary evidence regarding the hypothesis that shared group membership with a source would serve as a peripheral cue for those who highly identify with the group more than for those who are less identified with the group, college undergraduates who were either high or low in identification with their fellow undergraduates (Ohio State University [OSU] undergraduate students) as measured by the identity subscale of the Collective Self-Esteem scale (CSES; Luhtanen & Crocker, 1992) were recruited for a study (Fleming & Petty, 1997a). Participants were told the attitudes of both ingroup members (OSU students) and outgroup members (either University of Michigan students or Pennsylvania State University stu-

< previous page

page_185

next page >

< previous page

page_186

next page > Page 186

dents) about eight consumer products and then were asked to report their attitudes toward the products. The attitudes of ingroup members and outgroup members were always opposing. That is, if the ingroup was positive toward a product, the outgroup was negative. The eight items were chosen out of a larger set of products because they were shown in pretesting to be the most neutral (i.e., pretest participants reported attitudes that were closest to zero). Ingroup and outgroup attitudes were said to be an average of + 2 or -2 on a scale ranging from + 5 (very good) to -5 (very bad) and participants reported their attitudes on the same scale. Elaboration likelihood was kept low by deliberately not providing participants with any additional information about the products, other than the attitudes of ingroup and outgroup members, and by using attitude objects about which they had little or no previous knowledge (i.e., ''new and revived products'': banana-flavored Cheerios; balloon artwork; purple datebooks; theater in the round; painted barretts; sitcom about orphaned children; reruns of the TV show "Get Smart"; and a plastic car antenna). These attitude objects were also chosen to be irrelevant to either the ingroup or outgroups. Thus, the experiment replicated the set of conditions under which shared group membership has previously been found to act as a peripheral cue. In sum, the only information participants had available on which to base their judgments was the attitudes of ingroup and outgroup members. The design of the experiment was a 2 (Identification with ingroup: High vs. low) X 2 (Ingroup attitude: Favorable vs. unfavorable) factorial, with one between and one within participant factor. Participants' attitudes toward the four objects for which the ingroup attitude was favorable and their attitudes toward the four objects for which the ingroup attitude was unfavorable were each averaged for each participant, resulting in two composite attitude scores. A two-way mixed analysis of variance (ANOVA) with one between- and one withinparticipant factor conducted on these scores revealed two main effects. First, those high in identification reported more favorable attitudes (M = .70) than did those low in identification (M = -1.20), F(1, 22) = 6.65, p < .05. Second, the attitudes of participants were more favorable when the ingroup attitude was positive (M = .51) than negative (M = .60), F(1, 22) = 12.38, p < .01. More importantly, these main effects were qualified by the anticipated interaction, F(2, 22) = 4.53, p < .05 (see Fig. 10.1). This interaction suggested that whereas participants who were high in identification with the ingroup were more persuaded when the ingroup position was positive (M = 1.16) than when the ingroup position was negative (M = -.46), F(1, 13) = 15.77, p < .01, the attitudes of those who were low in identification with the ingroup were not affected by ingroup position (Mingroup positive = -.40, Mingroup negative -.80, F(1, 9) = 1.20, p =.30, n.s.).

< previous page

page_186

next page >

< previous page

page_187

next page > Page 187

These findings provide initial evidence that individual differences in the level of identification with the ingroup moderates when an ingroup source serves as a peripheral cue. Specifically, those high in identification with the ingroup reported more favorable attitudes when the ingroup attitude was favorable toward the attitude object than when the ingroup attitude was unfavorable. On the other hand, those low in identification with the ingroup did not appear to use the ingroup position as a cue in determining their attitudes. Their attitudes were not more favorable when the ingroup position was favorable than when it was unfavorable. Identity Under Conditions of High Elaboration Likelihood In stark contrast to when the elaboration likelihood is low, when the elaboration likelihood is high, individuals are motivated and able to process the issue-relevant arguments presented and evaluate their merits. According to

FIG. 10.1. Effect of ingroup positivity and group identity on attitudes.

< previous page

page_187

next page >

< previous page

page_188

next page > Page 188

the ELM, under high elaboration conditions, identity can serve two roles. First, identity can serve as an argument for the merits of the attitude object if it is relevant to the perceived merits of the object, such as in our previous example of perceiving an ingroup source as a favorable argument for a hang out location. In this example, it is assumed that identification with the source was explicitly relevant to the merits of the attitude object and thus could be used as an item of issue-relevant information. Identity can play a second role in persuasion under high elaboration conditions, even if it is substantively irrelevant to the merits of the attitude object. Specifically, it can bias the thoughts that are generated in response to a persuasive message. This biased processing explanation has been offered as one account for why messages that match the functional basis of one's attitude are more persuasive than messages that do not match. For example, Lavine and Snyder (1996) found that messages with content that matched an aspect of the recipients' personal identity (i.e., image arguments for high selfmonitors, and quality appeals for low self-monitors) elicited more favorable thoughts and were rated as more persuasive than did messages with content that mismatched recipients' personal identities. In addition, thoughts and ratings of persuasiveness were found to mediate the impact of identity-match on post-message attitudes. Consistent with the prediction of the ELM that a variable will serve to bias elaboration under high elaboration likelihood conditions, Lavine and Snyder (1996) suggested that it was likely that in their research "processing motivation was relatively high" (p. 600). To examine whether identity serves to bias message processing, we designed a study based on the procedure Lord, Ross, and Lepper (1979) developed to examine bias. Lord et al. (1979) exposed participants who initially either supported or opposed capital punishment to two studies on capital punishment, one seemingly confirming and one disconfirming their existing attitude. They found that participants rated the report that supported their attitude as more convincing than the report disconfirming their belief and reported corresponding attitude change in the direction of their initial attitude (i.e., attitude polarization). Lord et al. (1979) interpreted these results as suggesting that individuals evaluated the evidence presented in a biased manner, accepting evidence confirming their initial attitude at face value but criticizing and finding flaws in evidence disconfirming their initial attitude. However, unlike Lavine and Snyder (1996), Lord et al. (1979) did not obtain evidence that such biased message processing occurred (e.g., in thought listings) but rather showed biased outcomes in the form of judgments of the evidence presented on each side of the issue and reports of attitude change. Thus, there are several possible explanations for Lord et al.'s results. That is, the biased outcomes could have resulted from (a) favorably biased processing of the confirming or unfavorably biased processing of the

< previous page

page_188

next page >

< previous page

page_189

next page > Page 189

disconfirming evidence; (b) increased objective processing (e.g., due to greater personal relevance) of only the confirming information which would lead to enhanced persuasion if the arguments were strong; or (c) using one's prior attitude as a cue to judge the worth of the evidence on each side without extensive scrutiny of it. In our experiment, we attempted to demonstrate two points. First, we attempted to replicate and extend the findings of Lord et al. (1979) and Lavine and Snyder (1996) by investigating whether identity could lead to a biased attitudinal outcome. Specifically, we investigated whether associating some information with message recipients' ingroup but associating other conflicting information with message recipients' outgroup would lead to a biased attitudinal outcome in the direction supported by the information associated with the ingroup. Second, we examined whether this effect would be most apparent for those who were highly identified with the ingroup. Third, we attempted to tease apart the mechanism for the effect. In our study, male and female college undergraduates who were either high or low in identification with their gender group (males or females, respectively) as measured by the identity subscale of the CSES (Luhtanen & Crocker, 1992) were presented with two messages of equal strength about a product, one positive and one negative (Fleming & Petty, 1997b). Background elaboration likelihood was kept relatively high by keeping the persuasion context free from distraction and explicitly instructing participants that (a) they had ample time to complete the study (they had only one brief packet to complete during the session), and (b) they would be asked about their reactions to the persuasive messages contained in the booklet. Hastie and Park (1986) found that informing individuals they will be asked to make an evaluative judgment before they receive information elicits online elaboration, whereas informing individuals about the judgment after they receive information elicits memory-based processing. In addition, it was thought that telling participants they would be asked to answer questions about the messages would elicit at least a moderate level of accountability, which has been found to increase elaboration (e.g., Petty, Harkins, & Williams, 1980; Tetlock, 1983). Once they had been exposed to the persuasive messages, participants were asked to report their attitudes toward the product and the thoughts they had while reading the messages. Of the two messages that participants received, one message described women's reactions to the product and one described men's reactions (e.g., "Snickerdoodles are [wo]men's favorite snack food." "[Wo]men are concerned that snickerdoodles are unhealthy."). This identity framing was counterbalanced across message valence. Thus, each message recipient (male or female) received two messages consisting of either (a) one message giving the reasons why women were positive towards the product and one giv-

< previous page

page_189

next page >

< previous page

page_190

next page > Page 190

ing the reasons why men were negative, or (b) one message giving the reasons why women were negative towards the product and one giving the reasons why men were positive. Pretesting of the positive and negative messages, without the identity framing, revealed that each message was equally persuasive and elicited the same proportion of thoughts equal in valence to the message valence (i.e., positive attitudes and thoughts in response to the positive message and negative attitudes and thoughts in response to the negative message that were as discrepant from neutrality in an unfavorable direction as the positive attitudes and thoughts were in a favorable direction). Thus, if participants engaged in objective processing of both the positive and negative messages, regardless of identity framing, they should report relatively neutral attitudes and the same number of positive as negative thoughts. However, if participants engaged in biased processing, then a two-way interaction between participant gender and identity frame should be obtained on both the attitude and thought measures, indicating that identity framing biased thoughts as well as attitude outcomes. Specifically, men should be more persuaded and report a greater proportion of positive thoughts when the positive message is framed for men than women, and women should be more persuaded and report a greater proportion of positive thoughts when the positive message is framed for women than men. Finally, if biased processing of identity-framed information is due to identification with the ingroup, then a three-way interaction of participant gender, identity frame, and identification with gender should be found on the attitude and thought measures, indicating that biased processing and a biased attitudinal outcome is greater for highly identified individuals than those who are relatively unidentified with their gender. Participants reported their attitudes on five attitude scales (one 11-point scale and four semantic differentials), with higher numbers on each indicating more positive attitudes. Responses to the five attitude items were standardized and averaged for each participant. To provide a more sensitive test of moderation by identification level, analyses were conducted including only those whose level of identification with gender fell in the top or bottom quartile of scores of those who participated in the experiment. Thus, a 2 (Participant gender: Male vs. female) X 2 (Identity framing: Men positive vs. women positive) X 2 (Message order: Positive first vs. negative first) X 2 (Identification with gender: Highest vs. lowest) ANOVA was conducted on the resulting attitude index. This analysis revealed a significant threeway interaction, F(1, 62) = 10.82, p < .01. (see Fig. 10.2) This interaction was examined by analyzing the Participant gender X Identity framing interactions for high and low identified participants separately. For participants who were highly identified with their gender, a two-way interaction of Gender X Identity framing was found, F(1, 34) = 11.80, p


< previous page

page_191

next page > Page 191

.01 (see Fig. 10.2, top). This interaction showed that whereas men were more persuaded when the positive message was framed for men (M = .57) than women (M = -.27), women were more persuaded when the positive message was framed for women (M = .43) than men (M = -.58). For people who were low in identity with their gender, no effects were significant (all ps > .20; see Fig. 10.2, bottom). Thus, it appears that a biased attitudinal outcome was produced by identity framing of messages, such that individuals were more persuaded by the identity-matched than the identitymismatched

FIG. 10.2. Effect of gender, gender identification, and message type on attitudes.

< previous page

page_191

next page >

< previous page

page_192

next page > Page 192

messages, and that this biased outcome occurred primarily for highly identified individuals. To determine whether this biased attitudinal outcome was due to biased elaboration, a thought positivity index was created by subtracting the number of negative from the number of positive thoughts for each participant and dividing by their total number of thoughts. Thus, higher numbers indicated a greater proportion of positive thoughts about the product. Analysis of this measure produced a marginal three-way interaction, F(l, 61) = 2.78, p < .10. Separate analyses of the Participant gender X Identity framing interaction for high and low identified participants revealed that, paralleling the attitude data, the two-way interaction of Participant gender X Identity framing was marginal for highly identified participants, F(1, 33) = 3.19, p < .09. This interaction suggested that whereas men had a greater proportion of positive thoughts when the positive message was framed for men (M = .19) than women (M = -.11), women had a greater proportion of positive thoughts when the positive message was framed for women (M = .12) than men (M = .24). For low identified participants, no effects were significant. This investigation suggests that identity can serve to bias processing of mixed (i.e., both positive and negative) persuasive information, thus leading to a biased attitudinal outcome. If biased elaboration occurred for at least some of our highly identified participants, there remain four possible explanations that can account for the results of this study and the research by Lavine and Snyder (1996), because there are four general ways in which a variable can bias elaboration. One way is by influencing which of the available pieces of issue-relevant information in the persuasion context is attended to and elaborated. If some information in a persuasion context is associated with an identity, but other information in the same context is not (as occurred in both experimental paradigms), the former might be given greater attention than the latter. This situation could lead to a biased outcome through objective processing of only part of the available information. The bias would be in the selective attention given to some information over other equally available information. If evidence of both sides of an issue is equally strong, people would mostly notice the strength of the information they chose to process (see subsequent discussion of identity and extent of processing). In this study (and in that of Lord et al., 1979), it is possible that highly identified recipients elaborated the information that was framed for ingroup members more than the information framed for outgroup members and thus were more persuaded by it. A second way a variable can bias elaboration is to bias the type or direction of thoughts that come to mind, that is, to bias the elaboration itself. As suggested by others (Mackie et al., 1990), perception of shared identity might affect the type of thoughts that come to mind most quickly; specifi-

< previous page

page_192

next page >

< previous page

page_193

next page > Page 193

cally, it might increase the accessibility of category- or trait-relevant information. Because neither the topic used in our study nor that used by Lavine and Snyder (1996) were relevant a priori to the aspect of the self-concept being appealed to, this explanation seems less plausible. In addition, however, identity might motivate a certain direction of thoughts in response to a message. Perception of shared identity might lead to predominantly favorable thoughts in response to a message (bolstering), while perception of unshared identity might lead to predominantly unfavorable thoughts (counterarguing), leading to a biased attitudinal outcome. This type of biased processing should be especially evident if message arguments are somewhat ambiguous and open to multiple interpretations (e.g., either positive or negative), and the message recipient does not have highly accessible attitudes and beliefs about the attitude object that are counter to the position of the identity-matched message (see Petty, Schumann, Richman, & Strathman, 1993, for an example of mood biasing elaboration under these conditions; see Chaiken & Maheswaran, 1994, for an example of source expertise biasing elaboration under these conditions). It is plausible that, in our study, an appeal to identity with ingroup-framed versus outgroup-framed messages biased the direction of the thoughts of highly identified individuals (see also Cacioppo, Petty, & Sidera, 1982). For example, individuals might have thought about both messages but were motivated to agree with the ingroup (and/or disagree with the outgroup) source because they wanted to identify with the ingroup, leading them to have thoughts that were predominantly biased in a direction consistent with the ingroup framed position and in disagreement with the outgroup framed position. The final two possible explanations for the results obtained are that biased elaboration could have occurred in response to the matched message only (i.e., bolstering of matched arguments but no counterarguing of mismatched arguments), or that biased elaboration occurred in response to the mismatched message only (i.e., counterarguing of mismatched arguments, but no bolstering of matched arguments). Each of the four possible ways of engaging in biased elaboration would result in the pattern of results obtained; that is, more positive attitudes and thoughts in response to the matched than mismatched message. Future research is needed to determine which of these processes is responsible for the biased attitudinal outcomes we have observed for group identity. Identity Under Conditions of Moderate Elaboration Likelihood Often, the elaboration likelihood is not constrained to be high or low by other variables in the persuasion context, such as personal relevance or distraction. When the elaboration likelihood is moderate, identity is predicted

< previous page

page_193

next page >

< previous page

page_194

next page > Page 194

to influence the extent of elaboration. This enhanced thinking could be relatively objective, or it could be biased as explained previously. This prediction differs from the theories of Conover (1988) and van Knippenberg and Wilke (1992), which predict that identity influences the amount of thinking in which recipients engage, regardless of the elaboration likelihood, and do not postulate other roles for identity. Recent studies by Mackie et al. (1990), Mackie et al. (1992), and Petty and Wegener (1998b), have shown that, under moderate elaboration likelihood conditions, a match between persuasion setting and recipient characteristic can determine the extent of thinking rather than serve as a peripheral cue. Specifically, it appears that individuals process a persuasive message that is matched with an aspect of their self-concept more than a message that is mismatched with an aspect of their self-concept. First, Mackie et al. (1990) found that, when participants were exposed to a message from an ingroup source (a fellow UCSB student) or an outgroup source (a distant UNH student) about an ingroup-relevant topic (oil drilling off the southwestern coast of the United States), a two-way interaction of source group membership and argument quality on postmessage attitudes occurred. This interaction revealed that individuals showed greater attitudinal differentiation in response to strong than weak arguments presented by an ingroup than an outgroup source. In addition, they found that the favorability of participants' thoughts as assessed by a thought-listing predicted the extent of pre- to postmessage attitude change for participants who had been exposed to the message by an ingroup source but did not predict the extent of attitude change for those who received the message by an outgroup source. Thus, it appears that participants processed a message by an ingroup source more than an outgroup source when the topic was ingroup-relevant. Importantly, and consistent with the ELM, shared group membership served to determine elaboration likelihood when the background elaboration likelihood was relatively moderate (an ingroup-relevant topic, source issue-position stated before the message), but, as reviewed in the previous section, served as a peripheral cue when the background elaboration likelihood was relatively low (an ingroup-irrelevant topic, source issue-position stated before the message). In one set of conditions in a second investigation, Mackie et al. (1992) found that greater processing of an ingroupirrelevant message from an ingroup versus outgroup source can occur when the advocated position is stated after the message. Under these conditions, people must process the message to determine the position taken by the ingroup member. Thus, the ingroup position cannot easily be used as a cue when it is not stated prior to the message arguments. Two studies by Petty and Wegener (1998b) suggest that a personal identity-appeal can also increase the extent of message processing under condi-

< previous page

page_194

next page >

< previous page

page_195

next page > Page 195

tions of moderate elaboration likelihood. They gave strong or weak persuasive messages about four products (a shampoo, coat, shoe, and toothpaste) to high or low self-monitors that either matched an aspect of their self-concept (image arguments for high self-monitors and quality arguments for low self-monitors) or mismatched an aspect of their self-concept (quality arguments for high self-monitors and image arguments for low self-monitors). The persuasive messages contained pieces of information about the products and did not contain a summary position statement. They found greater differentiation of attitudes in response to strong and weak arguments when the message matched rather than mismatched recipient personal identity, especially when elaboration likelihood was not at a ceiling (i.e., for low but not high need for cognition individuals; Cacioppo & Petty, 1982). Thus, the existing literature provides evidence that both group and personal identity-appeals can determine elaboration likelihood when the likelihood of thinking is not constrained to be high or low by other variables. A match between persuasion setting and recipient characteristic can lead to increased processing because it increases the motivation to process, the ability to process, or both. Specifically, consistent with Festinger (1950) and Turner (1982, 1991; Turner & Oakes, 1989), Mackie et al. (1990) suggested that the attitudes of similar others might be more interesting because they provide information about the subjective validity of people's views, increasing motivation to process. It is also possible that matching the persuasion setting with a recipient characteristic increases the personal relevance of the persuasive information, increasing motivation to process (Petty & Cacioppo, 1990). Alternatively, a match could activate category- or trait-relevant information, increasing the ability to process other category or traitrelevant information contained in the message. People who strongly identify with a group (or trait) would presumably find information associated with the group or trait to be especially relevant (thereby motivating processing) and might also have more readily accessible information relevant to the group or trait (thereby enabling processing). Future research should explore these issues. Summary It appears that there is evidence both across separate studies, and within a single study, for the view that identity, both personal and group, can serve multiple roles in persuasion settings. Specifically, the two studies reported here suggest that identity can serve (a) as a peripheral cue under low elaboration conditions (e.g., Fleming & Petty, 1997a), and (b) to bias elaboration under high elaboration conditions (Fleming & Petty, 1997b). In addition, two investigations by Mackie and her colleagues suggest that shared group

< previous page

page_195

next page >

< previous page

page_196

next page > Page 196

membership, and thus perhaps identity, can serve more than one role in persuasion depending on the elaboration likelihood (Mackie et al., 1990, 1992). Specifically, they showed that shared group membership acted as a cue under low elaboration conditions (when the topic was ingroup-irrelevant and participants did not have to elaborate the message to determine the ingroup source's position), and served to motivate processing under moderate elaboration conditions (when the topic was ingroup-relevant, or elaboration was necessary to determine the ingroup source's position). It seems plausible to suggest that recipients who are highly identified with the ingroup might be especially likely to process information by ingroup more than outgroup sources under these conditionsjust as they are more likely to show cue and biased processing effectsbut future research will have to verify this. Future Research on the Multiple Roles of Identity on Persuasion As described in this chapter, preliminary support for the view that identity can serve in at least three roles in persuasion situations is now available. Even though the existing research provides a coherent yet complex pattern of results, considerable additional work is needed before the multiple roles for identity in persuasion are understood completely. The conceptualization and evidence presented here have emphasized the overall elaboration likelihood of a persuasion context as an important determinant of the process by which identity modifies attitudes. Other variables not previously applied to this domain also deserve attention as factors influencing identity in its multiple roles in persuasion (i.e., impact on its effectiveness as a peripheral cue, in motivating elaboration, in serving as an argument, and in biasing elaboration). For example, research suggests that groups that are optimally distinctive induce higher levels of identification than do groups that are too broad or too narrow (Brewer, 1991). Thus, appeals to group identity could be more effective in these roles when the group identity appealed to is one that message recipients feel is optimally distinctive than when the group identity appealed to is either too inclusive or too distinctive. Alternatively, appeals to group identity might be more effective when message recipients themselves are satisfying both the need for affiliation and the need for distinctiveness within a group than when individual message recipients either feel a loss of individuality, or a lack of belonging within a group. Thus, optimal distinctiveness as both a group characteristic and as an additional individual difference variable deserves attention as a factor that can influence the effectiveness of group identity in its multiple roles in persuasion.

< previous page

page_196

next page >

< previous page

page_197

next page > Page 197

In addition to optimal distinctiveness, the studies we described each investigated recipients' identification with actual group memberships or with traits they actually held. Investigation of whether actually possessing a group membership or trait is a necessary determinant for identity to serve in its multiple roles would be interesting. Specifically, Higgins' (1989) self-discrepancy theory proposes that discrepancies between the actual self (the qualities or traits one possesses) and the ideal self (the qualities or traits one would ideally like to possess), lead to negative affective states (e.g., sadness, disappointment, and dissatisfaction) and a motivation to attain a match between the actual and ideal selves that is manifested in action or increased information processing about information that might help to reduce the discrepancy. Although Higgins' theory focuses on discrepancies between ideal and actual traits one holds (i.e., personal identities), discrepancies between the actual group identity of an individual (the extent to which a group membership actually forms part of the self-concept) and her or his ideal group identity (the extent to which an individual would like a group membership to form part of the self-concept) might occur as well. These group identity discrepancies also might motivate individuals to behave, process information, and ultimately adopt attitudes that facilitate attainment of a match between the actual and ideal group identity. If so, it is possible that individuals who are not members of a group or who do not hold a trait (or those who are low in identification with a group membership or trait) but who ideally would like to be highly identified with the group or trait might show the effects outlined here. That is, we might find identity acting as a cue, as an argument, to bias or to motivate processing for those who do not have group membership or a trait or who are low rather than high in actual identification with the particular group or trait (but who would ideally like to be highly identified), because these individuals are motivated to reduce the discrepancy between their actual and ideal identity(ies). Unthinkingly agreeing with those of a particular identity, thinking more about information associated with those of a particular identity, and being positively biased toward information associated with those of a particular identity could each be mechanisms by which low identified individuals or those who do not possess a trait or group membership might strive to attain their preferred identity state. Thus, both situational (e.g., group optimal distinctiveness) and individual difference (e.g., large actual-ideal identity discrepancy) factors concerning the particular characteristic being appealed to could determine the effectiveness of identity in persuasion. The overview provided here can be viewed as a model answering the question of how identity affects persuasion. One set of questions left to be answered is why identity serves each of its multiple roles in persuasion. Several mechanisms have been proposed in the

< previous page

page_197

next page >

< previous page

page_198

next page > Page 198

literature, and several might operate, one or more for each role. Although we have reviewed a number of possible mechanisms in this chapter to account for why identity serves in the roles that it does, other mechanisms are possible. For example, identity might affect persuasion processes because (a) identity is like a value (e.g., I value my association with this group or trait), and making any value salient might lead to greater persuasion through multiple roles; (b) message recipients hold a belief that similar others' view of reality is more accurate, and any variable associated with greater accuracy will lead to greater persuasion through multiple roles (e.g., Festinger, 1950); (c) similarity of any kind between a source, message, or persuasion context and a message recipient, not just in identity (e.g., in characteristics that do not form part of the self-concept such as owning a green toothbrush, or temporary similarity such as having the same outcome on a particular task), leads to greater persuasion through multiple roles, perhaps through eliciting positive affect; (d) a match of any kind between a persuasion setting and aspect of the message recipient, not just in identity (e.g., between the affective or cognitive nature of an appeal, and the basis of recipients' attitude; Edwards, 1990) leads to greater persuasion through multiple roles, perhaps because the ''root'' of the attitude is being attacked; (e) liking a source or set of individuals an appeal is framed towards for any reason, not just shared identity, leads to greater persuasion through multiple roles, because we generally feel more comfortable agreeing with those we like (e.g., Heider, 1958); or finally, (f) referencing the self in any way in the persuasion setting (e.g., by embedding the pronoun you instead of one in message arguments; Burnkrant & Unnava, 1989) increases persuasion through multiple roles, through increasing self-relevance. By answering some of the questions remaining regarding the multiple roles of identity in persuasion processes, we gain a clearer understanding of the role our self-concepts (at the group and individual levels) serve in forming and modifying our judgments and in guiding our behavior. Finally, when addressing the broader question of how the persuasion and identification literatures relate to one another, it becomes evident that the persuasion literature has much to add to our understanding of the psychological mechanisms through which identification itself occurs. Following traditional attitude theory, Deaux (1996) suggested that identification has three components: cognitive (e.g., self-categorization as holding a particular group membership or trait), affective (e.g., the value and emotional significance placed on the particular membership or trait; Tajfel, 1981), and behavioral (e.g., interdependence of action of those sharing the membership or trait). If identity is like an attitude, then it should be possible for identification with a group or personal identity to be reached either through relatively thoughtful or nonthoughtful processes. Identifications that are

< previous page

page_198

next page >

< previous page

page_199

next page > Page 199

adopted through a relatively thoughtful process then might be more likely than identifications adopted through a relatively nonthoughtful process to persist over time, resist counterattack (e.g., appeals to disidentify and/or adopt a different identity), and predict behavior.

< previous page

page_199

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_201

next page > Page 201

11 Altercasting as an Influence Tactic Anthony R. Pratkanis University of California, Santa Cruz Remember, a con artist can be anyone: your minister, doctor, or travel agent, a loyal employee, business consultant, or janitor, or perhaps, the love of your life. A con artist will play any role in order to win your confidence and trust. Whitlock, (1997, p. 25). Every day, people of all walks of life fall prey to the swindles and scams of con artists. In the United States alone, consumers lose an estimated $40 billion a year to just telemarketing fraud. The successful con artist is one who is skilled at being an impostor, assuming whatever role or identity needed to fleece the mark (or victim). Consider these classic but common con games (see Henderson, 1994; Schulte, 1995; Whitlock, 1994, 1997). Some Classic Cons Three-Card Monte A con and a shill (confederate) play a betting game in which the shill must pick a certain card among three face-down playing cards. The mark watches the game until he or she is lead to "discover" a trick that will guarantee a winfor example, one of the cards is bent slightly or the con always deals in a certain way. Feeling as if he or she is the expert, the mark takes a turn, bets big, and then, of course, loses. Through a sleight of hand, the con had removed the bent card or dealt the cards in a different manner. The mark slinks away embarrassed and with a lighter wallet.

< previous page

page_201

next page >

< previous page

page_202

next page > Page 202

Bank Examiner Swindle In this scam, the con artist poses as a bank official or a police officer (appearing in uniform with a badge). The con asks the mark (typically an elderly woman with a large bank account) to assist in capturing a bank teller who is defrauding customers. The "police" need her to remove a certain sum of money from her account, place the cash in an unmarked envelope, and turn it over so that it can be checked for finger prints and serial numbers. Eager to be a good citizen and to assist an authority, the mark complies with the request, and the con, money in hand, beats it to the next town. The Payoff The mark may be sitting at a bar when he or she overhears two con artists talking about all the loot they won on the horses. Curiosity piqued, the mark finds a way to enter the conversation, and, after much hemming and hawing, the cons reveal that the races are fixed and that they are receiving inside information from a messenger. (In a variant called "the rag" the stock market replaces the horses). Feeling like an insider with special knowledge, the mark begs to enter the game. The cons let the mark win a few low-stakes races and then encourage the mark to bet big. The mark then loses a rigged bet. The payoff and rag are often coupled with the phony cop blow-off. In this scam, confederates of the con artists dress up as cops (or government agents) and raid the betting parlor (or stock broker office) just after the mark loses. By luck the mark is able to escape the long-arm of the phony law. Feeling like a criminal, the mark never mentions again how he or she was taken nor reports it to the real police for fear of incrimination. The Sweetheart Scam In this caper, the con artist finds a mark who is vulnerable to the overtures of romancesay a new widow(er) or divorce(e). The con then sweeps the mark off her or his feet. The con listens attentively to the mark for hours, brings flowers and candy, builds the mark's ego, agrees with anything the mark says, and does anything necessary to get the mark to think, "Finally, I have found my Mr. or Ms. Right." With "love" blossoming, the con artist then asks for a loan or some money or a shared credit card or other favors. Blinded by love, the mark delivers. Who can refuse a lover and soulmate? Once the mark is bled dry, the con leaves town and goes to the next scam. The Pyramid or Ponzi Scheme The mark is offered the business opportunity of a lifetimefast, easy money from oil futures, playing the foreign stamp market, or selling anything from cosmetics to cleansers to New Age herbs. All the mark needs to do is to join

< previous page

page_202

next page >

< previous page

page_203

next page > Page 203

the company by paying an investment fee and buying a sales kit. The fee will be returned as the mark recruits more and more victims to invest in the phony company. Co-opted into the role of company representative, the mark usually speaks highly of the investment scheme to friends and neighbors, urging them to get in on the plan. Eventually, the scheme collapses, and all the marks lose their money. Why do these scams work? Each of these swindles employs a number of effective influence devices identified by social psychologists (Cialdini, 1984; Pratkanis & Aronson, in press)for example, deals are presented as scarce, communications are overheard, the con does a favor for the mark to instigate the norm of reciprocity, commitments are induced, and emotions such as loneliness, greed, guilt, confidence, trust, and fear are played on. However, at the heart of each scam is the altercastthe placing of the mark in a role that requires the mark to do just what the con artist wants. What Is Altercasting? In 1963, Weinstein and Deutschberger introduced the term altercasting as a means of understanding interpersonal control. According to Weinstein and Deutschberger (1963, 1964), altercasting describes a social interaction in which an ego (in our examples, the con artist) adopts certain lines of action (the con game) to place an alter (the mark) into a role that specifies an interpersonal task (the scam's payoff). Altercasting works as an influence tactic because social roles serve to structure interaction in a particular manner (Biddle, 1979; Biddle & Thomas, 1966; Stryker & Statham, 1985). A role is "a set of mutual (but not necessarily harmonious) expectations of behavior between two or more actors, with reference to a particular type of situation" (Goode, 1968, p. 249); it represents what a person is supposed to do in a given situation because that person occupies a particular status in society (e.g., teacher or daughter), holds an informally defined position in interpersonal relationships (e.g., leader or hothead), or is identified with a particular value in society (e.g., tidy student or patriot; Coutu, 1951; R. Turner, 1956). When a person accepts a social role (either through altercasting or other means) a number of social pressures are brought to bear to insure the role is enacted. First, others in the social environment expect the alter to behave in a manner consistent with the role; in other words, the role establishes social processes associated with the self-fulfilling prophecy (Darley & Fazio, 1980; Merton, 1957a). Second, the role provides the alter with selective exposure to information consistent with the role (e.g., the police officer understands crime differently from the way that the criminal and the professor of criminology do); this differential knowledge serves as the basis of differential ac-

< previous page

page_203

next page >

< previous page

page_204

next page > Page 204

tion. Finally, a role is associated with a social identity that provides the alter with a stake in a given social system. Altercasting can serve as an influence tactic by using these pressures to have the alter accept and engage in an interpersonal task or set of role behaviors that are exactly the sort of behavior desired by the ego. For example, in threecard monte, the payoff, and the rag, the alter is like an expert who cannot lose. As an expert, the alter is confident in her or his judgments and does not question what is going on (after all, the alter is the expert). Alter acts on this false expertise and is conned. In the bank swindle, the alter is placed in the role of agent to authorities. We have learned not to question authorities and to be helpful to them. In the sweetheart scam, alter is placed in the role of intimate and does what every lover doesgives unquestioningly to the beloved. In pyramid schemes, alter is co-opted into being a representative of the company. As a representative, alter is required to sell the company to others, thus making a public commitment to the role. In the phony cop blow-off, the alter becomes a criminal. Criminals never go to the police (especially if it implicates oneself) and thus the con's purposes are served. The Goals of This Chapter This chapter has four purposes. Original research by Weinstein, Deutschberger, and their colleagues (Beckhouse, Tanur, Weiler, & Weinstein, 1975; Weinstein & Deutschberger, 1963, 1964; Weinstein, Wiley, & DeVaughn, 1966) focused primarily on the nature and dimensions of altercasting. However, over the last two decades, considerable research evidence has accrued to support the proposition that altercasting is an effective influence tactic (although much of this research was not done to test altercasting theory). One purpose of this chapter is to review some of this research. Second, Weinstein and Deutschberger (1963, 1964) viewed altercasting as a fundamental social process not limited to, say, the scams of con artists. This chapter illustrates the power of an altercasting theory for explaining a wide range of human behavior by using it to tackle some unresolved questions from the literature on source credibility and persuasion. Third, this chapter uses recent research on social roles, self-presentation, and social cognition to develop further the theory of altercasting. Finally, the power of altercasting as an influence device raises ethical questions about its usean issue we address in the last section of the chapter. We begin by looking at two ways of altercastingby mand and by tact. Manded Altercasting: Telling You Who You Are One day, Mr. Branch Rickey, President of the Brooklyn Dodger baseball club, was having lunch with Red Barber, the announcer for the club, when

< previous page

page_204

next page >

< previous page

page_205

next page > Page 205

he made the following pronouncement about Barber: "You think I don't know much about you. But there is more about you that I know than you have any idea. You have a potential for civic value that has not been touched" (Barber & Creamer, 1997, p. 265). Barber recounts that up until that day he had never been involved in any sort of community service. But from that day forward, Barber muses, he seemed always to be contributing to charities and community groups. This example from the annals of baseball history illustrates the manded altercast. Skinner (1957) defined the mand as a verbal operant or expression that specifies its own reinforcement. For example, the mand "Pass the salt" specifies what the listener should do. The term mand is derived from the words command, demand, and countermand. Thus, manded altercasting is a verbal expression that specifies the role to be taken by the audience or alter. For example, Mr. Rickey used the manded altercast by explicitly labeling Red Barber civic-minded, thus placing him in the role of community leader. How effective is the manded altercast as an influence device? Social psychological research from four general areas demonstrates that it can be quite successful in guiding and determining attitudes and behavior. Making Social Roles and Identities Salient One way to altercast someone into a role is to take an already existing role and make it salient and accessible. For example, Kelley (1955) made the role of being Catholic salient by having Catholics read stories about either Pope Pius XII or the missionary work of the Catholic Church. Compared to non-Catholics and Catholics who read articles about Dwight Eisenhower or UNESCO, participants whose Catholic identity was made salient were more likely to support the positions of the Catholic Church (see also, Charters & Newcomb, 1958). Similarly, Festinger (1950a) found that Catholics and Jews showed no biases in voting for leaders of their group when they did not know their religion. However, when their religious identities were made salient, members were more likely to vote for those from their own religion. Biddle et al. (1985) observed that students with highly salient identities of "alcohol user" or "smoker" were more likely to drink or to smoke. Piliavin and Callero (1991) also observed that those who gave blood and developed a salient blood donor identity were more likely to continue to donate blood. Collins, Litman, Hadsall, and Gross (1994) found that pharmacists completing a questionnaire labeled as a survey of "Health Professionals'' were more consistent with their social role as pharmacists compare to pharmacists who received the questions labeled as a survey of "Community Opinion." Abrams, Wetherell, Cochrane, Hogg, and J. Turner (1990) found that conformity to a group was a function of identification

< previous page

page_205

next page >

< previous page

page_206

next page > Page 206

with membership in the group. Finally, Haslam, Hutchinson and J. Turner (1994) found that an ingroup was more successful in persuading than an outgroup when group memberships were made salient. Placing Alter in a Pre-Existing Role Manded altercasting can take the form of simply placing the alter into a pre-existing role that then constrains and modifies the target's behavior. Lieberman (1956) provided a classic example of this type of altercasting. In a natural experiment, Lieberman measured the attitudes of rank-and-file factory workers toward management and unions. Some of these workers were then promoted to foremen, whereas others were elected union stewards. When placed in the role of foremen, workers became more favorable toward management, whereas those placed in the roles of union stewards became more favorable toward the union. Interestingly, when the foreman and stewards returned to their old role of rank-and-file worker, they showed a tendency to revert back to their old attitudes. (For other examples see Berger & Cummings, 1979; Frederickson, 1969; Herman & Hulin, 1972; Porter & Lawler, 1965; Samuelson & Allison, 1994.) Many of the classic experiments in social psychology use manded altercasting by placing participants in roles that produce behavior that appears on the surface to be quite extreme. For example, in the Stanford prison experiment (Haney, Banks, & Zimbardo 1973; Zimbardo, 1991) normal students were assigned the roles of "prisoner" or "guard" by the flip of a coin. After 6 days, the experiment was prematurely concluded because the students assigned to the role of guard became abusive whereas those assigned to be prisoners suffered humiliation. Milgram's (1974) studies on obedience to authority placed participants in the role in which they became agents of authority. Most participants obeyed by ostensibly providing extreme shock to another human. Interestingly, the level of obedience varied as a function of minor alterations in the role of the authority (i.e., changes in location and dress). By subtly inducing participants to lie about a boring experiment, Festinger and Carlsmith (1959) placed them in a role that conflicted with their normal self-concept as good and moral persons; the theory of cognitive dissonance is a theory of how people justify the consequences of taking a role that is discrepant with how they normally view themselves. A variant of this form of altercasting is co-optation (Pfeffer, 1981), or placing a powerful (and sometimes enemy) alter in a position or role that supports one's own cause. For example, a firm facing financial difficulties may place a banker on the board of directors (Pfeffer, 1972), a utility with a bad public image may place a media representative on an advisory council (Pfeffer, 1974), or a university under attack for its discriminatory hiring

< previous page

page_206

next page >

< previous page

page_207

next page > Page 207

practices may place women on a key committee (Pfeffer, 1981). Co-optation is an effective tactic because it provides the alter with new information sympathetic to the ego's cause and places the alter in a new role that constrains behavior to be congenial to the organization. In a series of experiments, Lawler and his colleagues (Lawler, 1983; Lawler & Thompson, 1979; Lawler, Youngs, & Lesh, 1978) demonstrated that a leader can co-opt a subordinate by offering her or him a symbolic position, by promoting him or her to "advisor" with some, but nonetheless, minimum power, or by giving a pay bonus. Regardless of the form, co-opted participants were much less likely to rebel against the leader. Attributing to the Alter a New Identity or Role A third way of using the manded altercast is to attribute to the alter a new role that the alter may not have previously considered. (This is how Mr. Rickey increased Red Barber's civic-mindedness). Two experiments by Miller, Brickman, and Bolen (1975) illustrated the power of this technique. In their first study, Miller et al. told a class of fifth graders that they were neat and tidy people; for example, the students were commended by their teacher, principal, and janitor for picking up candy wrappers, keeping their room orderly, and not leaving litter around the school. Relative to control groups who either were told nothing or received a lecture on being neat and tidy, the students who were told they were neat and tidy exhibited less littering over a 2-week period. In their second experiment, Miller et al. told second graders that they were talented in math; these students outscored students in the control groups on a subsequent test of math skills. Similarly, Strenta and DeJong (1981) placed participants in a prosocial role by telling them that a personality test had shown them to be more kind and thoughtful than others. These participants were more likely to help a confederate that dropped a stack of cards compared to participants who were told that they were intelligent, that it is important to be kind, or told nothing (see also Jensen & Moore, 1977; Kraut, 1973; McArthur, Kiesler, & Cook, 1969). A variant of this form of altercasting is called the hidden self technique (see Hilgard, 1977). The technique was used by Spanos, Radtke, and Bertrand, (1985) to distinguish a special process from a role interpretation of hypnotic amnesia. According to a dissociative interpretation, hypnotic amnesia is the result of the special qualities of hypnosis whereby memories become dissociated from conscious control (see Kihlstrom & Shor, 1978). In contrast, a role interpretation suggests that amnesia is part of the hypnotic role (Spanos, 1996; Spanos & Chave, 1989). Indeed, hypnosis can be seen as a double altercast; first the hypnotist takes the role of authority

< previous page

page_207

next page >

< previous page

page_208

next page > Page 208

thereby placing the participant in a role of agent and then mands a certain pattern of behavior, say, forgetting a word list. In the Spanos et al. (1985) experiments, the hypnotic role was modified to include a hidden self or observer who could recall even though the hypnotized part of the self could not. This simple role modification eliminates hypnotic amnesia and illustrates the power of casting an alter into a well-specified role. Asking the Alter to Play at a Role A final area of research supporting the effectiveness of manded altercasting is research on role-playing and attitude change. For example, Janis and King (1954) asked participants to pretend to be a sincere advocate for a position opposite to the one they normaly hold. The results showed that role-playing a position led to significantly more acceptance of that position (relative to controls who merely heard the speeches). This basic finding has been replicated repeatedly (Elms, 1967; Greenwald, 1969, 1970; Janis & Mann, 1965; King & Janis, 1954). How does role-playing lead to attitude change? In actuality, the procedure developed by Janis and King should probably be termed playing-at-a-role instead of role-playing, because the alter is asked to pretend to play a role as opposed to actually performing the role (see Coutu, 1951, for more on this distinction). In this situation, the participant who accepts the task becomes motivated to seek arguments to support the assigned positiona process Elms and Janis (1965) termed biased scanning. Playing-at-a-role leads to a psychological mindset of open-mindedness in which a premium is placed on the fair evaluation of information opposing one's own position (Greenwald, 1969). This induction of open-mindedness separates the effects of playing-at-a-role from placing participants in a role that arouses dissonance (Festinger & Carlsmith, 1959) in which self-justification is required. Tact Altercasting: A Theory of Source Credibility Two social psychologists (call them A and Z) happened to be visiting the Monterey Bay Aquarium with A's family. Upon arriving at the parking lot, Social Psychologist A begins to lecture his 4-year-old son on why he should not dart into the street but should walk calmly on the sidewalk, and so on. Before the lecture goes too far, Social Psychologist Z interrupts, calls the boy by name and says, "I am afraid. All this traffic is going too fast, and I don't know the way to the aquarium. Will you hold my hand and show me how to get there?" A's son grabs Z's hand and walks safely to the aquarium.1 1Anthony Pratkanis thanks Mark Zanna for this example.

< previous page

page_208

next page >

< previous page

page_209

next page > Page 209

This true example illustrates altercasting by means of a tact. According to Skinner (1957), a tact is a verbal operant or expression in which a response of a given form is evoked or strengthened by a particular object or event. The term tact comes from the phrase make contact with the world. Thus, in tact altercasting the alter is placed in a role through mere contact with the social world; in other words, contact with a role or a set of roles places alter in a complementary role. By taking the role of someone needing help, Psychologist Z placed the child in the role of helper and adult. Note that this role requires the child to act like an adult and not run into the street, which is exactly what the child did. In contrast, Psychologist A used an ineffective manded altercasting, in effect placing his son in two conflicting rolespleading with him to act like an adult and reminding him that he is a child. Tact altercasting can be seen in an experiment by Snyder, Tanke, and Berscheid (1977; for additional examples see Dweck, Davidson, Nelson, & Bradley, 1978; Word, Zanna, & Cooper, 1974). In this study, pairs of opposite-gender students interacted over the telephone. Before the conversation, male students, unbeknownst to the females, received a photo that made their partner appear to be physically attractive or unattractive. In other words, the male participants were altercast through contact with a photo of an attractive partner into the role of admirer. When men believed they were interacting with an attractive female, they were more sociable, sexually warm, interesting, and socially adept than males who thought they were interacting with an unattractive partner. This behavior then altercasted the females into attractive and unattractive roles with those female students who were perceived to be attractive exhibiting behavior that was more poised, sociable, warm, and outgoing. Altercasting by tact is a theory of source credibility, that is, a theory of the conditions under which it will appear plausible to believe the source of a message. Roles typically come in role-sets (Merton, 1957b) consisting of pairs of complementary roles (e.g., teacher-student, lawyer-client, parent-child, outgroup-ingroup) or in systems of interlocking roles that Durkheim (1933) termed divisions of labor (e.g., positions on an assembly line, in the justice system, in a bureaucracy). As the ego adopts a certain role (e.g., teacher), the alter must accept the complementary role (of student) or attempt to come up with another role acceptable to the ego in order to allow the continuation of the interaction. Source credibility is created as a function of the interaction; ego and alter are credible (e.g., they successfully influence each other) to the extent that each takes the offered role and their message makes use of the expectations associated with their adopted roles. An altercasting theory can be contrasted with the two major and current theories of source credibilityHovland, Janis, and Kelley's (1953) reward or reinforcement model and Kelman's (1958,1961) three-factor model. Ac-

< previous page

page_209

next page >

< previous page

page_210

next page > Page 210

cording to Hovland et al., credibility is based on two attributes of the communicator: perceived trustworthiness and expertise. In general, people find it rewarding to hold correct and valid opinions. One way to ensure such rewards is to accept the opinions of those who are in the know (experts) and those who are not likely to be biased (trustworthiness). According to Kelman, effective communicators can be of three types: powerful, attractive, and expert. Each type of source invokes a different psychological process: The powerful source induces compliance by controlling rewards and punishments; the attractive source persuades because the message recipient seeks to identify with the communicator; the expert source is knowledgeable and trustworthy (in the manner described by Hovland et al.) and changes opinion through internalization. There are at least two major differences between the Hovland et al. and the Kelman models and an altercasting theory of credibility. First, Hovland et al. and Kelman locate credibility in the communicator; it is an attribute of the source. The source is described by adjectives such as trustworthy, expert, knowledgeable, ugly, low credibility. These attributes are then used to specify when a source will be more or less persuasive. An altercasting theory locates credibility in the social relationship between the source and the audience; perceptions of credibility thus can vary across persons and situations. For example, a cult leader is seen as a great spiritual master by those within the social relationship of ''member" of the cult but is seen as a flim-flam artist by those outside the group. Second, altercasting theory specifies a different psychological reason for why a communicator is effective; a communicator is not effective because of the control of rewards (as Hovland et al. claimed), nor because of one of Kelman's three processes, but because of the nature of the role relationship between source and message recipient. Such relationships can be dynamic; thus source credibility can change over time as a function of role negotiation and development. When dealing with expert, attractive, and powerful sources, altercasting theory generally makes the same predictions as those derived from theories by Hovland et al. and by Kelman but for different reasons. The expert is not persuasive for any reward value but because he or she places the recipient in a role (unknowing public) that requires acceptance of the message. Similarly, an attractive source casts the alter in the role of admirer, whereas the powerful source elicits the complementary role of a coward who is afraid of punishment and afraid to appear to be afraid. However, altercasting theory makes two classes of predictions not made by Hovland et al. and by Kelman. First, altercasting theory predicts that some nontraditional sources can be persuasive under certain conditions. By nontraditional, I mean communicators such as a child or a member of an outgroup that do not possess the attributes specified by Hovland et al. and

< previous page

page_210

next page >

< previous page

page_211

next page > Page 211

by Kelman. Second, altercasting theory also specifies when a communicator should be most effective (i.e., when the acceptance of the message is consistent with the demands of the role-set). For example, an expert is most effective when arguing within a domain of expertise and with the trappings of an expert. To gain an understanding of how an altercasting theory of credibility diverges from past approaches, we examine three sets of communicators to see when and how they persuade. Through the Eyes of a Child One of the classic films of the U.S. civil rights era is To Kill a Mockingbird (Mulligan, 1962). The film stars Gregory Peck as a southern lawyer who defends a Black man charged with the rape of a White woman. The trial exposes some of the racist norms that often guide behavior. Interestingly, the story is told through the eyes of a child, Scout, the motherless daughter of Peck's character. Does this literary device increase the impact of the film? According to both Hovland et al. and Kelman, it should not, because a child lacks the attributes of a credible source. A child is not powerful and controls few rewards. Although a child can be cute and attractive, most adults do not seek to identify with and be like the child. Children lack world experience and are thus not very knowledgeable; further their expression of opinions can be variable, making their communications unreliable and untrustworthy. In contrast, an altercasting theory specifies conditions under which a child is an effective communicator. A child places the alter into the role of generativitya parent or an adult who is responsible for nurturing and protecting the child (see discussion of helper-helpee). Any communication that uses these desires should increase in effectiveness if it is attributed to a child. Gliner and Pratkanis (1997) tested these propositions in a pair of experiments. In their first study, participants were asked to evaluate one of four public service announcements. Half of the messages were attributed to "Sarah Whitmore, second grader" and accompanied by a photo of Sarah as a young girl with long ponytails. The other half were attributed to "Dr. Whitmore, PhD" and accompanied by a picture of a distinguished looking man wearing a suit and tie. The message argued for one of two conclusions: "Nobody wins a nuclear arms race. Disarm for the future," (a message emphasizing the need for protecting the future) or "There must be more than the nine currently known planets in our solar system'' (a message requiring the acceptance of a technical conclusion). The results provided considerable support for an altercasting interpretation. The child was most effective in arguing for nuclear disarmament and ineffective in arguing for the existence of a 10th planet. In contrast, the ex-

< previous page

page_211

next page >

< previous page

page_212

next page > Page 212

pert with the PhD was most effective in arguing for the 10th planet and was ineffective as a promoter of nuclear disarmament. According to an altercasting theory, a child invokes a social role of protector; accepting a message arguing for nuclear disarmament is one way to protect the child's future. In contrast, the role of protector does little to encourage the acceptance of a message based on technical, specialized knowledge of astronomy that the general public, much less a child, is not likely to possess. Such a message would be more acceptable from an expert source who is assumed to possess such knowledge and who casts the alter in the role of a member of the unknowing public. In contrast, even the unknowing public has a ready store of information and strong feelings concerning nuclear arms, and the expert's perceived technical competencies are of little value in arriving at a decision on this issue. It should be noted that neither the Hovland et al. nor the Kelman theories can predict the complex interaction obtained in this experiment. Gliner and Pratkanis replicated their results in a second experiment that again used the Sarah Whitmore and Dr. Whitmore communicators. The message in this study argued for the purchase of fictitious Darelli tires. The message could emphasize one of two themes: safety ("Don't be the cause of yet another road casualty") or the special performance ability of Darnelli tires. As in the first experiment, and consistent with an altercasting interpretation, the child was most effective when the message emphasized safety compared to performance. The opposite pattern was obtained for the expert. African Americans as Effective Communicators African Americans are perhaps the United States' most visible and negatively stereotyped outgroup. For example, still today many Whites believe that African Americans possess stereotypic attributes, such as less sensitivity to pain, thicker heads, and longer arms than Whites (Plous & Williams, 1995). Given these cultural stereotypes, from the standpoint of the theories of Hovland et al. and of Kelman, African Americans should not be effective sources of communications directed at Whites; these cultural stereotypes ensure that African Americans are not perceived as powerful, limit Whites' ability to identify with African Americans, and prevent Whites from perceiving the expertise and trustworthiness of African Americans. Yet, under certain conditions, African Americans have been used successfully in commercial persuasion. Historically, one of White America's most curious obsessions came in the form of "Black collectibles"common household items such as housewares, toys, games, kitchen decorations, and postcards that depicted African Americans in a derogatory and stereotyped manner (Goings, 1994).

< previous page

page_212

next page >

< previous page

page_213

next page > Page 213

Produced primarily from the 1880s (the end of Reconstruction) until the late 1950s (the beginning of the dismantling of Jim Crow laws), tens of thousands of Black collectibles were produced and displayed in White American homes including such items as potholder hangers featuring big-eyed Black children eating watermelon, Black servant salt and pepper shakers, cast-iron "jolly nigger" banks, and "Coon Chicken" ashtrays. Also popular during the period was the use of Black spokespersons such as Aunt Jemina, the Gold Dust twins, Rastus, and Giddap Uncle in advertising (KernFoxworth, 1994). Ironically, at a time when White Americans passed laws promoting segregation, the image of African Americans were used to sell everything from pancakes to washing powder to tourism. According to an altercasting theory, the use of African American communicators can be effective when the ad makes use of the underlying social roles. For example, consider the myth used to establish Aunt Jemina pancakes. According to advertising legend, during the Civil War a southern officer was separated from his troops and managed to evade the Union forces. However, he grew hungry and weak until he came upon Aunt Jemina's cabin. He had a wonderful meal of pancakes and quickly regained his health. After the war, he could not get those pancakes out of his mind and so he went back to that cabin with some friends from the north. Aunt Jemina fixed them an incredible batch of pancakes, and the southern officer and his northern friends decided to market them. Notice how Aunt Jemina happily takes a servant role to altercast the White American into the role of head of the plantation. She is happy to make pancakes for the one big happy family of White Americans from the north and south with the past sins of war and slavery forgotten. Of course, Aunt Jemina as "mammy" can be viewed as an expert and trustworthy source in the terms of Hovland et al. and of Kelman; a mammy is an expert in domestic matters and, according to stereotype, can be trusted. However, such explanations do not apply to the Gold Dust Twins, who appeared as fools to sell washing powder, or the Giddap Uncle portrayed as a pathetic servant to sell cereal. Both images altercasted Whites into the role of superior. Since the birth of the modern civil rights era, new patterns of African American-White social roles have emerged (Gaertner & Dovidio, 1986). One key development is that some Whites now feel that bigotry is wrong and thus attempt to appear nonprejudiced. In this attempt, Whites may accept a message, or at least, pay more attention to a message from an African American communicator. For example, Aronson and Golden (1962) found that an African American engineer was more effective than a White engineer in convincing low-prejudice White students about the value of arithmetic. The opposite relationship held, however, if the students were prejudiced against African Americans. More recently, White and Harkins

< previous page

page_213

next page >

< previous page

page_214

next page > Page 214

(1994) found that Whites were more highly motivated to process a message from an African American as opposed to a White communicator. This resulted in more persuasion if the message was strong and cogent. White and Harkins also provided evidence that a desire to avoid appearing prejudiced against African Americans was driving the motivation of Whites to process the message. Brains Versus Beauty In an attempt to distinguish the Hovland et al. model from the Kelman model of source credibility, persuasion researchers have attempted to answer the question of who is the more convincing communicatorthe expert or the physically attractive source? Unfortunately, the answer this research provides is far from clear. For example, Maddux and Rogers (1980) found that the expert is more effective than the physically attractive source, whereas Mills and Harvey (1972) and Norman (1976) found, under certain conditions, just the opposite; Horai, Naccari, and Fatoullah (1974) found little difference. Altercasting theory provides a means of reconciling these results. A communicator should be more or less persuasive depending on how he or she uses the demands inherent in the roles adopted by the source and audience. In other words, the expert and the physically attractive can be more or less effective depending on the ways social roles are used in the influence attempt. For example, an expert is someone with specialized knowledge in a given domain. An expert casts alter into the role of "not knowing;" in other words, the alter feels as if he or she does not know enough to evaluate the issues at hand and thus must trust the expert. Experts are most effective at persuasion when they (a) argue within their domain of expertise, (b) address an audience that feels incompetent or lacks confidence concerning the issue, (c) put on the trappings of the role, and (d) develop alter's trust (McGinnies & Ward, 1980). For example, Maddux and Rogers (1980) found that source expertise was more persuasive than beauty when arguing that people should sleep only 4 hours a nightan issue that probably requires knowledge that the average person is not likely to possess. Similarly, Gliner and Pratkanis (1997) found that an expert was more effective when arguing matters of technical fact (the existence of a 10th planet in the solar system) than of social opinion (the fate of nuclear arms). Pallak, Murroni, and Koch (1983) found that an expert was more effective if the message had a rational as opposed to an emotional tone. On the other hand, physically attractive persons are often viewed to be socially competent (Egly, Ashmore, Makhijani, & Longo, 1991) and thus are socially prestigious; we admire them and listen to them on matters of social opinion because they occupy a prominent position in the social network.

< previous page

page_214

next page >

< previous page

page_215

next page > Page 215

The physically attractive source, as Chaiken (1979) noted in her review, appears to be associated with greater influence on a wide array of issues as opposed to the limited domain of the expert (cf. Chaiken, 1979; Horai et al., 1974; Snyder & Rothbart, 1971). For example, the attractive communicator (in contrast to the expert) is effective regardless of the strength of the message (Norman, 1976) and the timing of source identification (Mills & Harvey, 1972). However, a role interpretation also suggests the occasions that physical attraction is less persuasive. For example, the attractive source is at a disadvantage when the issue involves technical matters as opposed to social opinion (Maddux & Rogers, 1980; Pallak et al., 1983). Attractive sources are less persuasive when they fail to act consistently with their role of social competence. For example, Mills and Aronson (1965) found that an attractive source was more persuasive than an unattractive one when she volunteered to answer questions, thus demonstrating social competence. In contrast, when the attractive source was reluctant to take a visible position in the group, she was less persuasive than the unattractive source. Finally, attractive sources are less persuasive when their audiences actually think for themselves (Chaiken, 1986)after all, physical attractiveness does not represent any real achievement and thus, represents a poor foundation on which to rest a decision. Some General Remarks As we have seen, an altercasting theory of source credibility is particularly useful in identifying effective but nontraditional communicators and for delineating the conditions under which a given communicator will be more or less effective. An altercasting theory offers at least two additional advantages. First, it is useful for stimulating new research. For example, looking at credibility as an altercast generates new hypotheses about the characteristics of effective communicators and when they are effective. In addition, altercasting theory stresses that audience roles are important factors in persuasion, thus raising the question of how an influence attempt is received when the audience (as opposed to the source) takes roles involving expertise, trustworthiness, attractiveness, power, and so on. Second, the altercasting approach complements other theories of social influence. For example, according to Petty and Cacioppo's (1986a) Elaboration Likelihood Model Postulate 3, a source of a message can serve as a cue to accept the message, as a message argument, or to motivate thought about the message; however, specifying the function of any given source is exogenous to the theory. A theory of altercasting can provide a partial answer. For example, some social roles invite immediate acceptance of a command (authorities, the socially prestigious), others can serve as an argument (e.g.,

< previous page

page_215

next page >

< previous page

page_216

next page > Page 216

"experts agree"), yet others invite scrutiny either as a nature of their roles (lawyers in court) or because the communicators behave inconsistently with role expectations. In a similar vein, one of the longstanding issues in dissonance theory is identifying the factors that determine the route of dissonance reduction (Abelson, 1959). Social roles structure social behavior and thus influence the manner of dissonance reduction. For example, agents of authority can claim they did a misdeed as part of following orders, experts may lose their expert status by admitting to a mistake and thus may try to deny the wrongdoing or blame others (scapegoats), and charismatic figures may not need to justify misdeeds at all (after all, they are godlike). Some Common Role Sets Used in Influence A complete theory of altercasting and persuasion requires the specification of common social roles and their relation to influence. Previous attempts to specify the nature and organization of social roles have focused on (a) developing procedures for measuring respondents' understanding of specific roles (e.g., Burke & Tully, 1977; Motz, 1952) and (b) creating taxonomies of dimensions of roles (e.g., Gold, 1997; Marwell & Hage, 1970). These attempts have met with limited success, perhaps because, as symbolic interactionists pointed out, people can actively remake the nature of roles to suit the requirements of the situation (Stryker & Statham, 1985). Nevertheless, some role sets frequently are used in influence attempts, and it is worthwhile to describe how these roles are used in persuasion. Authority-Agent An authority is a person who is perceived to have the legitimate right to command others to be her or his agent. For example, the doctor easily can have others shed their clothes, a therapist can have clients accept all sorts of treatment regimens, and the police officer's command to "pull over" often is obeyed (as in the bank officer swindle). Authority can be obtained legitimately by occupying a position in a division of labor (see Weber 1956/1978) or merely by putting on the trappings of authority. For example, Bickman (1974) found that passersby would readily obey a command to pick up a paper bag or give a dime to a stranger or move away from a bus stop from someone in a guard's uniform compared to either a civilian or someone dressed as a milkman (see also Bushman, 1984, 1988). As Milgram (1974) has shown, authority can be a powerful role for inducing obedience. Alternatively, if one wants to justify immoral or negative behavior, the ego can cast the alter into the role of authority and claim the evil deed was required by the authority.

< previous page

page_216

next page >

< previous page

page_217

next page > Page 217

Charismatic-Devotee The charismatic is one who is seen as having special gifts (Weber, 1956/1978); the person is believed to be endowed with supernatural, superhuman, or, at least, exceptional qualities that set the charismatic individual apart from the common lot of humanity. The role can be obtained by appearing to perform great feats (as Homer's public relations did for Odysseus) or by taking on the trappings of God or God's messenger (as do many cult leaders). The charismatic leader casts the alter into the role of devotee who is required to have unquestioning faith in the leader and to worship those with charismatic gifts. One's salvation must needs this devotion. What better way to demonstrate faith than through blind obedience? Defector-Attentive Confidant A defector is someone who takes a position opposite to the apparent self-interest of her or his social role. For example, previous research demonstrates the effectiveness of these communicators: a criminal supporting tougher courts (Walster, Aronson, & Abrahams, 1966), a U.S. general claiming the Vietnam War was going badly, an African American stating that African Americans lack initiative (Koeske & Crano, 1968), a scientist who quits working for a pro-UFO (or anti-UFO) organization arguing that UFOs do not (or do) exist (Dutton, 1973), a manufacturer selling a product by pointing out some of the brand's weaker features (Settle & Golden, 1974), an ex-alcoholic arguing against alcohol use (Levine & Valle, 1975), and a politician arguing for a position opposite to ones he or she had taken previously (Eagly, Wood, & Chaiken, 1978). A defector casts alter into the role of an attentive confidant. When someone behaves inconsistently with role expectations, it attracts attention and encourages others to make dispositional attributions to explain the behavior (ones, Davis, & Gergen, 1961). Because the defector acts in seeming opposition to the self-interest of the role, the defector's behavior creates in alter a sense of trust and a feeling that the defector speaks out of a need to set theed to have special, inside knowledge about the issue, the alter feels privileg record straight. Because the defector has occupied the role and thus is presumed to get this information. Expert-Unknowing Public An expert is an individual with specialized knowledge in a given domain (e.g., the lawyer, doctor, college professor, or nuclear physicist). Someone can become an expert through the possession of objective skills, by being socially recognized by the appropriate institution (e.g., licensing and credentialing), or by faking expertise by putting on the trappings of an expert (Gerver & Bensman, 1954). As previously noted, an expert casts the

< previous page

page_217

next page >

< previous page

page_218

next page > Page 218

alter into the role of "not knowing" and is thus more effective when arguing within a domain or specialization and especially in a domain in which the alter lacks information and comes to trust the expert. Interestingly, casting the alter as an expert may also result in influence as the alter exudes overconfidence and thus may not look at the issue in a critical manner (as in three-card monte, the payoff, and rag). Fool-Normal The role of fool is conferred on those who are perceived to depart from group norms through a deficiency in person (e.g., physical deformity, or artificial distortions) or in conduct (e.g., lack of intelligence, wit, or courage) or through excesses of conduct (e.g., pranks, recklessness, or undignified behavior; see Klapp, 1949). The status of fool is primarily negativethe fool is despised, degraded, and ridiculed. As such, it is a role to be avoided. Casting an alter as a fool is a means of sanctioning behavior and social control. In contrast, the presence of a fool makes the "normal" ego feel superior. The fool often appears in commercial advertising to show what disastrous effects can occur if the advertised brand of product is not purchased immediately. Typically, the fool loses the date, ruins the laundry, or destroys home and automobile because he or she is too foolish to use a particular product. The recipient of the ad feels superior to the hapless fool and can maintain these feelings simply by buying the product. Two variants of this role is the redeemed fool (a fool who escapes the role by doing something heroic) and the wise and rustic wit disguised as a fool (Klapp, 1949). One of the most successful direct advertisements of all time used a redeemed fool when it announced: "They laughed when I sat down at the piano, but when I started to play!" (Watkins, 1959). The ad features a man ready to make a fool of himself at the piano, save that he purchased a correspondence course that taught him to play. The disguised fool is often used by the purveyors of pseudosciences such as inventors of perpetual motion machines or marketers of "alternative" psychotherapies and health regimens when they claim: ''They made fun of Galileo, too." Such altercasting forestalls critical assessments (who wants to persecute the next Galileo?) and elevates the fool to a higher level of greatness. Granfalloon According to the novelist Kurt Vonnegut, the granfalloon is a "proud and meaningless association of human beings" such as a Hoosier, a Buckeye, a member of a pyramid scheme, or a devotee of Klee or Kandinsky. Sometimes the granfalloon can take on special meaning such as Nazis, social psychologists, or Fraternal Order of Police. Once an individual is altercasted

< previous page

page_218

next page >

< previous page

page_219

next page > Page 219

into a social identity or granfalloon (which appears to be a remarkably easy task; Richardson & Cialdini, 1981; Tajfel, 1981) influence follows in at least two ways (Abrams & Hogg, 1990; Mackie, Worth, & Asuncion, 1990; van Knippenberg, Lossie, & Wilke, 1994). First, the social identity provides a simple rule to tell the individual what to believe; in other words, "I am a _____ [fill in the blank with an identity] and we do and believe _____ [fill in with appropriate identity-related behavior and belief]." Second, some identities become very important as a source of selfesteem. Propagandists can play on emotions by focusing threats to this source of self-esteem and then providing a means, compatible with the propagandist's goals, for dealing with these threats; for example, the ego can remind the alter that, "If you are really one of us, you'll do this" or ''Our enemies are attacking us; to defeat them you must do this." Helper-Dependent The ego can cast alter into the role of helper through supplication or announcing one's weaknesses and dependencies on another (Jones & Pittman, 1982). For example, one may feign ignorance, youthful inexperience, or just plain incompetence to cast others into the role of helper. Once cast into that role, helpers are expected to place the ego's problems at the fore of their concerns, ease discomfort, maintain positive communications, and assist in solving the ego's problems (Thomas, Polansky, & Kounin, 1955). Although the supplicant may receive help, it is not without psychological costs. The dependent is expected to take the aid unquestioningly and with gratitude (see discussions of altercasting in therapy by Davison, 1973; Senour, 1982). In situations in which help violates social norms (e.g., among equals, against norms of self-sufficiency), the recipient of aid is seen as inferior and the subject of ridicule (Nadler & Fisher, 1986; M. Turner & Pratkanis, 1994). Intimates (Friends and Lovers) As the sweetheart scam illustrates, it is hard to refuse a request from someone we like, love, and feel intimate toward. We do not want to hurt those we like or love by refusing their plea. Close relations create obligations to place the needs of others before one's own (Roloff, Janiszewski, McGarth, Burns, & Manrai, 1988). Failure to meet these obligations generates guilt which in turn produces more motivation to comply with the request of an intimate (Vangelisti, Daly, & Rudnick, 1991). According to Sternberg's (1986) triangular theory of love, the requirements of an intimate role can vary as a function of degree of intimacy, passion, and commitment, each of which creates a type of close relationship with their own rules of influence.

< previous page

page_219

next page >

< previous page

page_220

next page > Page 220

Intimidator-Coward The goal of the intimidator is to strike fear into the alter by reducing the alter's role to that of a quivering coward (Jones & Pittman, 1982). The intimidator gains power by attempting to control critical resources (such as money, reputation, physical well-being, status, etc.) needed by the alter. This can be done through threats to take resources (punishment) or to deliver them (rewards). For example, a dean may reprimand a faculty member without cause ''to show who is boss," or a department chair may deny a faculty's achievements to justify taking away lab space needed for research and promotions. Of course, not all intimidators are high-powered. For example, the student who fails an exam may spread rumors and lies about the teacher in hope that destroying the professor's reputation will "teach her or him a lesson." The intimidator persuades simply because the alter seeks to get some reward or escape the frightening consequences threatened by the intimidator (as the mark does in the phony cop blow-off). In general, by controlling punishers and rewards, the intimidator produces ready compliance that dissipates once the intimidator leaves the scene (Kelman, 1958). However, accepting the role of coward can be painful and dissonance-arousing; to help restore a positive selfconcept, the alter may attempt to identify with the oppressor. A possible response to the intimidator is to take the role of victim, thereby altercasting the intimidator into the socially undesirable role of "bully." Just Plain Folks One of the propaganda devices identified by the Institute of Propaganda Analysis (1937) was termed plain folksattempts by politicians, leaders, ministers, business officials, and others to take on the role of being common folk. This can be done by emphasizing similarities such as claiming to like the same foods or practice the same religion or engage in the same leisure activities as do everyday people. The ego thus appears as neighbor and casts the alter into the role of neighbor. Considerable research demonstrates that source-recipient similarity increases persuasion. For example, Brock (1965) found that a paint salesman was more influential when he had performed a job similar to the one facing the alter compared to when he was highly experienced. Berscheid (1966) found that similarity was effective in promoting influence only if the shared similarity was directly relevant to the issue (a limitation that does not apply to the typical propagandist who often has free range in what similarity to emphasize). Berscheid's work indicates that the plain folks altercast works via Festinger's (1954) social comparison process; independent of other attributes of the source, people have a tendency to use similar others as referents for their opinions.

< previous page

page_220

next page >

< previous page

page_221

next page > Page 221

Opinion Deviant-Majority The opinion deviant advocates a counternormative position and thereby casts the alter into the role of protecting the majority and the status quo. The response of alter is to direct frequent communications to the opinion deviant (which gives the opinion deviant the opportunity to set the discussion agenda) followed by shunning, avoiding, and ignoring the now-disliked deviant (Schacter, 1951). On occasion, an ego may cast the alter into a role of opinion deviant (by assuming to be the majority position), so others will not like the alter and not take what the alter has to say seriously. If the opinion deviant wants to have more influence than just creating discussion, the deviant role needs to be modified. Two common adaptations include (a) finding ways to control critical resources as an intimidator (Alinsky, 1971) and (b) adopting Moscovici's (1976) behavioral style (consistency of response, flexibility in negotiation, self-confidence, linking position to objective standard), thus appearing more like a prophet (see following discussion) than like a deviant. Prophet-Unwashed The prophet (or saint) takes on the mantle of integrity and of moral greatness and thus altercasts the rest of us into the role of "unwashed" and "unredeemed" (see Jones & Pittman's, 1982, exemplification). Some examples of effective change agents who have taken this role include Ghandi, Branch Rickey, Martin Luther King, and Mother Teresa. The prophet obtains the role by seemingly engaging in self-sacrifice and by placing moral principles above all else. The prophet uses moral power to persuade by shaming the unwashed into supporting a cause and by offering a chance at redemption through acts of charity and sacrificing for the moral good. Scapegoat-Sinners In this role set, an often innocent person is made to bear the sins and frustrations of the rest of the community (see Berkowitz & Green, 1962; Richardson & Cialdini's 1981 concept of BLASTing). The scapegoat has negative persuasion value. Because this individual or group must be absolutely rejected from the community (as a symbolic cause of sin), so too any message from the scapegoat must be rejected. Indeed, the goal is to reject the scapegoat's message as quickly as possible to further bolster the role of the scapegoat as the sinner. Sinners can cause the rejection of a message merely by attributing it to the scapegoat. In contrast, sinners can increase the acceptance of their message by linking it to the eradication of the scapegoat. In

< previous page

page_221

next page >

< previous page

page_222

next page > Page 222

others words sinners say, "We must do X to prevent [the scapegoat] from being in our community." Socially Prestigious-Admirer The celebrity, the popular kid in the class, the movie star, the famous person, the star athlete, the television personality, the fictitious cartoon character, and the physically attractive are all socially attractive and prestigious. The socially prestigious are socially competent; thus we listen to them on matters of social opinion. In the words of (1961), the socially prestigious are well known for being well known. Because they occupy a prominent position in the social hierarchy, we give those with prestige preferential treatment; their words are treated with honor as if they speak the collective wisdom. As such, prestige allows the communicator to give opinions on a wide array of topics. Prestigious communicators are particularly effective when they address their fans who greatly admire and seek to be like their idolsa process described by Kelman's (1958, 1961) identification. As noted, there are some limits to the persuasiveness of the prestigious source including (a) arguing in technical as opposed to domains of social opinion, (b) when they violate the expectation of social competence inherent in the role, (c) when their audiences are motivated and able to think for themselves, and (d) when they betray the basis for the identification relationship with their fans. Why Is Altercasting Such a Powerful Influence Device? Our review of the research literature indicates that altercasting can be a powerful means of influence. Why? What makes altercasting so effective? At least four complementary answers can be given to this question. 1. The social role is a basic unit in people's everyday social cognition. A number of researchers and theorists have argued that people naturally think of other people in terms of roles (e.g., Brewer, 1988; Cantor & Mischel, 1979). Given that social roles are one of the basic units of social cognition, thinking in terms of roles is fairly simple and easy for most alters, thereby giving roles a phenomenology of "given-ness" and a quality of "taken-for-granted." This, of course, adds to the power of altercasting as an influence device. 2. Presenting oneself in a social role that can then be used to cast the alter is relatively easy. One of the key findings from research on impression management is the myriad of ways that people can present themselves to others. For example, Leary (1996) listed the following tactics for

< previous page

page_222

next page >

< previous page

page_223

next page > Page 223

self-presentation: favorable self-descriptions; attitude expressions of agreement; self-serving attributions; biased recall; emotional displays; gestures; modification of physical appearance with clothes, cosmetics, sunbathing, diets, and exercise; development of social associations; conformity to the group; behavioral expressions of helpfulness, aggression, risk taking, etc.; and arranging one's physical environment. With so many available tactics, putting on the trappings of any given social role, and thus altercasting others into complementary roles, is relatively easy. 3. Constructing roles that trap others in a course of action also is relatively easy. For any given issue, a role can be cast that makes the alter more predisposed to accept the ego's position. Consider these issues and possible roles that, once accepted, trap the alter into a given position: pro-national health insurance (patient or possible future patient), antinational health insurance (taxpayer), increases in education spending (parent), cuts in education spending (taxpayer), increases in welfare spending (helping neighbor), cuts in welfare spending (taxpayer), overpriced wine (a connoisseur more concerned with quality than price), spiritual firewalking (curious adventurer), and Valentine candy (lover who cares). Two aspects of this list are worthy of comment. First, some roles can be used to persuade on more than one issue, which can be an advantage in promoting a program of issues. For example, the conservative can argue against various forms of increases in government spending with one role (taxpayer), whereas the liberal must use a different role with each issue. Second, the dynamics of role taking not only traps the alter into a course of action but traps the ego as well. For example, a conservative who casts alter into the role of taxpayer to argue for a cut in social spending later may have difficulty arguing for increased spending in other areas, such as defense. 4. People often accept with ease the social roles offered to them. One of the major lessons learned by Allen Funt (1994), the producer of television's "Candid Camera," and by social psychologists in experiments such as Milgram's obedience to authority, the counter-attitudinal role play of dissonance research, the Stanford mock prison, and Tajfel's minimal group is that placing people in social roles is relatively easy. The situations created by "Candid Camera" and by social psychologists capture our imagination because what they find is seemingly so amazing: People readily get into roles and then engage in behavior that they probably should not and would not do otherwise. We should learn from these studies that role taking is quite common and relatively easy. This point is underscored in a recent study by Antaki, Condor, and Levine (1996) on the presentation of social identities in talk. They found that in everyday discussions, speakers' self-described social identities change rapidly as a function of the demands of the situation even to

< previous page

page_223

next page >

< previous page

page_224

next page > Page 224

the point where a speaker may avow contradictory identities within a single discussion. Instead of being amazed at how easy it is for people to "get into a role," perhaps we should be looking for the conditions under which they will not accept a proffered role or identity (see Burger & Guandagno, 1997, for an example of this type of research). Is Altercasting a Moral and Ethical Influence Tactic? This chapter began with five examples of altercasting taken from the world of the con artist. Certainly, these examples demonstrate that altercasting can be a nasty and despicable influence tactic. But is this always the case? What about Mr. Rickey's use of the tactic to get Red Barber to contribute to society? Or Psychologist Z's use of the tactic to ensure the safety of a child? Or the positive results obtained by Miller et al. (1975)? Or the use of altercasting in experiments such as those by Milgram and by Festinger? Can altercasting be used in a manner that is considered fair and ethical? Before answering this question, I should note that there is at least one interpersonal defense against being altercast into a role that overprescribes one's actions:role-making. In other words, the alter can take steps to modify and recreate the proffered role so that her or his behavior is less constrained. For example, in the sweetheart scam, the mark could respond to requests for loans and money with a counter-request; after all, true love would not put the loved one in a delicate position that would risk her or his financial future. Of course, the fact that people can remake roles limits the predictive ability of role theory in general, as symbolic interactionists made clear (Stryker & Statham, 1985), as well as the present theory in specific. However, it does raise an important and interesting question of theoretical and practical value: Under what conditions will people engage in role-making as oppose to merely accepting and taking a role? Returning to the task at hand, one way of evaluating the ethics of a communication is the means used to deliver the message. It is generally accepted that one should avoid knowingly dispensing false information. This would clearly identify con artists' use of altercasting as immoral. However, social roles are often social constructions, making it difficult to known wherein lies the truthwhat is one person's god is another's flim-flam artist. A variant of this ethical rule is Cialdini's (1996) distinction between an immoral smuggler (who illicitly and artificially imports an influence tactic into a situation in which it does not naturally reside) and the more ethical sleuth (who uses influence tactics that exist naturally in the situation).

< previous page

page_224

next page >

< previous page

page_225

next page > Page 225

Again, this would indict the con artist because false roles are imported into the situation. However, the ethics of other uses of altercasting are less clear. For example, Hitler made use of natural roles inherent in the situation (anti-Semitism as granfalloon; Jew as scapegoat) with results that were immoral. A similar argument can be made concerning the use of Black collectibles to sell consumer goods. In contrast, smuggling social roles into a situation may result in some very positive results. For example, Psychologist Z's feigning of helplessness created a set of very positive consequences including child safety, a more pleasant social exchange, and the opportunity for a child to develop adult behaviors. A similar argument can be made for altercasting in the Miller et al. (1975) experiments. Another method for evaluating the ethics of an influence tactic, which Cialdini (1996) and Pratkanis and Aronson (in press) advocate, is based on the fact that the means often determine the ends, or, as the Chinese philosopher Mencius put it, immediate goals may be accomplished with devices that produce long-term evil. Here, once again, the con artist's insincere use of roles is deemed unethical because its long term consequences strike at the very trust needed to ensure social interaction in any society. As Fromm (1972/1997) noted, when everything and everybody is for sale including persons, convictions, a feeling, and a smile, then fewer people can be trusted because fewer people have a character and an identity that can be depended on. In addition, Mencius's observation reminds us that social roles can trap us into behavior that may be unwanted and undesirable long after the social patterns were established. Finally, Pratkanis and Aronson (in press) offered another means for evaluating the fairness of a communication, especially ones used in a democracy: Persuasion tactics such as altercasting should lead us to think more deeply about an issue and to explore and develop personal options for dealing with any given situation. In this case, the con artist's use of altercasting that plays on emotions to truncate thought and to trap the mark into a course of action would be deemed unethical. In contrast, the use of altercasting by Mr. Rickey opened up new ways for Red Barber to think about himself, his role in society, and the importance of contributing to his community. It didn't trap him into a course of action but instead offered new challenges that expanded his self-concept. (A similar case can be made for the use ofaltercasting by Psychologist Z, Miller et al., and in high-impact experiments such as those by Milgram and by Festinger, if these experiments are accompanied by a debriefing that allows the participants to gain self-insight.) This emphasis on the use of altercasting to expand one's self-concept and to gain insight into an issue points to one of the more interesting uses of altercasting: role-playing multiple identities and roles. Bowman (1949) ar-

< previous page

page_225

next page >

< previous page

page_226

next page > Page 226

gued that people gain insight into their lives and an understanding of the situations facing others by playing new roles and seeing things from multiple points of view. Fisher, Ury, and Patton (1991) marshaled evidence that superior negotiationsones that provide more benefits to each of the participantscan occur if each of the negotiators attempts to take the role of others. Finally, Maier (1952) found that groups made better decisions when members assumed the role of other members and relevant audience members. In such cases, altercasting is not only a powerful influence tactic but also is one that is capable of moving beyond the phony con to help us develop an authentic and more extensive understanding of the issues we face.

< previous page

page_226

next page >

< previous page

page_227

next page > Page 227

12 Cognitive Dissonance and the Social Group Joel Cooper Jeff Stone Princeton University Cognitive dissonance theory (Festinger, 1957) has been a major player in the field of social psychology for more than 40 years. In that time, more than 1,000 studies have appeared in the literature, but rarely has the person experiencing dissonance been conceived of as a member of a group. Rather, as Zanna and Sande (1987) pointed out, virtually all dissonance research has focused on the individual acting alone. In this chapter, we examine some of the ways that an individual's membership in social groups can affect dissonance and examine some of the evidence that suggests that group membership can play an important role in how dissonance is reduced. Festinger conceived of dissonance as arising from the relation among cognitions. In Festinger's view, cognitions are "knowledges" about the environment, about other people, and about one's own attitudes and behaviors. In more modern terminology, Festinger recognized that people create cognitive representations of themselves and their social world, and that the relations among these cognitive representations is a critical determinant of social behavior. Most such relations are satisfying when they are consistent, when they "fit" together in a psychological sense. Sometimes, however, cognitions do not fit with one another. Festinger proposed that when two cognitions share an inconsistent relation, dissonance is present, creating an uncomfortable drive state that needs to be reduced. In the face of inconsistent cognitions, a person is motivated to alter his or her cognitive representations to reduce the discomfort and restore greater consistency among the relevant cognitions.

< previous page

page_227

next page >

< previous page

page_228

next page > Page 228

Dissonance occurs in a variety of circumstances in everyday life. Making a choice between two attractive alternatives (e.g., Brehm, 1956), working hard to obtain membership in a group (e.g., Aronson & Mills, 1959), or harming an innocent victim (e.g., Davis & Jones, 1960) have all been shown to arouse dissonance. The most frequently studied experimental situation in which dissonance has been aroused has been termed induced compliancea situation in which people advocate a belief that is inconsistent with their own attitudes and behavior. For example, in the original experiment, Festinger and Carlsmith (1959) had people participate in a task designed to be dull and boring. They were then asked to tell a person waiting in an adjoining room that the task was really exciting and interesting. It was predicted that freely producing a counterattitudinal statement that is at variance with one's true attitudes would create dissonance. Furthermore, it was predicted that people would reduce their dissonance by changing their attitudes toward the task. Indeed, relative to people who did not make a counterattitudinal statement, people who said the task was interesting came to believe it was interesting . . . with one important exception: If participants had been paid a considerable amount of money to make the counterattitudinal statement, they did not experience dissonance. As predicted, the large amount of money served as a cognition that was consonant with their counterattitudinal behavior and thus reduced the total magnitude of dissonance. Through its long history, dissonance theory has evolved. Although the original theory was characterized by its elegant simplicity, the subsequent research revealed that it did not completely capture all of the phenomena identified experimentally. As a result, a number of researchers proposed revisions to the original notion of psychological inconsistency. For example, Aronson (1969) proposed that people experience dissonance only if their self-concept is implicated in their counterattitudinal and inconsistent behavior. More recently, Cooper and Fazio (1984) proposed that the uncomfortable drive state of dissonance occurs when people hold themselves responsible for bringing about an aversive or unwanted event. In their view, dissonance begins with behavior and an assessment of the consequences that the behavior produced. If those consequences are judged to be aversive in that the person would rather that those consequences had not occurred, then the potential for dissonance exists. It causes the person to search for whether he or she is truly responsible for those consequences. If so, then the unpleasant state of dissonance is aroused. We can see how this combination of variables occurred in Festinger and Carlsmith's experiment. Participants not only said something they did not believe to the fellow student who was waiting in the next room, but they also duped him into being excited about the experiment in which he was about

< previous page

page_228

next page >

< previous page

page_229

next page > Page 229

to participate, despite their knowing that the experimental task was dreadfully dull. In Cooper and Fazio's view, this unwanted consequence is the primary cause of dissonance. Indeed, Cooper and Worchel (1970) systematically varied whether the waiting confederate was or was not convinced by the participants' statements. When he was convinced, dissonance-produced attitude change occurred. When the waiting confederate was not convinced (i.e., no unwanted consequence ensued), then no attitude change resulted. Furthermore, Scher and Cooper (1989) showed that if the consequences of an act are systematically disentangled from the attitudinal inconsistency of the act, the aversiveness of the consequence and not the inconsistency creates dissonance. Responsibility for the outcome is also critical for dissonance to occur. If a person feels that he or she was forced into the behavior (Davis & Jones, 1960) or could not foresee the consequences of the behavior (Goethals, Cooper, & Naficy, 1979), then dissonance does not occur. Thus, from aversive consequences perspective (Cooper & Fazio, 1984), inconsistency between attitude and behavior produces dissonance, but only if the behavior produces, or has the potential to produce, an unwanted outcome for which the person feels responsible. Dissonance at the Group Level Ironically, the first published dissonance research involved a very real social group. Prior to the publication of the formal theory, Festinger, Riecken, and Schachter (1956) introduced dissonance theory to the world in their study called, When Prophecy Fails. While preparing to publish his dissonance book, Festinger noticed a small article in the newspaper that announced that Mrs. Marion Keech, and a small group of believers known as The Seekers, were preparing to leave Earth for the planet Clarion on December 21. The article reported that Mrs. Keech had received messages that the Earth would end in a cataclysmic flood but that the Seekers would be escorted from Earth by a Clarion spaceship just hours before the cataclysm. Festinger et al. (1956) reasoned that cognitive dissonance was certain to develop on the afternoon of December 21 because the Seekers' firm belief in the early morning destruction would be discrepant from the likely observation that Earth still exists. Moreover, members of the group had undergone difficult personal consequences to adhere to their beliefs. Many had sold their possessions, left their occupations and even separated from their families to be prepared for the cataclysm. How would the members of the Seekers reduce their dissonance? Without experiencing massive denial, they could not change the cognitive representation of the reality that Earth was still intact on the afternoon of the 21st. They also could not change their knowl-

< previous page

page_229

next page >

< previous page

page_230

next page > Page 230

edge that they had believed in the cataclysm and had suffered serious personal and economic consequences in support of that belief. Festinger et al. predicted that this group, which had shunned publicity throughout its existence, would suddenly seek social support. They would abandon their policy of avoiding other people, the press and so on, and would avidly try to get others to believe in the correctness of their beliefs. What happened was striking. When the Seekers gathered during the early morning hours of December 21, they sat in shocked horror because they had been abandoned by their prophesied saviors from the planet Clarion. But suddenly, Mrs. Keech received a message. It stated, in part, that ''this little group, sitting all night long, had spread so much light, that God had saved the world from destruction.'' Immediately, the Seekers went to the phones to call the press, to tell them what had happened. They arranged speaking engagements, wrote flyers and press releases, all designed to convince others that their efforts had not been in vain. In psychological terms, the members reduced the negative consequences of their failed prophesy by creating cognitions that were consonant with their doomsday beliefs. Dissonance reduction was accomplished by convincing others that, through their actions and their faith, they had saved the world from destruction. Group Dissonance: The Case for Responsibility Diffusion It is easy to overlook the fact that the dissonance experienced by Seekers' members was very much a group phenomenon. For example, it is likely that having a cohesive social group made it easier for members to make the sacrifices they made to pursue their beliefs in the days prior to December 21. According to Festinger et al. (1956), group members supported each other during the weeks and months leading up to the expected cataclysm. Each member was encouraged by other members to engage in the behaviors necessary to be true Seekersbehaviors that meant an end to their prior social relationships, financial security, and identities as accountants, physicians, and so forth. We might expect these behaviors to arouse dissonance before December 21after all, the group members were committing acts that undoubtedly were non-normative. However, from a dissonance perspective, the group may have served as a means of avoiding taking personal responsibility for the consequences of these actions in two ways. First, part of the responsibility for behaving in consequential ways could be attributed to the group. If a member were to withdraw his or her savings, for example, the responsibility would be shared with other group membership. Second, individuals may have recognized that, in the remote possibility that the cataclysm did not occur, the responsibility for the very unwanted consequences of having ruined their individual lives could be diffused through

< previous page

page_230

next page >

< previous page

page_231

next page > Page 231

the entire group. We know that people will diffuse responsibility for potentially costly behaviors (Latané & Darley, 1970; Mynatt & Sherman, 1975). If anticipated responsibility diffusion occurred, it would explain why group members were so willing to take such extreme measures and risk such aversive consequences. The discomfort of the resulting dissonance might be reduced by diffusing responsibility to other members of the group. Even in the absence of anticipation, diffusion of responsibility may offer a plausible explanation for the bolstering of the group's beliefs that followed the failed prophesy on December 21. By proselytizing and seeking new group members, in effect, the group may have attempted to reduce dissonance by increasing social support for their beliefs. According to the account by Festinger et al. (1956), social support was sought not only from the general public but also from one another. Group members supported and reinforced the cognitive representation of social support as a satisfactory means of dissonance reduction. Without the internal support, the external support from nongroup members might not have functioned in a dissonance-reducing capacity. If they had not shared a common identity, the individuals might not have had a satisfactory way of dealing with the dissonance that arose from the failed prophesy. Group Dissonance Processes in the Laboratory Because the Seekers were an ongoing social group that, at least during its precataclysm days, avoided publicity and was averse to being studied, we cannot be certain how the events that transpired affected the processes that eventually resulted in the bolstering of the group's beliefs. The notion that social support can serve as a consonant cognition capable of reducing the magnitude of dissonance was speculative and had never been tested in well controlled research until about 25 years after When Prophecy Fails was published (e.g., Stroebe & Diehl, 1981; Zanna & Sande, 1987). In one test of how diffusion of responsibility might attenuate dissonance, Zanna and Sande (1987) invited three university students to come to the laboratory at the same time. All of the participants were asked to write essays that took attitude-discrepant positions. For example, students were asked to write persuasive essays advocating cutbacks in university funding, knowing that the essays might convince administrators to do just that. In the alone conditions, each individual was asked to write a separate essay. Although the three participants were in the same room, they were not formed into a group by a common task or common social identity. In the group condition, participants were asked to discuss their arguments together, decide which points would be used in the essay, and then write a common written document. Following the essay writing, participants filled out an attitude questionnaire on the issue about which they had written.

< previous page

page_231

next page >

< previous page

page_232

next page > Page 232

As suggested in our analysis of the Seekers' group, being in a group might enable individuals to feel less responsible for any unwanted consequence that accrues. A person may reason that, even if the administration of the university should use the essay to raise tuition, his or her own contribution to bringing about the event is diminished by the fact that he or she was only partially responsible for the creation of the essay. Zanna and Sande considered an alternate hypothesis based on the idea that "three heads are better than one." That is, if members of a group are asked to arrive at a common essay, and they anticipate that the consequence of a group product will be superior to an individual product, they will experience more dissonance compared to individuals who construct a counterattitudinal essay. To the degree the group is effective, a person may evaluate the potential consequence of participation in group activity as being more aversive than if it were merely his or her own individual product. The results of Zanna and Sande's experiment (1987) provide support for both hypotheses. They found that people in the group condition did consider it more likely that they would produce an essay that was more convincing than individual participants believed their essays would be. Nonetheless, attitude change in the alone condition was significantly greater than attitude change in the group condition. Apparently, the ability to diffuse responsibility to the rest of the group rendered dissonance reduction less likely in the group condition than in the alone condition. The implication of this finding is that group members successfully diffused the responsibility for the consequence of their essays. In fact, further evidence for the diffusion of responsibility in group dissonance situations was obtained when Zanna and Sande asked participants in the group condition how responsible each member of the group was for the final product. The results showed participants who scored high on responsibility diffusionthat is, who saw others in the group as more responsible than they wereshowed far less attitude change than did participants who saw themselves as more responsible for the group product. Thus, diffusion of responsibility seems to mediate dissonance reduction when people produce aversive consequences as members of a group. Social Identity and the Reduction of Dissonance People are members of multiple groups. Some of those groups have discrete membership boundaries (e.g., the Boy Scouts), others have nearly limitless boundaries (e.g., humans), and still others have fuzzy boundaries (e.g., extroverts). Tajfel and his colleagues (Hogg, 1996; Hogg & Abrams, 1988; Tajfel, 1982; Turner, 1982) emphasized the importance of social categories to the self-concept. We have a vast repertoire of such categories, and these categories help to define who we believe we are. They also help us to evalu-

< previous page

page_232

next page >

< previous page

page_233

next page > Page 233

ate ourselves, with our self-concepts depending on the evaluation of our social categories (Doise & Sinclair, 1973). Social categories vary in their importance to the self-concept in particular situations and at particular moments in time. According to Hogg (1996), a person's memberships in social categories form his or her social identity, which, in turn, both describes and prescribes his or her attributes as a group member. When a specific social identity becomes salient, both behavior and self-perceptions become stereotypic and normative. That is, when a particular social identity is salient, people tend to act as the stereotypical member of that group. They adhere more closely to the norms of that particular group and tend to evaluate themselves as a function of the evaluation of their group. They are motivated, therefore, to evaluate their group more highly, to favor it in the allocation of resources, and to derogate alternative groups, or outgroups. One of the advantages of social identity theory is that it makes clear the notion that people constantly negotiate their identity and that their identity is inextricably intertwined with the multiplicity of their social groups. Thus, group membership is a consistently available cognitive category that may come into play as a person prepares to act in a variety of situations. Although it has rarely been studied, a person's social identity as a group member may determine the way in which dissonance is resolved. People have several identities available to them. A fictitious Mr. A. is a male, a father, a rugged individualist, a corporate executive, a Republican, a Little League coach, and a member of the Lions Club, among other social identities. He attends a dinner party at which many of the high-ranking executives are rubbing elbows with policymaking legislators of the Democratic Party. When idle conversation turns to talk about the president of the United States, Mr. A. voices the opinion that the Democratic president is doing a fine job. This attitude discrepant statement, made freely in front of other executives and politicians, should result in the unpleasant tension state of dissonance. How will dissonance be resolved? The most typical prediction to make for this induced compliance situation is that Mr. A. will change his attitude to favor Democrats and to support the Democratic president. We propose, however, that the typical prediction must be qualified by social identities that become accessible in the situation. If Mr. A thinks of himself as a corporate executive, dissonance may well be reduced through attitude change. However, if Mr. A.'s social identity as a Republican is salient, he may not be able to reduce his considerable dissonance by becoming more of a Democrat. More generally, the cognitive representations of group membership vary in importance and in accessibility. We propose that important and accessible social identities fundamentally may alter the way in which dissonance is

< previous page

page_233

next page >

< previous page

page_234

next page > Page 234

resolved. Some social identities may be critically important in determining one's self esteem and self concept (Doise & Sinclair, 1973). Yet, those social identities may not be accessible in a given situation. Mr. A. knows that he is a member of the Republican Party and that this group is critically important. But, while shopping, taking his children to school, or performing his job as an executive, his social categorization as a Republican is not immediately accessible in memory. On the other hand, if something about the situation makes his group membership salient, then social categorization processes will make his standing as a Republican accessible. The accessibility of a critically important group membership may preclude his reducing dissonance in the manner that more typically characterizes people for whom relevant social identities are less accessible or less important. The process of social categorization relies on a determination of the dimensions that distinguish one's own group from other possible group. Cooper and Mackie (1983) suggested a distinction between definitional and associated attributes. Definitional attributes are fundamental to the group category and are a necessary condition for group membership. Associated attributes may be possessed by most, or even all, members of the group but failure to possess them does not automatically preclude membership. For example, being Christian is a definitional attribute of being a member of a campus extracurricular group of Evangelical Christians. On the other hand, having an attitude that supports prayer in public schools may be an associated attitude. One cannot be an Evangelical Christian without being Christian. However, one can be an Evangelical Christian while nonetheless believing that Christianity should be espoused in places other than publicly supported educational institutions. If a person is induced to make a statement that is contrary to the definitional attribute of group membership, how will dissonance be resolved? Assuming the group is important and that membership in the group has been made salient, then we propose that changes of definitional attitudes will not occur. Rather, the individual needs to find alternate methods to reduce the unpleasant arousal state. The Evangelical who makes an anti-Christian speech will not change his attitude about Christianity. An officer of the Sierra Club will not change her attitude about environmental pollution. Group membership, when it is important and accessible, precludes straightforward attitude change. Social categorization and social identity contribute to dissonance theory the important insight, implicit in Festinger's original formulation, that attitude change is only one way of accomplishing dissonance reduction. It is not an effortless way to reduce dissonance (Zanna & Azizza, 1976), but it is typically the cognition least resistant to change (Wicklund & Brehm, 1976). Under circumstances in which important and accessible social categoriza-

< previous page

page_234

next page >

< previous page

page_235

next page > Page 235

tion precludes attitude change, another method must be found to deal with the unpleasant arousal. We suggest, and show evidence to support, two such alternatives to attitude change when group membership precludes attitude change: hostility toward the outgroup and repression. Dissonance and Hostility Toward the Outgroup A number of studies have supported the proposition that the uncomfortable tension state of dissonance can be misattributed to other factors in the environment, especially if circumstances make it difficult for someone to change attitudes or if a plausible cause of uncomfortable arousal is made salient in the environment. For example, Zanna and Cooper (1974) had college students volunteer to ingest a pill believing that the pill was part of a memory study. They were also led to believe that the pill had side effects, one of which was to create a mildly unpleasant state of arousal. (In fact, the pill was only milk powder and had no real side effect.) Soon thereafter, the participants were induced to write a counterattitudinal essay whose aversive consequences made dissonance arousal likely. Participants who believed that the pill they had taken caused the uncomfortable arousal did not manifest attitude change as their dissonance reduction strategy. Instead, with the side effects of the pill made salient, the arousal due to dissonance was mistakenly attributed to the pill. Similarly, Fazio, Zanna, and Cooper (1977) demonstrated that political conservatives who advocated liberal positions and liberals who advocated conservative positions also preferred to believe that their discomfort was due to some disturbing aspect of the room in which they were questioned (see also Fried & Aronson, 1995). If people in a cohesive group are asked to make an attitude-discrepant statement on an issue that is definitional for their group membership and thus threatening to their social identity, they will be in a state of uncomfortable arousal. The prediction is that they will find an alternate source to which to attribute that arousal. Previous research has shown that the source has to be a hedonic match to the tension state for it to serve as a way to reduce dissonance arousal. That is, the source itself must be capable of producing negative affect (Higgins, Rhodewalt, & Zanna, 1979; Losch & Cacioppo, 1990). A neutral source or a stimulus that produces positive affect will not do. One source of negative affect to which group members can misattribute their dissonant state is the outgroup. We know that social categorization makes the attractiveness of members of outgroups suffer (Hogg, 1996; Tajfel, 1982). A group whose purpose and identity is not only different from, but contradictory to, a person's ingroup is likely to cause negative affect. As such, if it is salient when people are in the throes of cognitive dissonance, the outgroup may serve as a convenient source of misattribution. Placing

< previous page

page_235

next page >

< previous page

page_236

next page > Page 236

additional negative arousal (due to dissonance) onto a group that already is unattractive and the source of its own negative affect is likely to lead to additional derogation of, and hostility toward, the outgroup and its members. Cooper and Mackie (1983) established an experimental test of the hypothesis that, when group membership is made salient, people find alternate modes of dissonance reduction rather than change an attitude that is a definitional attribute of their social group. Specifically, we examined the hypothesis that dissonance reduction would lead to greater hostility and derogation. We sought a cohesive group that had clear definitional attributes. The year was 1980 and people in the United States were deciding between electing Ronald Reagan as president or re-electing Jimmy Carter to a second term. University campuses had their own versions of political organizations ranging from the far right to the far left of the political spectrum. We chose to invite student members of Youth for Reagan into our laboratory for a study. Participants volunteered for two sessions. The first consisted of their filling out a number of attitude scales related to the election. They indicated their opinions about their group, about the two major candidates, and about a number of opinion items that were related to the positions taken by the candidates. About a week later, the participants returned to respond to a second questionnaire. They were greeted by a second experimenter who explained that the original researcher was delayed but that she was also doing research about the elections and asked the participants if they wouldn't mind participating in her research while they waited for the original experimenter to arrive. Participants were told that the research was about communicating political positions and that their task was to make strong and forceful statements on one side of a particular issue. Half of the participants were randomly assigned to the definitional attitude condition. The experimenter explained, "What I would like you to do is to write out the arguments supporting the re-election of President Carter." In the associated attitude condition, the participants were asked to write in support of extending government-funded health care programs. The testing in the first session had revealed that participants were highly negative about Carter's re-election and equally negative about extending government health care. However, the former was thought to be at variance with the pro-Reagan attitude that defined the participants' social group. The health care issue, on the other hand, although clearly correlated with attitudes toward Ronald Reagan, was not definitional. It was possible, although unlikely, for a Reagan supporter to agree with expanded health care. At the very least, one could still be a member in good standing of Youth for Reagan, even if one supported government health care.

< previous page

page_236

next page >

< previous page

page_237

next page > Page 237

The participants' personal responsibility for the attitude-discrepant act was varied by systematically manipulating the degree of choice a person had for writing the attitude-discrepant statements. Half the participants were asked if they would help out by writing the statements. The other half were merely told that their task was to write arguments in support of President Carter (or expanded health care.) As in much previous research, dissonance should be aroused for those participants who freely wrote counterattitudinal arguments. It should not exist for participants whose choice was low. After the participants wrote the arguments, the original experimenter returned, armed with her battery of questionnaires. For our purposes, the key questions were participants' attitudes toward the election of Ronald Reagan versus Jimmy Carter, government-sponsored health care, their Youth for Reagan group, Democrats in general, and a specific campus outgroup, Students for Carter. The results of the attitude measures were as predicted. As Figure 12.1 indicates, members of Youth for Reagan who wrote counterattitudinal arguments in favor of expanded health care showed increased support for governmentsponsored health care when they freely wrote their arguments compared to when they were told to do so. However, when the attitude was

FIG. 12.1 Favorability of attitude toward expanded health care, as a function of nature of argument and degree of choice in writing argument. Lower numbers indicate greater favorability.

< previous page

page_237

next page >

< previous page

page_238

next page > Page 238

an issue that was defining for group identity, attitudes did not change. There was no difference between the high-choice participants, who were presumably experiencing dissonance arousal, and low-choice participants who had no dissonance arousal. We predicted hostility toward the outgroup would characterize high-choice participants who wrote essays in support of Jimmy Carter. Figure 12.2 shows that this is what occurred. The figure shows that the Youth for Reagan members came to derogate Carter supporters, but only if they had written pro-Carter essays under conditions of high choice. Apparently, writing dissonance-producing essays on attitudes that define group membership precludes attitude change as a means of dissonance reduction and leads instead to derogation of an already negatively valued outgroup. Repression as an Alternate Form of Dissonance Reduction The study by Cooper and Mackie (1983) demonstrates the tremendous difficulty of changing attitudes that define membership in important social categories. In Cooper and Mackie's study, the alternate form of dissonance

FIG. 12.2 Favorability of attitude towards Carter supporters as a function of nature of argument and degree of choice in writing argument.

< previous page

page_238

next page >

< previous page

page_239

next page > Page 239

reduction was derogation of the outgroup. Note that the dissonance-reducing phenomenon created by social identity concerns contained a certain degree of irony. People wrote essays that were pro-Carter and, in contrast to the usual effect of dissonance, the participants ended up disliking Carter's supporters and any group that supported Carter. This attests to the power of group membership in altering the dissonance reduction process. The battle between changing attitudes to reduce dissonance and maintaining attitudes that anchor one's social identity is tipped in favor of the latter. But that, in turn, only leads to greater dissonance and the need for the dissonance to be resolved in some other less direct way. We now look at a somewhat different situation in which group membership and social identity force dissonance to be resolved in yet a different way. In this situation, people are motivated to forget discrepant, dissonance producing information. In an early dissonance study reported by Brock (1962), people who were not of the Catholic faith were asked to write an essay on "Why I would like to be a Catholic." The results showed that, under high-choice condition, the nonCatholic students became somewhat more favorable to Catholicism than they had been previously. From this study, we know that people can express attitude discrepant religious statements and manifest attitude change as a result. However, we can make Brock's situation more akin to the Cooper and Mackie (1983) example by involving people whose social identity is characterized in strong, coherent religious sentiments. It is unlikely, for instance, that Evangelical or fundamentalist Christians would change their attitudes toward Christianity if they were induced to make statements that questioned the supremacy of Christ. Similarly, it is unlikely that members of the orthodox Hassidic Jewish sect would question their own religious beliefs even if they were induced to write statements supporting the contention that Jesus was the son of God. These group members cannot change their attitudes toward their Christian or Jewish premises any more than the Youth for Reagan members could change their attitudes about the election, for those attitudes define their identities as group members. Suppose a person is exposed to counterattitudinal information on a topic that is a defining attitude for his or her social identity. If the exposure is voluntary and dissonance is high, we nonetheless suspect that the person will not change his or her attitude. One way for the person to deal with the dissonance created by the counterattitudinal message is to deny or forget the content of the message. Forgetting may serve as a means of reducing the unpleasant tension state of dissonance. Forgetting in the service of avoiding an unpleasant or unacceptable affect is what Freud meant by repression (Freud, 1921). Therefore, we may predict that the ego defense mechanism of repression can serve as a mode of dissonance reduction in situations in which attitude

< previous page

page_239

next page >

< previous page

page_240

next page > Page 240

change is precluded by a person's social identity of his or her group membership. In the context of a memory experiment, participants were asked to read, memorize and then recite two paragraphs of written material. One paragraph contained relatively benign material on the heliocentric nature of the universe. It contained statements such as ''while there have been people who would have objected . . . science can now tell us that the earth is indeed round and that it revolves around the sun.'' The second paragraph was about religion and questioned the supremacy of Christ and the Christian religion. It posited that the Buddhist religion is the most convincing, the purest, and "comes closer than any other religion in offering salvation to people." Each paragraph consisted of six sentences totaling 60 words and judged to be equal in reading difficulty. The participants were selected in two different ways. Half of the participants were recruited from a campus group that described itself as "Evangelical Christian." The group was unified in a strong belief in a personal religious conversion, the efficacy of prayer, and the unique expiatory nature of Christ. The other participants were people who identified themselves as church-going Christians but as unaffiliated with the Evangelical movement. All of the participants considered themselves to be Protestant. The task of the participants was to read the pro-Buddhist and heliocentric messages, memorize them, and then tape record them as accurately as possible, as though they were personally making the speech contained in the message. To create dissonance in half of the participants, they were asked if they were willing to participate and given the opportunity to decline. Low-choice participants, on the other hand, could absolve themselves of any responsibility because they were not given any opportunity to decline. The dependent measures in the study included the participants' attitudes toward Buddhism, their attitudes toward the position that the solar system is heliocentric, and their memory for both paragraphs. First, it should be mentioned that a control group of evangelical and nonevangelical Christians was shown the two paragraphs. They did not differ in their degree of disagreement about the supremacy of the Buddhist religion (nor about the heliocentric nature of the universe). Figure 12.3 presents the results of the attitude measure for the experimental groups. As expected, nonevangelical Christians who read, memorized and recorded counterattitudinal statements under high choice showed significantly more agreement with the paragraph than they did when choice was low, thus replicating a typical dissonance result. However, the Evangelicalslike the Reaganites of the previous studycould not change their defining attitudes regardless of choice condition. This did not mean that high-choice, Evangelical participants did not experience dissonance. If we

< previous page

page_240

next page >

< previous page

page_241

next page > Page 241

FIG. 12.3 Agreement with counterattitudinal statements as a function of salience of attitude toward social identity and degree of choice in reading statements. examine the memory measure presented in Figure 12.4, we can see that high-choice Evangelical participants could not remember as much of the paragraph as could each of the other groups of participants. Note that the low-choice Evangelicals also showed diminished memory for the paragraph, but they had significantly better memory than their high-choice counterparts. Notice, also, that none of the four groups differed in their recall of the neutral paragraph. Thus, the failure to recall the religious, counterattitudinal message was not a function of poor memory, but it appears to be motivated forgettingthat is, repression in the service of reducing, or possibly avoiding, dissonance arousal. Like the data of the Cooper and Mackie (1983) study, the results of the current experiment show that people whose social identity is threatened by the counterattitudinal message find an alternate way of dealing with cognitive dissonance.

< previous page

page_241

next page >

< previous page

page_242

next page > Page 242

FIG. 12.4 Word recall as a function of nature of passage (high vs. low group salience), group status (evangelical vs. nonevangelical), and degree of choice in reading statements (high vs. low).

< previous page

page_242

next page >

< previous page

page_243

next page > Page 243

Conclusions and Speculations We have noted at least two ways that group membership and social identity can affect the dissonance process. First, as the research by Zanna and Sande suggests, groups can help individuals to diffuse responsibility for aversive consequences they have created and thus reduce the magnitude of dissonance arousal. An interesting suggestion is that the diffusion of responsibility may occur prior to the advent of the consequences. As Festinger (1957) noted in his original presentation of the theory, people may anticipate experiencing dissonance. Dissonance anticipation may be partially accountable for people limiting the aversive consequences they are willing to risk setting in motion. If people anticipate that, as members of a group, they will be able to diffuse responsibility and experience less dissonance, then people in groups or collectives may be willing to instigate more aversive consequences than they would as individuals. This may offer a partial explanation for the tendency of some groups to engage in more negative behavior than the individuals would if acting alone. The second influence of the group on cognitive dissonance is the impact of group membership in causing social categorization and social identity. By anchoring one's social identity in one's group, it may be impossible for an individual to reduce dissonance in a way that would threaten that social identity. As we have seen in two studies, people who produced or were exposed to counterattitudinal arguments on issues that were defining characteristics of their important ingroups did not resolve their dissonance by attitude change. The dissonance reducing vehicle that operates in countless numbers of induced compliance situations did not serve as potential routes to dissonance reduction. Instead, people used the motivated forgetting of repression or attributed their negative arousal to the outgroup, coming to dislike them more as a way to alleviate their dissonance. Some Thoughts About Process Further research is needed to identify the processes by which social identity leads to alternate forms of dissonance reduction. In this chapter, we have assumed that the influence of the social group occurred because group membership was important and the group, its values and its norms, were made accessible in memory. These variables can usefully be studied in subsequent research. By using Youth for Reagan members in the height of a presidential election campaign, and using evangelical Christians whose members had devoted several hours per week to the cause of proselytizing for fundamentalist beliefs, we had chosen groups that were high in importance to their members. The assumption that group importance is a key variable in blocking attitude change on definitional items can be put to the empirical test.

< previous page

page_243

next page >

< previous page

page_244

next page > Page 244

It is equally important to study the assumption of accessibility of the cognitive representation of social identity, for this strikes at the core of the mental processes that may produce social identity's effect on dissonance. Accessibility of one's social identity may affect dissonance in two ways. One is chronic accessibility. There may be some people for whom group-anchored social identity is always accessible. One can think of religious, political, or social group members who always view themselves as the prototype of their group. They live, breathe, and sleep as, say, militia members, fundamentalists, Boy Scouts, and so on. For such people, group membership is always accessible, constantly guiding their thoughts and their actions. Perhaps more interesting (at least, to students of social cognition) is the accessibility of a social identity that is caused by a person's immediate situation. As we indicated earlier, people have multiple social identities. They may be parents, recreational athletes, Boy Scouts, Republicans, and professors. The cognitive representations of their social groups may be equally available but not equally accessible. To become accessible, situational cues might bring a particular social identity to the foreground of memory. The procedures of the studies that were presented here may have made social identity cognitions accessible, although accessibility was not systematically varied in that research. Members of an evangelical Christian group were specifically recruited for one study and the nature of the attitudinal material was directly relevant to their group membership. In the prior study, members of a specific campus political group were recruited and the counterattitudinal essay they wrote was directed pointedly at their core political beliefs. It is quite reasonable that the procedures made their religious and political social identities immediately accessible and that those cognitions then guided their attitudes and behaviors. Varying accessibility of social identity systematically is important. This might be done by priming procedures in which alternate representations are made accessible. The relevant group for which social identity is expected to preclude attitude change can be primed, as can an alternate group for which one's social identity would not be threatened by counterattitudinal behavior. One's role as an individual can also be primed by experimental manipulation. In that way, we can obtain evidence for the mediating process by which social identity alters the course of dissonance reduction. In conclusion, the often ignored roles played by group membership, social categorization, and social identity offers the promise of expanding our knowledge of cognitive dissonance. The possibility that responsibility diffusion can alter dissonance arousal and that social identity can systematically alter the potential modes of dissonance reduction are fruitful avenues for future research.

< previous page

page_244

next page >

< previous page

page_245

next page > Page 245

13 Social Influence and Identity Conflict Juan Manuel Falomir Gabriel Mugny University of Geneva Juan Antonio Pérez University of Valencia One of the findings most regularly observed in social psychology is that certain kinds of behavior do not always relate to attitudes. From the point of view of social influence, this observation brings us back to the fact that changes in attitudes and beliefs do not necessarily involve a change in the corresponding behaviors. Indeed, attitude change, of the kind often obtained with influence strategies modeled on learning or information processing, seldom translates into a change in behavior and on this point information campaigns seem to be characterized by a relative lack of effectiveness (cf. Hyman & Sheatsley, 1978; Lewin, 1978; McGuire, 1985; Roberts & Maccoby, 1985). This shortcoming assumes a particular importance when modification of behavior is the primary goal of social influence, especially when the behaviors of concern are health-threatening. One notable example is smoking and the campaigns aimed at eradicating this habit, the focus of this chapter's analysis. Ever since the first discoveries of a link between tobacco consumption and the likelihood of developing cancer (e.g., Doll & Hill, 1950, 1952), bodies with a responsibility for public health have been putting their efforts behind campaigns to ensure that people, and particularly those who are smokers, become aware of this link and consequently decide not to smoke (e.g., Generalitat Valenciana, Direcció General de Salut Pública, 1995; World Health Organization [WHO], 1988, 1993; see also Roemer, 1993).

< previous page

page_245

next page >

< previous page

page_246

next page > Page 246

These campaigns, like others that have had the aim of inducing healthy behaviors (e.g., sexual practices that reduce the risk of contracting human immunodeficiency virus), have been devised around, and primarily inspired by, models of persuasion (e.g., McGuire, 1985; Petty & Cacioppo, 1986a) and fear appeals (e.g., Janis & Feshbach, 1953; Rogers, 1975) as means for motivating smokers to give up smoking and motivating nonsmokers to maintain their abstinence (cf. Leventhal & Cleary, 1980; Maibach & Parrott, 1995; Stroebe & Stroebe, 1995). Although it would be no easy task to summarize and evaluate all such public information campaigns, they do appear to have a significant impact insofar as awareness of the risks linked to tobacco consumption is concerned. For example, whereas, in 1964, 66% of Americans claimed to know that smoking increased the risk of contracting lung cancer, by 1987, this proportion exceeded 90% of the population (cf. American Lung Association, 1987; see also Dawley, Fleischer, & Dawley, 1985; U.S. Department of Health and Human Services [USDHHS], 1989). This recognition of the health risk also seems to have occurred among the smokers who largely shared the views of the nonsmokers in this respect (cf. Cecil, Evans, & Stanley, 1996; GVDGSP, 1993; Kristiansen, Harding, & Eiser, 1983; Pervin & Yatko, 1965), both groups acknowledging the benefits for health in giving up smoking (cf. Brownson et al., 1992). Finally, the same opinions seem also to have penetrated to a much younger section of the population (down to and including 10year-olds) who display an awareness of the negative consequences of smoking and for the most part concede that smoking represents a health problem (cf. O'Rourke, Smith, & Nolte, 1983). Furthermore, smokers seem to accept the discrimination to which they are subjected (cf. Echebarria Echabe, Fernandez Guede, & Gonzalez, 1994), accepting a ban on smoking in public places (cf. GVDGSP 1993) or else favoring the introduction of antismoking measures (e.g., campaigns of persuasion, restrictive measures, fines; cf. Eroski, 1988). But now let us consider the other side of the coin. Smokers do show some reluctance to accept the antismoking argument in its entirety. Thus, they have less exposure than nonsmokers to negative information about tobacco (cf. Becoña, 1995; Brock & Balloun, 1967; Feather, 1962), they are disinclined to accept the inevitability of the negative consequences of smoking (cf. Becoña, 1995) or to recognize the probability of specific illnesses (cf. Cecil et al., 1996; Eiser, Sutton, & Wober, 1979), and they regard the relation between the arguments for and against tobacco consumption as more evenly balanced than do nonsmokers (cf. Eiser & Harding, 1983). This resistance also can be observed with respect to giving up smoking. Although the majority of smokers have attempted to give up smoking at least once in their lives (cf. Schachter, 1982), the percentage of smokers who try to quit

< previous page

page_246

next page >

< previous page

page_247

next page > Page 247

seems to have diminished more recently. For example, 70% of Spanish smokers have no intention of giving up smoking in the near future (cf. Becoña, 1995; GVDGSP, 1993). The current level of tobacco consumption in the world population is estimated by the WHO to be 1,100 million people (cf. WHO, 1994, 1996). If some studies show that a proportion of smokers succeed in giving up smoking (e.g., 13% of the Spanish population are ex-smokers; cf. Ministerio Español de Sanidad y Consumo, 1993; see also USDHHS, 1989), the tendency since the beginning of the 1990s seems to show a relative stability in the number of smokers and the number of cigarettes consumed (more than 30% of the Spanish population are current smokers; cf. Becoña, 1995; Tabacalera, 1994). Furthermore, the rate of initiation of smoking among adolescents seems to be on the increase again, after reaching a plateau in the 1970s and remaining stable through the 1980s (cf. Altman, 1990; Lynch & Bonnie, 1994). In brief, antismoking campaigns seem to transmit antismoking information and to establish a social norm against smoking that are accepted by smokers even if they show some reticence about accepting the case in its entirety. It is noteworthy, however, that the effectiveness of persuasive campaigns at the level of beliefs and attitudes is not matched by the number of smokers who give up smoking or by the number of nonsmokers who never start (Chassin, Presson, & Sherman, 1990; Leventhal & Cleary, 1980; Lichtenstein, Biglan, Glasgow, Severson, & Ary, 1990; Office of Population Censures and Surveys, 1990; Russell, 1990; WHO, 1994). This fact shows that knowledge of the risks to health does not suffice to prevent smoking (Frydman, 1987; Johnston, O'Malley, & Bachman, 1989; Thompson, 1978) and suggests that antismoking campaigns must draw inspiration from other theoretical sources if they are to overcome this inconsistency at the behavioral level. This chapter describes a body of research on social influence that aims to theorize several of the limitations which antismoking campaigns can suffer in their efforts to convince smokers to abandon smoking. Social Influence and Identity Threat A common assumption of antismoking campaigns seems to be that smokers' becoming aware of the negative consequences of smoking is sufficient to cause them to give up smoking. However, these beliefs do have to come into direct conflict with the behavior involved in smoking (the act of smoking); that is to say, it is important that the beliefs do not remain segregated from concrete decisions smokers make with respect to their practice of smoking (cf. Eiser, 1982). The question arises, therefore, as to the nature of the factors that facilitate (or inhibit) such a conflict and that allow this conflict to be resolved in such a way that it changes the behavior.

< previous page

page_247

next page >

< previous page

page_248

next page > Page 248

Several factors could account for the failure to change behavior. One, for example, is the fact that smokers consider the subjective benefits associated with the act of smoking (e.g., the pleasure of smoking) are at least as important as the risks incurred (cf. Bauman & Chenoweth, 1984; Mausner & Platt, 1971); their basic ambivalence here could be resolved in favor of maintaining the habit. The absence of fear about negative consequences of smoking for health in the short term (Evans, 1976; Frydman, 1987), or underestimation of the perception of personal vulnerability (Janz & Becker, 1984; Witte, 1992) also could favor a lack of change in behavior. Expectations of risk to health seem in practice to predict neither the likelihood of taking up smoking nor the subsequent rate of consumption (Bauman & Chenoweth, 1984). Certain norms favorable to tobacco consumption to which the smoker may be exposed (such as the encouragement of friends or lack of disapproval by parents) also tend to work against any intention to stop smoking (cf. Fishbein & Ajzen, 1975). Furthermore, the perceived gravity of health problems is reduced to the degree that the individual observes some consensus in performing those actions that supposedly provoke the risks (ermott, Ditto, & Croyle, 1986). Finally, perceived inability to stop smoking (Ajzen, 1988; Eiser, van der Pligt, Martin, & Sutton, 1985); the habit of consumption itself (cf. Bentler & Speckart, 1979; Triandis, 1977); and the personal norms, values, and affective evaluations of the behavior (cf. Fishbein, 1967b; Manstead & Parker, 1995) also potentially determine the occurrence of the relevant behavior. Despite the general tendency to explain the persistence of tobacco consumption largely in terms of internal factors (nicotine dependence) rather than external factors (social pressure; cf. Chassin et al., 1990), various aspects of the social context in which antismoking campaigns occur also can cause smokers to resolve the conflict in ways that have repercussions for behavior. Social context here is taken to include the social system in which smoking is embedded (i.e., the values, norms and social representations of smoking) as well as the social relationships that smokers maintain with their acquaintances (i.e., with nonsmokers) and in particular with sources of antismoking influence. In this social context, the threat to identity of which the smoker becomes aware following exposure to the information disseminated in antismoking campaigns seems especially important in conceptualizing the passage from attitude to behavior. Two aspects define such a threat to the identity of the smoker. The first concerns the implications of smokers' own behavior and the attitudes that correspond to this. The act of smoking locates smokers in the social field by placing them within the visible social category of smokers. Thus, tobacco consumption must be regarded as associated with a "strong"

< previous page

page_248

next page >

< previous page

page_249

next page > Page 249

attitude (cf. Petty & Krosnick, 1995), by virtue of the fact that it contributes to the definition smokers have of themselves, that it will be linked to values that are important to smokers, and that it will therefore carry a high affective value. In addition, because of the habitual nature of the practice and the benefit that smokers derive from tobacco consumption, they will be especially concerned about restrictive measures (cf. Petty & Cacioppo, 1979a, 1979b). Finally, the habitual practice of smoking also defines the smoker as someone strongly committed to tobacco use (cf. Pomerantz, Chaiken, & Tordesillas, 1995). In brief, given that attitudes to smoking and the behavior associated with it are important to smokers and are linked to their self-concept, any attempt to change these is bound to introduce a threat to their identity. The second aspect derives from the fact that smokers find themselves in a socially stigmatized position. They are aware of the relatively weak case in favor of smoking and even smokers will produce more arguments against than for tobacco (cf. Pérez, Mugny, Roux, & Butera, 1991). They share with nonsmokers a dominant representation of tobacco smoking, which contains a rather negative image of smokers (Echebarría et al., 1994). Their legitimacy or social normativity has become progressively weakened (i.e., restrictive measures on behavior) to the point where nonsmokers have become a distinct group defending interests that conflict with their own (cf. Roemer, 1993). Finally, they are the targets of major campaigns against not only tobacco but also their behavior and their identity (e.g., Hirsch, Hill, Frossart, Tassin, & Pechabrier, 1988). This social change is therefore composed of various convergent elements that challenge the content and the value of the attitudes and behavior of smokers and consequently their identity. One must therefore recognize that smokers now find themselves in a social context that questions their identity. One might expect these circumstances would lead them to stop smoking and this is indeed the case for a certain proportion of smokers. In effect, one way of coping with this threat to identity is to accept the antismoking point of view, which means acknowledging the criticism of their identity. If this provides the necessary encouragement to adopt a more valued identity, then internalization of the positions advocated by an antismoking source represent one possible route to the restoration of a positive self-concept. Stopping smoking would in this case be an adequate strategy for facing the threat to identity. To sum up then, one means for restoring a positive image of the self comes down to changing attitudes and behavior via a kind of mobility of identity (cf. Tajfel & Turner, 1979), which is to say the acquisition of a new identity that allows escape from the negative attributes of the old.

< previous page

page_249

next page >

< previous page

page_250

next page > Page 250

From Identity Threat to Identity Conflict The threat to the identity of smokers derives its meaning, however, from contexts that tend to inhibit the decision to stop smoking, and the reality is that a majority of smokers do not quit. Why do not factors questioning the identity of smokers as a result lead them to change? A particular feature of the behavior and attitudes involved is their temporal stability (Eagly & Chaiken, 1993, 1995; Krosnick, 1988) as much as the motivation of the individual to preserve them (Chaiken, Giner-Sorolla, & Chen, 1996). Additionally, insofar as stopping smoking presupposes a change of identity (i.e., moving from the identity of smoker to that of nonsmoker or ex-smoker), then a strong identification with the membership group would inhibit social mobility (Tajfel & Turner, 1979, 1986) and should create resistance to change. In brief, accepting that antismoking campaigns seek an impact in part by questioning and stigmatizing the identity of smokers (i.e., seeking to convince them of the unreasonable and illegitimate character of their behavior), raises the possibility that smokers will be aroused to defend their identity. The issue we return to therefore is how to determine when smokers will be motivated to seek a more valued identity via a strategy of social mobility, renouncing their status as smokers, and when instead they will be more motivated to defend it. With respect to this latter alternative, the social context provides an explanation for the type of defense offered (cf. Breakwell, 1986). The existence of a threat inherent in the identity in effect adds a special element to the process of influence: The target of influence may relate the content of the threat to salient features of the social relationship (Falomir, Mugny, Sanchez-Mazas, Pérez, & Carrasco, 1998) and thus be motivated to protect his or her own identity as a response to the context of this relationship. Influence at the level of behavior will consequently be obstructed by the fact that the smoker responds in terms of a conflict of identity, a response characterized by a disposition to preserve initial attitudes and behaviors, and by the activation of strategies for the defense of identity (Breakwell, 1986). The threat to identity activated by an attempt to influence will introduce conflict into tobaccorelated behavior and attitudes. But smokers will work out this conflict in a fashion that is primarily sociorelational and will restore their identity as smokers as a response to this salient relational context. The general hypothesis is that the resistance induced by such a conflict of identity does not allow the integration of alternative viewpoints. To sum up, when their identity is threatened, smokers will defend it rather than abandon it, as they are concerned about their own relation to the social context, in this case, their relationship with the source of influence. Working out the conflict in this way prevents them making a decision to stop smoking.

< previous page

page_250

next page >

< previous page

page_251

next page > Page 251

When does the social influence context allow smokers to accept a challenge to their identity and thus work through the conflict in such a way that they are led to stop smoking, and when does it push them instead toward protecting their existing identity so that the conflict prevents any such change? Given that the conflict involves interpreting the threat to identity in relation to salient elements of the social influence relationship, such a resolution depends on the presence and the salience of these elements of the social relationship. In the remainder of this chapter we consider two elements of the social influence context that are likely to determine the nature of the conflict: the relation between the groups present in such a context, and the influence relationship established between the source and the target of influence. Intergroup Conflict and Identity Conflict We have proposed that behavior and attitudes toward tobacco consumption are determined by the position smokers occupy in an unequal social relationship, one in which they appear to belong to a socially stigmatized category (that of smokers) opposed by a dominant social category (that of nonsmokers). This allows us to advance various predictions concerning the way smokers interpret the threat to their identity. We can postulate that the salience of an intergroup conflict is more likely to motivate smokers to respond to this identity threat as a conflict of identity in which preservation and protection of their current identity is the principal objective. A study of social representations of smoking illustrates the effect of the intergroup conflict on the defensive motivation of smokers (Echebarria et al., 1994). In two experimental conditions, university students (smokers and nonsmokers) responded to a questionnaire, having been led to believe either that they were participating in ''a study about tobacco'' (no conflict condition) or else that they were participating in a "comparative study between smokers and nonsmokers about tobacco" (intergroup conflict condition). To reinforce this manipulation, in the intergroup conflict condition classes of students were asked to split into two groups. The smokers were asked to sit together on the right hand side of the classroom and the nonsmokers to sit together on the left hand side. First, let us note the existence of two kinds of social representation of tobacco use. The majority, which included 90.1% of the nonsmokers and 73.5% of the smokers in the study, described the consumption of tobacco in psychopathological terms and expressed negative feelings toward smokers but positive feelings about nonsmokers (negative psychopathological representation). The minority view, shared by 9.9% of nonsmokers and 26.5% of smokers, was to regard tobacco consumption in a nonstigmatized fashion, and to hold a positive stereotype of smokers (defensive positive representation).

< previous page

page_251

next page >

< previous page

page_252

next page > Page 252

Regarding intergroup conflict, the number of smokers who held a positive representation was found to be significantly greater (41.2%) when the social categorization was made salient. This result shows that a social context which reinforces the categorization between smokers and nonsmokers strengthens the defensive motivation of smokers because they shift significantly toward a social representation of tobacco consumption that provides more protection for their identity. Identification with smokers as a group also was found to be the principal factor underlying this representation. These results confirm, therefore, the importance of aspects of the social context (in this case the salience of the smoker vs. nonsmoker categorization) in determining the motivation to protect a socially threatened identity. It is notable that most antismoking campaigns tend to make this categorization salient. If categorization by itself is enough to prompt the activation of processes of resistance to change in behavior, then how can these difficulties be overcome? One means for achieving influence in a social context marked by a salient intergroup categorization is to employ sources of influence that are part of the ingroup, and this will at the same time induce conditions for self-categorization (cf. Turner, 1991) rather than categorical differentiation (cf. Doise, 1978). This idea was tested in a study which confronted smokers with a strongly argued case against tobacco use (Pérez et al., 1991; see also Pérez & Mugny, 1992). The text presented described the smoker as an individual dependent on an artificial need to smoke, manipulated by publicity, and serving the interests of the tobacco industry (cf. Pérez & Mugny, 1990a). Two variables were introduced. The first concerned the majority versus minority support for the message. The second the ingroup (smokers) versus outgroup (nonsmokers) identity of the minority or majority supporters. The principal dependent measure was change in intention to stop smoking, assessed before and after presentation of the antismoking message (on a 7-point scale). The results showed that a majority source presented as belonging to the ingroup was the only one to have any influence on the intention of smokers to give up smoking (cf. Fig. 13.1). Although its impact was weak, it differed significantly both from the outgroup majority and the ingroup minority, which both had negative influences, smokers in these cases strengthening their resolve to continue smoking. Thus, social categorization proves to be one cause of the failure of a majority source when this is perceived as an outgroup. One must recognize here that smokers respond in terms of a conflict of identity: The threat to identity is in this case associated with an intergroup relation that puts them at a disadvantage and any change in intention to stop smoking is inhibited by their purely defensive motivation. Obliged as they are to evaluate their identity in a symbolic competition (Tajfel &

< previous page

page_252

next page >

< previous page

page_253

next page > Page 253

FIG. 13.1 Change in intention to quit smoking. Turner, 1986) with nonsmokers, smokers are limited to a strategy of identity protection (Breakwell, 1986). This result has a particular importance; it suggests that the context of social categorization in which antismoking campaigns operate can in part explain their failure to change the behavior of smokers. Social Influence as a Factor in Identity Conflict Another possible reason for the failure to influence smoking behavior concerns the link that smokers can be led to make between the threat to their identity and the attempt to influence itself, and this can also produce a conflict of identity for smokers. In this case the influence source will be seen as the real cause of the threat to identity. Rather than accepting this external imposition, smokers will resort to protective strategies that allow them to reassert their identity. This occurs particularly when the influence source has high social status (e.g., expert or competent sources, or those with a certain power of decision over the individual). Furthermore, an influence situation can be understood as making salient a particular kind of relationship between two actors, namely a source who is supposed to influence and a target who is supposed to be influenced. In this sense, the source possesses an ascendancy over the target (i.e., a persuasive constraint), in particular by reason of his or her superior status and assumed role in the influence relation. A consequence of this is that the inherent threat to the identity of the target is more likely to be associated with this relation. The target will then

< previous page

page_253

next page >

< previous page

page_254

next page > Page 254

be more reluctant to treat the content of the source's message in an "objective" manner or to draw any implications for his or her own attitudes. When the influence relationship itself is made salient, treatment of information will be "biased" by the target's desire to defend his or her own beliefs and values (cf. Chaiken, Giner-Sorolla, & Chen, 1996; McGuire & Papageorgis, 1961, 1962). Most antismoking campaigns have been inspired by the principle of conformity (Leventhal & Cleary, 1980), in using high status sources (i.e., health or educational authorities) or expert sources (i.e., experts in the domain of health, cf. Karsenty & Hirsch, 1992) who are assumed to increase the informational dependence of smokers (cf. Deutsch & Gerard, 1955). The logic is that if sources are more credible they can as a result enhance smokers' recognition of the negative consequences of smoking, leading them to give up the habit. If this seems at first glance to represent an advantage in convincing smokers to stop smoking, it can nevertheless be transformed into a handicap by virtue of the way it draws attention to the threat to identity. This will give rise to an identity conflict for smokers motivating them to protect their identity as smokers. Two characteristic aspects of the influence of high status sources suggest the hypothesis that high status sources will give rise to an identity conflict among smokers. The first is that identity conflict is related to the pressure introduced by an influence relation with a high status source. This stems from the fact that smokers perceive their own response, that is, their feeling that their identity is threatened, to be a reaction to the message advanced by the source. Moreover, this will be the case specifically for high status sources. We would argue that when the source has low status there will be no threat to identity associated with the influence relation, conflict being here perceived as independent of or dissociated from the source's message (cf. Pérez & Mugny, 1990b). This point can be illustrated with the results of a study (Falomir, Mugny, & Pérez, 1996) in which smokers, after having been exposed to the same antismoking case as in the study described previously (Pérez et al., 1991), had to indicate the first five ideas that came into their minds with respect to the text they had read. For each idea, we asked them to assess, in percentages, the degree to which they had thought of this before (an idea predating the text), to what degree it had been directly suggested by the text (idea derived from the text) and to what degree the idea had come to them at that moment (self-generated idea). The three percentages had to total to 100%. The results revealed an interaction between the status of the source and the origin of the ideas (cf. Fig. 13.2). First, degree of novelty (self-generated ideas) did not differ as a function of the status of the source; this was also the smallest category of response. Depending on the source, however, the prom-

< previous page

page_254

next page >

< previous page

page_255

next page > Page 255

inence of the message in the conflict varied. Ideas were regarded as originating mainly from the text when the latter had a high status source than when it emanated from a low status source. In contrast, ideas were perceived to have their origin prior to reading the text when the source of this text had low status. These results suggest that smokers' response to the conflict provoked by the challenge to their identity contained in the source's message is more closely associated with the influence relationship when the source has high status, while it is dissociated from the influence relation when the source has low status. The same normative position contained in a message therefore is more compelling when this position is attributed to a high status source than when it is attributed to a low status source. The strength and salience of the social comparison involved with a high status source (cf. Moscovici, 1980; see also Guillon & Personnaz, 1983) therefore is decisive in prompting smokers to respond to the attack on their identity by relating this to a conspicuous element in the influence relationship, namely the source itself. Another issue is that high status sources give particular legitimacy to the point of view they advocate, and this further enhances smokers' perception that their identity is under threat. Consider, for example, a study (Falomir et al., 1998; Experiment 2) in which smokers were first exposed to an antismoking argument attributed either to a high or to a low status source, and then had to indicate the importance they accorded (on a 7-point scale) to four factors commonly associated with the consumption of tobacco, two favorable to smokers (the pleasure of smoking and the reduction of stress) and two unfavorable (health and nicotine dependence) and the importance of granting respect to nonsmokers and to smokers.

FIG. 13.2 Origins of ideas generated.

< previous page

page_255

next page >

< previous page

page_256

next page > Page 256

In relation to the factors associated with the consumption of tobacco (cf. Fig. 13.3), we can see that the favorable dimensions are judged less important than the unfavorable dimensions, which indicates "acceptance" of the stigmatization of smokers by smokers themselves. This is particularly true when the source has high status, for the dimensions favorable to smokers are considered more important when the source has low status. Conversely the unfavorable dimensions are judged more important when confronted with a high status source than when faced with a low status source. Along similar lines, of the two categories, nonsmokers are rated as meriting more respect than smokers, but this effect again is modified by the status of the source (cf. Fig. 13.3). When the source has high status, respect for nonsmokers is rated higher than respect for smokers, whereas this difference is nonsignificant when the source has low status. The high status source therefore appears to force more recognition of the illegitimacy of the act of smoking and to produce a stronger differentiation in the treatment regarded as appropriate for the category of smokers versus nonsmokers. In brief, it makes salient the threat to the identity of smokers for smokers themselves. Social Influence, Identiity Conflict and Defensive Motivation The persuasive constraint associated with the threat to identity introduced by high status sources could be argued to activate strategies for protecting identity and to obstruct the process of change. One way of testing this idea involves manipulating, on the one hand, the salience of the persuasive con-

FIG. 13.3 Importance of factors associated with tobacco consumption and of respect for smokers and nonsmokers (7 = very important).

< previous page

page_256

next page >

< previous page

page_257

next page > Page 257

straint and, on the other, the availability to the smoker of an identity-protection strategy. The hypothesis is that when persuasive constraint is salient, such availability will reduce the influence a high status source has on the intention to stop smoking. When the persuasive constraint is not salient, the presence of such a strategy should not inhibit influence. This idea was tested in another study (Falomir, 1996), in which smokers were exposed to the usual antismoking argument, attributed this time only to a high status source (a group of experts). First, the experiment involved varying the style (imperative vs. optional; cf. Pérez & Mugny, 1991) in which the message challenged the smokers. For example, in the strong persuasive constraint condition, the source declared: "As university professors, our conclusion is that it is absolutely essential that every smoker has these ideas put in his head, and that he accepts them without discussion. There can be no question here of leaving to each smoker the freedom to draw conclusions as he chooses." In the weak persuasive constraint condition, the conclusion is formulated in terms of desires, each smoker explicitly being free to draw his or her own conclusions. The manipulation is thus very similar to those used in the psychological reactance paradigm (cf. Brehm & Brehm, 1981). Second, the possibility of defending identity was manipulated by the presence versus the absence of social support for the participants (cf. Breakwell, 1986), on the assumption that one is more likely to make use of social support and be resistant to change when confronted with a strong threat to identity. In the social support condition, participants were informed that the majority (90%) of smokers were not in favor of antismoking measures. In the lack of social support condition, they were informed that the majority of smokers were in favor of such measures. Recourse to social support here is regarded as an identity protection strategy to the degree that it serves as a heuristic for the social validity of one's position (cf. Chaiken, 1980; Festinger, 1950b; Turner, 1991), and this has elsewhere been shown to allow the individual to resist change (cf. Allen, 1975; Asch, 1951; Doms, 1987; Milgram, 1974). Furthermore, the availability of social support, in the form of an exaggerated consensus, proves to be functional in the search for a positive identity (Agostinelli, Sherman, Presson, & Chassin, 1992; Goethals, 1987; Ross, Green, & House, 1977; Sherman, Presson, & Chassin, 1984). The principal dependent measure here again is represented by change in intention to give up smoking, assessed (on 7point scales) before and after presentation of the antismoking message. Two further measures were included to provide a better assessment of the processes activated. One of these was a measure (on a 7-point scale) of change in agreement with the antismoking arguments contained in the message. If the dynamics predicted are observed only for the measure of intention to stop smoking, one would

< previous page

page_257

next page >

< previous page

page_258

next page > Page 258

be in a position to claim that identity conflict operates beyond the mere acceptance (or rejection) of the source's position; that is, it operates at the level of the personal decision a smoker takes to stop smoking. The other measure was of the identification of smokers with their own group of fellow smokers. The expectation here was that smokers who increase their resolve to stop smoking should display a reduced identification with smokers as a category. Analysis of variance, with the two measures of changein agreement with the antismoking message and in intention to stop smokingas repeated measures and the two experimental variables as between-subject factors, revealed an interaction between the three variables. The mean change scores are presented in Table 13.1. In relation to change in intention to stop smoking, the results confirm the predictions. When persuasive constraint is strong, the presence of social support limits change in intention to stop smoking while its absence facilitates this change. Moreover, the availability of social support reduces change only when the persuasive constraint is strong. These results confirm that social support tends to reduce the influence of a high-status source but that this effect only occurs when the smoker is exposed to a strong persuasive constraint that would otherwise cause the smoker to take the threat to identity into consideration. In relation to changes in agreement with the arguments contained in the message, no significant effects were found for either of the experimental manipulations. Nonetheless, a comparison between the two measures of change shows that when the persuasive constraint is strong and smokers do not have the support of their group, the positive change in intention to stop smoking differs significantly from the negative change in agreement with the message. TABLE 13.1 Change in Agreement with Antismoking Arguments and in Intention to Stop Smoking Persuasive Constraint Strong Weak Social Support Absent Present Absent Present Agreement with antismoking arguments: -O.80b 0.09ab 0.02ab 0.08ab Intention to stop smoking: 0.81a -0.36b -0.08ab 0.80a aPositive values indicate increase in agreement; increase in intention to stop smoking. bmeans with distinct subscripts differ at p < .05.

< previous page

page_258

next page >

< previous page

page_259

next page > Page 259

The fact that the predicted dynamics are not observed in the change in agreement with the message but in the change in intention to quit smoking suggests that the conflict of identity is not a mere response to an influence attempt but that it implies a particular elaboration of the identity threat. What is the beneficial effect of social support when the smoker is not in a constraining influence relation? The results of this study indicate that social support aids the decision to stop smoking when the persuasive constraint is weak. Because this supportin the form of an exaggeration of the size of the ingroupreduces the ingroup favoritism classically found in intergroup relations (Blanz, Mummendey, & Otten, 1995), one might suppose that the support of the group therefore suffices to protect the identity of smokers in the influence relation, and this then allows them to process the position advocated by the source. Complementary results do indeed show that, in this condition, smokers attach more weight to the source's antismoking message. Identification with smokers as a group represents an important factor in tobacco consumption. For example, identification with the stereotype of the smoker explains the subsequent initiation of young people into the smoking habit (Barton, Chassin, Presson, & Sherman, 1982; Chassin et al., 1981) and also explains the practice of smoking itself (Echebarria et al., 1994). Finally, therefore, correlational analyses were executed to see if identification with smokers as a group explains any of the variance in resistance to change. The results show that group identification is associated with the intention to maintain tobacco consumption when smokers do not have the support of their group and are confronted by a strong persuasive constraint to accept propositions that threaten their identity as smokers. In other words, change in intention is of the same order as reduced identification: Conspicuous absence of support increases smokers' intention to stop smoking, the persuasive constraint forcing them to recognize the negative implications for identity that the lack of a protective strategy prevents them from defending and consequently maintaining. Coping with Identity Conflict in Social Influence We have, therefore, a particular kind of influence situation in which the identity of the target (the smoker) is socially threatened and in which elements of the influence contextsocial categorization or the status of the influence sourceensure that this conflict with the smoker's identity will be recognized. One way of avoiding the activation of a conflict of identity is by allowing smokers to respond in terms of a threat to their identity but without leading them to relate this threat to features of the influence context. This, for example, would be the case if, when faced with a high status influence source, a smoker is able to assert autonomy and independence. A

< previous page

page_259

next page >

< previous page

page_260

next page > Page 260

self-affirmation process (Steele, 1988) would reduce the persuasive constraint associated with an unequal influence relationship of this kind. Let us consider some observations on this point. In a study by Pérez, Falomir, and Mugny (1995), smokers were exposed to the usual antismoking message, attributed either to a source with superior social status (a group of university professors) or to one of inferior status (a minority group). During the experimental session, they either did or did not have the opportunity to smoke. The results showed that when they were faced with a superior status source, the smokers increased their intention to give up smoking if they could smoke. This effect was specific to the high status source condition; when the source had low status, the act of smoking had a negative effect on change. In the context of this chapter, our interpretation is that smoking represents a positive expression of the smoker's identity in the face of a high status source. The modification of this relationship with the source of influence, manipulated in the experiment by allowing the smoker to redress the balance of this otherwise unequal relationship, produced the increased influence on the intention to quit smoking. Another study supports this interpretation (Falomir et al., 1996), and also helps to resolve an undecided question: Is the opportunity to affirm identity provided by participants' being able to smoke or the act of smoking itself? The option to smoke or its absence, as it was operationalized in the Pérez et al. (1995) study, allows another explanation. In this study the possibility of smoking in effect confounded permission to smoke with the act of smoking itself. The possibility of smoking could play a dual role. On the one hand, it might diminish the salience of constraint by rendering temporarily inoperative a social norm regulating behavior, in this case the normal prohibition of smoking associated with public places. On the other hand, it allows the performance of an act that is against the persuasive intention of the source, so that by this act the target regains a certain autonomy in the face of the source. Also in the preceding study, participants in the condition that allowed smoking were in fact encouraged to smoke, and all had smoked. The procedure in this experiment was different. In the condition allowing smoking, the experimenter neither encouraged smokers to light up nor did he discourage them if they were inclined to do so, and this allowed a distinction to be made between those participants who did and those who did not smoke. The hypothesis tested was that it is the act of smoking and not simply the opportunity to smoke that should allow participants some control in the relationship with a high status source, because only in smoking do the participants preserve their identity (if the participant does not smoke, he or she submits to the source). For a low status source, the participants do not interpret the conflict of viewpoints in terms of the influence relation, and consequently, the act of smoking should in this condition be unimportant in the

< previous page

page_260

next page >

< previous page

page_261

next page > Page 261

expression of identity. In this case, smoking becomes merely an expression of disdain for a point of view that a priori has no particular expertise or credibility, and it should therefore inhibit influence. In this study, participants first read an antismoking text (identical to that used in the previous study), attributed in half the conditions to an expert source (university professors) and in the other half to a nonexpert source (high school students). They were next asked to memorize a list of 16 adjectives that could be used to describe the source (half positive and half negative). As in the previous study, they then had to wait in silence for 4 minutes during which they either did or did not have the opportunity to smoke. In the nonsmoking condition, there were no ashtrays in the room and in addition notices were posted indicating the prohibition. When asked if they could smoke, the experimenter indicated he had no power in the matter and emphasized that the prohibition was due to the lack of ventilation in the room. In the condition in which smoking was possible, ashtrays were available so that those participants who wished to smoke could do so, and the notices forbidding smoking were removed. The experimenter recorded whether or not participants did actually smoke. Analyses of variance were executed with status of the source as a variable and also in terms of whether participants were not allowed to smoke, were able to do so but did not, or were able to smoke and did do so. The two measures considered here concern change in agreement with the antismoking claims contained in the text, and change in personal intention to give up smoking. In relation to the first of these measures, change in agreement with extracts drawn from the message, the only significant effect was for the status of the source, the expert source having more influence than the nonexpert source. For changes in intention, depicted in Fig. 13.4, there was an interaction between the two variables that confirms that, when participants could not smoke, the nonexpert source produced more change than the expert source. Decomposition of effects for the superior status source shows primarily that participants who did smoke tended to change more than the nonsmoking participants. When the source had low status, participants who could smoke and did so changed less than those who could not smoke and less than those who could smoke but did not. Finally, among the participants who had the option of smoking, those who took advantage of this option changed, as predicted, more when the source had high status than when the source had low status, whereas those who chose not to smoke under these conditions did not differ as a function of source status. These results clearly reflect a distinct way of working out the conflict between divergent points of view. On the one hand, an expert source produces an increase in agreement with its claims, in contrast to a nonexpert source, but not necessarily in intention to give up smoking. Indeed, only

< previous page

page_261

next page >

< previous page

page_262

next page > Page 262

FIG. 13.4 Change in intention to quit smoking. self-affirmation contributed to a conflict that was worked out at an identity level leading also to a change in intention. Faced with a low status source, the target would locate the debate primarily outside (upstream, as it were, from) the confrontation with this source (see Fig. 13.2). This reaction facilitates an internalization of the conflict created by the message (Pérez et al., 1995) that is dissociated from the relationship with the source (Pérez & Mugny, 1990b). The change in intention found when smokers are unable to smoke reflects the consequences at a latent level of treatment of the threat to identity; a conversion effect occurs (cf. Moscovici, 1980). In contrast, when the participant is faced with an expert source, the conflict would be interpreted more in terms of the relationship with the source (cf. Fig. 13.2). When the influence context excludes a positive response to this influence relation, which in this case is when participants do not smoke, the persuasive constraint associated with the source is found to lead to a defensive attitude characteristic of an identity conflict. The results also support the corollary of this dynamic: Individuals placed in a situation involving identity threat need some positive way of adjusting to the influence relationship if they are to be able to work out their own personal attitude (i.e., the intention to give up smoking). In effect, when the source has high status and when the participant takes advantage of the freedom to smoke, overt conformity is accompanied by a change in personal intention, a sign of a genuine internalization (Kelman, 1958). Finally, the simple fact of allowing smoking appears to be insufficient to induce these effects. Only the act of smoking is in itself a regulative element.

< previous page

page_262

next page >

< previous page

page_263

next page > Page 263

The action of smoking would contribute to self-affirmation, allowing the smoker to reassert freedom and autonomy in the context of a salient and constraining influence relationship. Conclusion We have proposed that the social context in which antismoking campaigns operate includes factors that counteract the translation of attitudes and beliefs about tobacco use into behavior such as quitting smoking. On this basis we have argued that the prevailing social norms, values, and practices against tobacco consumption introduce a threat to the identity of smokers that can explain a part of this difficulty. In effect, if this threatening social context is intended to encourage smokers to seek out a more valued identity, then a defensive interpretation of this threat in terms of an identity conflict prevents internalization of the antismoking perspective at the level of action; that is, it prevents the smoker deciding for himself or herself to stop smoking. The ineffectiveness of antismoking campaigns in changing the behavior of smokers can partly be attributed to this kind of dynamics. On the one hand, we have seen that the social context is characterized by a categorization of people into smokers and nonsmokers, and this is moreover one of those rare categorizations that is not subject to the taboo against discrimination. This leads smokers to preserve a social identity distinct from that of the category of nonsmokers. The emergence of a conflict of identity is less likely here to the extent that the identity of smokers is guaranteed in the relation with nonsmokers, or else if such a comparison is not salient (e.g., if the influence source is an ingroup). On the other hand, the kind of high status sources that are most frequently associated with antismoking campaigns (e.g., experts) encourage an interpretation of the threat as an identity conflict and this motivates smokers to defend their identity within the influence relationship. A conflict of identity is less likely to arise if smokers can preserve their autonomy and independence in the influence relationship or if the influence source does not impose any persuasive constraint (e.g., because it has low status or is a minority source). In pursuing measures to overcome the marginalization of various ethnic minorities (e.g., Gypsies in Europe, Native Americans in America), or social minorities (e.g., unemployed, smokers, or the obese), society develops influence strategies intended to convert them to dominant norms, values, and practices. Influence on these stigmatized groups is limited by the very status ascribed to them, a status that is intrinsically threatening. Individuals belonging to these groups may accept and meet such a threat by trying to acquire a new majority identity (Tajfel & Turner, 1986). But such a process presupposes that individuals acknowledge the implications of their current

< previous page

page_263

next page >

< previous page

page_264

next page > Page 264

stigmatized social status, and this is not always the case. Even if these individuals are aware of or accept such a threat in some way, they still may be motivated to defend it in the social relationship rather than to change it. If these stigmatized groups respond in terms of an identity conflict, any influence on them would be more likely to fail. By way of conclusion, we note that Moscovici (1985) distinguished three ways of confronting the conflict introduced by a difference of opinion: (a) a rational demonstration of the validity or invalidity of a point of view, (b) the use of power to impose a solution, and (c) reciprocal influence to negotiate the conflict. Up to now, antismoking campaigns seem to have drawn primarily on the first two of these, which is to say they try to convince smokers that smoking is bad, and they introduce legislation that will lead smokers to stop smoking. However, the validity of a point of view is not inherent solely in the facts on which it draws, and the relation an individual maintains with other social actors affects the validity he or she accords to opposing points of view as well as the way he or she treats them. In effect, even though the credibility or power of a source of influence can in themselves reinforce the validity of their claims, this is not always the case. To put it in other terms, the norms regulating social relationships (e.g., the reciprocity rule; cf. Cialdini, 1993) also determine the processes of change. The results reported in this chapter are then more in accord with the notion that expert sources or those possessing power may need to negotiate the conflict they create.

< previous page

page_264

next page >

< previous page

page_265

next page > Page 265

14 The Perceived Impact of Persuasive Messages on ''Us'' and "Them" Julie M. Duck Michael A. Hogg Deborah J. Terry University of Queensland The mass media potentially play a significant role in shaping attitudes and behavior, not only through coverage of persuasive advertising campaigns that are specifically designed to influence the audience but also through the portrayalin entertainment, news, and current affairs programsof people, their opinions, and behavior. Moreover, apart from the direct, persuasive effects of media content on attitudes and behavior, people's beliefs or assumptions about media influences on others also may have a substantial indirect effect. That is, attitudinal and behavioral change may arise from the perception that the opinions of others have been influenced by the media. This chapter reviews research that deals not with the direct effects of persuasive messages on people's attitudes and behavior but with the indirect effects that result from people's beliefs about the impact of persuasive messages on others. Specifically, we describe research on the third-person effect (Davison, 1983) in which people expect the media, and persuasive communications in general, to have a greater effect on others than on themselves. Against a dominant tendency to treat persuasive communication and social influence within groups as separate areas of inquiry, recent research has sought to re-emphasize the role of social groups in the persuasion process (see Mackie, Gastardo-Conaco, & Skelly, 1992; Mackie, Worth, & Asuncion, 1990; McGarty, Haslam, Hutchinson, & Turner, 1994; van Knippenberg &

< previous page

page_265

next page >

< previous page

page_266

next page > Page 266

Wilke, 1992; Wilder, 1990). For instance, Turner (1991) stressed the social-normative aspect of information processing, arguing that, when social identity is salient, the validity of persuasive information psychologically is established by ingroup norms. Our account of perceived media effects is argued from a similar, social-normative perspective. We assume that perceptions of persuasive impact on self and other are also dependent on salient group memberships or social identities (e.g., gender identity, political identity, student identity)that is, that perceived self-other differences in media effects are affected by the social psychological relationships between self and others in varying social contexts and reflect ingroup norms regarding the acknowledgment of influence. Specifically, we aim to emphasize an "us-them" or group-based distinction that was implicit in Davison's (1983) original formulation of the third-person effect but has since been underplayed. The Third-Person Effect First identified by Davison (1983), the third-person effect describes a systematic perceptual bias by which people assume that media messages and persuasive communications in general have a greater impact on others' beliefs, attitudes, and behavior than on their own. Much empirical evidence supports this perceptual bias (see Perloff, 1993, for a review). For instance, perceived self-other discrepancies have been identified in people's perceptions of media reports about apartheid (Mutz, 1989), news broadcasts of Middle East conflict (Perloff, 1989; Vallone, Ross, & Lepper, 1985), political advertising (Rucinski & Salmon, 1990), product advertising (Gunther & Mundy, 1993; Gunther & Thorson, 1992), dramatic television series with political overtones (Lasorsa, 1989), media pornography (Gunther, 1995), and defamatory newspaper articles (Cohen, Mutz, Price, & Gunther, 1988; Gunther, 1991). Although research on the behavioral component of the third-person effect is more limited, evidence suggests that people may act or behave on the basis of these perceptions. For instance, in a desire to protect vulnerable others, people may call for censorship of media pornography although they have no qualms about viewing such content themselves (Gunther, 1995). Third-person perceptions also may reduce people's willingness to voice their own opinions on controversial issues. People may perceive that the media are biased against their points of view (i.e., hostile media bias; Giner-Sorolla & Chaiken, 1994; Perloff, 1989; Vallone et al., 1985) and, assuming that others will be influenced by "biased" coverage, may reason that public opinion is against them (i.e., the spiral of silence; Mutz, 1989; NoelleNeumann, 1974, 1984).

< previous page

page_266

next page >

< previous page

page_267

next page > Page 267

The Basis of Third-Person Perceptions Third-person perceptions usually are attributed to a tendency for individuals to overestimate persuasive influence on others, although people also may underestimate persuasive influence on themselves (see Cohen et al., 1988; Perloff, 1993). Both motivational and cognitive factors may play a part. People may be less inclined to make external attributions to media influences on their own behavior than on others' behavior to retain a sense of personal control and to enhance self-esteem. In addition, people may possess an acquired script about the gullibility of the mass audience (Smith, 1986) that invokes a stereotyped image of the kind of person who seems particularly likely to be influenced by media content. In comparison to this image, people may judge themselves as impervious to influence (cf. Weinstein, 1980). Consistent with these theoretical accounts, research suggests that certain conditions such as negative media content, personal involvement in the issue, and perceived source bias facilitate third-person effects. For instance, research has indicated that third-person effects are more pronounced with respect to negative media content (e.g., violence, pornography, defamatory news articles, and product commercials) than with respect to positive media content (e.g., public service advertisements for health and safety issues; see Duck & Mullin, 1995; Gunther & Mundy, 1993; Gunther & Thorson, 1992; Innes & Zeitz, 1988). Indeed, research suggests that the third-person effect can be eliminated, or even reversed, when the intended influence is perceived as desirable, intelligent, or beneficial. For instance, in a study that examined the perceived effects of commercial TV advertisements, we found that respondents perceived themselves as less influenced than others would be by the least-liked product ads but as more influenced than others would be by the most-liked product ads (Duck, Hogg, & Terry, 1998a). Similarly, in research on the perceived impact of Acquired Immunodeficiency Syndrome (AID)/human immunodeficiency virus (HIV) advertising (Duck, Terry, & Hogg, 1995), we found that student respondents perceived themselves as less vulnerable than others to low-quality AIDS advertisements, but as more influenced than others by high-quality AIDS advertisements. These findings suggest that third-person perceptions reflect the motivation to deny personal persuasibility in contexts where media influence is judged to produce negative rather than positive outcomes. Presumably, widespread support for third-person perceptions in the empirical literature reflects the predominance of negative, rather than positive, media content. Who Are the Third Persons? One important factor that also may influence the magnitude of third-person effects is the nature of the comparison other. Third-person

< previous page

page_267

next page >

< previous page

page_268

next page > Page 268

perceptions typically have been demonstrated by comparing people's judgments of media impact on self with their judgments of media impact on a single and generalized comparison other (e.g., "voters in general," Davison, 1983; "other viewers," Lasorsa, 1989; "most Americans," Rucinski & Salmon, 1990). Yet, evidence suggests that the magnitude of the perceived self-other discrepancy may depend, at least in part, on who the comparison others are. For instance, perceived effects on others are magnified as the definition of others becomes progressively more broad or socially distant (e.g., other Stanford students, other Californians, the public at large, see Cohen et al., 1988, Gibbon & Durkin, 1995; Gunther, 1991). In two studies, Duck and Mullin (1995) varied the nature of the comparison other along two independent dimensionsvagueness and closeness. Our results indicated a marked third-person effect when the comparison other was both vague and distant (i.e., the average person) but no significant self-other difference when the comparison other was both specific and close (i.e., the respondent's closest friend). We reasoned that motivated selfserving biases may extend to include close friends and relatives as an extension of self (e.g., Burger, 1981; Schlenker & Miller, 1971). Alternatively, vague comparison others may facilitate downward comparisons with vulnerable stereotypes, whereas identifying specific others may inhibit such a process (Perloff & Fetzer, 1986). Group processes, as opposed to interpersonal processes (e.g., acquaintanceship), also may underlie perceived self-other differences. Taken literally, the third-person effect draws a distinction between perceptions of media influence on first and second persons ("me" and "you") and third persons ("they"), that is between perceived persuasive impact on ''us'' and "them." Indeed, Davison (1983) argued that people believe communications exert their greatest impact "not on 'me' or on 'you', but on 'them'the third-persons" (p. 3) and that "the concept of reference groups may prove useful in explaining the third-person effect" (p. 11). He reasoned that people may overestimate media impact on some sections of the audience but not on others. For instance, there might be little exaggeration of the perceived impact of a communication on members of a group that constitutes a normative reference group for the perceiver. The possibility that the third-person effect reflects the nature of the social psychological relationship between self and comparison other within varying social contexts has not been explored adequately and is the focus of the research described in this chapter. In particular, we draw on current theories of group membershipsocial identity theory (SIT) and selfcategorization theory (Hogg & Abrams, 1988; Tajfel & Turner, 1979; Turner, 1982, 1985; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987)to provide a theoretical framework in which to examine the influence of group membership on the third-person effect.

< previous page

page_268

next page >

< previous page

page_269

next page > Page 269

A Social Identity Perspective Both SIT and self-categorization theory suggest that an important part of the self-concept is derived from membership in social groups. Although there are situations in which people view themselves as individuals or unique persons, different from all others, there are many social contexts in which people view themselves predominantly as group members with a common or shared social identity. In these contexts a process of self-categorization "depersonalizes" perception of self and others in terms of relevant group prototypes (norms or stereotypes). People view themselves and others as group members rather than individuals"I" becomes "we,'' and ''me" versus "you" becomes "us" versus "them." When social identity is salient, people accentuate prototypical similarities between self and ingroup members and prototypical differences between ingroup and outgroup members. Further, because social identities have important selfevaluative consequences (Hogg & Abrams, 1990; Turner, 1982), in these situations group members tend to make intergroup comparisons that favor the ingroup (e.g., Brewer, 1979; Brewer & Kramer, 1985; Tajfel & Turner, 1979). The process of depersonalization is responsive to contextual factors that influence the degree to which people identify themselves as group members rather than individuals (see Hogg, 1992). The stronger the social identification, the more pronounced the tendencies toward perceptions of similarity or equivalence between self and ingroup members (ingroup assimilation) and toward perceptions of difference between ingroup and outgroup members (intergroup contrast). From a self-categorization perspective, we proposed that third-person perceptions would be sensitive to the categorization of self and other into relevant ingroup and outgroup categories and would reflect the subsequent accentuation of similarities within, and differences between, those categories. Take the typical case in which media influence is perceived to have a negative impact. To the extent that comparison others are judged as outgroup members ("them"), they are contrasted to the perceiver's identity, evaluated negatively, and represented unfavorablyas relatively vulnerable to media influence. In contrast, to the extent that comparison others are judged as ingroup members, they would be assimilated to the perceiver's identity ("us"), evaluated positively, and represented favorably, like the selfas relatively invulnerable to influence. Moreover, the tendency to categorize self and others as ingroup and outgroup members depends on features of the social comparative context, and the extent of assimilation and contrast displayed vary according to the perceiver's level of identification with the salient social category.

< previous page

page_269

next page >

< previous page

page_270

next page > Page 270

Evidence for a Social-Normative Perspective on Third-Person Perceptions A selection of research that provides evidence for the social-normative basis of third-person perceptions is reviewed below. This research includes studies examining political identity and perceptions of media campaign impact, gender identity and perceptions of the impact of commercial advertisements, and student identity and perceptions of the impact of AIDS/HIV advertisements. A study that examines the perceived effects on members of an attitudinal ingroup and attitudinal outgroup of a newspaper editorial on the issue of life imprisonment and that addresses the behavioral as well as the perceptual component of the third-person effect is also reviewed. Political Identification and Perceptions of Media Campaign Impact Support for predictions derived from SIT and self-categorization theory was found in two naturalistic studies conducted during the course of campaigns for the Australian federal elections in 1993 and in 1996. In these studies, we examined political identification and the perceived effects on self and other of political media campaigns, first in a one-shot survey 3 days prior to the 1993 election (Duck, Hogg, & Terry, 1995) and second, in a two-wave study conducted before and after the 1996 election (Duck, Terry, & Hogg, 1998b). Perceived Media Impact in the Lead Up to a Federal Election. The 1993 Australian federal election, contested against the background of fierce media advertising, provided an ideal naturalistic context for an initial study for two reasons. First, a precedent existed for examination of the third-person effect in the context of election campaigningin a study conducted during the 1988 U.S. presidential election campaign, Rucinski and Salmon (1990) found that people perceived others as the vulnerable voters, more influenced than self by media campaign content. Second, with political identity salient, the Australian federal election provided a natural opportunity to extend previous theoretical and empirical work through a focus on the social-normative basis of perceived self-other differences. Specifically, this study allowed examination of the effects of political group membership on perceptions of media impact not only on self and on voters in general but also on political ingroup and political outgroup members. In Australia, voting is compulsory for Australian citizens 18 years or over. The Australian political sphere is dominated by two major partiesthe Australian Labor Party and the conservative Liberal-National Coalition.

< previous page

page_270

next page >

< previous page

page_271

next page > Page 271

Other political parties (e.g., the Australian Democrats) have a lower profile, with the bulk of media campaigning focusing on the relative merits of the two major parties and their policies. At the time of the 1993 election, the Labor Party was in government, led by Paul Keating. The Liberal-National Coalition was in opposition. This study sought to examine the perceptions of voters who said they identified with one of the two major parties and thus had a clearly defined political ingroup and political outgroup. Thus, 3 days prior to the election, Australian university students who identified with one of the two major political parties were surveyed regarding their perceptions of media campaign impact on self and on others (political ingroup members, political outgroup members, and voters in general). First, the study explored the effects of group membership on respondents' perceptions of the impact of general campaign contentnews reports, polls, political advertisements promoting party policies, and negative political advertisements attacking opposing candidates and their policies. In accordance with the notion of a basic self-other distinction in perceptions of persuasive impact, respondents were predicted to judge political communications to have a greater influence on the voting decisions of other people than on themselves. Further, in line with a social-normative perspective on third-person perceptions, political ingroup members were predicted to be perceived, like the self, as less influenced than political outgroup membersparticularly by those who identified strongly with their political ingroup. Compared with respondents who were low in political identification, those high in identification were expected to perceive less media influence on self and ingroup members, less difference between the level of media influence on self and on ingroup members, and more difference between the level of media influence on ingroup and outgroup members. These predictions follow from the assumption that the salience of social categorization increases perceived similarities between self and ingroup members ("us") and perceived differences between ingroup and outgroup members ("them") on dimensions relevant to the categorization and that self-categories tend to be evaluated positively (i.e., as relatively uninfluenced). Further it was assumed that these perceptions might reflect, in part, a belief that media coverage of the election campaign was biased in favor of the ''other" side (the hostile media phenomenon) but that these third-person perceptions would still exist after controls for the effects of perceived media bias and thus reflect an independent selfcategorization dynamic. As expected, respondents perceived others as more influenced by the election campaign than they were, indicating a basic perceived self-other difference in persuasive impact (see Table 14.1). Moreover, consistent with predictions derived from theories of group membership, political ingroup

< previous page

page_271

next page >

< previous page

page_272

next page > Page 272

members also were perceived as less influenced by campaign content than political outgroup members. That is, thirdperson perceptions were more pronounced with respect to the political outgroup than to the political ingroup. Respondents who identified strongly with their preferred party (high identifiers) perceived self and political ingroup members as less influenced by campaign content than did other respondents, and showed more evidence of positive intergroup differentiation. At the same time, however, high identifiers exaggerated self-ingroup differences, challenging the theoretical assumption that intergroup differentiation is necessarily associated with ingroup assimilation. Additional regression analyses indicated that strength of political identification made a unique contribution to prediction of thirdperson perceptions over and above that made by other measures of political involvement (e.g., general political interest, specific interest in the election campaign). As expected, respondents perceived the media as biased against their preferred party. However, follow-up analyses also indicated that perceptions of media impact on ingroup and outgroup members were unrelated to the level of perceived media bias and that differences in perceived media impact according to level of political identification were not eliminated by controls for this factor. Second, the study explored the effects of group membership on perceptions of the impact of media campaign content that explicitly favored the political ingroup or the political outgroup (e.g., political advertisements promoting the Liberal-National Coalition)a context in which it was assumed that group-based perceptions of media influence would be most pronounced. Based on self-categorization theory, and in line with findings on the hostile media bias that suggest that people perceive a disproportionate effect on other people of material favoring the "wrong" side, respondents were predicted to judge self and ingroup members as less influenced than TABLE 14.1 Mean Perceived Influence on Self, Ingroup Members, Outgroup Members and Voters in General as a Function of Strength of Political Identification Target Political Identification Self Ingroup Outgroup Voters in General Low 4.46 5.66 5.95 6.19 High 2.69 4.82 5.44 5.73 Overall 5.96 3.56 5.23 5.69 Note. Scale range is 1 (not influenced at all) to 9 (a large influence).

< previous page

page_272

next page >

< previous page

page_273

next page > Page 273

outgroup members and voters in general were by material favoring the political outgroup. However, in accordance with the notion of self-serving tendencies, self and ingroup members were predicted to be perceived as more influenced than outgroup members by material promoting the political ingroup. Consistent with predictions, results indicated that perceived similarities between self and ingroup members and perceived differences between ingroup and outgroup members were more pronounced when respondents considered the effects of campaign material that explicitly favored either their political ingroup or the political outgroup. Moreover, perceived us-them distinctions in impact of persuasion were most pronounced when respondents considered the impact of persuasive material that favored the political outgroup (see Fig. 14.1). Respondents perceived that voters in general were swayed in line with the intent of persuasive political messages, irrespective of which party was promoted; outgroup members were perceived to be influenced by material favoring their own party but were not perceived to be influenced by material favoring the opposing party; and, although self

FIG. 14.1 Perceived effect of pro-ingroup and pro-outgroup campaign content on support for the promoted party (for self, ingroup, outgroup, and voters in general).

< previous page

page_273

next page >

< previous page

page_274

next page > Page 274

and ingroup members also were judged to be influenced by material favoring their own party, they were judged to react against material favoring the opposing partya pattern of responses that was significantly stronger for self than for the political ingroup. In addition, where campaign material supported the respondent's political ingroup, and influence could be construed as socially desirable, there was evidence that the self and ingroup were judged as more influenced than outgroup members, although a self-serving self-ingroup distinction was still maintained. Taken together, the results of this study confirmed the existence of a generalized third-person effect in the context of political media campaigns. In addition, the study extended previous theoretical and empirical work on the third-person effect by demonstrating that political identity affects people's perceptions of media impact on self and others and by offering an explanation for variations in the magnitude of the third-person effect according to the nature of the comparison other against whom judgments of media impact are made. Results suggested that political identifications are powerful anchoring points for people's perceptions of media campaign effects on self and others (cf. Perloff, 1989)for instance, influence on the political outgroup was routinely exaggerated relative to influence on the political ingroup. Yet, the tendency to portray self as uniquely invulnerable to influence dominated the pattern of perceptions obtained, and there was little evidence of a simple, mechanical assimilation of self and political ingroup. We argued that this may not be incompatible with a social identity perspective but may be due, in part, to the fact that political identification involves a broad and overinclusive social category that does not satisfy individual needs for distinctiveness (e.g., optimal distinctiveness theory; Brewer, 1991, 1993a, 1993b). Perceived Media Impact and the Changing Intergroup Context. Of course, from the perspective of SIT, the form and content of group behavior depend on the nature of the intergroup context and are responsive to changes in intergroup relations. For instance, in the course of an election campaign, the responses of politically aligned individuals to media campaigning should reflect their membership in either a dominant or subordinate group pre-election as well as their status as winners or losers postelection. Thus, in a second study conducted during the volatile 1996 Australian federal election, we sought to extend the results from our first study on political identification and perceptions of media campaign impact by demonstrating that ingroup-serving biases in communication perceptionlike other forms of group behaviorreflect the (changing) nature of the intergroup context.

< previous page

page_274

next page >

< previous page

page_275

next page > Page 275

According to SIT, an important part of an individual's sense of self comes from membership in social groups, value connotations associated with the ingroup are the result of social comparison with a relevant outgroup, and individuals strive to achieve positive self-esteem through a belief in the positive distinctiveness of the ingroup. It is further proposed that the particular strategies by which a positive social identity is achieved or maintained depend on social beliefs concerning the relative status of the ingroup, the stability and legitimacy of its status, and the permeability of group boundariesthat is, the form and content of group behavior depend on subjective understandings of the nature of the relations between the groups. This "intergroup" or "macrosocial" aspect of SIT has stimulated research both in the laboratory and the field (see Brown, 1978; Brown, Condor, Mathews, Wade, & Williams, 1986; Giles & Johnson, 1987; Hogg & Hains, 1996; Oaker & Brown, 1986; Sachdev & Bourhis, 1993; Turner & Brown, 1978; Wright, Taylor, & Moghaddam, 1990). In particular, recent experimental research has focused on the strategies that members of low status groups use to enhance their social identity (see van Knippenberg & Ellemers, 1993, for a review) and on the important distinction between intergroup comparisons that occur in a consensually legitimate and stable framework and those that occur when there is no consensus or stability (see Tajfel, 1974). In a stable intergroup structure in which the status position of a group is fixed and cannot be improved, members of low status or subordinate groups are assumed to be motivated to explore individual mobility strategies (i.e., strategies that involve leaving the low status group and joining the high status group). However, when group status differences are unstable or insecure, and an alternative status position for the group as a whole is likely to be realized, subordinate group members are assumed to respond more collectively and work toward group mobility or social change (see Ellemers, van Knippenberg, de Vries, & Wilke, 1988; Ellemers, van Knippenberg, & Wilke, 1990). In these "insecure" contexts, low status group membership may not inevitably result in low ingroup identification. Rather, members of low status groups may emphasize their ingroup identity and display a competitive intergroup attitude, closing ranks to increase the chances of "winning" as a group. At the same time, members of groups that have already achieved positive distinctiveness from comparison groups (i.e., dominant or high status groups) may be actively engaged in trying to preserve the status quo (e.g., by downplaying the achievements of the subordinate group; Tajfel, 1975). These status protection motives are assumed to operate most strongly in contexts in which there is a possibility of demotion to a lower status position (see Ellemers, Doojse, van Knippenberg, & Wilke, 1992). The competitive responses of subordinate group members and the defensive re-

< previous page

page_275

next page >

< previous page

page_276

next page > Page 276

sponses of dominant group members in unstable or insecure intergroup contexts also have been demonstrated in naturalistic research on intergroup mergers (e.g., Haunschild, Moreland, & Murrell, 1994; Skevington, 1981; Terry, Callan, & Sartori, 1996) and on long-term social or cultural change (e.g., Vaughan, 1978), and SIT has been used to explain the "creative" identity-enhancing strategies that members of losing sports teams employ (see Lalone, 1992; Lalonde, Moghaddam, & Taylor, 1987). In this study we sought to apply this theoretical ideas to research on group-based media perceptions during an election campaigna context in which the status or power structure is under challenge and is likely to change. Our interest was in changes to three types of media perceptionsperceptions of hostile media bias, perceptions of media impact on self and other (third-person perceptions), and attributions of media influence on self and other. As in the 1993 study, we examined the perceptions of voters who identified with one of the two major parties and thus had a clearly defined political ingroup and political outgroup. However, in the 1996 study, these perceptions were examined twice3 days prior to the election and 1 week later (4 days after the election result had been announced). Before the election, the Australian Labor Party (ALP) was in government and the Liberal-National Coalition was in opposition. The ALP had been in government for 13 years and, despite the fact that nationwide polls during the 6-week campaign consistently predicted a change of government (e.g., Butler, 1996; Steketee, 1996), the ALP leader, Paul Keating, remained publicly confident of a win up until election day (Taylor, 1996; Tingle, 1996). In fact, the Liberal-National Coalition defeated the ALP in an emphatic 45-seat majority win (Steketee, 1996)one of the biggest since World War II ("Howard's 100 days of Power," 1996). Overnight, the power structure was reversed and John Howard, the leader of the Liberal Party, became prime minister. First, given the salience of political identity in the period surrounding a general election, we predicted that politically aligned respondents from both sides would show ingroup-serving biases in their perceptions of the media campaign and its impact and that these biases would be more pronounced for respondents who identified strongly with their political ingroup. We reasoned that these biases would be reflected in a perception that media coverage of the election was hostile to their side, that self and political ingroup others were less influenced by campaign content than political others and voters in general ("us" versus "them"), and that dispositional explanations in terms of gullibility were more likely to account for any media influence on outgroup others and voters in general ("them") than on ingroup others and self ("us''). According to the notion of group-serving attributions or the ultimate attribution error (Pettigrew, 1979; Taylor & Jaggi,

< previous page

page_276

next page >

< previous page

page_277

next page > Page 277

1974), when individuals make attributions for themselves as group members and others as either ingroup or outgroup members, intergroup attributions may be characterized by ethnocentric or ingroup-serving bias. For instance, socially undesirable behavior by outgroup members (e.g., media influence) is likely to be attributed to internal dispositions (e.g., gullibility). Second, and more important, we predicted that the magnitude of these biases would depend on the changing nature of the intergroup context and thus on both group membership and time. We reasoned that the state of flux in the lead-up to the election would raise subordinate group members' concern with group goals, collective enhancement, and social competition, producing strong ingroup-serving biases, but that this competitive orientation would decline once their party had won. By contrast, we reasoned that members of the dominant-losing group would show a reciprocal pattern in which group-serving perceptions were more pronounced after, rather than before, the election. We assumed that although members of the dominant group would be less socially competitive before the election, the threat to status resulting from their election loss would promote defensive attempts to restore positive social identity through derogation of the media, the outgroup, and the electorate in general. Specifically, we predicted that evaluative biases in communication perception would be more pronounced for members of the subordinate group than for members of the dominant group prior to the election, but more pronounced for members of the dominant-losing group rather than the subordinate-winning group after the election. Implicit in this prediction was the assumption that evaluative biases would decrease over time for members of the subordinate-winning group but increase over time for members of the dominant-losing group. Finally, we predicted that these patterns would be more pronounced for high rather than low identifiers. As expected, we found that politically aligned respondents tended to perceive media coverage of the election campaign as biased against and hostile towards their sideat least when they identified strongly with their preferred party. Moreover, as predicted on the basis of social identity theory, the magnitude of this perception of hostile bias depended on both group membership and time (see Fig. 14.2), although these effects were not moderated by level of identification. In the lead-up to the election, perceptions of hostile media bias were more pronounced among members of the subordinate group and, at this stage, members of the dominant group perceived media coverage to have been quite fair. Following their election win, members of the subordinate-winning group were less inclined to perceive that the media had been biased against their side, whereas members of the dominant-losing group were somewhat more inclined to perceive hostile intent. Indeed, after the election, members of the dominant-losing group perceived

< previous page

page_277

next page >

< previous page

page_278

next page > Page 278

FIG. 14.2 Hostile media bias according to group membership and time. Note. Asterisks indicate means that differ significantly from zero; * p < .05, ** p < .001. that the media had been significantly biased against their sidea clear re-evaluation of their assessment just 1 week earlier. Apparently, the winners, from their newfound position of dominance, had less need to be critical of the media and its potential role in the election outcome, whereas the losers sought to blame their party's loss on biased media coverage. Consistent with previous research (e.g., Duck, Hogg, & Terry, 1995; Rucinski & Salmon, 1990), we found that respondents perceived media campaigning for the 1996 election to have had more impact on others' voting decisions than on their own. Also as expected, ingroup members were perceived as less influenced than outgroup members and voters in general (at least by high identifiers) providing some evidence for the predicted group-serving tendency to perceive that persuasive impact is on "them," not on "us." However, as in our 1993 election study, respondents perceived themselves as least influenced overall and evidence for ingroup-outgroup differentiation was stronger than was evidence for self-ingroup assimilation.

< previous page

page_278

next page >

< previous page

page_279

next page > Page 279

Presumably, self-serving tendencies resulted both in a belief in the positive distinctiveness of the ingroup and in one's own distinctiveness within the ingroup (see Brewer, 1991, 1993a, 1993b; Codol, 1984; Lemaine, 1974; Maslach, 1974; Snyder & Fromkin, 1980). As expected, perceptions of media influence on ingroup and outgroup others also varied in predictable ways according to the changing nature of the intergroup context. For instance, results indicate that respondents who identified strongly with their preferred party were keen to emphasize the relative invulnerability of the ingroup after the election result had been announcedpresumably in an attempt to confirm that ingroup members had voted on the basis of strong convictions rather than as a result of media campaigning. Results also showed that the intergroup contrast increased over time for members of the dominant-losing group. A further analysis focusing on the perceived impact of negative political advertising provided clearer evidence of intergroup contrasts and indicated that the magnitude of the intergroup contrast varied according to group membership and time for high identifiers, but not for low identifiers (see Fig. 14.3). As expected, prior to the election the ingroup-serving intergroup contrast was more pronounced for high-identifying members of the subordinate group than for high-identifying members of the dominant group. However, the magnitude of the intergroup contrast increased over time for high-identifying members of the dominant-losing group and, after the election, the losers perceived that, compared to ingroup others, outgroup others had been significantly more influenced by negative political advertising. These results suggest that members of the dominant-losing group tried to rationalize the election outcome not simply by perceiving that media coverage of the election campaign had been biased after all but also by perceiving that the media campaignand especially negative media contenthad had a disproportionate effect on their political rivals. Analyses of the attributions for media influence on self and other also demonstrated basic self- and group-serving biases that were affected in predictable ways by group membership, time, and strength of identification. Overall, respondents were most willing to make dispositional attributions for media influence on outgroup others and voters in general and least willing to make dispositional attributions for media influence on themselves. Moreover, consistent with both social identity theory and the notion of group-serving attributions, respondents who identified strongly with their political ingroup generalized this self-serving tendency to members of their ingroup"we are not gullible but they are." Further analyses show the important role of the changing intergroup context. Prior to the election, members of the subordinate group were more willing than their dominant-group counterparts were to make us-them distinctions in their perceptions of per-

< previous page

page_279

next page >

< previous page

page_280

next page > Page 280

FIG. 14.3 Perceived ingroup-outgroup difference in influence of negative political advertising according to group membership and time for high identifiers. Note. Asterisks indicate means that reflect a significant ingroup-outgroup difference; *p < .001. ceived gullibility. However, after the election, high-identifying members of the dominant-losing group showed evidence of pronounced us-them distinctions when making dispositional attributions for media impact. Indeed, their willingness to say that outgroup others and voters in general were influenced because they were gullible and easily influenced suggests that the losers' dissatisfaction with the voting population was widespread. By contrast, after the election, high-identifying members of the subordinate-winning group were less critical of voters in general and, although making intergroup differentiations between the ingroup and outgroup, tended to perceive voters in general more favorablylike the ingroup rather than like the outgroup. Presumably, the winners were more willing to embrace as "us" the electorate that had just voted their party into government.

< previous page

page_280

next page >

< previous page

page_281

next page > Page 281

Taken together, the responses of politically aligned voters who identified with one of the two major parties reflected their membership in a dominant or subordinate group pre-election and in a losing or winning group postelection. Respondents who identified with the party in power (i.e., dominant group members) were initially less biased in their views of the campaign and its impact on the electorate. However, after the election result was announced, members of the dominant-losing party sought to blame their party's loss on media bias and on the gullibility of political outgroup members and voters in general. Subordinate group members initially showed strong ingroup-serving biases (i.e., stronger perceptions of hostile media bias, clearer third-person perceptions) but were less critical of the media and the electorate after their party had won. The results of this study highlight a number of important ways in which political partisanship can influence perceptions of a political media campaign and emphasize the dynamic and reactive quality of communication perceptions. Specifically, our results indicated asymmetries in the communication perceptions of politically aligned voters that reflected the differing social identity concerns implicit in their respective positions as the dominant and subordinate group in an unstable pre-election context and as losers and winners postelection. Our results suggest that the way we apprehend and respond to persuasive communications on controversial issues depends not only on ''with whom one stands" (see Price, 1989) but also on the changing nature of relations between the sides. Moreover, our results indicated not only that communication perceptions are dynamic rather than static or fixed but also that communication perceptions can change suddenly and dramatically over time in response to changes in the nature of the intergroup context. Student Identity and the Perceived Effects of AIDS/HIV Advertisements In a further study (Duck, Hogg, & Terry, in press), we examined students' perceptions of the impact of AIDS advertisements on self, students (ingroup), nonstudents (outgroup), and people in general. Results from our 1996 election study had indicated that, politically identified respondents perceived self and ingroup others as less influenced than others by campaign material that explicitly favored the "wrong" side, although they perceived self and ingroup others to be more influenced than political outgroup others by campaign content that explicitly favored the "right" side (in a reversal of the typical third-person effect). These findings suggest that perceived self-other differences in persuasibility depend not only on how we perceive others in terms of relevant group memberships (i.e., as ingroup or

< previous page

page_281

next page >

< previous page

page_282

next page > Page 282

outgroup others) but also on the extent to which persuasibility on a specific content dimension is considered normative or desirable for the relevant ingroup. When resistance to persuasion is socially acceptable (e.g., when political partisans are faced with a persuasive message promoting their political opponents), people should see themselves and members of their ingroup as highly resistant and others as less so (a third-person effect). But, when it is socially acceptable to think of oneself as influenced (e.g., when political partisans are faced with a persuasive message promoting their preferred party), people should see themselves and members of their ingroup as quite yielding and others as less so (a reverse third-person effect). Compared with other theoretical accounts of the third-person effect (e. g., Gunther & Mundy, 1993), this proposal emphasizes that the social desirability of influence on a particular content dimension is defined in terms of salient group norms and, therefore, that perceived self-other differences in media influence reflect social-normative processes. This study was designed to explore further this issue. It examined university students' perceptions of the impact of televised AIDS advertisements not only on self and people in general (see Duck, Terry, & Hogg, 1995) but also on students in general (ingroup others) and on nonstudents (outgroup others). Acquired Immunodeficiency Syndrome is one of the most pressing current health problems and is an issue of particular relevance to university students. Research shows that students engage in sexual practices that lead to increased exposure to HIV infection (e.g., Crawford, Turtle, & Kippax, 1990; Rosenthal, Hall, & Moore, 1992; Struckman-Johnson, Gilliland, Struckman-Johnson, & North, 1990) and many HIV campaigns promoting safe sex have explicitly targeted people in this age bracket. The issues of AIDS and safe sex are frequently discussed in student newspapers and student unions promote safe sex through the distribution of free condoms and information about AIDS/HIV during university orientation weeks. Hence, we reasoned that students who identified strongly with the student community would perceive acknowledging the impact of AIDS advertisements as socially acceptable. However, given a general reluctance to admit persuasibility, especially to mass media messages, we reasoned that students who did not identify with the student community would perceive acknowledging such influence as personally undesirable. The target issue of AIDS/HIV advertising presented an interesting scenario in which identification with a relevant reference group might serve to reverse typical third-person perceptions. Specifically, based on self-categorization theory, we predicted that judgments of media impact on self, and on other (students, nonstudents and people in general) would vary with the perceiver's level of identification with the student ingroup, that is, with the salience of student identity.

< previous page

page_282

next page >

< previous page

page_283

next page > Page 283

Respondents who did not identify strongly with the student ingroup were expected to perceive themselves to be less influenced than others (students, nonstudents, and people in general), whereas respondents who identified strongly with the student ingroup would perceive self and students (ingroup) as more influenced than nonstudents (outgroup). Compared with respondents who did not identify strongly with the student ingroup, students who identified strongly with the student ingroup would perceive more media influence on self and students (ingroup), less difference between the level of media influence on self and on students (ingroup members), and more difference between the level of media influence on students (ingroup) and nonstudents (outgroup). We also reasoned that the effects of social identity would be clearest when there was a norm for acknowledgment of personal influence and, given wide variation in the quality and character of AIDS advertisements, this norm was expected to be stronger for some advertisements than for others. Hence, we predicted that our hypotheses would be supported more clearly when we focused on the perceived impact of a subset of AIDS advertisements that students identified as being good to be influenced by, and that this improvement in support for our predictions would be unique to this subset of advertisements and not hold for other subsets of advertisements judged as, say, the better or more informative advertisements. As predicted, perceived self-other differences in impact varied with the salience of student identity (see Fig. 14.4). Respondents who did not identify strongly with the student ingroup displayed the typical third-person effect, judging themselves as less influenced by AIDS advertising than others. Students who identified strongly with the student ingroup were more willing to acknowledge the impact of AIDS commercials on themselves and, to a lesser extent, on students in general. They differentiated less clearly between their judgments of impact on self and students and more clearly between their judgments of impact on students and nonstudents, especially when considering the impact of a subset of advertisements for which influence was normatively acceptable and, in this context, they perceived self and students as somewhat more influenced than nonstudents. Thus, our results also suggest that, when social identity is salient, perceived self-other differences in influence reflect both the social psychological relationship between self and comparison others (ingroup-outgroup) and ingroup norms concerning the acceptability of the acknowledgment of the persuasive influence of specific content. When the message content is related to the salient categorization, perceivers accentuate the similarities between self and ingroup others in terms of level of impact (an assimilation effect) and accentuate the differences between ingroup and outgroup others in terms of level of impact (a contrast effect), perceiving their ingroup as

< previous page

page_283

next page >

< previous page

page_284

next page > Page 284

FIG. 14.4 Perceived influence on self, students, nonstudents, and people in general of five AIDS commercials that students identified as being good to be influenced by. positively distinct relative to the outgroup. This is consistent with social identity theory and, more specifically, with the notion of positive distinctiveness (e.g., Tajfel & Turner, 1979) and the self-esteem hypothesis (Hogg & Abrams, 1990), in which evaluative bias in favor of own-group identities can be viewed as a consequence of social identificationan extension of self-esteem at the group level (e.g., Brown, Collins, & Schmidt, 1988; Crocker, Blaine, & Luhtanen, 1993; Crocker & Luhtanen, 1990). Students who identified strongly with the student community were more willing to acknowledge the impact of AIDS advertisements on themselves and, to a lesser extent, on students in general and, in fact, judged themselves and ingroup members as more influenced than outgroup others by the advertisements that students say it is good to be influenced by. These results are consistent with our assumption that acknowledgment of the impact of AIDS advertisements within the student community is socially acceptable; hence students who identify strongly with the student ingroup perceive the student community (and themselves) as more responsive to safe-sex messages than are nonstudent others.

< previous page

page_284

next page >

< previous page

page_285

next page > Page 285

Clearly, these results indicate that a perception of self as less persuaded than others is not an inevitable and universal response, at least where the media message is positive in nature (cf. Perloff, 1993). Moreover, in demonstrating that ingroup norms for the acknowledgment of personal influence can reverse typical third-person perceptions, these results complement and extend previous findings that have identified contexts in which the third-person effect is eliminated or reversed (e.g., Duck & Mullin, 1995; Duck, Terry, & Hogg, 1995; Gunther & Mundy, 1993; Gunther & Thorson, 1992). It must be stressed that, although students who identified strongly with the student ingroup were more willing to acknowledge the personal impact of AIDS advertisements in general, a perception of self and ingroup others as more influenced than outgroup others was only apparent in the context of the five advertisements for which normative acceptance of influence was clearestthe us-them distinction was not apparent in judgments of the 11 advertisements overall or in judgments of subsets of advertisements classified as effective-informative or as emotional-unpleasant. This accords with self-categorization theory in suggesting that evidence of ingroup favoritism is more pronounced on evaluative dimensions that are relevant to the social categorization (e.g., see Turner et al., 1987) and parallels findings from our 1993 election study (Duck, Hogg, & Terry, 1995) that showed that us-them distinctions in judgments of political campaign impact were more pronounced when respondents considered the impact of campaign content that explicitly favored one or other political side. Respondents who identified strongly with the student ingroup did not merely perceive self and students as more responsive than nonstudents to AIDS advertisements in general, perhaps because the social identity, student, was only indirectly related to the issue at hand, AIDS advertising. However, respondents perceived self and students as more responsive than nonstudents to AIDS advertisements which students say it is good to be influenced byadvertisements that focused on young people with whom university students, as opposed to nonstudent others, might identify. Gender Identity and the Perceived Impact of Commercial Advertising We found further support for predictions from SIT and self-categorization theory and for the flexible nature of thirdperson perceptions, using an experimental or situational manipulation of group salience (Duck, 1997). In this study, male and female respondents evaluated the effect of commercial advertising on a range of targets including self, women in general, and men in general. They did so in either a control condition or in an experimental

< previous page

page_285

next page >

< previous page

page_286

next page > Page 286

condition in which gender identity was made salient in a competitive, intergroup context. Specifically, in the experimental condition, the ingroup was derogated by the outgroup for their vulnerability to commercial advertising, thus providing a direct threat to social identity on the focal dimension (cf. Lee & Ottati, 1995). In accordance with the notion of a third-person effect, respondents were expected to perceive themselves as less influenced than others (''men in general" and "women" in general). Consistent with SIT and self-categorization theory, perceptions of influence on self and others were expected to vary according to the salience of social identity, producing significant differences between perceptions of persuasive effects on self, ingroup others, and outgroup others in the experimental and control conditions. Specifically, and, relative to the control condition, in the experimental conditions in which gender identity was threatened, respondents were predicted to perceive the gender ingroup as less vulnerable to commercial advertising, to perceive less difference in advertising impact on self and ingroup, and to present the ingroup more favorably relative to the outgroup. Gender identifications as males and females, also are recognized as profound and complex identifications that reflect the sociohistorical context of relationships between the sexes. As emphasized by Hogg and Abrams (1988), social identifications as men and women are made against an objective background structure of power, status, and prestige asymmetries between the two groups. Men have traditionally been the breadwinners, competitors, providers, and achievers, whereas women, traditionally have been the dependents, supporters, consumers, and emotional responders (see Archer & Lloyd, 1982; Deaux, 1985; Eagly, 1987; Greenglass, 1982). To this extent, men constitute a dominant group, whereas women comprise a subordinate, dominated, or "minority" group (see Lorenzi-Cioldi, 1993). Moreover, powerful gender stereotypes prevail (see Bakan, 1966; Spence, Helmreich, & Stapp, 1974; Williams, 1984; Williams & Best, 1982). Accordingly, perceptions of persuasive impact on "women in general" and "men in general" were assumed to reflect sex stereotypes (e.g., of women as the weaker, more vulnerable and more impressionable sex) and that both male and female respondents would perceive women in general as more influenced by commercial advertising than men in general. The responses of male and female respondents to the experimental manipulations also were assumed to differ in accordance with their respective status as members of dominant and dominated groups. Research suggests that membership criterion is highly salient and cognitively accessible for dominated group members (see Lorenzi-Cioldi, 1988, 1993). According to Doise (1988), "members of dominated groups define themselves more, and are also defined more by others in terms of social categorizations imposed on them'' (p. 105). Moreover, "minority" group members are particularly motivated to counter-

< previous page

page_286

next page >

< previous page

page_287

next page > Page 287

act the threat to self-esteem associated with group membership by accentuating their positive social identity (see Simon, 1992). Hence, female respondents were predicted to react more strongly to the experimental manipulation of gender identity and especially to the threat to self-esteem embodied in the suggestion that men think women are the more vulnerable sex. Although there was clear evidence of perceived self-other distinctions in both the control and experimental conditions (see Fig. 14.5), when gender identity was threatened, there was increased evidence of group-based responses and especially of self-ingroup assimilation. In the experimental condition (when outgroup members derogated the ingroup for vulnerability to commercial advertising), respondents were motivated to evaluate their gender ingroup more positively (i.e., as less influenced) and to acknowledge somewhat more influence on themselves. Thus, consistent with the notion of ingroup assimilation, perceived self-other differences were significantly reduced when the comparison other constituted a salient self-category or ingroup. Planned analyses further indicated that these patterns were more pronounced for females than for males. In the control condition, both male and female respondents perceived womenthe subordinate groupas

FIG. 14.5 Perceived influence of commercial advertising on self, ingroup, and outgroup members as a function of experimental context.

< previous page

page_287

next page >

< previous page

page_288

next page > Page 288

more vulnerable than men were to the effects of commercial advertising, ostensibly based on the rationale that women are targeted more by commercial advertisers. However, in the experimental condition, perceptions of influence on the ingroup relative to the outgroup reflected more favorably on the ingroup than they had in the control condition. For male respondents, the tendency to perceive women (the outgroup) as more vulnerable than men was exaggerated in the experimental condition, whereas, for females, this tendency was negated in the experimental condition. In this competitive, intergroup context female respondents asserted their positive social identity by suggesting that women and men are equally vulnerable to commercial advertising. This study provided further evidence that third-person perceptions are moderated by the social psychological relationship between self and otherthat is, by salient group memberships. Moreover, in keeping with the macrosocial perspective in SIT that argues that sociohistorical factors should be incorporated theoretically to furnish an adequate explanation of intergroup behavior (see Hogg & Abrams, 1988), results indicated differences in the responses of male and female respondents that are consistent with their respective sociopolitical status as dominant and dominated groups. Attitudinal Position on Life Imprisonment and Perceived Effects of a Newspaper Editorial on the Issue: Third-Person Perceptions and Behavior As noted at the beginning of this chapter, the third-person effect Davison (1983) described has two components: People not only expect the media and persuasive communications in general to have a greater effect on others than on themselves but they also act on this assumption. A final study (Morgan & Duck, 1996) was designed to show that salient social identities affect both perceptions of persuasive impact on self and other as well as behavior based on those beliefs. Students were recruited for a study on attitudes toward current issues. On the basis of their attitude on a focal issuethe issue of automatic life imprisonment following three offensesthey were exposed to a balanced or counterattitudinal article on the issue (a mock newspaper editorial) and were asked to judge the potential impact of this article on themselves and on others. This was done in the context of manipulations designed to make salient either their personal identity and unique opinions or their social identity as a member of an attitudinal ingroup. Half the students were also provided with an ingroup norm for censorshipfor the others no norm was implied. Specifically, respondents in the norm condition read an article that

< previous page

page_288

next page >

< previous page

page_289

next page > Page 289

ostensibly was from a recent campus newspaper in another major capital city describing protests against publication of the article by students who shared their opinion on life imprisonment. Thus the experiment comprised a 2 (personal or group orientation) x 2 (norm or no norm context) x 2 (balanced or counterattitudinal article) between-subjects design. Participants judged perceived impact on self and on others who did and did not share the opinion (i.e., the attitudinal ingroup and the attitudinal outgroup) and the association between perception of persuasive impact on self and other (third-person perceptions) and behavior was assessed. Behavioral measures included measures of both behavioral intentions (e.g., intention to recommended censorship of the article, intention to engage in public and private protest on the issue) and actual behavior (e.g., voting to censor publication of the article in the local student newspaper, signing a petition on the issue). We reasoned that respondents would perceive members of the attitudinal outgroup to be more influenced than were self and members of the attitudinal ingroup and that this pattern would be magnified for respondents in the social orientation condition and when respondents considered the impact of the counterattitudinal versus the balanced article. Furthermore, we predicted that the presence of an ingroup norm for censorship and exposure to a counterattitudinal article would strengthen behavioral intentions aimed at censorship of the article, especially among respondents in the social orientation condition. Finally, we reasoned that there would be a link between third-person perceptions and behavioral intentions and that the effects of the experimental manipulations on behavior would be mediated by thirdperson perceptions. As predicted, respondents judged outgroup others as significantly more influenced by the newspaper editorial than the self and other ingroup members providing further evidence for an us-them or group-based distinction in perceptions of persuasive influence. As predicted, the magnitude of this third-person effect also varied with norm and article, although, contrary to predictions, these effects were not moderated by the manipulation of orientation. Respondents who judged the impact of a counterattitudinal article perceived a larger self-other discrepancy than respondents who judged the impact of a balanced article, although the self-other difference was still significant with respect to the balanced article. The third-person effect also was more pronounced when there was an ingroup norm for censorship. Both the provision of an ingroup norm for censorship and exposure to a counterattitudinal article also led to stronger intentions to act to censor the article. More important, results provided evidence of a significant link between thirdperson perceptions and behavior, both actual and intendedthe greater the magnitude of third-person perceptions, the stronger the calls for censorship. Third-person perceptions (i.e., perceived

< previous page

page_289

next page >

< previous page

page_290

next page > Page 290

self-outgroup differences in influence) added significantly to the prediction of 9 out of 10 behavioral measures, over and above the contributions of the experimental manipulations and perceptions of outgroup influence. Mediational analyses further indicated that, in the prediction of behavior (e.g., intention to censor the article), third-person perceptions fully mediated the effect of ingroup norm, and partially mediated the effect of article. The results of this study provide important evidence not only for the behavioral component of third-person perceptions but also for the social-psychological basis of the effect. As noted earlier, evidence for the behavioral component of the third-person effect is limited by comparison to evidence for the perceptual component of the effect. Hence, this experimental test of the behavioral component provided additional support for Davison's (1983) original formulation. Moreover, this research constitutes the first attempt to demonstrate the important mediational role of third-person perceptions in behavioral responses to persuasive messages (e.g., media censorship). Given a number of limitations with this experimental study, including the absence of significant effects involving the manipulation of social orientation, further work of this nature is required. Conclusions Our results highlight the need to re-emphasize the role of social groups, not only in understanding persuasion per se (see chapters in this volume by Mackie & Queller; van Knippenberg; and Fleming & Petty) but also in understanding perceptions of the persuasive impact of messages on self and other. Our research findings suggest that perceptions of persuasive impact on self and other are sensitive to the social psychological relationship between self and other in ways that are consistent with the theoretical assumptions of SIT. That is, third-perceptions are influenced by the subjectively salient social relationship between self and other and are governed by motivational needs, such as self-esteem, social identity, and differentiation from others (cf. Brewer, 1991; Hogg & Abrams, 1993). In this sense, third-person perceptions often represent not simply a self-serving perceived self-other difference in persuasive impact but also a group-serving us-them distinction. It is important to note, however, that even when social identity is salient, respondents often perceive themselves as uniquely invulnerable to influencethat is, they perceive themselves as positively distinct from, rather than interchangeable with, ingroup members. Intergroup contrast in perceived persuasive impact is more readily observed than is self-ingroup assimilation. Perhaps as Simon (1993) suggested, "the cognitive differentiation between oneself and other people is much clearer than the

< previous page

page_290

next page >

< previous page

page_291

next page > Page 291

cognitive differentiation between different others" (p. 133), and the basic distinction the perceiver makes in the social world is between "me" and "not me." This may be partly because as individuals, we have has superior knowledge about our own behavior, but the distinction between self and others may be sought for its own sakebecause a feeling of uniqueness is central to the self-image (see Codol, 1984). Indeed, self-ingroup differentiation may reflect the primus inter pare, or first among equals, effect (Codol, 1984) wherein people perceive themselves as conforming most closely to the prevailing group normin this case, of invulnerability to influence. Nonetheless, it has become increasingly clear that research and theorizing on third-person perceptions does benefit from a recognition of the role of reference groups and salient group identities and, specifically, from a recognition of the usthem distinction implicit in a literal interpretation of the phenomenon. Future research will explore more fully the ways in which salient social identities affect perceptions of the impact of persuasive messages on self, ingroup, and outgroup others and will consider how the belief that others are more influenced than oneself might inform research on groups and attitude change (e.g., Mackie et al., 1992; Mackie et al., 1990; van Knippenberg & Wilke, 1992; Wilder, 1990). For instance, our research indicates that, even where beneficial messages are concerned, "invulnerability to influence" is commonly a part of the contextually salient ingroup prototypepresumably limiting the direct persuasive impact of a message. People who are willing to acknowledge impact are more likely to be amenable to desired behavioral and attitudinal change, and a particular aim of future research might be to identify the intergroup contexts in which ingroup persuasibility, rather than invulnerability, could operate as the norm. At present, research on perceived, rather than actual, persuasion provides an interesting dimension that is not addressed in more traditional accounts of the attitudebehavior relation. But attempts to integrate more closely findings in the two domains promises to provide a more general account of the role of social-normative factors in the persuasion process.

< previous page

page_291

next page >

< previous page

page_i

next page > Page i

Attitudes, Behavior, and Social Context The Role of Norms and Group Membership

< previous page

page_i

next page >

< previous page

page_293

next page > Page 293

REFERENCES Abelson, R. P. (1959). Modes of resolution of belief dilemmas. Journal of Conflict Resolution, 3, 343-352. Abelson, R. P. (1981). Psychological status of the script concept. American Psychologist, 36, 715-729. Abelson, R. P. (1982). Three models of attitude-behavior consistency. In M. P. Zanna, E. T Higgins, & C. P. Herrmann (Eds.), The Ontario Symposium: Vol. 2. Consistency in social behavior (pp. 131-146). Hillsdale, NJ: Lawrence Erlbaum Associates. Abelson, R. P., Kinder, D. R., Peters, M. D., & Fiske, S. T (1982). Affective and semantic components in political person perception. Journal of Personality and Social Psychology, 42, 619-630. Abrams, D., & Hogg, M. A. (1990). Social identification, self-categorization, and social influence. European Review of Social Psychology, 1, 195-228. Abrams, D., & Hogg, M. A. (1999). Social identity and social cognition. Oxford, UK: Blackwell. Abrams, D., Wetherell, M., Cochrane, S., Hogg, M. A., & Turner, J. C. (1990). Knowing what to think by knowing who you are: Self-categorization and the nature of norm formation, conformity and group polarization. British Journal of Social Psychology, 29, 97-119. Acock, A. C., & DeFleur, M. L. (1972). A configurational approach to contingent consistency in the attitude-behavior relationship. American Sociological Review, 37, 714-726. Agostinelli, G., Sherman, S. J., Presson, C. C., & Chassin, L. (1992). Self-protection and self-enhancement biases in estimates of population prevalence. Personality and Social Psychological Bulletin, 18, 631-642. Ainslie, G. (1992). Picoeconomics: The strategic interaction of successive motivational states within the person. Cambridge, England: Cambridge University Press. Ajzen, I. (1987). Attitudes, traits, and actions: Dispositional prediction of behavior in personality and social psychology. Advances in Experimental Social Psychology, 20, 1-64. Ajzen, I. (1988). Attitudes, personality, and behavior. Chicago: The Dorsey Press. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 170211. Ajzen, I., & Fishbein, M. (1972). Attitudes and normative beliefs as factors influencing behavioral intentions. Journal of Personality and Social Psychology, 21(1), 1-9. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall. Akers, R. L. (1977). Prisonization in five countries: Types of prison and inmate characteristics. Criminology, 14, 527742. Alinsky, S. D. (1971). Rules for radicals. New York: Vintage.

< previous page

page_293

next page >

< previous page

page_294

next page > Page 294

Allen, V. L. (1975). Social support for nonconformity. Advances in Experimental Social Psychology, 8, 2-43. Allen, V., & Wilder, D. (1977). Social comparison, self-evaluation, and group conformity. In J. M. Suls & R. L. Miller (Eds.), Social comparison processes: Theoretical and empirical perspectives (pp. 187-206). New York: Wiley. Allen, V., & Wilder, D. (1978). Perceived persuasiveness as a function of response style: Multi-issue consistency over time. European Journal of Social Psychology, 8, 289-296. Allen, V., & Wilder, D. (1979). Group categorization and attribution of belief similarity. Small Group Behavior, 10, 73-80. Allport, F H. (1924). Social psychology. Boston: Houghton Mifflin. Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), Handbook of social psychology (pp. 798-844). Worcester, MA: Clark University Press. Altman, D. G. (1990). The social context and health behavior: the case of tobacco. In S. A. Shumaker, E. B. Schron, & J. K. Ockene. (Eds.), The handbook of health behavior change. New York: Springer. American Lung Association (1987). Smokers know risks but don't say "no" to cigarettes. [press release]. New York: Author. Andersen, K. E., & Clevenger, T., Jr. (1963). A summary of experimental research in ethos. Speech Monographs, 30, 59-78. Andersen, S. M., & Klatzky, R. L. (1987). Traits and social stereotypes: Levels of categorization in person perception. Journal of Personality and Social Psychology, 53, 235-246. Andrews, K. H., & Kandel, D. B. (1979). Attitude and behavior: A specification of the contingent-consistency hypothesis. American Sociological Review, 44, 298-310. Antaki, C., Condor, S., & Levine, M. (1996). Social identities in talk: Speakers' own orientations. British Journal of Social Psychology, 35, 473-492. Archer, J., & Lloyd, B. (1982). Sex and gender. Harmondsworth: Penguin. Arie, O. G., Durand, R. M., & Bearden, W. O. (1979). Attitudinal and normative dimensions of opinion leaders and nonleaders. The Journal of Psychology, 101, 305-312. Armitage, C., & Conner, M. (1997). Distinguishing perceptions of control from self-efficacy: Predicting low-fat diet using the theory of planned behavior. Manuscript submitted for publication. Aronson, E. (1969). The theory of cognitive dissonance: A current perspective. Advances in Experimental Social Psychology, 4, 2-34. Aronson, E., & Golden, B. W. (1962). The effects of relevant and irrelevant aspects of communicator credibility on opinion change. Journal of Personality, 30, 135-146. Aronson, E., & Mills, J. M. (1959). The effect of severity of initiation on liking for a group. Journal of Abnormal and Social Psychology, 59, 177-181. Asch, S. (1951). Effects of group pressure upon the modification and distortion of judgments. In H. Guetzkow (Ed.), Groups, leadership and men (pp. 177-190). Pittsburgh, PA: Carnegie Press. Asch, S. E. (1952). Social psychology. Englewood Cliffs, NJ: Prentice-Hall. Asch, S. E. (1956). Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs: General and Applied, 70, 1-70. Back, K. (1951). Influence through social communication. Journal of Abnormal and Social Psychology, 46, 9-23. Bacon, C., & Lerner, R. M. (1975). Effects of maternal employment status on the development of vocational-role perception in females. The Journal of Genetic Psychology, 126, 187-193. Bagozzi, R. P., Baumgartner, H., & Yi, Y. (1992). State versus action orientation and the theory of reasoned action: An application to coupon usage. Journal of Consumer Research, 18, 505-518.

< previous page

page_294

next page >

< previous page

page_295

next page > Page 295

Bagozzi, R. P., & Schnedlitz, P. (1985). Social contingencies in the attitude model: A test of certain interaction hypotheses. Social Psychology Quarterly, 48, 366-373. Bakan, D. (1966). The duality of human existence. Chicago: Rand McNally. Bandura, A. (1996). Social foundations of thought & action: A social cognitive theory. Englewood Cliffs, NJ: PrenticeHall. Barber, R., & Creamer, R. (1997). Rhubarb in the catbird seat. Lincoln, NE: Bison Books. Bargh, J. A. (1990). Auto-motives: Preconscious determinants of social interaction. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2, pp. 93-130). New York: Guilford. Bargh, J. A., Chen, M., & Burrows, L. (1996). 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. A., & Thein, R. D. (1985). Individual construct accessibility, person memory, and the recall-judgment link: The case of information overload. Journal of Personality and Social Psychology, 49, 1129-1146. Baron, J. (1988). Thinking and deciding. Cambridge, England: Cambridge University Press. Baron, J. (1992). The effect of normative beliefs on anticipated emotions. Journal of Personality and Social Psychology, 63, 320-330. Barton, J., Chassin, L., Presson, C. C., & Sherman, S. J. (1982). Social image factors as motivators of smoking initiation in early and middle adolescence. Child Development, 53, 1499-1511. Bassili, J. N. (1995). Response latency and the accessibility of voting intentions: What contributes to accessibility and how it affects vote choice. Personality and Social Psychology Bulletin, 21, 686-695. Bauman, K. E., & Chenoweth, R. L. (1984). The relationship between the consequences adolescents expect from smoking and their behavior: A factor analysis with panel data. Journal of Applied Social Psychology, 14(1) 28-41. Beach, L. R. (1990). Image theory: Decision making in personal and organizational contexts. Chichester, England: John Wiley. Beattie, J., Baron, J., Hershey, J. C., & Spranca, M. D. (1994). Psychological determinants of decision attitude. Journal of Behavioral Decision Making, 7, 129-144. Beck, L., & Ajzen, I. (1991). Predicting dishonest actions using the theory of planned behavior. Journal of Research in Personality, 25, 285-301. Beckhouse, L., Tanur, J., Weiler, J., & Weinstein, E. (1975). And some men have leadership thrust upon them. Journal of Personality and Social Psychology, 31, 557-566. Becoña, E. (1995). Evaluación de la eficacia de las acciones emprendidaspor la Consellería de Sanidade e Servicios Sociales en el Día Mundial sin Tabaco (31 de Mayo) del año 1995. Consellería de Sanidade e Servicios Sociales (Dirección Xeral de Organización Sanitaria) and University of Santiago de Compostela Bell, D. E. (1982). Regret in decision making under uncertainty. Operations Research, 21, 961-981. Benaquisto, L., & Freed, P. J. (1996). The myth of inmate lawlessness: The perceived contradiction between self and other in inmates' support for criminal justice sanctioning norms. Law & Society Review, 30, 481-511. Bentler, P. M., & Speckart, G. (1979). Models of attitude-behavior relations. Psychological Review, 86, 452-464. Bentler, P. M., & Speckart, G. (1981). Attitudes "cause" behavior: A structural equation analysis. Journal of Personality and Social Psychology, 40, 226-238. Berger, P. (1966). Identity as a problem in the sociology of knowledge. Archives Europeennes De Sociologie, 7, 105115.

< previous page

page_295

next page >

< previous page

page_296

next page > Page 296

Berger, C. J., & Cummings, L. L. (1979). Organizational structure, attitudes, and behavior. Research in Organizational Behavior, 1, 169-208. Berkowitz, L., & Green, J. A. (1962). The stimulus qualities of the scapegoat. Journal of Abnormal and Social Psychology, 64, 293-301. Berscheid, E. (1966). Opinion change and communicator-communicatee similarity and dissimilarity. Journal of Personality and Social Psychology, 4, 670-680. Berscheid, E. (1985). Interpersonal attraction. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 2, pp. 413-484). New York: Random House. Bickman, L. (1974). The social power of a uniform. Journal of Applied Social Psychology, 4, 47-61. Biddle, B. J. (1979). Role theory: Expectations, identities, and behavior. New York: Academic Press. Biddle, B. J., Bank, B. J., Anderson, D. S., Hauge, R., Keats, D. M., Keats, J. A., Marlin, M. M., & Valantin, S. (1985). Social influence, self-referent identity labels, and behavior. Sociological Quarterly, 26, 159-185. Biddle, B. J., Bank, B. J., & Slavings, R. L. (1987). Norms, preferences, identities and retention decisions. Social Psychology Quarterly, 50(4), 322-337. Biddle, B. J., & Thomas, E. J. (1966). Role theory: Concepts and research. New York: Wiley. Blake, J., & Davis, K. (1964). Norms, values, and sanctions. In R. E. Faris (Ed.), Handbook of modern sociology (pp. 456-484). Chicago: Rand McNally. Blanz, M., Mummendey, A., & Otten, S. (1995). Perceptions of relative group size and group status: effects on intergroup discrimination in negative evaluations. European Journal of Social Psychology, 25, 231-247. Bobrow, D. G., & Norman, D. A. (1975). Some principles of memory schemata. In D. G. Bobrow & A. Collins (Eds.), Representation and understanding: Studies in cognitive science (pp. 131-150). New York: Academic Press. Bocock, R. (1993). Consumption. London: Routledge. Boninger, D. S., Krosnick, J. A., & Berent, M. K. (1995). Origins of attitude importance: Self-interest, social identification, and value relevance. Journal of Personality and Social Psychology, 68(1) 61-80. Bontempo, R. (1993). Translation fidelity of psychological scales. Journal of Cross-Cultural Psychology, 24, 149-166. Boorstin, D. J. (1961). The image: A guide to pseudo-events in America. New York: Atheneum. Bourdieu, P. (1984). Distinction: A social critique of the judgment of taste. London: Routledge & Kegan Paul. Bowman, C. C. (1949). Role-playing and the development of insight. Social Forces, 28, 195-199. Boyanowsky, E. O., & Allen, V. L. (1973). Ingroup norms and self-identity as determinants of discriminatory behavior. Journal of Personality and Social Psychology, 25, 408-418. Boyd, B., & Wandersman, A. (1991). Predicting undergraduate condom use with the Fishbein and Ajzen and the Triandis attitude-behavior models: Implications for public health interventions. Journal of Applied Social Psychology, 21, 1810-1830. Breakwell, G. (1986). Coping with threatened identities. London: Methuen. Breckler, S. J. (1984). Empirical validation of affect, behavior, and cognition as distinct components of attitude. Journal of Personality and Social Psychology, 47, 1191-1205. Breckler, S. J., & Wiggins, E. C. (1989). Affect versus evaluation in the structure of attitudes. Journal of Experimental Social Psychology, 25, 253-271. Breed, W., & Ktsanes, T. (1961). Pluralistic ignorance in the process of opinion formation. Public Opinion Quarterly, 25, 382-392.

< previous page

page_296

next page >

< previous page

page_297

next page > Page 297

Brehm, J. W. (1956). Post-decision changes in desirability of alternatives. Journal of Abnormal and Social Psychology, 52, 384-389. Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and control. San Diego, CA: Academic Press. Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-motivational analysis. Psychological Bulletin, 86, 307-324. Brewer, M. B. (1988). A dual process model of impression formation. In. T. K. Srull & R. S. Wyer (Eds.), Advances in social cognition, (Vol. 1, pp. 1-36). Hillsdale, NJ: Lawrence Erlbaum Associates. Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17, 475-482. Brewer, M. B. (1993a). Social identity, distinctiveness, and in-group homogeneity. Social Cognition, 11, 150-164. Brewer, M. B. (1993b). The role of distinctiveness in social identity and group behaviour. In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 1-16). London: Harvester Wheatsheaf. Brewer, M. B., & Kramer, R. M. (1985). The psychology of intergroup attitudes and behaviour. Annual Review of Psychology, 36, 219-243. Brock, T. C. (1962). Cognitive restructuring and attitude change. Journal of Abnormal and Social Psychology, 64, 264271. Brock, T. C. (1965). Communicator-recipient similarity and decision change. Journal of Personality and Social Psychology, 1, 650-654. Brock, T. C., & Balloun, J. C. (1967). Behavioral receptivity to dissonant information. Journal of Personality and Social Psychology, 6, 413-428. Brown, J. D., Collins, R. L., & Schmidt, G. W. (1988). Self-esteem and direct versus indirect forms of selfenhancement. Journal of Personality and Social Psychology, 55, 445-453. Brown, R. (1988). Group processes: Dynamics within and between groups. Oxford: Blackwell. Brown, R., Condor, S., Mathews, A., Wade, G., & Williams, J. (1986). Explaining intergroup differentiation in an industrial organization. Journal of Occupational Psychology, 59, 273-286. Brown, R. J. (1978). Divided we fall: An analysis of relations between a factory workforce. In H. Tajfel (Ed.), Differentiation between social groups (pp. 395-429). London: Academic Press. Brownson, R. C., Jackson-Thompson, J., Wilkerson, J. C., Davis, J. R., Owens, N. W., & Fisher, E. B. (1992). Demographic and socioeconomic differences in beliefs about the Health effects of smoking. American Journal of Public Health, 82(1), 99-102. Burger, J. M. (1981). Motivational biases in the attribution of responsibility for an accident: A meta-analysis of the defensive-attribution hypothesis. Psychological Bulletin, 90, 496-512. Burger, J. M., & Guadagno, R. E. (1997). Self-concept clarity, responsiveness to false feedback, and the foot-in-thedoor procedure. Unpublished manuscript, Santa Clara University, Santa Clara, CA. Burke, P. J., & Tully, J. C. (1977). The measurement of role identity. Social Forces, 55, 881-897. Burnkrant, R. E., & Unnava, H. R. (1989). Self-referencing: A strategy for increasing processing of message content. Personality and Social Psychology Bulletin, 15, 628-638. Burnkrant, R. E., & Unnava, H. R. (1995). Effects of self-referencing on persuasion. Journal of Consumer Research, 22, 17-26. Bushman, B. J. (1984). Perceived symbols of authority and their influence on compliance. Journal of Applied Social Psychology, 14, 501-508.

< previous page

page_297

next page >

< previous page

page_298

next page > Page 298

Bushman, B. J. (1988). The effects of apparel on compliance: A field experiment with a female authority figure. Personality and Social Psychology Bulletin, 14, 459-467. Butler, D. (1996, February 22). What difference does it make who wins? The Australian, p. 11. Buzzelli, C. A. (1991). Morality in context: A sociocultural approach to enhancing young children's moral development. Child and Youth Care Forum, 20, 375-386. Byrne, D. (1969). Attitudes and attraction. Advances in Experimental Social Psychology, 4, 35-89. Cacioppo, J. T, & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116131. Cacioppo, J. T., & Petty, R. E. (1985). Central and peripheral routes to persuasion: The role of message repetition. In L. F. Alwitt & A. A. Mitchell (Eds.), Psychological processes and advertising effects (pp. 91-111). Hillsdale, NJ: Lawrence Erlbaum Associates. Cacioppo, J. T., & Petty, R. E. (1989). Effects of message repetition on argument processing, recall, and persuasion. Basic and Applied Social Psychology, 10, 3-12. Cacioppo, J. T., Petty, R. E., Kao, C. F, & Rodriguez, R. (1986). Central and peripheral routes to persuasion: An individual difference perspective. Journal of Personality and Social Psychology, 51, 1032-1043. Cacioppo, J. T., Petty, R. E., & Sidera, J. A. (1982). The effects of a salient self-schema on the evaluation of proattitudinal editorials: Top-down versus bottom-up message processing. Journal of Experimental Social Psychology, 18, 324-338. Cacioppo, J. T., Priester, J. R., & Berntson, G. G. (1993). Rudimentary determinants of attitudes II: Arm flexion and extension have differential effects on attitudes. Journal of Personality and Social Psychology, 42, 16-131. Campbell, D. T. (1963). Social attitudes and other acquired behavioral dispositions. In S. Koch (Eds.), Psychology: A study of a science (pp. 94-172). New York: McGraw-Hill. Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105. Cancian, F. M. (1975). What are norms ? A study of beliefs and action in a Maya community. Cambridge, England: Cambridge University Press. Cantor, N., & Mischel, W. (1979). Prototypes in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 12, pp 3-52). New York: Academic Press. Cantor, N., Mischel, W., & Schwartz, J. (1982). A prototype analysis of psychological situations. Cognitive Psychology, 14, 45-77. Cecil, H., Evans, R. I., & Stanley, M. A. (1996). Perceived believability among adolescents of health warning labels on cigarette packs. Journal of Applied Social Psychology, 26(6), 502-519. Chaiken, S. (1979). Communicator physical attractiveness and persuasion. Journal of Personality and Social Psychology, 37, 1387-1397. Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-756. Chaiken, S. (1986). Physical appearance and social influence. In C. P. Herman, M. P. Zanna, & E. T. Higgins (Eds.), Physical appearance, stigma, and social behavior (pp. 143-177). Hillsdale, NJ: Lawrence Erlbaum Associates. Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna, J. M. Olson, & C. P. Herman (Eds.), The Ontario symposium: Vol. 5. Social influence (pp. 3-39). Hillsdale, NJ: Lawrence Erlbaum Associates.

< previous page

page_298

next page >

< previous page

page_299

next page > Page 299

Chaiken, S., Giner-Sorolla, R., & Chen, S. (1996). Beyond accuracy: defense and impression motives in heuristic and systematic information processing. In P. M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action: Linking motivation and cognition to behavior. New York: Guilford. Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic information processing within and beyond the persuasion context. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thoughts: Limits of awareness, intention and control (pp. 212-252). New York: Guilford. Chaiken, S., & Maheswaran, D. (1994). Heuristic processing can bias systematic processing: Effects of source credibility, argument ambiguity, and task importance on attitude judgment. Journal of Personality and Social Psychology, 66, 460-473. Chaiken, S., Wood, W., & Eagly, A. H. (1996). Principles of persuasion. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 702-742). New York: Guilford. Charng, H. W, Piliavin, J. A., & Callero, P. L. (1988). Role identity and reasoned action in the prediction of repeated behavior. Social Psychology Quarterly, 51(4), 303-317. Charters, W. W., & Newcomb, T. M. (1958). Some attitudinal effects of experimentally increased salience of a membership group. In E. E. Maccoby, T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psychology (3rd ed., pp. 276-281). New York: Holt, Rinehart & Winston. Chassin, L., Presson, C. C., Bensenberg, M., Corty, E., Sherman, S. J., & Olshevsky, R. W. (1981). Predicting adolescents' intentions to smoke cigarettes. Journal of Health and Social Behavior, 22, 445-455. Chassin, L., Presson, C. C., & Sherman, S. J. (1990). Social psychological contributions to the understanding and prevention of adolescent cigarette smoking. Personality and Social Psychology Bulletin, 16(1), 133-151. Chung, J. (1985). A study of social influence: Effects of type of referent, source of power, and level of compliance. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign. Cialdini, R. B. (1984). Influence. New York: Morrow. Cialdini, R. B. (1993). Influence: Science and practice (3rd ed.). Glenview, IL: Harper Collins. Cialdini, R. B. (1996). Social influence and the triple tumor structure of organizational dishonesty. In D. M. Messick & A, E. Tenbrunsel (Eds.), Codes of conduct (pp. 44-58). New York: Russell Sage. Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). A focus theory of normative conduct: A theoretical refinement and reevaluation of the role of norms in human behavior. Advances in Experimental Social Psychology, 24, 201-234. Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58, 1015-1026. Codol, J. P. (1984). Social differentiation and non-differentiation. In H. Tajfel (Ed.), The social dimension (pp. 314337). Cambridge, England: Cambridge University Press. Cohen, J., Mutz, D., Price, V., & Gunther, A. (1988). Perceived impact of defamation: An experiment on the thirdperson effects. Public Opinion Quarterly, 52, 161-173. Collins, D. M., Litman, T. J., Hadsall, R. S., & Gross, C. R. (1994). An investigation of role bias using a mail survey of AIDS-related issues. Journal of Personality and Social Behavior, 9, 775-788. Conner, M. T., & Sparks, P. (1996). The theory of planned behaviour and health behaviors. In M. T. Conner & P. Norman (Eds.), Social cognition in health: Predicting behaviour (pp. 121-162). Milton Keynes, England: Open University Press.

< previous page

page_299

next page >

< previous page

page_300

next page > Page 300

Conover, P. J. (1988). The role of social groups in political thinking. British Journal of Political Science, 18, 51-76. Cooley, C. H. (1964). Human nature and the social order. New York: Scribners. (Original work published 1902) Cooper, J., & Fazio, R. H. (1984). A new look at dissonance theory. Advances in Experimental Social Psychology, 17, 229-265. Cooper J., & Mackie, D. (1983). Cognitive dissonance in an intergroup context. Journal of Personality and Social Psychology, 44, 536-544. Cooper, J., & Worchel, S. (1970). Role of undesired consequences in arousing cognitive dissonance. Journal of Personality and Social Psychology, 16, 199-206. Coutu, W. (1951). Role-playing vs. role-taking: An appeal for clarification. American Sociological Review, 16, 180187. Crawford, J., Turtle, A., & Kippax, S. (1990). Student-favoured strategies for AIDS avoidance. Australian Journal of Psychology, 42, 123-137. Crites, S. L., Fabrigar, L. R., & Petty, R. E. (1994). Measuring the affective and cognitive properties of attitudes: Conceptual and methodological issues. Personality and Social Psychology Bulletin, 20, 619-634. Crocker, J., Blaine, B., & Luhtanen, R. (1993). Prejudice, intergroup behaviour and self-esteem: Enhancement and protection motives. In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 52-67). London: Harvester Wheatsheaf. Crocker, J., & Luhtanen, R. (1990). Collective self-esteem and ingroup bias. Journal of Personality and Social Psychology, 58, 60-67. Danto, A. (1963). What we can do. Journal of Philosophy, 40, 435-445. Darley, J. M., & Fazio, R. H. (1980). Expectancy confirmation processes arising in the social interaction sequence. American Psychologist, 35, 867-881. Davidson, A. R., & Morrison, D. M. (1983). Predicting contraceptive behavior from attitudes: A comparison of withinversus across-subjects procedures. Journal of Personality and Social Psychology, 45, 997-1009. Davis, K. E., & Jones, E. E. (1960). Changes in interpersonal perception as a means of reducing cognitive dissonance. Journal of Abnormal and Social Psychology, 61, 402-410. Davis, R. W. (1986). Pretrial publicity, the timing of the trial, and mock jurors' decision processes. Journal of Applied Social Psychology, 16(7), 590-607. Davis, J. H., Stasson, M. F, Ono, K., & Zimmerman, S. (1988). Effects of straw polls on group decision making: Sequential voting pattern, timing, and local majorities. Journal of Personality and Social Psychology, 55(6), 918-926. Davison, G. C. (1973). Counter control in behavior modification. In L. A. Hamerlynck, L. C. Handy, & E. J. Mash (Eds.), Behavior change: Methodology, concepts, and practice (pp. 153-167). Champaign, IL: Research Press. Davidson, D. (1980). Intending. In D. Davidson, Essays on Actions and Events. Oxford: Clarendon Press. Davison, W. P. (1983). The third-person effect in communication. Public Opinion Quarterly, 47, 1-15. Dawley, H. H., Fleischer, B., & Dawley, L. T. (1985). Attitudes toward smoking and smoking rate: implications for smoking discouragement. International Journal of Addictions, 20, 483-488. Deaux, K. (1985). Sex and gender. Annual Review of Psychology, 36, 49-81. Deaux, K. (1996). Social identification. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 777-798). New York: Guilford. 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.

< previous page

page_300

next page >

< previous page

page_301

next page > Page 301

DeBono, K. G., & Packer, M. (1991). The effects of advertising appeal on perceptions of product quality. Personality and Social Psychology Bulletin, 17, 194-200. DeCourville, N. H., & Zanna, M. P. (1993a, May). The role of affective beliefs and perceived immorality in predicting frequency of intoxification: An extension of the theory of planned behavior. Paper presented at the annual meeting of the Canadian Psychological Association, Montreal, Canada. DeCourville, N. H., & Zanna, M. P. (1993b, August). Theory of planned behavior: Application and extension to alcohol intoxification. Paper presented at the annual meeting of the American Psychological Association, Toronto, Canada. Dennison, C. M., & Shepherd, R. (1995). Adolescent food choice: An application of the theory of planned behaviour. Journal of Human Nutrition and Dietetics, 8, 9-23. Deshpande, R., Hoyer, W. D., & Donthu, N. (1986). The intensity of ethnic affiliation: A study of the sociology of hispanic consumption. Journal of Consumer Research, 13, 214-220. De Tocqueville, A. (1955). The old regime and the French Revolution (S. Gilbert, Trans.), New York: Doubleday/Anchor. (Original work published 1856) Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational influences upon individual judgment. Journal of Abnormal and Social Psychology, 51, 629-636. Devine, P. G., Hamilton, D. L., & Ostrom, T. M. (Eds.). (1994). Social cognition: Impact on social psychology. San Diego, CA: Academic Press. Doise, W. (1978). Groups and individuals. Cambridge: Cambridge University Press. Doise, W. (1988). Individual and social identities in intergroup relations. European Journal of Social Psychology, 18, 99-111. Doise, W., & Sinclair, A. (1973). The categorization process in intergroup relations. European Journal of Social Psychology, 3, 145-157. Doll, J., & Ajzen, I. (1992). Accessibility and stability of predictors in the theory of planned behavior. Journal of Personality and Social Psychology, 63, 754-763. Doll, R., & Hill, A. B. (1950). Smoking and carcinoma of the lung. British Medical Journal, 2, 739-748. Doll, R., & Hill, A. B. (1952). A study of the aetiology of carcinoma of the lung. British Medical Journal, 2, 12711286. Doms, M. (1987). Support social et innovation [Social support & innovation]. In S. Moscovici & G. Mugny (Eds.), Psychologie de la conversion. Couset: Delval. Doob, L. W. (1947). The behavior of attitudes. Psychological Review, 54, 135-156. Downing, J. W., Judd, C. M., & Brauer, M. (1992). Effects of repeated expressions on attitude extremity. Journal of Personality and Social Psychology, 63, 17-29. Duck, J. M. (1999). Gender identity and the perceived impact of commercial advertising. Manuscript submitted for publication. Duck, J. M., Hogg, M. A., & Terry, D. J. (1995). Me, us and them: Political identification and the third-person effect in the 1993 Australian federal election. European Journal of Social Psychology, 25, 195-215. Duck, J. M., Hogg, M. A., & Terry, D. J. (1998a). Perceived self-other differences in: The effects of interpersonal and group-based similarity. European Journal of Social Psychology, 28. 1-21. Duck, J. M., Hogg, M. A., & Terry, D. J. (in press). Social identity and perceptions of media persuasion: Are we always less influenced than others? Journal of Applied Social Psychology. Duck, J. M., & Mullin, B. A. (1995). The perceived impact of the mass media: Reconsidering the third person effect. European Journal of Social Psychology, 25, 77-93. Duck, J. M., Terry, D. J., & Hogg, M. A. (1995). The perceived influence of AIDS advertising: Third-person effects in the context of positive media content. Basic and Applied Social Psychology, 17, 305-325.

< previous page

page_301

next page >

< previous page

page_302

next page > Page 302

Duck, J. M., Terry, D. J., & Hogg, M. A. (1998b). Perceptions of a media campaign: The role of social identity and the changing intergroup context. Personality and Social Psychology Bulletin, 24. 3-16. Duran, A., & Trafimow, D. (in press). Cognitive organization of beliefs for and against performing a behavior. Manuscript submitted for publication. Durkheim, E. (1933). The division of labor in society. New York: MacMillian. Durkheim, E. (1982). The rules of sociological method. (W. D. Halls, Trans.). London: Macmillan. Dutton, D. G. (1973). The maverick effect: Increased communicator credibility as a result of abandoning a career. Canadian Journal of Behavioral Sciences, 5, 145-151. Dweck, C. S., Davidson, W., Nelson, S., & Bradley, E. (1978). Sex differences in learned helplessness: II. The contingencies of evaluative feedback in the classroom and III. An experimental analysis. Developmental Psychology, 14, 268-276. Eagly, A. H. (1987). Sex differences in social behavior: A social-role interpretation. Hillsdale, NJ: Lawrence Erlbaum Associates. Eagly, A. H. (1992). Uneven progress: Social psychology and the study of attitudes. Journal of Personality and Social Psychology, 63, 693-710. Eagly, A. H., Ashmore, R. D., Makhijani, M. G., & Longo, L. C. (1991). What is beautiful is good, but . . . : A metaanalytic review of research on the physical attractiveness stereotype. Psychological Bulletin, 110, 109-128. Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. New York: Harcourt Brace Jovanovich. Eagly, A. H., & Chaiken, S. (1995). Attitude strength, attitude structure, and resistance to change. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 413-432). Mahwah, NJ: Lawrence Erlbaum Associates. Eagly, A. H., Mladinic, A., & Otto, S. (1994). Cognitive and affective bases of attitudes toward social groups and social policies. Journal of Experimental Social Psychology, 30, 113-137. Eagly, A. H., Wood, W., & Chaiken, S. (1978). Causal inferences about communicators and their effect on opinion change. Journal of Personality and Social Psychology, 36, 424-435. Echebarría Echabe, A., Fernandez Guede, E., & Gonzalez, J. L. (1994). Social representations and intergroup conflicts: who's smoking here? European Journal of Social Psychology, 24, 339-355. Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology, 59, 202-216. Edwards, W. (1967). Behavioral decision theory. In W. Edwards & A. Tversky (Eds.), Decision making (pp. 65-95). Harmonsworth, England: Penguin. Eiser, J. R. (1982). Attitudes and applied research. In P. Stringer (Ed.), Confronting social issues: some applications of social psychology (Vol. 1). London: Academic Press. Eiser, J. R., & Harding, C. M. (1983). Smoking, seat-belt use and perception of health risks. Addictive behaviors, 8(1), 75-78. Eiser, J. R., Sutton, S. R., & Wober, M. (1979). Smoking, seat belts and beliefs about health. Addictive Behaviors, 4, 331-338. Eiser, J. R., van der Pligt, J., Martin, R., & Sutton, S. R. (1985). Trying to stop smoking: effects of perceived addiction, attributions for failure and expectancy of success. Journal of Behavioral Medicine, 8(4), 321-341. Ellemers, N., Doojse, B., van Knippenberg, A., & Wilke, H. (1992). Status protection in high status minority groups. European Journal of Social Psychology, 22, 123-140. Ellemers, N., van Knippenberg, A., de Vries, N., & Wilke, H. (1988). Social identification and permeability of group boundaries. European Journal of Social Psychology, 18, 497-513.

< previous page

page_302

next page >

< previous page

page_303

next page > Page 303

Ellemers, N., van Knippenberg, A., & Wilke, H. A. (1990). The influence of permeability of group boundaries and stability of group status on strategies of individual mobility and social change. British Journal of Social Psychology, 29, 233-246. Elms, A. C. (1967). Role playing, incentive, and dissonance. Psychological Bulletin, 68, 132-148. Elms, A. C., & Janis, I. L. (1965). Counter-norm attitudes induced by consonant versus dissonant conditions of roleplaying. Journal of Experimental Research in Personality, 1, 50-60. Eroski, (1988). Tabaco: Encuesta (Special Issue No. 15). Elorrio, VI: Eroski Publicaciones. Etzioni, A. (1988). The moral dimension: Toward a new economics. New York: The Free Press. Evans, R. I. (1976). Smoking in children: Developing a social psychological strategy of deterrence. Preventive Medicine, 5, 122-127. Falomir, J. M. (1996). Influence sociale et menace de l'identité: le recours au support social comme stratégie défensive de l'identité [Social influence and identity threat: revert to social support as an identity management strategy]. Unpublished manuscript presented to obtain Diplôme d'Études Supérieures en Psychologie Sociale, University of Geneva, Switzerland. Falomir, J. M., Mugny, G., & Pérez, J. A. (1996). Social Influence and threat to identity: Does the fight against tobacco use require a ban on smoking? International Review of Social Psychology, 2, 95-108. Falomir, J. M., Mugny, G., Sanchez-Mazas, M., Pérez, J. A., & Carrasco, F (1998). Influence et conflit d'identité: de la conformité à l'intériorisation [Influence and identity conflict: from conformity to internalization]. In J. L. Beauvois, R. V. Joule & J. M. Monteil (Eds.), Perspectives cognitives et conduites sociales (Vol. 6). Neuchâtel: Delachaux & Niestlé. Farley, J. U., Lehmann, D. R., & Ryan, M. J. (1981). Generalizing from ''imperfect'' replication. Journal of Business, 54, 597-610. Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition, Vol. 1: Foundations of social behavior (pp. 204-243). New York: Guilford. Fazio, R. H. (1989). On the power and functionality of attitudes: The role of attitude accessibility. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.), Attitude structure and function (pp. 153-179). Hillsdale, NJ: Lawrence Erlbaum Associates. Fazio, R. H. (1990a). Multiple processes by which attitudes guide behavior: The MODE model as an integrative framework. Advances in Experimental Social Psychology, 23, 75-109. San Diego, CA: Academic Press. Fazio, R. H. (1990b). A practical guide to the use of response latency in social psychological research. In M. S. Clark (Ed.), Review of personality and social psychology: Vol. 11. Research methods in personality and social psychology (pp. 74-97). Newbury Park, CA: Sage. Fazio, R. H., Chen, J-M, McDonel, E. C., & Sherman, S. J. (1982). Attitude accessibility, attitude-behavior consistency, and the strength of the object-evaluation association. Journal of Experimental Social Psychology, 18, 339357. Fazio, R. H., Powell, M. C., & Herr, P. M. (1983). Toward a process model of the attitude-behavior relation: Accessing one's attitude upon mere observation of the attitude object. Journal of Personality and Social Psychology, 44, 723-735. Fazio, R. H., Powell, M. C., & Williams, C. J. (1989). The role of attitude accessibility in the attitude-to-behavior process. Journal of Consumer Research, 16, 280-288.

< previous page

page_303

next page >

< previous page

page_304

next page > Page 304

Fazio, R. H., & Williams, C. J. (1986). Attitude accessibility as a moderator of the attitude-perception and attitudebehavior relations: An investigation of the 1984 presidential election. Journal of Personality and Scoail Psychology, 51, 505-514. Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude-behavior consistency. Advances in Experimental Social Psychology, 14, 161-202. Fazio, R. H., Zanna, M. P., & Cooper, J. (1977). Dissonance and self-perception: An integrative view of each theory's proper domain of application. Journal of Experimental Social Psychology, 13, 464-479. Feather, N. T. (1962). Cigarette smoking and lung cancer: A study of cognitive dissonance. Australian Journal of Psychology, 14, 55-64. Festinger, L. (1950a). Experiments in group belongingness. In J. Miller (Ed.), Experiments in social process, (pp. 3635). New York: McGraw-Hill. Festinger, L. (1950b). Informal social communication. Psychological Review, 57, 271-282. Festinger, L. (1953). An analysis of compliant behavior. In M. Sherif & M. O. Wilson (Eds.), Group relations at the crossroads (pp. 232-256). New York: Harper. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140. Festinger, L. (1957). A theory of cognitive dissonance. Palo Alto, CA: Stanford University Press. Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of forced compliance. Journal of Abnormal and Social Psychology, 58, 203-210. Festinger, L., Riecken, H., & Schachter, S. (1956). When prophecy fails. Minneapolis: University of Minnesota Press. Festinger, L., Schachter, S., & Back, K. (1950). Social pressures in informal groups. New York: Harper & Row. Fielding, K. S., & Hogg, M. A. (1997). Social identity, self-categorization, and leadership: A field study of small interactive groups. Group Dynamics: Theory, Research, and Practice, 1, 39-51. Fields, J. M., & Schuman, H. (1976). Public beliefs and the beliefs of the public. Public Opinion Quarterly, 40, 427448. Finlay, K. A., Trafimow, D., & Jones, D. (1997). Predicting health behaviors from attitudes and subjective norms: Between-subjects and within-subjects analyses. Journal of Applied Social Psychology, 27, 2015-2031. Finlay, K. A., Trafimow, D., & Villarreal, A. (1998). [Attitudes and subjective norms in sports behaviors: Betweensubjects and within-subjects analyses]. Submitted data. Fischler, C. (1988). Food, self and identity. Social Science Information, 27(2), 275-292. Fishbein, M. (1967a). Attitude and the prediction of behavior. In M. Fishbein (Ed.), Readings in attitude theory and measurement (pp. 477-492). New York: Wiley. Fishbein, M. (1967b). Readings in attitude theory and measurement. New York: Wiley. Fishbein, M. (1980). Theory of reasoned action: Some applications and implications. In H. Howe & M. Page (Eds.), Nebraska Symposium on Motivation, 1979 (pp. 65-116). Lincoln: University of Nebraska Press. Fishbein, M. (1997). Predicting, understanding, and changing socially relevant behaviors: lessons learned. In McGarty & Haslam (Eds.), The message of social psychology (pp. 77-91). Oxford, England: Blackwell. Fishbein, M., & Ajzen, I. (1974). Attitudes toward objects as predictors of single and multiple behavioral criteria. Psychological Review, 81, 59-74. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley.

< previous page

page_304

next page >

< previous page

page_305

next page > Page 305

Fishbein, M., Middlestadt, S. E., & Trafimow, D. (1993). Social norms for condom use: Implications for HIV prevention interventions of a KABP survey with heterosexuals in the eastern Caribbean. Advances in Consumer Research, 20, 292-296. Fishbein, M., Trafimow, D., Francis, C., Helquist, M., Eustace, M. A., Ooms, M., & Middlestadt, S. E. (1993). AIDS knowledge, attitudes, beliefs, and practices (KABP) in two Caribbean countries: A comparative analysis. Journal of Applied Psychology, 23, 687-702. Fishbein, M., Trafimow, D., Middlestadt, S. E., Helquist, M., Francis, C., & Eustace, M. A. (1995). Using an AIDS KABP Survey to identify determinants of condom use among sexually active adults from St. Vincent and The Grenadines. Journal of Applied Social Psychology, 25, 1-20. Fisher, R, Ury, W., & Patton, B. (1991). Getting to yes. New York: Penguin. Fiske, S. T., & Taylor, S. E. (1991). Social cognition (2nd ed.). New York: McGraw-Hill. Flavell, J. H. (1985). Cognitive development (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall. Fleming, M. A., & Petty, R. E. (1997a). [Social identification with ingroup moderates biased processing of social identity framed persuasive messages]. Unpublished raw data. Ohio State University, Columbus. Fleming, M. A., & Petty, R. E. (1997b). [Social identification with ingroup moderates the effectiveness of an ingroup source as a persuasive cue]. Unpublished raw data. Ohio State University, Columbus. Forgas, J. P., & Fiedler, K. (1996). Us and them: Mood effects on intergroup discrimination. Journal of Personality and Social Psychology, 70(1), 28-40. Fredricks, A. J., & Dossett, D. L. (1983). Attitude-behavior relations: a comparison of the Fishbein-Ajzen and the Bentler-Speckart models. Journal of Personality and Social Psychology, 45, 501-512. Frederickson, H. G. (1969). Role occupancy and attitudes toward labor relations in government. Administrative Science Quarterly, 14, 595-606. Freedman, P. (1967). Race as a factor in persuasion. Journal of Experimental Education, 35, 48-51. French, J. R. P., & Raven, B. (1959). The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150-167). Ann Arbor: University of Michigan. Freud, S. (1921). A general introduction to psychoanalysis. New York: Pocket books. Frideres, J. S., Warner, L. G., & Albrecht, S. L. (1971). The impact of social constraints on the relationship between attitudes and behavior. Social Forces, 50, 102-112. Fried, C. B., & Aronson, E. (1995). Hypocrisy, misattribution, and dissonance reduction. Personality and Social Psychology Bulletin, 21(9), 925-933. Fromm, E. (1997). In On being human. New York: Continuum. (Original work published 1972) Frydman, M. (1987). Les habitudes tabagiques: comment les démystifier. Bruxelles: Labor. Funt, A. (1994). Candidly, Allen Funt: A million smiles later. New York: Barricade Books. Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In J. F Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 61-89). Orlando, FL: Academic Press. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall. Generalitat Valenciana, Direcció General de Salut Pública. (1993). Encuesta de salud de la Comunidad Valenciana. Plan de Salud de la Comunidad Valenciana. Generalitat Valenciana n°2: Conselleria de Sanitat i Consum. Generalitat Valenciana, Direcció General de Salut Pública. (1995). 31 de Maig, Dia Mundial Sense Tabac: El Tabac és més car del que es creu. Generalitat Valenciana: Conselleria de Sanitat i Consum.

< previous page

page_305

next page >

< previous page

page_306

next page > Page 306

Gergen, K. J. (1971). The concept of self. New York: Holt, Rinehart and Winston. Gerver, I., & Bensman, J. (1954). Towards a sociology of expertness. Social Forces, 32, 226-235. Gibbon, P., & Durkin, K. (1995). The third person effect: Social distance and perceived media bias. European Journal of Social Psychology, 25, 597-602. Gibbs, J. C. (1991). Toward an integration of Kohlberg's and Hoffman's moral development theories. Human Development, 34, 88-104. Gibbs, J. P. (1981). The sociology of deviance and social control. In M. Rosenberg & R. H. Turner (Eds.), Social psychology: Sociological perspective (pp. 483-522). New York: Basic Books. Giles, H., & Johnson, P. (1987). Ethnolinguistic identity theory: A social psychological approach to language maintenance. International Journal of Social Language, 68, 69-99. Giner-Sorolla, R., & Chaiken, S. (1994). The causes of hostile media judgments. Journal of Experimental Social Psychology, 30, 165-180. Gleason, P. (1983). Identifying identity: A semantic history. Journal of American History, 69, 910-931. Gliner, M. D., & Pratkanis, A. R. (1997). And when shall a little child lead them ? Towards an altercasting theory of source credibility. Unpublished manuscript. University of California, Santa Cruz. Goethals, G. R. (1972). Consensus and modality in the attribution process: The role of similarity and information. Journal of Personality and Social Psychology, 21, 84-92. Goethals, G. R. (1987). Fabricating and ignoring social reality: self-serving estimates of consensus. In J. H. Olson, C. P. Herman & M. P. Zanna (Eds.), Relative deprivation and social comparison: The Ontario symposium (Vol. 4, pp.). Hillsdale, NJ: Lawrence Erlbaum Associates. Goethals, G. R., Cooper, J., & Naficy, A. (1979). Role of foreseen, foreseeable, and unforeseeable behavioral consequences in the arousal of cognitive dissonance. Journal of Personality and Social Psychology, 37, 1179-1185. Goethals, G. R., & Nelson, R. E. (1973). Similarity in the influence process: The belief-value distinction. Journal of Personality and Social Psychology, 25, 117-122. Goffman, E. (1959). The presentation of self in everyday life. Harmondsworth: Penguin. Goings, K. W. (1994). Mama and Uncle Mose: Black collectibles and American stereotyping. Bloomington: Indiana University Press. Gold, M. (1997). A new outline of social psychology. Washington, D.C.: American Psychological Association. Goldstein, M. J. (1959). The relationship between coping and avoiding behavior and response to fear-arousing propaganda. Journal of Abnormal and Social Psychology, 58, 247-257. Goode, W. J. (1968). Norm commitment and conformity to role-status obligations. American Sociological Review, 33, 246-258. Gorsuch, R. L., & Ortberg, J. (1983). Moral obligations and attitudes: Their relation to behavioral intentions. Journal of Personality and Social Psychology, 44, 1025-1028. Granberg, D., & Holmberg, S. (1990). The intention-behavior relationship among U.S. and Swedish voters. Social Psychology Quarterly, 53(1), 44-54. Greenglass, E. R. (1982). A world of difference: Gender roles in perspective. Toronto: Wiley. Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and attitude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological foundations of attitudes (pp. 147-170). San Diego, CA: Academic Press. Greenwald, A. G. (1969). The open-mindedness of the counterattitudinal role player. Journal of Experimental Social Psychology, 5, 375-388.

< previous page

page_306

next page >

< previous page

page_307

next page > Page 307

Greenwald, A. G. (1970). When does role playing produce attitude change? Toward an answer. Journal of Personality and Social Psychology, 16, 214-219. Greenwald, A. G., Carnot, C. G., Beach, R., & Young, B. (1987). Increasing voting behavior by asking people if they expect to vote. Journal of Applied Psychology, 72, 315-318. Grube, J. W., & Morgan, M. (1990). Attitude-social support interactions: Contingent consistency effects in the prediction of adolescent smoking, drinking, and drug use. Social Psychology Quarterly, 53, 329-339. Grube, J. W., Morgan, M., & McGree, S. T. (1986). Attitudes and normative beliefs as predictors of smoking intentions and behaviors: A test of three models. British Journal of Social Psychology, 25, 81-93. Guillon, M., & Personnaz, B. (1983). Analyse de la dynamique des représentations au cours d'une interaction d'influence avec une minorité et une majorité [Analysis of the dynamics of representations during an interaction of influence with a minority and majority]. Cahiers de Psychologie Cognitive, 3, 65-87. Gunther, A. (1991). What we think others think: Cause and consequence in the third-person effect. Communication Research, 18, 355-372. Gunther, A. (1995). Overrating the X-rating: The third-person perception and support for censorship of pornography. Journal of Communication, 45, 27-38. Gunther, A., & Mundy, P. (1993). Biased optimism and the third-person effect. Journalism Quarterly, 70, 58-67. Gunther, A., & Thorson, E. (1992). Perceived persuasive effects of product commercials and public service announcements: Third-person effects in new domains. Communication Research, 19, 574-596. Guthrie, C. A., & Sparks, P. (1996, July). Moral issues and food choice: An empirical investigation. Poster presented at the 11th General Meeting of the European Association for Experimental Social Psychology, Gmunden, Austria. Guthrie, C. A., & Sparks, P (1997). Perceived behavioural control: Interpretation of the construct and elicitation of its proximal predictors. Manuscript submitted for publication. Hains, S. C., Hogg, M. A., & Duck, J. M. (1997). Self-categorization and leadership: Effects of group prototypicality and leader stereotypicality. Personality and Social Psychology Bulletin, 23, 1087-1099. Haney, C., Banks, C., & Zimbardo, P. G. (1973). Interpersonal dynamics in a simulated prison. International Journal of Criminology and Personology, 1, 69-97. Harkness, A. R., DeBono, K. G., & Borgida, E. (1985). Personal involvement and strategies for making contingency judgments: A stake in the dating game makes a difference. Journal of Personality and Social Psychology, 49, 22-32. Harrison, D. A. (1995). Volunteer motivation and attendance decisions: Competitive theory testing in multiple samples from a homeless shelter. Journal of Applied Psychology, 80, 371-385. Harsanyi, J. C. (1977). Morality and the theory of rational behavior. In A. Sen & B. Williams (Eds.), Utilitarianism and beyond (pp. 39-62). Cambridge, England: Cambridge University Press. Haslam, S. A., Oakes, P. J., McGarty, C., Turner, J. C., & Onorato, R. S. (1995). Contextual changes in the prototypicality of extreme and moderate outgroup members. European Journal of Social Psychology, 25, 509-530. Hass, R. G. (1981). Effects of source characteristics on cognitive responses and persuasion. In R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive responses in persuasion (pp. 141-172). Hillsdale, NJ: Lawrence Erlbaum Associates. Hastie, R. (1984). Causes and effects of causal attribution. Journal of Personality and Social Psychology, 46, 44-56.

< previous page

page_307

next page >

< previous page

page_308

next page > Page 308

Hastie, R., & Park, B. (1986). The relationship between memory and judgment depends on whether the judgment task is memory-based or on-line. Psychological Review, 93, 258-268. Haugtvedt, C. P, & Petty, R. E. (1992). Message order effects in persuasion: An attitude strength perspective. Journal of Consumer Research, 21, 205-218. Haugtvedt, C. P., Petty, R. E., Cacioppo, J. T., & Steidley, T. (1988). Personality and ad effectiveness. Advances in Consumer Research, 15, 209-212. Haugtvedt, C. P., & Wegener, D. T. (1994). Personality and persuasion: Need for cognition moderates the persistence and resistance of attitude changes. Journal of Personality and Social Psychology, 62, 308-319. Haunschild, P. R., Moreland, R. L., & Murrell, A. J. (1994). Sources of resistance to mergers between groups. Journal of Applied Social Psychology, 24, 1150-1178. Heider, F.(1958). The psychology of interpersonal relations. New York: Wiley. Henderson, M. A. (1994). How con games work. New York: Citadel Press. Herek, G. M. (1986). The instrumentality of attitudes: Toward a neofunctional theory. Journal of Social Issues, 42(2), 99-114. Herman, J. B., & Hulin, C. L. (1972). Studying organizational attitudes from individual and organizational frames of reference. Organizational Behavior and Human Performance, 8, 84-108. Higgins, E. T. (1989). Self-discrepancy theory: What patterns of self-beliefs cause people to suffer? Advances in Experimental Social Psychology, 22, 93-136. Higgins, E. T, Rhodewalt, F., & Zanna, M. P. (1979). Dissonance motivation: Its nature, persistence and reinstatement. Journal of Experimental Social Psychology, 15, 16-34. Hilgard, E. R. (1977). Divided consciousness: Multiple controls in human thought and action. New York: Wiley. Hinton, G. E., & Anderson, J. A. (Eds.). (1981). Parallel models of associative memory. Hillsdale, NJ: Lawrence Erlbaum Associates. Hirsch, A., Hill, C., Frossart, M., Tassin, J. P., & Pechabrier, M. (1988). Lutter contre le tabagisme. Paris: La documentation Française. Hoffman, M. L. (1983). Affective and cognitive processes in moral internalization. In E. T. Higgins, D. Ruble, & W. W. Hartup (Eds.), Social cognition and social development: A sociocultural perspective (pp. 236-274). New York: Cambridge University Press. Hogg, M. A. (1992). The social psychology of group cohesiveness: From attraction to social identity. London: Harvester Wheatsheaf. Hogg, M. A. (1993). Group cohesiveness: A critical review and some new directions. European Review of Social Psychology, 4, 85-111. Hogg, M. A. (1996). Intragroup processes, group structure and social identity. In W. P. Robinson (Ed.), Social groups and identities: Developing the legacy of Henri Tajfel (pp. 65-93). Oxford, England: Butterworth-Heinemann. Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations and group processes. London: Routledge. Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem, and the social identity. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 28-47). London: Harvester Wheatsheaf. Hogg, M., & Abrams, D. (1993). Towards a single-process uncertainty-reduction model of social motivation in groups. In M. A. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 173-190). New York: Harvester Wheatsheaf. Hogg, M. A., Cooper-Shaw, L., & Holzworth, D. W. (1993). Group prototypicality and depersonalized attraction in small interactive groups. Personality and Social Psychology Bulletin, 19, 452-465.

< previous page

page_308

next page >

< previous page

page_309

next page > Page 309

Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group solidarity: Effects of group identification and social beliefs on depersonalized attraction. Journal of Personality and Social Psychology, 70, 295-309. Hogg, M. A., & Hardie, E. A. (1991). Social attraction, personal attraction, and self-categorization: A field study. Personality and Social Psychology Bulletin, 17, 175-180. Hogg, M. A., & Hardie, E. A. (1992). Prototypicality, conformity and depersonalized attraction: A self-categorizational analysis of group cohesiveness. British Journal of Social Psychology, 31, 41-56. Hogg, M. A., Terry, D. J., & White, K. M. (1995). A tale of two theories: A critical comparison of identity theory with social identity theory. Social Psychology Quarterly, 58, 255-269. Hogg, M. A., & Turner, J. C. (1987a). Intergroup behavior, self-stereotyping and the salience of social categories. British Journal of Social Psychology, 26, 325-340. Hogg, M. A., & Turner, J. C. (1987b). Social identity and conformity: A theory of referent informational influence. In W. Doise & S. Moscovici (Eds.), Current issues in European social psychology (Vol. 2, pp. 139-182). Cambridge, England: Cambridge University Press. Hogg, M. A., Turner, J. C., & Davidson, B. (1990). Polarized norms and social frames of reference: A test of the selfcategorization theory of group polarization. Basic and Applied Social Psychology, 11, 77-100. Holzner, B., & Robertson, R. (1980). Identity and authority: A problem analysis of processes of identification and authorization. In R. Robertson & B. Holzner (Eds.), Identity and authority: Explorations in the theory of society (pp. 139). Oxford, England: Basil Blackwell. Hom, P. W., & Hulin, C. L. (1981). A competitive test of the prediction of reinlistment by several models. Journal of Applied Psychology, 66, 23-29. Hom, P. W., Katerberg, R., & Hulin, C. L. (1979). Comparative examination of three approaches to the prediction of turnover. Journal of Applied Psychology, 64, 280-290. Homans, G. (1961). Social behavior: Its elementary forms. New York: Harcourt Brace and World. Horai, J., Naccari, N., & Fatoullah, E. (1974). The effects of expertise and physical attractiveness upon opinion agreement and liking. Sociometry, 37, 601-606. Houston, D. A., & Fazio, R. H. (1989). Biased processing as a function of attitude accessibility: Making objective judgments subjectively. Social Cognition, 7, 51-66. Hovland, C. L., Janis, L., & Kelley, H. H. (1953). Communication and persuasion. New Haven, CT: Yale University Press. Howard's 100 days of power. (1996, June 8-9). The Weekend Australian, p. 18. Humphreys, M. S., Bain, J. D., & Pike, A. R. (1989). Different ways to cue a coherent memory system: A theory for episodic, semantic, and procedural tasks. Psychological Review, 96, 208-233. Huston, A. C. (1983). Sex-typing. In M. Hetherington (Ed.), Handbook of child psychology: Vol. 4. Socialization, personality, and social development (pp. 387-467). New York: Wiley. Huston, T. L., & Levinger, G. (1978). Interpersonal attraction and relationships. Annual Review of Psychology, 29, 115156. Hyman, H., & Sheatsley, P. B. (1978). Quelques unes des raisons de l'échec des campagnes d'information [Social psychology: Basic English and American texts]. In A. Lévy (Ed.), Psychologie sociale. Textes fondamentaux anglais et américains (Vol. 1). Paris: Dunod. Infante, D. A. (1978). Similarity between advocate and receiver: The role of instrumentality. Central States Speech Journal, 29, 187-193.

< previous page

page_309

next page >

< previous page

page_310

next page > Page 310

Innes, J. M., & Zeitz, H. (1988). The public's view of the impact of the mass media: A test of the 'third person' effect. European Journal of Social Psychology, 18, 457-463. Institute for Propaganda Analysis. (1937). How to detect propaganda. Propaganda Analysis, 1(2), 5-7. Jaccard, J. J. (1981). Attitudes and behavior: Implications of attitudes toward behavioral alternatives. Journal of Experimental Social Psychology, 17, 286-307. James, W. (1983). The principles of psychology. Cambridge, MA: Harvard University Press. (Original work published 1890) Jamieson, D. W., & Zanna, M. P. (1989). Need for structure in attitude formation and persuasion. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.), Attitude structure and function (pp. 383-406). Hillsdale, NJ: Lawrence Erlbaum Associates. Janis, I. L., & Feshbach, S. (1953). Effects of fear-arousing communications. The Journal of Abnormal and Social Psychology, 48(1), 78-92. Janis, 1. L., & King, B. T. (1954). The influence of role playing on opinion change. Journal of Abnormal and Social Psychology, 49, 211-218. Janis, I. L., & Mann, L. (1965). Effectiveness of emotional role playing in modifying smoking habits and attitudes. Journal of Experimental Research in Personality, 1, 17-27. Janis, I. L., & Mann, L. (1977). Decision making: A psychological analysis of conflict, choice, and commitment. New York: Free Press. Janz, N. K., & Becker, M. H. (1984). The health belief model: A decade later. Health Education Quarterly, 11, 1-47. Jensen, R. E., & Moore, S. G. (1977). The effect of attribute statements on cooperativeness and competitiveness in school-age boys. Child Development, 48, 305-307. Jermott, III, J. B., Ditto, P. H., & Croyle, R. T. (1986). Judging health status: effects of perceived prevalence and personal relevance. Journal of Personality and Social Psychology, 50, 899-905. Johnson, L., O'Malley, P., & Bachman, J. (1989). Drug use, drinking and smoking: national survey results from high school, college, and young adult populations, 1975-1988. Washington, DC: National Institute on Drug Abuse. Johnston, V. S., & Oliver-Rodriguez, J. C. (1997). Facial beauty and the late positive component of event related potentials. Journal of Sex Research, 34, 188-198. Jones, E. E., Davis, K. E., & Gergen, K. J. (1961). Role playing variations and their informational value for person perception. Journal of Abnormal and Social Psychology, 63, 302-310. Jones, E. E., & Pittman, T. (1982). Toward a general theory of strategic self-presentation. In J. Suls (Ed.), Psychological perspectives on the self (pp. 231-262). Hillsdale, NJ: Lawrence Erlbaum Associates. Kaplan, M. F., & Miller, L. E. (1978). Reducing the effects of juror bias. Journal of Personality and Social Psychology, 36, 1443-1455. Karsenty, S., & Hirsch, A. (1992). Une lutte contre le tabagisme fondée sur l'expertise scientifique. In K. Slama, S. Karsenty, & A. Hirsch (Eds.), La lutte contre le tabagisme est-elle efficace? Paris: INSERM (Institut National de la Santé et de la Recherche Médicale) and CFES (Comité Français d'Education pour la Santé). Kashima, Y., & Gallois, C. (1993). The theory of reasoned action and problem-focused research. In D. J. Terry, C. Gallois, & M. McCamish (Eds.), The theory of reasoned action: Its application to AIDS-preventive behaviour (pp. 207226). Oxford: Pergamon. Kashima, Y., Gallois, C., & McCamish, M. (1993). The theory of reasoned action and cooperative behaviour: It takes two to use a condom. British Journal of Social Psychology, 32, 227-239. Kashima, Y., & Kashima, E. S. (1988). Individual differences in the prediction of behavioral intentions. Journal of Social Psychology, 128, 711-720.

< previous page

page_310

next page >

< previous page

page_311

next page > Page 311

Kashima, Y., & Kerekes, A. R. Z. (1994). A distributed memory model of averaging phenomena in person impression formation. Journal of Experimental Social Psychology, 30, 407-455. Kashima, Y., Woolcock, J., & King, D. (1998). The dynamics of group impression formation: The tensor product model of exemplar-based social category learning. In S. J. Read & L. C. Miller (Eds.), Connectionist models of social reasoning and behavior (pp. 71-109). Mahwah, NJ: Lawrence Erlbaum Associates. Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly, 24, 163-204. Katz, D., & Allport, F. H. (1928). Student attitudes: A report of the Syracuse University research study. Syracuse, NY: Craftsman Press. Katz, D., & Schanck, R. L. (1938). Social psychology. New York: Wiley. Katz, D., & Stotland, E. (1959). A preliminary statement to a theory of attitude structure and change. In S. Koch (Ed.), Psychology: A study of a science (Vol. 3, pp. 423-475). New York: McGraw-Hill. Kauffman, K. (1981). Prison officer attitudes and perceptions of attitudes. Journal of Research in Crime Delinquency, 18, 272-294. Kelley, H. H. (1952). Two functions of reference groups. In G. E. Swanson, T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psychology (2nd ed., pp. 410-414). New York: Holt. Kelley, H. H. (1955). Salience of membership and resistance to change of group-anchored attitudes. Human Relations, 3, 275-289. Kelley, H. H. (1967).Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium on motivation (Vol. 15, pp. 192-238). Lincoln: University of Nebraska Press. Kelley, H. H., & Volkart, E. H. (1952). The resistance to change of group-anchored attitudes. American Sociological Review, 17, 453-465. Kelly, C., & Breinlinger, S. (1995). Attitudes, intentions and behavior: A study of women's participation in collective action. Journal of Applied Social Psychology, 25(16), 1431-1446. Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2, 51-60. Kelman, H. C. (1961). Processes of attitude change. Public Opinion Quarterly, 25, 57-78. Kern-Foxworth, M. (1994). Aunt Jemina, Uncle Ben, and Rastus: Blacks in advertising, yesterday, today, and tomorrow. Westport, CT: Praeger. Kihlstrom, J. F., & Shor, R. E. (1978). Recall and cognition during posthypnotic amnesia. International Journal of Clinical and Experimental Hypnosis, 26, 330-349. King, B. T., & Janis, I. L. (1959). Comparison of the effectiveness of improvised versus non-improvised role-playing in producing opinion changes. Human Relations, 9, 177-186 King, S. W., & Sereno, K. K, (1973). Attitude change as a function of degree and type of interpersonal similarity and message type. Western Speech, 37, 218-232. Klapp, O. E. (1949). The fool as a social type. American Journal of Sociology, 55, 157-162. Klassen, A. D., Williams, C. J., & Levitt, E. E. (1989). Sex and morality in the U. S.: An empirical enquiry under the auspices of the Kinsey Institute. Wesleyan, CT: Wesleyan University Press. Klofas, J., & Toch, H. (1982). The guard subculture myth. Journal of Research in Crime and Delinquency, 19, 238254. Koeske, G. F., & Crano, W. D. (1968). The effect of congruous and incongruous source-statement combinations upon the judged credibility of a communication. Journal of Experimental Social Psychology, 4, 384-399. Kohlberg, L. (1984). The psychology of moral development: Essays on moral development (Vol. 2). San Francisco, CA: Harper & Row.

< previous page

page_311

next page >

< previous page

page_312

next page > Page 312

Korte, C. (1972). Pluralistic ignorance about student radicalism. Sociometry, 35, 576-587. Kraut, R. E. (1973). The effects of social labeling on giving to charity. Journal of Experimental Social Psychology, 9, 551-562. Kristiansen, C. M., Harding, C. M., & Eiser, J. R. (1983). Beliefs about the relationship between smoking and causes of death. Basic and Applied Social Psychology, 4(3), 253-261. Krosnick, J. A. (1988). Attitude importance and attitude change. Journal of Experimental Social Psychology, 24, 240255. Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A parallel constraint satisfaction theory. Psychological Review, 103, 284-308. Kurland, N. B. (1996). Sales agents and clients: Ethics, incentives, and a modified theory of planned behavior. Human Relations, 49, 51-74. Lalonde, R. N. (1992). The dynamics of group differentiation in the face of defeat. Personality and Social Psychology Bulletin, 28, 336-342. Lalonde, R., Moghaddam, F M., & Taylor, B. M. (1987). The process of group differentiation in a dynamic intergroup setting. Journal of Social Psychology, 127, 273-287. Lane, I. M., Mathews, R. C., & Prestholdt, P. H. (1988). Determinants of nurses' intentions to leave their profession. Journal of Organizational Behavior, 9, 367-372. Lane, I. M., Mathews, R. C., & Prestholdt, P. H. (1990). Educational background, marital status, moral obligation, and the nurse's decision to resign from her hospital. Journal of Applied Social Psychology, 20, 1432-1443. LaPiere, R. T. (1934). Attitudes vs. actions. Social Forces, 13, 230-237. Lasorsa, D. L. (1989). Real and perceived effects of 'Amerika'. Journalism Quarterly, 66, 373-378. Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help? Englewood Cliffs, NJ: PrenticeHall. Lavine, H., & Snyder, M. (1996). Cognitive processing and the functional matching effect in persuasion: The mediating role of subjective perceptions of message quality. Journal of Experimental Social Psychology, 32, 580-604. Lawler, E. J. (1983). Cooptation and threats as "divide and rule" tactics. Social Psychology Quarterly, 46, 89-98. Lawler, E. J., & Thompson, M. E. (1979). Subordinate response to a leader's cooptation strategy as a function of the type of coalition power. Representative Research in Social Psychology, 9, 69-80. Lawler, E. J., Youngs, G. A., & Lesh, M. D. (1978). Cooptation and coalition mobilization. Journal of Applied Social Psychology, 8, 199-214. Leary, M. R. (1996). Self-presentation. Boulder, CO: Westview Press. Leavitt, C., & Kaigler-Evans, K. (1975). Mere similarity versus information processing: An exploration of source and message interaction. Communication Research, 2, 300-306. Lee, Y. T., & Ottati, V. (1995). Perceived ingroup homogeneity as a function of group membership salience and stereotype threat. Personality and Social Psychology Bulletin, 21, 610-619. Lemaine, G. (1974). Social differentiation and social originality. European Journal of Social Psychology, 4, 17-52. Leventhal, H., & Cleary, P. D.. (1980). The smoking problem: a review of the research and theory in behavioral risk modification. Psychological Bulletin, 88(2), 370-405. Levine, J. M., & Valle, R. S. (1975). The convert as a credible communicator. Social Behaviour and Personality, 3, 8190. Lewin, K. (1978). Décision de groupe et changement social [Group decision and social change]. In A. Lévy (Ed.), Psychologie sociale. Textes fondamentaux anglais et américains (Vol. 1). Paris: Dunod.

< previous page

page_312

next page >

< previous page

page_313

next page > Page 313

Lichtenstein, E., Biglan, A., Glasgow, R. E., Severson, H., & Ary, D. (1990). The tobacco use research program at Oregon Research Institute. British Journal of Addiction, 85, 715-724. Lieberman, S. (1956). The effects in changes in role on the attitudes of role occupants. Human Relations, 9, 385-402. Liska, A. (1984). A critical examination of the causal structure of the Fishbein and Ajzen attitude-behavior model. Social Psychology Quarterly, 47, 61-74. Loomes, G., & Sugden, R. (1982). Regret theory: An alternative theory of rational choice under uncertainty. Economic Journal, 92, 805-825. Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098-2109. Lorenzi-Cioldi, F. (1988). Self-other discrimination and social categorization. Revenue Internationale de Psychologie Sociale, 1(2), 239-256. Lorenzi-Cioldi, F. (1993). They all look alike, but so do we . . . sometimes: Perceptions of in-group and out-group homogeneity as a function of sex and context. British Journal of Social Psychology, 32, 111-124. Losch, M. E., & Cacioppo, J. (1990). Cognitive dissonance may enhance sympathetic tonus, but attitudes are changed to reduce negative affect rather than arousal. Journal of Experimental Social Psychology, 26, 289-304. Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302-318. Lumsdaine, A., & Janis, 1. (1953). Resistance to ''counterpropaganda'' produced by one-sided and two-sided propaganda presentations. Public Opinion Quarterly, 17, 311-318. Lynch, B. S., & Bonnie, R. J. (Eds.). (1994). Growing up tobacco free: Preventing nicotine addiction in children and youth. Washington, DC: National Academy Press. MacCoun, R. J., & Kerr, N. L. (1988). Asymmetric influence on mock jury deliberation: Jurors' bias for leniency. Journal of Personality and Social Psychology, 54, 21-33. Mackie, D. M. (1986). Social identification effects in group polarization. Journal of Personality and Social Psychology, 50, 720-728. Mackie, D. M., Gastardo-Conaco, M. C., & Skelly, J. J. (1992). Knowledge of the advocated position and the processing of in-group and out-group persuasive messages. Personality and Social Psychology Bulletin, 18, 145-151. Mackie, D. M., Worth, L. T., & Asuncion, A. G. (1990). Processing of persuasive in-group messages. Journal of Personality and Social Psychology, 58, 812-822. Maddux, J. E., & Rogers, R. W. (1980). Effects of source expertness, physical attractiveness, and supporting arguments on persuasion. A case of brains over beauty. Journal of Personality and Social Psychology, 39, 235-244. Maibach, E., & Parrott, R. L. (Eds.). (1995). Designing health messages. London: Sage. Maier, N. R. F. (1952). Principles of human relations. New York: Wiley. Manstead, A. S. R. (1996). Attitudes and behavior. In G. R. Semin & K. Fiedler (Eds.), Applied social psychology (pp. 3-29). London: Sage. Manstead, A. S. R., & Parker, D. (1995). Evaluating and extending the theory of planned behavior. European Review of Social Psychology, 6, 69-95. Manstead, A. S. R., Parker, D., & Stradling, S. G. (1997). Adding affect to the theory of planned behavior. Unpublished manuscript, University of Amsterdam. March, J. G. (1978). Bounded rationality, ambiguity, and the engineering of choice. Bell Journal of Economics, 9, 587608. Markus, H. (1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35, 63-78.

< previous page

page_313

next page >

< previous page

page_314

next page > Page 314

Markus, H., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224-252. Markus, H., & Zajonc, R. B. (1985). The cognitive perspective in social psychology. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 1, pp. 137-229). Reading, MA: Addison-Wesley. Marwell, G., & Hage, J. (1970). The organization of role-relationships: A systematic description. American Sociological Review, 35, 884-900. Maslach, C. (1974). Social and personal bases of individuation. Journal of Personality and Social Psychology, 29, 411425. Mausner, B., & Platt, E. S. (1971). Smoking: A behavioral analysis. New York: Pergamon. McArthur, L. A., Kiesler, C. A., & Cook, B. P. (1969). Acting on an attitude as a function of self-percept and inequity. Journal of Personality and Social Psychology, 12, 295-302. McGarty, C., Haslam, S. A., Hutchinson, K. J., & Turner, J. C. (1994). The effects of salient group memberships on persuasion. Small Group Research, 25, 267-293. McGarty, C., Turner, J. C., Hogg, M. A., David, B., & Wetherell, M. S. (1992). Group polarization as conformity to the most prototypical group member. British Journal of Social Psychology, 31, 1-20. McGinnies, E., & Ward, C. D. (1980). Better liked than right: Trustworthiness and expertise as factors in credibility. Personality and Social Psychology Bulletin, 6, 467-472. McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & Aronson E. (Eds.), The handbook of social psychology (3rd ed.). New York: Random House. McGuire, W. J. (1986). The vicissitudes of attitudes and similar representational constructs in twentieth century psychology. European Journal of Social Psychology, 16,(2) 89-130. McGuire, W. J., & Papageorgis, D. (1961). The relative efficacy of various types of prior belief-defense in producing immunity against persuasion. Journal of Abnormal and Social Psychology, 62, 327-337. McGuire, W. J., & Papageorgis, D. (1962). Effectiveness of forewarning in developing resistance to persuasion. Public Opinion Quarterly, 26, 24-34. Merton, R. K. (1957a). Social theory and social structure. Glencoe, IL: Free Press. Merton, R. K. (1957b). The role-set: Problems in sociological theory. British Journal of Sociology, 8, 106-120. Merton, R. (1968). Social theory and social structure. New York: The Free Press. Merton, R. (1976). Sociological ambivalence and other essays. New York: The Free Press. Milgram, S. (1974). Obedience to authority: An experimental view. New York: Harper & Row. Millar, M. G., & Tesser, A. (1986). Effects of affective and cognitive focus on the attitude-behavior relation. Journal of Personality and Social Psychology, 51, 270-276. Miller, D. T, & McFarland, C. (1987). Pluralistic ignorance: When similarity is interpreted as dissimilarity. Journal of Personality and Social Psychology, 53, 298-305. Miller, D. T., & McFarland, C. (1991). When social comparison goes awry: The case of pluralistic ignorance. In J. Suls & T. Wills (Eds.), Social comparison: Contemporary theory and research (pp. 287-313). Hillsdale, NJ: Lawrence Erlbaum Associates. Miller, D. T, & Prentice, D. A. (1994). Collective errors and errors about the collective. Personality and Social Psychology Bulletin, 20, 541-550. Miller, D. T, & Prentice, D. A. (1996). The construction of social norms and standards. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 799-829). New York: Guilford. Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the structure of behavior. London: Holt, Rinehart & Winston.

< previous page

page_314

next page >

< previous page

page_315

next page > Page 315

Miller, L. E., & Grush, J. E. (1986). Individual differences in attitudinal versus normative determination of behavior. Journal of Experimental Social Psychology, 22, 190-202. Miller, R. L., Brickman, P., & Bolen, D. (1975). Attribution versus persuasion as a means of modifying behavior. Journal of Personality and Social Psychology, 31, 430-441. Mills, J., & Aronson, E. (1965). Opinion change as a function of the communicator's attractiveness and desire to influence. Journal of Personality and Social Psychology, 1, 173-177. Mills, J., & Harvey, J. (1972). Opinion change as a function of when information about the communicator is received and whether he is attractive or expert. Journal of Personality and Social Psychology, 21, 52-55. Mills, J., & Kimble, C. E. (1973). Opinion change as a function of perceived similarity of the communicator and subjectivity of the issue. Bulletin of the Psychonomic Society, 2, 35-36. Miniard, P. W., & Cohen, J. B. (1981). An examination of the Fishbein behavioral intentions model's concept and measures. Journal of Experimental Social Psychology, 17, 309-329. Ministerio Español de Sanidad y Consumo. (1993). Encuesta Nacional de Salud de España. Ministerio de Sanidad y Consumo: Secretaría General Técnica, Centro de Publicaciones [Press paper]. Mischel, W. (1977). The interaction of person and situation. In D. Magnuson & N. S. Endler (Eds.), Personality at the crossroads: Current issues in interactional psychology (pp. 333-352). Hillsdale, NJ: Lawrence Erlbaum Associates. Mitchell, T. R., & Beach, L. R. (1990). " . . . Do I love thee? Let me count . . . " Toward an understanding of intuitive and automatic decision-making. Organizational Behavior and Human Decision Processes, 47, 1-20. Moreland, R. L., Hogg, M. A., & Hains, S. C. (1994). Back to the future: Social psychological research on groups. Journal of Experimental Social Psychology, 30, 527-555. Morgan, K., & Duck, J. M. (1996, August). Third-context of media censorship. Paper presented at the 26th International Congress of Psychology, Montréal, Canada. Moscovici, S. (1976). Social influence and social change. London, England: Academic Press. Moscovici, S. (1980). Towards a theory of conversion behavior. Advances in Experimental Social Psychology, 13, 209239. Moscovici, S. (1985). Innovation and minority influence. In S. Moscovici, G. Mugny & E. Van Avermaet (Eds.), Perspectives on minority influence. Cambridge, England: Cambridge University Press. Moscovici, S., & Faucheux, C. (1972). Social influence, conformity bias and the study of active minorities. Advances in Experimental Social Psychology, 6, 149-202. Moscovici, S., & Lage, E. (1976). Studies in social influence: III. Majority and minority influence in a group. European Journal of Social Psychology, 6, 149-174. Moscovici, S., Lage, E., & Naffrechoux, M. (1969). Influence of a consistent minority on the responses of a majority in a color perception task. Sociometry, 32, 365-379. Moscovici, S., & Neve, P. (1972). Studies in social influence: I. Those absent are in the right; convergence and polarization of answers in the course of social interaction. European Journal of Social Psychology, 2, 201-214. Motz, A. B. (1952). The role conception inventory: A tool for research in social psychology. American Sociological Review, 17, 465-471. Mugny, G., & Pérez, J. A. (1991). The social psychology of minority influence. Cambridge, England: Cambridge University Press. Mulligan, R. (Director). (1962). To kill a mockingbird [Film]. Universal City, CA: MCA Home Video.

< previous page

page_315

next page >

< previous page

page_316

next page > Page 316

Mummundey, A., & Mummundey, H. D. (1982). Selbstkonsistenz vs. Gruppenkonformitat bei Selbstufmerksamkeit: Die rolle des einflusses von ingroup vs. outgroup [Self-consistency versus group conformity under self-attention conditions: The role of ingroup versus outgroup influence]. Zeitschrift fur Sozialpsychologie, 13, 88-96. Mutz, D. C. (1989). The influence of perceptions of mass influence: Third person effects and the public expression of opinions. International Journal of Public Opinion research, 1, 3-23. Myers, D. G. (1982). Polarizing effects of social interaction. In H. Brandstatter, J. H. Davis, & G. Stocker-Kreichgauer (Eds.), Group decision-making. London: Academic Press. Mynatt, C., & Sherman, S. J. (1975). Responsibility attribution in groups and individuals: A direct test of the diffusion of responsibility hypothesis. Journal of Personality and Social Psychology, 32, 1111-1118. Nadler, A., & Fisher, J. D. (1986). The role of threat to self-esteem and perceived control in recipient reaction to help: Theory development and empirical validation. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 19, pp. 81-122). Orlando, FL: Academic Press. Newcomb, T. M. (1943). Personality and social change: Attitude formation in a student community. New York: Holt, Rinehart & Winston. Noelle-Neumann, E. (1974). The spiral of silence: A theory of public opinion. Journal of Communication, 24, 43-51. Noelle-Neumann, E. (1984). The spiral of silence: A response. In K. R. Sanders, L. L. Kaid, & E. Nimmo (Eds.), Political Communication Yearbook 1984 (pp. 66-94). Carbondale: Southern Illinois University Press. Norman, D. A. (1981). Categorization of action slips. Psychological Review, 88, 1-15. Norman, R. (1976). When what is said is important: A comparison of expert and attractive sources. Journal of Experimental Social Psychology, 12, 294-300. Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill. Nye, J. L., & Bower, A. M. (Eds.). (1996). What's social about social cognition: Research on socially shared cognition in small groups. Thousand Oaks, CA: Sage. Oaker, G., & Brown, R. (1986). Intergroup relations in a hospital setting: A further test of social identity theory. Human Relations, 39, 737-778. Office of Population Censures and Surveys. (1990). OPC Monitor Government Statistical Service, London. O'Gorman, H. J. (1975). Pluralistic ignorance and white estimates of white support for racial segregation. Public Opinion Quarterly, 39, 313-330. O'Gorman, H. J. (1986). The discovery of pluralistic ignorance: An ironic lesson. Journal of History of the Behavioral Sciences, 22, 333-347. O'Gorman, H. J., & Garry, S. L. (1976). Pluralistic ignorancea replication and extension. Public Opinion Quarterly, 40, 449-458. O'Keefe, E. S. C., & Hyde, J. S. (1983). The development of occupational sex-role stereotypes: The effects of gender stability and age. Sex Roles, 9, 481-492. Oliver, R. L., & Bearden, W. O. (1985). Crossover effects in the theory of reasoned action: A moderating influence attempt. Journal of Consumer Research, 12, 324-340. O'Rourke, T, Smith, B. J., & Nolte, A. E. (1983). Health risk attitudes, beliefs and behaviors of students grades 7-12. Journal of School Health, 54(5), 210-214. Packard, J. S., & Willower, D. J. (1972). Pluralistic ignorance and pupil control ideology. Journal of Education Administration, 10, 78-87. Pallak, S. R. (1983). Salience of a communicator's physical attractiveness and persuasion: A heuristic versus systematic processing interpretation. Social Cognition, 2, 158-170. Pallak, S. R., Murroni, E., & Koch, J. (1983). Communicator attractiveness and expertise, emotional versus rational appeals, and persuasion: A heuristic versus systematic processing interpretation. Social Cognition, 2, 122-141.

< previous page

page_316

next page >

< previous page

page_317

next page > Page 317

Parker, D., Manstead, A. S. R., & Stradling, S. G. (1995). Extending the theory of planned behaviour: The role of personal norm. British Journal of Social Psychology, 34, 127-137. Parsons, T. (1949). The structure of social action. New York: The Free Press. Pérez, J. A., Falomir, J. M., & Mugny, G. (1995). Internalisation of conflict and attitude change. European Journal of Social Psychology, 25, 117-124. Pérez, J. A., & Mugny, G. (1990a). Changement d'attitude, crédibilité et influence minoritaire: interdépendance et indépendance de la comparaison sociale [Attitude change, credibility & minority influence interdependence and independence of social comparison]. Revue Suisse de Psychologie, 49(3), 150-158. Pérez, J. A., & Mugny, G. (1990b). Minority influence, manifest discrimination and latent influence. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances. London: Harverster Wheatsheaf. Pérez, J. A., & Mugny, G. (1992). Social impact of experts and minorities, and smoking cessation. In M. von Cranach, W. Doise & G. Mugny, (Eds.), Social representations and the social bases of knowledge. Lewinston, Hogrefe & Huber. Pérez, J. A., Mugny, G., Roux, P., & Butera, F. (1991). Influences via la comparaison sociale, influences via la validation Influence by social comparison, influence through validation]. In J. L. Beauvois, R. V. Joule, & J. M. Monteil (Eds.), Perspectives cognitives et conduites sociales (Vol. 3, pp. 184-188). Cousset: Delval. Perloff, L. S., & Fetzer, B. K. (1986). Self-other judgments and perceived vulnerability to victimization. Journal of Personality and Social Psychology, 50, 502-510. Perloff, R. M. (1989). Ego-involvement and the third person effect of televised news coverage. Communication Research, 16, 236-262. Perloff, R. M. (1993). Third-person effect research 1983-1992: A review and synthesis. International Journal of Public Opinion Research, 5, 167-184. Pervin, L. A., & Yatko, R. J. (1965). Cigarette smoking and alternative methods of reducing dissonance. Journal of Personality and Social Psychology, 2, 30-36. Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport's cognitive analysis of prejudice. Personality and Social Psychology Bulletin, 5, 461-476. Petty, R. E. (1996). The evolution of theory and research in social psychology: From single to multiple effect and process models. In C. McGarty & S. A. Haslam (Eds.), The message of social psychology: Perspectives on mind in society (pp. 268-290). Oxford, England: Blackwell. Petty, R. E., & Cacioppo, J. (1979a). Effects of forewarning of persuasive intent and involvement on cognitive response and persuasion. Personality and Social Psychology Bulletin, 5, 173-176. Petty, R. E., & Cacioppo, J. (1979b). Issue involvement can increase or decrease persuasion by enhancing messagerelevant cognitive responses. Journal of Personality and Social Psychology, 37, 1915-1926. Petty, R. E., & Cacioppo, J. T. (1980). Involvement and persuasion: Tradition versus integration. Psychological Bulletin, 107, 367-374. Petty, R. E., & Cacioppo, J. (1981 a). Attitudes and persuasion: Classic and contemporary approaches. Dubuque, IA: William C Brown. Petty, R. E., & Cacioppo, J. T. (1981 b). Issue involvement as a moderator of the effects on attitude of advertising content and context. Advances in Consumer Research, 8, 20-24. Petty, R. E., & Cacioppo, J. (1984). The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion. Journal of Personality and Social Psychology, 46, 69-81. Petty, R. E., &, J. T. (1986a). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer-Verlag.

< previous page

page_317

next page >

< previous page

page_318

next page > Page 318

Petty, R. E., & Cacioppo, J. T (1986b). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123-205). New York: Academic Press. Petty, R. E., & Cacioppo, J. T. (1990). Involvement and persuasion: Tradition versus integration. Psychological Bulletin Bulletin, 107, 367-374. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. Advances in Experimental Social Psychology, 19, 123-205. Petty, R. E., Cacioppo, J. T, & Goldman, R. (1981). Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology, 41, 847-855. Petty, R. E., Cacioppo, J. T., & Haugtvedt, C. (1992). Involvement and persuasion: An appreciative look at the Sherifs' contribution to the study of self-relevance and attitude change. In D. Granberg & G. Sarup (Eds.), Social judgment and intergroup relations: Essays in honor of Muzafer Sherif (pp. 147-174). New York: Springer-Verlag. Petty, R. E., Cacioppo, J. T, & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research, 10, 134-148. Petty, R. E., Harkins, S. G., & Williams, K. D. (1980). The effects of group diffusion of cognitive effort on attitudes: An information processing view. Journal of Personality and Social Psychology, 38, 81-92. Petty, R. E., Haugtvedt, C. P., & Smith, S. M. (1995). Elaboration as a determinant of attitude strength: Creating attitudes that are persistent, resistant, and predictive of behavior. In R. E. Petty R. & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 93-130). Hillsdale, NJ: Lawrence Erlbaum Associates. Petty, R. E., & Krosnick, J. A. (Eds.) (1995). Attitude strength: Antecedents and consequences. Hillsdale, NJ: Lawrence Erlbaum Associates. Petty, R. E., Priester, J. R., & Wegener, D. T. (1994). Cognitive processes in attitude change. In R. Wyer & T Srull (Eds.), Handbook of Social Cognition (Vol. 2,2nd ed., pp. 69-142). Hillsdale, NJ: Lawrence Erlbaum Associates. Petty, R. E., Schumann, D. W., Richman, S. A., & Strathman, A. J. (1993). Positive mood and persuasion: Different roles for affect under high- and low-elaboration conditions. Journal of Personality and Social Psychology, 64, 5-20. Petty, R. E., & Wegener, D. T. (1997). Attitude change: Multiple roles for persuasion variables. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), Handbook of Social Psychology (4th ed., pp. 323-390). New York: McGraw-Hill. Petty, R. E., & Wegener, D. T. (1999). The Elaboration Likelihood Model: Current status and controversies. In S. Chaiken & Y. Trope (Eds.), Dual process theories in social psychology. New York: Guilford Press. Petty, R. E., & Wegener, D. T. (1998). Matching versus mismatching attitude functions: Implications for scrutiny of persuasive messages. Personality and Social Psychology Bulletin, 24, 227-240. Petty, R. E., Wells, G. L., & Brock, T. C. (1976). Distraction can enhance or reduce yielding to propaganda. Journal of Personality and Social Psychology, 34, 874-884. Pfeffer, J. (1972). Size and composition of corporate boards of directors: The organization and its environment. Administrative Science Quarterly, 17, 218-228. Pfeffer, J. (1974). Cooptation and the composition of electrical utility boards of directors. Pacific Sociological Review, 17, 333-363. Pfeffer, J. (1981). Power in organizations. Cambridge, MA: Ballinger. Pfister, H-R., & Böhm, G. (1992). The function of concrete emotions in rational decision making. Acta Psychologica, 80, 105-116.

< previous page

page_318

next page >

< previous page

page_319

next page > Page 319

Pike, A. R. (1984). Comparison of convolution and matrix distributed memory systems for associative recall and recognition. Psychological Review, 91, 281-294. Piliavin, J. A., & Callero, P. L. (1991). Giving blood. Baltimore, MD: Johns Hopkins University Press. Plous, S., & Williams, T. (1995). Racial stereotypes from the days of slavery: A continuing legacy. Journal of Applied Social Psychology, 25, 795-817. Plutzer, E., & Zipp, J. F. (1996). Identity politics, partisanship, and voting for women candidates. Public Opinion Quarterly, 60, 30-57. Pomazal, R., & Jaccard, J. J. (1976). An informational approach to altruistic behavior. Journal of Personality and Social Psychology, 33, 317-326. Pomerantz, E. M., Chaiken, S., & Tordesillas, R. (1995). Attitude strength and resistance processes. Journal of Personality and Social Psychology, 69, 408-419. Porter, L. W., & Lawler, E. E. (1965). Properties of organizational structure in relation to job attitudes and job behavior. Psychological Bulletin, 64, 23-51. Posavac, S. S., Sanbonmatsu, D. M., & Fazio, R. H. (1997). Considering the best choice: Effects of the salience and accessibility of alternatives on attitude-decision consistency. Journal of Personality and Social Psychology, 72, 253261. Powell, M. C., & Fazio, R. H. (1984). Attitude accessibility as a function of repeated attitudinal expression. Personality and Social Psychology Bulletin, 10, 139-148. Pratkanis, A. R., & Aronson, E. (in press). Age of propaganda: The everyday use and abuse of persuasion (rev. ed.). New York: Freeman. Pratkanis, A. R., Breckler, S. J., & Greenwald, A. G. (Eds.); (1989). Attitude structure and function. Hillsdale, NJ: Lawrence Erlbaum Associates. Prelec, D., & Herrnstein, R. (1991). Preferences or principles: Alternative guidelines for choice. In R. J. Zeckhauser (Ed.), Strategy and choice (pp. 319-340). Cambridge, MA: MIT. Press. Prentice, D. A., Girgus, J. S., & Miller, D. T. (1997). Pluralistic ignorance and children's sex-role stereotypes. Unpublished manuscript, Princeton University, Princeton, NJ. Prentice, D. A., & Miller, D. T. (1993). Pluralistic ignorance and alcohol use on campus: Some consequences of misperceiving the social norm. Journal of Personality and Social Psychology, 64, 243-256. Prentice, D. A., & Miller, D. T. (1996). Pluralistic ignorance and the perpetuation of social norms and unwitting actors. Advances in Experimental Social Psychology, 28, 161-210. Prestholdt, P. H., Lane, I. M., & Mathews, R. C. (1987). Nurse turnover as reasoned action: Development of a process model. Journal of Applied Psychology, 72, 221-227. Price, R. H., & Bouffard, D. L. (1974). Behavioral appropriateness and situational constraint as dimensions of social behavior. Journal of Personality and Social Psychology, 30, 579-586. Price, V. (1989). Social identification and public opinion: Effects of communicating group conflict. Public Opinion Quarterly, 53, 197-224. Puckett, J., Petty, R. E., Cacioppo, J. T., & Fisher, D. (1983). The relative impact of age and attractiveness stereotypes on persuasion. Journal of Gerontology, 38, 340-343. Raats, M. M., Shepherd, R., & Sparks, P. (1995). Including moral dimensions of choice within the structure of the theory of planned behavior. Journal of Applied Social Psychology, 25, 484-494. Rabow, J., Neuman, C. A., & Hernandez, A. C. R. (1987). Contingent consistency in attitudes, social support and the consumption of alcohol: Additive and interactive effects. Social Psychology Quarterly, 50, 50-63. Radcliffe-Brown, A. R. (1952). Structure and function in primitive society. New York: The Free Press.

< previous page

page_319

next page >

< previous page

page_320

next page > Page 320

Raden, D. (1977). Situational thresholds and attitude-behavior consistency. Sociometry, 40(2), 123-129. Randall, D. M., & Gibson, A. M. (1991). Ethical decision making in the medical profession: An application of the theory of planned behavior. Journal of Business Ethics, 10, 111-122. Read, S. J., & Marcus-Newhall, A. R. (1993). Explanatory coherence in social explanations: A parallel distributed processing account. Journal of Personality and Social Psychology, 65, 429-447. Read, S. J., Vanman, E. J., & Miller, L. C. (1997). Conncectionism, parallel constraint satisfaction processes, and Gestalt principles: (Re)Introducing cognitive dynamics to social psychology. Personality and Social Psychology Review, 1, 26-53. Reid, A., & Deaux, K. (1996). The relationship between social and personal identities: Segregation or integration? Journal of Personality and Social Psychology, 71, 1084-1091. Richard, R., van der Pligt, J., & de Vries, N. (1995). Anticipated affective reactions and prevention of AIDS. British Journal of Social Psychology, 34, 9-21. Richard, R., van der Pligt, J., & de Vries, N. (1996). Anticipated affect and behavioral choice. Basic and Applied Social Psychology, 18, 111-129. Richardson, K. D., & Cialdini, R. B. (1981). Basking and blasting: Tactics of indirect self-presentation. In J. T. Tedeschi (Ed.), Impression management theory and social psychological research (pp. 41-53). New York: Academic Press. Roberts, D. F., & Maccoby, N. (1985). Effects of mass communication. In G. Lindzey & Aronson E. (Eds.), The handbook of social psychology (3rd ed.). New York: Random House. Robinson, C. E. (1932). Straw votes. New York: Columbia University Press. Robinson, W. P. (1996). Social groups and identities, Developing the legacy of Henri Tajfel. International series in Social Psychology. Oxford, England: Butterworth Heinemann. Robinson, R. J., Keltner, D., Ward, A., & Ross, L. (1995). Actual versus assumed differences in construal: "Naive realism" in intergroup perception and conflict. Journal of Personality and Social Psychology, 68, 404-417. Roemer, R. (1993). Legislative action to combat the world tobacco epidemic (2nd ed.). Geneva: World Health Organization. Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change. Journal of Psychology, 91, 93-114. Roloff, M. E., Janiszewski, C. A., McGarth, M. A., Burns, C. S., & Manrai, L. A. (1988). Acquiring resources from intimates: When obligation substitutes for persuasion. Human Communication Research, 14, 364-396. Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization (pp. 2748). Hillsdale, NJ: Lawrence Erlbaum Associates. Rosenberg, M. (1981). The self concept: Social product and social force. In M. Rosenberg & R. H. Turner (Eds.), Social psychology: Sociological perspectives (pp. 593-624). New York: Basic Books. Rosenstock, I. M. (1974). The health belief model and preventive health behavior. Health Education Monographs, 2, 354-386. Rosenthal, D. A., Hall, C., & Moore, S. M. (1992). AIDS, adolescents, and sexual risk taking: A test of the Health Belief Model. Australian Psychologist, 27, 166-171. Ross, L., Greene, D., & House, P. (1977). The "false consensus effect": An egocentric bias in social perception and attribution processes. Journal of Experimental Social Psychology, 13, 279-301. Ross, L., & Ward, A. (1996). Naive realism: Implications for misunderstanding and divergent perceptions of fairness and bias. In T. Brown, E. Reed, & E. Turiel (Eds.), Values and knowledge (pp. 103-135). Mahwah, NJ: Lawrence Erlbaum Associates.

< previous page

page_320

next page >

< previous page

page_321

next page > Page 321

Ruble, D. N., Balaban, T., & Cooper, J. (1981). Gender constancy and the effects of sex-typed televised toy commercials. Child Development, 52, 667-673. Rucinski, D., & Salmon, C. T. (1990). The "other" as the vulnerable voter: A study of the third-person effect in the 1988 U.S. presidential campaign. International Journal of Public Opinion Research, 2(4), 345-368. Rumnelhart, D. E., McClelland, J. L., & PDP Research Group. (1986). Parallel distributed processing: Explorations in the microstructure of cognition, (Vol. 1). Cambridge, MA: Bradford. Russell, M. A. H. (1990). How to enhance national rates of smoking cessation. World Conference on Tobacco & Health, The Global War. Health Department of Western Australia. Ryan, A. (Ed.). (1987). Utilitarianism and other essays. Harmondsworth, England: Penguin. Ryan, M. J. (1978). An examination of the alternative form of the behavioral intention model's normative component. In H. Keith Hunt (Ed.), Advances in Consumer Research (Vol. 5, pp. 282-289). Ann Arbor, MI: Association for Consumer Research. Ryan, M. J. (1982). Behavioral intention formation: The interdependency of attitudinal and social influence variables. Journal of Consumer Research, 9, 263-278. Sachdev, I., & Bourhis, R. Y. (1993). Ethnolinguistic vitality: Some motivational and cognitive considerations. In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 33-51). London: Harvester Wheatsheaf. Sagoff, M. (1988). The economy of the earth. Cambridge, England: Cambridge University Press. Samuelson, C. D., & Allison, S. T. (1994). Cognitive factors affecting the use of social decision heuristics in resourcesharing tasks. Organizational Behavior and Human Decision Processes, 58, 1-27. Sanbonmatsu, D. M., & Fazio, R. H. (1990). The role of attitudes in memory-based decision-making. Journal of Personality and Social Psychology, 59, 614-622. Schacter, S. (1951). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology, 46, 190207. Schachter, S. (1982). Recidivism and self-cure of smoking and obesity. American Psychologist, 37, 436-444. Schanck, R. L. (1932). A study of community and its group institutions conceived of as behavior of individuals. Psychological Monographs, 43, 1-133. Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ: Lawrence Erlbaum Associates. Scheibe, T. (1985). Historical perspectives on the presented self. In B. R.. Schlenker (Ed.), The self and social life. New York: McGraw-Hill. Schelling, T. C. (1992). Self-command: A new discipline. In G. Loewenstein & J. Elster (Eds.), Choice over time (pp. 167-209). New York: Russell Sage Foundation. Scher, S. J., & Cooper, J. (1989). Motivational basis of dissonance: The singular role of behavioral consequences. Journal of Personality and Social Psychology, 56, 899-906. Schlenker, B. R., & Miller, R. S. (1977). Egocentrism in groups: Self-serving bias or logical information processing? Journal of Personality and Social Psychology, 35, 755-764. Schoemaker, P. J. H. (1982). The expected utility model: Its variants, purposes, evidence and limitations. Journal of Economic Literature, 20, 529-563. Schofield, J. W. (1975). Effects of norms, public disclosure, and need for approval on volunteering behavior consistent with attitudes. Journal of Personality and Social Psychology, 31, 1126-1133. Schroeder, C. M., & Prentice, D. A. (1997). Dispelling pluralistic ignorance to reduce excessive drinking on college campuses. Unpublished manuscript, Princeton University, Princeton, NJ.

< previous page

page_321

next page >

< previous page

page_322

next page > Page 322

Schuette, R. A., & Fazio, R. H. (1995). Attitude accessibility and motivation as determinants of biased processing: A test of the MODE model. Personality and Social Psychology Bulletin, 21, 704-710. Schulte, F. (1995). Fleeced! Amherst, NY: Prometheus. Schuman, H., & Johnson, M. P. (1976). Attitudes and behavior. Annual Review of Sociology, 2, 161-207. Schwartz, S. H., & Tessler, R. (1972). A test of a model for reducing measured attitude-behavior discrepancies. Journal of Personality and Social Psychology, 24, 225-236. Senour, M. N. (1982). How counselors influence clients. The Personnel and Guidance Journal, 60, 345-349. Settle, R. B., & Golden, L. L. (1974). Attribution theory and advertising credibility. Journal of Marketing Research, 11, 181-185. Shafir, E. (1993). Choosing versus rejecting: Why some options are both better and worse than others. Memory and Cognition, 21(4), 546-556. Shavitt, S. (1989). Operationalizing functional theories of attitude. In A. R. Pratkanis, S. J. Breckler & A. G. Greenwald (Eds.), Attitude structure and function (pp. 311-337). Hillsdale, NJ: Lawrence Erlbaum Associates. Shavitt, S. (1990). The role of attitude objects in attitude functions. Journal of Experimental Social Psychology, 26, 124-148. Shepherd, G. J., & O'Keefe, D. J. (1984). Separability of attitudinal and normative influences on behavioral intentions in the Fishbein-Ajzen model. Journal of Social Psychology, 122, 287-288. Sheppard, B. H., Hartwick, J., & Warshaw, P R. (1988). The theory of reasoned action: A meta-analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15, 325-343. Sherif, M. (1936). The psychology of social norms. New York: Harper and Brothers. Sherif, M., & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in communication and attitude change. New Haven, CT: Yale University Press. Sherman, S. J. (1980). On the self-erasing nature of errors of perdiction. Journal of Personality and Social Psychology, 39, 211-221. Sherman, S. J., & Fazio, R. H. (1983). Parallels between attitudes and traits as predictors of behavior. Journal of Personality, 51, 308-345. Sherman, S. J., Presson, C. C., & Chassin, L. (1984). Mechanisms underlying the false consensus effect: The special role of threats to the self. Personality and Social Psychology Bulletin, 10(1), 127-138. Shimp, T. A., & Kavas, A. (1984). The theory of reasoned action applied to coupon usage. Journal of Consumer Research, 11, 795-809. Shrauger, J. S., & Shoenemann, T. J. (1979). Symbolic interactionist view of self-concept: Through the looking-glass darkly. Psychological Bulletin, 86(3), 549-573. Shultz, T. R., & Lepper, M. R. (1996). Cognitive dissonance reduction as constraint satisfaction. Psychological Review, 103, 219-240. Siegel, A. E., & Siegel, S. (1957). Reference groups, membership groups, and attitude change. Journal of Abnormal and Social Psychology, 55, 360-364. Simon, B. (1992). The perception of ingroup and outgroup homogeneity: Reintroducing the intergroup context. European Review of Social Psychology, 3, 1-30. Simon, B. (1993). On the asymmetry in the cognitive construal of ingroup and outgroup: A model of egocentric social categorization. European Journal of Social Psychology, 23, 131-147. Simons, H. W., Berkowitz, N. N., & Moyer, R. J. (1970). Similarity, credibility and attitude change: A review and a theory. Psychological Bulletin, 73, 1-16.

< previous page

page_322

next page >

< previous page

page_323

next page > Page 323

Simonson, I. (1992). The influence of anticipating regret and responsibility on purchase decisions. Journal of Consumer Research, 19, 105-118. Singelis, T. M. (1994). The measurement of independent and interdependent selfconstruals. Personality and Social Psychology Bulletin, 20, 580-591. Skevington, S. M. (1981). Intergroup relations and nursing. European Journal of Social Psychology, 11, 43-59. Skinner, B. F (1957). Verbal behavior. New York: Appleton-Century-Crofts. Skinner, W. F., & Catarello, A. M. (1989). Understanding the relationships among attitudes, group norms, and behavior using behavioral commitment: A structural equation analysis of marijuana use. Journal of Applied Social Psychology, 19, 1268-1291. Smith, E. R. (1996). What do connectionism and social psychology offer each other? Journal of Personality and Social Psychology, 70, 893-912. Smith, E. R., & Zaraté, M. A. (1992). Exemplar-based model of social judgment. Psychological Review, 99, 3-21. Smith, R. (1986). Television addiction. In J. Bryant & D. Zillman (Eds.), Perspectives on media effects (pp. 109-128). Hillsdale, NJ: Lawrence Erlbaum Associates. Smolensky, P. (1990). Tensor product variable binding and the representation of symbolic structures in connectionist systems. Artificial Intelligence, 46, 159-216. Snyder, C. R., & Fromkin, H. L. (1980). Uniqueness: The human pursuit of difference. New York: Plenum. 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., & DeBono, K. G. (1987). A functional approach to attitudes and persuasion. In M. P. Zanna, J. M. Olson, & C. P Herman (Eds.), The Ontario Symposium: Vol. 5. Social influence (pp. 107-125). Hillsdale, NJ: Lawrence Erlbaum Associates. Snyder, M., & DeBono, K. G. (1989). Understanding the functions of attitudes: Lessons from personality and social behavior. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.), Attitude structure and function (pp. 361-381). Hillsdale, NJ: Lawrence Erlbaum Associates. Snyder, M., & Ickes, W. (1985). Personality and social behavior. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 2, pp. 883-947). New York: Random House. Snyder, M., & Rothbart, M. (1971). Communicator attractiveness and opinion change. Canadian Journal of Behavioral Science, 3, 377-387. Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and interpersonal behavior: On the self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666. Spanos, N. P. (1996). Multiple identities and false memories: A sociocognitive perspective. Washington DC: American Psychological Association. Spanos, N. P., & Chave, J. F. (1989). Hypnosis: The cognitive-behavioral perspective. Buffalo, NY: Prometheus Books. Spanos, N. P., Radtke, H. L., & Bertrand, L. D. (1985). Hypnotic amnesia as a strategic enactment: Breaching amnesia in highly susceptible subjects. Journal of Personality and Social Psychology, 47, 1155-1169. Sparks, P., James, R., Conner, M., Shepherd, R., & Povey, R. (1997). Ambivalence about health-related behaviors: An exploration in the domain of food choice. Manuscript submitted for publication. Sparks, P., & Guthrie, C. A. (1998). Self-identity and the Theory of Planned Behavior: a useful addition or an unhelpful artifice? Journal of Applied Social Psychology, 28, 15, 1394-1411.

< previous page

page_323

next page >

< previous page

page_324

next page > Page 324

Sparks, P., & Shepherd, R. (1992). Self-identity and the theory of planned behavior: Assessing the role of identification with ''green consumerism.'' Social Psychology Quarterly, 55(4), 388-399. Sparks, P., Shepherd, R., & Frewer, L. J. (1995). Assessing and structuring attitudes toward the use of gene technology in food production: The role of perceived ethical obligation. Basic and Applied Social Psychology, 16, 267-285. Sparks, P., Shepherd, R., Wieringa, N., & Zimmermanns, N. (1995). Perceived behavioral control, unrealistic optimism and dietary change: An exploratory study. Appetite, 24(3), 243-255. Spence, J. T., Helmreich, R., & Stapp, J. (1974). Ratings of self and peers on sex role attributes and their relation to self-esteem and conceptions of masculinity and femininity. Journal of Personality and Social Psychology, 32, 29-39. Staats, A. W., & Staats, C. K. (1958). Attitudes established by classical conditioning. Journal of Abnormal and Social Psychology, 57, 37-40. Stasser, G., & Davis, J. H. (1977). Opinion change during group discussion. Personality and Social Psychology Bulletin, 3, 252-256. Stasson, M., & Fishbein, M. (1990). The relation between perceived risk and preventive action: Within-subject analysis of perceived driving risk and intentions to wear seat belts. Journal of Applied Social Psychology, 20, 1541-1547. Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. Advances in Experimental Social Psychology, 21, 261-302. Stein, R. I., & Nemeroff, C. J. (1995). Moral overtones of food: Judgments of others based on what they eat. Personality and Social Psychology Bulletin, 21, 480-490. Steketee, M. (1996, March 4). Howard's way forward. The Australian, p. 17. Sternberg, R. J. (1986). A triangular theory of love. Psychological Review, 93, 119-135. Strack, F., Martin, L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the human smile: A nonobtrusive test of the facial feedback hypothesis. Journal of Personality and Social Psychology, 54, 768-777. Strenta, A., & DeJong, W. (1981). The effect of a prosocial label on helping behavior. Social Psychology Quarterly, 44, 142-147. Stroebe, W., & Diehl, M. (1981). Conformity and counterattitudinal behavior: The effect of social support on attitude change. Journal of Personality and Social Psychology, 41, 876-889. Stroebe, W., & Stroebe, M. S. (1995). Social psychology and health. Buckingham, England: Open University Press. Struckman-Johnson, C. J., Gilliland, R. C., Struckman-Johnson, D. L., & North, T. C. (1990). The effects of fear of AIDS and gender on responses to fear-arousing condom advertisements. Journal of Applied Social Psychology, 20, 1396-1410. Stryker, S. (1986). Identity theory: Developments and extensions. In K. Yardley & T Honess (Eds.), Self and identity: Psychosocial perspectives. Chichester: Wiley. Stryker, S., & Statham, A. (1985). Symbolic interaction and role theory. In G. Lindzey & E. Aronson (Eds.), Handbook of Social Psychology, (3rd ed., Vol. 1, pp. 311-378). New York: Random House. Tabacalera. (1994). Press release of Gabinete de Comunicación Externa. Tabacalera S. A., Madrid. Tajfel, H. (1974). Social identity and intergroup behaviour. Social Science Information, 13, 65-93. Tajfel, H. (1975). The exit of social mobility and the voice of social change. Social Science Information, 14, 101-118. Tajfel, H. (1978). Differentiation between social groups. London: Academic Press.

< previous page

page_324

next page >

< previous page

page_325

next page > Page 325

Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. Cambridge, England: Cambridge University Press. Tajfel, H. (1982). Social identity and intergroup relations. Cambridge, England: Cambridge University Press. Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33-147). Pacific Grove, CA: Brooks/Cole. Tajfel, H., & Turner, J. C. (1986). An integrative theory of intergroup conflict. In S. Worchel & W. Austin (Eds.), Psychology of intergroup relations (pp. 2-24). Chicago: Nelson-Hall. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (2nd ed., pp. 7-24). Chicago: Nelson-Hall. Taylor, D. M., & Jaggi, V. (1974). Ethnocentrism and causal attributions in a S. Indian context. Journal of CrossCultural Psychology, 5, 162-171. Taylor, L. (1996, February 29). Optimistic Keating dismisses private party polling. The Australian, p. 7. Terry, D. J., Callan, V. J., & Sartori, G. (1996). Employee adjustment to organizational mergers: Stress, coping and intergroup differences. Stress Medicine, 12, 105-122. Terry, D. J., Gallois, C., & McCamish, M. (1993). The theory of reasoned action and health behaviour. In D. J. Terry, C. Gallois, & M. McCamish, (Eds.), The theory of reasoned action: Its application to AIDS-preventive behaviour (pp. 1-27). Oxford: Pergamon. Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitude-behavior relationship: A role for group identification. Personality and Social Psychology Bulletin, 22, 776-793. Terry, D. J., & Hogg, M. A., & McKimmie, B. M. (in press). Group salience, norm congruency, and mode of behavioral decision-making: The effect of group norms on attitude-behavior relations. British Journal of Social Psychology. Terry, D. J., Hogg, M. A., & White, K. M. (In press). The theory of planned behavior: Self-identity, social identity, and group norms. British Journal of Social Psychology. Terry, D. J., & O'Leary, J. E. (1995). The theory of planned behaviour: The effects of perceived behavioral control and self-efficacy. British Journal of Social Psychology, 34, 199-220. Tetlock, P. E. (1983). Accountability and complexity of thought. Journal of Personality and Social Psychology, 45, 7483. Theodorakis, Y. (1994). Planned behavior, attitude strength, role identity, and the prediction of exercise behavior. The Sport Psychologist, 8, 149-165. Thomas, E., Polansky, N., & Kounin, J. (1955). The expected behavior of a potentially helpful person. Human Relations, 8, 165-174. Thomas, W. I., & Znaniecki, F. (1918). The Polish peasant in Europe and America (Vol. 1). Boston, MA: Badger. Thompson, E. (1978). Smoking education programs: 1960-1976. American Journal of Public Health, 68, 250-260. Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1996). Let's not be indifferent about (attitudinal) ambivalence. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 361-386). Mahwah, NJ: Lawrence Erlbaum Associates. Tingle, L. (1996, March 2-3). Coalition a 'fraudulent' regime, says defiant PM. The Weekend Australian, p. 9. Toch, H., & Klofas, J. (1984). Pluralistic ignorance, revisited. In G. M. Stephenson & J. H. Davis (Eds.), Progress in applied social psychology (Vol. 2, pp. 129-159). New York: Wiley.

< previous page

page_325

next page >

< previous page

page_326

next page > Page 326

Trafimow, D. (1994). Predicting intentions to use a condom from perceptions of normative pressure and confidence in those perceptions. Journal of Applied Social Psychology, 24, 2151-2163. Trafimow, D. (1997a). The moderating effect of habit on determinants of intentions to use a condom. Manuscript submitted for publication. Trafimow, D. (1999). Predicting condom use: The importance of confidence in normative and attitudinal perceptions. Manuscript submitted for publication. Trafimow, D. (1999). Attitudinal and normative processes in health behavior. Psychology and Health, 13, 307-317. Trafimow, D., & Duran, A. (1998). Some tests of the distinction between attitude and perceived behavioral control. British Journal of Social Psychology, 37, 1-14. Trafimow, D., & Finlay, K. (1996). The importance of subjective norms for a minority of people. Personality and Social Psychology Bulletin, 22, 820-828. Trafimow, D., & Finlay, K. (1998). The relationship between normatively versus attitudinally controlled people and normatively versus attitudinally controlled behaviors. Manuscript submitted for publication. Trafimow, D., & Fishbein, M. (1994a). The importance of risk in determining the extent to which attitudes affect intentions to wear seat belts. Journal of Applied Social Psychology, 24, 1-11. Trafimow, D., & Fishbein, M. (1994b). The moderating effect of behavior type on the subjective norm-behavior relationship. The Journal of Social Psychology, 134, 755-763. Trafimow, D., & Fishbein, M. (1995). Do people really distinguish between behavioral and normative beliefs? British Journal of Social Psychology, 34, 257-266. Trafimow, D., & Sheeran, P. (1998). Some tests of the distinction between cognitive and affective beliefs. Journal of Experimental Social Psychology, 34, 378-397. Trafimow, D., Silverman, E. S., Fan, R. M., & Law, J. S. F (1997). The effects of language and priming on the relative accessibility of the private self and the collective self. Journal of Cross-Cultural Psychology, 28, 107-123. Trafimow, D., & Smith, M. D. (1998). An extension of the "two-baskets" theory to Native Americans. European Journal of Social Psychology. Trafimow, D., & Trafimow, J. H. (1998). Predicting back pain sufferers' intentions to exercise. The Journal of Psychology, 132, 581-592. Trafimow, D., Triandis, H. C., & Goto, S. G. (1991). Some tests of the distinction between the private self and the collective self. Journal of Personality and Social Psychology, 60, 649-655. Triandis, H. C. (1977). Interpersonal behavior. Monterey, CA: Brooks/Cole. Triandis, H. C. (1980). Values, attitudes, and interpersonal behavior. In H. E. Howe & M. M. Page (Eds.), Nebraska Symposium on Motivation, 1979 (pp. 195-259). Lincoln: University of Nebraska Press. Triandis, H. C. (1989). Self and social behavior in differing cultural contexts. Psychological Review, 96, 269-289. Triandis, H. C. (1990). Cross-cultural studies of individualism and collectivism. In J. Berman (Ed.), Nebraska Symposium on Motivation 1989 (pp. 43-133). Lincoln: University of Nebraska Press. Triandis, H. C. (1994). Culture and social behavior. New York: McGraw-Hill, Inc. Triandis, H. C., McCusker, C., & Hui, C. H. (1990). Multimethod probes of individualism and collectivism. Journal of Personality and Social Psychology, 59, 1006-1020. Turner, J. (1982). Toward a cognitive redefinition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 15-40). Cambridge, England: Cambridge University Press.

< previous page

page_326

next page >

< previous page

page_327

next page > Page 327

Turner, J. C. (1985). Social categorization and the self-concept: A social-cognitive theory of group behaviour. In E. J. Lawler (Ed.), Advances in group processes: Theory and research (Vol. 2, pp. 77-122). Greenwich, CT: JAI Press. Turner, J. C. (1991). Social influence. Bristol, England: Open University Press. Turner, J. C., & Brown, R. (1978). Social status, cognitive alternatives and intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups (pp. 201-226). London: Academic Press. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell. Turner, J. C., & Oakes, P. J. (1989). Self-categorization theory and social influence. In P. B. Paulus (Ed.), The psychology of group influence (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Turner, M. E., & Pratkanis, A. R. (1994). Affirmative action as help: A review of recipient reactions to preferential selection and affirmative action. Basic and Applied Social Psychology, 15, 43-69. Turner, R. H. (1956). Role-taking, role standpoint, and reference-group behavior. American Journal of Sociology, 61, 316-328. U.S. Department of Health and Human Services. (1989). Reducing the health consequences of smoking: 25 years of progress. A report of the Surgeon-General. Rockville, MD, U.S. Department of Health and Human Services, Offices on Smoking and Health, DHHS Publication No. (CDC) 89-8411. Vallacher, R. R., & Wegner, D. M. (1987). What do people think they're doing? Action identification and human behavior. Psychological Review, 94, 3-15. Vallacher, R. R., Wegner, D. M., & Somoza, M. P (1989). That's easy for you to say: Action identification and speech fluency. Journal of Personality and Social Psychology, 56, 199-208. Vallerand, R. J., Deshaies, P., Cuerrier, J-P., Pelletier, L. G., & Mongeau, C. (1992). Ajzen and Fishbein's theory of reasoned action as applied to moral behavior: A confirmatory analysis. Journal of Personality and Social Psychology, 62, 98-109. Vallone, R. P., Ross, L., & Lepper, M. R. (1985). The hostile media phenomenon: Bias in the coverage of the Beirut massacre. Journal of Personality and Social Psychology, 49, 577-585. van der Pligt, J. (1996). Judgment and decision making. In G. R. Semin & K. Fiedler (Eds.), Applied social psychology (pp. 30-64). London: Sage. Vangelisti, A. L., Daly, J. A., & Rudnick, J. R. (1991). Making people feel guilty in conversations: Techniques and correlates. Human Communication Research, 18, 3-39. van Knippenberg, A., & Ellemers, N. (1993). Strategies in intergroup relations. In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 17-32). London: Harvester Wheatsheaf. van Knippenberg, D., & van Knippenberg, B. (1996). Polarized group perceptions in intergroup contexts: A group is more extreme than its average member. Unpublished manuscript, University of Leiden. van Knippenberg, D., Lossie, N., & Wilke, H. (1994). In-group prototypicality and persuasion: Determinants of heuristic and systematic message processing. British Journal of Social Psychology, 33, 289-300. van Knippenberg, D., & Wilke, H. (1991). Sociale categorisatie, verwerking van argumenten en attitudeverandering [Social categorization, processing of arguments and attitude change]. In J. van der Pligt, W. van der Kloot, A. van Knippenberg & M. Poppe (Eds.), Fundamentele sociale psychologie, 5, (pp. 96-111). Tilburg University Press. van Knippenberg, D., & Wilke, H. (1992). Prototypicality of arguments and conformity to ingroup norms. European Journal of Social Psychology, 22, 141-155.

< previous page

page_327

next page >

< previous page

page_328

next page > Page 328

Vaughan, G. M. (1978). Social change and intergroup preferences in New Zealand. European Journal of Social Psychology, 8, 297-314. Vygotsky, L. S. (1978). Mind in society. Cambridge, MA: Harvard University Press. Vygotsky, L. S. (1981). The genesis of higher mental functions. In J. V. Wertsch (Ed. & Trans.), The concept of activity in Soviet psychology (pp. 144-188). White Plains, NY: Sharpe. Wagner, W. (1984). Social comparison of opinions: Similarity, ability, and the value-fact distinction. Journal of Psychology, 117, 197-202. Walster, E., Aronson, E., & Abrahams, D. (1966). On increasing the persuasiveness of a low prestige communicator. Journal of Experimental Social Psychology, 2, 325-342. Watkins, J. L. (1959). The 100 greatest advertisements. New York: Dover Press. Watson, J. B. (1930). Behaviorism. New York: Norton. Weber, M. (1978). Economy and society (Vols. 1-2). Berkeley: University of California Press. (Original work published 1958). Wegner, D. M., & Vallacher, R. R. (1986). Action identification. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: Foundations of social behavior (pp. 550-582). New York: Guilford. Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and Social Psychology, 39, 806-820. Weinstein, E. A., & Deutschberger, P. (1963). Some dimensions of altercasting. Sociometry, 26, 454-466. Weinstein, E. A., & Deutschberger, P. (1964). Task, bargains, and identities in social interaction. Social Forces, 42, 451-456. Weinstein, E. A., Wiley, M. G., & DeVaughn, W. (1966). Role and interpersonal style as components of social interaction. Social Forces, 45, 210-216. Wetherell, M. (1987). Social identity and group polarization. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. D. Reicher, & M. Wetherell (Eds.), Rediscovering the social group: A self-categorization theory (pp. 142-170). Oxford: Blackwell. Wheatley, J. J., & Oshikawa, S. (1970). The relationship between anxiety and positive and negative advertising appeals. Journal of Marketing Research, 7, 85-89. Wheeler, S. (1961). Role conflict in correctional communities. In D. R. Cressey (Ed.), The prison: Studies in institutional organization and change (pp. 75-98). New York: Holt, Rinehart & Winston. White, J. D., & Carlston, D. E. (1983). Consequences of schemata for attention, impressions, and recall in complex social interactions. Journal of Personality and Social Psychology, 45, 538-549. White, K. M., Terry, D. J., & Hogg, M. A. (1999). Attitude-behavior relations: The role of ingroup norms and mode of behavioral decision-making. Basic and Applied Social Psychology. White, K. M., Terry, D. J., & Hogg, M. A. (1994). Safer sex behavior: The role of attitudes, norms, and control factors. Journal of Applied Social Psychology, 24, 2164-2192. White, P. H., & Harkins, S. G. (1994). Race of source effects in the elaboration likelihood model. Journal of Personality and Social Psychology, 67, 790-807. Whitlock, C. R. (1994). Easy money. New York: Kensington Books. Whitlock, C. R. (1997). Scam school. New York: Macmillan. Wicker, A. W. (1969). Attitude versus actions: The relationship of verbal and overt behavioral responses to attitude objects. Journal of Social Issues, 25, 41-78. Wicklund, R. A., & Brehm, J. W. (1976). Perspectives on cognitive dissonance. Hillsdale, NJ: Lawrence Erlbaum Associates.

< previous page

page_328

next page >

< previous page

page_329

next page > Page 329

Wilder, D. A. (1990). Some determinants of the persuasive power of ingroups and outgroups: Organization of information and attribution of independence. Journal of Personality and Social Psychology, 59, 1202-1213. Williams, J. A. (1984). Gender and intergroup behaviour: Towards an integration. British Journal of Social Psychology, 23, 311-316. Williams, J. E., Bennett, J. M., & Best, D. L. (1975). Awareness and expression of sex stereotypes in young children. Developmental Psychology, 11, 635-642. Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes: A thirty nation study. Beverly Hills, CA: Sage. Witte, K. (1992). The role of threat and efficacy in AIDS prevention. International Quarterly of Community Health Education, 12, 225-249. Wood, W., Kallgren, C., & Priesler, R. (1985). Access to attitude-relevant information in memory as a determinant of persuasion. Journal of Experimental Social Psychology, 21, 73-85. Woodside, A. G., & Davenport, J. W., Jr. (1974). The effect of salesman similarity and expertise on consumer purchasing behavior. Journal of Marketing Research, 11, 198-202. Word, C. O., Zanna, M. P., & Cooper, J. (1974). The nonverbal mediation of self-fulfilling prophecies in interracial interaction. Journal of Experimental Social Psychology, 10, 109-120. World Health Organization (1988). A 5 year action plan. Copenhagen: WHO, Regional Office for Europe. World Health Organization (1993). A tobacco-free Europe, an action plan. Copenhagen: WHO, Regional Office for Europe. World Health Organization (1994). World No-Tobacco Day 1994: Over one billion smokers in the world. Press release (WHO/44), 30 May 1994. World Health Organization (1996). The tobacco epidemic: A global public health emergency first global report on tobacco. Press release (WHO/4), 23 May 1996. Worth, L. T., Smith J., & Mackie, D. M. (1992). Gender schematicity and preference for gender-typed products. Psychology and Marketing, 9, 17-30. Wright, S. C., Taylor, D. M., & Moghaddam, F. M. (1990). Responding to membership in a disadvantaged group: From acceptance to collective protest. Journal of Personality and Social Psychology, 58, 994-1103. Wrong, D. H. (1961). The oversocialized conception of man in modern sociology. American Sociological Review, 26, 183-193. Ybarra, O., & Trafimow, D. (1998). How priming the private self or collective self affects the relative weights of attitudes and subjective norms. Personality and Social Psychology Bulletin, 24, 362-370. Zanna, M. P., & Azizza, C. (1976). On the interaction of repression-sensitization and attention in resolving cognitive dissonance. Journal of Personality, 44, 577-593. Zanna, M. P., & Cooper, J. (1974). Dissonance and the pill: An attribution approach to studying the arousal properties of dissonance. Journal of Personality and Social Psychology, 29, 703-709. Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar-Tal & A. W. Kruglanski (Eds.), The social psychology of knowledge (pp. 315-334). Cambridge, England: Cambridge University Press. Zanna, M. P., & Sande, G. N. (1987). The effects of collective action on the attitudes of individual group members: A dissonance analysis. In M. P. Zanna, J. M. Olson, & C. P. Herman (Eds.), The Ontario Symposium: Vol. 5. Social Influence (pp. 151-163). Hillsdale, NJ: Lawrence Erlbaum Associates.

< previous page

page_329

next page >

< previous page

page_330

next page > Page 330

Zeelenberg, M., Beattie, J., van der Pligt, J., & de Vries, N. K. (1996). Consequences of regret aversion: Effects of expected feedback on risky decision making. Organizational Behavior and Human Decision Processes, 65, 148-158. Zipp, J. F., & Plutzer, E. (1985). Gender differences in voting for female candidates: Evidence from the 1982 election. Public Opinion Quarterly, 49, 179-197. Zimbardo, P. G. (Executive Producer). (1991). Quiet Rage: Stanford Prison. [Video recording]. New York: Insight Media. Zuckerman, M., & Reis, H. T. (1978). Comparison of three models for predicting altruistic behavior. Journal of Personality and Social Psychology, 36, 498-510.

< previous page

page_330

next page >

< previous page

page_331

next page > Page 331

AUTHOR INDEX A Abelson, R. P., 39, 42, 60, 121, 216 Abrahams, D., 217 Abrams, D., 2, 53, 58, 67, 69, 71, 72, 82, 87, 109, 158, 160, 161, 166, 169, 176, 205, 219, 232, 268, 269, 284, 286, 288, 290 Acock, A. C., 71 Agostinelli, G., 257 Ainslie, G., 41, 42 Ajzen, I., 11, 12, 13, 14, 16, 17, 31, 32, 34, 37, 40, 47, 49, 51, 63, 67, 68, 69, 70, 71, 73, 93, 95, 99, 100, 103, 115, 116, 117, 127, 248 Akers, R. L., 109, 111 Albrecht, S.,L., 74 Alinsky, S. D., 221 Allen, V. L., 143, 154, 257 Allison, S. T., 206 Allport, F. H., 103 Allport, G. W., 1, 45, 115 Altman, D. G., 247 American Lung Association, 246 Andersen, K. E., 136 Andersen, S. M., 153 Anderson, D. S., 205 Anderson, J. A., 118 Andrews, K. H., 71, 91 Antaki, C., 223 Archer, J., 286 Arie, O. G., 57 Armitage, C., 38 Aronson, E., 203, 213, 215, 217, 225, 228, 235 Ary, D., 247 Asch, S. E., 137, 138, 140, 157, 257 Ashmore, R. D., 214 Asuncion, A. G., 83, 92, 93, 149, 151, 169, 184, 185, 192, 194, 195, 196, 219, 265, 291 Azizza, C., 234 B Bachman, J., 247 Back, K., 157, 174 Bacon, C., 111 Bagozzi, R. P., 57, 71, 91 Bain, J. D., 119, 130 Bakan, D., 286 Balaban, T., 111

Balloun, J. C., 246 Bandura, A., 63 Bank, B. J., 35, 36, 37, 205 Banks, C., 206 Barber, R., 205 Bargh, J. A., 49, 143 Baron, J., 14, 33 Barton, J., 259 Bassili, J. N., 77 Bauman, K. E., 248 Baumgartner, H., 57 Beach, L. R., 43 Beach, R., 116 Bearden, W. O., 50, 57, 64, 70 Beattie, J., 14 Beck, L., 16, 17, 40 Becker, M. H., 248 Beckhouse, L., 204 Becona, E., 246, 247 Bell, D. E., 14 Benaquisto, L., 109, 110, 111 Bennett, J. M., 111 Bensenberg, M., 259 Bensman, J., 217 Bentler, P. M., 47, 49, 116, 126, 248 Berent, M. K., 45 Berger, C. J., 206 Berger,P., 35 Berkowitz, L., 221 Berkowitz, N. N., 136 Bemtson, G. G., 180

< previous page

page_331

next page >

< previous page

page_332

next page > Page 332

Berscheid, E., 136, 177, 209, 220 Bertrand, L. D., 207, 208 Best, D.L., 111, 286 Bickman, L., 216 Biddle, B. J., 35, 36, 37, 203, 205 Biglan, A., 247 Blaine, B., 87, 284 Blake, J., 124 Blanz, M., 259 Bobrow, D. G., 153 Bocock, R., 38 Bohm, G., 60 Bolen, D., 207, 224, 225 Boninger, D. S., 45 Bonnie, R. J., 247 Bontempo, R., 54 Boorstin, D. J., 222 Borgida, E., 153 Bouffard, D. L., 101 Bourdieu, P, 37, 38 Bourhis, R. Y., 275 Bower, A. M., 2 Bowman, C. C., 225 Boyanowsky, E. O., 154 Boyd, B., 20 Bradley, E., 209 Brauer, M., 77, 79 Breakwell, G., 250, 253, 257 Breckler, S. J., 2, 60, 123 Breed, W, 108 Brehm, J. W., 228, 234, 257 Brehm, S. S., 257 Breinlinger, S., 39 Brewer, M. B., 196, 222, 269, 274, 279, 290 Brickman, P, 207, 224, 225 Brock, T C., 136, 180, 181, 220, 239, 246 Brown, J. D., 284 Brown, R. J., 69, 275 Brownson, R. C., 246 Burger, J. M., 224, 268 Burke, P J., 216 Burnkrant, R. E., 198 Burs, C. S., 219

Burrows, L., 49 Bushman, B. J., 216 Butera, F., 249, 252, 254 Butler, D., 276 Buzzelli, C. A., 24, 25 Byrne, D., 136 C Cacioppo, J., 82, 135, 141, 142, 143, 153, 162, 163, 165, 169, 172, 176, 179, 180, 181, 182, 183, 193, 195, 215, 235, 246, 249 Callan, V. J., 276 Callero, P. L., 35, 36, 37, 205 Campbell, D. T., 60, 95, 98, 99 Cancian, F. M., 125 Cantor, N., 124, 222 Carlsmith, J. M., 101, 206, 208, 228 Carlston, D. E., 143 Carnot, C. G., 116 Carrasco, F., 250, 255 Catarello, A. M., 74 Cecil, H., 246 Chaiken, S., 2, 14, 32, 34, 35, 37, 39, 82, 123, 141, 142, 153, 158, 162, 163, 169, 180, 193, 215, 217, 249, 250, 254, 257, 266 Charng, H., W., 35, 36, 37 Chassin, L., 247, 248, 257, 259 Chave, J. F., 207 Chen, M., 49 Chen, S., 250, 254 Chenoworth, R. L., 248 Chung, J., 62 Cialdini, R. B., 125, 203, 219, 221, 224, 225, 264 Cleary, P. D., 246, 247, 254 Clevenger, T. Jr, 136 Cochrane, S., 161, 166, 169, 205 Codol, J. P., 279, 291 Cohen, J., 266, 267, 268 Cohen, J. B., 50, 70, 71 Collins, D. M., 205 Collins, R. L., 284 Condor, S., 223, 275 Conner, M. T., 32, 38, 41 Conover, P. J., 177, 178, 194 Cook, B. P., 207 Cooley, C., H., 34 Cooper, J., 111, 209, 228, 229, 234, 235, 236, 238, 239, 240 Cooper-Shaw, L., 167 Corty, E., 259

Coutu, W., 203 Crano, W. D., 217 Crawford, J., 282 Creamer, R., 205 Crites, S. L., 60, 61 Crocker, J., 87, 185, 189, 284 Croyle, R. T, 248 Cuerrier, J. P., 26, 50, 64, 70 Cummings, L. L., 206 D Daly, J. A., 219 Danto, A., 122 Darley, J. M., 101, 107, 203, 231

< previous page

page_332

next page >

< previous page

page_333

next page > Page 333

Davenport, J. W. Jr., 136 David, B., 161, 167 Davidson, A. R., 116 Davidson, D., 41, 169, 170 Davidson, W., 209 Davis, J. H., 87 Davis, J. R., 246 Davis, K., 124 Davis, K. E., 217, 228, 229 Davis, R. W., 87 Davison, W. P., 219, 265, 266, 268, 288, 290 Dawley, H, H., 246 Dawley, L. T., 246 de Vries, N. K., 14, 20, 275 Deaux, K., 173, 198, 286 DeBono, K. G., 43, 44, 153, 178, 184 DeCourville, N. H., 21 DeFleur, M. L., 71 DeJong, W., 207 Dennison, C. M., 38 Deshaies, P., 26, 50, 64, 70 Deshpande, R., 171 Deutsch, M., 71, 139, 140, 158, 159, 175, 183, 254 Deutschberger, P., 203, 204 DeVaughn, W., 204 Devine, P. G., 2 Diehl, M., 231 Ditto, P. H., 248 Doise, W., 233, 234, 252, 286 Doll, J., 117, 127 Doll, R., 245 Doms, M., 257 Donthu, N., 171 Doob, L. W., 129 Doojse, B., 275 Dossett, D. L., 47, 49 Dovidio, J. F., 213 Downing, J. W., 77, 79 Duck, J. M., 170, 267, 268, 270, 278, 281, 282, 285, 288 Duran, A., 65 Durand, R. M., 57 Durkheim, E., 124, 209 Durkin, K., 268

Dutton, D. G., 217 Dweck, C. S., 209 E Eagly, A. H., 2, 14, 32, 33, 34, 35, 37, 39, 61, 82, 115, 116, 123, 158, 162, 163, 169, 214, 217, 250, 286 Echebarria Echabe, A., 246, 249, 251, 259 Edwards, K., 198 Edwards, W., 33 Eiser, J. R., 246, 247, 248 Ellemers, N., 275 Elms, A. C., 208 Eroski, 246 Etzioni, A., 43, 44 Eustace, M. A., 57 Evans, R. I., 246, 248 F Fabrigar, L. R., 60, 61 Falomir, J. M., 250, 254, 255, 257, 260, 262 Fan, R. M., 54, 55 Farley, J. U., 51, 63, 69 Fatoullah, E., 214, 215 Faucheaux, C., 160 Fazio, R. H., 49, 59, 68, 69, 75, 77, 79, 80, 82, 83, 86, 90, 95, 116, 117, 118, 126, 203, 228, 229, 235 Feather, N. T., 246 Fernandez Guede, E., 246, 249, 251, 259 Feshbach, S., 246 Festinger, L., 101, 138, 139, 140, 142, 143, 157, 159, 174, 176, 177, 195, 198, 205, 206, 208, 220, 224, 227, 228, 229, 230, 231, 243, 257 Fetzer, B. K., 268 Fiedler, K., 83 Fielding, K. S., 170 Fields, J. M., 108 Finlay, K., 52, 54, 63, 125 Fischler, C., 37, 38 Fishbein, M., 11, 12, 13, 14, 31, 32, 34, 37, 39, 47, 48, 49, 50, 51, 52, 56, 57, 59, 60, 62, 63, 6567, 68, 69, 70, 71, 73, 95, 99, 100, 103, 115, 116, 248 Fisher, D., 142, 181 Fisher, E. B., 246 Fisher, J. D., 219 Fisher, R., 226 Fiske, 38 Fiske, S. T., 1, 2, 60 Flavell, J. H., 23 Fleischer, B., 246 Fleming, M. A., 185, 189, 195 Forgas, J. P., 83 Francis, C., 57

Frederickson, H. G., 206 Fredricks, A. J., 47, 49 Freed, P. J., 109, 110, 111 Freedman, P., 179 French, J. R. P., 159 Freud, S., 239 Frewer, L. J., 22, 27, 40, 45 Frideres, J. S., 74 Fried, C. B., 235

< previous page

page_333

next page >

< previous page

page_334

next page > Page 334

Fromkin, H. L., 279 Frossart, M., 249 Frydman, M., 247, 248 Funt, A., 223 G Gaertner, S. L., 213 Galanter, E., 129 Gallois, C., 68, 125, 126 Garfinkel, H., 125 Garry, S. L., 108 Gastardo-Conaco, M. C., 134, 169, 184, 194, 196, 265, 291 Generalitat Valenciana, Dirreccio General de Salut Publica, 245, 246, 247 Gerard, H. B., 71, 139, 140, 158, 159, 175, 183, 254 Gergen, K. G., 35 Gergen, K. J., 217 Gerver, I., 217 Gibbon, P., 268 Gibbs, J. C., 22, 23, 24 Gibbs, J. P., 125 Gibson, A. M., 19 Giles, H., 275 Gilliland, R. C., 282 Giner-Sorolla, R., 250, 254, 266 Girgus, J. S., 111 Glasgow, R. E., 247 Gleason, P., 34 Gliner, M. D., 211, 212, 214 Goethals, G. R., 136, 176, 177, 229, 257 Goffman, E., 34, 35, 96, 97, 98 Goings, K. W., 212 Gold, M., 216 Golden, B. W., 213 Golden, L. L., 217 Goldman, R., 135, 163, 182 Goldstein, M. J., 135 Gonzalez, J. L., 246, 249, 251, 259 Goode, W. J., 203 Gorsuch, R. L., 16, 40, 47 Goto, S. G., 53, 55 Granberg, D., 35, 36 Green, J. A., 221 Greene, D., 143, 257 Greenglass, E. R., 286

Greenwald, A. G., 2, 116, 123, 208 Griffin, D. W., 41 Gross, C. R., 205 Grube, J. W., 47, 49, 50, 64, 70, 71, 91, 125 Grush, J. E., 57 Guadagno, R. E., 224 Guillon, M., 255 Gunther, A., 266, 267, 268, 282, 285 Guthrie, C. A., 37, 38 H Hadsall, R. S., 205 Hage, J., 216 Hains, S. C., 2, 78, 170, 275 Hall, C., 282 Hamilton, D. L., 2 Haney, C., 206 Hardie, E. A., 78, 167 Harding, C. M., 246 Harkins, S. G., 189 Harkins, S. J., 179, 213 Harkness, A. R., 153 Harrison, D. A., 16 Harsanyi, J. C., 41, 43 Hartwick, J., 49 Harvey, J., 142, 214, 215 Haslam, S. A., 169, 170, 178, 179, 266 Hass, R. G., 142 Hastie, R., 143, 189 Hauge, R., 205 Haugtvedt, C., 172, 176, 181, 183 Haunschild, P. R., 276 Heider, F., 143, 198 Helmreich, R., 286 Helquist, M., 57 Henderson, M. A., 201 Herek, G. M., 43 Herman, J. B., 206 Hernandez, A. C. R., 91 Herr, P. M., 77, 127 Herrnstein, R., 40 Hershey, J. C., 14 Higgins, E. T., 197, 235 Hilgard, E. R., 207 Hill, A. B., 245 Hill, C., 249 Hinton, G. E., 118

Hirsch, A., 249, 254 Hoffman, M. L., 3, 22, 23, 24, 25, 26, 28, 29 Hogg, M. A., 2, 3, 34, 40, 43, 53, 58, 59, 63, 64, 67, 69, 70, 71, 72, 75, 76, 78, 79, 81, 82, 83, 87, 90, 91, 93, 97, 100, 109, 126, 127, 128, 158, 160, 161, 164, 166, 167, 169, 170, 176, 185, 205, 219, 232, 233, 235, 267, 268, 269, 270, 275, 278, 281, 282, 284, 285, 286, 288, 290 Holmberg, S., 35, 36 Holzner, B., 35 Holzworth, D. W., 167 Hom, P. W., 18, 27 Homans, G., 124 Horai, J., 214, 215

< previous page

page_334

next page >

< previous page

page_335

next page > Page 335

House, P., 143, 257 Houston, D. A., 77 Hovland, C. L., 70, 73, 175, 209, 210, 211, 214 Hoyer, W. D., 171 Hui, C. H., 53 Hulin, C. L., 18, 27, 206 Humphreys, M. S., 119, 130 Huston, A. C., 110 Huston, T. L., 136 Hutchinson, K. J., 169, 178, 179, 266 Hyde, J. S., 111 Hyman, H., 245 I Ickes, W., 102 Infante, D. A., 136 Innes, J. M., 267 Institute for Propaganda Analysis, 220 J Jaccard, J. J., 15, 116 Jackson-Thompson, J., 246 Jaggi, V., 277 James, W., 34 Jamieson, D. W., 82 Janis, I., 135 Janis, I. L., 14, 208, 246 Janis, L., 70, 73, 209, 210, 211, 214 Janiszewski, C. A., 219 Janz, N. K., 248 Jensen, R. E., 207 Jermot, J. B. III, 248 Johnson, L., 247 Johnson, M. P., 102 Johnson, P., 275 Johnston, V S., 62 Jones, D., 52 Jones, E. E., 217, 219, 220, 221, 228, 229 Judd, C. M., 77, 79 K Kaigler-Evans, K., 136 Kallgren, C. A., 125, 153 Kandell, D. B., 71, 91 Kaplan, M. F., 87 Karsenty, S., 254

Kashima, E. S., 125 Kashima,Y., 118, 119, 121, 125, 126 Katerberg, R., 18, 27 Katz, D., 44, 104, 106, 123 Kaufmann, K., 109, 111 Kavas, A., 50, 64 Keats, D. M., 205 Keats, J. A., 205 Kelley, H. H., 70, 71, 73, 138, 155, 159, 176, 205, 209, 210, 211, 214 Kelly, C., 39 Kelman, H. C., 139, 140, 141, 142, 148, 152, 159, 175, 178, 180, 183, 209, 210, 211, 214, 220, 222, 262 Keltner, D., 110 Kerekes, A. R. Z., 119 Kem-Foxworth, M., 213 Kerr, N. L., 87 Kiesler, C. A., 207 Kihlstrom, J. F., 207 Kimble, C. E., 136 Kinder, D. R., 60 King, B. T., 208 King, D., 119, 121 King, S. W., 136 Kippax, S., 282 Kitayama, S., 53 Kitsanes, T., 108 Klapp, O. E., 218 Klassen, A. D., 109 Klatzky, R. L., 153 Klofas, J., 109, 110, 111, 112 Koch, J., 214, 215 Koeske, G. F., 217 Kohlberg, L., 3, 22, 23, 24, 26, 28 Korte, C., 112 Kounin, J., 219 Kramer, R. M., 269 Kraut, R. E., 207 Kristiansen, C. M., 246 Krosnick, J. A., 2, 45, 249, 250 Kunda, Z., 119 Kurland, N. B., 19 L Lage, E., 108 Lalonde, R. N., 276 Lane, I. M., 18, 27 LaPiere, R. T, 97, 98, 100, 101, 102, 104, 107 Lasorsa, D. L., 266, 268

Latane, B., 101, 107, 231 Lavine, H., 188, 189, 192, 193 Law, J. S. F., 54, 55 Lawler, E. E., 206, 207 Leary, M. R., 222 Leavitt, C., 136 Lee, Y. T, 286 Lehmann, D. R., 51, 63, 69 Lemaine, G., 279 Lepper, M. R., 119, 188, 189, 192, 266

< previous page

page_335

next page >

< previous page

page_336

next page > Page 336

Lerner, R. M., 1101 Lesh, M. D., 207 Leventhal, H., 246, 247, 254 Levine, J. M., 217 Levine, M., 223 Levinger, G., 136 Levitt, E. E., 109 Lewin, K., 245 Liberman, A., 82, 158, 162, 169 Lichenstein, E., 247 Lieberman, S., 206 Liska, A., 50, 70, 71 Litman, T. J., 205 Lloyd, B., 286 Longo, L. C., 214 Loomes, G., 14 Lord, C. G., 188, 189, 192 Lorenzi-Cioldi, F., 286 Losch, M. E., 235 Lossie, N., 164, 167, 169, 219 Luhtanen, R., 87, 185, 189, 284 Lumsdaine, A., 135 Lynch, B. S., 247 M Maccoby, N., 245 MacCoun, R. J., 87 Mackie, D. M., 83, 92, 93, 143, 149, 151, 154, 166, 169, 184, 185, 192, 194, 195, 196, 219, 234, 235, 238, 239, 241, 265, 291 Maddux, J. E., 214, 215 Maheswaran, D., 163, 193 Maibach, E., 246 Maier, N. R. F., 226 Makhijani, M. G., 214 Mann, L., 14, 208 Manrai, L. A., 219 Manstead, A. S. R., 14, 17, 32, 40, 248 March, J. G., 41, 43 Marcus-Newhall, A. R., 119 Markus, H., 1, 43, 53, 143, 173 Marlin, M. M., 205 Martin, R., 248 Marwell, G., 216 Maslach, C., 279 Matthews, A., 275

Matthews, R. C., 18, 27 Mausner, B., 248 McArthur, L. A., 207 McCamish, M., 68, 126 McClelland, J. L., 118 McCusker, C., 53 McFarland, C., 103, 107 McGarth, M. A., 219 McGarty, C., 161, 167, 169, 170, 178, 179, 266 McGinnies, E., 214 McGree, S. T, 47, 49, 50, 64, 70, 71, 125 McGuire, W. J., 1, 245, 246, 254 McKimmie, B. M., 83, 87 Merton, R., 35, 42 Merton, R. K., 203, 209 Middlestadt, S. E., 57 Milgram, S., 101, 206, 216, 257 Mill, J. S., 33 Millar, M. G., 60, 61 Miller, D. T., 96, 103, 105, 107, 111 Miller, G. A., 129 Miller, L. C., 119 Miller, L. E., 57, 87 Miller, R. L., 207, 224, 225 Miller, R. S., 268 Mills, J. M., 136, 142, 214, 215, 228 Miniard, P. W., 50, 70, 71 Ministerio Espanol de Sanidad y Consumo, 247 Mischel, W., 101, 124, 222 Mitchell, T. R., 43 Mladinic, A., 61 Moghaddam, F. M., 275, 276 Mongeau, C., 26, 50, 64, 70 Moore, S. G., 207 Moore, S. M., 282 Moreland, R. L., 2, 276 Morgan, K., 288 Morgan, M., 47, 49, 50, 64, 70, 71, 91, 125 Morrison, D. M., 116 Moscovici, S., 108, 139, 140, 158, 160, 221, 255, 262, 264 Motz, A. B., 216 Moyer, R. J., 136 Mugny, G., 249, 250, 252, 254, 255, 257, 260, 262 Mulligan, R., 211 Mullin, B. A., 267, 268, 285 Mummendey, A., 259

Mundy, P., 266, 267, 282, 285 Murrell, A. J., 276 Murroni, E., 214, 215 Mutz, D. C., 266, 267, 268 Myers, D. G., 101 Mynatt, C., 231 N Naccari, N., 214, 215 Nadler, A., 219 Naffrechoux, M., 108 Naficy, A., 229 Nelson, R. E., 136, 176, 177 Nelson, S., 209 Nemeroff, C. J., 38

< previous page

page_336

next page >

< previous page

page_337

next page > Page 337

Neuman, C. A., 91 Neve, P., 108 Newcomb, T. M., 112, 157, 178 Noelle-Neumann, E., 266 Nolte, A. E., 246 Norman, D. A., 121, 122, 153 Norman, R., 214, 215 North, D. C., 282 Nunnally, J. C., 103 Nye, J. L., 2 O O'Gorman, H. J., 108 O'Keefe, D. J., 70 O'Keefe, E. S. C., 111 O'Leary, J. E., 63 O'Malley, P., 247 O'Rourke, T., 246 Oaker, G., 275 Oakes, P. J., 67, 97, 109, 126, 158, 160, 161, 164, 167, 170, 176, 185, 268, 285 Office of Population Censures and Surveys, 247 Oliver, R. L., 50, 64, 70 Oliver-Rodriguez, J. C., 62 Olshevsky, R. W., 259 Ono, K., 87 Onorato, R. S., 170 Ooms, M., 57 Ortberg, J., 16, 40, 47 Oshikawa, S., 135 Ostrom, T. M., 2 Ottati, V., 286 Otten, S., 259 Otto, S., 61 Owens, N. W., 246 P Packard, J. S., 109, 112 Packer, M., 178 Pallak, S. R., 142, 214, 215 Papageorgis, D., 254 Park, B., 189 Parker, D., 14, 17, 40, 248 Parrott, R. L., 246 Parsons, T., 124 Patton, B., 226

PDP Research Group, 118 Pechabrier, M., 249 Pelletier, L. G., 26, 50, 64, 70 Perez, J. A., 249, 250, 252, 254, 255, 257, 260, 262 Perloff, L. S., 268 Perloff, R. M., 2, 266, 267, 274, 285 Personnaz, B., 255 Pervin, L. A., 246 Peters, M. D., 60 Pettigrew, T., 8, 276 Petty, R. E., 2, 60, 61, 82, 135, 141, 142, 143, 153, 162, 163, 165, 169, 172, 174, 176, 179, 180, 181, 182, 183, 185, 189, 193, 194, 195, 215, 246, 249 Pfeffer, J., 206, 207 Pfister, H. R., 60 Pike, A. R., 119, 130 Piliavin, J. A., 35, 36, 37, 205 Pittman, T., 219, 220, 221 Platt, E. S., 248 Plous, S., 212 Plutze, E., 171, 172 Polansky, N., 219 Pomazal, R., 15 Pomerantz, E. M., 249 Porter, L. W., 206 Posavac, S. S., 118 Povey, R., 41 Powell, M. C., 77, 79, 127 Pratkanis, A. R., 2, 203, 211, 212, 214, 219, 225 Prelec, D., 40 Prentice, D. A., 96, 103, 105, 106, 111 Presson, C. C., 247, 248, 257, 259 Prestholdt, P. H., 18, 27 Pribram, K. H., 129 Price, R. H., 101 Price, V., 266, 267, 268, 281 Priesler, R., 153 Priester, J. R., 158, 162, 180 Puckett, J., 142, 181 R Raats, M. M., 22, 40, 45 Rabow, J., 91 Radcliffe-Brown, A. R., 124 Raden, D., 99 Radtke, H. L., 207, 208 Randall, D. M., 19 Raven, B., 159

Read, S. J., 119 Reicher, S. D., 67, 97, 109, 126, 158, 160, 161, 164, 167, 185, 268, 285 Reid, A., 173 Reis, H. T., 15, 40 Rempel, J. K., 40 Reno, R. R., 125 Rhodewalt, F., 235 Richard, R., 20 Richardson, K. D., 219, 221 Richman, S. A., 193

< previous page

page_337

next page >

< previous page

page_338

next page > Page 338

Riecken, H., 229, 230, 231 Roberts, D. F., 245 Robertson, R., 35 Robinson, C. E., 106 Robinson, R. J., 110 Robinson, W. P., 2 Roemer, R., 245, 249 Rogers, R. W., 214, 215, 246 Roloff, M. E., 219 Rosch, E., 161 Rosenberg, M., 35 Rosenstock, I. M., 69 Rosenthal, D. A., 282 Ross, L., 110, 143, 188, 189, 192, 257, 266 Rothbart, M., 215 Roux, P., 249, 252, 254 Ruble, D. N., 111 Rucinski, D., 266, 268, 270, 278 Rudnick, J. R., 219 Rumelhart, D. E., 118 Russel, M. A. H., 247 Ryan, A., 33 Ryan, M. J., 51, 63, 69, 70, 71 S Sachdev, I., 275 Sagoff, M., 43 Salmon, C. T., 266, 268, 270, 278 Samuelson, C. D., 206 Sanbonmatsu, D. M., 59, 82, 117, 118 Sanchez-Mazas, M., 250, 255 Sande, G. N., 227, 231, 232, 243 Sartori, G., 276 Schachter, S., 157, 221, 229, 230, 231, 246 Schanck, R. L., 104, 106 Schank, R. C., 121 Scheibe, T., 34 Schelling, T. C., 41, 42 Scher, S. J., 229 Schlenker, B. R., 268 Schmidt, G. W., 284 Schnedlitz, P., 71, 91 Schoemaker, J. H., 33, 43 Schoenemann, T. J., 35

Schofield, J. W., 74 Schroeder, C. M., 106 Schuette, R. A., 59, 82 Schulte, F., 201 Schuman, H., 102, 108 Schumann, D. W., 193 Schwartz, J., 124 Schwartz, S. H., 15 Senour, M. N., 219 Sereno, K. K., 136 Settle, R. B., 217 Severson, H., 247 Shafir, E., 41 Shavitt, S., 43, 44 Sheatsley, P. B., 245 Sheeran, P., 61 Shepherd, G. J., 70 Shepherd, R., 22, 27, 36, 37, 38, 39, 40, 41, 45 Sheppard, B. H., 49 Sherif, M., 137, 138, 140, 157, 160, 175 Sherman, S. J., 95, 116, 231, 247, 248, 257, 259 Shimp, T. A., 50, 64 Shor, R. E., 207 Shrauger, J. S., 35 Shultz, T. R., 119 Sidera, J. A., 193 Siegel, A. E., 173, 178 Siegel, S., 173, 178 Silverman, E. S., 54, 55 Simon, B., 287, 290 Simons, H. W., 136 Simonson, I., 14 Sinclair, A., 233, 234 Singelis, T. M., 54 Skelly, J. J., 134, 169, 184, 194, 196, 265, 291 Skevington, S. M., 276 Skinner B. F., 205, 209 Skinner, W. F., 74 Slavings, R. L., 35, 36, 37 Smith, B. J., 246 Smith, E. R., 119, 121 Smith, J., 154 Smith, M. D., 54 Smith, R., 267 Smith, S. M., 183 Smolensky, P., 119

Snyder, C. R., 279 Snyder, M., 43, 44, 102, 178, 188, 189, 192, 193, 209, 215 Somoza, M. P., 122, 123 Spanos, N. P., 207, 208 Sparks, P., 22, 27, 32, 36, 37, 38, 39, 40, 41, 45 Speckart, G., 47, 49, 116, 126, 248 Spence, J. T., 286 Spranca, M. D., 14 Staats, A. W., 180 Staats, C. K., 180 Stanley, M. A., 246 Stapp, J., 286 Stasser, G., 87 Stasson, M. F., 56, 87 Statham, A., 203, 216, 224 Steele, C. M., 260 Steidley, T., 181 Stein, R. I., 38 Steketee, M., 276 Sternberg, R. J., 219 Stotland, E., 123

< previous page

page_338

next page >

< previous page

page_339

next page > Page 339

Stradling, S. G., 1, 7, 40 Strathman, A. J., 193 Strenta, A., 207 Stroebe, M. S., 246 Stroebe, W., 231, 246 Struckman-Johnson, C. J., 282 Struckman-Johnson, D. L., 282 Stryker, S., 42, 203, 216, 224 Sudgen, R., 14 Sutton, S. R., 246, 248 T Tabacalera, 247 Tajfel, H., 58, 67, 69, 72, 87, 109, 136, 172, 198, 219, 232, 235, 249, 250, 252, 263, 268, 269, 275, 284 Tanke, E. D., 209 Tanur, J., 204 Tassin, J. P., 249 Taylor, B. M., 276 Taylor, D. M., 275, 277 Taylor, S. E., 1, 2 Terry, D. J., 3, 34, 40, 43, 53, 58, 59, 63, 64, 68, 70, 75, 76, 78, 79, 81, 83, 87, 90, 91, 100, 109, 127, 128, 267, 270, 276, 278, 281, 282, 285 Tesser, A., 60, 61 Tessler, R., 15 Tetlock, P. E., 189 Thagard, P., 119 Thein, R. D., 143 Theodorakis, Y., 38 Thomas, E. J., 203 Thomas, W. I., 1, 219 Thompson, E., 247 Thompson, M. E., 207 Thompson, M. M., 41 Thorson, E., 266, 267, 285 Toch, H., 109, 110, 111, 112 Tordesillas, R., 249 Trafimow, D., 50, 51, 52, 53, 54, 55, 57, 58, 59, 61, 62, 63, 65, 125 Trafimow, J. H., 63 Triandis, H. C., 47, 49, 53, 55, 57, 59, 60, 63, 121, 248 Tully, J. C., 216 Turner, J. C., 2, 35, 58, 67, 69, 71, 72, 82, 87, 97, 109, 126, 136, 143, 153, 154, 158, 160, 161, 164, 166, 167, 169, 170, 172, 176, 178, 179, 185, 195, 205, 232, 249, 250, 252, 253, 257, 263, 266, 238, 269, 275, 284, 285 Turner M. E., 219 Turner, R., 203 Turtle, A., 282

U United States Department of Health and Human Services, 246, 247 Unnava, H. R., 198 Ury, W., 226 V Valantin, S., 205 Vallacher, R. R., 122 Valle, R. S., 217 Vallerand, R. J., 26, 50, 64, 70 Vallone, R., 266 van der Pligt, J., 14, 33, 248 van Knippenberg, D., 164, 165, 166, 167, 168, 169, 177, 194, 219, 266, 275, 291 Vangelisti, A. L., 219 Vanman, E. J., 119 Vaughan, G. M., 276 ven der Pligt, J., 20 Villarreal, A., 52 Vygotsky, L. S., 3, 24, 25, 26, 28, 29 W Wade, G., 275 Wagner, W., 136 Walster, E., 217 Wandersman, A., 20 Ward, A., 110 Ward, C. D., 214 Warner, L. G., 74 Warshaw, P. R., 49 Watkins, J. L., 218 Watson, J. B., 1 Weber, M., 216, 217 Wegener, D. T., 158, 162, 179, 181, 183, 194 Wegner, D. M., 122 Weiler, J., 204 Weinstein, E. A., 203, 204 Weinstein, N. D., 267 Wells, G. L., 180, 181 Wetherall, M. S., 67, 82, 97, 109, 126, 158, 160, 161, 164, 166, 167, 169, 185, 205, 268, 285 Wheatley, J. J., 135 Wheeler, S., 109, 111 White, J. D., 143 White, K. M., 3, 34, 40, 43, 53, 59, 70, 76, 78, 79, 81, 91 White, P. H., 179, 213 Whitlock, C. R., 201 Wicker, A. W., 67, 115

< previous page

page_340

next page > Page 340

Wicklund, R. A., 234 Wieringa, N., 37, 38, 39 Wiggins, E. C., 60 Wilder, D. A., 82, 143, 178, 179, 266, 291 Wiley, M. G., 204 Wilke, H., 164, 165, 166, 167, 168, 169, 177, 194, 219, 266, 275, 291 Wilkerson, J. C., 246 Williams, C. J., 109 Williams, J., 275 Williams, J. A., 286 Williams, J. E., 111, 286 Williams, K. D., 189 Williams, T., 212 Willover, D. J., 109, 112 Witte, K., 248 Wober, M., 246 Wood, W., 153, 158, 162, 217 Woodside, A. G., 136 Woolcock, J., 119, 121 Worchel, S., 229 Word, C. O., 209 World Health Organization, 245, 246, 247 Worth, L. T., 83, 92, 93, 149, 151, 154, 169, 184, 185, 192, 194, 195, 196, 219, 265, 291 Wright, S. C., 275 Wrong, D. H., 125 Y Yatko, R. J., 246 Ybarra, O., 55 Yi, Y., 57 Young, B., 116 Youngs, G. A., 207 Z Zajonc, R. B., 1 Zanna, M. P., 21, 40, 41, 82, 95, 209, 227, 231, 232, 234, 235, 243 Zarate, M. A., 121 Zeelenberg, M., 14 Zeitz, H., 267 Zimbardo, P. G., 206 Zimmerman, S., 87 Zimmermanns, N., 37, 38, 39 Zipp, J. F., 171, 172 Znaniecki, F., 1 Zuckerman, M., 15, 40

< previous page

page_341

next page > Page 341

SUBJECT INDEX A Action identification theory, 122-123 Affect, 60-62 Affect-cognition distinction, 60-62 Alcohol consumption, 21-22 Alter, 203 Altercasting, 6, 203-224 African Americans as source, 212-214 authority-agent, 216 charismatic-devotee, 217 children as source, 211-212 common role sets used in, 216-222, 206-207 defector-attentive confidant, 217 definition, 203 expert vs. attractive source, 214-215 expert-unknowing public, 217-218 fool-normal, 218 future research, 215-216, 218-219 helper-dependent, 219 hidden self technique, 207-208 how it works, 222-224 intimates (friends and lovers), 219, 220 just plain folks, 220 making roles/identities salient, 205-206, 204-208 morality and ethicality of, 224-226 new identity/role, use of, 207 opinion deviant-majority, 221 pre-existing roles, use of, 206-207 prophet-unwashed, 221 role making, 224 role playing, 208 scapegoat-sinners, 221-222 socially prestigious-admirer, 222 source credibility, 209-216 tact, 208-210 Altruistic behaviors, 15-16 Ambivalence, 41-42 Anticipated postbehavioral feelings, 14 Antismoking campaigns, 7 Antismoking campaigns; see smoking Antisocial behaviors, 16-18

Associated attributes, 234, 236-238 Attitude, 47, 49, 116 definition, 47, 49, 116 Attitude accessibility, 68, 77-80, 82, 92, 116-117 problems with research methodology, 117 Attitude change, 5, 107-108 trends in research, 5 stable behavior, 107-108 Attitude-behavior consistency, 4, 8, 57, 74-90, 92-93, 95-99, 101-102, 107-112, 245-264 effect of self-consciousness, 57 effect of self-monitoring, 57 future research, 92-93 hurdle, 98-99 impact of reference group norms, 74-90 past literature, 102 sources of inconsistency, 107-112 threshold, 98-99 Attitude-behavior inconsistency, 103-107 pluralistic ignorance, 103-107 Attitude-behavior relations, 35-37, 74-91 proposed reconceptualisation of norms, 74-91 self-identity, 35-37 Attitude-intention relations, 59-60 effect of habit, 59-60 Attitudes, 1-5, 8-9, 40-44, 50-51, 57-62, 96, 107-109, 115, 123-124 as defining identity, 9 as distinct from subjective norms, 50-51, 57 as predictors of behavior, 3

< previous page

page_341

next page >

< previous page

page_342

next page > Page 342

conflict of, 41-42 effect of sociocontextual variables, 58-59 energizing, 8 functions of, 43-44 inferred from behavior, 4 inferred from behavior, 96 influence of affect, 60-62 influence of cognition, 60-62 problems with measurement of, 40-42 temporal decoupling of behavior, 107-109 trends in research, 1, 2, 5, 115 tripartite model of, 123-124 utilitarian function, 44 Attitudinal congruence, 74 Attitudinal control, 50-60, 63 as determined by collective self, 53-55 as determined by culture, 55-57 as determined by private self, 53-55 determinants of, 53-60 individual differences in, 51-52 Autokinetic effect, 137, 157 B Behavior, 49-50, 60-63, 95, 107-109, 112, 115-118, 121-123 influence of affect, 60-62 influence of cognitions, 60-62 mental representations of, 121-123 prediction, 95, 112 priming of, 49 source of, 115-118 temporal decoupling of attitudes, 107-109 under attitudinal control, 50, 63 under normative control, 50, 63 Behavior change and stable attitudes, 108-109 Behavior generation, 131-33 mathematical representation, tensor product model, 131-133 Behavior generation model, 118-134 Behavior options; see mental representations of behavior Behavior specific referents, 62-63 Behavioral beliefs, 12-14, 32, 68-69 Behavioral decision theory, 43 Behavioral norms, 49, 125 Biased message processing, 188-193 future research, 193

role of identity, 188-193 Blood donation, 15 Business ethics, 19 C Categorization, 72 Category membership, 136 Central route processing, 5, 6, 8, 141-143, 145-152, 162-169, 180-185 consequences, 182-183 Classic cons, 201-203 bank examiner swindle, 202 payoff, the, 202 pyramid or ponzi scheme, the, 202-203 sweetheart scam, the, 202 three card monte, 201 Cognition, 60-62, 227 Cognition-affect distinction, 60-62 Cognitive developmental theory, 22-24, 28-29 Cognitive dissonance, 206, 228-244 future research, 243-244 groups within the laboratory, 231-232 hostility towards outgroup, 235-238 induced compliance, 228 repression, 238-242 responsibility diffusion, 230-232 social categories, 232-244 social identity, 232-244 social identity accessibility, 244 social identity processes, 243-244 within groups, 229-244 Cognitive dissonance theory, 7, 216, 227-229 Cognitive processing, 2 Cognitive-affective model, 177 Collective self, 3, 4, 53-55 Commitment, 42 Compliance, 139 Con, 203 alter, 203 ego, 203 Con artist, 201 Conflicting attitudes, 41-42 Conformity, 175 three process model, 175 Conformity, 137-138, 157, 254 Connectionism, 118-119 Connectionist approach, 115-133 Connectionist model of behavior generation, 118-134

Connectionist modeling, 5 Cons, 201-203 why they work, 203 Context representation, 124 Contingent consistency hypothesis, 71, 74, 76 Control beliefs, 32 Correlational consistency, 102-103, 112 effect of norms, 103 Culture, 55-57 D Decentration, 23 Definitional attributes, 234, 236-238

< previous page

page_342

next page >

< previous page

page_343

next page > Page 343

Deliberative decision making, 68 Deliberative processing, 82-85, 87-88 Depersonalization, 4, 109, 126, 269 Descriptive norm, 125 Differential action, 203-204 Diffusion of responsibility, 7, 230-232 Direct measures, 13, 14 Dishonest behaviors, 16-18 Dissonance reduction; see cognitive dissonance Dissonance; see cognitive dissonance theory Divisions of labor, 209 Downward comparison, 268 Dual process models of persuasion, 160-161 challenging, 160-161 E Eating behavior, 21-22 Ego, 203 Elaboration likelihood model (ELM), 6, 82, 142, 158, 162-164, 166, 169, 178-200, 215-216 biased message processing, 188-193 future research, 193, 196-200 impact of variables, 181 role of group membership, 185-187 role of identity, 180-811, 183-195 role of identity, high elaboration, 187-193 role of identity, low elaboration, 183-187 role of identity, moderate elaboration, 193-195 source attractiveness, 181 Employee behaviors, 18-19 Encoding, 130-131 mathematical representation, tensor product model, 130-131 Energizing, 8 Exemplar theory of social judgment, 121 F Focus theory, 125 Food, 3 Food choice, 37-38 self-identity, 37-38 Functions of attitudes, 43-44 G Gender stereotypes, 286 General referents, 62-63 Group identity, 172 Group locomotion, 138-139

Group mediated persuasion, 136-140, 142-155 definition, 136 early perspectives, 137-140 empirical support, 143-152 Group membership, 2, 3, 4, 68, 72 Group mobility, 275 Group norms, 2, 68, 74-90 Group polarization, 82 Group processes, 2 Group prototype, 78, 83, 97, 109, 111-112, 126, 161, 166-170, 176-178, 269 H Habit, 49, 59-60 Heuristic processing; see peripheral route processing Heuristic-systematic model, 82, 142, 158, 162-164, 169 Hidden self technique, 207-208 Hostile media bias; see hostile media phenomenon Hostile media phenomenon, 271-274, 276 I Identity, 6, 34-35, 172-174, 178-179, 183-195, 198-199 alternative explanations for impact on persuasion, 198-199 definition of, 172 empirical evidence for impact on persuasion, 178-179, 183-195 Operationalization of, 174 Identity conflict, 250-251, 253-256 Identity defense, 257-259 Identity protection, 253, 257 Identity theory, 42 Identity threat, 247-249, 252 Illegal behaviors, 16-18 Image theory, 43 Indirect measures, 13, 14 Individual mobility strategies, 275 Induced compliance, 228 Influence, 216-222 authority-agent, 216 charismatic-devotee, 217 common role sets used in, 216-222 defector-attentive confidant, 217 expert-unknowing public, 217-218 fool-normal, 218 granfallloon, 218-219 helper-dependent, 219 intimates (friends and lovers), 219 intimidator-coward, 220

< previous page

page_344

next page > Page 344

just plain folks, 220 opinion deviant-majority, 221 prophet-unwashed, 221 scapegoat-sinners, 221-222 socially prestigious-admirer 222 Influence internalization, 140 Informational influence, 139, 142, 157-158, 158-160 Ingroup norms, 92-93 future research, 92-93 Injunctive norm, 125 Inner speech, 25 Intention, 11, 49 Intergroup conflict, 251-253 Internalization; see also central route processing, 141 L Learning, 130-131 mathematical representation, tensor product model, 130-131 Literal consistency-effect of norms, 103 Literal consistency, 102 M Mand, 205 definition, 205 Manded altercasting, 204-208 co-optation, 206-207 effectiveness of, 205-208 hidden self technique, 207-208 making roles/identities salient, 205-206 new identity/role, use of, 207 pre-existing roles, use of, 206-207 role playing, 208 Mass media, 265 Mass persuasion, 6-7 Material self, 34 Matrix, 130 Mental representations of behavior, 121-123 Message quality, 143-153 Modally salient beliefs, 14 MODE model, 59, 82-90, 92, 117 Mode of decision making, 82-90, 92, 117 Moral dilemma, 12 Moral norms, 2, 3, 11-30, 40 altruistic behaviors, 15-16 as predictors of intention, 15-22, 26-30, 40

business ethics, 19 development of, 22-26 eating and drinking behavior, 21-22 employee behaviors, 18-19 future research, 28 illegal, antisocial, or dishonest behaviors, 16-18 internalization of; see norms-internalization of research evidence, 15-22, 26-30 sexual behavior, 20-21 Moral socialization theory, 22-24, 28-29 Motivation to comply, 12, 14, 32, 68-69 N Norm congruency, 74-90 Normative behavior, 4 Normative beliefs, 12, 14, 32, 68-69 Normative control, 50-60, 63 as determined by collective self, 53-55 as determined by culture, 55-57 as determined by private self, 53-55 determinants of, 53-60 individual differences in, 51-52 Normative influence, 139, 158-160 Normative social influence, 175 Norms, 2-3, 23-26, 28-29, 36, 67-72, 92-93, 100, 266 as predictors of behavior, 3 definition, 69 future research, 92-93 internalization of, 23-26, 28-29 lack of support for, 69 need to reconceptualize, 69, 100 O Obedience, 206 Object representation, 123-124 Optimal distinctiveness, 196 Optimal distinctiveness theory, 274 Outcome evaluations, 12, 14, 32, 68-69 Outgroup hostility-cognitive dissonance, 235-238 Outgroup stereotyping; see stereotyping P Parallel processing, 129 Perceived behavioral control, 32, 49, 63, 68-69, 75-76 Perceived self-efficacy, 63 Peripheral route processing, 5, 6, 8, 142-143, 145-152, 162-169, 180-185 consequences, 182-183 Personal identity, 173

Persuasion, 5-6, 135-136, 139-141, 143-155, 160-164, 166-172, 174-179, 182-200

< previous page

page_344

next page >

< previous page

page_345

next page > Page 345

alternative explanations for the influence of identity, 198-199 cognitive approaches, 141 determining route of, 182 empirical evidence for impact of social identification, 166-168 empirical evidence for role of identity, 178-179, 183-195 future research, 154-155, 170, 195 identification, 139, 141, 152 lack of focus on groups in research, 140-141 listing thoughts generated, 182 message processing 143-153 message quality, 143-153 motivation to process, 163 previous research, 135-136 prototypicality, impact of, 161-164 quality of argument, 182 reference groups, 136 role of identity, 171-172 role of identity, future research on, 196-200 role of identity, theories explaining, 174-178 self-categorizational approach, 160-161, 169 trends in research, 5, 162 Persuasion, group mediated; see group mediated persuasion Persuasive messages, 5, 265 impact on others, 265 Pluralistic ignorance, 4, 96, 103-112 among children, 110-111 dispelling of, 105-107 gender roles, 110-111 Positive distinctiveness, 284 Postbehavioral feelings, 14 Precommitting devices, 42 Prescriptive norms, 36, 125 Previous behavior, 49 Priming, 49 Primus inter pare effect, 291 Private attitude, 102 Private self, 3, 4, 53-55 Prototype; see group prototype Prototypicality, 164-168 as a moderator of systematic processing, 164-168 empirical support for influence on processing, 164-168 Psychological reactance paradigm, 257 Public behavior, 96-103

Pure ego, 34 Q Quality of argument, 182 R Reference group norms, 74-90 Reference group theory, 173 Reference groups, 136, 138, 268 comparative function, 138 normative function, 138 Referent informational influence, 72, 176-177 Referents, 62-63 Repression, 239-242 cognitive dissonance, 238-242 Response latencies, 77 Responsibility diffusion, 230-232 Reverse third person effect, 285 Role, 203 definition, 203 Role making, 224 Role playing, 208 Route of processing, 182-183 consequences, 182-183 S Salience, 42 Salience of group membership, 75-80, 82-83, 87-88, 90-91, 270, 283, 288-290 Same category sources; see group mediated persuasion Scams; see cons Schema, 121-122 Script, 121-122 Selective exposure, 203 Selective perception, 77 Self, 34-35 Self concept, 35 Self perception, 35 Self presentation, 96-97 Self-affirmation, 260 Self-categorization, 4, 5, 72 Self-categorization theory, 67-69, 72-76, 78, 91-93, 125-126, 158, 268-270, 272, 282-283, 285-286 Self-concept, 3 Self-consciousness, 57 Self-discrepancy theory, 197 Self-enhancement, 72 Self-esteem hypothesis, 284 Self-fulfilling prophecy, 203 Self-identity, 3, 8, 34-43, 45-46

attitude-behavior relations, 35-37 choice of food, 37-38 conflicting facets, 42-43 explaining the effects of, 40-41 future research, 45-46 problems with measures, 39

< previous page

page_345

next page >

< previous page

page_346

next page > Page 346

research evidence, 35-37 Self-monitoring, 57 Self-serving biases, 268 Sexual behavior, 20-21 Smoking, 245-64 conformity, 254 factors maintaining the behavior, 249 identity conflict, 250-251, 253-256 identity conflict, coping with, 260-263 identity defense, 257-259 identity protection, 253 identity threat, 247-249, 252 intergroup conflict, 251-253 self-affirmation, 260 social influence, 247-249, 253-256 social mobility, 250 social representations, 251-253 social support, 258-259 Smoking; see anti-smoking campaigns Social categories, 232-244 associated attributes, 234, 236-238 definitional attributes, 234, 236-238 Social change, 275 Social cognition, 2, 5 Social identit,y 2, 4, 5, 109-112, 172, 204, 232-244, 266 accessibility, 244 cognitive dissonance, 232-244 Social identity theory, 35, 58-59, 67-69, 72-76, 78, 80, 83, 90-93, 97, 172-173, 232-235, 268-270, 274-275, 277, 279, 283-286, 288, 290 Social influence; see also persuasion, 5, 71, 72, 137, 157-158, 160-161, 169, 247-249, 253-256 Self-categorizational approach, 160-161, 169 Social judgment theory, 175 Social mobility, 250 Social norms, 5, 71, 97 Social perceivers, 96 Social psychology, 1, 2 Social reality, 138-139, 143 Social reality testing, 174, 176-177 Social representations, 251-253 Smoking, 251-253 Social role, 203 Social roles, 216-222 authority-agent, 216 charismatic-devotee, 217

defector-attentive confidant, 217 expert-unknowing public, 217-218 fool-normal, 218 granfalloon, 218-219 helper-dependent, 219 intimates (friends and lovers), 219 intimidator-coward, 220 just plain folks, 220 opinion deviant-majority, 221 prophet-unwashed, 221 scapegoat-sinners, 221-222 Socially prestigious-admirer, 222 Social self, 34 Social support, 258-259 Social support, 7 Socially dependent influence; see normative influence Socially independent influence; see informational influence Societal norms, 124-125 definitions, 124 past research, 124-125 Sociocontextual variables, 58-59 Sociocultural theory, 24, 25 Sociological ambivalence, 42-43 Source attractiveness, 181 Source characteristics, 142 Source credibility, 209-216 African Americans, 212-214 altercasting, 209-10 children, 211-212 expert vs. attractive source, 214-215 future research, 215-216 reward/reinforcement model, 209-210 three factor model, 209-210 Spiritual self, 34 Spontaneous decision making, 68 Spontaneous processing, 82-85, 87-88 Stanford prison experiment, 206 Status protection, 275 Stereotyping, 110-111 gender, 110-111 Strength of group identification, 75-77, 83, 85, 87-91 Subjective expected utility, 33-34, 43 limited descriptive value of approach, 43 Subjective expected utility-based (SEU) attitude-behavior models, 31-46 Subjective norm, 12-13, 15, 32, 47, 49-51, 57-60, 68-73, 125 as distinct from attitudes, 50, 51, 57

effect of self-consciousness, 57 effect of self-monitoring, 57 effect of sociocontextual variables, 58-59 Subjective norm-intention relations, 59-60 effect of habit, 59-60 Subjective value; see utility Symbolic attitudes, 39-40 Systematic processing; see also central route processing, 164-168 empirical evidence for impact of prototypicality, 164-168 as a moderator of, 164-168

< previous page

page_346

next page >

< previous page

page_347 Page 347

T Tact, 209 definition, 209 Tact altercasting, 208-210 source credibility, 209-210 Tensor, 119, 130 Tensor product model, 5, 118-121, 123, 125-133 advantages of, 119 assumptions of, 123 behavior generation, mathematical representation of, 131-133 direct experience, an explanation for the effects of, 127 encoding, mathematical representation of, 130-131 future research, 129 group norms, an explanation for the effects of, 127-128 implications of, 126-128 learning, mathematical representation of, 130-131 mathematical details, 130-133 mathematical representation of an event, 120-121 notations, 130 past behavior, an explanation for the impact of, 126 predictions of, 121 Theory of planned behavior, 11, 14-22, 27-46, 64-65, 67-90, 99 control beliefs, 32 future research, 64-65 impact of reference group norms, 74-90 limitations of normative component, 69-74 moral norms, 15-22, 27-30 perceived behavioral control, 32 self-identity as a predictor of intention, 31-46 SEU attitude-behavior models, 31-46 strength of group identification, 74-90 Theory of reasoned action, 11-22, 26-47, 64-65, 67, 68-90, 99, 115-117, 121, 125 future research, 64-65 impact of reference group norms, 74-90 limitations of normative component, 69-74 moral norms, 15-22, 26-30 problems with standard research methodology, 116-117 self-identity as a predictor of intention, 31-46 SEU attitude-behavior models, 31-46 strength of group identification, 74-90 subsequently added variables, 47, 64 Third person effect, 7-8, 29, 265-291 AIDS advertisements, 281-285

behavior, 288-90 commercial advertising, 285-288 definition, 266 desirability of persuasion, 281-285 dominant vs. subordinate groups, 274-281, 285-828 downward comparison, 268 election campaigning, 270-281 future research, 29 gender identity, 285-288 group mobility, 275 individual mobility strategies, 275 ingroup vs. outgroup members, 271-274, 277-279 intergroup context, 274-281 level of identification, 270-281 life imprisonment, 288-290 nature of comparison other, 267-268, 271-274 negative vs. positive content, 267 positive distinctiveness, 284 primus inter pare effect, 291 reference groups, 268 reversal, 285 self-esteem hypothesis, 284 self-serving biases, 268 social change, 275 social identity theory, 274-281 social psychological basis, 290 status protection, 275 support for social-normative perspective, 270-290 ultimate attribution error, 276-277 Third person perceptions basis of, 267 Thought generation listing, 182 Three process model, 175 Tobacco; see smoking Triangular theory of love, 219 Tripartite model of attitudes, 123-124 U Ultimate attribution error, 276-277 Utilitarian function of attitudes, 44 Utility, 33 W When prophecy fails, 229

< previous page

page_347