Towards an Interdisciplinary Perspective on the Life Course, Volume 10 (Advances in Life Course Research)

TOWARDS AN INTERDISCIPLINARY PERSPECTIVE ON THE LIFE COURSE i ADVANCES IN LIFE COURSE RESEARCH Series Editor: Timoth...

Author: René Levy | Paolo Ghisletta | Jean-Marie Le Goff | Dario Spini | Eric Widmer

55 downloads 1222 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

TOWARDS AN INTERDISCIPLINARY PERSPECTIVE ON THE LIFE COURSE

i

ADVANCES IN LIFE COURSE RESEARCH Series Editor: Timothy Owens Recent Volumes: Volume 1:

Work, Retirement and Social Policy, 1986 Edited by Zena Smith Blau

Volume 2: Family Relations in Life Course Perspective, 1986 Edited by David I. Kertzer Volume 3: Personal History Through the Life Course, 1993 Edited by R. S. Olusegun Volume 4: Delinquency and Disrepute in the Life Course, 1995 Edited by Zena Smith Blau and John Hagen Volume 5: Self and Identity Through the Life Course in Cross-Cultural Perspective, 2000 Edited by Timothy J. Owens Volume 6: Children at the Millennium: Where Have We Come From, Where Are We Going?, 2001 Edited by Sandra L. Hofferth and Timothy J. Owens Volume 7: New Frontiers in Socialization, 2002 Edited by Richard A. Settersen, Jr. and Timothy J. Owens Volume 8:

Changing Life Patterns in Western Industrial Societies, 2004 Edited by Janet Zollinger Giele and Elke Holst Volume 9: The Structure of the Life Course: Standardized? Individualized? Differentiated?, 2005 Edited by Ross Macmillan ii

ADVANCES IN LIFE COURSE RESEARCH VOLUME 10

TOWARDS AN INTERDISCIPLINARY PERSPECTIVE ON THE LIFE COURSE EDITED BY

RENE´ LEVY Universite´ de Lausanne

PAOLO GHISLETTA Universite´ de Gene`ve

JEAN-MARIE LE GOFF Universite´ de Lausanne

DARIO SPINI Universite´ de Lausanne

ERIC WIDMER Universite´ de Lausanne

Amsterdam – Boston – Heidelberg – London – New York – Oxford Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo iii

ELSEVIER B.V. Radarweg 29 P.O. Box 211 1000 AE Amsterdam, The Netherlands

ELSEVIER Inc. 525 B Street, Suite 1900 San Diego CA 92101-4495 USA

ELSEVIER Ltd The Boulevard, Langford Lane, Kidlington Oxford OX5 1GB UK

ELSEVIER Ltd 84 Theobalds Road London WC1X 8RR UK

r 2005 Elsevier Ltd. All rights reserved. This work is protected under copyright by Elsevier Ltd, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax (+44) 1865 853333, e-mail: [email protected]. Requests may also be completed on-line via the Elsevier homepage (http://www.elsevier.com/locate/permissions). In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of the Publisher is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. First edition 2005 British Library Cataloguing in Publication Data A catalogue record is available from the British Library. ISBN: 0-7623-1251-3 ISSN: 1040-2608 (Series)

∞ The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.

Working together to grow libraries in developing countries www.elsevier.com | www.bookaid.org | www.sabre.org

iv

CONTENTS PREFACE

ix

LIST OF CONTRIBUTORS

xi

INTRODUCTION WHY LOOK AT LIFE COURSES IN AN INTERDISCIPLINARY PERSPECTIVE? Rene´ Levy and the Pavie Team

3

PART I: AGENCY AND STRUCTURE STRUCTURE, AGENCY, AND THE SPACE BETWEEN: ON THE CHALLENGES AND CONTRADICTIONS OF A BLENDED VIEW OF THE LIFE COURSE Richard A. Settersten, Jr. and Lynn Gannon

35

AGENCY, EVENTS, AND STRUCTURE AT THE END OF THE LIFE COURSE Victor W. Marshall

57

LOOKING AT AMBIVALENCES: THE CONTRIBUTION OF A ‘‘NEW-OLD’’ VIEW OF INTERGENERATIONAL RELATIONS TO THE STUDY OF THE LIFE COURSE Kurt Lu¨scher

93

v

vi

CONTENTS

PART II: TRANSITIONS AGENCY AND STRUCTURE IN EDUCATIONAL ATTAINMENT AND THE TRANSITION TO ADULTHOOD Jeylan T. Mortimer, Jeremy Staff and Jennifer C. Lee

131

NON-NORMATIVE LIFE COURSE TRANSITIONS: REFLECTIONS ON THE SIGNIFICANCE OF DEMOGRAPHIC EVENTS ON LIVES Frank F. Furstenberg

155

THE SECRET OF TRANSITIONS: THE INTERPLAY OF COMPLEXITY AND REDUCTION IN LIFE COURSE ANALYSIS Katherine Bird and Helga Kru¨ger

173

PART III: BIOGRAPHICAL RECONSTRUCTION LIFE COURSE TRANSITIONS AND SOCIAL IDENTITY CHANGE Nicholas Emler

197

THE IMPACT OF PERSONALITY AND LIVING CONTEXT ON REMEMBERING BIOGRAPHICAL TRANSITIONS Pasqualina Perrig-Chiello and Walter J. Perrig

217

STUDYING LIVES IN TIME: A NARRATIVE APPROACH Dan P. McAdams

237

vii

Contents

PART IV: METHODOLOGICAL INNOVATIONS LIFE COURSE ANALYSIS: TWO (COMPLEMENTARY) CULTURES? SOME REFLECTIONS WITH EXAMPLES FROM THE ANALYSIS OF THE TRANSITION TO ADULTHOOD Francesco C. Billari

261

LIFE COURSE DATA IN DEMOGRAPHY AND SOCIAL SCIENCES: STATISTICAL AND DATAMINING APPROACHES Gilbert Ritschard and Michel Oris

283

FIVE STEPS IN LATENT CURVE MODELING WITH LONGITUDINAL LIFE-SPAN DATA John J. McArdle

315

AFTERTHOUGHTS INCITATIONS FOR INTERDISCIPLINARITY IN LIFE COURSE RESEARCH Rene´ Levy, Paolo Ghisletta, Jean-Marie Le Goff, Dario Spini and Eric Widmer

361

This page intentionally left blank

viii

PREFACE This volume has grown out of an international colloquium on interdisciplinarity in life course research organized by the Center for life course and life style studies (PAVIE) at the Universities of Lausanne and Geneva (Switzerland) in October 2003. The contributors were asked to react to a working document prepared by the organizers but had entire freedom as to how they wanted to do it. The authors represent four disciplines directly involved in life course research: developmental and personality psychology, social psychology, sociology, and social demography. The focus on these four social-science disciplines in a venture into interdisciplinarity should certainly not be read as a programmatic limitation of our wished-for disciplinary scope of collaboration. On the contrary, we fully acknowledge that other disciplines have made important contributions to life course research and that they are by no means located at its margins – let us only think of social history or political science. The potential role of biology and social medicine may be more controversial, but there again, our selection is pragmatic and provisional, not exclusive and even less ideological. The aim of this volume is to contribute to the enlargement and strengthening of collaboration between all disciplines able to develop an interest in life course or life span research. The colloquium as well as some of the subsequent work necessary for this volume to materialize has been financed by the Swiss National Science Foundation, and a series of services and facilities have been contributed by the University and the Municipality of Lausanne, an indispensable help for which we are very grateful. Beyond the contributors who are, of course, the main actors of this volume, we owe special thanks to Tatiana Lazzaro who did a great job all along this enterprise, beginning with the organization of the colloquium and ending with a large amount of administrative and editing services she delivered with exceptional enthusiasm, competence and efficacy. We also thank Tim Owens, the editor of this series, who was highly interested in our endeavor from the outset and extremely helpful for its taking the shape our readers can appreciate now.

Rene´ Levy, Paolo Ghisletta, Jean-Marie Le Goff, Dario Spini and Eric Widmer (Editors) Lausanne/Geneva ix


x

LIST OF CONTRIBUTORS Francesco C. Billari

Istituto di Metodi Quantitativi, Universita` Bocconi and Innocenzo Gasparini Institute for Economic Research, Milano, Italy

Katherine Bird

Universita¨t Bremen, EMPAS, Bremen, Germany

Nicholas Emler

Department of Psychology, University of Surrey, Guildford, UK

Frank F. Furstenberg

University of Pennsylvania, Department of Sociology, Philadelphia, PA, USA

Lynn Gannon

Department of Sociology, Case Western Reserve University, Cleveland, OH, USA

Paolo Ghisletta

Centre Interfacultaire de Ge´rontologie (CIG) et Faculte´ de Psychologie et des Sciences de I’e´ducation, Universite´ de Gene`ve, Centre le´manique d’e´tude des parcours et modes de vie, Universite´s de Lausanne et de Gene`ve, Gene`ve, Switzerland

Helga Kru¨ger

Universita¨t Bremen, EMPAS, Bremen, Germany

Jennifer C. Lee

Life Course Center, Department of Sociology, University of Minnesota, Minneapolis, MN, USA

Jean-Marie Le Goff

Centre le´manique d’e´tude des parcours et modes de vie, Universite´ de Lausanne, Lausanne, Switzerland

Rene´ Levy

Centre le´manique d’e´tude des parcours et modes de vie, Universite´ de Lausanne, Lausanne, Switzerland xi

xii

LIST OF CONTRIBUTORS

Kurt Lu¨scher

Fachgruppe Soziologie, Universita¨t Konstanz, Germany

Victor W. Marshall

UNC Institute on Aging, University of North Carolina at Chapel Hill, NC, USA

Dan P. McAdams

Northwestern University, Evanston, IL, USA

John J. McArdle

Department of Psychology, University of Virginia, Charlottesville, VA, USA

Jeylan T. Mortimer

Life Course Center, Department of Sociology, University of Minnesota, Minneapolis, MN, USA

Michel Oris

Department of Economic History, University of Geneva, Geneva, Switzerland

Walter J. Perrig

Institut fu¨r Psychologie, Universita¨t Bern, Bern, Switzerland

Pasqualina PerrigChiello

Institut fu¨r Psychologie, Universita¨t Bern, Bern, Switzerland

Gilbert Ritschard

Department of Econometrics, University of Geneva, Geneva, Switzerland

Richard A. Settersten, Jr.

Department of Sociology, Case Western Reserve University, Cleveland, OH, USA

Dario Spini


Jeremy Staff

Department of Sociology and Crime, Law and Justice, The Pennsylvania State University, University Park, PA, USA

Eric Widmer


INTRODUCTION

1


2

WHY LOOK AT LIFE COURSES IN AN INTERDISCIPLINARY PERSPECTIVE? Rene´ Levy and the Pavie Team1 CHALLENGES OF LIFE COURSE RESEARCH After decades of a rather marginal existence and little coherence in its development, life course research is definitely coming of age. There recent signs of consolidation and first attempts at reaping the scattered harvest of research in various disciplines, especially in the form of a first handbook of life course research (Mortimer & Shanahan, 2003), of first attempts at interdisciplinary dialogue around specific approaches such as Baltes’ lifespan psychology2 (Staudinger & Lindenberger, 2003) and of a specialized annual review (in which this volume is published). Nevertheless, life course scholars still seem to be a small handful ‘‘digging’’ on the fringe of their disciplinary mainstreams, as yet with little influence on more established fields of research. Why are we, i.e., life course researchers, so keen on life courses? What is there so special about life course research? We feel in fact that there are a number of specific challenges life course researchers have to confront and answer. A first, global and not very differentiated reason for finding it particularly interesting could be that it encompasses all we find important about human life and that everything humanly relevant is in the life course. Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 3–32 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10014-8

3

4

RENE´ LEVY AND THE PAVIE TEAM

A second answer, of the same global kind but a bit more specific, could be that only a life course approach takes fully into account the fact that our lives are ongoing processes and not just single states or events that can be adequately captured and understood using snapshots. A third reason of attractiveness, even more fascinating but also demanding, is the cross-cutting and integrative nature of the life course perspective with respect to most of our more conventional, institutionalized disciplinary specialties. To illustrate this statement, take the case of sociology.3 There are special sociologies of childhood, of youth or adolescence, (maybe soon also of post-adolescence,) of retirement, and of aging. Each ‘‘age group’’ is treated in a more or less static perspective and therefore reified as a somehow homogenous social category and not as a phase in the process of life course unfolding. There are special sociologies of the family, of work, of labor markets, of leisure, of stratification and mobility, of voluntary associations, of social movements and so on. Each social field is mainly treated as an isolated entity with its inner logic and specific pathways for the individuals who participate in these fields. Life-phase specific and sector specific sociologies have developed their specialized perspectives, slicing up the lives and contexts of the ‘‘whole individuals’’ we pretend to be into various aspects, types of social relations, phases, or fields of action. Likewise, in psychology, ontogeny has been regarded by many as occurring almost differently in distinct fragments of the life span. Much as in sociology, the American Psychological Association, among its 53 divisions, has Division 7 for Developmental Psychology and Division 20 for Adult Development and Aging. Similarly, Division 12 is for the Society of Clinical Psychology, while Division 53 for the Society of Clinical Child and Adolescent Psychology. Again, the fragmentation of the life course has been, for some aspects at least, institutionalized. Against this reductionistic, fragmenting tendency, Settersten (2003b, p. 196) has rightly made a strong point of ‘‘The importance of understanding people in whole (over time) and as wholes (studying larger profiles of traits and characteristics rather than single variables)’’. Without an encompassing perspective, life course research cannot meet the challenge of its very raison d’eˆtre. On a more fundamental level, there is a fourth challenge which is already included in these remarks but not explicitly spelled out: life course analysis is one of the rare ways social sciences have developed so far to conceptualize time, not just as a physical happening whose whereabouts escape our theoretical understanding, but as something that is culturally, socially and also individually molded, reworked and ‘‘constructed’’, something that has both a subjective and an objective existence, and whose very objectivity is the

Why Look at Life Courses in an Interdisciplinary Perspective?

5

result of social objectiveness rather than of mere ‘‘being there’’ like the eternal rhythm of atomic oscillation. Settersten (1999) has entitled his recent stock-taking book ‘‘Lives in Time and Place’’, referring to two basic dimensions of the real world that both are not just physical ‘‘givens’’, but socially constructed: social space and social time. In classical developmental psychology, most research was concerned with charting along the age axis the newly acquired skills and mastered developmental tasks by children. Refined descriptions of what the child could achieve at what age flourished, and in various domains age-related norms were established. Wohlwill (1970) was one of the first to question the limitations of the mere descriptive role of chronological age. (We shall take up the question of what status to give to age in our theorizing about life courses in the final chapter.) In short, in both a humanistic and in a scientific view, life course research is fascinating because it forces us to break down traditional limitations of understanding.

THE NECESSITY OF INTERDISCIPLINARITY If adopting a life course perspective in a single study does not exclude focusing on a specific aspect of peoples’ lives, the perspective as such is necessarily integrative: it has to bridge a number of institutionalized chasms in and between social science disciplines. These chasms are sometimes openly discussed, even questioned, and sometimes rather taken for granted. At any rate they tend to limit our awareness, inducing what we may call an internalistic perspective, artificially bounded by the boundaries of the specialty in which our scholarly activity is organized. We have already mentioned the two dimensions life course researchers have to bridge in their work, the age-group axis and the life-spheres axis. At least three others are of equal importance: What sociologists, and others as well, are used to call the macro–micro link, i.e., more generally, the fact that one basic dimension in the complexity of reality is the nesting of encompassing systems of various levels;4 this dimension (we might call it systemic differentiation to distinguish it from two others, hierarchization or parallelism) takes all its importance in the debate about the institutionalization of life course patterns as opposed to actorial agency although it is more often invoked as important than really taken into account. Gender differences or more generally the social relationships between the sexes,5 but also a number of other cases of strong social differentiation of both a cultural and a structural kind that have strong incidences on the

6


level of individual identities (like ethnic or religious group membership, etc.), that are often leveled out by static analyses or even ignored by genderblind or more generally ‘‘differentiation-blind’’ theoretical models.6 Linked lives, the strong interdependencies between the life courses of related persons that remind us that the real object of most of the social sciences are relations between actors and not just monadic individuals as some current mainstream paradigms would have it.7 If we are to bridge successfully all these gaps, with their thematical, conceptual and methodological implications, one may object that we have to become scientific geniuses. As social scientists we know that individual mastery is not all, that groups are frequently wiser than the ‘‘addition’’ of their individual members, so teamwork is certainly a way out of this problem. Working integratively in the sense just mentioned calls for teamwork. Among the gaps that work on life courses intrinsically needs to bridge figures also the one between scientific disciplines, above all in the social sciences (including, of course, psychology). The arguments for interdisciplinarity in this area may not be fundamentally different from those applying to other themes – as soon as we get interested in a real world problem and not only an epistemological slice of it, we cannot be satisfied by adopting one disciplinary perspective only. Most often, interdisciplinarity is claimed for problem-oriented or ‘‘applied’’ research. If the research objective is to resolve a problem, disciplinary purity and coherence is of no interest, we mobilize all promising resources whatever their disciplinary origin and theoretical ramifications. This may not be the primary orientation of life course research. But there is another important and sufficient motive for interdisciplinarity in this area that we have already encountered: the holist perspective on individuals and their development in social context which cannot be contented by any single aspect alone, be it sectoral or disciplinary. It is no accident that interdisciplinarity is frequently called for in this area. When formulating his ‘‘life course paradigm’’, Elder (1995) included interdisciplinarity explicitly, but the overriding research praxis remains monodisciplinary.8 A further question is then: what should we mean by interdisciplinarity? We usually use the term in a very general sense, meaning just going beyond the limits of any one of our habitual disciplines (monodisciplinarity). In order to be more precise, we may join an emergent distinction between multidisciplinarity (or pluridisciplinarity), meaning the cooperation of several disciplines that work on complementary aspects of a common overall question,


7

interdisciplinarity, meaning the integrated cooperation of disciplines on the basis of common concepts, and transdisciplinarity, meaning the scientific work on the basis of common concepts that do not belong to any specific discipline and that may also include the points of view of extra-scientific actors concerned by the question under study.9 There is no general, absolute value hierarchy to be established between these possibilities, ‘‘trans’’ is not necessarily better than ‘‘multi’’ and there may very well be problems whose ‘‘mono’’ study is quite adequate. The proper level between mono- and transdisciplinarity must be identified for each research enterprise, and the optimum may very well be an evolving mix and not a one-level solution fixed once and forever. But what is certainly necessary is the capacity of people and teams to work on these different levels. In order for such an enterprise to work, a more or less clearly stated common goal is of course essential, but other conditions are of equal importance: the real desire of the participants to work together, their acknowledgment of the potential of mutual learning rather than competition, even less the pursuit of disciplinary hegemony, and a fundamental working knowledge allowing for an understanding and appreciation of the conceptual and methodological panoply of neighboring disciplines. There is a huge gap between the almost unisonous choir of public voices asking for interdisciplinarity, and scientific everyday practice that is needed to discover the real difficulties of such an endeavor to begin with. A realistic approach must probably start from the principle with a view to interdisciplinary convergence or at least cooperation, differences should not be silenced, but discussed and integrated. Well-placed protagonists assure us that this is no easy business – witness, as one example, Mayer (2002) who coolly states on the encounter of life course sociology and life-span psychology (in which he has been engaged himself): ‘‘In retrospect, despite all the strong mutual recognition and reinforcement, surprisingly little convergence and integration has actually occurred’’. There is definitely still some way to make; the present volume is meant to bring us a step ahead on this way.

DIFFICULTIES OF INTERDISCIPLINARITY As long as not many social scientists themselves have a really interdisciplinary education and working profile, interdisciplinary work is bound to be teamwork. Along with more current organizational questions, interdisciplinary

8


teams have to confront quite real problems in their everyday functioning, even if these may often seem somewhat trivial when looked at from a theoretical vantage point. These practical difficulties need to be taken seriously in order to overcome them. Let us briefly look into some of them. But an assured rootedness in their own disciplines of scholars working together in an interdisciplinary team is most probably a precondition for successful interdisciplinarity, even though this may appear as a paradox. A first kind of difficulty to be clarified and surpassed on the way to interdisciplinary teamwork is in the area of vocabulary. Our disciplines use some identical terms with quite different meanings (e.g., ‘‘norm’’ has an entirely cultural, obligational ring in sociology, but the objectivist meaning of statistical prevalence – or even biological ‘‘normality’’ – in developmental psychology; or the different semantic extension of the notion of ‘‘micro’’ and ‘‘macro’’ between the same two disciplines). The other way round, we sometimes use different terms for the same things (e.g., in the methodological area). One example seems to pose few problems: the more sociological or demographical term of life course seems to overlap very largely with the more psychological one of life span. We therefore consider them in this chapter as synonyms. Another, probably minor, example is different names used in different disciplines for the same methods (e.g., survival analysis, event history analysis, or, in French, ‘‘demographical analysis of biographies’’10 in the current terminology of demography all designate the same or at least highly similar analytical methods; the same is true for multilevel models – random effect models – mixed effect models and also hierarchical models). Other differences, that cannot be ruled out by mere lexical agreements, are related to the typical frames of attention of the disciplines. Mainstream (or traditional) developmental psychology focuses largely on very early (and, to a lesser extent, very old) years of life while the bulk of sociological work concerns the adult years (mostly without spelling out explicitly this horizon). At first sight, social psychology does not seem to be very specific in this respect; in fact, it shows only rarely an outspoken awareness of and interest in processes of growing older or in the life course. But there is a hidden and rarely acknowledged specificity inherent in much research in this discipline: a large fraction of empirical research in social psychology, at least about attitudes and representations, is done with the most accessible category of experimental subjects, i.e., with students, who represent quite a specific age group as well as a set of rather specific locations in social stratification. Little is known about the impact of this double background specificity on the results and scientific knowledge in this discipline. Two transitions have


9

attracted the attention of social demographers, the first from adolescence to adulthood and the second the end of life (mortality). There are, of course, many conceptual differences that may constitute as many difficulties for mutual understanding and cooperation. Let us only mention one example: the weight that different social science disciplines usually attribute to biology vs. environment (or nature vs. nurture in a more traditional formulation). While psychology has largely put aside earlier apprehensions with respect to close contact with biology, sociology and even more political sciences or social and cultural anthropology have up to now maintained a greater distance; demography is hard to locate on this dimension, but would certainly have little hesitations to be situated near psychology, and potentially even could stretch its interests as far as to reach not only psychology or sociology, but also biology in the sense that it is currently interested in biologically structured events like birth and death. This basic opposition has a close relative in the difference between internal and external explanations of the ‘‘socialization vs. institutional channeling’’ variety (e.g., concerning the probable impact of age-normative conceptions vs. context or environmental conditions). Confronted to such differences in basic assumptions that characterize the often implicit scientific cultures of different disciplines, interdisciplinary collaboration needs above all the renouncement of rigorous a priori stands and a pragmatic curiosity for the other disciplines’ way of looking at things. Another important and potentially dividing difference concerns the distinction between factual and representational aspects that may be highlighted by the far-reaching attachment to statistical description by demography on the one hand and the heavy emphasis of most of social psychology on individual and collective representations, on the other. If we push this line of thinking further, there are of course situations where it may become very difficult to find common ground between disciplines, like psychology and demography, because they are normally interested in many different topics – but then, we are also leaving the array where interdisciplinarity really makes sense.

PATHWAYS TO INTERDISCIPLINARITY Beyond such unavoidable practical questions and their solutions, whose importance must not be underestimated, we should also think about more theoretically grounded pathways to real interdisciplinarity. In this section, we wish to propose three such routes for practical exploration: direct theoretical linkages between concepts of different disciplines, identification of

10


formal characteristics of life courses that may acquire similar meanings for different disciplines and specific substantial themes that may link different disciplinary outlooks due to their transversal relevance. Our presentation will necessarily remain abstract, with only limited illustrations, as our aim cannot go beyond sketching some promising directions; their exploration is yet to come, but hopefully, this outline can provoke some fruitful discussions that help us progress in this sense. We feel moreover that systematic scrutiny of the extant literature would produce a wealth of examples that have already arisen out of research practice. Our proposal is probably little more than a possible systematization of thoughts already present in our field.

THEORETICAL LINKAGES A first possible pathway to the construction of interdisciplinarity in life course research consists of creating theoretical linkages between concepts of different disciplines. If this is not an explicit part of our everyday practice when doing research, it is often implicitly present. If we take the example of the relationship between sociology and social psychology, we may say that many – perhaps all – sociological hypotheses located on the microinteractional level and concerning behavior of individuals and groups are based on implicit assumptions of social–psychological mechanisms relating the (sociological) concepts used. The most trivial such mechanism is some kind of social rationality, be it along the lines of rational choice theory or according to more complex paradigms. Less trivial examples are social comparison, social learning, the pursuit of social recognition, reciprocity and advantage, or conformity to social norms. Another, quite different mechanism of that kind would be the one operating the transformation of internal into external attribution in the course of social interactions leading the implied actors to realize that each of them is not alone to experience a given problem but that there is a social category of people with whom they share the problem in question (classical example: the formation of the labor movement, and social movements more generally), and the personally ‘‘alleviating’’ consequence of such processes of collectivization of subjectively felt problems. Many hypotheses about social behavior, e.g., concerning biographical decisions with a view to social mobility, currently assume the functioning of such mechanisms without normally making them explicit, even less testing them. Similar relationships may exist or be developed between other disciplines.


11

A slightly different example is the formulation of direct interdisciplinary hypotheses, i.e., hypotheses linking dimensions ‘‘belonging’’ to different disciplines. Stating more explicitly Baltes’ (1987, 1997) thesis of contextuality in psychological development directly leads to such hypotheses, for instance in postulating how structural locations of individuals (in the larger social structure, in their family, in one of their peer groups) explain elements of their identity formation, ways of psychological or interactional styles, etc.

FORMAL ASPECTS OF LIFE COURSES: TRAJECTORIES, STAGES, TRANSITIONS AND EVENTS We know from research that life course transitions (such as entering or leaving school, entering professional life or parenthood, marriage, divorce, retirement, moving into nursing homes in old age) have a great significance for individual identities and for the connection between people and institutions. The stages and transitions of life courses are significant in terms of cognitive abilities, representations of the self, relations with significant others and with the institutional and the societal order. Four current concepts are directly related to any process perspective on the life course: trajectory, stage, transition and event; they are central for the understanding of human action, relatively common to various disciplines, and may therefore constitute another inroad to facilitating the construction of interdisciplinary cooperation.

Trajectory The meaning of trajectory can be defined as ‘‘model of stability and longterm changes’’ (George, 1993) or ‘‘sequence of profiles of insertion’’ participation (Levy, 2001). In this sense, the notion of trajectory is mainly used to describe the movements or developments occurring during the whole span of life, i.e., all that takes place between the two ultimate life boundaries – or, in our context, transitions – that are birth and death. If this explication remains especially close to sociological use, one finds a similar one in social psychology where trajectories are conceived of as all the movements of an individual in the social space (see among others Viaud, 1999, p. 80) and in social demography with the notion of biography (Courgeau & Lelie`vre,

12


1989, p. 56).11 On a more formal level, we may view trajectories as composed by a sequence of transitions (change) and stages (stability). Cognitive-developmental psychology proposes an in-depth analysis of the models of change and stability in the long run (life span). The idea of the integration of a lifelong trajectory results from the reformulation of the concept of development. The principle of a definite succession of stages in a classical structuralist perspective is completed by that of a sequence of compromises and settlements (growth, maintenance/resilience, regulation) arising from heterogeneous resources (biological, social, cultural). This, considered under the angle of individual development (or ontogeny) and cognitive capacities in particular, contributes to shaping an individual trajectory. Baltes and Schaie (1976) discuss cohort effects on cognitive performance. Such differences were often confounded with age effects. These results justify the distinction of typical trajectories that correlate strongly with institutional influences and societal transformations while these may in turn affect cognitive performance. Baltes (1987) discussed the relative regularity of processes of change during the first years of life. Baltes interpreted these results by stating that the social fabric is more solid, substantial and homogeneous during childhood than during adulthood, the latter showing less regularity due to a greater variability of situational factors. Moreover, one could add that the more relevant influences of biological nature during childhood (maturation, etc.) imply more regularity in that section of the developmental trajectory. The shift to a social-psychological and sociological perspective on this question happens in the first place by considering the principle of ‘‘linked lives’’ formulated by Elder (1974): trajectories of the individual members of the same primary group (family, friends, workmates) are interdependent. For instance and in principle, it is necessary to take into account the trajectory of the father and the mother of any individual to understand the logic underlying that individual’s trajectory. In sociological research, this principle is widely applied, notably in research on strategies of social mobility, socialization, domestic functioning, etc. The same principle could bring important results in interdisciplinary research. In order to build a more complete model, it is advisable to add institutional, historical and geographical dimensions (Kohli, 1985; Elder, 1985; Mayer & Schoepflin, 1989; Heinz, 1992) to these individual and relational factors. Ideally, the ambition of such a global model would be to capture and interpret the interdependences between these various developmental dynamics. In practice, however, one often prefers to distinguish specific


13

trajectories (marital, professional, health-related, residential) within the global trajectory.12 This raises two important questions: on the one hand we need to know how to take account of the mutual influences between these particular trajectories instead of studying them separately. On the other, is it after all possible and/or desirable to merge the separate trajectories into a unique global trajectory? In the perspective of modeling multidimensional trajectories, Abbott (1992), a historian and sociologist working on both the empirical and theoretical level of trajectories, proposes a ‘‘narrative’’ approach where the notion of narration is substituted to that of causality. According to Abbott, one of the major problems of research concerning trajectories lies in the fact that one is mostly confronted with a variety of sequences of events occurring at different speeds. These multiple temporal horizons constitute a theoretical barrier that prevents researchers from raising formalized stories beyond the ‘‘simplistic’’ analyses of stage process (stories which develop relatively independently) and sequences of rational actions. The possibility of matching sequences of events occurring at various speeds might prove very useful for life course research. The study of life courses might require the integration of seemingly quite heterogeneous dimensions in a unique interpretational model. Ontogenesis does not simply follow a linear succession of ordered stages. Thus trajectories result from the interaction of various factors of different kinds, such as biological, psychological, relational and institutional ones, all probably occurring at different paces and with different impacts on the life course. This fact raises important theoretical and methodological questions and surely requires an interdisciplinary collaboration.

Stage In sociology, demography, social psychology, psychology as well as in other disciplines, the notion of stage refers to a life period of variable length, characterized by a relative stability and often something like ‘‘balance’’. At the very least and on the most abstract level, a stage can be defined as a ‘‘stable state between two transitions.’’ This definition is rather descriptive and may constitute an empirical cut-out of reality based on the criteria defined beforehand. Various terms (episode, state, phase, etc.) are based on a similar logic or similar conceptions across the disciplines. At times, the duration of a stage offers a better insight. Jean Piaget has greatly contributed to the definition of the concept of stage, especially in psychology. He defines stages as relatively long periods

14


resting on the same underlying structures and equilibrium states.13 In cognitive developmental psychology, neofunctionalists (or neo-Piagetians) and contextualists have reconsidered this concept by means of notions such as ‘‘equifinality’’ (Kruglanski, 1996) and ‘‘testing the limits’’ (Baltes, Lindenberger, & Staudinger, in press). Neostructuralists have maintained certain Piagetian principles and added the following modifications: (1) the redefinition of cognitive structures (other structures than Piaget’s ‘‘logicomathematical’’ ones), not necessarily seen as universal; moreover, cultural and linguistic factors are also considered; (2) a stage is now defined by an upper cognitive limit, rather than by certain behaviors in different situations; the upper limit moves during ontogenesis, hence, maturation and aging are of chief importance; (3) individual differences and characteristic patterns of individual development are considered; and (4) it is assumed that affective and social development are similar to cognitive development. Moreover, the notion of stage has been extended and applied to adult life (Labouvie-Vief, 1980, 1982; Edelstein & Noam, 1982; Pascual-Leone, 1983; Riegel, 1976; but see already Erikson, 1950). In social psychology the concept of stage is used to refer to bases of retrospective biographical memories (McAdams, 1993, 1999). However, in many applications of research, the Piagetian conception of stage has been maintained. Moral development, as conceived by Kohlberg and his followers (Turiel, 1983; Gilligan & Attanucci, 1988), has kept the universal deterministic flavor. However, this conception has been criticized for theoretical and methodological reasons (Emler, 1999; Tostain, 1999). In social demography, the concepts of episode or phase are usually used to refer to well-defined life periods (e.g., employment phases and phases of professional inactivity for women). Sometimes, the term of age is used to designate a broader phase of the life course (e.g., adult age).14 However, the application of the term of stage in a sense close to the Piagetian tradition seems to be recent (De Bruijn, 1999); its universalistic and deterministic features are mostly avoided. Instead, emphasis is put on human agency. Furthermore, the deterministic feature of a stage is opposed to the stochastic character attached to events. In sociology, the concept of stage refers to a stable state of various durations. The social constraints and normative expectations typical of each stage are generally emphasized. Historically, in a macro-sociological perspective, models directly relying on the conception of stage (Comte, Marx, Spencer, Parsons) have been strongly criticized because of their evolutionist, linear and deterministic features. In a micro-sociological perspective, the notion of stage has been applied to the description of family development


15

and, of course, to individual life courses where it is sometimes used to designate age-groups (e.g., adolescents, middle-aged, etc.), sometimes also categories defined by a specific institutional participation (like preschoolers, employed, retired, etc.). Close terms such as period are often used to avoid the restrictions of the more assumption-loaded term of stage.

Transition Across disciplines, the concept of transition refers to the idea of change: change from one state or situation to another, from one life period to another, from one status or role to another. In this vein and by inverting the above definition of stage, we might define a transition as the (short) period of change relating consecutive stages. More precisely and less circularly, transitions are moments within a particular trajectory characterized by accelerated changes, compared to the relative stability of stages. Examples of transitions are quite different according to the type of change considered: change from a limited to a broadened capacity of information processing (possibly leading from one type of reasoning to another), from single life to marriage, from adolescence to adulthood, from the status of an employee to that of a manager. Three characteristics attributed to the concept of transition seem to be rather consensual across the different approaches: (1) a transition always refers to its outcome, a novel status of relative stability (in psychology often characterized by a higher equilibrium or stability, but not in social demography or sociology); the different approaches, depending on their objects of study, their objectives and their theoretical assumptions, do or do not have precise criteria allowing an evaluation of a transition in terms of the developmental progress it conveys; (2) transitions are processes more or less clearly limited in time (although they may have long-term consequences); and (3) most often, the concept of transition is applied to the changes in an individual life course, but it appears also in notions of social transition, collective transition or even demographical transition. Defining a phenomenon as transitional cannot be done in an absolute way since by definition, a transition relates a state ‘‘before’’ to a state ‘‘after’’. The use of the term depends on the extension and on the specific processes implied by the transition and may sometimes be a specific way of looking at an object rather than this object’s inherent characteristics. For instance, adolescence can either be qualified as a transitional stage, in particular in a perspective of development from childhood to late adulthood (Durkin,

16


1995), or as a life stage in itself, with its own developmental tasks, at the end of which the transition to adulthood takes place (Erikson, 1950). Three related concepts, each referring to a particular kind of transition, are: (1) ‘‘revolution’’ (Mounoud, 1982), which refers to a discrete transition, to the emergence of a new ‘‘structure d’ensemble’’ and thus does not apply to the acquisition of a new isolated capacity; (2) ‘‘turning point’’ (Gotlib & Wheaton, 1997), which refers to a transition (or perhaps more frequently to an event) that implies a change in the orientation of a trajectory and not just a mere confirmation of this trajectory by a transition that fits into a general pattern; and (3) ‘‘social mobility’’, upward, downward or horizontal, which is only related to transitions that affect the social status of a person, mostly implying an increase or a reduction of his economic, social or cultural capital (in the case of vertical mobility). A number of theorists have attempted to apply to the study of individual transitions the mathematical ‘‘catastrophe theory’’ (Thom, 1975) and the concept of ‘‘bifurcation’’ (e.g., van der Maas & Molenaar, 1992). Such a modeling approach has the advantage of combining quantitative with qualitative changes. For instance, a series of small quantitative changes can eventually lead to an abrupt transition, resulting in what proves to be qualitatively very different from the preceding periods. Although commonalities of the concept of transition exist across disciplines, each discipline differentiates its approaches to and uses of the notion of transitions. Social demography and sociology share a focus on the study of the effects of various parameters – socio-demographical (e.g., fertility, life expectancies, migration, social and geographical mobility) or institutional (compulsory school, labor market, retirement system, etc.) on the occurrence, timing and variability of transitions (Hogan, 1981; Rindfuss, Swicegood, & Rosenfeld, 1987; Settersten & Mayer, 1997). In social psychology, the same transitions are most often conceived of as independent variables whose effects are studied (e.g., in studies interested in the effects of particular transitions on social representations (Viaud, 1999), and selfconcept (Hagestad & Neugarten, 1985; Havighurst, 1972; Kling, Ryff, & Essex, 1997)). Developmental cognitive psychologists are more interested in explaining the mechanisms and explanatory factors of transitions characterizing the individual cognitive development (de Ribaupierre, 1989). In a Piagetian perspective, these mechanisms are described as depending on processes of maturation, cognitive conflict, equilibration and the interaction with the social environment. Here, transitions are usually conceived as dependent variables and at a more microscopic level than in sociology or social demography.


17

The idiosyncrasies of the various disciplines have not hindered interdisciplinary work. For instance, some researchers in social psychology are interested in life choice orientations (which directly influence normative transitions) through the analysis of the impact of values or of social identities on transitions considered as dependent variables (Cinnirella, 1998). Furthermore, acknowledging the role of the social environment and the individual behavior on cognitive transitions requires that the disciplines interact and provide complementary contributions to the study of the life course. In this regard, Doise and Mugny (1997) illustrate the facilitating role of social interactions and social marking in the resolution of Piagetian tasks. Indeed, this line of research provides insights into the influence of social dynamics on the transition from one cognitive stage to another, and it also shows how these dynamics can be studied with an interdisciplinary approach. Another possible link between social psychology, sociology and social demography is their common interest as to how transitions are regulated by normative representations as well as by life experiences and their interpretation on the one hand, and on the other how these are shaped by changes in their socio-historical background (Neugarten, Moore, & Lowe, 1965; Elder, 1974).

Event According to a general definition, an event is ‘‘what happens at a given time in a given place’’ (Encyclopaedia Universalis, 1999). This notion is ambiguous since some events are characterized by their singularity and unexpectedness, others by their regularity and expectedness. This ambiguity holds for several events of the life course. For instance, life events (such as a second marriage) can be considered as unique at the level of individuals. They ‘‘are particular moments in a particular time and place, complete with particular characters, actions, thoughts and feelings’’ (McAdams, 1993, p. 258). But they can also be characterized by their repetition at the level of a social group or a population. In this case, the event could be defined as a ‘‘fact which concerns an individual and which affects the structure of populations and its development’’ (Pressat, 1979, p. 68). Scientific disciplines could be distinguished depending on the emphasis they give either to the singularity or to the expectedness of an event. In this way, one can oppose, e.g., sociodemographical and socio-psychological approaches.15 Social demography emphasizes regularity (i.e., repetition). This is achieved by: (1) describing the event (e.g., marriage, birth, death); (2) considering this

18


event as part of a collective phenomenon (e.g., nuptiality, natality, mortality, mobility and migration); (3) using quantitative methods for this description; and (4) eventually predicting the event with the tools of demographic projection. The meaning of the event is defined on the basis of its statistic regularity. For instance, in the frequently cited paper titled ‘‘The changing meaning of cohabitation and marriage’’ of Manting (1996), the meaning of cohabitation or marriage in the Netherlands is not derived from Dutch values and opinions about these events, but from the development of partnerships starting as cohabiting unions and marriages of cohabiting couples. Entry into cohabitation was rare during the 70 s, and cohabiting couples rarely married. However, this kind of union became more and more frequent during the 80s and 90s, as cohabiting couples married more often. According to Manting, the role of cohabitation evolved from a substitute to marriage to a preliminary period before marriage. Social psychology insists on the singularity of events. This is achieved by assuming that a person defines an event and then by focusing the research on the possible disturbance brought about by the event on the individual identity. The event represents a mark between a ‘‘before’’ and an ‘‘after.’’ Because the event is defined by the individual (or the group), social psychology insists on its subjective character. For example, the concept of attribution captures the idea that various degrees of causal relations could be established between behaviors and ‘‘reinforcements’’ (e.g., the interpretation of an event as positive or negative; Deschamps & Beauvois, 1996). Attribution leads to the notion of allocation (i.e., how persons can explain events in their everyday life). Here, a distinction is made between ‘‘internal and external causalities.’’ Internal causality means that an individual attributes the event to him- or herself and external causality that he or she names an external, environmental cause for the event. The current hypothesis is that the type of attribution performed by the individuals has major consequences on their future behaviors. Sociology and social demography have a certain extent of common ground as far as the definition of events is concerned, especially of events perceived by the individuals as intrinsically motivated (e.g., marriage). Furthermore, sociology also insists on the expectedness of the event (normative life events vs. stressful events). In the case of stressful events, the interest lies especially on their ‘‘consequences’’ for the life courses. In cognitive psychology, the event is mainly apprehended as a stimulus or a novel phenomenon or a disturbance the individual has to react to, and is thus usually defined by the experimenter (i.e., simulated and manipulated as an independent variable). Typically, the research is not interested in the


19

event itself, but in the individual reaction to the event understood as a stimulus. The meaning of the term ‘‘disturbing event’’ is sometimes only slightly different in each discipline. In social demography, the succession of events is addressed by most researchers. The questions circle around the effects of a perturbing event on the occurrence of other events (e.g., the effect of migration on marriage; Courgeau & Lelie`vre, 1989). In this perspective, disturbing events are similar to the ‘‘turning points’’ in social psychology or sociology. These events coincide with trajectory changes (‘‘bifurcation’’ as described in sociology by de Coninck and Godard (1989) is a closely related concept). An example of a turning point (or event) is a disabling illness or accident and its eventual consequences in the form of a deep and global reorientation of life. Events occurring in a ‘‘coherent’’ way within the overall orientation of a trajectory can be opposed to turning points. In sociology and social psychology, disturbing events are unexpected and often considered as ‘‘stressful’’, at least potentially. These events can be directly related to persons (e.g., illness) or to external events (e.g., death of a significant relative, parents’ divorce, collective dismissal). In this last case, the interest lies in the amount of stress caused by the event depending on its social context and meaning. In the present context, a special mention should be made of the research field of life event research (Dohrenwend & Dohrenwend, 1982; Brown & Harris, 1989), an area of study that rests on the basic hypothesis that unforeseen events in the life course can alter its further progress and have more or less important consequences on the persons who experience them. It is multidisciplinary since at least social psychiatry, social psychology and sociology are being mobilized by life-event researchers, and it represents one extreme, ‘‘empiricist’’ option of interdisciplinarity in that it is often practiced with particularly little theoretical preconceptions, at least concerning the nature of significant events and of the mechanisms that relate them to their possible effects on the life course. Even though it seems to be rarely integrated by more outspoken life course or lifespan-oriented research, some results of this line of research confirm basic tenets of life course research precisely because of its atheoretical conception, such as the finding that the positive or negative value of life events cannot be found in themselves, but depend strongly on their interpretation by the concerned actors and their environment, that this evaluation depends on the coping resources concerned actors can mobilize, and that the abruptness of their occurrence may be as consequential as their qualitative noxiousness (which reminds one of Durkheim’s notion of ‘‘happy crises’’ that may create, according to his hypothesis, as much anomy as unhappy ones).

20


A disturbing or stressful event may act as a turning point. In a rigorous conceptual perspective, we may even ask whether the notion of events should be located on the same level as the other three terms, trajectory, stage and transition. We might also consider events to be potential transitions (of positive or negative social value) depending on the possibilities of life course actors to cope with them. Finally, what an event is may vary as a function of the fine-grainedness we choose for our interrogation: divorce can be considered as an event, unique and momentary, e.g., in a demographic analysis. But it can as well be studied as a rather long-term process in which the formal act of legalized separation is only one relatively late step, and maybe not even the most consequential one.

TRANSVERSAL SUBSTANTIVE THEMES: CONTENTS OF THIS VOLUME A third line of interdisciplinary collaboration can be found in transversal themes or research dimensions that appear in several disciplines, even though they are typically approached differently by them. In a way, the formal aspects treated in the previous section – especially transitions and events – have also very substantive meanings; they can therefore also be subsumed under this heading and may be considered as first examples. Examples of other such dimensions are, in an arbitrary and imperfect order, gender differentiation, subjective vs. objective aspects of the life course and their relationships (‘‘subjective’’ including questions of experience, representations, identity, projects, cultural models, etc.), agency vs. structure in and about life courses, linked lives, interinstitutional links forming life course regimes. We may also think of thematic fields such as gerontology, which, by definition, is inter- or at least multidisciplinary. Various academic departments (especially human development and family studies) as well as centers of gerontology typically count among their member scholars from different disciplines, such as demography, sociology, psychology, biology, kinesiology, nurse practice, health policy, communication science, social work, educational sciences, ecology, counseling, anthropology, philosophy, theology, etc.16 It is, however, certainly not sufficient to have different disciplines attached to the same organizational unit for interdisciplinarity to develop, even if this can become a favorable condition. Direct collaboration is a much more promising situation, especially in projects explicitly designed to be interdisciplinary from the outset.


21

We have chosen to organize this volume around four such themes: agency and structure, transitions, biographical reconstruction and methodological innovations. The first three themes are substantive and take up some of the transversal lines we have sketched earlier in this essay. The disciplinary origins of their authors reflect that they are not yet equally treated in the disciplines we solicited for this volume.

Agency and Structure Agency and structure has become an important issue in sociological debates since the 1980s, perhaps less so in other disciplines. The first chapter takes up this theme, locating it clearly in the substantive area of life course research. Settersten and Gannon focus on the joint impact of social structure and human agency, in proposing a life course model of agency within structure. They consider how individuals actively create their own lives and maximize their own development within parameters set by their social worlds (some of which may constrain them, and some of which may enable them), and how individuals interact with, and even make proactive attempts to alter, those worlds. Several examples from three different life periods are provided: childhood and adolescence, early adulthood through midlife, and old age. The tension between the theses of standardization or destandardization of life courses is reconceptualized as not to be resolved by one thesis winning out over the other, but as being acknowledged to be complementary processes, loosely associated with agentic and structural dimensions of the life course. Marshall tackles the topic of structure and agency first by reviewing the literature for explicit and implicit definitions of these two crucial terms and by revisiting, in an autobiographic return, some earlier results of his comparative study of aging communities. He comes up with a theoretical model containing both aspects and recovers Clausen’s concept of planful competence as one straightforward way to conceptualize life course agency. Like other contributors, he reminds us of important complexity that must not go unnoticed in life course research, especially a necessary differentiation of the concept of identity, and expresses hope for a rejuvenating theoretical input by European life course researchers. Lu¨scher looks into a specific and often neglected aspect of social relations, especially intergenerational relations, tending to rehabilitate ambivalence as a basic psychological ingredient for sociological analysis. This is a timely

22


hint at the fact that the links between the lives life course analysis is bound to take into account are not just arrows on network charts, but real social relations with their affective depth and complexity. In fact, the concept of linked lives is extremely important, but still awaits theoretical and empirical development, not only concerning the type of people that can lead linked lives (like successive generations or partners in a couple), but also referring to the nature of their relationship (besides the emotional valence, their strength is also a variable that merits more consideration: are linked lives based only on strong ties or may weak ties also play this role?).

Transitions While life course research makes a strong argument for considering the whole life span instead of limiting its interest to specific parts of it (especially single phases), specific aspects of life courses may be more fertile for research than others. Transitions are particularly revealing of life course dynamics on several accounts and are at the center of four contributions. Although, for the time being, they seem to be principally investigated by sociologists, they look promising for psychology and social psychology as well. Mortimer and her co-authors attempt to disentangle the respective influences of social structures and individual agency, in focusing on the issue of educational attainment in the transition to adulthood. Some scholars highlight cultural values, normative timetables, stratification processes and institutional career lines as determinants of the contents and pacing of role changes through the life course. Others, in contrast, emphasize the exercise of human agency as a central causal force in shaping the life course, including the expression of values and identities, self-regulative processes, decision-making and striving to achieve personal objectives through goal selection, strategic planning and action. Mortimer and colleagues focus on the interactions existing between these two dimensions. Using data from the Youth Development Study, a 15-year panel study of work experience and the transition to adulthood, they show that early goals and values are linked to work behavior during high school, which in turn has predictive power with respect to subsequent trajectories of work, schooling and educational attainment. The study presents promising ways for opening the ‘‘black box’’ of the processes and mechanisms, both structural and agentic, underlying life paths of transition to adulthood.


23

With an autobiographical twist, Furstenberg takes up the thesis of many life course researchers that chronological and sequential ‘‘order’’ is important, that deviation from this order (like being out of schedule) constitutes a problem and is bound to provoke undesirable outcomes. Reviewing recent analyses, he shows that results seeming to confirm this somewhat structurodeterministic idea are in fact less than convincing and remain open to contrary, more optimistic and ‘‘agentic’’ interpretations; according to his argument, ‘‘life course deviance’’ may as well lead to growth in life course managing capacities. A major role in determining whether non-normative trajectories, be they out-of-time or out-of-order, become problematic for later stages in life seems to be reserved to social and psychological resources and thus index the structural environment of ‘‘life course passengers’’. Looking into The Secret of Transitions, Bird and Kru¨ger develop a strong argument about the danger of substantive blinders brought along by methodological and conceptual reductions, for example, in the case of the study of life course transitions by help of event-history analysis. They advise us not to forget the complexity of life courses in the double sense that they are made up of several parallel and related trajectories and that transitions belonging to different trajectories may interact without necessarily being synchronized. Moreover, even in one trajectory, transitions are often multilayered and last longer than their treatment as ‘‘events’’ would suggest. On the basis of empirical examples concerning sex-differentiated life courses, they propose a timely typology of transitions based on their socio-dynamic features that can help avoid excessive reduction of social complexity on the conceptual and interpretive level.

Biographical Reconstruction Life course and biographical analysis (in the sociological, not the demographic use of the term as we pointed out before) have long been considered to be strangers ignoring each other – if not hostile antagonists – especially in some European research traditions. Heirs of the fundamental debate between the ‘‘two cultures’’ in the social sciences, life course analysis is often purported to be positivist, explanatory and quantitative; biographical analysis constructivist, interpretive and qualitative. While this division still exists, it has been questioned by more recent pragmatic positions, preferring mixed methodologies (Tashakkori & Teddlie, 1998) and quali–quanti triangulation (Erzberger & Prein, 1997) to methodological fundamentalism. Even though this debate has probably raged more in sociology than in most

24


other social science disciplines, it is highly relevant for all of them, especially in the life course area. The contributions concerning biographical reconstruction are instructive in this respect: the first relates identity changes directly to life course transitions, the second approaches directly and empirically the crucial theme of biographical memorization and the third is an example of interpretive work in a discipline that has heretofore privileged quantitative approaches. Biographical accounts have become a field more currently shared at least by social psychologists and sociologists. In a chapter relating this part of the volume to the previous one, Emler presents in a social psychological perspective how identity shifts take place during life transitions. A first section is devoted to the presentation of the classical view of cognitive development (Piaget, Kohlberg) as abrupt qualitative changes resulting in a succession of qualitatively different stages. Then social identity is defined on the basis of self-categorization theory, taking into account the relationships between social identity and social networks. One central issue of this chapter is to understand if identity development is a gradual process or if it corresponds to a stage model with sudden and discrete changes. Based on different examples in the areas of political identity and changes in personal relationships, it opens new directions of thought and research in the study of shifts of identity across life transitions. Perrig-Chiello and Perrig bring together different important psychology fields in the study of personality across the life span: autobiographical memory, well-being and personality traits. Psychologists have developed different theories on how the individuals construct their autobiographical memory. This chapter presents empirical evidence showing that personality traits like neuroticism and extraversion have a consistent relationship with recollection of autobiographical episodes and well-being. Results also indicate that the impact of personality traits may be more important than positive or negative life events. These results raise important questions about how individuals adapt and reconstruct life events across their life course, and how subjective experiences, and in particular personality traits, may have objective effects on life trajectories. McAdams is a personality psychologist who has developed a narrative approach to identity across the life span – a highly welcome complement to the personality trait approach. Building on the idea that identity is constructed in a continuous process across the life course, he shows that life stories are developed by individuals following distinct ‘‘story lines’’. For example, he shows that life stories may be narratively organized as contamination (good to bad) or redemption (bad to good) sequences.


25

The various antecedents and consequences of these types of life stories are described on the basis of empirical examples. A promising development of this approach is the possible link between these life stories and the cultural context in which they are expressed and constructed.

Methodological Innovation Methods of analysis have a peculiar role in our thematic field, especially so in quantitative methods.17 They are not very numerous and well-known and they are based on technical and also conceptual implications that have to be scrutinized all the more seriously for their implicit assumptions and basic logics for the interpretation of results as they are practically all imported from more or less distant fields or disciplines (this is probably the most multidisciplinary sector in life course research!). Moreover, the danger of one method dominating the field and the substantive outlook of researchers merits particular attention (one obvious candidate for becoming a ‘‘hegemonic method’’ in quantitative life course research is event history analysis, with optimal matching a new-coming junior competitor that remains at the margins of the arena for the time being). Therefore, we find it useful to contribute to methodological diversification with a set of promising and innovative contributions presenting methods for exploration and analysis that have not yet made decisive inroads into this area. The three chapters of this volume nicely complement each other, largely due to the different, but sometimes converging and highly complementary, research fields of the authors. Francesco Billari’s applications are mainly in demography, especially in the research fields of fertility and transition to adulthood. Michel Oris, a demographic historian, and Gilbert Ritschard, an econometrician, have been collaborating for several years on projects requiring both an acute historical sense for the analyses of demographical issues and advanced analytical tools, which statistically address the theoretical questions of interest. The work of Jack McArdle, a psychologist, has been focused on agesensitive methods for psychological and educational measurement, and longitudinal data analysis. Together, the four authors provide an exceptional overview of novel life course methodologies of great potential and invite us to think about future developments. Each chapter briefly summarizes the state of the art of its research field before heading into the most recent advances and promising extensions. In his contribution, Billari compares the two major approaches of quantitative methods used in life course research, the event-oriented and the

26


holistic approach. The former quantitative approach is largely represented by the freshly popular analytical techniques of event history (or survival) analysis and also newer techniques stemming from program evaluation work. The latter approach is mostly represented by techniques used to analyze any sequence of events and trajectories over the whole life course, and the prime among these is optimal matching analysis. Both approaches are used in turns to answer different but complementary life course research questions. The event-oriented approach is often used in a causal perspective of analysis, and the author describes all different kinds of causal factors susceptible to affect the occurrence of an event. The holistic approach is adopted to analyze different factors influencing trajectories. Billari then invites the reader to consider that the two perspectives are couched in the two statistical cultures of data vs. holistic modeling discussed in the very influential paper of Breiman (2001). The data modeling culture naturally pairs up with the event-oriented approach, the algorithmic modeling culture with the holistic approach. Billari provides several examples to clarify his arguments and to illustrate how the two traditions and approaches complement each other in addressing life course phenomena. The chapter by Ritschard and Oris discusses how the fields of historical demography and contemporary demography promisingly joined forces, especially under the life course research paradigm. They discuss three analytical approaches and illustrate them by relevant examples. First, event history analyses are again taken up, but this time in relation to very recent advances allowing to account for shared heterogeneity or frailty in the presence of multiple groups. The combination of event history analysis with multilevel modeling will undoubtedly witness much success in several research fields. The second set of statistical techniques presented by the authors is that of Markov transition models used to analyze state sequences, or individual trajectories across quantitatively and possibly qualitatively different phases in the life course. The categorical changes can be dependent of observed or latent covariates and differ with respect to their temporal expression, or lags. Again, this technique is progressing at a quick rate and the range of its applications is growing accordingly. The third set of analyses consists in longitudinal data mining techniques based on an induction tree approach. While more exploratory in nature than the previous two techniques, induction trees ideally describe the change trajectories of complex multivariate systems, in which the elements are ordered by importance and predictability power. As illustrated by the authors, this newer set of techniques may complement the other two and add much insight into life course research objects, over and above what was already learnt with more classical techniques.


27

McArdle accompanies the reader in exploring a unique data set of cognitive measures obtained on seven occasions from 1931 to 1998 on the same individuals. The data are first described and observed with respect to their presence as well as absence (i.e., incomplete data). Then individual and group change characteristics are determined and possible predictors of change are tested. Latent growth curve models, a particular application of structural equation modeling, are used to this end, which are known in the educational and biostatistical literature under different names (hierarchical linear modeling, multilevel modeling, random or mixed effects modeling, etc.). Group differences in developmental trajectories are then explored, on an a priori level (via traditional multigroup structural equation modeling) as well as on an a posteriori level (with more recent latent mixture models). Finally, McArdle introduces an advanced class of structural equation models based on previous work in econometrics, biostatistics and behavior genetics to explore dynamics of change. In other words, we are now invited to examine the data in a non-static perspective, going from description to explanation of change phenomena. The advanced latent difference score models introduced by McArdle represent a very promising set of analytical tools capable of addressing classical as well as innovative theories of change in several research fields. Finally, in their afterthought chapter, the editors bundle together the novel impulses for fostering interdisciplinary life course research that can be gleaned from the contributions to this volume and suggest directions for future development.

NOTES 1. The Pavie Team, as it existed when the interdisciplinary venture that produced this volume was launched, comprised a founding group of six professors and eight researchers and research assistants. They all participated in the elaboration of a working document that was the basis of an international colloquium held in October 2003 and that has been largely integrated into this introduction. In order to underscore the team character of this work, we list their names in strict alphabetical order: Jean-Claude Deschamps, Guy Elcheroth, Yannic Forney, Jacques-Antoine Gauthier, Paolo Ghisletta, Jean Kellerhals, Christian Lalive d’Epinay, Jean-Marie Le Goff, Rene´ Levy, Anik de Ribaupierre, Claudine Sauvain-Dugerdil, Dario Spini, Manuel Tettamanti, and Eric Widmer. 2. Coming from two – at times – different scientific traditions, the terms life course and life span share so much meaning at many levels (conceptual and methodological)

28


that we can safely focus on the similarities between the two terms and consider them as basically synonymous. Instead of using them interchangeably, we rather settle on the use of life course for the sake of simplicity (for a discussion of these two terms, along with the less adequate one of life cycle, see Settersten (2003a) who takes the same terminological stand as we do). 3. An entirely parallel argument could be made for psychology and probably other disciplines of the social sciences. 4. One can lead lengthy discussions about the proper use of terms like macroscopic or microscopic – what is micro for a sociologist may be very macro for a psychologist, what is micro for a psychologist may be way underneath the systemic level for which other social sciences have analytical resources. Instead of doing this, let us just illustrate the notion of encompassing system levels by citing three examples in ascending order: different psychological systems (like memory, emotion, etc.) – interindividual interactions – relations between social institutions (e.g., in organizing subsequent or simultaneous stages of the life course). 5. This is an uneasy translation of the standard French expression ‘‘rapports sociaux de sexe’’. 6. It goes without saying that a similar argument could be made for all classical sociological variables supposed to capture some aspect of social differentiation, first of all ‘‘class’’ or position in social hiearchies, but several of these are socially less ‘‘thick’’. We do, of course, not mean to relegate them to a secondary position for analysis, but rather feel that in current research, there is a danger of too exclusive a concentration on them. 7. The most general paradigm of this sort in sociology is rational choice theory or ‘‘methodological individualism’’. But there are many more specialized areas where predominant approaches at least implicitely adopt an individualist stance, for instance in the form of the status attainment paradigm in mobility research. In psychology, the same problem exists but in a more fundamental sense, considerations of ‘‘contextuality’’ being more intrinsically out of theoretical bounds than, say, in sociology, social psychology or political science (but see Bronfenbrenner’s ecological model (Bronfenbrenner, 1979), Baltes’ contextualism (Baltes, Lindenberger, & Staudinger, in press), or even Erikson’s epigenetic model (1950). 8. There are some special research fields where interdisciplinarity is already practiced in a more or less regular way, such as gerontology. But even here, collaboration is often more of a coordinated parallelism than real cooperation. Moreover, in this case, research is not systematically placed in a full life course framework. 9. Some authors suggest that the vocabulary of truly transdisciplinary analysis should be based outside the boundaries of scientific terminology. This may be useful in a highly problem-oriented research, but should not be seen as a general rule. Systems theory provides a scientific vocabulary for transdisciplinary conceptualization which is, to the least, not less promising than everyday or other non-scientific language. 10. The current French expression is ‘‘analyse de´mographique des biographies’’ and has nothing to do with the qualitative study of biographical reconstruction that sociologists might have in mind. 11. We should note that in some contexts, as here, ‘‘biography’’ is practically synonymous of ‘‘life course’’ whereas in others, the distinction of these two concepts


29

is almost as important as that of sex and gender in gender studies, with ‘‘biography’’ meaning the narrated, principally self-accounted history of an individual, and ‘‘life course’’ the factual unfolding of life as it can be registered by an outside observer (e.g., the Research Committee Nr. 38 ‘‘Biography and Society’’ of the International Sociological Association is clearly interested in biographical research in the sense of this distinction; cf. Rosenthal & Ko¨ttig, 2004). 12. One can distinguish the interdependences between processes belonging to different spheres in the life of an individual, or between processes which concern an individual and his significant others, or a process which concerns an individual and a transforming context. 13. In most disciplines, the concept of stage, even if conceived as an element of a continuous process, is not inherently linked to notions of finality and universality although these notions have often been intimately associated with the definition of stage. This misconception has been strongly criticized, especially by anthropologists. An important example of such a debate about implicit and scientifically unfounded directionality assumptions on a supraindividual level has concerned modernization theory. 14. For some researchers interested in fertility issues, the reproductive age (career) might be divided in several phases or stages, each of them characterized by a specific knowledge, a childbearing motivation, and a style of decision making (Forrest, 1988): (1) before first sexual intercourse; (2) from first sexual intercourse to marriage; (3) from marriage to first birth; (4) first birth to desired family size and (5) from family completion to mother’s menopause. 15. We do not explicitly integrate the specific area of life-event research that has developed mainly in social medicine and psychiatry, but would like to mention it as an interesting source of empirical findings that strongly underscore the relevance of specific and especially of singular events for subsequent periods of life. At the same time, the diversity and sometimes even contradictory characters of these findings point to the necessity of explicitly taking into account conceptual tools of various disciplines, particularly with respect to the non-reductible complementarity of objective and subjective features (i.e., their personal meanings) of such events if we are to understand their consequences. 16. This is often possible because faculty members hold joint or courtesy appointments. Hence, they belong to distinct departments and only occasionally collaborate on common projects. This sort of institutional structure, although representing progress, is not the ideal environment for fostering common research grounds. The sharing of information is undoubtedly facilitated by centers of this kind. However, interdisciplinarity, as we claimed before, aspires to deeper linkages among the disciplines. This is reflected by the lack of Ph.D. programs in gerontology. To our knowledge, various postgraduate diplomas exist in gerontology, but only few Ph.D. titles; the same seems to hold for life course studies. 17. Given their greater openness, qualitative methods do not seem to present the same potential of substantively orienting research and of blinding it toward a phenomena for which the methods we use have no outright provision. Even though, a similar danger could be related to an exclusive use of qualitative methods, especially if accompanied by a conceptual reduction of the range of research questions to those qualitative methods can answer.

30


REFERENCES Abbott, A. (1992). From causes to events: Notes on narrative positivism. Sociological Methods & Research, 20, 428–455. Baltes, P. B. (1987). Theoretical propositions of life-span developmental psychology: On the dynamics between growth and decline. Developmental Psychology, 23, 611–626. Baltes, P. B. (1997). On the incomplete architecture of human ontogeny: Selection, optimization, and compensation as foundation of developmental theory. American Psychologist, 52, 366–380. Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (in press). Life-span theory in developmental psychology. In: R. M. Lerner (Ed.), Handbook of child psychology, Theoretical models of human development, (Vol. 1, pp. 1029–1143). New York: Wiley. Baltes, P. B., & Schaie, K. W. (1976). On the plasticity of intelligence in adulthood and old age: Where Horn and Donaldson fail. American Psychologist, 31(10), 720–725. Breiman, L. (2001). Statistical modeling: The two cultures (with comments and rejoinder). Statistical Science, 16, 199–231. Bronfenbrenner, U. (1979). The ecology of human development. Cambridge: Harvard University Press. Brown, G. W., & Harris, T. O. (Eds) (1989). Life events and illness. New York: Guilford. Cinnirella, M. (1998). Exploring temporal aspects of social identity: The concept of possible social identities. European Journal of Social Psychology, 28, 227–248. Courgeau, D., & Lelie`vre, E. (1989). L’approche biographique en de´mographie. Revue franc- aise de sociologie, 31(1), 55–74. De Bruijn, B. J. (1999). Foundations of demographic theory. Choice, process, context. Amsterdam: NethurD Publications. de Coninck, F., & Godard, F. (1989). Les formes temporelles de la causalite´. Revue franc- aise de sociologie, 31(1), 23–53. de Ribaupierre, A. (Ed.) (1989). Transition mechanisms in child development. Cambridge: Cambridge University Press. Deschamps, J.-C., & Beauvois, J.-L. (Eds) (1996). Des attitudes aux attributions – Sur la construction de la reálite´ sociale (Vol. II). Grenoble: Presses universitaires de Grenoble. Dohrenwend, B. S., & Dohrenwend, B. P. (Eds) (1982). Stressful life events and their contexts. New York: Neale Watson. Doise, W., & Mugny, G. (1997). Psychologie sociale & de´veloppement cognitif. Paris: Armand Colin. Durkin, K. (1995). Developmental social psychology from infancy to old age. Malden: Blackwell. Edelstein, W., & Noam, G. (1982). Regulatory structures of the self and ‘‘postformal’’ stages in adulthood. Human Development, 6, 407–422. Elder, G. H. (1974). Children of the great depression. Chicago: University of Chicago Press. Elder, G. H. (Ed.) (1985). Life course dynamics: Trajectories and transitions, 1968–1980. Ithaca: Cornell University Press. Elder, G. H. (1995). The life course paradigm: Social change and individual development. In: P. Moen, G. H. Elder & K. Lu¨scher (Eds), Examining lives in context: Perspectives on the ecology of human development (pp. 101–139). Washington: APA Press. Emler, N. (1999). What does principled versus conventional moral reasoning convey to others about the politics and psychology of the reasoning? European Journal of Social Psychology, 29(4), 455–468.


31

Encyclopaedia Universalis Multimedia version 8. (1999). Paris: Encyclopaedia Universalis. Erikson, E. H. (1950). Childhood and society. New York: Norton. Erzberger, C., & Prein, G. (1997). Triangulation: Validity and empirically based hypothesis construction. Quality and Quantity, 31(2), 141–154. Forrest, J. D. (1988). Contraceptive needs through stages of women’s reproductive lives. Contemporary Obstetrics Gynaecology, 32, 12–22. George, L. K. (1993). Sociological perspectives on life transitions. Annual Review of Sociology, 19, 353–373. Gilligan, C., & Attanucci, J. (1988). Two moral orientations: Gender differences and similarities. Merril-Palmer Quarterly, 34(3), 223–237. Gotlib, I. H., & Wheaton, B. (Eds) (1997). Stress and adversity over the life course. Cambridge: Cambridge University Press. Hagestad, G. O., & Neugarten, B. L. (1985). Age and the life course. In: E. Shanas (Ed.), Handbook of aging and the social sciences, (Vol. 2, pp. 35–61). New York: Van Nostrand Reinhold. Havighurst, R. H. (1972). Developmental tasks and education (3rd ed.). New York: David McKay. Heinz, W. (Ed.) (1992). Institutions and gatekeeping in the life course. Weinheim: Deutscher Studienverlag. Hogan, D. P. (1981). Transitions and social change: The early lives of American men. New York: Academic Press. Kling, K. C., Ryff, C. D., & Essex, M. J. (1997). Adaptative changes in the self-concept during a life transition. Personality and Social Psychology Bulletin, 23, 981–990. Kohli, M. (1985). Die Institutionalisierung des Lebenslaufs. Ko¨lner Zeitschrift fu¨r Soziologie und Sozialpsychologie, 37, 1–29. Kruglanski, A. W. (1996). Goals as knowledge structures. In: J. A. Bargh (Ed.), The psychology of action: Linking cognition and motivation to behavior (pp. 599–618). New York: Guilford. Labouvie-Vief, G. (1980). Beyond formal operations: Uses and limits of pure logic in life-span development. Human Development, 23, 141–161. Labouvie-Vief, G. (1982). Growth and aging in life span perspective. Human Development, 25(1), 65–79. Levy, R. (2001). Regard sociologique sur les parcours de vie. In: P. Domininice´ (Ed.), Regards pluriels sur l’approche biographique : Entre discipline et indiscipline (pp. 1–20). FAPSE, UNIGE: Cahiers de la section des sciences de l’e´ducation, No 95. Manting, D. (1996). The changing meaning of cohabitation and marriage. European Sociological Review, 12, 53–65. Mayer, K. U. (2002). The sociology of the life course and life-span psychology: Diverging or converging pathways? In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Lifespan psychology in exchange with other disciplines. Dordrecht: Kluwer. Mayer, K. U., & Schoepflin, U. (1989). The state and the life course. Annual Review of Sociology, 15, 187–209. McAdams, D. P. (1993). The stories we live by: Personal myths and the making of the self. New York: Guilford. McAdams, D. P. (1999). Personal narratives and the life story. In: O. John (Ed.), Handbook of personality: Theory and research. (2nd ed., pp. 478–500), New York: Guilford.

32


Mortimer, J. T., & Shanahan, M. J. (Eds) (2003). Handbook of the life course. New York: Kluwer/Plenum. Mounoud, P. (1982). Revolutionary periods in early development. In: T. G. Bever (Ed.), Regression in mental development (pp. 119–131). Hillsdale, NJ: Erlbaum. Neugarten, B. L., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints, and adult socialization. American Journal of Sociology, 70(6), 710–717. Pascual-Leone, J. (1983). Growing into human maturity: Toward a metasubjective theory of adulthood stages. In: O. G. Brim Jr. (Ed.), Life-span development and behaviour, (Vol. 5, pp. 117–156). New York: Academic Press. Pressat, R. (1979). Dictionnaire de de´mographie. Paris: PUF. Riegel, K. F. (1976). The dialectics of human development. American Psychologist, 31, 689–700. Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How common and does it matter? American Sociological Review, 49, 359–372. Rosenthal, G., & Ko¨ttig, M. (2004). Biography and society– Reading list. Internal paper of ISA RC 38 ‘‘Biography and Society’’. Settersten, R. A. (1999). Lives in time and place. The problems and promises of developmental science. Amityville: Baywood. Settersten, R. A. (2003a). Propositions and controversies in life-course scholarship. In: R. A. Settersten (Ed.), Invitation to the life course: Toward new understandings of later life (pp. 15–45). Amityville: Baywood (Chapter 1). Settersten, R. A. (2003b). Rethinking social policy: Lessons of a life-course perspective. In: R. A. Settersten (Ed.), Invitation to the life course: Toward new understandings of later life (pp. 191–222). Amityville: Baywood (Chapter 7). Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring, and the life course. Annual Review of Sociology, 23, 233–261. Staudinger, U. M., & Lindenberger, U. (Eds) (2003). Understanding human development. Dialogues with lifespan psychology. Boston: Kluwer. Tashakkori, A., & Teddlie, C. (1998). Mixed methodology. Combining qualitative and quantitative approaches. Thousand Oaks: Sage. Thom, R. (1975). Structural stability and morphogenesis: An outline of a general theory of models. Reading, MA: Benjamin-Cummings Publishing. Tostain, M. (1999). Psychologie, morale et culture. L’e´volution de la morale de l’enfant a` l’aˆge adulte. Grenoble: Presses Universitaires de Grenoble. Turiel, E. (1983). The development of social knowledge: Morality and convention. Cambridge: Cambridge University Press. van der Maas, H. L., & Molenaar, P. C. (1992). Stagewise cognitive development: An application of catastrophe theory. Psychological Review, 99, 395–417. Viaud, J. (1999). Principes organisateurs et repre´sentations sociales de l’ećonomie: gene`se et dynamique. Revue Internationale de Psychologie Sociale, 12(2), 79–105. Wohlwill, J. F. (1970). The age variable in psychological research. Psychological Review, 77, 49–64.

PART I: AGENCY AND STRUCTURE

33


34

STRUCTURE, AGENCY, AND THE SPACE BETWEEN: ON THE CHALLENGES AND CONTRADICTIONS OF A BLENDED VIEW OF THE LIFE COURSE Richard A. Settersten, Jr. and Lynn Gannon The field of life course studies has at its core two propositions for which there is an inherent tension: one emphasizing that the life course is the product of social forces (broadly construed as ‘‘social structure’’), and the other emphasizing individual capacities and effort (broadly construed as ‘‘human agency’’). A wide array of perspectives on structure and agency can be found in the literature. At one extreme are models of structure without agency. More common in the discipline of sociology and in European scholarship, these models take the life course to be largely constrained, if not determined, by the characteristics of and processes in social settings, and by the locations of individuals within those settings. Politically, these models can be viewed as problematic, at least if they are carried too far, because they ‘‘externalize’’ blame and leave little or no room for personal responsibility. At the other extreme are models of agency without structure. More common in the discipline of psychology and in North American scholarship, Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 35–55 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10001-X

35

36

RICHARD A. SETTERSTEN, JR. AND LYNN GANNON

these models take the life course to be largely fostered, if not determined by, individual decisions and actions. Life, as the adage goes, is largely what one makes of it. While few of these latter models take the life course to be completely devoid of social obstacles and barriers, these models often downplay the effects of social forces and assume that good planning and hard work go a long way in overcoming barriers. In the political sphere, these models can be dangerous, at least if they are carried to an extreme, because they blame people for negative outcomes and suggest that individuals need not be extended support from the state because their problems and circumstances are their own doing. In both cases, these models have led to the question ‘‘What matters more, structure or agency?’’ and to discussions of agency versus structure. While these two prevalent perspectives offer different and important lenses for understanding human development, a third and underdeveloped set of models instead seem necessary to advance interdisciplinary life course research. These are blended models of agency within structure, which explicitly seek to understand how individuals set goals, take action, and create meanings within – and often despite – the parameters of social settings, and even how individuals may change those parameters through their own actions. These models bridge over-structured and under-structured views of the life course by asking how the characteristics of and processes in social settings interact with the characteristics, capacities, and actions of individuals to jointly affect life trajectories and outcomes. They involve incorporating interactions with a wide range of social settings both near to and far from individuals – from proximal settings of everyday life such as families, peer groups and friendships, neighborhoods, schools, or work organizations, to more distal settings such as the labor market, the state and its policies, and historical events and periods of social change. Modes of agency within structure bring significant challenges because they demand that boundaries between disciplines be crossed, especially between life-span psychology and life course sociology; that the concepts and measures of ‘‘structure’’ and ‘‘agency’’ be clarified; and that more sophisticated theories and methods be developed to frame and analyze them. Models of agency within structure also demand a critical evaluation of the unique nature and effects of structure–agency dynamics within and across distinct life periods. The tensions between structure and agency need not be resolved as much as capitalized upon to build new social theories and research on specific life periods and on the life course as a whole. Finally, there is mounting and conflicting evidence that the life course has become both more standardized (with regularity in life course patterns being

Structure, Agency, and the Space Between

37

driven by the increased ‘‘institutionalization’’ through norms, laws, and social policies) and de-standardized (or ‘‘individualized,’’ with variability in life course experiences being driven by greater freedom to ‘‘live a life of one’s own’’). The chapter ends with some thoughts on how these contradictions may be explained by more explicitly tending to structure–agency dynamics.

STRUCTURE, AGENCY, AND THE SPACE BETWEEN Structure The concept of social structure is difficult to grasp and make precise. The things indexed by the notion of social structure are recognized as the central sociological contribution to the study of lives, yet there is little agreement about what exactly structure is (see also Alwin, 1995). Indeed, sociologists not only find it difficult to define structure adequately, but they often cannot do so without using the word ‘‘structure’’ or a variant of it (see also Sewell, 1992). Conventional approaches treat structure as a powerful set of stable top-down forces that impinge upon individuals and cannot be (easily) altered. As an example, consider Alwin’s (1995, p. 218) definition of social structure as a set of ‘‘opportunities and constraints within networks of roles, relationships, and communication patterns, which are relatively patterned and persisting (emphasis added). These opportunities and constraints may, at one extreme, refer to ‘‘large, organic institutional structures, such as bureaucracies, which structure and orient human activities,’’ or they may, at the other extreme, refer to a set of ‘‘dyadic norms negotiated between two individuals for the purposes of social exchange.’’ This definition of social structure is typical in that it emphasizes stability, but it is unusual in that it acknowledges both constraints and opportunities, and both macro- and micro forces. The emphasis on stability, however, has led to the neglect of dynamic aspects of structure. Few approaches have emphasized the fact that people have the ability to change structures (that is, that the relationship between people and structures is reciprocal) and that both social structures and human lives and the connection between them, are dynamic (an important exception to this trend has been the age stratification framework long advocated by Riley and colleagues; e.g., Riley & Riley, 1999). On the surface, the notion of ‘‘dynamic’’ structures seems to negate the very concept of structure, which has stability at its core. But in a fast-paced world of rapid social change, it may be transformations in social structure, and the po-

38


tentially disruptive effects they have on human lives, that pose new challenges for growing up and growing older today. Sociologists of ‘‘social stratification’’ have emphasized the fact that the social structure of a society can be described through key dimensions of social organization: age (or cohort), race (or ethnicity), sex (or gender), and social class (education, occupation, income, or some combination thereof). Some of these are ‘‘ascribed’’ statuses into which individuals are born or over which individuals have little control. Others are ‘‘achieved’’ statuses that are largely the result of performance, effort, or things over which individuals presumably have some control. Much social thought related to stratification has often been tied to the question of who gets what in society. In most complex societies, certain individuals and groups hold disproportionate shares of social resources (e.g., property, power, prestige), and sociologists have traditionally been interested in how and why the distribution of resources varies as a function of the dimensions noted above. Similarly, both life course sociology and life-span psychology continue to be dominated by what Bronfenbrenner (1988) once called ‘‘personal attribute,’’ ‘‘social address,’’ and ‘‘social niche’’ models. Personal attribute models group and compare individuals based on biological or physical features (e.g., age, sex, body type). Social address models group and compare individuals by geographical or social group (e.g., urban or rural, social class, race or ethnic group). While social address models focus on geographical or social dimensions, they nonetheless often rely on personal attributes as a means for grouping individuals. Social niche models group and compare individuals based on intersections between multiple statuses (e.g., poor, young, unmarried mothers versus other groups). All of these models can be described as class-theoretical models because they assume that the characteristics of individuals serve as important proxies for ‘‘social structure’’ and individuals’ experiences in it, that they index some aspect of inequality, and that the phenomena under study are somehow explained by the categories themselves. In reality, however, these variables provide little or no direct information about the characteristics of or processes in the social worlds that individuals inhabit, nor about the experiences of individuals in those worlds. When investigators find significant differences between classes, they are then faced with the challenge of having to explain these differences – and it is with respect to explanation that classtheoretical models are not sufficient. Because these models are based on the characteristics of individuals, they do not tap what most sociologists and ecologically minded psychologists think about as social structure. A big and problematic leap of faith is required when we assume that individuals who


39

share a set of personal characteristics also share similar social worlds and experiences. At best, such a leap leads to suspect interpretations; at worst, it results in misleading and false ones. This is especially true when classtheoretical approaches are used to explore what Bronfenbrenner calls fieldtheoretical questions – that is, when we want to understand the ecology of human development. Because ‘‘social structure’’ is such an elusive concept, it is not surprising that its measurement and modeling are so difficult. Important strides must be made in conceptualizing social settings, measuring their characteristics and the processes that occur in them, and analyzing additive and multiplicative effects across settings. This includes attention to the proximal and distal settings noted earlier. Of course, the more distal the environment, the more difficult it is to describe, and the harder it is to trace the processes and mechanisms through which it affects specific developmental outcomes. Yet even the measurement of seemingly simple personal characteristics on which investigators have long relied can quickly become complicated. Indeed, the measurement of race and ethnicity, socioeconomic status, sex and gender, and age and cohort are all highly controversial practices in the social sciences (see Oakes & Rossi, 2003; Passel, 2001; Settersten, 1999).

Agency Matters of agency are as complicated as matters of structure. While the ‘‘structure–agency’’ debate has long been central to the discipline of sociology, the ‘‘structure’’ side has been emphasized more than the ‘‘agency’’ side – except perhaps at the boundary between sociological social psychology and psychological social psychology. In psychology, in contrast, the bundle of concepts associated with the notion of agency is recognized as one of its central contributions to the study of lives. Despite this, there is little agreement about what exactly agency is, how it matters theoretically, or how it should be measured. A wide range of concepts – such as ‘‘self-efficacy,’’ ‘‘self-determination,’’ ‘‘locus of control,’’ ‘‘effort,’’ ‘‘mindfulness,’’ ‘‘resourcefulness,’’ ‘‘mastery,’’ and ‘‘autonomy’’ – is often used to index, or used interchangeably with, ‘‘agency.’’ New depictions of individuals as active and self-aware are especially the result of a growing ‘‘constructivist’’ view of human development, in which individuals are seen as the primary architects of their own lives – making their own decisions, creating their own opportunities, and generating their own meanings. This reflects a growing recognition of the need to integrate

40


action perspectives in the field of human development, a trend which is also mirrored in the discipline of sociology (Giele, 2002). These views, for example, have transformed old views of children and childhood, in which the self was treated as something given, and have instead promoted new views of the self as something created and reflexive (e.g., James, Jenks, & Prout, 1998). Constructivist views are equally important for understanding adulthood. Activities of self-regulation and intentional self-development typically become more differentiated and concrete in the transition to adulthood, as independence and autonomy gain in importance and as personal goals and ‘‘identity projects’’ are expected to become clearer (Brandtsta¨dter, 1998). Personal goals and identity projects also undergo a process of continual revision and readjustment throughout adult life, including advanced old age (e.g., Freund & Smith, 1999). They are not only shaped by the input of important others, but are dependent on the plans of other intimates (e.g., spouse or partner, children) with whom one’s life is interwoven (that is, lives must be jointly negotiated and coordinated). They are also shaped by a larger system of cultural norms, as representations of the ‘‘normal’’ or ‘‘expectable’’ life course are incorporated into and presumably guide personal plans. Cultural scripts for life in modern societies seem unclear, and whether these scripts actually affect personal plans seems even less certain (Settersten, 2003). The pervasive focus on personal control and agency in developmental science corresponds with the growing emphasis on internal motivation, planning, decision-making, and open and flexible pathways in societies (Diewald, 2001). An important agency-based concept in sociology has been Clausen’s (1993) notion of ‘‘planful competence.’’ Clausen envisioned the life course as largely the result of personal choice, and positive trajectories as largely the result of planful competence. Clausen argued that planful competence is characterized by three things: dependability, intellectual involvement, and self-confidence. Based on Q-sort methods, Clausen’s measurement of these components has been the source of some debate. But as a concept, planful competence is simple and intuitive: It means knowing your strengths, limitations, and interests, and knowing what options are available and how to take advantage of them. It means being able to assess the actions and feelings of others, and to take these into account when interacting with others. Most importantly, it means having goals and the self-confidence to carry them out, coupled with a high degree of flexibility and openness to new experiences.


41

Clausen’s work is important theoretically because it yields interesting insights, especially when combined with recent theorizing on cumulative advantage and disadvantage over the life course (e.g., Dannefer, 2003). Clausen suggested that planful competence can be discerned by the adolescent years (but that it develops naturally as adults mature), and that it has powerful effects on subsequent life. Adolescents who are planfully competent make good decisions and have successes early in life, the benefits of which cumulate and lead to further good decisions and successes over the life course. If planful competence is a skill that is at least partially learned, interesting interventions might be designed to help instill it in children and young adults. This also raises the question of whether it is ever too late to acquire planful competence, for even individuals in midlife or old age might have much to gain from it, despite their limited time horizons. Of course, even the most planfully competent individuals may not be able to act on opportunities if they do not have the social resources to express these competencies or if other social barriers prevent them from doing so. This is a reminder that forms of agency are constrained by both barriers in the social world and personal resources and capacities. Models of agency are, in part, models of rationality. They take humans to be capable of computing probabilities and joint distributions, and to have substantial knowledge of their environments. But these models must be ‘‘faithful to the actual cognitive capacities of human beings’’ – to real limits in ‘‘knowledge, attention, memory, and other resources’’ (Gigerenzer, 2003, p. 425). Attention to agency is important because the life course is to some extent a personal construction. But the life course also entails selective social processes that sift and sort people into and out of various settings, and open or close opportunities depending on the characteristics of people and the contexts that surround them. These settings and opportunities are not entirely their own doing but are systematically allocated by social forces related to race, sex, age, socio-economic status, and other factors. It is for this reason that joint models of agency and structure, to which we now turn, represent such important ventures for future scholarship.

Agency within Structure Advances in life course research require greater attention to the joint impact of social structure and human agency. Models of agency within structure consider how individuals actively create their own lives and maximize their

42


own development within parameters set by their social worlds (some of which may constrain them, and some of which may enable them), and how individuals interact with and make proactive attempts to alter those worlds. These models require a stronger partnership between life-span psychology and lifecourse sociology (see Diewald, 2001; Mayer, 2003; Settersten, 2005a). Sociologists, for example, have often lost sight of the person and overlooked the roles of personality traits and characteristics, motivations, and action as life course determinants. From the vantage point of sociology, however, psychological factors are often viewed as being outside its realm, given that the purpose of sociology is often understood to be the explanation of the social by the social, a tradition dating back to Durkheim (1895/ 1964). Or, if psychological factors are viewed as relevant, they are not taken to have independent causal significance if they are partially created through social processes. Psychologists, by contrast, have often neglected the powerful social and historical forces that promote or limit development. Given the emphasis on matters of agency in psychology, a working assumption would seem to be that individuals are capable of altering structures, or at least are reciprocally affected by them. Yet modern psychology has focused on family and interpersonal relationships to the exclusion of more distal settings. In addition, perceptions, beliefs, and goals are often viewed as separate and independent from structural opportunities and constraints, which they are not, for social forces may already be present in psychological states, as noted above. The life course is the result of social institutions, culture, and history and the result of decision-making, action, control, and personality (Diewald, 2001). New models of the life course must capture a more complete range of these factors. The latter factors, however, traditionally the domain of psychology, will become increasingly important as the strength of institutions and norms weakens, and as individuals have greater latitude to select or develop their own life scripts. This freedom results in both new possibilities and risks, as the experimental nature of ‘‘do-it-yourself’’ biographies makes them more fragile (Beck, 2000). When individuals choose or find themselves on pathways that are not widely shared by others or reinforced in institutions or policies, they may lose important sources of support and struggle with institutional barriers built on more dominant models. Research in life course sociology and life-span psychology is more compatible than it seems at the first sight. Indeed, the more fine-grained work of life-span psychology, which seeks to unearth the quality and quantity of ‘‘intra-individual plasticity,’’ can offer an empirical base to reinforce


43

sociological research by decomposing macro patterns into their more proximal and micro-genetic causes (Settersten, 2005a). Together, the two fields offer tremendous potentials for moving bodies of knowledge from one discipline to another, and for providing mutually supportive explanations from different levels of aggregation and with different views on causality. The concepts of structure and agency, and the dynamics between them, can be more actively combined in the design and execution of collaborative interdisciplinary work (see also Sibeon, 1999). Several models offer excellent starting points for improving knowledge of agency–structure dynamics, especially if they are developed in more contextual ways. For example, the ‘‘developmental regulation’’ models of Heckhausen (2003) give extensive attention to the selective and compensatory control strategies of individuals, but relatively limited attention, at least empirically, to larger socio-structural forces that constrain or enable developmental potential. Great possibilities also exist in extending the models of ‘‘selective optimization and compensation’’ of Baltes and colleagues (for applications to intellectual functioning, see Baltes, Staudinger, & Lindenberger, 1999), and those of ‘‘assimilative’’ and ‘‘accommodative’’ coping developed by Brandtsta¨dter (e.g., Brandtsta¨dter & Rothermund, 2003), to better incorporate socio-structural forces. Modes of agency within structure cannot be general, but must be understood within particular domains of functioning and particular environments. They must also consider agency not only as an individual phenomenon, but as a collective one – as when many individuals who dare to make innovative life decisions end up creating new options for others, or when entire groups of people acting in concert attempt to forge new patterns through social change and collective movements. These models must also explore how the nature and balance of agency and structure may change within and between individuals as they move through different periods of life. This includes a need to assess individuals’ subjective understandings of their own capacities and resources, as well as those that exist in the settings around them, for these understandings affect how life is interpreted and projected forward. It is important to note, however, that life periods themselves are important elements of social structure. They are a constant part of societies (though their boundaries and content may change); they are widely recognized and have shared meanings; they are reflected and reinforced in law, policies, and institutions; and they are given practical forms in everyday life. We now briefly explore concerns related to agency and structure in three periods: childhood and adolescence; early adulthood through midlife; and old age.

44


STRUCTURE, AGENCY, AND SPECIFIC LIFE PERIODS Childhood and Adolescence Four immediate social settings have been central to understanding the development of children and adolescents: families, schools, neighborhoods, and peer groups and friends (for a review, see Cook, Herman, Phillips, & Settersten, 2002). This literature also contains debates about the relative importance of these four settings on child development, the changing strength and nature of their influence as children become adolescents (especially what the growing influence of peers and friends, and the emergence of romantic relationships and sexual awakening, mean for the other settings), and their additive and multiplicative effects. Given the significance of these four settings and the natural links between them, research on this period of life has examined person-context interactions more than any other period. Family characteristics and processes are probably best articulated, followed by schools, neighborhoods, and peer groups. Much of the focus has been on the negative forms of these settings – the problems that ‘‘bad’’ neighborhoods, schools, peer groups, or families pose for development, and how these settings may be improved to promote positive outcomes (or at least minimize negative ones). The interesting question becomes how and why some kids in bad environments manage to do well, despite these disadvantages, while many others do not (e.g., Furstenberg, Cook, Eccles, Elder, & Sameroff, 1999). There has been much less attention to the equally interesting flip-side of these dynamics: how and why some kids in good environments do not do well, despite their advantages. Given the emphasis on settings of poor quality, views of ‘‘structure’’ often carry negative connotations, while those of ‘‘agency’’ carry positive ones. Here, the agency of children and adolescents, often in combination with that of adults, is seen as being critical to overcoming the challenges of these environments – and the lack of agency, in contrast, is seen as devastating. For example, it is popular belief that success in school, ensured through hard work and effort, is the key to a better life. In reality, of course, success in school is more complicated than this. Hard work and effort alone are often not enough to guarantee success, especially in poor schools and if coupled with other settings of poor quality. Nevertheless, a core component of agency in any of these settings relates to how well children are able to activate whatever resources exist in their environments. This includes building positive relationships with adults in those settings who have the ability to


45

take action on behalf the child, or who can protect or promote a child who might be at risk in some way. The renegotiation of family relationships as children become adolescents and young adults is also an important aspect of agency during this period. Indeed, typical ‘‘acting out’’ in the family or other settings may even be considered an exertion of agency against structure. Many adolescents also begin to develop attachments to the world of work through partemployment and volunteer activities, and become politically and civically aware. These, too, offer important opportunities to actively develop and express one’s self. Of course, agency can also be self-destructive. For example, when parenting strategies are too permissive, when neighborhoods have too little social cohesion, or when schools are disorganized, children may have too much autonomy – which can lead to negative outcomes. There is therefore a tricky balance between structure and agency in promoting positive outcomes for children and adolescents. The literature suggests the strong influence of structural forces during this period, expressed through these four settings, which are created and controlled by adults. The presence and power of these settings takes place against, and must be responsive to, the child’s growing need for independence and autonomy. Much literature, especially in North America, leaves one with the impression that children and adolescents, as dependents, are not yet capable of making their own decisions. The emerging field of childhood studies in Europe is altering this view. This field has been heavily influenced by the United Nations (1989) Convention on the Rights of the Child. It takes seriously the notion that children are competent social actors who can – and should be granted opportunities to – participate in decisions that affect them, define the directions and processes of their own development, and participate in the social world (e.g., James, Jenks, & Prout, 1998). This field has produced innovative studies of how children inhabit and negotiate social settings, including a range of both public and private spaces with varying levels of adult control and supervision, and with varying levels of direct involvement of children in designing and creating those spaces (see, for example, Philo, 2000).

Early Adulthood through Midlife Dramatic changes in transitions to adulthood have given rise to a host of questions about whether current generations of young people are more

46


dependent on parents, less interested in growing up, and more wary of making commitments and taking action for themselves and with others. As movement into adulthood has become more prolonged and complex, parents are now providing extensive support to children well into their 30 s (Settersten, Furstenberg, & Rumbaut, 2005). The types and levels of investments needed to launch even well-positioned young people only exacerbate the needs of vulnerable populations who have little or no family support to back them (Foster & Gifford, 2005; Schoeni & Ross, 2005). Despite the often extraordinary actions required of families to see their children into adulthood today, young people nonetheless express an understanding that they are responsible for their life directions, and that ultimately their own determination and efforts are necessary to achieve their goals (Furstenberg, Kennedy, McLoyd, Rumbaut, & Settersten, 2004). While young people today may reach legal adulthood at the age of 18 or 21, at least in most Western societies, they are often not adults psychologically, socially, or economically, until much later. A brand new challenge to understanding this period is how individuals develop a sense of autonomy amidst increasingly long periods of dependence on others, without strong or clear scripts to guide them, and when the institutions are based on models of early adulthood that no longer reflect the realities of the modern world. Modes of political expression also emerge as concerns during the early adult years, especially as legal rights of adulthood and responsibilities of citizenship are granted (Settersten, 2005b). There is convergent evidence that young adults who wrestle with social issues and participate in civic matters are more likely to be engaged citizens throughout life (Flanagan, 2004). With these shifts come new experiences with collective agency, especially through youth movements, which often coalesce around challenging the status quo and the power to ‘‘make history.’’ Unlike children, adults can more easily mobilize and advocate on their own behalf. In the most extreme of social movements – revolutions – ‘‘structure’’ is something not to be reshaped but to be toppled (e.g., Blechler, 2000). The growing autonomy of youth begins an emphasis on both the ability and need to set developmental goals, harness one’s resources, and exert control over the environment. Attention to these processes have propelled agency-related matters to center stage in literature on adult development, as they are necessary for the successful performance and management of multiple adult roles. With longer and more certain lives, experiences are potentially more predictable and controllable, and careful planning may be increasingly necessary in a world with seemingly unlimited opportunities and time. As suggested in the previous section, adults not only strive for


47

agency in their own lives, but they also hold positions and take actions that directly affect the agency of others, especially children. Recent research, however, has stressed the significant need to understand the ecology of adulthood – how adults shape and are themselves shaped by the social spaces in which they exist (for illustrations, see Gecas, 2003; Settersten & Owens, 2002). New theories and research in these directions are necessary to compensate for the heavy emphasis on individual capacities and resources in research on adult development and aging. In midlife, individuals must confront many new challenges, including changes or stressors related to physical health, memory, personality, emotional development, adaptation and resilience, work and retirement, and family relationships (for overviews, see Lachman, 2001). These challenges, like those of old age, bring new concerns about how to maintain forms or levels of agency in the face of growing personal constraints, whether real or anticipated. Yet some of this literature has also emphasized the new potentials of midlife and the chance to reclaim aspects of the self that were lost or put aside in early adulthood or to develop the self in whole new ways.

Old Age Like earlier periods of adulthood, scholarship on old age continues to emphasize agency through planning, goal-setting, decision-making, even through the end of life. What is unique to old age is that these matters are now assumed to be heavily conditioned by losses in physical, cognitive, psychological, and social capacities, and by increased dependence on others. The changing configuration of capacities in old age constrains the degree to which old people have or can express agency. Indeed, maintaining control over one’s life is understood to be a (if not the) central task of old age, and fears about losing control preoccupy many old people. Many of the dominant models in life-span psychology noted earlier (e.g., Baltes, Heckhausen, Brandtsta¨dter, and their colleagues) have guided this literature. In the light of reduced capacities and a limited time horizon, opportunities for action – and those selectively chosen – become especially meaningful to old people (see also Lang & Carstensen, 2002). The strong emphasis on individual capacities and characteristics in old age has resulted in a surprisingly acontextual view of this period – what Hagestad and Dannefer (2001) call the ‘‘microfication’’ of gerontology – with the exceptions of the vast bodies of literature on health care environments, transitions to assisted living or nursing homes, giving and receiving

48


care, and old age policies (for overviews, see Binstock & George, 2001). In these cases, agency is a significant concern as old people interface with the family, institutions, and the state, and as they jointly walk the blurry and shifting line between dependence and independence. End-of-life issues also result in a widespread interest in spirituality and religion in old age. In some cases, this may signal the relinquishment of one’s own control over life and the placement of control in another entity, or it may signal a shift away from the self and toward generative actions and ideas meant to improve the lives of others or humanity. Gerontology has in the last few decades seriously challenged the belief that old age is a bleak dark period of great losses. Apart from the areas noted above, gerontologists seem committed to promoting positive images of old people and combating negative stereotypes about aging and old age, especially in promoting models of ‘‘successful’’ aging (e.g., Rowe & Kahn, 1998). These models have helped produce portraits of active elders enrolled in school, involved in neighborhoods, productive at work, engaged in politics, and enjoying leisure. All of these spheres offer important opportunities for expressing agency, especially for those who have health and wealth. The lengthening period of old age, like that of early adulthood, should prompt interest in how old people forge pathways through the final decades of life without strong or clear scripts to guide them, how they navigate institutions that have no or outdated scaffolding to support them, and how they manage to retain a sense of autonomy against potentially long periods of dependence on others. We must also ask whether we do ourselves and old people a disservice when we downplay the very real hardships encountered in old age. There is no document parallel to the U.N. Convention on the Rights of the Child, mentioned earlier, that enforces special rights for old people. But like children, many old people around the world do have special concerns related to independence, participation, care, self-fulfillment, and dignity, all of which were recently outlined in the United Nations (1999) Principles for Older Persons as part of the International Year of Older Persons. Finally, there is growing empirical support for the notion that the ‘‘architecture’’ of the life span becomes increasingly ‘‘incomplete,’’ and that the relative influence of biology and culture changes, over the course of adulthood, with cultural influences diminishing and biological influences increasing over time (Baltes, 1997; Li, 2003). The intersections between socio-cultural and bio-genetic forces in different life periods offer many possibilities for exploring structure–agency dynamics, especially during childhood and old age (see Settersten, 2005c). In this section, we have briefly


49

considered some of the ways in which the nature and relative balance of agency and structure may differ across particular life periods. We now turn to a few observations on the life course as a whole.

STRUCTURE, AGENCY, AND THE LARGER LIFE COURSE There is mounting and conflicting evidence that the life course has become both more standardized (with regularity in life course patterns being driven by the increased ‘‘institutionalization’’ brought about by norms, laws, and social policies) and de-standardized (or ‘‘individualized,’’ with variability in life course experiences being driven by the greater choices and control individuals have over their lives). Much contemporary life course research emphasizes the latter, with widespread belief that a wide range of macroand micro-level factors in the last few decades have resulted in life courses that are less conventional, patterned, and predictable, and more risky in private and public spheres alike (Mayer, 2004). This seems especially true of American research, but it is also increasingly true of Western European research. Forms of individual agency are assumed to be central to the emergence of pluralistic life courses. Extending Prout (2005), this diversity is ‘‘locally constructed’’ through repeated interactions between the self and others in immediate environments. The fragility of everyday life in the modern world demands that individuals focus constantly on maintaining and repairing themselves and social relationships. Only rarely do individuals realize that the contour of their lives and nature of their experiences may be shared by many others, wrapped up in larger patterns produced by resources and constraints of settings well beyond their immediate environments – what Mills (1959) called the ‘‘sociological imagination.’’ Even individual agency, as the centerpiece of these models, is often glossed over and taken to be an essential characteristic that requires no explanation. The standardization thesis, which suggests the opposite, rests especially on the strength of welfare states and social policies to regulate particular transitions (e.g., marriage) or the structure and content of life periods. This thesis has mostly been based on Western European research, though international findings are highly variable and heavily conditioned by the type of welfare-state ‘‘regime’’ and its benefits (for an overview, see EspingAndersen, 2002; Mayer, 2001). For example, at one extreme are ‘‘Liberal

50


Market States,’’ such as the United States or United Kingdom, which provide only temporary and limited support under specific circumstances. At the other extreme are ‘‘Scandinavian Social Democratic Welfare States,’’ which provide high degrees of social protection and support across life. In between are ‘‘Continental Conservative Welfare States’’ such as Germany, and ‘‘Southern European Welfare States’’ such as Italy. These views pay special attention to the significance of the state, as a distal force, in determining the structure and content of the life course. Again extending Prout (2005), these views are short-sighted because they take nations and welfare-state regimes to be ‘‘stable and bounded entities.’’ Descriptions of welfare state regimes are important in that these regimes are frames for understanding how the life course is organized in particular societies, and are helpful for explaining cross-national variability in life course patterns. But they reveal little about the dynamic nature of boundaries within and between societies, or of exchanges across these boundaries. More importantly, these regimes ‘‘homogenize’’ forms of the life course within societies because they describe life course patterns more than they explain how the patterns are produced or maintained. They assume that large-scale patterns ‘‘trickle down’’ and explain the action of individual and collective agents, or the options from among which they must choose, rather than seek to understand how individual and collective activities ‘‘percolate up’’ to explain large-scale patterns. Evidence of standardization or institutionalization should also not be interpreted to mean that the decisions and actions of individuals do not matter. Indeed, one could argue that whatever decisions and actions individuals are able to make or take will become even more precious under conditions of standardization or institutionalization. The evidence for the two theses also differs depending on whether the historical view is narrow or wide, and on the phenomena of interest. For example, while the timing of many American life course transitions became more uniform over the course of the twentieth century, especially midcentury, their sequencing simultaneously became more diverse. This is especially true of transitions typically associated with entry into adulthood (Shanahan, 2000). Yet for American women – and contrary to contemporary discussions of the emergence of ‘‘non-traditional’’ family patterns – the timing and sequencing of family transitions has been high, since at least the early decades of the twentieth century (Wu & Li, 2005). This also serves as a reminder that views of institutionalization must span and be differentiated across multiple domains. Kru¨ger and Levy’s (2001) distinctions between three types of ‘‘institutional framing’’ will yield fresh insights into points of positive and negative synergy between institutions,


51

and into the ‘‘not so visible nexus’’ between men and women. Sequential institutionalization structures particular life periods and prompts movement from one period to another. The traditional, but dissolving, lock-step organization of men’s lives – with education early in life, continuous employment in the middle, and retirement at the end – symbolizes this type of institutionalization. Simultaneous institutionalization refers to the attachment of individuals to multiple organizational forms within particular periods. Popular discussions of the difficulties of balancing work and family demands speak to the complexities of simultaneous institutionalization during the early and middle adult years. Adjacent institutionalization relates to the constraints that other institutions bring for managing work and family life, such as those posed by schools, public administration, transportation services, or businesses. Kru¨ger and Levy also remind us that much institutionalization of the life course is not intended or direct, but rather unintended and secondary. These dimensions of institutionalization also warrant greater attention in life course research. The tension between standardization and de-standardization need not be resolved, with one thesis winning out over the other, as much as actively seized to more creatively theorize the life course. Indeed, the evidence for each thesis need not be incompatible and can be simultaneously true, depending on the level of analysis and the target domain or phenomenon under study. Joint attention to these matters – to the tensions between structure and agency, to evidence for standardization or de-standardization, and to the possible connections between them – will advance interdisciplinary research. The most obvious connections to be explored are how (a) structural factors produce standardization and uniformity (at least for subgroups exposed to common forces), and (b) forms of agency produce de-standardization and variability. But consideration should also be given to the ‘‘off-diagonals’’ – to how (c) forms of agency might produce standardization and uniformity (such as when cohort differences in attitudes and values prompt new and widespread decisions about marriage or parenting), and (d) forms of structure might produce de-standardization and variability (such as when disorganization in, or poorly coordinated connections between, educational institutions and the labor market results in disjointed or incoherent experiences). Models of agency within structure, described earlier, are central to understanding how the life course is partly the result of active and free choices, partly created within a fixed set of possibilities and partly imposed from outside – all of which come with consequences, some good and some bad, for individuals, depending on how far their paths stray from others or

52


deviate from those assumed in social institutions and policies. These models will demand clearer definitions and more precise measurement of ‘‘agency’’ and ‘‘structure,’’ their characteristics and processes, and their sources and determinants. A central challenge here is that the lives of individuals and successive cohorts have changed rapidly, but the assumptions that underlie social institutions and policies are often based on outdated models of life (see also Settersten, 2005b). These mismatches may bring serious risks for the functioning of individuals and societies, and there is significant need to rearchitect social institutions and policies so that they better meet the changing needs and realities of individuals and societies. New commitments must simultaneously improve and make more flexible the institutions through which individuals move (including endorsing or permitting a wider range of paths), as well as improve and make more flexible the connections between them. New commitments must also strengthen the skills and resources of individuals so that they can better navigate the life course, for these capacities are vital to ensuring positive outcomes amidst the rapid social dramatic change and great uncertainty of the contemporary world.

REFERENCES Alwin, D. (1995). Taking time seriously: Studying social change, social structure, and human lives. In: P. Moen, G. H. Elder Jr. & K. Lu¨scher (Eds), Examining lives in context: Perspectives on the ecology of human development (pp. 211–262). Washington, DC: American Psychological Association. Baltes, P. B. (1997). On the incomplete architecture of human ontogeny: Selection, optimization, and compensation as foundation of developmental theory. American Psychologist, 52, 366–380. Baltes, P. B., Staudinger, U. M., & Lindenberger, U. (1999). Life-span psychology: Theory and application to intellectual functioning. Annual Review of Psychology, 50, 471–507. Beck, U. (2000). Living your own life in a runaway world: Individualisation, globalisation, and politics. In: W. Hutton & A. Giddens (Eds), Global capitalism (pp. 164–174). New York: The New Press. Binstock, R. H., & George, L. K. (Eds) (2001). Handbook of aging and the social sciences. San Diego: Academic Press. Blechler, M. (2000). Structure and agency, intellectual ‘‘nationalism,’’ and method: Tang Tsou’s contributions to China studies and social science. Modern China, 26(2), 239–247. Brandtsta¨dter, J. (1998). Action perspectives on human development. In: R. M. Lerner (Ed.), Handbook of child psychology Vol. 1. Theoretical models of human development (5th ed., pp. 807–863). New York: Wiley.


53

Brandtsta¨dter, J., & Rothermund, K. (2003). Intentionality and time in human development and aging: Compensation and goal adjustment in changing developmental contexts. In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dialogues with lifespan psychology (pp. 105–124). Norwell, MA: Kluwer Academic Publishers. Bronfenbrenner, U. (1988). Interacting systems in human development: Research paradigms, present and future. In: N. Bolger, A. Caspi, G. Downey & M. Moorehouse (Eds), Persons in context: Developmental processes (pp. 25–49). New York: Cambridge University Press. Clausen, J. A. (1993). American lives: Looking back at the children of the great depression. New York: Free Press. Cook, T. D., Herman, M. R., Phillips, M., & Settersten, R. A., Jr. (2002). Some ways in which neighborhoods, nuclear families, friendship groups, and schools jointly affect changes in early adolescent development. Child Development, 73(4), 1283–1309. Dannefer, D. (2003). Cumulative advantage/disadvantage and the life course: Cross fertilizing age and social science theory. Journals of Gerontology: Social Sciences, 58B(6), S327–S337. Diewald, M. (2001). Unitary social science for causal understanding: Experiences and prospects of life course research. Canadian Studies in Population, 28(2), 219–248. Durkheim, E. (1895/1964). The rules of sociological method (8th ed.). New York: Free Press. Esping-Andersen, G. (2002). Towards the good society, once again? In: G. Esping-Andersen (with D. Gallie, A. Hemerijck & J. Myles), (Ed.) Why we need a new welfare state. Oxford: Oxford University Press. Flanagan, C. A. (2004). Volunteerism, leadership, political socialization, and civic engagement. In: R. M. Lerner & L. Steinberg (Eds), Handbook of adolescent psychology (pp. 721–746). New York: Wiley. Foster, E. M., & Gifford, E. J. (2005). The transition to adulthood for youth leaving public systems: Challenges to policies and research. In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy (pp. 501–533). Chicago: University of Chicago Press. Freund, A., & Smith, J. (1999). Content and function of self-definition in old age and very old age. Journals of Gerontology: Psychological Sciences, 54B, P55–P67. Furstenberg, F. F., Jr., Cook, T. D., Eccles, J., Elder, G. H., Jr., & Sameroff, A. (1999). Managing to make it: Urban families and adolescent success. Chicago: University of Chicago Press. Furstenberg, F. F., Jr., Kennedy, S., McLoyd, V., Rumbaut, R., & Settersten, R. A., Jr. (2004). Growing up is harder to do. Contexts, 3(3), 33–41. Gecas, V. (2003). Self-agency and the life course. In: J. Mortimer & M. Shanahan (Eds), Handbook of the life-course (pp. 369–388). New York: Kluwer Academic/Plenum Publishers. Giele, J. (2002). Life careers and the theory of action. In: R. A. Settersten Jr. & T. Owens (Eds), Advances in life-course research: New frontiers in socialization (pp. 65–88). London: Elsevier Science, Ltd. Gigerenzer, G. (2003). The adaptive toolbox and lifespan development: Common questions? In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dialogues with lifespan psychology (pp. 423–436). Norwell, MA: Kluwer Academic Publishers.

54


Hagestad, G. O., & Dannefer, D. (2001). Concepts and theories of aging: Beyond microfication in social science approaches. In: R. Binstock & L. George (Eds), Handbook of aging and the social sciences (5th ed., pp. 3–21). San Diego, CA: Academic Press. Heckhausen, J. (2003). The future of lifespan developmental psychology: Perspectives from control theory. In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dialogues with lifespan psychology (pp. 383–400). Norwell, MA: Kluwer Academic Publishers. James, A., Jenks, C., & Prout, A. (1998). Theorising childhood. Cambridge: Polity Press. Kru¨ger, H., & Levy, R. (2001). Linking life courses, work, and the family: Theorizing a not so visible nexus between women and men. Canadian Journal of Sociology, 26(2), 145–166. Lachman, M. E. (Ed.) (2001). Handbook of midlife development. New York: Wiley. Lang, F. R., & Carstensen, L. L. (2002). Time counts: Future time perspective, goals, and social relationships. Psychology and Aging, 17(1), 125–139. Li, S.-C. (2003). Biocultural orchestration of developmental plasticity across levels: The interplay of biology and culture in shaping the mind and behavior across the life span. Psychological Bulletin, 129(2), 171–194. Mayer, K. U. (2001). The paradox of global social change and national path dependencies: Life course patterns in advanced societies. In: A. Woodward & M. Kohli (Eds), Inclusions and exclusions in European societies. New York: Routledge. Mayer, K. U. (2003). The sociology of the life course and lifespan psychology: Diverging or converging pathways? In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dialogues with lifespan psychology (pp. 463–481). Norwell, MA: Kluwer Academic Publishers. Mayer, K. U. (2004). Whose lives? How history, societies, and institutions define and shape life courses. Research in Human Development, 1(3), 161–187. Mills, C. W. (1959). The sociological imagination. New York: Oxford University Press. Oakes, J. M., & Rossi, P. H. (2003). The measurement of SES in health research: Current practice and steps toward a new approach. Social Science and Medicine, 56(4), 769–784. Passel, J. S. (2001). Censuses: Demographic issues. In: N. J. Smelser & P. B. Baltes (Eds), International encyclopedia of the social and behavioral sciences (pp. 1599–1605). Oxford: Elsevier. Philo, C. (2000). The cornerstones of my world: Editorial introduction to special issue on spaces of childhood. Childhood, 7(3), 243–256. Prout, A. (2005). The future of childhood. London: Routledge Falmer. Riley, M. W., & Riley, J. W., Jr. (1999). Sociological research on age: Legacy and challenge. Ageing and Society, 19(1), 123–132. Rowe, J., & Kahn, R. (1998). Successful aging. New York: Pantheon. Schoeni, R., & Ross, K. (2005). Material assistance from families during the transition to adulthood. In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy (pp. 396–416). Chicago: University of Chicago Press. Settersten, R. A., Jr. (1999). Lives in time and place: The problems and promises of developmental science. Amityville, NY: Baywood. Settersten, R. A., Jr. (2003). Age structuring and the rhythm of the life course. In: J. Mortimer & M. Shanahan (Eds), Handbook of the life course (pp. 81–98). New York: Kluwer Academic/Plenum Publishers.


55

Settersten, R. A., Jr. (2005a). Toward a stronger partnership between life-course sociology and life-span psychology. Research in Human Development, 2(1–2), 25–41. Settersten, R. A., Jr. (2005b). Social policy and the transition to adulthood: Toward stronger institutions and individual capacities. In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy (pp. 534–560). Chicago: University of Chicago Press. Settersten, R. A., Jr. (2005c). Linking the two ends of life: What gerontology can learn from childhood studies. Journal of Gerontology, 60(4), S173–S180. Settersten, R. A., Jr., Furstenberg, F. F., Jr., & Rumbaut, R. G. (Eds) (2005). On the frontier of adulthood: Theory, research, and public policy. Chicago: University of Chicago Press. Settersten, R. A., Jr., & Owens, T. (Eds) (2002). New frontiers in socialization. London: Elsevier. Sewell, W. H., Jr. (1992). A theory of structure: Duality, agency, and transformation. American Journal of Sociology, 98, 1–29. Shanahan, M. J. (2000). Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective. Annual Review of Sociology, 26, 667–692. Sibeon, R. (1999). Agency, structure and social chance as cross-disciplinary concepts. Politics, 19(3), 139–144. United Nations. (1989). Convention on the rights of the child. New York: Author. http:// www.uniceff.org.crc United Nations. (1999). Principles for older persons. New York: Author. http://www.un.org/sea/ socdev/iyop/iyoppop.htm Wu, L., & Li, A. (2005). Historical roots of family diversity: Marital and childbearing trajectories of American women. In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy (pp. 110–149). Chicago: University of Chicago Press.


56

AGENCY, EVENTS, AND STRUCTURE AT THE END OF THE LIFE COURSE$ Victor W. Marshall My purpose is to address a number of related conceptual issues that deal with the basic, conceptual building blocks of the life course perspective. These building blocks have been described, over time and from different perspectives, in different languages. There is the language of the psychology of human development, with its stages, crises, and developmental transformations. There are different languages within sociology, describing careers, status passages, events, transitions, trajectories, and so forth. Many of these terms appear to be synonyms, different words for the same underlying concept. There is also at least one case in life course scholarship where one word is used with many different underlying concepts: that is the word ‘agency’,

$

This paper draws in part on an earlier paper, ‘‘Agency, Structure, and the Life Course in the Era of Reflexive Modernization’’ presented at the American Sociological Association meetings, Washington, DC, August 2000. I have had the pleasure to discuss many of these ideas with my colleague, Glen Elder at UNC; with Margaret Mueller, our former student at UNC; with two of my former students and continuing colleagues, Philippa Clarke (now at Duke University) and Julie McMullin (now at The University of Western Ontario). I am indebted to them for their insightfulness, intellectual honesty and collegial support. I also thank the students in my graduate seminar in current issues in sociological theory, Department of Sociology, University of North Carolina, with whom many of these ideas were discussed.

Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 57–91 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10002-1

57

58

VICTOR W. MARSHALL

which takes on many meanings in different disciplines, but also within the same discipline and indeed within the writings of the same author. I will address three questions with reference to my own early research on the sociology and social psychology of aging and dying. Thus, the paper reinterprets work that I did, for the most part prior to the emergence and formalization of the life course paradigm, in order to address conceptual and methodological issues of today in life course terminology. The first question that I shall address is conceptual that asks whether there exists, in a given society, values for human development and the life course, that is, a shared view of what is a good life course. I will examine this with reference to the concept of the ‘good death’. The second, methodological, question asks how the different concepts (such as trajectory, stage, transition, and event) are usually operationalized, and with what consequences. I will examine this with respect to the comparison of ‘objective’ with ‘subjective’ data about the dying trajectory. The third, results, question asks to what extent changes in status result in the redefinition of identity. I will address this question by referring to the distinction between self and identity, and to the action of the aging and dying individual, in managing identity. I am thus revisiting my early scholarship undertaken before the life course perspective came to be institutionalized in the social sciences. I am personally curious to see how the life course perspective can shed new light on my scholarship of 30 years ago; and I hope that this exercise will help to clarify some of the concepts of the life course that are currently in vogue. Older individuals face objective changes in their life situation and their sense of themselves in time. Events and the entire past and anticipated life course have taken on a new meaning in an interplay between what is increasingly referred to as agency and social structure. I find that terminology problematic and, before turning to examine these three research questions in light of my data on aging and dying, I will deal at some length with the concept of agency and the agency–social structure linkage in sociology.

1. AGENCY Every social theory, whether implicitly or explicitly, has to deal with two things – energy and direction. Without energy there would be no action. One metaphor for this is heating a house. You need to have a furnace to produce heat. But to understand the heating of a house you also have to account for direction – how and where that warm air moves from the furnace, through ducts, to the various levels and rooms of the house (this metaphor rests on a

Agency, Events, and Structure at the End of the Life Course

59

more North American than European concept of how heating systems work). Turning to social theory, the two easiest examples would be psychoanalytic thought and symbolic interactionism. For Freud and his followers, the id is the furnace, it is the source of energy. But that energy needs channeling, and the complexities of psychoanalysis provide an explanation for how this energy is channeled into socially acceptable behavior (or not). In symbolic interactionism, the ‘I’, acting in the dialectic of the self process, is a necessary hypothetical construct if the perspective is to move beyond stasis to process. One variety of symbolic interactionism, the Iowa School, downplays the ‘I’ to focus on the ‘me’ (which is akin to the duct work in the furnace metaphor in steering behavior). In emphasizing the me, it loses the sense of process or dynamism that is such an important contribution of the symbolic interactionist perspective. Agency lies somewhere in the area of energy. The term has increasingly popped up in the sociology of the life course, but rarely is it defined or explored in great detail. Even more rarely is agency measured. Is agency just a new name for something old, such as the concept of action or the concept of voluntarism? The term agency is used in conflicting ways, but always has something to do with choice. Dannefer and Uhlenberg decry the way choice is handled in life course research and theorizing: In the study of action, choice is a problem to be analyzed, not an accomplishment to be asserted (Dannefer, 1999). Given the problematic epistemological and ontological status of ‘‘choice’’ in the wider social science literatureyits remarkably unproblematic appearance in life course theory cannot be defended. What is almost always measured in such discussions is behavior, and it is simply presumed that behavior is based on choice (Dannefer & Uhlenberg, 1999, p. 312).

The fuzzy and conflicting treatment of choice and agency has also been recognized as a major problem for more general sociological theory. Thus, Meyer and Jepperson (2000, p. 101) charge that ‘‘ythere is more abstract metatheory about ‘actors’ and their ‘agency’ than substantive arguments about the topic.’’ It is a concept more often invoked than measured.

1.1. Agency as a Principle of the Life Course Glen Elder has been more specific than anyone else in the aging and life course domain to define this concept. For Elder, agency is one of five defining principles of the life course. Here is the principle: ‘‘Individuals construct their own life course through the choices and actions they take within the opportunities and constraints of history and social circumstances’’

60

VICTOR W. MARSHALL

(Elder & Johnson, 2003). Settersten (2003, p. 39) lists the same as ‘‘an ‘emerging proposition’ of the life course perspective: The life course is a partial product of human action. It is actively created within the confines of the social worlds in which individuals exist. Life course scholarship therefore promotes models of human agency within structure.’’ But this principle has a long history in developmental psychology and life course sociology. I would therefore like to review a bit of that history.

1.2. Agency as Production of a Life Of many arbitrary starting points,1 I will begin with a largely psychological publication in the human development area. In the Foreword to Lerner and Busch-Rossnagel’s (1981a) important edited volume, Individuals as Producers of their Development: A Life-Span Perspective, sociologist Orville Brim notes that ‘‘The idea that organisms act to create environments to elicit responses from themselves is not new.’’ However, Brim argues that the 1981 volume is the first to treat the idea broadly and in the context of the theory of life-span development. While the term, ‘agency’, does not seem to be used in the volume, the idea of agency is there. As Brim (1981, pp. xv–xvi) puts it: Behind this idea, to be sure, is the view that the organism is dynamic, powered by curiosity, growth, expansion, and a drive toward mastery over itself and its world; and also by the development during the first two years of life of a sense of self as a distinctive being, and the construction of images of future selves that are different from what one is now. Behind the idea is also the view that organisms are open to change, are much more malleable than heretofore thought, and that the consequences of early experience and biological endowment are transformed by later experience.

The Lerner and Busch-Rossnagel volume lays a broad foundation, from psychology, evolutionary biology, and anthropology, for a view that individuals are not only produced by, but also produce their world. This work suggests that the capacity for organisms to produce their own world varies both ontogenetically and phylogenetically. In terms of individual human lives and ontogenetic development, the capacity to act on the world and to have greater ‘plasticity’ of function increases with human development, and this capacity can be described in terms of physiological changes in the brain and, more broadly, the entire organism (Lerner & Busch-Rossnagel, 1981b).2 I will turn to the sociological foundations of this same notion later, but in the life course perspective, which is explicitly interdisciplinary, this is an important line of theoretical reasoning related to the concept of agency, and one that continues to the present (e.g., Diehl, 1999).


61

Without using the term agency, in his Foreword to the volume Brim raises the possibility that agency might be considered a variable as contrasted with simply an aspect of human nature – the ability to choose: ‘‘Throughout this volume the subject is humanity and the concern is with the species rather than with individual differences. Certainly there may be individual differences in degree of mastery of, and re-creation of, the environment because of individually endowed or acquired differences, but the idea refers to the main thrust of the human animal, not an occasional remarkable human being’’ (Brim, 1981, p. xvi).

1.3. Agency as Environmental Proactivity or Adaptation Another line of research in psychology, at times related to the exploration of Erikson’s developmental framework, either explicitly uses the concept of agency or uses related terms, which are themselves conceptualized in terms of agency, and which considers agency to be a variable, and measurable, property of individuals. Lawton (1989), whose earlier work on person– environment fit and the ‘environmental docility hypothesis’ viewed the individual as largely reactive to environmental limitations and pressures, subsequently introduced the hypothesis of ‘environmental proactivity’, to recognize ‘‘action and agency – the person’s competence as a determinant of environment’’ (Lawton, 1989, p. 140). Building on Lawton’s conceptualization, Kahana and Kahana (1996) used the term agency in a paper promoting the concept of ‘proactive adaptation’. Here is what they say: Dependency models of aging have emphasized the propensity of older adults to be passive respondents to environmental influencesy. However, there has been a small but growing group of gerontological researchers, who have recognized that older persons can play significant proactive roles and behave in ways that draw upon and can generate resources in their environment (Lawton, 1989). This orientation parallels theoretical developments in the broader field of sociology (Giddens, 1983) that increasingly recognize the role of agency, reflecting progress, intentionality, and responsibility in the actions of human beings.

1.4. Agency as Masculine Trait The term agency has been used by psychologists investigating the concept of ‘ego strength’, and studying the relationship between agency and Eriksonian ‘generativity’. Gutmann (1965) noted that the psychological concept of ‘ego

62

VICTOR W. MARSHALL

strength’ had decidedly masculine properties. Building on Gutman, Bakan distinguished between ‘agency’ and ‘communion’, with the former referring to ‘‘the existence of an organism as an individual’’ and the latter to ‘‘the participation of an individual in some larger organism of which the individual is a part. Agency manifests itself in self-protection, self-assertion, selfexpansiony’’ (Bakan, 1966, p. 15). The life course sociologist Alice Rossi (1980, p. 9) uses the term ‘affiliation’ instead of ‘communion’, in order to ‘‘avoid any theological association with the concept’’, but retains Bakan’s concept of agency. In a more recent paper, psychologists measured agency as both a psychological trait and a characterization of behavior (Ackerman, Zuroff, & Moskowitz, 2000). Although an independent definition of the construct, agency, was not given, the conceptualization relates it to ‘masculine’ as opposed to ‘feminine’ traits and behavior. The self-report measure of agency is based on the measurement of psychological traits through Likert scales to assess ‘self-assertive-instrumental traits’ versus ‘interpersonal-expressive traits’. These generate ‘masculinity’ and ‘femininity’ scores that are considered to represent agency (masculine) or communion (feminine). An alternative adjective-rating scale contrasts ‘love’ and ‘status’, the latter measure a conceptual domain ‘‘from dominance to submission and is thought to represent agency’’ (Ackerman et al., 2000, p. 30). Behaviorally, self-described behaviors were coded on four dimensions: dominance and submission, and agreeableness and quarrelsomeness. Dominance minus submissiveness is considered to represent agency, while agreeableness minus quarrelsomeness is considered to represent ‘communion’. Agency, measured in these ways, is implicated in predicting generativity (as conceptualized by Erikson), and this relationship, the authors speculate, is because ‘‘creating is a primarily agentic form of generativity’’, such that ‘‘highly agentic individuals manifest generative concern focused on creatingy’’ (Ackerman et al., 2000, p. 37).3 In this psychological formulation, it is clear that agency is considered to be a variable personality trait of the individual. The issue of measurement of agency, an issue that is boldly addressed in the article by Ackerman et al., cannot be dismissed. For example, George (1999) who is sympathetic to an ‘agentic’ perspective on the self in relation to society, points out that ‘‘Researchers typically assume that behavior observed in natural settings is triggered by self-perceptions and self-related motives, but that view remains more assumption or interpretation than documented fact’’ (p. 47). George is here arguing that the aspect of agency asserted by many, often qualitative, researchers is nothing but the invocation of an unmeasured, hypothetical construct. The construct is inferred


63

from the behavior that it is presumed to cause. If Ackerman’s measurement approach is problematic (as it seems to me), this may be more in its execution than its intent.

1.5. Agency as Making Possible ‘Loose Coupling’ (Agency as ‘Unexplained Variance’) Let us return to Glen Elder. Elder (1985, p. 42) has noted that ‘‘An agentic concept of individuals in shaping their own trajectory has been a central principle of the life course framework.’’ Indeed, Elder has made it so. He notes that conceptions of human agency have characterized ‘life-history’ studies at least since the classic study of the Polish peasant in Europe and America, by Thomas and Znaniecki (1918), Elder (1997), and Elder and Johnson (2003). Elder sees agency as important in selection processes:4 ‘‘Within the constraints of their world, competent people are planful and make choices among alternatives that form and can recast their life course’’ (Elder, 1997, pp. 964–965). Agency refers to choices among available options, and Elder relates this to his concept of ‘loose coupling’: ‘‘Loose coupling reflects the agency of people even in constrained situations as well as their accomplishments in rewriting their journeys in the course of aging’’ (Elder, 1997, p. 965). Constraint is accounted for, in Elder’s formulation of the life course perspective, by ‘‘The principle of historical time and place: The life course of individuals is embedded in and shaped by the historical times and places they experience over their lifetime’’ (Elder & Johnson, 2003, p. 62). Elder’s own work has emphasized the impact of the depression and WW II as structuring choices. Agency can only manifest itself through choice, and choice is possible only if there are alternatives. In Elder’s usage, agency makes possible ‘loose coupling’, which in turn is a ‘principle’ of life course theory that makes room for departures from structure. In a way, agency functions in this theoretical perspective in the same way that ‘unexplained variance’ functions in statistical models: if behavior is not patterned structurally, then it must reflect resistance to structure. As Elder and O’Rand (1995) put it, ‘‘Loose coupling reflects the agency of individuals even in constrained situations as well as their achievements in rewriting past journeys in the course of aging’’ (p. 456). The relationship of agency to structure is complex for Elder. Agency is presumably necessary if individuals are to operate either within or outside the boundaries of social structure. As Elder and O’Rand (1995, p. 457) argue, ‘‘Age grades and loose coupling exemplify two sides of the adult life course – its social regulation

64

VICTOR W. MARSHALL

and the actor’s behavior within conventional boundaries, and even outside of them’’.

1.6. Agency as Overcoming Resistance While Elder seems to argue that humans always exercise agency (from which we might infer agency is a property of human nature), the way in which Elder describes agency sometimes leaves the implication that it is manifested only at critical turning points of the life course, or in resistance to the established social order. I will first deal with the latter: agency as overcoming resistance. Thus, Elder (1997) argues that loose coupling reflects the antithesis of age grading, and that the agency involved in loose coupling exemplifies ‘‘the actor’s initiatives and interpretations that press for individuality and deviations from convention’’ (p. 965). This would imply that agency is not exemplified if someone follows convention, complies with social norms or social control pressures, or leads a life in conformity with social institutions. This is a much narrower view of agency than, for example, Giddens’ arguments. The same view is found in Heinz (1996, p. 57), who notes that while biography is constructed with guidance from ‘‘institutional standards and the unequal distribution of resources for building continuous life paths’’, ‘‘Such limitations must not be seen as fateful constraints, they can be overcome by human agency.’’5 In setting agency as against social structure, Elder and others adopt a specific, and questionable, stance about the link between the individual and social structures. This point has been vigorously made by Dannefer and Uhlenberg, and can also be pursued through Giddens, Gubrium and others. Dannefer (1999) links agency in a fundamental way to the processes of world construction without which there would be no social system. As he puts it (p. 73), ‘‘yhuman behavior is purposeful; it is not guided by instincts but by intentions (Weber, 1978).’’ The interaction of intentional actors coproduces the social system at both micro and macro levels and the self-hood or social (human) nature of the actors themselves (Dannefer, 1999).6

1.7. Agency as Evidenced in Life Transitions Moreover, at times in Elder’s writings, agency is indicated not in the smooth flow or the routines of everyday life, but in life’s more dramatic moments, moments of transition rather than of continuity, even if the notion of


65

resistance is not implied. This theme runs through Elder’s work. Thus, ‘‘No idea better illustrates the contemporary link between social context and the agency of the individual than the concept of life transition, which defines the problem as a change in states – social and psychological. Adults bring a history of life experiences to each transition, interpret the new circumstances in terms of this legacy, and work out adaptations that can alter their life course. When transitions disrupt habitual patterns of behavior, they provide options for new directions in life, a turning point’’ (Elder & O’Rand, 1995, p. 456). This notion that agency is found only at transition points or, as Giddens (1991a, p. 113) calls, them, ‘‘fateful moments,’’7 or the related notion that it is found only in resistance to social barriers, is at odds with the notion that agency is a property of human nature. Three different notions of agency are found, I believe, in Elder’s work (capacity, resistance, transition), leading to the question of whether different terms should be used to more clearly indicate these three meanings.

1.8. Agency as Responsibility As the last in this long list of usages of agency, let us turn to the notion of ‘being an agent’ for someone or something. While this is perhaps the most commonsensical, lay-usage meaning of agency, it is little found in life course studies. Rather, it is considered to be a sub-theory of rational choice theory (Kiser, 1999) that draws on Max Weber’s sociology and the ‘new institutionalism’ in economics. According to Kiser (1999, p. 146), ‘‘An agency relation is one in which a ‘principal’ delegates authority to an ‘agent’ to perform some service for the principal’’. This notion can, however, be related to life course theorizing. Meyer and Jepperson (2000) have recently provided a detailed theoretical examination of the social construction of agency, in which they argue that the modern actor is a cultural construction and is ‘‘yan authorized agent for various interests (including those of the self)’’ (p. 101). For them, agency is ‘‘legitimated representation of some legitimated principal, which may be an individual, an actual or potential organization, a nation-state, or abstract principlesythe concept ‘agency’ draws attention to the devolution of external authority, and to the external legitimation and chartering of activity’’ (Meyer & Jepperson, 2000, note 2). I can act as an agent for someone else or someone else can act as my agent. But I can also be constructed as having authority to act for myself, to be my own agent, to act on behalf of myself. This may be formulated in terms of the I–me distinction found in G. H. Mead (as the authors acknowledge) and

66

VICTOR W. MARSHALL

symbolic interactionism (p. 111). In life course parlance, Elder and O’Rand (1995, p. 465) note that ‘‘ypeople function as agents of their own life course and development’’. In terms of the sociology of the risk society and its implications for the life course, as in so much European life course theorizing (see Marshall & Mueller, 2003), a move to the risk society may be characterized by an increasing delegation of authority for life course management from society to the individual. Moreover, Meyer and Jepperson (2000) maintain, in the modern world I am more and more called upon to assume such agency, in large measure because of a decline in the attribution of agency to spiritual forces: ‘‘Some agency is built into modern pictures of the agentic authority and responsibility of the state and other organizations; much devolves to the modern individual, who is empowered with more and more godlike authority and vision’’ (p. 105). This notion of agency therefore extends to ‘being responsible for your self’, in which case it points to the reflexivity of the self – taking responsibility for the self as an object. In summary, I have reviewed many ways in which the concept of agency, either directly so named or denoted by a closely related term, has enriched life course and aging studies. In a very general way, agency has been seen as the production of a life. The agent is the producer; human development, the lived life, the narrative, is produced by agency. In more specific formulations along the same lines, agency has been viewed in terms of environmental proactivity or adaptation. This recognizes that people not only react but act and, in acting, produce their biographical selves. Agency has also, in a few studies by psychologists, been seen as a masculine trait – doing rather than being, taking charge, making things happen. This is clearly not logically consistent with the previously discussed notions, which can all be grouped as making an assertion about the nature of being human. Rather, it postulates that agency is a variable property of humans, albeit one much more often found among men than women. In the most widely accepted theoretical statements about the life course, by Elder and those who have adopted his language, agency has been seen as the force making possible ‘loose coupling’, thereby somehow accounting for the fact that all people do not follow standardized, institutionalized life courses. But I question whether this adds anything to an explanation. Is it not the same as depicting ‘unexplained variance’ in multivariate models? One can draw a ‘causal arrow’ to show where variance is unexplained, but the arrow emanates from something unexplained. If you push that kind of argument too far, you end up with unexplained variance as the ‘uncaused cause’, which St. Thomas Aquinas argued was proof for the existence of God.


67

While the above usages of agency are quite general, some uses are highly specific, referring to agency only if the individual is overcoming or resisting social structural barriers, or only if the behavior of the individual is with respect to a major or important life transition. In such formulations, people would not be ‘agentic’ most of the time, but only at critical points in their lives. And, finally, I noted the usage that suggests agency refers to culturally legitimated responsibility to act – on behalf of others, of organizations or ideas, or of one’s own self. This is quite a general conceptualization of agency, but more restricted than the first ones, which would view the capacity to act as agency, regardless of its cultural or social legitimation. This usage of the concept of agency also treats it as a variable, and the authors I drew on specifically trace a historical shift of agency from spiritual forces to institutions and to individuals.

2. MAKING SENSE OF AGENCY: TOWARD TERMINOLOGICAL CLARITY To summarize, I have noted that the term agency has many meanings in the sociology and the psychology of aging and the life course. I find it generally useful to use specific terms for specific meanings. It may be necessary to distinguish between four different sorts of things: (1) the human capacity to make a choice, that is, to be intentional; (2) the resources within the individual or at the command of the individual that can be brought to bear in intentional or agentic behavior; (3) behavior of individuals that reflects intention; and (4) the social and physical structuring of choices. As I turn to elaborate on these, I note that each of these four constructs has a metaphorical place in the analogy of the furnace that I developed earlier. By the first, I mean that a social theorist may propose that a capacity to exercise choice is a fundamental aspect of human nature. In terms of the metaphor, this corresponds to the physical fact of the furnace. Without the furnace there is no capacity to heat the house. This is not to deny that this capacity develops in the human over time. Such a capacity would include awareness and thus requires cognitive capacities to give identity to objects and events in the world. This can be seen as agency in this first sense and it can be viewed as a developmental capacity of (virtually) all humans.8 I recommend using the term agency in this way to refer to the human capacity (as aspect of what it is to be human), to act intentionally, planfully,

68

VICTOR W. MARSHALL

and reflexively and in a temporal or biographical mode. Except in explicit developmental work, most life course researchers would not need to treat or measure this as a variable. Rather, it is an assumption about human nature, of the order, ‘‘all human beings have free will’’. By the second construct, the resources that can be brought to bear in agentic behavior, I refer to both personal capacities of the individual and resources at his or her command. Metaphorically, this corresponds to the fuel for the furnace. Some furnaces might have more, or better, fuel than others. Personal capacities might be intelligence or stage of cognitive development, learned skills or abilities, knowledge, or physical strength and talent. Resources might be economic (wealth), social (social capital, networks, contractual or informal social ties, and alliances). I do not think it is helpful to use the term agency to describe this variable property, which is better called resources or perhaps assets. Resources can be measured, although with care to include those ‘within’ the individual (e.g., education), and external to the individual but at his or her ‘command’ (e.g., social capital). Clausen calls ‘‘planful competence’’ the ability to make informed, rational decisions, and set realistic short- and long-term goals (Clausen, 1991). I would consider ‘planful competence’ to be a resource, rather than a capacity. Clausen (1991, 1993) saw planfulness as indicated by self-confidence, intellectual investment, and dependability. Shanahan, Elder, and Miech (1997, p. 59), drawing on Clausen’s concept of planfulness, defined ‘planful competence’ as ‘‘The extent to which an individual is aware of his or her knowledge, abilities, and interests; pursues goals; and reflects on important decisions’’. They found ‘‘reasonable approximations’’ for these dimensions in the Stanford–Terman data they used in parents’ assessments of these traits in their children. These are not basic traits of human nature, but variables in the ability to do the things that humans do. There is no reason why measurement of these dimensions of planful competence, as a resource, could not be further developed. Presumably, more objective measures would include not only the awareness of an individual of his or her knowledge, abilities and interests, but the actual existence of these things regardless of awareness of them. One can have an ability without knowing it, and this might have resource consequences even if not at the level of awareness. By the third construct, the actual behavior that reflects intention, I refer to what has been called ‘action’ or ‘voluntaristic action’ or ‘social action’ by a host of sociologists who rarely used the term ‘agency’. I think the term ‘action’ is appropriate here, although I have a trained incapacity, as a sociologist, to conceive of action as other than social (see, Campbell, 1996). This term is itself manifestly used in various contradictory ways, as Colin


69

Campbell (1996), Derek Layder (1994), and others have shown, but I think it will be useful if specified in this way. In terms of the furnace metaphor, this construct would be captured by adjusting or programming the thermostat and opening and closing duct openings and valves. By the fourth construct, the social and physical structuring of choices, I refer to the de facto structure of opportunities or life chances that is open within the range of action of the actor. Metaphorically, the construct corresponds to the duct work, the degree of insulation, the leaky windows, and other structural aspects that pose barriers to effective home heating – regardless of how good the furnace, the fuel, and the thermostat. The concept of ‘life chances’ (Dahrendorf, 1979; Weber, 1978) is useful here, so long as it is recognized that life chances are not static but emanate from social processes. It is useful to emphasize the co-constitution of self and society, through which action creates social structure just as social structure constrains, or opens up, possibilities for choice and, thereby, ‘structures’ action. A large number of life course theorists have enjoined their colleagues to avoid theorizing ‘‘agency without structure’’ or ‘‘structure without agency’’ (George, 1999; Ryff, Marshall, & Clarke, 1999; McMullin & Marshall, 1999; Marshall, 1995, 1996; Settersten, 1999, p. 223). The broader challenge is to theoretically address the linkages between agency and structure or self and society (Campbell, 1996; Layder, 1994; Ryff & Marshall, 1999).9 Moreover, if agency is difficult to define, so too is social structure (Alwin, 1995).10 I turn next to that conceptual issue.

3. DEFINING SOCIAL STRUCTURE Despite its centrality to the very core of sociology as a discipline, a plethora of definitions exists for the term and its conceptualization varies greatly. I relate a conceptualization of social structure directly to the aforementioned Weberian concept of life chances, and I draw very heavily on Sewell for this. If it is assumed that human beings have agency (the capacity for choice) and that they manifest agency through their actions, then the dilemma is to find a way to define social structure in a way that recognizes agency and action without sacrificing the notion of social constraint. Berger and Luckmann (1966, p. 18) approach the action-structure problem by outlining three dialectical processes in which ‘‘subjective meanings become objective facticities’’. The first of these is ‘externalization’, the process through which individuals create their social worlds through their action.

70

VICTOR W. MARSHALL

The second is ‘objectivation’, through which once created, these social constructions take on a reality of their own, independent of their constructors. The third process, ‘internalization’ refers to the process by which the ‘‘objectivated social world is retrojected into consciousness in the course of socialization’’ (Berger & Luckmann, 1966, p. 61). As Smelser (1997a, b) has noted, this concept does not fully capture the ways in which reality constrains action. Rather, objectivated reality ‘acts back’ upon individuals through a number of ‘reality maintenance’ devices, often making use of language and ritual, and with institutional and physical traces as well (see McMullin & Marshall, 1999). It is the latter – ‘institutional and physical traces’ – that moves the conceptualization somewhat beyond Giddens’ (1991b) concept of the duality of action and structure. There is something more enduring than the instantiation of reproduced social structure through action, and this ‘something’ can take a number of forms, including legal arrangements, rules and regulations, and the capacity to mobilize resources.11 For example, an army is more than the activities of its personnel. It includes its authority and reporting structure and, not the least, its personnel, weapons, and supplies. It is true, as Giddens (1984) argues, that these other things come to be through human action over time, but they are ‘real’ and constraining of action all the same, and independent of the action that constitutes them and certainly more than ‘‘memory traces, the organic basis of human knowledgeability, and as instantiated in action’’ (p. 377). As Sewell (1992, p. 3) remarks, ‘‘ythe notion of structure does denominate, however problematically, something very important about social relations: the tendency of patterns of relations to be reproduced, even when actors engaging in the relations are unaware of the patterns or do not desire their reproduction’’. While Berger and Luckmann are helpful in emphasizing ‘objectivated’ social reality as one moment in the dialectic, I would emphasize an insight from one of their intellectual sources, Alfred Schutz (1964), which is that people are not fully free to fashion their own lives because they are born into a world of predecessors. Archer (1995) suggests that we have to analytically stretch out the three modes of a dialectic similar to that of Berger and Luckmann’s in order to recognize that there are objective (material) conditions that already exist prior to the initiation of action.12 For her the modes are structure, interaction, and structural elaboration. For Sewell, ideas of social structure are more than the internalized ‘rules’ of Giddens (and perhaps Berger & Luckmann), but they are socially shared ideas, which he prefers to call ‘schemas’, and which I would consider to be cultural; and for Sewell (1992, p. 13), ‘‘Sets of schemas and resources may properly be


71

said to constitute structures only when they mutually imply and sustain each other over time.’’ And in this framework, which I endorse, as Sewell states it, ‘‘Agencyyis the actor’s capacity to reinterpret and mobilize an array of resources in terms of cultural schemas other than those that initially constituted the array’’ of resources available to them. Agency is then the human capacity to make something new, to contribute to innovative social production of the world.13

4. AGENCY AND STRUCTURE IN THE LAST CHAPTERS OF LIFE At this point, I will use these constructs to briefly summarize the findings of my research on the social psychology of aging and dying, and relate the discussion to the specific research questions mentioned at the beginning.

4.1. My Early Research in Aging and Dying I will briefly outline my research in the sociology of aging and dying, which began with my doctoral dissertation work. Georg Simmel has set out a basis for my research questions early in the last century ywe are, from birth on, beings that will die. We are this, of course, in different ways. The manner in which we conceive this nature of ours and its final effect, and in which we react to this conception, varies greatly. So does the way in which this element of our existence is interwoven with its other elements (Simmel, 1950).

I set out, in my dissertation, to find some answers to the questions implied in Simmel’s statement, to wit: How is it that people die in different ways? How is it that people conceive their nature and its ‘final effect’? How do people react to their conception of death? In what way is the individual’s conception of his mortality interwoven with other elements of his life? This study focuses on the aging, and attempts to assess the implications of the fact that the aging are also dying. It is, therefore, not a study of death or dying per se, but rather of living under the growing realization of impending death (Marshall, 1972, p. 2).

In contemporary life course terms, I was interested in the final trajectory before the transition to dying, and to investigate this trajectory I would have to identify the transition that led into this trajectory. We are all dying of course, but my interest was in that period in which the individual became more highly aware of this fact, and I used the term ‘awareness of finitude’ to

72

VICTOR W. MARSHALL

describe it. I pursued this research in a field study using participant observation and semi-structured interviews in a middle-class retirement community and a nursing home that had a clientele of a much lower social class background.14 I was not using terms that predominate in current life course usage. Leonard D. Cain Jr. (1964) had already published the first major statement about the life course in an important essay, ‘‘Life course and social structure’’, in which he aimed ‘‘to identify, isolate, and systematize a life course, or age status, frame of referencey’’, as well as to ‘‘contribute to the advancement of a sociology of age status’’ (Cain, 1964, p. 273). But I did not cite it. He had been influenced by Anselm Strauss, and I did draw on Strauss, and his concept (with Glaser) of status passage, which is a direct steal from the career concept of Everett Hughes (1971). But it was only in developing publications from my dissertation that I moved deeply into using the career and status passage concepts (Marshall, 1978–1979, 1980). This was before much of the current life course terminology had been reintroduced by Elder (1975) and Riley (1979). After more than a decade of ‘chewing on’ my data, I developed a metaphor to describe the main line of my theorizing and published this in a book, Last Chapters: A Sociology of Aging and Dying (Marshall, 1980). In this metaphor, I saw aging individuals as if they were about to author the last few chapters of their lives. With the inevitability of death approaching, I saw them looking at their own biographies, and motivated to develop a stance that their lives made sense. (See, McAdams, this volume, for a general approach to narrative and the quest to make sense of one’s biography. My metaphor is based on similar psychological assumptions about motivation, but the elements of the metaphor are specific to the situation of heightening awareness of finitude.) This meant making sense of the earlier chapters of the autobiography, but also of the end of the biography in death itself. Moreover, I saw people in such circumstances as wanting to assume responsibility for their lives as a whole. These metaphorical elements stood for the theoretical dimensions of reconstructing and legitimating one’s past biography, developing a legitimation for the self as dying, and assuming control, or responsibility for the life as lived and as ending.

4.2. Is there a Shared Conception of what is a Good Life Course? I turn now to the first question that I shall address. This conceptual question asks whether there exist in a given society values for human development


73

and the life course, that is, a shared view of what is a good life course. I will examine this with reference to the concept of the ‘good death’. At one level, this is a question of culture; at another, it is a question of the individual. I drew on the anthropologist, David Counts (1976) for a cultural conception of ‘the good death’. The Kaliai of Western New Britain hold a cultural conception of the ‘good death’ that is quite similar to the process of aging and dying that is held to be ideal in the disengagement theory of aging, a North American model (Cumming & Henry, 1961). As in many cultures, the Kaliai see death not as the end of all experience, but as a transition between different life states – a transition that takes time and is aided by funerary rituals. Later life is seen as a progressive severing of ties and putting one’s house in order. When the Kaliai sense that death is coming they try to ward it off, to give time for bringing social relationships to a close. In the idealized good death, the dying person ‘‘called all his kinsmen to gather around him, disposed of his possessions after repaying the obligations owed by him and forgiving any obligations of others to him, and then informed those gathered that it was time for him to die’’ (Counts, 1976–1977, p. 370).15 There are no mortuary rites for a good death – all is appropriate and nothing needs to be ‘made right’ through ritual intervention. In Kaliai, however, there are virtually no ‘good death’s, which means that almost everyone has a funeral. The good death in Kaliai, as the good death in terms of the disengagement theory of aging, is one in which the individual times withdrawal from society to coincide with actual biological decline and cessation. The individual ceases to be a social being at the same time that he or she ceases to be a living biological being.16 The ‘good death’ has been little studied in western societies but in the terminology of the life course perspective, a good death would mean a correspondence between objective and subjective aspects of the career (Marshall, 1978–1979) in which, as with other transitions, rituals may be required in order to bring about this concordance in the biographical experience of individuals and in the social definition of the transition by members of that individual’s social group. It is not just that cultural ideals or schemas of the good death influence people’s action; their social and collective actions develop such cultural ideals or schemas. In the retirement village (or congregate living complex for older people)17 that I studied, the residents, or at least a leadership group among them, recognized that they were all in the final trajectory of their lives, one that leads to death. I drew on Berger and Luckmann’s concept of the legitimating function of symbolic universes for this analysis. As they noted in their classic book, The Social Construction of Reality (Berger & Luckmann, 1966, p. 101), ‘‘A strategic legitimating function of symbolic

74

VICTOR W. MARSHALL

universes for individual biography is the ‘location’ of death. The experience of the death of others and, subsequently, the anticipation of one’s own death posit the marginal situation par excellence for the individual’’. In this retirement village (pseudonym Glen Brae), I gathered persuasive evidence that the residents were highly accepting of the frequency of death in this community of old people, and also of their own impending death (Marshall, 1975a, b). However, in a nursing home (pseudonym, St. Joseph’s Home) that I studied at the same time, there was much less evidence of the acceptance of death (Marshall, 1975b). In the retirement village, the administration had not made plans for the management of dying and death as a community event, but by the time the community had been in existence for only one year, the residents had begun to organize as a community of the dying. An editorial in the newsletter published by the ‘residents’ forum’ is worth quoting at length. It noted that life was taking on a new meaning at Glen Brae, And new responsibilities are ours too. Fifteen deaths have occurred to date, which was the predicted actuarial rate. The rate will increase as we grow older. With 100 new residents arriving next year, it is forecast that we can expect a death amongst us as frequently as one every two weeks. This is a sober thought. Our responsibility, therefore, involves a point of view, a determination. Either Glen Brae will turn into a place shrouded in a funeral parlor atmosphere of tears and perpetual sadness, or it will play its intended role – the best place to be when crises occur. It is suggested that each of us look toward the future and be prepared, that we respect the faith of others, the wishes of the survivor, and above all else that we reduce to a minimum the prolongation of sorrow, the discussion of pain, loss, tragedy. It is up to us, not management, to make Glen Brae the haven we desire.

Deaths at Glen Brae are acknowledged by a brief bulletin-board notice and a name-only listing in the newspaper. Funerals are held off-site. There is an infirmary at the facility and many resident deaths occur there, thus reducing the visibility of death. At congregate mealtimes, the hostess sometimes directed sympathetic residents to the tables of recently bereaved residents. In a number of ways, then, the residents developed a culture internal to their community, which gave them a set of guidelines and procedures to handle the frequent occurrence of death among them. In contrast, at St. Joseph’s Home, where death was an equally frequent visitor, I found no resident-developed cultural beliefs or organizational procedures that suggested they had themselves taken control of this final stage of their life course. This was consistent with a general organizational contrast between it and the retirement village (Marshall, 1975b). At Glen Brae there was little administrative intervention into the lives of residents;


75

St. Joseph’s much more strongly represented a total institution in Goffman’s (1961) terminology, where the whole round of life was organized by the administration in ways that limited spontaneity and control by the residents over their lives. Time was more regulated. For example, residents lined up outside the dining hall and entered when a bell was rung. At St. Joseph’s the term ‘dying’ seems to be reserved for the very last stages. As one nursing aide noted, ‘‘When they start dying, they don’t last for but a day. Some of them take longer’’. However, the trajectory toward death begins earlier and is marked by physical movement from a dormitory or private room on an upper-level floor to an infirmary on the main level. Then, if or when the prognosis of ‘dying’ is made, the resident is moved to a special room, which is referred to by staff and residents as ‘the dying room’, but also known to both as ‘St. Peter’s Room’. In a Catholic institution this clearly signifies the last stage prior to death and, ironically, this is also Room 13. Residents are acutely aware of this trajectory. One resident moved to St. Peter’s room with a ‘dying’ prognosis returned on his own accord to his old room in the middle of the night. Another resident, not ‘dying’ but placed in the room when it was vacant because he had been ‘acting out’ and disturbing roommates, also abruptly left the room when he realized where he was. Almost all of the 15–20 deaths per year were commemorated by funerals at St. Joseph’s chapel, with an open coffin and at which the priest deliberately delivered the funeral oration in a booming voice that could be heard throughout the home. The death of residents is also more visible in the nursing home than the retirement community because residents eat every meal at fixed dining hall places, marked by a name card. In my observation, these name cards, and the empty place, remained for some time following a resident’s death. It might be said that people move to the retirement community, but have organized a collective way to live with the fact that they are a community of the dying as well as the living; but that the nursing home is a place where people go to die, where an important aspect of daily life centers around the fact that people die there, and where death receives considerable ritual treatment. In some senses, judging from a staff-resident interaction patterns and the low level of visitation of residents by family members or friends, many of the nursing home residents were ‘socially dead’, to use Kalish’s (1966) terminology: Social death occurs when an individual is thought of as dead and treated as dead, although he remains medically and legally alive. Any given person may be socially dead to one individual, to many individuals, or to virtually everyone, and perhaps to himself as well.

76

VICTOR W. MARSHALL

Referring back to my earlier treatment of agency as delegation – authorizing parties to act on behalf of something or someone, including themselves – agency is not delegated to residents of St. Joseph’s home. Decisions are made for them more than by them. Kalish elsewhere notes that ‘‘The selfperceived socially dead individual has accepted the notion that he is ‘as good as dead’, or that he is, for all practical purposes, dead’’: (Kalish, 1966) Such characterizations reflect the social construction of the life course, which creates life course trajectory categories through which some segments of the aging population will pass. The contrast between these two social settings shows the differentiation of the life course and the fact that its organizational properties and biographical experience by individuals can vary greatly. In summary on this point, the research question was, Is there a shared conception of what is a good life course? I approached this very narrowly in terms of the concept of the good death. There is cultural and subcultural variability in concepts of how people should end their life course in death. This can be seen cross-culturally and within the subcultures of two residential facilities for the aged in the same geographical area, but which differed by social class, religion, and many other characteristics. But the comparison also says something about agency and social structure. The social organization and administration of the two communities reflected extreme positions with respect to structural barriers to social action. The two community populations differed greatly in both human and social capital (e.g., in education, socioeconomic status, and viable family and friendship networks), and these differences were reflected in the administrative stances of the administrations of the two communities. At Glen Brae, social action by residents was encouraged by administrative principles and practices; at St. Joseph’s home, it was not. It would have been much more difficult to organize around the social construction of the last trajectory of the life course in the latter milieu than in the former. Formulated much more abstractly and extending from this example to other aspects of the life course, my answer to the research question is three-fold. First, let me return to the general conceptual discussion in the first half of this paper. The issue is framed as one of culture, which I take to mean the beliefs and values shared by the members of a collectivity. Cultures are schemas developed by social groups and communities. Together with resources, they constitute social realities, which have their own existence over and above those of the individuals who came together to construct them, and this (rules and resources) is social structure, with its potential to either liberate or constrain social action.


77

Second, there is cultural variability in conceptions of the good life course; this variability may lead to conflicts between constituencies – societies, communities, interest groups – with different idealized views of the life course. Third, social structures, such as the organizational features of the two settings I studied, can have a broader impact in either constraining or facilitating social action. In a longer time frame, some social structures recognize the agency of societal or group members, and this can produce new social institutions and structures (shared meanings, schema, and values plus patterns of social behavior and accompanying resources) that meet member needs. Other social structures place agency in the hands of a minority such as the administrators at St. Joseph’s home, and fail to reward social action by members. In terms of this research question, it is not simply a case of the existence or not of conceptions of the good life course, but rather it is also a case of the existence of structural opportunities or constraints on both the social construction of such life courses and the realization of these life courses in individual biographies.

4.3. Operationalizing the Life Course The second, methodological, question asks how the different concepts (such as trajectory, stage, transition, and event) are usually operationalized, and with what consequences. I will examine this with respect to the comparison of ‘objective’ with ‘subjective’ data about the dying trajectory. The previous discussion has shown how the objective dying trajectory was differentially constructed in different societies and in the two settings in the United States that I had studied. Here I want to first introduce the notion of differential objective and subjective life courses among those near to death, and then to describe a situation in which objective and subjective realities conflict, with intriguing social consequences. Just as the changing nature of work and retirement has dramatically altered the objective nature of the working life course, as well as its biographical or subjective experience through the adult years and the transition to retirement (Marshall, Heinz, Kru¨ger, & Verma, 2001), changes in life expectancy and life expectancy in good health have dramatically altered health transitions and the transition from life to death – both objectively and subjectively. Objectively, people are living longer and could expect to do so. I thought it would be interesting to ask people how long they expected to

78

VICTOR W. MARSHALL

live. The findings point to the subjectivity of such responses and to the importance of social psychological comparison processes. I asked respondents at Glen Brae how old they would live to be in three different ways, with varying success. Of 79 respondents, only 31 gave a definitive answer to a direct question, ‘‘How old do you think you will live to be’’; 60 placed a mark on a line to indicate where they saw themselves between birth and death; and 50 answered a four-alternative question, ‘‘Which one of these would you say about your own future?’’, ranging from ‘‘I shall be around for some time yet; more than ten years’’ to ‘‘The end may be any time now’’. All indicators correlated quite strongly with age, but there was a lot more going on in their estimates of what I called ‘‘awareness of finitude’’ (Marshall, 1975c). Focusing on the third, fixed alternative measure, responses to it could be better predicted in relation to a variable describing whether the respondent was currently younger than the age at death of both parents, one parent or none. That relationship was also much stronger for men than women. Another strong predictor was the number of dead brothers or sisters. Qualitative data also suggested that awareness of finitude is higher (i.e., anticipated life expectancy is shorter) with the death of friends. Awareness of finitude is thus a partial function of age, but it is also influenced by a number of social comparison processes, and the conviction (which is of course supported by data) that ‘longevity runs in families’. This said, it is also the case that many Glen Brae respondents found themselves in the position of having lived beyond the age they had anticipated. Here the objective life course and the subjective life course were at odds. In the particular living circumstances of this retirement community, this created problems for many residents – problems made more complex because the individual residents were not the only ones with a stake in the accuracy of these estimates. Glen Brae was one of the early examples of a retirement community organized on the ‘life care principle’. On entry, a resident would pay a founder’s fee that guaranteed his or her right to remain in the retirement community for life, paying rents that would be set by the corporation. Since the founder’s fee was quite significant, a rational person would not move to this community unless he or she expected to live long enough to make this a good investment. In addition, the resident had to be satisfied that, after making the substantial founder’s fee payment on entry, he or she had enough assets to generate a revenue base to pay the monthly rental fees as long as they lived. On the other hand, the administration of Glen Brae used the founder’s fees to amortize the large mortgage on the new, costly physical plant, while counting on monthly rental fees to meet


79

operating expenses. Once all units in Glen Brae were occupied, the only way for the corporation to generate funds to pay the mortgage was through new founder’s fees, but these could be generated only when a resident died and the apartment became vacant. Thus, in an abstract sense, two parties – resident and administration – were making bets based on life-expectancy estimates. I examined this situation from the perspective of game theory, and found that both parties had made significant errors in the life estimates. As noted earlier, residents made subjective estimates of their life expectancy based on their age and other social comparisons. They mostly neglected to take into account general cohort increases in life expectancy, and cohort differences in socioeconomic status, which correlate with health and life expectancy. In other words, they were not good life course theorists in that they ignored cohort factors. (I did not characterize this in life course terms at the time, but it is a good example of one of the most basic of life course principles). The corporation, on the other hand, used age alone. Rather, it relied on the insurance company that held the mortgage to estimate turnover of the apartment units, and that company did so based on life-expectancy tables for its insurance policy holders. However, it did not take into account two things: first that the typical resident of Glen Brae is of a higher social class, and therefore likely to live longer than the typical life insurance policyholder; and second, that the person moving to the retirement community may be thought of as acting in a different strategic or decision-making mode than the typical insurance policyholder. The former may be seen to be making a very strategic bet to live long enough to make this a good investment, and people who think they will not live very long may very well opt not to pay that expensive founder’s fee. To cast this situation as a game is to go well beyond the data, reconstructing actions of the administration and residents in terms of imputed rationality. However, the exercise calls attention to a few points related to the research question I have been discussing. That question, which is put to us as a methodological question, asks, how are different concepts in the life course operationalized in different disciplines, and with what consequences? As with the first question (and the next), I am addressing this very narrowly with respect to two concepts – trajectory, and that troublesome conceptual relationship between agency and social structure. The methodology here is principally in my crude attempt to measure and to understand the anticipated final trajectory or, more precisely, its length or duration; i.e., awareness of finitude. In other aspects of my research on aging and dying, I used awareness of finitude as an independent variable, to

80

VICTOR W. MARSHALL

assess as best I could its causal relationship to other phenomena, such as concerns about death and dying, time perspective, and the extent to which the individual is preoccupied with self and identity issues. These investigations, which I cannot go into here, and those just reported, which try to explain how people come to make estimates of anticipated life expectancy, all help us to understand the concept of awareness of finitude (through construct validity). In terms of the life course perspective per se, there has been less work done on the anticipated life course than on the experienced life course (see Markus & Nurius, 1986; Ryff, 1991; Mueller, 2002). The general points I would make with respect to this methodological question are: First, the distinction between objective and subjective aspects of trajectories should be maintained and investigated, and discrepancies between the two are likely to be theoretically interesting – as work on Neugarten’s notion of ‘social clocks’ attests (Neugarten & Hagestad, 1976; Ryff, 1991). Anticipated and realized careers or trajectories are ideally investigated with longitudinal data (Mueller, 2002), and there are great concerns that retrospective data will be biased through self-serving reconstruction. However, reconstructions of the past as well as changing constructions of the future are core materials in studies of the life course. Second, and more generally, there can be multiple objective and subjective life courses ‘in play’ at the same time, and objective life courses can include largely biological aspects of the life course including physiological changes such as puberty, menopause, physiological declines, and finally death.18 Third, the investigations I have described, and particularly the use of game theory, raises questions about limits to Clausen’s (1991) concept of planful competence, discussed earlier under agency. As noted earlier, ‘planful competence’ is the ability to make informed, rational decisions and set realistic short- and long-term goals. I suggested earlier that this can be seen as agency in this first sense of the metaphor – the existence of the furnace, and it can be viewed as a developmental capacity of (virtually) all humans. However, using game theoretical analysis is a methodological trick to show the difference between capacity and outcome or, in my preferred terminology, between agency and action. Life courses are far from ‘rational’ in any pure sense, and related to this departure from rationality is the fact that people live their life courses socially, negotiating their way through life with other actors also exercising planful competence, but with often-conflicting definitions of the situation, goals, and resources.


81

4.4. Status Changes and Changes in Identity The third, results question asks to what extent changes in status result in the redefinition of identity. I will address this question by referring to the distinction between self and identity, and to the action of the aging and dying individual, in managing identity. The question begs for a definition of identity, which is a term used in many different ways in the social sciences.19 In my work, identity is that sense of sameness and continuity in the organization of one’s selves over time. Following Goffman (1963) I also distinguish between ego of felt identity, imputed or social identity, and personal or presented identity. The former refers to the individual’s own sense of selves over time, the second to the views of others, and the third to the presented self (which as Goffman so well shows, does not always correspond closely to the former and is a way in which people seek to influence others to change the identity they impute to them). At the social or collective level, I described above how the residents at Glen Brae sought explicitly to shape their social identity. The residents had, in fact, changed their social status when they moved to the retirement community, which, as a ‘life-care community’ they anticipated would be their collective home until each died. But they recognized options in how they were to view themselves and collectively committed to an option that focused on themselves as living, rather than as dying. There are good social psychological concepts to deal with this phenomenon, of which the concept of role distance is as insightful as it is difficult to measure (Goffman, 1961). Role distance is setting a distance between one’s self and a specific role. It is a way of remaining more than the specific role one is playing at the time. At Glen Brae, residents collectively took the position, ‘I know I am dying but my whole being is not wrapped up in that. I am many other things than a dying person’. At a more individual level, I was interested in my early research in how the recognition of the self as dying leads to changes in ego or felt identity. This is the metaphor of the individual in the last chapters of life (Marshall, 1980). Long ago, the hermeneutic scholar Wilhelm Dilthey described the selectivity of autobiographical reflection as follows: The person who seeks the connective threads in the history of his life has already, from different points of view, created a coherence in that lifey. He has created it by experiencing values and realizing purposes in his life, making plans for it, seeing his past in terms of development and his future as the shaping of his lifey. He has in his memory, singled out and accentuated the moments which he experienced as significant; others he has allowed to sink into forgetfulness (Dilthey, 1962, p. 86).

82

VICTOR W. MARSHALL

To summarize my findings briefly, I found a relationship between awareness of finitude, and reminiscence. While scholars such as Erikson (1959) and Butler (1963) had focused on awareness of finitude as initiating focused or intense individual reminiscence and reconstruction of one’s identity, I found a strong relationship with social reminiscence as well. Asked, ‘‘Do you often talk about things that have happened in your past life with anybody else?, 36% of respondents estimating more than ten years to live, 32% of those estimating five to ten years, but fully 62% of those estimating less than five years said they do so ‘at least once a week’, rather than less frequently (Marshall, 1975c, p. 125). Higher social reminiscence was associated with a greater intensity of reminiscence (they reported more turning points when asked to describe their lives), with reporting turning points that were coded as reflecting internal rather than external locus of control (things they did rather than things that happened to them), and with a greater likelihood of reporting satisfaction with their life as a whole (summarized in Marshall, 1980, pp. 115–119). It is impossible to sort out causation from these cross-sectional data. Only a longitudinal study could establish a clear temporal relationship among changes in awareness of finitude, changes in the extent to which the individual engages in individual or social reminiscence, and changes in the content of the remembered biography as well as satisfaction with the biography as a whole. My cross-sectional data let me talk about differences but not changes. My interpretation that social reminiscence assists the individual in the reconstruction of biography leans more strongly on theory than on my data. Moreover, in attending to social in addition to individual reminiscence, it contrasts with more individualistic, psychological approaches to biographical narrative, such as those described in McAdam, this volume. Let me return to the third question, about results, is it possible to demonstrate that a change in status implies a redefinition of the identities of the persons involved? From my limited study of three decades ago, I infer the following answer to this question: First, life course scholars should devote theoretical and methodological attention to the complexity of identity. If a person has as many selves as there are groups about whose opinion he or she cares, then a concept of identity is needed to deal with how the individual prioritizes, situates, manages, and organizes these many selves so as to give some sense of sameness and continuity over time. We need first to distinguish self from identity. Second, we need to distinguish identity as experienced by the individual from identity as attributed to the individual by various groups of audiences.


83

Third, it is easier to describe differences in identity than to explain changes in identity, and we would all benefit from a sustained attempt to theorize the mechanisms of identity change. This task calls for multidisciplinary skills because these processes, my work suggests, take place at biological, psychological, and social levels, and the intersection or interaction of these levels.

5. AGENCY, EVENTS, AND STRUCTURE AT THE END OF THE LIFE COURSE I want to quickly conclude with some general remarks based on my visit to earlier research that I conducted more than three decades ago. I have tried not to reconstruct the language of this work, while at the same time showing its relevance to contemporary issues of the life course. Perhaps, the major conclusion of this exercise is that it reminds us of the long chains of scholarship that have contributed to what we now think of as the life course perspective; and perhaps in addition we can be reminded that the life course perspective of today is highly diverse. I have addressed both these issues elsewhere (Marshall & Mueller, 2003), focusing on the similarities and differences between North American and European approaches to the life course. In some senses I came to the ‘life course perspective’ late, but in other senses I came to it early. Many theoretical strands come together to form what is now the life course approach, and I had made use of some of these strands. Some of them, grounded in symbolic interactionism, gave us concepts like career and status passage that gained currency in parts of Europe, but were largely supplanted in North America by the terminology subsequently introduced by Riley, Elder, and their associates. Status passages and careers became transitions and trajectories in most American scholarship, and increasingly sophisticated statistical techniques were applied to analyses that paid more attention to complex descriptions of the life course than to how people actually made their way – as individuals and in collective action – through life courses that they were also helping to construct. The concept ‘agency’ came to be a catch-all phrase that most of the time simply paid respect to the human capacity for voluntaristic action, and told the reader that the author was not a complete determinist (despite using deterministic statistical methods). Agency often meant, ‘unexplained variance’ – if the

84

VICTOR W. MARSHALL

model did not explain much variance, then this must mean that people were making innovative choices or struggling against structural barriers. Revisiting my early, ‘pre-life course perspective’ scholarship, I did not feel uncomfortable with the symbolic interactional and phenomenological sources and language I had used. Instead, I think this approach to life course theorizing made it easier than the more orthodox language that predominates today, to address the negotiated character of life course construction and transitions. I hope that the Europeans will reinvigorate international life course theorizing by continuing to make good use of the life course–concepts they initially borrowed from Chicago school sociology. Revisiting my early work produced a stronger sense than I would have anticipated of the importance of bringing together objective and subjective aspects of the life course. When cultural ideas of the life course are combined with resource allocation mechanisms and resources, we have social structure, which is objective at the macro level. Individuals then create, experience, and endure their life courses in relation to these objective dimensions of social structure. As they do so they may change the social structure or they may reinforce it, by submitting to its constraints or capitalizing on the opportunities it provides (these two together constitute life chances). Their careers or life courses may turn out to be some jointly produced product of objective, structural constraints, and opportunities, the assets and resources they bring to the choices they make, and the micro-level interactions they have with others who, like them, are trying to create, experience, and endure their own life courses. A given individual’s vision of his or her life course may be realistic or not in terms of objective or structural life course conditions. It may coincide, conflict, or synergize with that of other members of his or her social group or collectivity and this might in turn influence the nature of the attained life course. In any case, people will experience their own objective life courses, the dimensions of which can be measured and analyzed in life course research. But the full nature of their life experiences includes a subjective life course as well, one with anticipations and plans, but also memories and reconstructions or reinterpretations of the lives they have led.

NOTES 1. Another beginning point might well be the work of the late Klaus Riegel (e.g., Riegel, 1975, 1976; see Lerner & Busch-Rossnagel, 1981b for a brief discussion of his contributions in this area).


85

2. The work of Piaget is grounded in, and empirically supports, such a notion, with the individual seen as actively seeking novelty so that, through accommodation, schemata will change so as to permit assimilation, enabling increasing mastery of one’s environment. 3. It may be instructive to examine creativity across the life course in light of agency and developmental theory. See Kastenbaum (1992). 4. On this point he is criticized by Dannefer (1999). 5. At this point Heinz is in fact drawing on Giddens (1984), but elsewhere in Giddens, e.g., 1991b) the notion of agency is much more broadly drawn. I discuss this below. 6. This conceptualization seems to include agency as an aspect of human nature. However, in the same article, Dannefer treats agency as a variable. Elaborating on a case described by Gubrium, Holstein, and Buckholdt (1994), of a child labeled as low in ability and placed in the lowest of three ability groups, Dannefer shows how labeling creates a self-fulfilling prophecy that keeps this child in his or her place. Here is how Dannefer interprets this situation: Even when the outcome is positiveywhat is being described here is a social system process in which the labeled individual participates in a subordinate manner. Thus, the agentic force of the individuals in question, including the positively labeled individual is, in principle, no more in evidence in the case of positive than negative labeling. This raises the question of whether and how genuine human agency might be augmented – a question that, in any domain, is ultimately political. (Dannefer, 1999, p. 78).

7. Giddens (1991a, p. 112) contrasts ‘fateful moments’ with daily routines of life. ‘‘Fateful moments are those when individuals are called on to take decisions that are particularly consequential for their ambitions or more generally for their future lives. Fateful moments are highly consequential for a person’s destiny’’. 8. I limit the statement as this is not the place to consider whether persons born with limited or no cognitive capacity are fully human. 9. This task is related to, but quite distinct from, the conceptual and methodological challenges of linking ‘micro’ and ‘macro’ levels of analysis; where challenges are addressed absent the notion of individual agency. See, for example, Collins (1992); Heinz (1996); Marshall (1995) and O’Rand and Campbell (1999). 10. Alwin (1995, p. 217) observes of a panel that tried to define it: ‘‘There was little theoretical consensus on what structure was and a confusion of meanings, but virtually everyone agreed that the conceptual apparatus conveyed by the concept of social structure was what made that sociological contribution to the study of individual lives a possibility’’. The same cannot be said of agency or even of action, as there are sociologists, such as some structuralists, who argue such concepts are not needed. 11. Going beyond Giddens’ social structure as virtual ‘rules’ and ‘resources’, Sewell (1992, p. 8) ‘‘would in fact argue that publicly fixed codifications of rules are actual rather than virtual and should be regarded as resources rather than as rules in Giddens’ sense’’. 12. While she postulates this view as against Giddens, Berger and Luckmann, there are many similarities to both. As for possible differences, Archer sees a constant possibility for emergence, from unanticipated consequences of social action,

86

VICTOR W. MARSHALL

but also from the innovative nature of individuals. The similarity to Schutz is apparent in the following: ‘‘The activity-dependence of structures is in no way compromised by the argument that a given structure was issued in by a particular generation/cohort of actors as an unintended yet emergent consequence of their activities, whilst it then necessarily pre-existed their successors. This is the human condition, to be born into a social context (of language, beliefs, and organization) which was not of our making: agential power is always restricted to re-making, whether this be reproducing or transforming our social inheritance.’’ (Archer, 1995, p. 72). Because Giddens’ structuration theory denies the existence of emergence, and also (p. 87) because Giddens is interested in a short time frame, Archer argues that structuration theory has ‘‘an inability to examine the interplay between structure and agency over longer temporal tracts because the two presuppose one another so closely’’. Archer’s desire to make time an actual component in the theory may be compared to the importance of time in most life course perspective approaches, such as Elder. 13. I cannot do justice to Sewell’s full argument here, including his use of Bourdieu, but I think he has given the most satisfactory conceptualization to the agency-structure discussion in current sociological theorizing. In my own thinking I would draw further on Berger and Luckmann and Archer (whom Sewell does not resource). 14. Data were collected through participant observation, interviewing, and archival methods as my doctoral dissertation research. I was in the field in these two settings over the period 1969–1970. The bulk of my data were based on fieldwork and semi-structured interviews in the retirement community, where I completed at least one lengthy focused interview with 105 of just under 400 residents. Of these, I interviewed 68 respondents three times, and 92 respondents twice. In the nursing home, I conducted fieldwork over a 3-month period, mostly using participant observation, but also unsystematic focused interviews. 15. The Limbu of Nepal (Jones, 1974) and the LoDagaa of West Africa (Goody, 1962, pp. 208–209) have a similar conception of the good death and Meyerhoff (1978) describes the dying of an elderly Jewish American in a similar light (Marshall, 1980, pp. 33–34, 151–152). 16. Hertz (1960, p. 76) notes that in many parts of the world, deaths of children, strangers and slaves ‘‘arouse no emotion, occasion no ritual’’, and with respect to the aged, he says, ‘‘In various Australian tribes, old people who, because of their great age, are incapable of taking part in the totemic ceremonies, who have lost their aptitude for sacred functions, are buried immediately after deathy. This is so because, due to the weakening of their faculties, they have ceased to participate in social life; their death merely consecrates an exclusion from society which has in fact already been completed, and which every one has had time to get used to’’ (Hertz, 1960, p. 85). 17. Current terminology in the USA for such a facility would be ‘‘Continuing Care Retirement Community’’. 18. I am not able to address the importance of catastrophe and diversity in life courses. Many things simply cannot be predicted and these range from major events such as wars, terrorist attacks and natural disasters to individual-level events such as motor vehicle accidents and heart attacks. Linda George (2003) in a recent essay on life course perspectives in the area of aging and health, notes that health traumas often lead to divergence in life course pathways. She also reminds us of the tremendous


87

divergence of pathways in the absence of trauma, as noted, for example, in the classic study by Rindfuss, Swicegood, and Rosenfeld (1987). 19. Often the terms self and identity are used interchangeably. I argue that it is important to distinguish the two. Self, drawing on symbolic interactionism, is the process of reflection and cognitive orientation in which the I responds to the self – as object (the me) based on both internal cues and external feedback from others.Following James, I have as many selves as there are groups of people about whom I care. These many selves develop in relations with others in role relationships and individual selves become situated in social relationships with (types of) others. However, some selves are more important than others, and moreover the social scientist needs to be able to account for the individual’s ability to select and present selves.

REFERENCES Ackerman, S., Zuroff, D. C., & Moskowitz, D. S. (2000). Generativity in midlife and young adults: Links to agency, communion, and subjective well-being. International Journal of Aging and Human Development, 50(1), 17–41. Alwin, D. F. (1995). Taking time seriously: Studying social change, social structure, and human lives. In: P. Moen, G. H. Elder Jr. & K. Lu¨scher (Eds), Examining lives in context. Washington, DC: American Psychological Association. Archer, M. S. (1995). Realist social theory: The morphogenic approach. Cambridge: Cambridge University Press. Bakan, D. (1966). The duality of human existence. Chicago: Rand McNally & Co. Berger, P. L., & Luckmann, T. (1966). The social construction of reality. New York: Doubleday. Brim, O. G. (1981). Foreword. Introduction to R. M. Lerner & N. A. Busch-Nagel (Eds), Individuals as producers of their development: A life-span perspective (pp. xv–xvii). New York: Academic Press. Butler, R. N. (1963). The life review: An interpretation of reminiscence in the aged. Psychiatry: Journal for the Study of Interpersonal Processes, 26, 65–76. Cain, L. D., Jr. (1964). Life course and social structure. In: R. E. L. Faris (Ed.), Handbook of modern sociology (pp. 273–309). Chicago: Rand McNally. Campbell, C. (1996). The myth of social action. Cambridge: Cambridge University Press. Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American Journal of Sociology, 96, 805–852. Clausen, J. S. (1993). American lives: Looking back at children of the great depression. New York: Free Press. Collins, R. (1992). The romanticism of agency/structure versus the analysis of micro/macro. Current Sociology, 40, 77–97. Counts, D. (1976–1977). The good death in Kaliai: Preparation for death in Western New Britain. Omega, 7(4), 367–372. Cumming, E., & Henry, W. E. (1961). Growing old: The process of disengagement. New York: Basic Books. Dahrendorf, R. (1979). Life chances. Chicago: The University of Chicago Press.

88

VICTOR W. MARSHALL

Dannefer, D. (1999). Neoteny, naturalization, and other constituents of human development. In: C. D. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 67–93). New York: Springer (Chapter 3). Dannefer, D., & Uhlenberg, P. (1999). Paths of the life course: A typology. In: V. L. Bengtson & K. W. Schaie (Eds), Handbook of theories of aging (pp. 306–326). New York: Springer (Chapter 17). Diehl, M. (1999). Self-development in adulthood and aging: The role of critical life events. In: C. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 150–183). New York: Springer. Dilthey, W. (1962). In: H. P. Rickman (Ed.), Pattern and meaning in history. New York: Harper Torchbooks. Elder, G. H., Jr. (1975). Age differentiation and the life course. Annual Review of Sociology, 1, 165–190. Elder, G. H., Jr. (1985). Perspectives on the life course. In: G. H. Elder (Ed.), Life course dynamics: Trajectories and transitions, 1968–1980 (pp. 23–49). Ithaca: Cornell University Press (Chapter 1). Elder, G. H., Jr. (1997). The life course and human development. In: R. M. Lerner (Ed.), Handbook of child psychology: Theoretical models of human development (Vol.1, pp. 939–991). New York: Wiley (Chapter 1). Elder, G. H., Jr., & Johnson, K. M. (2003). The life course and aging: Challenges, lessons, and new directions. In: R. A. Settersten Jr. (Ed.), Invitation to the life course: Towards new understandings of later life (pp. 49–81). Amityville, NY: Baywood Publishing Company. Elder, G. H., Jr., & O’Rand, A. (1995). Adult lives in a changing society. In: K. S. Cook, G. A. Fine & J. S. House (Eds), Sociological perspectives on social psychology (pp. 452–475). Boston: Allyn & Bacon (Chapter 17). Erikson, E. (1959). Identity and the life cycle. Psychological Issues, 1(1). George, L. K. (1999). Social perspectives on the self in later life. In: C. D. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 42–66). New York: Springer (Chapter 2). George, L. K. (2003). What life-course perspectives offer the study of aging and health. In: R. A. Settersten Jr. (Ed.), Invitation to the life-course: Toward new understandings of later life (pp. 161–188). Amityville NY: Baywood. Giddens, A. (1983). Profiles and critiques in social theory. Berkeley, CA: University of California Press. Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Berkeley and Los Angeles: University of California Press. Giddens, A. (1991a). Modernity and self-identity: Self and society in the late modern age. Stanford: Stanford University Press. Giddens, A. (1991b). New rules of sociological method (2nd ed.). Stanford, CA: Stanford University Press. Goffman, I. (1961). Role distance. In: I. Goffman (Ed.), Encounters (pp. 85–152). Indianapolis: Bobbs-Merrill. Goffman, I. (1963). Stigma. Harmondsworth: Penguin. Goody, J. (1962). Death, property and the ancestors. Palo Alto, CA: Stanford University Press. Gubrium, J. F., Holstein, J. A., & Buckholdt, D. R. (1994). Constructing the life course. Dix Hills, NY: General Hall.


89

Gutmann, D. (1965). Women and the conception of ego strength. Merrill-Palmer Quarterly, 11, 229–240. Heinz, W. R. (1996). Status passages as micro–macro linkages in life course research. In: A. Weymann & W. R. Heinz (Eds), Society and biography: Interrelationships between social structure, institutions and the life course (pp. 51–65). Weinheim: Deutscher Studien Verlag. Hertz, R. (1960). In: R. C. Needham (Trans.), Death and the right hand. Aberdeen: Cohen and West. Hughes, E. C. (1971). The sociological eye: Selected papers. Chicago: Aldine, Atherton. Jones, R. L. (1974). Religious symbolism in Limbu death-by-violence. Omega, 5(3), 257–266. Kahana, E., & Kahana, B. (1996). Conceptual and empirical advances in understanding aging well through proactive adaptation. In: V. L. Bengtson (Ed.), Adulthood and aging: Research on continuities and discontinuities (pp. 18–40). New York: Springer (Chapter 2). Kalish, R. A. (1966). A continuum of subjectively perceived death. The Gerontologist, 6(2), 73–76. Kastenbaum, R. (1992). The creative process: A life-span approach. In: T. R. Cole, D. D. Van Tassel & R. Kastenbaum (Eds), Handbook of the humanities and aging (pp. 285–306). New York: Springer. Kiser, E. (1999). Comparing varieties of agency theory in economics, political science, and sociology: An illustration from state policy implementation. Sociological Theory, 17(2), 146–170. Lawton, M. P. (1989). Environmental proactivity and affect in older people. In: S. Spacapan & S. Askamp (Eds), The social psychology of aging (pp. 135–163). Newbury Park: Sage. Layder, D. (1994). Understanding social theory. London: Sage. Lerner, R. M., & Busch-Rossnagel, N. A. (Eds) (1981a). Individuals as producers of their development: A life-span perspective. New York: Academic Press. Lerner, R. M., & Busch-Rossnagel, N. A. (1981b). Individuals as producers of their own development: Conceptual and empirical bases. In: R. M. Lerner & N. A. Busch-Nagel (Eds), Individuals as producers of their development: A life-span perspective (pp. 1–36). New York: Academic Press. Markus, H. R., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969. Marshall, V. W. (1972). Continued living and dying as problematical aspects of old age. Unpublished doctoral dissertation in sociology, Princeton University. Marshall, V. W. (1975a). Socialization for impending death in a retirement village. American Journal of Sociology, 80, 1124–1144. Marshall, V. W. (1975b). Organizational features of terminal status passage in residential facilities for the aged. Urban Life, 4(3), 349–368. Marshall, V. W. (1975c). Age and awareness of finitude in developmental gerontology. Omega, 6(2), 113–129. Marshall, V. W. (1978–1979). No exit: A symbolic interactionist perspective on aging. International Journal of Aging and Human Development, 9(4), 345–358. Marshall, V. W. (1980). Last chapters: A sociology of aging and dying. Monterey, CA: Brooks/ Cole. Marshall, V. W. (1995). The micro–macro link in the sociology of aging. In: C. Hummel & C. L. D’Epinay (Eds), Images of aging in western societies (pp. 337–371). Geneva: Centre for Interdisciplinary Gerontology, University of Geneva.

90

VICTOR W. MARSHALL

Marshall, V. W. (1996). The state of theory in aging and the social sciences. In: R. H. Binstock, & L. George (Eds), Handbook of aging and the social sciences (4th ed., pp. 12–30). San Diego, CA: Academic Press. Marshall, V. W., Heinz, W. R., Kru¨ger, H., & Verma, A. (Eds) (2001). Restructuring work and the life course. Toronto: University of Toronto Press. Marshall, V. W., & Mueller, M. M. (2003). Theoretical roots of the life-course perspective. In: W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course: Transitions, institutions, and interrelations (pp. 3–32). New York: Aldine de Gruyter. McMullin, J. A., & Marshall, V. W. (1999). Structure and agency in the retirement process: A case study of Montreal garment workers. In: C. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 305–338). New York: Springer. Meyer, J. W., & Jepperson, R. L. (2000). The ‘‘actors’’ of modern society: The cultural construction of social agency. Sociological Theory, 18(1), 100–120. Meyerhoff, B. G. (1978). A symbol perfected in death: Continuity and ritual in the life and death of an elderly Jew. In: B. G. Myerhoff & A. Simic (Eds), Life’s career–aging: Cultural variations on growing old (pp. 163–205). Beverly Hills and London: Sage. Mueller, M. M. (2002). Work, family and well-being over the life course: Continuities and discontinuities in the lives of American women. Doctoral dissertation in sociology, University of North Carolina at Chapel Hill. Neugarten, B. L., & Hagestad, G. O. (1976). Age and the life course. In: R. Binstock, E. Shanas, with the assistance of associate editors, V. L. Bengtson, G. L. Maddox, & D. Wedderburn (Eds), Handbook of aging and the social sciences (pp. 35–55). New York: Van Nostrand Reinhold. O’Rand, A., & Campbell, R. T. (1999). On reestablishing the phenomenon and specifying ignorance: Theory development and research design in aging. In: V. L. Bengtson & K. W. Schaie (Eds), Handboook of theories of aging (pp. 59–78). New York: Springer. Riegel, K. R. (1975). Toward a dialectical theory of development. Human Development, 18, 50–64. Riegel, K. R. (1976). The dialectics of human development. American Psychologist, 31, 689–700. Riley, M. W. (1979). Introduction: life-course perspectives. In: M. W. Riley (Ed.), Aging from birth to death. Interdisciplinary Perspectives (pp. 3–13). Boulder, CO: Westview Press for American Association for the Advancement of Science. Rindfuss, R. R., Swicegood, C., & Rosenfeld, R. A. (1987). Disorder in the life course: How common and does it matter? American Sociological Review, 52, 785–801. Rossi, A. S. (1980). Life-span theories and women’s lives. Signs, 6(1), 4–32. Ryff, C. D. (1991). Possible selves in adulthood and old age: A tale of shifting horizons. Psychology and Aging, 6, 286–295. Ryff, C. D., & Marshall, V. W. (Eds) (1999). The self and society in aging processes. New York: Springer. Ryff, C. D., Marshall, V. W., & Clarke, P. (1999). Linking the self and society in social gerontology: Crossing new territory via old questions. In: C. D. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 3–41). New York: Springer. Schutz, A. (1964). The problem of rationality in the social world. In: A. Brodersen (Ed.), A Schutz, collected papers II: Studies in social theory (pp. 64–88). The Hague: Martinus Nijhoff. Settersten, R. A., Jr. (1999). Lives in time and place: The problems and promises of developmental science. Amityville, NY: Baywood Publishing Company.


91

Settersten, R. A., Jr. (2003). Propositions and controversies in life-course scholarship. In: R. A. Settersten Jr. (Ed.), Invitation to the life course: Towards new understandings of later life (pp. 15–45). Amityville, NY: Baywood Publishing Company. Sewell, W. H., Jr. (1992). A theory of structure: Duality, agency, and transformation. American Journal of Sociology, 98(1), 1–29. Shanahan, M. J., Elder, G. H., Jr., & Miech, R. A. (1997). History and agency in men’s lives: Pathways to achievement in cohort perspective. Sociology of Education, 70, 54–67. Simmel, G. (1950). In: K. H. Wolff (Trans., Ed.), The sociology of Georg Simmel. New York: The Free Press of Glencoe. Smelser, N. J. (1997a). Problematics of sociology: The Georg Simmel lectures. Berkeley, CA: University of California Press. Smelser, N. J. (1997b). Social structure. In: N. J. Smelser (Ed.), Handbook of sociology. Newbury Park, London & New Delhi: Sage. Thomas, W. I., & Znaniecki, F. (1918). The Polish peasant in Europe and America, volumes 1–2. Boston: Badger. Weber, M. (1978). In: G. Roth & C. Wirth (Eds), Economy and society, Vol. I: An outline of interpretive sociology. Berkeley: University of California Press.


92

LOOKING AT AMBIVALENCES: THE CONTRIBUTION OF A ‘‘NEW-OLD’’ VIEW OF INTERGENERATIONAL RELATIONS TO THE STUDY OF THE LIFE COURSE Kurt Lu¨scher 1. INTRODUCTION This chapter has its origins in the kind invitation to present, at the PaVieColloquium, an idea that is receiving increasing attention in the study of intergenerational relations. Its essence can be summarized in the following hypothesis: Intergenerational relationships, especially among adult children and their parents, imply the experience of ambivalences and, consequently, require dealing with ambivalences.1 Thus, my point of departure does not seem to be a major issue of life course research. However, at second glance, one may recall that embeddedness in intergenerational relations is crucial for personal development. Most human beings are conceived in and born into familial contexts, and parent–child relationships – as diverse as they may be – are in many ways important for the unfolding of personal abilities Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 93–128 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10003-3

93

94

KURT LU¨SCHER

and the consciousness of the self. Youth is a formative phase in the life course where intergenerational relationships are of importance, simply because their dominance may be challenged by other relationships, such as those among siblings and peers. This is also true for early adulthood. Later, through partnerships and marriages, and thus the acquisition of in-laws, there is an increase in the number of elders with whom close and intimate relationships become possible or are even expected and required. In mid-life, nowadays, most adults belong to genealogical networks involving three or even four generations. Later, obligations for the care of the very old may emerge. The rules and the practice of inheritance once more accentuate the social and material importance of intergenerational relationships and their impact for the conduct of personal lives. In addition, the institution of inheritance reminds us that any life course may also be comprehended as a link in a chain of generations. Indeed, the study of the life course may profit from taking into account the interplay with the study of intergenerational relationships, and consequently from recent developments in this field. To this obvious statement, I would like to add two points. First, because of their omnipresence, intergenerational relations are at the core of the processes of socialization and of human sociability. This is why insights from the study of intergenerational relationships are of foremost interest for the analysis of social relationships in general, be it with regard to what they have in common with other relationships, or to where they differ from them, for instance from market relationships. My second remark is meta-theoretical. Because of the great relevance of intergenerational relations, their understanding is usually bound to moral judgments. Such normative views often penetrate scholarly descriptions. For instance, it is quite common to idealize intergenerational relations – positively – with reference to the concept of solidarity, or to deplore them – negatively – as a notorious source of conflict. As I will show, a well-grounded theory of ambivalence allows us to overcome these biases, because it simultaneously takes into account and analyzes both perspectives. In this way, a high degree of social authenticity can be achieved, and respective normative orientations can become a deliberate topic of analysis. Moreover, we recall that general assumptions about human nature underlie the concepts used in social science research, especially about such fundamental issues as the conduct of human lives and their social organization. However, at this point I cannot present a comprehensive account of the importance of ambivalence for the study of the life course. I must limit myself to outlining the meanings of this concept as such, and I will present the conceptual frame that I and other researchers have developed. Taking

Looking at Ambivalences

95

this as a point of reference, I will also illustrate the usefulness of this approach by presenting some exemplary results of recent research. I shall concentrate on issues closely related to the study of the life course and of connected lives, and I will suggest further applications in this field.

2. AMBIVALENCE IN INTERGENERATIONAL RELATIONS: THE REDISCOVERY OF AN OLD EXPERIENCE The idea of drawing on the concept of ambivalence for the study of intergenerational relationships has two sources. First, an awareness of the usefulness of ambivalence as a theoretical concept arose from a critical evaluation of the existing literature on intergenerational relationships, which in the 1990s was aptly characterized as data-rich and theory-poor (Lu¨scher & Pillemer, 1998). In particular, we criticized the dominance of the so-called solidarity perspective, because it presents a picture of intergenerational relationships that pays too much attention to positive aspects and too little to the innately darker ones. The solidarity perspective arose in reaction to Talcott Parsons’s (1942, 1949) portrayal of the nuclear family as isolated. It holds that, to the contrary, extensive family solidarity does exist. (Shanas et al., 1968; Littwak, 1965; Sussman, 1959). Since the early 1970s, Bengtson and co-workers have continued to develop this approach in an influential series of articles and books (cf. Roberts, Richards, & Bengtson, 1991; Bengtson & Harootyan, 1994; Bengtson, Giarusso, Mabry, & Silverstein, 2002). The solidarity perspective has also been adopted by other researchers in the United States (Rein, 1994; Rossi & Rossi, 1990) and serves as a reference point for many European authors, although not without critical reservations (AttiasDonfut, 1995; Bawin-Legros, Gauthier, & Strassen, 1995; Donati, 1995; Finch & Mason, 1993; Szydlik, 2000). However, at the same time as scholars in the solidarity tradition have emphasized mutual support and value consensus, another line of research has focused on isolation, caregiver stress, family problems, conflict and abuse (Marshall, Matthews, & Rosenthal, 1993). The image of weakened family ties and the abandonment of the elderly continues to be widely held in popular opinion and in portrayals of the family in contemporary fiction and theater. Thus, some scholars, as well as the public at large, appear reluctant to accept that intergenerational relationships include solidarity and are characterized by shared values and reciprocal help. As Marshall et al. (1993, p. 47) have succinctly put it, ‘‘the

96

KURT LU¨SCHER

substantive preoccupations in gerontology over the past 30 years point to a love–hate relationship with the family.’’ In a somewhat different mode, Lalive d0 Epinay and Bickel (1994), summarizing their comprehensive description of the aged and their familial networks in Switzerland, refer to the tensions created between the potentials of family solidarity and the limitations imposed by contemporary social conditions. In view of such accounts, Karl Pillemer and I have proposed that the study of parent–child relations in later life must move beyond a ‘‘love–hate relationship’’ (Lu¨scher & Pillemer, 1998). The vacillation between images of mistreatment and neglect, on the one hand, and comforting images of solidarity, on the other, are not two sides of an academic argument that will ultimately be resolved in favor of one viewpoint. Second, parallel to this theoretical evaluation, we conducted a research project at the University of Konstanz on the reorganization of families after divorce in later life, e.g. an important event in the life course (Lu¨scher & Pajung-Bilger, 1998). Data were collected in semi-structured interviews with 103 persons in 65 families. These interviews included questions about the way all the subjects experience intergenerational relations. Our goal was to distinguish different degrees of mutual solidarity in the aftermath of what in many cases represents a ‘‘turning point’’ in the lives of the individuals involved and their experience of intimate relationships. Yet, even a differentiation in terms of everyday concerns, and by content and types of relationships, did not yield conclusive results regarding the relevance of solidarity. Family members reported both instances of support and of neglect. This led us to search for a concept with which we could take into account the existence of both solidarity and conflict in the process and the understanding of intergenerational relations. The notion of ambivalence in the everyday sense (being torn in two directions) was a first and natural choice. In the course of work along these lines, we became aware, however, that references to the experience of ambivalence in social relationships, and especially in personal relationships, which involve dependency and intimacy, have long been a topic of popular wisdom and of literary writings, even before the term existed. Indeed, insights into what we call in modern language ‘‘ambivalence’’ between parents and adult children can be traced back to the beginnings of human society. In Greek mythology, some of the greatest sagas depict what we now refer to as ambivalence. The best known of these is the tragic drama of the relationship between Oedipus and his father and mother. Reinharz (1986) gives an informative overview on ‘‘loving and hating one’s elders’’ as ‘‘twin themes in legend and literature.’’


97

She refers, among other examples, to the tragedy of Uranus and his sons. Hamlet as well, she tells us, can be read as a portrait of intergenerational relations. Peter von Matt (1995) presents a comparable and very colorful overview of the theme under the provocative title ‘‘Verkommene So¨hne, missratene To¨chter’’ (Degenerate Sons, Misguided Daughters). He draws a line from the biblical story of Absalom to the admonitory children’s book Der Struwwelpeter (Shock-headed Peter – a classic of moralizing German children’s literature) and recalls the complex relationships described in Theodor Fontane’s Effie Briest and in Kafka’s tale ‘‘The Metamorphosis.’’ We may add, as one more illustration certainly known to many readers, Philip Roth’s novel American Pastoral as an example of ambivalence in recent American literature.2 Furthermore, ambivalence can be seen as an ongoing theme in the life-script or biography. Kierkegaard could serve as one of many examples. An impressive study with ambivalence as a latent theme is Lee’s (1998) study of generativity in the life course of the dancer Martha Graham. In everyday life, ambivalences are often experienced, for example, in negotiations over caregiving. They can also be found by examining the overall history of a given relationship. Seen this way, ambivalence is a conceptual tool for evaluating specific situations, as well as for studying the development and institutionalization of the self in the life course. This brief account of recent approaches to the study of intergenerational relations (and given the already-mentioned interplay: to the study of the life course) that draw upon the idea of ambivalence illustrate why it is appropriate to speak of a ‘‘new–old perspective.’’ However, in order to become a useful tool for contemporary social research, a complete, detailed conceptualization is needed.

3. CONCEPTUALIZING AMBIVALENCE 3.1. Elements of a Comprehensive Definition In the light of the foregoing, it seems reasonable to start with a brief look at the original formulation of the term. As far as we know, ambivalence was conceived and first introduced by the Swiss psychiatrist Eugen Bleuler (1910) as one of four core symptoms of schizophrenia. Yet, soon thereafter he argued that ambivalence is not merely a symptom of mental illness, but can also be experienced and thus observed in everyday life. He distinguishes between affective and cognitive ambivalence and points out that the two are closely intertwined (Bleuler, 1914, p. 98). His text already contains a reference to ambivalence in intergenerational relationships (p. 103). Freud first

98

KURT LU¨SCHER

used the concept in an article on the theory of transference (e.g. also with regard to social relationships!). Later, he included it in his theory of the Oedipus complex, as is concisely and clearly demonstrated in a short essay, ‘‘Some reflections on schoolboy psychology’’ (Freud, 1914). Freud thus applied ambivalence to the analysis of an exemplary intergenerational phenomenon, as well as assigning it a role in the life course. This is not the place for a more detailed history of the concept, its reception and its adaptation in different scholarly discourses. Taking into account the major contributions and arguments in the existing literature,3 I would list the following elements as constituents of a comprehensive understanding of ambivalence: The experience of diametrically opposed (polarized) structures and forces in the dynamic fields of individual (and collective) actions and respective relationships. The insight that these experiences are relevant for the identities (selves) of the actors (individuals, in certain contexts also collective actors). In other words, the experience of ambivalence and the ability to cope with it can be understood as an aspect of human agency. The assumption that these polarizations will be interpreted as irreconcilable as long as the actors belong to a certain field of action (or situation) and are concerned, in this context, with the reflection of these tasks. This field of action can be brief, e.g. a turning point, or extend over a longer period of time (for instance becoming a parent).4 The assumption that the experience of ambivalences and the ways of dealing or coping with them can be systematically connected with the aspects of psychological functioning, of the logic of social relations and social structures, including the regulation of social control and power. In view of the background of the concept’s history and its acceptance in the social sciences, I would like to propose the following definition: For purposes of sociological research on intergenerational relations, it is useful to speak of ambivalence when polarized simultaneous emotions, thoughts, volitions, social relations and structures that are considered relevant for the constitution of individual or collective identities are (or can be) interpreted as temporarily or even permanently irreconcilable. Taking this attempt at a comprehensive analytical definition as a reference point, we find, in scholarly texts, two different usages. First, the term can serve as an interpretative (or explanatory) concept. This is, in fact, its primary use in macro-sociological texts as, for instance, in the widespread characterization of ‘‘post-modernity’’ as pervaded by ambivalence.


99

References to social reality are confined to generalizations, based mostly on highly aggregated, generalized data. Descriptions are sometimes presented in the form of ‘‘ideal-types’’ or ‘‘model personalities’’ such as Bauman’s (1997) proposed ‘‘tourist’’ or ‘‘player.’’ This usage is also common in research reviews, for instance in Cohler’s text about young adults ‘‘coming out’’ as gay or lesbian and their parents (see below). Second, the concept of ambivalence may be used as a ‘‘research construct.’’ Here, the goal is to apply the concept in research, such as in surveys, experiments, observations and the analysis of documents. For this purpose, an explicit definition is necessary – one that can serve as the reference point for formulating specific hypotheses and constructing research instruments. We can hypothesize that people must live with ambivalences and that they can cope with them in more or less competent, productive ways. People can even create ambivalences, as mentioned above with regard to the works of creative writers and artists. Deliberately constructing ambivalences can also be a strategy in social interaction. This possibility is another reason to view ambivalences as both opportunities and as burdens. In this regard, the understanding of ambivalence suggested here differs from other usages where – more or less explicitly – the term bears a negative connotation. This is true, for instance, of the term’s usage in characterizing styles of attachment between mothers and children, as well as in other typologies. Closeness and intimacy may reinforce or strengthen the susceptibility to ambivalence. An important precondition of ambivalence is dependency (Smelser, 1998), which begins with birth (or even during pregnancy), continues through childhood and youth into adulthood, and in many cases even into the later phases of the life course. It manifests itself very early in the needs for nurture, care, protection and education. Beyond these immediate obligations, and in the course of fulfilling them, parents develop and acquire specific information and particular knowledge about their individual child as a person. This knowledge reinforces the parents’ power to control and to discipline the child, not only while he or she is young, but also in later life phases. Over the intergenerational life course, the direction of dependency between children, parents and older or younger generations may become more complicated – support and care are specific instances explored in this book. Yet the authority of older persons, established early in life, may persist as another source of ambivalence, even as situations arise that lead to a potential or actual reversal of dependency. Cohler and Grunebaum’s (1981, pp. 120ff., 197ff.) studies of the relationships of mothers and daughters in Italian immigrant families provide many convincing illustrations of this process (see below). More generally, ambivalences in the past and the

KURT LU¨SCHER

100

present may offer an interesting topic in the study of life reviews, both in scholarly work (Staudinger, 1989) and in the curricula of courses offered on practical gerontology. The contemporary relevance of ambivalence can be deduced from a close examination of the structural and cultural conditions of present Western (postmodern) societies. On the macro-sociological level, population dynamics have created a frame in which ambivalence easily emerges. The rise in life expectancy, attributable to improved living conditions for increasingly large segments of the population, was accompanied by a decrease in infant mortality. As a child’s chances of survival increased, the possibility of seeing it as an individual person also increased. A decrease in the birth rate was a logical consequence. Childhood and youth soon came to be seen as specific phases of the life course calling for their own institutions – for instance, public schooling. The same observation can be made with respect to the other end of the life course via the recognition of aging as a life stage calling for its own institutions. The demarcation of different periods or segments of the life course has led to a heightened consciousness of the importance of relationships between age groups, or in other words, between generations. This has been true especially in the realm of the family, and also in society as a whole. The development of social welfare was another factor contributing to this demarcation of life stages and of intergenerational relationships. In many instances, structural conditions for both dependence and autonomy were thereby created. Seen in this way, the concept of ambivalence is another possibility to relate the analysis of the life course to the study of contemporary society and the dynamic interplay of generations and their cultural manifestations (see for example Edmunds & Turner, 2002a, b; Blossfeld, this volume).

3.2. Proposal for a Research Module The foregoing discussion represents a background for new applications in research and respective operationalizations.5 The concern shared by the study of intergenerational relations and life course analysis for the development of personal identity (or the self) through interaction and institutionalization is a major point of reference and allows us to concomitantly pay attention to social relationships. This approach is compatible with a two-dimensional view of personal identity, particularly with G.H. Mead’s (1938) notion of the self as emerging from the interplay between ‘‘I’’ and ‘‘me,’’ where ‘‘I’’ refers to spontaneous subjectivity and ‘‘me’’ refers to


101

generalized others or, more generally speaking, to the interplay between a subjective and an institutional component of the self. Many interpersonal models of personality explicitly refer to Mead. For example, Leary (who developed a circumplex model that describes personality as located between the poles of love vs. hate and dominance vs. submission) speaks of Mead as a ‘‘creative watershed to which later theories of interpersonal relations can trace their sources’’ (Leary, 1957, p. 101). We can see in the juxtaposition between the subjective and the institutional dimensions a primary condition for the experience of ambivalences. In addition, within the module presented below, a secondary condition is suggested by hypothesizing that both dimensions of an intergenerational relationship, the subjective as well as the individual, can be influenced and shaped by fundamental polarizations. Thus, the module is based on a ‘‘twofold’’ notion of ambivalence. This implies a departure from the everyday understanding of the term. The ‘‘personal’’ or ‘‘subjective’’ dimension can be characterized as follows: Parents, children and the members of other involved generations share a certain degree of similarity. While some of this similarity can be attributed to biological inheritance, no inheritance is total, insofar as individual parents and individual children are never genetically identical. Their similarity is reinforced by the intimacy of interactive learning processes, which creates a potential for closeness and subjective identification. At the same time, the biological equipment of each organism is different. Sociologically speaking, processes of maturation increase difference and diversity. Ultimately, children develop different personal identities than their parents. In order to create a schematic representation that can be used in different contexts, two rather abstract labels are needed. To account for not only the socio-spatial, but also for the socio-temporal aspects, we propose – for the subjective component – the terms ‘‘convergence’’ and ‘‘divergence.’’ These two polarities can serve as umbrellas for a variety of attributes. Convergence includes such relational attributes as loving, warm, solicitous, reliable and close. Divergence is characterized as cool, easy-going, indifferent and superficial. For the structural–institutional component, we can conceive of a polar opposition between a desire to preserve the traditional social forms or structures of relationships and a desire for dramatic change. Neither is fully realizable. For instance, although children may choose a way of organizing their private lives that is vastly different from that customary in their family of origin, some ties to childhood experiences may remain, even if only in that they provide a negative background. As technical designations, taking

102

KURT LU¨SCHER

into account again the socio-temporal as well as the socio-spatial aspects, the terms ‘‘reproduction’’ and ‘‘innovation’’ appear useful to express the idea of a dynamic polarization. Here, reproduction includes relational attributes such as inflexible, restrictive and ‘‘stuck in a rut.’’ Innovation is expressed by terms such as open to new experiences, changeable and so on. We can represent these considerations in the form of a module (or diagram). In this way, it is possible to analytically deduce four basic modes of experiencing and dealing with intergenerational ambivalences. Referring to empirical findings and their discussion, as well as to conceptual considerations, we went through different phases of representation.6 We also took into account criticisms that representation in the form of a circumplex-model suggests a static typology, in other words, one where a certain way of dealing with ambivalences is viewed as finite. Overcoming this limitation is highly desirable in the field of life course studies. It seems likely that individual modes of experiencing ambivalences and coping with them change as people move through different contexts and segments of their lives. In order to visualize the dynamics of development e.g. the possibility to move from one type of experience and of coping to another, we suggest using the geometric form of a spiral. As for characterizations of the modes of ambivalence, the already-existing descriptions seem still useful. Thus, the modified module (graphic representation) can be presented and commented on in the following way (see also Lu¨scher & Pajung-Bilger, 1998; Lu¨scher & Lettke, 2002, 2004; as well as Lang, 2004; Brannen, 2003): 1. Solidarity refers to reliable support, or the willingness of the generations to provide each other with services of a not necessarily reimbursable sort. This involves the exercise of authority, but not in the sense of a one-sided exertion of influence and power. Rather, it is understood as representative action including empathy. The maxim of action can be characterized as to ‘‘preserve consensually.’’ The members of a family feel committed to their traditions and get along with one another quite well. Thus, ‘‘solidarity’’ is one possible mode of dealing with intergenerational ambivalences, which in this case may be more covert than overt. (It should be noted that this term implies a specific notion of solidarity and that the term ‘‘loyalty’’ may also be appropriate for this dynamic.) 2. Where family members strive for emancipation, actions predominate that support mutual emotional attachment (convergence) and openness toward institutional change (innovation). Relationships between parents and children are organized in such a way that the individual development and personal unfolding of all family members is furthered without losing

103

Looking at Ambivalences Convergence

Solidarity

Emancipation

To preserve consensually

To mature reciprocally

To conserve reluctantly

To separate conflictingly

Reproduction

Innovation

Captivation

Atomization

Divergence

Subjective (personal) dimension: Convergence vs. Divergence Institutional dimension: Reproduction vs. Innovation Intergenerational Ambivalence: A research module

sight of their mutual interdependence. This general setting contains a certain amount of direct, common purpose pursued by efforts to ‘‘mature reciprocally.’’ Tensions can be discussed openly, and temporary practical solutions can be continually negotiated. 3. Atomization takes into account that family cohesiveness is no longer assured by institutional ties and the subjective experiences of relational histories. The concept expresses the fragmentation of the family unit into its smallest components, specifically individual family members who ‘‘separate conflictingly.’’ Apart from the unalterable fact that family members are parents and children, they otherwise have very little in common. Actions follow a line of conflicting separation, although an awareness of generational bonds remains.

KURT LU¨SCHER

104

4. Captivation designates cases where the family as an institution is invoked to support the claims of one family member against another. A fragile relationship of subordination and superiority thereby arises in which moral claims and moral pressure are used to exert power. Usually one generation, predominantly the parental, attempts – by invoking the institutional order – to assert claims on the other or to bind them by means of moral appeals without, however, basing its claims on a sense of personal solidarity. The guiding maxim here is to ‘‘conserve reluctantly,’’ whereby family members may try to ‘‘instrumentalize’’ each other, not respecting each other as subjects, but using each other as ‘‘means to an end’’ or as objects. I would like to underscore the heuristic character of the module. It is used in an attempt to synthesize and visualize certain basic assumptions about intergenerational ambivalence and to suggest a first set of labels for the poles that characterize the dimension of simultaneously experienced juxtapositions. It also suggests ways to see how the micro- and macro-systems are embedded in a social ecology of action. The module, so far, emphasizes the experience of ambivalences in relationships. Metaphorically, we can evoke the image of a ‘‘dialogue with significant others.’’. Along this line, we can think of other modi. Thus, we can comprehend the experience of ambivalences in the form of a ‘‘dialogue with oneself,’’ and furthermore as a ‘‘dialogue with generalized others,’’ namely as a quarrel with general normative (societal) expectations, or prescriptions. As a general schematic representation, the module encourages further differentiations and adaptations to specific research topics. Such specifications seem to be necessary, especially in applications to life course analysis. Thus, I offer the foregoing conceptual ideas as a proposal to analytically structure the field of research in terms of the concept of ambivalence, particularly in studying intergenerational relationships. Existing studies can be characterized by the way, and to the extent that, they refer to elements of this conceptualization, or use alternatives. The conceptualization represents one of several possible approaches.

4. CURRENT STATUS OF RESEARCH 4.1. Methodological Preliminaries Although this is not the place for a detailed methodological discussion (for this see Lettke & Klein, 2004 and the literature discussed there), I will start


105

with a brief comment on the possibilities to assess the experience of ambivalences and respective actions. In general, it seems more reasonable to use qualitative methods. But we should not ignore the fact that they require highly elaborate interpretative strategies in order to achieve inter-subjective validity, especially when studying accounts given in everyday language and experiences that are not always conscious. Beside the well-established research techniques in the social sciences, advances may also be possible through cooperation with literature studies. For instance, Zima (2002) provides a complex demonstration of ambivalence on the level of syntax, on one hand, and on the level of semantics and content, on the other. In quantitative research, a major obstacle lies in the general orientation of many scaling techniques, insofar as they strive for clarity, in an effort to strictly avoid contradictions. In the available research on ambivalence, the following approaches, techniques and methods are found: 1. Interview techniques addressing the awareness of ambivalence: Respondents can be asked about their awareness of ambivalences in a more or less direct way, by using the term itself or by presenting circumscriptions such as ‘‘feeling torn in two directions.’’ 2. Assessment of relationships with regard to covert ambivalence: Subjects can be invited to characterize their relationships with polarized attributes presented separately, such as warm or loving for convergence, indifferent or superficial for divergence. If the answers are contradictory, because both of the two opposing attributes are simultaneously judged applicable, they can be transformed into indicators of ambivalence. Currently, the most widely used procedure is one proposed by Thompson, Zanna, and Griffin (1995). 3. Use of vignettes: Subjects are presented with situations in which they have to make ambivalent choices. In the following overview, I concentrate on contents. It is not meant to be comprehensive, but rather illustrative. Its focus is on findings and studies, mostly of a quantitative nature, which highlight aspects that may be especially relevant for transfer from the analysis of intergenerational relations to life course research. The systematization is not a strict one, insofar as some studies obviously concern different topics.

4.2. Assessment and Differentiation of Ambivalences Ambivalences, formulated in direct or circumscribed ways, are part of everyday life and are therefore commonplace experiences for men and women,

106

KURT LU¨SCHER

parents and (adult) children. This finding has frequently been confirmed. For instance, an exploratory study by Pillemer and Suitor (2002, p. 609) demonstrates, ‘‘that direct measures of ambivalence toward children can be used effectively...and that ambivalent assessments of the relationship are sufficiently widespread to be of scientific interest.’’ In another analysis of the same data, concerning mothers’ general assessments of parent–child relationships, Pillemer (2004, p. 128) concludes that the ‘‘data offer convincing evidence that parental ambivalence regarding adult children is sufficiently widespread to be of scientific interest.’’ Similar conclusions can be drawn from studies by Connidis (2001), Jekeli (2002), Spangler (2002), and Willson, Shuey, and Elder (2003) and others. Coenen-Huther, Kellerhals, and von Allmen (1994) made a survey of the relations among kin in a representative sample of families. They discovered that a majority of relations, approximately 60%, were experienced and judged positively. However, one third (36%) referred to ambivalences, and a small minority (4%) judged their relationships negatively. More interestingly, the intensity of dilemmas rose with the frequency of mutual help. Ambivalent judgment that are considered important can be detected in about half of the cases. The authors conclude: ‘‘Intensive solidarity is not self-evident’’ (Coenen-Huther et al., 1994, p. 334). Reluctance is apparent, especially in long-term relations. The ongoing studies at Konstanz (Lu¨scher & Lettke, 2004) confirm that if one asks about them directly, using everyday expressions, experiences of ambivalence turn out to be almost commonplace. A similar picture emerges from data concerning the answers to contradictorily formulated statements about relationships, such as, for instance, the following statement: ‘‘[Name of other person] and I often get on each other’s nerves, but nevertheless we feel very close and like each other very much.’’7 In addition, these studies yield a finding that is particularly relevant for life course research: The experience of ambivalence is not judged, per se, as negative. Of importance seems to be the level, the intensity and perhaps the context of ambivalent experiences. In other words, dealing with ambivalences may be understood as a challenge, hence in the context of the life course as a ‘‘developmental task.’’ Here, a connection exists to the origins of the concept and its elaboration in psychotherapy, where several authors see the acceptance or the ‘‘tolerance of ambivalence’’ as a criterion of growth and maturity and stipulate it as a goal of therapeutic efforts. We also find the idea of an optimal level in the experience of ambivalence, for example in a study by Mayer and Filipp (2004). This questionnaire study explored middle-aged adults’ perceptions of their parents’ generativity and


107

the interpersonal consequences of these perceptions. The subjects assessed the typicality of behaviors indicating generativity for their mother or father and evaluated the parent–child relationship on several measures (affection, manifest and latent conflicts). Some of those relations were moderated by adult children’s positive regard for parental advice. Affection was highest at intermediate levels of perceived generativity, but was also linked with moderate levels of manifest parent–child conflict. In the understanding of the authors, these results ‘‘suggest to analyze effects of generativity under the aspect of intergenerational ambivalence’’ (Mayer & Filipp, 2004, p. 166).8 The idea of an optimum level is useful to interpret nonlinear variations and correlations as the expression of the interplay between contradictory forces. Such a view encourages a secondary analysis of existing research. Empirical research on kin networks shaping the life course suggests that the effect of support networks on conjugal quality is curvilinear (Holman, 1981), i.e., extremely cohesive networks might be detrimental to conjugal functioning. The interference model (Johnson & Milardo, 1984; Julien, Markman, Leveille, Chartrand, & Begin, 1994) states that social networks and conjugal relationships may actually compete. Developing relationships create anxiety in social networks, because the time and energy devoted to other relationships are thereby reduced. Thus, social network members may try to hold or regain some influence on their ego by interfering with conjugal relationships. In this perspective, strong networks may not buffer the effects of conjugal conflict, but may actually increase them, because the emergence of conjugal problems opens doors to further interference by network members with a couple’s relationship. These examples also invite us to look at the dynamics of conjugal relationships as a field of overt and covert ambivalent feelings and behaviors. In the Konstanz studies, as outlined in the conceptual part of this chapter, we emphasize the analytical distinction between an institutional and a subjective dimension of ambivalence. The data suggest evidence for the fruitfulness of this idea. In general, ambivalences on the institutional dimension seem to be more pronounced than on the subjective dimension (Lettke & Lu¨scher, 2001, p. 527ff.). This is true for both parents and adult children, a finding which suggests, in addition, that the so-called ‘‘generational stake’’ hypothesis is questionable with regard to ambivalences. Overall, then, ambivalent experiences seem commonplace, yet they differ in character. In other words, the concept of ambivalence should be differentiated. This is an idea that can be traced back to Bleuler, who distinguished ambivalences of feelings, cognitions and volitions. Other authors also adopt this view in their current work (see for instance, Lorenz-Meyer, 2004).

KURT LU¨SCHER

108

Brannen (2003), in a small-scale study of four-generation families, provides a typology of intergenerational relations with respect to the transmission of material assets, childcare and elder care, sociability, emotional support and values. It examines two a fortiori conditions that are thought to shape intergenerational relations: (a) occupational status continuity/mobility and (b) geographical proximity/mobility. Four types of intergenerational relations are generated by this examination: traditional solidaristic; differentiated; incorporation of difference; and reparation in estrangement. The authors look at families holistically and draw on the concept of ambivalence to describe the forces which encourage family members to preserve family patterns and divisive forces that lead them to strike out on their own. It shows how, whatever the type of intergenerational pattern, each generational unit seeks to make its own particular mark.

4.3. Diversification of Contexts In the wider horizon of a comparative study, Fingerman and Hay (2004, p. 145ff.) ‘‘revealed that parents and their offspring do seem to experience greater ambivalence toward one another than they experience in many other social ties.’’ However, other relationships are also considered ambivalent, in particular ties to romantic partners and ties to siblings. The authors’discussion hints at another topic of interest in the possible application of the ambivalence perspective to the study of the life course. Since nearly all the romantic partners of adults older than 20 in the Fingerman and Hay study were spouses or cohabiting partners, they hypothesize that ‘‘proximity may play a role in the experience of ambivalence with romantic partners and siblings. When siblings grow up and no longer live in the same household, there is a precipitous drop in the likelihood that they will be classified as ambivalent; teenagers classified their ties to siblings as ambivalent, whereas individuals in their 20 s did not. It may simply be the case that individuals are more likely to experience ambivalence when they occupy the same life space. This pattern regarding proximity was not the same for parents and children, however. Adult children in their 20 s who do not reside in their parents’ households were more likely to consider their ties to their parents ambivalent than were teenagers who lived with their parents. Therefore, ambivalence between parents and children may reflect different factors than does ambivalence in other social ties’’ (ibid.). The conclusion that suggests itself is plausible: The experience of ambivalences may change over the life


109

course, but this is certainly only the starting point for a range of propositions still to be developed. In an extension and follow-up of the survey done at Konstanz (see above) using as far as appropriate the same instruments, interviews have been made of two types of families facing specific tasks and difficulties. In one group, an adult child suffers from schizophrenia, in the other group, an adult child is on drugs. In both instances, the child was living in a clinical institution at the time of the research. This design allows, among others, a comparison between statements concerning the relationship to the sick child and to other children in the same family. The data show, as hypothesized, a higher frequency of ambivalence in the relationship with the sick child, and a lower relationship quality. Surprisingly enough, there is no significant difference in feelings of connectedness to the children in the families (Brand, 2004; Rudorf, 2004; Burkhardt, 2005). Taking into account additional findings, the general conclusion for this study is the conclusion is justifiable that most parents distinguish among their children in many ways, yet they feel close to and committed to all of them. These results give rise to certain doubts and criticisms of the holistic view of families propagated by some popular systemic approaches used in family therapy. More generally, we may again observe that the usage of indicators of ambivalence, i.e. the ambivalence perspective, promises an understanding of families that reflects their internal dynamics and therefore comes close to real life. The subjective attitudes and orientations of family members are taken into account without neglecting the role of institutionalized bonds. The concern for parent–child relationships in exceptional families is also reflected in studies of families with gay or lesbian children. A large body of research is available; Cohler (2004) offers a comprehensive overview drawing upon the interpretative power of the concept of ambivalence. Among the many topics covered, of particular interest in the life course perspective is the process of ‘‘coming out.’’ It is subject to several forms of ambivalence and requires different strategies of coping, e.g. with regard to personal sameness and difference, to traditional and new life styles. Parents may also have the task of revealing their child’s sexual orientation to kin and friends. On another level, a kind of institutional ambivalence may be implied in the way legislation deals with homosexual partnerships. Should they be treated as just another form of marriage, or should a special legal institution be created (e.g. civil partnerships or civil unions, as is the case in most European countries)? Do gays and lesbians themselves want to accept rules derived from traditional marriage, especially with regard to the dissolution

110

KURT LU¨SCHER

of the relationship? Quite to the point, the German author Lautmann (1996) uses the notion of ‘‘ambivalences of the law.’’ Extending the horizon, it is easy to propose other family configurations as breeding grounds for latent and manifest ambivalences. In single-parent families, relationships with the absent father or mother and struggles for custody may bear all the features of an enduring conflict, putting the child in an ambivalent position. In the case of foster families, the child as well as those who have institutional responsibilities for the arrangement, such as social workers, may find themselves caught up in struggles between the biological mother and the so-called social parents (or legal parents). Here too, legal regulations and procedures may be relevant to the search for a way of pragmatically coping with ambivalences. In Germany, this is the case for the legal obligations of adult children to support their parents when they are poor and need institutional care (Hoch & Lu¨scher, 2002). Divorce at all stages of marital and generational biographies may accentuate, often over a longer period or for an entire lifetime, overt and covert ambivalences. One is reminded of the proposal by Cherlin (1981) to view remarriage as an incomplete institution. In these and comparable cases a specific and elaborate operationalization of the concept of ambivalence is needed if one wants to go beyond simple plausibility. As a result of these efforts, one can expect, as mentioned above, at least a higher level of authenticity with regard to the diversities and the dramas of everyday life. One should also strive for findings, which systematically illuminate the consequences of different levels of awareness and of different strategies in dealing with ambivalences. Practical interests may lie in the evaluation of therapeutic interventions that strive to heighten the awareness of ambivalences and to establish specific ways of dealing with them.

4.4. Ambivalences at Turning Points and Transitions The notion of turning points refers to phenomena, experiences and actions where the awareness of ambivalences may be especially promising and where the interplay between generations and the life course is quite pertinent. A turning point may be understood, metaphorically speaking, as an interruption in a person’s development. It coincides with the necessity, or at least the possibility, to reflect upon personal relationships and the commitments they involve. Changes may be requested and importance attributed to particular relationships, or persons may be asked to restructure their relationships. New commitments and obligations may emerge that compete with


111

ongoing concerns and ties. In reality, ‘‘turning points’’ may extend a certain period of orientation and search, hence it is also appropriate to speak of transitions. They can be seen as fields of action entailing an accentuated experience of ambivalences. Perhaps the most obvious turning point at the intersection between intergenerational relationships, the life course and the social context is the transition to parenthood. This appears in many ways and, not surprisingly, there is still no comprehensive theory of generative behavior and decisionmaking. Several attempts, however, refer to the notion of ambivalence, mostly using the word in an everyday meaning. More elaborate studies along this line point out that decisions are reached only through a lengthy process that takes the form of oscillations typical of ambivalences. A good illustration is the phenomenon of late first motherhood (see Engstler & Lu¨scher, 1991). The experience of ambivalences (as defined above) is bound to the self and personal identity. In addition to their search for the subjective meaning of motherhood, many women are confronted with or exposed to normative expectations, traditional or progressive, by others who are close to them, and also by society at large, as represented by subcultures such as religions and ethnic groups, not to speak of economic pressures and the contemporary organization of the labor market. This topic also illustrates what is referred to above as the experience of ambivalence ‘‘in the dialogue with generalized others.’’ An attempt to draw upon the concept of ambivalence and to further explore its relevance for a typological differentiation of generative behaviors is offered as part of in an analysis of the Swiss Family Survey (Le Goff, Sauvain-Dugerdil, Rossier, & Coenen-Huther, 2005). Ambivalence is used as an alternative to the notion of rational choice in discussing fertility behavior in low-fertility countries like Switzerland. It serves as a key concept to distinguish between four main types of the fertility project: The familialist subculture, either sequential or simultaneous articulation between labor market participation and motherhood, and childlessness. Future trends are discussed in the light of the pressure to change exerted by those women who experience a high degree of ambivalence between their own life aspirations and normative expectations, while also possessing high levels of personal resources. With regard to motherhood as such, a treatise by Parker, with the suggestive title ‘‘Mother Love, Mother Hate,’’ written from a psychoanalytical perspective, merits special attention. Parker (1995, p. 6) refers to Melanie Klein, who ‘‘considered that ambivalence had a positive part to play in

112

KURT LU¨SCHER

mental life as a safeguard against hate.’’ Parker adds: ‘‘I want to go further and claim a specifically creative role for manageable maternal ambivalence. I suggest that it is in the very anguish of maternal ambivalence itself that a fruitfulness for mothers and children resides.’’ The major mechanism can be described as follows: Given the fundamental dichotomy and the awareness of love and hate, mothers are able even in desperate situations to reactivate the forces of love. More generally, mothers search continuously, even under difficult situations, for arrangements that serve the well-being of their children. This fundamental ability to cope with ambivalence creatively can be seen as a genuine cultural and social contribution of mothers to civilization. Contributions like Parker’s make clear why – and also how – a focus on ambivalence can be compatible with feminist thinking. This field is sensitized to possible ambivalences in gender relations and to constructive, as well as destructive, strategies for dealing with them. Referring to a later phase in the life course, Pillemer and Suitor (2002) focus on the tension between autonomy and dependence and find that a key dilemma leading to intergenerational ambivalence is the conflict between the norm of solidarity with children and the normative expectation that children will develop independent lives in the case of the so-called ‘‘off-time transitions’’ – here in the lives of children. As a general finding, the authors showed, ‘‘that adult children’s failure to achieve and maintain normative adult statuses and financial independence, and mothers’ developmental stage predict ambivalent assessments of the relationship. Regression analyses supported these hypotheses and also revealed that the variables predicting ambivalence differed from those that predicted closeness and interpersonal stress’’ (Pillemer & Suitor, 2002, p. 602). In particular, heightened ambivalence can be anticipated when adult children have not attained (or maintained) adult statuses. When parents face such unexpected circumstances, they are likely to experience mixed emotions involving a desire to protect and assist the child, as well as disappointment at the child’s situation and self-doubt regarding parenting. This study, like the one mentioned before, makes explicit use of the concept of ambivalence. It is not difficult to imagine other turning points that display preconditions for the experience of ambivalences, such as occupational choice, or – at the end of a professional career – the period of retirement. Work in these areas would require – and could stimulate – further efforts in the conceptualization of ambivalence. The example suggests viewing non-normative (or even deviant) behavior as a cause of ambivalence. From a theoretical point of view, there may be a linkage with the analysis of stigma, such as that of Goffman (1963). Interestingly enough, although the latter does not use the term ambivalence,


113

he describes behaviors that can be interpreted as strategies for coping with ambivalences. If attention is directed toward specific features of the life course, trauma is certainly an experience that can generate ambivalences in several ways. Under the impact of personal and structural violence, the self is threatened, and this may remain so for a long time, or even lifelong. Thus, the traumatic experience becomes part of the personality. On one hand, it is so subjective that it cannot be shared with others, but on the other hand there may be a strong desire to share one’s experiences, not least of all in the hope of receiving therapeutic support. This holds true for personal traumatic experiences such as child abuse. Traumas can also be collective, as in the case of wars. The Holocaust is a unique case of the experience of collective trauma for which an extensive body of literature exists (see for example Ludewig-Kedmi, 2001, 2004). The twofold experience of ambivalence in connection with personal and collective trauma is concisely summarized by Smelser (2004, p. 53) in the following passage: One of the peculiarities that have been noticed in connection with acute psychological traumas is a very strong dual tendency: to avoid and to reliveyAt the ideational level one main defense is some form of amnesia (numbing, emotional paralysis)y, actual forgetting, denial, difficulty in recalling, or unwillingness to contemplate or dwell on the traumatic event. At the same time, the trauma has a way of intruding itself into the mind, in the form of unwanted thoughts, nightmares and flashbacks. These apparently antagonistic tendencies have presented themselves to some as a paradoxy At the behavioral level, the same double tendency has been observed: A compulsive tendency to avoid situations that resemble the traumatic scene or remind the victim of it, but at the same time an equally strong compulsion to repeat the trauma or to relive some aspect of ity When seeking an analogy at the socio-cultural level, we discover such dual tendencies – mass forgetting and collective campaigns on the part of groups to downplay or ‘put behind us’, if not actually to deny a cultural trauma on the one hand, and a compulsive preoccupation with the event, as well as group efforts to keep it in the public consciousness as a reminder that ‘we must remember’, or ‘lest we forget’, on the other. A memorial to an eventyhas both reactionsy[we can speak of] the compulsion to remember and the compulsion to forget.

4.5. Ambivalences Concerning Specific Fields of Action 4.5.1. Caring The experience of ambivalences may be greater in tasks where tensions and contradictions cumulate. This is certainly the case in caring. For caregivers, and in reference to the subjective component of relationships, sympathy and antipathy are at play, and many caring activities include intimate behaviors

114

KURT LU¨SCHER

that may be embarrassing. From an institutional perspective, normative expectations may exist which juxtapose the commitment of a woman as the daughter of elderly parents with the duties of husbands and wives. Men, too, may be burdened in this way, but caring is still considered a primarily female obligation. These traditional gender ideologies may add to the pressures and thereby further the likelihood of ambivalences. Seen from the point of view of the care-receivers, ambivalent feelings and attitudes may exist as well, since they realize the tensions between insight into apparent dependency and the wish for independency. This is the topic of a monograph by Cohler and Grunebaum (1981) which is cited here as an example of a study that analyzes the phenomenon of ambivalence without using the concept itself. The authors focus on mother– adult daughter relationships in four families of Italian Americans. Their point of departure is the ‘‘paradox in contemporary society where, on the one hand, it is believed that adults will strive to become both psychologically and economically autonomous and self-reliant, while, on the other, findings from systematic investigations of family life show that dependence across the generations is the typical mode of intergenerational relations, including the interdependence of very old parents on their middle-aged offspring’’ (ibid., p. 10). In the concrete case, for the mothers, the acceptance of the daughters commitments are in conflict with the mothers desire to continue to lead their own lives. The authors describe as an illustrative example the relationships of one mother (Mrs. Scardoni) and her daughter (Mrs. Russo) in the following way: Mrs. Russo’s continuing emotional involvement with her mother is both a source of support as well as a source of considerable discomfort and strain. Neither Mrs. Scardoni nor Mrs. Russo can tolerate any disagreement or disharmony, for neither mother nor daughter can admit to their own mixed feelings. On the one hand, Mrs. Russo is very dependent on her mother for help with even the most minute aspects of her life, such as recipes for supper or advice on her problems with her daughter or her husband. On the other hand, she is afraid that her mother will forget about her if she does not maintain continual contact. Burdened by her mother’s demand that she and her brother provide Mrs. Scardoni with the identity that she had never achieved for herself and unable to derive any sense of security or satisfaction from their relationship, Mrs. Russo feels frustrated, resentful, and then guilty. Finally, she becomes so distraught that she can only continue to function by swallowing large doses of the several ‘tranquilizers’ that her family doctor has prescribed for her. (ibid., p. 120)

A similar study of fathers and sons in later life has been published by Nydegger and Mitteness (1991). Their analysis also contains colorful descriptions of ambivalences without using the term itself. There are certainly more studies which contain an implicit and consequently not yet elaborated


115

reference to the idea of ambivalence. It may be worthwhile to reanalyze them in the light of the emergence of a theory of ambivalences. In this regard, a secondary analysis of data from the Berlin Aging Study (BASE) by Lang (2004) merits special attention. It also contains an explicit connection to the life course approach. Data are available from responses by adult children (mean age 54.4 years) to a mailed questionnaire on personal networks and the quality of relationships with parents. Ultimately, yfour distinct patterns of adult children’s relationship styles towards their parents were identified based on indicators of support exchange, personal norms and affective strength: close exchange, resilient giving, strained altruism, and detached distance. The four relationship styles were associated with motivations for seeking contact with parents and the inconsistency of relationship satisfaction with parents. Each of the four relationship styles reflects an individual response to the challenges of the filial task in midlife.

In the interpretation of the author (Lang, 2004, p. 199ff.), yfour observed styles of adult children’s relationships with their older parents are most consistent with the assumptions of the heuristic model of intergenerational ambivalence (Lu¨scher, 1998). According to this model, ambivalence is conceived as an implicit and underlying structure that may be experienced within any intergenerational relationship. For example, adult children may respond to ambivalence with detachment from their parents, referred to as atomization. This response is well reflected in the detached-distant relationship style of adult children’s attitudes towards their parents. Another prototypical response described in the heuristic model of ambivalence is captivation, which refers to feelings of being obligated to take responsibility, while at the same time feeling strained by such responsibility. This response pattern is well reflected in the strainedaltruistic relationship style of adult children. A third prototypical response to intergenerational ambivalence according to the heuristic model of ambivalence is the expression of normatively taking responsibility and close supportive exchanges with the aged parent. This response pattern may be characterized as solidarity and is best reflected in the group of adult children who display a style of close exchange with their parents characterized by strong emotional closeness and much supportive exchange with parents. The relationship style of close exchange with parents comes closest to the concept of family solidarity, at least with respect to the constructs of normative, functional and affective solidarity. Adult children in this group were mostly satisfied with their relationship to their parents and displayed the strongest level of consistency across different ratings of satisfaction. A fourth prototypical response pattern refers to emancipation, which involves a pragmatic attitude of keeping an affective distance to one’s parent while at the same time giving what is needed. Again, this response pattern is reflected in the group of adult children who displayed a style of resilient giving towards their parents. Adult children of this group gave much support because they felt obliged to do so, but also showed relative affective neutrality towards their parents.... Manifestations of personal ambivalence as indicated by the degree of inconsistency in ratings of satisfaction with parents were differently distributed across the four relationship styles. In particular, the strained-altruistic and the resilient-giving relationship styles

116

KURT LU¨SCHER

were found to have the greatest potential for perceptions of ambivalence (i.e. inconsistency). Both styles were associated with a basic and strong attitude towards giving support to one’s parents.

Lorenz-Meyer (2004) has explored, through narratives of young adults in Germany, the generation of ambivalences and strategies of dealing with them in relation to prospective parental care. In her own words, ‘‘the analysis shows that in contemporary Germany the (anticipated) transition of parents requiring personal care is perceived as a structurally ambivalent situation for many adult children that simultaneously values two opposing courses of actions and leads to decisional ambivalence of children between personally supporting their parents in old age and placing them in a nursing home. Participants’ reflections on viable and consensual care arrangements that can be interpreted as an attempt to deal with decisional ambivalence involved a multifaceted process of taking stock of (a) the personal relationship between parents and children, often in comparison with the relationship between parents and siblings; (b) the living situation of older parents; (c) the respondent’s own living situation; (d) past family care arrangements; (e) cultural-normative guidelines; (f) care institutions; and (g) expected commitments of other siblings (and partners).’’ (ibid., p. 246f). The interviews also show, ‘‘that research participants interpreted ambivalences not just in a biographical, but also in a socio-historic context. Participants’ localization of intergenerational positions and relationships in concrete historical conditions can serve to de-personalize and possibly mitigate personal ambivalences’’ (ibid., p. 248). In the context her analysis, Lorenz-Meyer also focused on points of connection and differentiation with the four strategies of dealing with ambivalence identified in the Konstanz studies. ‘‘Displaying inaction and not planning for parental care needs, for example, was not considered as contradicting a solidaristic orientation (and could even be interpreted as ‘‘emancipation’’ in the Konstanz typology, if previous familial care arrangements were not reproduced and personal contact maintained). This was a strategy of dealing with ambivalence that was used mainly by men. The assumption that other siblings, usually a sister, would provide co-residential care tended to facilitate inaction and mitigate decisional ambivalence. Conversely, it was exclusively women (with intermittent employment) who committed themselves to providing co-residential care (that can be interpreted as ‘‘solidarity’’ or, if the initiation of alternative arrangements had failed, as ‘‘captivation’’). Women were also the majority of those who


117

explicitly anticipated accommodating the parent in a home while committing themselves to complementary emotional care (which can be interpreted as ‘‘emancipation’’ if elder care had been provided in the family). For both groups of women the perceived absence of care commitments from other siblings increased decisional ambivalence. A crucial factor for planning residential (rather than co-residential) care was the availability of material resources to afford quality care among women (and some men) with more continuous employment that thereby had a mitigating effect on decisional ambivalence.’’ (ibid., p. 249). Lorenz-Meyer distinguishes between multiple, personal and structural ambivalences that underlie decisional ambivalence in the following way: Personal ambivalences refer to the simultaneity of opposing feelings and orientations such as closeness and distance that came to the fore when participants imagined co-residential living arrangements with their parents. Structural ambivalences refer to the simultaneity of opposing offerings, directives or guidelines for action inherent in institutional structures, such as state agencies or social policies. The notion of multiple ambivalences refers to overlapping personal and structural ambivalences that constitute multiple sources, rather than a single cause for decisional ambivalence. As part of the already-mentioned OASIS project, an extensive comparative study on the care of the elderly and the role of family support systems, complemented the traditional focus on solidarity with an analysis of ambivalences. The authors summarize the results of the quantitative and the qualitative analysis as follows (Lowenstein & Ogg, 2003, p. 223): Correspondence analysis of the ten questions relevant to inter-generational conflict, ambivalence and solidarity resulted in categorizing parent–child relationships into four distinct styles. Harmonious relationship styles were categorized, for example, by getting along extremely well but with an acceptance that conflict and ambivalent feelings could and did occur but without altering the essentially positive relationship experience. Distant family styles were conversely evidenced by emotional distancing, differences in view and the experience of conflict and ambivalent feelings in a way which could or did have a deleterious effect on family relationships. – In the qualitative data, dyads who experienced their relationships as effective and essentially harmonious tended to identify ambivalence or conflict as a part of the process of their relationship. Transitions created by changes in parental health for example, brought about the possibility of negotiating or redefining roles and responsibilities without impinging on participants’ views of the overall quality of the relationship.

118

KURT LU¨SCHER

4.5.2. Inheritance If one is searching for phenomena that seem in the light of experience to be breeding grounds for ambivalences, inheritance is undoubtedly a major candidate. Thus, we may use this topic as an illustration of how the new orientation, namely the interest in ambivalences, sheds light, encourages, stimulates new research interests, close to daily life, and also recalls the importance of interdisciplinary cooperation. Certainly a core phenomenon in the field of generations, inheritance has found surprisingly little attention in the field of social science. This is also true for its relevance for patterns of life courses, individual lives and personal ties. The chapter by Plakans (2004) in Pillemer and Lu¨scher (2004) is a good starting point. The author recalls how important the regulations concerning inheritance were in the past and how much they could influence the life courses of the rich, including aristocrats, as well as peasants and artisans. Major sources of ambivalences can be assumed on a structural level in the juxtaposition of institutional rules and customs, and the desire of the donators to express their personal sympathies, or to reward a child (or another person) for support and attention. Another conflict which most likely induced everyday ambivalences has to be seen in the self-interest of the old in their role as heads of households, as opposed to the desire of the young to have a family of their own and to become autonomous. Ambivalences may also be nourished by the rivalries among siblings. To this Plakans offers concrete illustrations. Ehmer and Gutschner (2000) confirm the overall fruitfulness of the concept of ambivalence for the study of inheritance and more generally speaking for the social history of the family and its implications for personal biographies. They see a major advantage or function of the concept in that it serves to deconstruct the idealizations that have long dominated family rhetoric. An attempt to include the concept of ambivalence in a study of presentday processes of inheritance has been made by Lettke in the Konstanz Inheritance Survey (Konstanzer Erbschafts Survey – KES), which is representative of the German population age 40 and above, using the method of telephone interviewing. His findings confirm, that about a third of the subjects refer to ambivalences – a number which seems lower than one would expect at first glance, and with regard to the usual socio-demographic variables, those with lower levels of education show a significantly higher rate of ambivalence. A more detailed analysis reveals that those who have already received an inheritance are significantly more ambivalent, which suggests that actual experience turns out to entail more difficulties than anticipated. Strong correlations exist between the experience of

119


ambivalences and the responses in terms of motivation. The following instances appear to be of significant importance: the intention to reward those who have provided care, who are especially sympathetic, by whom one wants to be remembered and with whom one shares common convictions and beliefs. Ambivalences also arise if a person wants to support children who have a family of their own and those who are in need. More generally, ambivalences seem to increase if the testator has reasons to deviate from the rules stipulated by the law and by a general societal idea of equity. With regard to the dimensions of the module suggested above, inheritance seems to be a field of action where the tensions between the subjective or personal and the institutional dimensions seem of particular relevance.

5. OUTLOOK In this section, I offer some proposals for a greater rapprochement between the study of intergenerational relationships and the study of the life course, especially with regard to its institutional embeddedness. Such an orientation refers back to the older issue of the interplay between biology and culture, which is fundamental both to the concept of human development and generational succession. A major focus is the understanding of personal identity and the self. In this connection, and also as an answer to recent calls for more there in the field of generational studies, the concept of ambivalence is appropriate. This is appropriate and attractive for at least three reasons. First, this concept too is relevant for a deeper understanding of personal identity in a nonmetaphysical and non-normative way. Second, if used in the sense of Mead (1938), identity development can be understood as advanced by ongoing dialogues with oneself and with significant others. Third, such dialogues imply the possible experiences of being torn in two opposed directions and oscillating between them. With regard to a life course perspective, ambivalences are presumed to activate, or at least to stimulate, the human potential for action in social structures. In other words, dealing with ambivalence requires ‘‘agency.’’ Thus, it is fruitful to view ambivalences as ‘‘neutral,’’ i.e. as possible preconditions for acting. Research on ambivalence should therefore focus on awareness and coping. We can hypothesize, first, that people cope with ambivalence in more or less competent, productive, or even creative ways. Second, the deliberate construction of ambivalences can be a strategy in shaping and organizing social interactions. Third, the personal experience of

KURT LU¨SCHER

120

ambivalences depends on aspects of interactions and social structures and on the embeddedness of ambivalences in role models and collective identities. We can expect that ambivalence will be especially manifest at ‘‘turning points’’ and that it will likewise be apparent throughout the biographical histories of the relationships between parents and their children. Ultimately, dealing with ambivalences can be conceptualized as a ‘‘meta-task’’ of the personal and social organization of intergenerational relations (and other kinds of social relations) over the life course (and vice versa). In addition, we may hypothesize (beyond the existing frameworks) that ambivalences, in a life course perspective, may be experienced in introspection (‘‘inner dialogues’’), as suggested by the idea of ‘‘life review’’ or ‘‘life reflection’’ (Staudinger, 2001). They may be experienced (and have to be dealt with) in social relationships (‘‘dialogues with others’’). Finally, one may consider the impact of generational politics (and politics in general) as creating conditions that can generate ambivalences (‘‘dialogues with generalized others’’). Recalling the frame of reference presented at the PaVie Colloquium Lausanne, we may ask where the experience of ambivalences can be expected to occur and where we may discover specific strategies of coping. I offer the following overview: Topics

Perspective of Subjects

Trajectory

Life reviews

Stage Transitions

Reproductive behavior Leaving home Retirement Caring

Events/tasks/ roles

Perspective of Researcher/Structures Socio-biological foundations Conflict nature/nurture Stages of development Developmental tasks Generativity (Erikson)

Grandparenthood Inheritance (multiple sense) Trauma

Newly introduced concepts also engage us to adopt a new perspective in examining existing theories and their interconnections. In this regard, further explorations within the field of intergenerational relationships, as well


121

as the field of life course analysis, may well be undertaken in regard, for instance, to Erikson’s well-known theory of identity. His schema of eight stages in the development of identity can certainly be read as a sequence of dilemmas with ambivalent qualities. However, Erikson’s theory would have to be linked systematically to descriptions of conduct, social relationships and roles and their possible relevance for the emergence of ambivalence. Another bridge can be built to the theory of generativity. In a recent, very concise summary of its substance by McAdams and Logan (2004), at least the second proposition points to a logical structure of the concept which comes close to ambivalence: ‘‘Generativity may spring from desires that are both selfless and selfish’’ (p. 18). If we want to strive for a closer integration, we should be aware that the focus here has been on relationships. This focus may be welcome in studies of the life course. The linkages between lives merit greater attention. Quite obviously, this draws attention to the dynamics of interpersonal relationships. The idea of ambivalence, as obvious as it may be in the case of intergenerational relationships, can certainly be enlightening for other personal relationships, such those between partners or husband and wives, siblings and even friends and comrades. Their dynamics over a life course may be quite meaningful and consequential. Despite the importance of the concept in relationships, a self-critical observation may well be appropriate. At the present stage of the development of the ambivalence perspective, concern with the consequences of ambivalent experiences is unexplored. The distinction of four different modes in dealing with ambivalences may well be a first step. Yet, more work is needed. As one direction, I would like to offer the following argumentation. The experience of ambivalences – it has been said – should be seen as relevant for the development of the self or personal identity. (This connection is also useful to distinguish ambivalences from trivial experiences of tensions and choices in daily life). Within the framework of a theory of social action, the reference to the self or personal identity implies a close connection to the concept of agency, insofar as it may be understood as the locus of action and of control (see also the chapter by Marshall, in this volume). Therefore, we should pay greater attention, on one side, to the extent to which the possibility (or the inability) to control one’s own behavior with regard to others, and therefore to shape relationships, is a source of ambivalences, and in what ways the mastering of ambivalences goes together with the exercise of power. Preliminary considerations along this line have been presented by Connidis and McMullin (2002). Other behavioral consequences of dealing with ambivalences may be considered as well. In short, a closer

KURT LU¨SCHER

122

interconnection between the study of ambivalences, agency and social control seems desirable and promising. I would like to finish on a more general note by referring to Smelser (1998, p. 13). Exploring the deeper meanings of the ambivalent, he states that, ‘‘we are dealing with a fundamental existential dilemma in the human condition. It is communicated in various dichotomies – freedom versus constraint, independence versus dependence, autonomy versus dependence, maturity versus infancy, and more – but ever the dichotomy, the dilemma appears to be insoluble.’’ In a time, when professional and even economic interests play a major role in the enterprise called ‘‘social science,’’ a reminder of some basic humanistic issues may well be appropriate – not least with regard to the question of how we organize and can organize our lives over the life course and master – as chances and as burdens – the ambivalences occurring in social relationships.

NOTES 1. Note the deliberate formulation as a general heuristic hypothesis: It is not suggested that intergenerational relationships are per se ambivalent, or that they always require dealing with ambivalences, but several reasons and observations, as shown below, speak for the assumption that this may often be the case. 2. From a systematic point of view, this reference to literature implies an important insight: Insofar as fictional works are or can be seen as constructions of imagined worlds, one also may see the ambivalences as deliberately constructed. This may be done on the assumption that these ambivalences are also experienced by readers or viewers in their personal lives. The deliberate creation of ambivalences is also used as a technique in certain psychotherapeutic methods. e.g the so-called ‘‘paradoxical intervention’’ 3. For a more detailed, yet still preliminary overview see Lu¨scher, 2004. Some references to the role of the concept in different discourses can be found in Smelser (1998). The reception of the concept in psychotherapy is outlined in Otscheret (1988) and Knellessen (1978). 4. Thus, we could also say that the concept of ambivalence refers to ‘‘decisionmaking as a process’’ for which the metaphor of ‘‘oscillation’’ seems quite appropriate, or as an alternative image, a ‘‘tug-of-war.’’ 5. Due to spatial limitations, I will focus only on the broad outlines. For a full presentation, see Lu¨scher & Lettke, 2004. The research instruments developed at Konstanz, in partial cooperation with Karl Pillemer, are available in English and German under: http://www.uni-konstanz.de/FuF/SozWiss/fg-soz/ag-fam/famsoz-i.html 6. Here, a note about the methodological status of a diagram may be in order. Following an idea by Bogen & Thu¨rlemann (2003), diagrams represent a unique category of ‘‘text,’’ which stems from the combination of words and graphics. Due to a certain degree of ambiguity and of openness, this kind of representation encourages further interpretations and can thus serve as a means to develop new ideas and even hypotheses.

123


7. This example is taken from the questionnaires used in the Konstanz studies, see footnote 5. 8. There is a parallel to the finding of Pyke and Bengston (1996). In a qualitative research project on family elder-care, they coined the concept of ‘‘overcare,’’ defined as care exceeding recipients’ actual needs which thus may have negative consequences, both relational and developmental. Close-knit networks may not always facilitate parent–child relationships, especially when the expectations of parents and other network members about the child are inconsistent (Belsky, 1984), or when network members are perceived by parents as competitors rather than as supporters in the parenting process (Robertson, Elder, Skinner, & Conger, 1991). It might well be worthwhile to reanalyze these studies in the light of the emerging theory of intergenerational ambivalence.

ACKNOWLEDGMENT I would like to thank James Stuart Brice for support in stylistic and editorial matters, and Denise Ru¨ttinger for general assistance.

REFERENCES Attias-Donfut, C. (1995). Le double circuit des transmissions [The twofold circle of transmissions]. In: C. Attias-Donfut (Ed.), Les solidarite´s entre geńe´rations. Veillesse, familles, e´tat [Solidarities between generations. Aging, families and the state] (pp. 41–81). Paris: E´dition Nathan. Bauman, Z. (1997). Flaneure, Spieler und Touristen. Essays zu postmodernen Lebensformen [Flaneur, gamblers and tourists. Essays about postmodern life-forms]. Hamburg: Hamburg Edition. Bawin-Legros, B., Gauthier, A., & Strassen, J.-F. (1995). Les limites de l’entreaide intergeńe´rationelle [The limits of intergenerational support]. In: C. Attias-Donfut (Ed.), Les solidarite´s entre geńe´rations. Vieillesse, familles, e´tat [Solidarities between generations] (pp. 117–130). Paris: Nathan. Belsky, J. (1984). The determinants of parenting. A process model. Child Development, 55, 83–96. Bengtson, V. L., & Harootyan, R. A. (1994). Generational linkages and implications for public policy. In: K. Kronebusch, L. Lawton, M. Schlesinger, M. Silverstein & R. E. Vorek (Eds), Intergenerational linkages (pp. 210–233). New York: Springer. Bengtson, V. L., Giarusso, R., Mabry, B., & Silverstein, M. (2002). Solidarity, conflict, and ambivalence. Complementary or competing perspectives on intergenerational relationships? Journal of Marriage and the Family, 64, 568–576. Bleuler, E. (1910). Zur Theorie des schizophrenen Negativismus [About the theory of schicophrenic negativsm]. Psychiatrisch-Neurologische Wochenschrift, 18, 171–176; 19, 184–187; 20, 189–191; 21, 195–198. Bleuler, E. (1914). Die Ambivalenz [Ambivalence]. In: U. Zu¨rich (Ed.), Festgabe zur Einweihung der Neubauten (pp. 95–106). Zu¨rich: Schulthess & Co.

124

KURT LU¨SCHER

Bogen, S., & Thu¨rlemann, F. (2003). Jenseits der Opposition von Text und Bild. U¨berlegungen zur Theorie des Diagramms und des Diagrammatischen [Beyond the opposition of text and image. Consideration for a theory of the diagram]. In: A. Patschovsky (Ed.), Die Bildwelt der Diagramme Joachims von Fiore [The images of the diagrams of Joachim von Fiore] (pp. 1–22). Ostfildern: Thorbecke. Brand, C. (2004). Generationenbeziehungen in Familien mit psychisch Kranken. Diplomarbeit [Intergenerational relationships between familys with mentally disordered persons]. Konstanz: Universita¨t Konstanz. Brannen, J. (2003). Towards a typology of intergenerational relations. Continuities and change in families. Sociological Research Online, 8, 1–17. Burkhardt, A. (2005). Generationenambivalenzen in Familien mit einem psychisch kranken erwachsenen Kind. Dissertation [Intergenerational ambivalences between familys with a mentally disordered adult child]. Konstanz: Universita¨t Konstanz. Cherlin, A. J. (1981). Marriage, divorce, remarriage. Cambridge, MA: Harvard University Press. Coenen-Huther, J., Kellerhals, J., & von Allmen, M. (1994). Les re´seaux de solidarite´ dans la famille [Networks of solidarity in the family]. Lausanne: Reálite´s Sociales. Cohler, B. J., & Grunebaum, H. U. (1981). Mothers, grandmothers and daughters. Personality and childcare in three-generation families. New York: Wiley. Cohler, B. J. (2004). The experience of ambivalence within the family. Young adults ‘‘coming out’’ gay or lesbian and their parents. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 255–284). Oxford: Elsevier Science Ltd. Connidis, I. A. (2001). The intergenerational ties of gay and lesbian adults and step kin. A conceptual discussion. Paper presented at the 54th meeting of the Gerontological Society of America. Chicago, IL: Gerontological Society of America. Connidis, I. A., & McMullin, J. A. (2002). Sociological ambivalence and family ties. A critical perspective. Journal of Marriage and Family, 64, 558–567. Donati, P. (1995). Quarto rapporto sulla famiglia in Italia [Fourth report on the family in Italy]. Cinisello Balsamo: Edizione San Paolo. Edmunds, J., & Turner, B. S. (Eds) (2002a). Generational consciousness, narrative and politics. Lanham, Boulder, New York, Oxford: Rowman & Littlefield Publishers. Edmunds, J., & Turner, B. S. (2002b). Generations, culture and society. Buckingham: Open University Press. Ehmer, J., & Gutschner, P. (2000). Das Alter im Spiegel der Generationen. Historische und sozialwissenschaftliche Beitra¨ge [Aging in the mirror of generations. Historical and social science contributions]. Wien: Bo¨hlau. Engstler, H., & Lu¨scher, K. (1991). Spa¨te erste Mutterschaft. Ein neues biographisches Muster der Familiengru¨ndung? [Late first motherhood. A new biographical pattern of family formation?]. Zeitschrift fu¨r Bevo¨lkerungswissenschaft, 17(4), 433–460. Finch, J., & Mason, J. (1993). Negotiating family responsibilities. London: Routledge. Fingerman, K. L., & Hay, E. (2004). Intergenerational ambivalence in the context of the larger social network. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 133–150). Oxford: Elsevier Science Ltd. Freud, S. (1914/1974). Some reflections on schoolboy psychology. In: J. Strachey & A. Freud (Eds), The standard edition of the complete psychological works of Sigmund Freud, (Vol. 13, pp. 241–244). London: Hogarth Press.


125

Goffman, I. (1963). Stigma. Notes on the management of spoiled identities. Englewood Cliff: Prentice-Hall. Hoch, H., & Lu¨scher, K. (2002). Familie im Recht [The Family in the law. A socio-ecological approach]. Konstanz: UVK-Verlagsgesellschaft. Holman, T. B. (1981). The influence of community involvement on marital quality. Journal of Marriage and Family, 43, 143–149. Jekeli, I. (2002). Ambivalenz und Ambivalenztoleranz [Ambivalence and tolerance of ambivalence]. Osnabru¨ck: Der Andere Verlag. Johnson, M. J., & Milardo, R. (1984). Network interference in pair relationships. A social psychological recasting of Slater’s theory of social regression. Journal of Mariage and Family, 46, 893–899. Julien, D., Markman, H. J., Leveille, S., Chartrand, E., & Begin, J. (1994). Networks’ support and interference with regard to marriage. Disclosure of marital problems to confidants. Journal of Marriage and Family, 56, 16–31. Knellessen, O. (1978). Ambivalenz und Doppelbindung. Eine Untersuchung des psychoanalytischen Ambivalenzbegriffes [Ambivalence and double bind. Analysis of the psychoanalytic concept of ambivalence]. Salzburg: Universita¨t Salzburg. Lalive d’Epinay, C., & Bickel, J.-F. (1994). Personnes aˆgeés et re´seau familial [Old people and familar networking]. Me´decine psychosomatique, (1/94), 12–23. Lang, F. R. (2004). The filial task in midlife. Ambivalence and the quality of adult children’s relationships with their old-aged parents. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 183–206). New York, Oxford: Elsevier. Lautmann, R. (1996). Ambivalenzen der Verrechtlichung. Die gleichgeschlechtlichen Partnerschaften im Gesetzgebungsverfahren [Ambivalences of the law. Gay and lesbian partnership in the legislative process]. Zeitschrift fu¨r Frauenforschung, 14(4), 121–128. Leary, T. (1957). Interpersonal diagnosis of personality. New York: The Roland Press Company. Lee, S. (1998). Generativity and the life course of Martha Graham. In: D. P. McAdams & E. de St. Aubin (Eds), Generativity and adult development. How and why we care for the next generation (pp. 429–448). Washington, DC: American Psychological Association. Le Goff, J.-M., Sauvain-Dugerdil, C., Rossier, C., & Coenen-Huther, J. (Eds) (2005). Maternite´ et parcours de vie. L’enfant a-t-il toujours une place dans les projets des femmes en Suisse? [Maternity and the life course. Is there still a place for the child in the life-projects of Swiss women?] Bern: Peter Lang. Lettke, F., & Lu¨scher, K. (2001). Wie ambivalent ‘‘sind’’ familiale Generationenbeziehungen? [How ambivalent ‘‘are’’ familiar intergenerational relationships?]. In: J. Allmendinger (Ed.), Gute Gesellschaft? Verhandlungen des 30. Kongresses der Deutschen Gesellschaft fu¨r Soziologie [Good society? Negotiations of the 30th congress of the German Society for Sociology] (pp. 519–540). Opladen: Leske+Budrich. Lettke, F., & Klein, D. (2004). Methodological issues in assessing ambivalences in intergenerational relations. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 85–113). Oxford: Elsevier Science Ltd. Littwak, E. (1965). Extended kin relations in a democratic industrial society. In: E. Shanas & G. Streib (Eds), Social structure and the family (pp. 290–323). Englewood Cliff, NJ: Prentice-Hall.

126

KURT LU¨SCHER

Lorenz-Meyer, D. (2004). The ambivalences of parental care among young German adults. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 225–252). Oxford: Elsevier Science Ltd. Lowenstein, A., & Ogg, J. (2003). OASIS. Old age and autonomy. The role of service systems and intergenerational family solidarity. Final report. Haifa: University Mimeo. Ludewig-Kedmi, R. (2001). Opfer und Ta¨ter zugleich? Moraldilemmata ju¨discher Funktionsha¨ftlinge in der Shoah [Victim and suspect at once? Moral dilemmas of jewish prisoners in the Shoah]. GieXen: Psychosozial-Verlag. Ludewig-Kedmi, R. (2004). Ambivalenz im Umgang mit der Shoa. Psychologische Perspektiven von Erza¨hlgeboten und Erza¨hlverboten [Ambivalence in exposure with the Shoa. Psychological perspectives of narrative commandments and narrative interdictions]. In: B. Bannasch & A. Hammer (Eds), Verbot der Bilder. Gebot der Erinnerung. Mediale Inszenierungen der Schoah [Ban of images. Commandment of memory. Medial staging of the Shoa] (pp. 1–18). Frankfurt am Main: Campus. Lu¨scher, K. (1998). A heuristic model for the study of intergenerational ambivalence. Konstanz: Forschungschwerpunkt ‘‘Gesellschaft und Familie’’, Universita¨t Konstanz. Lu¨scher, K. (2004). Conceptualizing and uncovering intergenerational ambivalence. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 23–62). Oxford: Elsevier Science Ltd. Lu¨scher, K., & Lettke, F. (2002). L’ambivalence, une cle´ pour l’analyse des relations intergeńe´rationelles [Ambivalence, a key to the study of intergenerational relationships]. Retraite et Societe´, 35, 140–169. Lu¨scher, K., & Lettke, F. (2004). Intergenerational ambivalence. Methods, measures, and results of the Konstanz Study. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent-child relations in later life (pp. 153–179). Oxford: Elsevier Science Ltd. Lu¨scher, K., & Pajung-Bilger, B. (1998). Forcierte Ambivalenzen. Ehescheidung als Herausforderung an die Generationenbeziehungen unter Erwachsenen [Forced ambivalences. Divorce as challenge to intergenerational relationships between adults]. Konstanz: Universita¨ts-Verlag. Lu¨scher, K., & Pillemer, K. (1998). Intergenerational ambivalence. A new approach to the study of parent–child relations in later life. Journal of Marriage and the Family, 60(2), 413–425. Marshall, V. W., Matthews, S. H., & Rosenthal, C. J. (1993). Elusiveness of family life. A challenge for the sociology of aging. In: G. L. Pladdox & M. P. Lawton (Eds), Annual Review of Gerontology and Geriatrics (pp. 39–72). New York: Springer. Mayer, A.-K., & Filipp, S.-H. (2004). Perzipierte Generativita¨t a¨lterer Menschen und die Qualita¨t der Eltern-Kind-Beziehung [Perceived Generativity of old people and the quality of the parent-child-relationship]. Zeitschrift fu¨r Soziologie der Erziehung und Sozialisation, 24, 166–181. McAdams, D. P., & Logan, R. L. (2004). What is generativity? In: E. de St. Aubin, D. P. McAdams & T.-C. Kim (Eds), The generative society. Caring for future generations (pp. 15–31). Washington, DC: American Psychological Association. Mead, G. H. (1938). The philosophy of the act. Chicago: University of Chicago. Nydegger, C. N., & Mitteness, L. S. (1991). Fathers and their adult sons and daughters. In: S. K. Pfeifer & M. B. Sussman (Eds), Families. Intergenerational and generational connections (pp. 249–265). London: The Haworth Press.


127

Otscheret, E. (1988). Ambivalenz. Geschichte und Interpretation der menschlichen Zwiespa¨ltigkeit [History and interpretation of human ambivalence]. Heidelberg: Roland Asanger. Parker, R. (1995). Mother love, mother hate. The power of maternal ambivalence. New York: Basic Books. Parsons, T. (1942). Age and sex in the social structure of the United States. American Sociological Review, 7(7), 604–616. Parsons, T. (1949). The social structure of the family. In: R. N. Anshen (Ed.), The family. It0 s function and destiny (pp. 173–201). New York: Harper. Pillemer, K. (2004). Can’t live with ‘em, can’t life without ‘em. Parents’ ambivalence about their adult children. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 115–132). Oxford: Elsevier Science Ltd. Pillemer, K., & Lu¨scher, K. (Eds). (2004). Intergenerational ambivalences. New perspectives on parent–child relations in later life. Amsterdam: Elsevier. Pillemer, K., & Suitor, J. J. (2002). Explaining mothers’ ambivalence toward their adult children. Journal of Marriage and the Family, 64(3), 602–613. Plakans, A. (2004). Intergenerational ambivalences in the past. A social-historiacal assessment. In: K. Pillemer & K. Lu¨scher (Eds), Intergenerational ambivalences. New perspectives on parent–child relations in later life (pp. 63–82). Oxford: Elsevier Science Ltd. Pyke, K. D., & Bengtson, V. L. (1996). Caring more or less. Individualistic and collectivist systems of family eldercare. Journal of Marriage and Family, 58, 379–392. Rein, M. (1994). Solidarity between generations. A five-country study of the social process of aging. Wien: Institut fu¨r Ho¨here Studien. Reinharz, S. (1986). Loving and hating one’s elders. Twin themes in legend and literature. In: K. Pillemer & R. S. Wolf (Eds), Elder abuse. Conflict in the family (pp. 25–48). Dover, MA.: Auburn House. Roberts, R. E. L., Richards, L. N., & Bengtson, V. L. (1991). Intergenerational solidarity in families. Untangling the ties that bind. Marriage and Family Review, 16, 11–46. Robertson, E. B., Elder, G. H., Skinner, M. L., & Conger, R. D. (1991). The costs and benefits of social support in families. Journal of Marriage and Family, 53, 403–416. Rossi, A. S., & Rossi, P. H. (1990). Of human bonding. Parent–child relations across the life course. New York: De Gruyter. Rudorf, S. (2004). Generationenambivalenzen in Familien mit einem substanzabha¨ngigen erwachsenen Kind. Diplomarbeit [Intergenerational ambivalences in families with a drug addicted adult child]. Konstanz: Universita¨t Konstanz. Shanas, E., Townsend, P., Weddeburn, D., Friis, H., Milhoj, P., & Stehouwer, J. (1968). Old people in three industrial societies. New York: Atherton Press. Smelser, N. J. (1998). The rational and the ambivalent in the social sciences. American Sociological Review, 63(1), 1–16. Smelser, N. J. (2004). Psychological trauma and cultural trauma. In: J. C. Alexander, R. Eyerman, B. Giesen, N. J. Smelser & P. Sztompka (Eds), Cultural trauma and collective identity (pp. 31–59). Berkley, Los Angeles, London: University of California Press. Spangler, D. (2002). Ambivalenzen in intergenerationalen Beziehungen. Hochaltrige Mu¨tter und deren To¨chter. Diplomarbeit [Ambivalences in intergenerational relationships. Old mothers and their daughters]. Berlin: Technische Universita¨t Berlin. Staudinger, U. (1989). The study of live review. An approach to the investigation of intellectual development across the life span. Berlin: Max-Planck-Institut fu¨r Bildungsforschung.

128

KURT LU¨SCHER

Staudinger, U. M. (2001). Life reflection. A social-cognitive analysis of life review. Review of General Psychology, 5, 148–160. Sussman, M. B. (1959). The isolated nuclear family. Fact or fiction. Social Problems, 6, 333–347. Szydlik, M. (Ed.) (2000). Lebenslange Solidarita¨t? Generationenbeziehungen zwischen erwachsenen Kindern und Eltern [Lifelong solidarity? Intergenerational relationships between adult children and parents] (Vol. 2). Opladen: Leske+Budrich. Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). 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. von Matt, P. (1995). Verkommene So¨hne. MiX ratene To¨chter. Familiendesaster in der Literatur [Degenerated sons. Failed doughters. Family disasters in literature]. Mu¨nchen: Hanser. Willson, A. E., Shuey, K. M., & Elder, G. H. J. (2003). Ambivalence in the relationship of adult children to aging parents and in-laws. Journal of Marriage and Family, 65(4), 1055–1072. Zima, P. V. (2002). L’ambivalence romanesque. Proust, Kafka, Musil [Romanesque ambivalence. Proust, Kafka, Musil]. Paris: L’Harmattan.

PART II: TRANSITIONS

129


130

AGENCY AND STRUCTURE IN EDUCATIONAL ATTAINMENT AND THE TRANSITION TO ADULTHOOD$ Jeylan T. Mortimer, Jeremy Staff and Jennifer C. Lee INTRODUCTION A macrostructural perspective highlights cultural values, normative timetables, stratification processes, and institutional career lines as determinants of the content and pacing of role enactment through the life course, and the acquisition of corresponding role identities. A social psychological perspective, in contrast, emphasizes the exercise of human agency as a central causal force in shaping the life course, including the expression of values and identities, self-regulative processes, decision-making, and striving to achieve personal objectives through goal selection, strategic planning and action. According to the macrostructural perspective, the life course consists of sets of interrelated trajectories (of education, work, family, community participation, and so forth) with an individual’s location in each at any $

This research was supported by a grant ‘‘Work Experience and Mental Health: A Panel Study of Youth,’’ from the National Institute of Chile Health and Human Development (HD44138) and the National Institute of Mental Health (MH42843).


131

132

JEYLAN T. MORTIMER ET AL.

given stage heavily dependent on positioning in prior phases. For example, in the US context the socioeconomic status of the family of origin affects children’s schooling in the very early years (placement in reading groups, marks received, grade retention), which in turn influences assignment to middle school and high school ‘‘tracks,’’ the opportunity to take college preparatory and ‘‘advance placement’’ courses, and consequently, the likelihood of entering a selective college, or even any institution of higher education (Entwisle, Alexander, & Olson, 2003). The educational trajectory is, in turn, closely linked to the trajectory of occupational achievement. A macrostructural perspective strongly informs sociological models of intergenerational occupational mobility. In these models, social psychological processes involving significant others’ influence and aspirations are considered essentially as intervening variables, mediating the relations between socioeconomic origins and destinations. At the extreme, expressed aspirations and goals for the future may be interpreted as signifying mere recognition of likely outcomes, given initial starting points, rather than psychological constructs that have their own motivating and determining force. The psychological constructs may predict subsequent outcomes, but for structural reasons alone. The second social psychological approach, while acknowledging that social location yields diverse and unequal opportunities and constraints at all stages of the life course, emphasizes the differential capacities of individuals to effectively exercise agency or initiative. Important individual differences enable some to take fuller advantage than others of the opportunities linked to their structural positions, or to more effectively overcome the constraints associated with disadvantage. Social scientists who take this point of view focus on psychological processes that lead to the crystallization of goal formation and channel individual action in preferred directions. Gecas (1991) argues that the self-concept itself is a motivating force because persons strive to protect and to enhance their self-esteem, sense of efficacy, and authenticity, and to validate their valued character identities. Individuals express their identities most notably through role selection and performance. Personal constructs that implicate future action, such as ‘‘the possible self’’ (Markus & Nurius, 1986), ‘‘self-efficacy’’ (Bandura, 1997), ‘‘planful competence’’ (Clausen, 1993), ‘‘locus of control’’ (Rotter, 1966), and ‘‘authenticity’’ (Gecas, 1991) are of central interest, since these are thought to guide strategic action. Processes of ‘‘goal selection,’’ ‘‘compensation,’’ ‘‘optimization,’’ ‘‘assimilation,’’ ‘‘accommodation,’’ planning, and related processes (see Wiese, Freund, & Baltes, 2000; Brandstaedter, 1998; Heckhausen, 1999) may be more or less adequate in bringing about the more favorable

Agency and Structure in Educational Attainment

133

life course outcomes. Individual orientations and actions are presumed to have significant consequences for life course trajectories of socioeconomic attainment, family formation and stability, psychological well-being, and physical health. Agentic strategies may be more effective among those in more favorable structural locations, but individual differences in these capacities are expected to influence life outcomes, net of the advantages or disadvantages associated with structural position. According to this point of view, agentic decision-making and behavior are especially important in late adolescence and early adulthood because the points of entry to socioeconomic trajectories are established at this time. These two fundamental paradigms, emphasized by macrosociologists and social psychologists, are surely not mutually exclusive, since structural and agentic processes operate in tandem. In fact, their coexistence and their codetermination of social processes make interdisciplinary collaboration essential. Greater mutual attention is especially called for between the ‘‘social structure and personality’’ branch of social psychology, concerned with the manifestations of social location in enduring individual psychological orientations and behavior; and the subareas of sociology concerned with the ‘‘micro–macro link.’’ The strong reliance on the rational choice paradigm among investigators interested in micro–macro analysis has them closer to economics and political science than to social psychology (House & Mortimer, 1990). Integration of social psychological and sociological frameworks is especially well suited to the increasingly prominent life course perspective (Mortimer & Shanahan, 2003). To fully comprehend the progression of individual lives, we must not only study the changing historical context and the variable and shifting structures of institutional trajectories and the ways these impinge on individual actors (the family life cycle, occupational careers, organizational career lines, etc.). We must also consider the ways persons subjectively orient themselves to the opportunities and constraints that they encounter in particular times and places, and the strategic actions they devise to reach their goals in particular structural contexts. However, in the face of persistent evidence that socioeconomic origin constrains life chances intra- and inter-generationally, it could be argued that the expression of agency makes little difference. Do adolescents’ aspirations in fact exert a meaningful influence upon their attainments? Or, alternatively, do their aspirations merely signify recognition, sometimes resignation, with respect to probable future destinations, given wellunderstood consequences of placement in class structures and school tracks? This question, in general, has perplexed scholars of stratification and the life

134


course for several decades (see, for example, Roberts, 1968, for an early formulation). Whereas associations between psychological dimensions (values, efficacy, and identity) at one point in time and social locations at a later date may be interpreted as providing support for the social psychological paradigm, the alternative structural interpretation of such relations is likewise tenable. It must be acknowledged at the outset that answers to these questions must be conditional, and especially, historically specific. That is, the relative importance of structural and social psychological determination of the life course is likely to vary across time and place. Shanahan, Elder, and Miech’s (1997) study illustrates this point very well. The Terman gifted men who came of age during the Great Depression, facing a quite limited employment market, showed strong proclivity toward postsecondary enrollment irrespective of their prior aspirations. But for a subsequent cohort, entering the labor force in circumstances of expanding economic opportunity, early educational values and aspirations were quite consequential for decisions to pursue postsecondary schooling. Educational aspirations were thus more predictive of higher educational attainment for the later cohort who encountered the postwar economic boom as they entered adulthood. Thus, the capacity of personal dispositions and goals to influence subsequent attainment depended on historically variable opportunities for choice. Our examination of the process of attainment is especially strategic in the contemporary US context where institutional bridges between school and work are notably undeveloped (Mortimer & Kru¨ger, 2000). High school education is focused on college preparation, with little emphasis on vocational certification, apprenticeship programs, or other structural connections between schools and employers. Young people develop their own strategies in positioning themselves for the labor market. When the time comes to find a full-time job, they must rely largely on their own resources, friends, and relatives. Of course, those who come from more advantaged families have greater resources at their disposal, which help them to complete college and compete successfully for highly coveted professional and managerial jobs. But even for these youths, pathways to college completion and high levels of occupational attainment are by no means transparent or certain. Youths’ efforts in school and their work behavior reveal two major strategies of human capital acquisition, the first, more successful one, through education; the second, constituting the more rapid route to fulltime employment, through early part-time work (Mortimer, 2003). Investigations of the process of attainment have generally taken the strategy of linking personal dispositions to outcomes measured at a much


135

later date (see Mortimer, 1996, for a review). For example, researchers have predicted the features of workers’ jobs by their occupational reward values, articulated as long as a decade prior (Mortimer & Lorence, 1979; Lindsay & Knox, 1984). In the classic status attainment model, educational and occupational aspirations measured during the senior year of high school predict first jobs and subsequent jobs, attainments that often occur many years later. Clausen (1993) reports that adolescents who were more ‘‘planfully competent’’ had more satisfying occupational careers, more stable marriages, and generally more gratifying experiences throughout their lives. Given this general research strategy, the processes linking the personal orientations, measured at Point A, and the outcomes, measured at Point B, are largely unknown. In this paper, we argue that attention could fruitfully be directed to what is inside this ‘‘black box,’’ to the processes, mechanisms, and behaviors between Points A and B. We contend that researchers should examine the actions that may express agency, actions that connect earlier psychological orientations, such as goals and values, and subsequent attainments in the life course. What are the immediate, day-to-day behaviors, provisional goals, and shorter-term outcomes that link the widely spaced orientations and outcomes that are the subject of most prior investigations? Moreover, we argue that psychological orientations provide clearer links to agentic processes, and are more likely to be predictive of the attainments and conditions of life that are of interest to social scientists, if they are domain specific. In support of this claim, we have presented evidence elsewhere that economic efficacy has stronger predictive power with respect to educational plans and postsecondary school attendance than global efficacy (Grabowski, Call, & Mortimer, 2001). We draw on findings from the Youth Development Study (YDS), a 15-year prospective study of work experience and the transition to adulthood, to illustrate our approach. These highlight the interrelations of educational orientations, patterns of employment during high school, and educational attainment; as well as the ways adolescent future orientations influence youth’s multifaceted pathways of transition to adulthood.

THE YOUTH DEVELOPMENT STUDY The Youth Development Study (YDS) is the only long-term longitudinal study designed to address the processes of structure, agency, development, and attainment that surround early work experiences in adolescence.

136


The study began with a randomly chosen community sample of ninth graders enrolled in the St. Paul Public School District in Minnesota;1 1,010 students agreed to participate, and 1,000 filled out surveys in the first data collection in 1988. United States Census data indicate that this site is quite comparable to the nation as a whole with respect to economic and social indicators (Mortimer, 2003). Most of the analyses reported here use data collected over the years from 1988, when the respondents were in the first year of high school (ninth grade), mostly at the age of 14 and 15; through 2000, when they were 26–27 years old. This panel study is still in process; the most recently collected data were obtained in 2005. Extensive tracking and follow-up procedures during each wave of data collection ensured that students who dropped out of school after ninth grade, or moved to another school district, continued to be followed. Data for the post-high school period, from 1992 to 2002, were collected via mailed surveys. The surveys contained large batteries of questions focused on work experiences in several domains – in the family, school, and neighborhood, as well as in formal work establishments. Extensive series of questions addressed the respondents’ psychological engagement, values, goals, and future orientations with respect to education, work, family, and community domains, as well as key activities in each of these spheres. As such, this study is well suited to examine the interrelations of structure and agency as youths make their transition to adulthood. Seventy-six percent of the initial panel remained through the 2000 data collection. The retained sample was somewhat more advantaged than the initial sample in terms of socioeconomic background and family composition (favoring the two-parent family), and retention was more likely for females than males. The distributions of demographic characteristics across the initial 1988 and 2000 samples, however, were substantially the same.

EDUCATIONAL ORIENTATIONS AS PREDICTORS OF EARLY WORK INVESTMENT The YDS enables assessment of lengthy pathways of attainment from orientations toward schooling at the time of entry to high school, to patterns of work investment during high school, to young adult educational and occupational trajectories, and to 4-year college degree receipt. Let us first consider adolescent educational achievement and engagement at the age of 14 and 15, as predictive of very early labor force involvement

137


during the succeeding 3 years of high school. Most prior studies of adolescent employment rely on the number of hours worked at a given time of observation, or over a period of months or years. Our more nuanced measure attempts to capture more long-term strategic involvement in the labor force, as it includes employment duration, measured in months, as well as intensity, or the average hours of work during the full period of employment. Each was dichotomized: months, so as to distinguish near-continuous employment from much shorter investment in work; and hours, so as to separate those who on average work more than 20 hours per week from those who work more moderately, limiting their hours to 20 or fewer. The 20-hour mark is widely considered the point at which employment becomes ‘‘excessive’’ for in-school American youth, interfering with other activities and increasing the likelihood of problem behaviors (Committee on the Health and Safety Implications of Child Labor, 1998). Cross-classifying the two dimensions yields five work patterns during the 10–12th grades of high school: (1) the most invested, high duration/high intensity workers; (2) sporadic, low duration/high intensity workers; (3) occasional, low duration/low intensity workers; (4) steady, high duration/ low intensity workers; and (5) non-workers. Table 1 shows the distribution Table 1.

Patterns of Labor Force Participation During High School by Gender. Percentage Distribution

Not working Occasional Low duration-low intensity Sporadic Low duration-high intensity Steady High duration-low intensity Most Invested High duration-high intensity Total N

Mean Months of Work

Mean Hours of Work

Total

Boys

Girls

Boys

Girls

Boys

Girls

7.0

9.9

4.6

0.0

0.0

0

0

23.7

23.2

24.1

9.8

11.7

578

650

18.4

23.2

14.3

10.4

11.8

1,216

1,376

24.9

18.2

30.6

22.0

22.0

1,263

1,328

26.0

25.6

26.4

21.9

22.2

2,678

2,587

100.0 887

100.0 406

100.0 481

138


of these patterns of work investment. Occasional, steady, and most invested workers are about one quarter each of the total; sporadic workers constitute 18%. The fact that only 7% of the panel reported no work experience at all demonstrates the high prevalence of part-time work experience, while school is in session, in the American context. Combining school and work is a nearuniversal experience in adolescent life. There are some notable gender differences; boys are more likely than girls to be sporadic workers and to be non-workers; girls are more likely than boys to be steady workers. For boys and girls in the same investment category, however, mean months and hours of work are quite similar. That is, high duration workers are employed approximately 22 of the 24 months of observation and low duration workers accrue only about half that number. From an agentic perspective, adolescent work may be conceived as instrumental action directed to enhancing future prospects for socioeconomic attainment (and other goals). Those who expect that they will accrue human capital mainly through their efforts in the educational system would likely downplay employment during high school. But those who think that they have limited chances for academic success, especially with respect to higher education, would likely seek alternative means of positioning themselves for adult full-time work. Disinterest and poor performance in school, as well as structural disadvantage (i.e., low socioeconomic status of origin) would likely foster attempts to acquire human capital through early intensive employment. If educational achievement has rather low priority in the adolescent peer culture, as Coleman (1961) suggested, one might also expect those who are more highly oriented to their peers to be less engaged in school and more attracted to work. School misconduct constitutes another indication of disengagement and difficulty in the educational realm. In prior investigations, we found that adolescents are not randomly selected (or allocated) to the work investment patterns (Mortimer, 2003; Mortimer, Oesterle, & Staff, 2003). For example, high levels of ‘‘educational promise’’ in the ninth grade increased the likelihood that youth would pursue steady or occasional patterns of employment, rather than more intensive work investment, during the following 24 months of high school. Educational promise referred to the ninth graders’ level of engagement and success in school: it included indicators of grade point average, academic self-esteem (the perception of self as intelligent, a good reader, and high in school ability), educational plans for the future, and intrinsic motivation toward schoolwork. High-promise youths scored above the median on at least three of these measures. Youths who pursued the ‘‘steady’’ work pattern during high school were particularly advantaged vis-a`-vis the educational realm.


139

They had high educational promise in the ninth grade, were unlikely to get in trouble at school, and had relatively low levels of engagement with their peers. Occasional workers were similarly characterized by high educational promise and low school misconduct. What is it about steady and occasional work that appears to attract young people who are more strongly engaged in the educational enterprise, who consider their classes more meaningful and interesting to them, and whose early efforts in the educational realm (as indicated by grade point average) are more successful? Surely, the ‘‘steady’’ pattern of working is the one that is the most ‘‘balanced,’’ as it involves high continuity of employment while at the same time, because hours of work are limited, permits simultaneous investment in homework and other school activities. Interestingly, youth who pursue this pattern are especially likely to say that they had sought employment ‘‘to save for my future education.’’ Our further analyses show that such ‘‘balanced’’ workers pursue multifaceted time use strategies throughout high school, indicating substantial investment in paid work, schoolwork, extracurricular activities, the peer group, and the family (Mortimer, 2003; Shanahan & Flaherty, 2001). Further indicative of agentic action, prior educational engagement not only influences the pattern of investment in work; it also is linked to the quality of work that is subsequently obtained. The features of their jobs suggest that teenagers with limited proclivity toward school seek human capital acquisition through employment. For example, low grade point average in the ninth grade is associated with greater advancement opportunity, higher earnings, more learning opportunities, and a perception that work confers greater status amongst one’s peers. Those youths who had lower academic self-esteem early on had higher earnings in their high school jobs and perceived greater work-derived status amongst their peers (Mortimer, 2003, p. 132). Clearly, a set of trade-offs between school and work is revealed. Young people who are more successful in school invest in work in a manner that yields less in the way of extrinsic and intrinsic occupational rewards, but enables greater investment in school and the capacity to pursue a wellrounded, multifaceted, and achievement-oriented adolescent life style. In contrast, youths who perform more poorly in this domain use their time so as to obtain more learning opportunities in the workplace, greater immediate economic reward, and higher status in the eyes of their peers (likely comprised of others who have similar orientations to school and work as they do themselves). Assessment of work investment and work quality after the ninth grade tells much about how the expression of academic interests,

140


achievements, and goals upon entry to high school affect immediately ensuing work-related behaviors that have quite divergent consequences for socioeconomic attainments during the post-high school years.

WORK INVESTMENT AND TRAJECTORIES OF SOCIOECONOMIC ACHIEVEMENT A growth curve analysis applied to annual survey data collected in the years following high school (1991–2000) has shown that youths who pursued less intensive, steady and occasional work patterns during high school make greater investments in higher education in subsequent years. That is, they accrue more months of higher education during the first year following high school, with background and socioeconomic origin controlled, and they manifest fewer months of full-time work (Mortimer et al., 2003, p. 451). High intensity work patterns were associated with lower initial levels of educational investment, and also with greater investment in full-time work. Furthermore, patterns of growth in months of education and months of full-time work for the work investment groups confirm the merit of thinking of these behavioral patterns in terms of strategic action. Low-intensity high school workers, starting from a higher initial educational investment level, have steeper losses in months of education during the 9 years following high school. They exhibit more pronounced increases in full-time work. In contrast, high-intensity workers during high school are more attached to the labor force during the first year after high school, and manifest a less steep increase in months of full-time work thereafter. Lower intensity (steady and occasional) workers start off with fewer mean months of full-time work. By 1998, however, they have caught up to the more intensively employed students during high school. Their gains in full-time work are steeper than those of their more intensively employed high school peers. To what extent do these distinct patterns of time use following high school-investment in higher education vs. full-time work – exhibited by the teenage work investment groups-yield differential achievement? We can now link early educational orientations, high school work patterns, and educational attainment. We have examined the youth’s capacity to achieve a prominent educational goal, the BA (bachelor of arts or sciences) degree (Mortimer, 2003; Mortimer et al., 2003). Consistent with classical models of status attainment, we found that early school performance and aspirations (indicated by the educational promise variable in the ninth grade), as well as

141


parents’ education, were linked with the receipt of a BA degree or higher in young adulthood. However, employment in a ‘‘steady’’ pattern of work during high school, in comparison to more intensive involvement, also conferred a net advantage in educational attainment in young adulthood, independent of socioeconomic background, educational promise, school misconduct, and other predictors. Based on the most recent wave of data collection (2002), Fig. 1 displays the percentage of respondents that currently have obtained at least a BA degree based on their prior engagement in work during the high school period. Forty-four percent of young people in the steady work pattern during high school had earned a BA degree by 2002, which was much higher than the 17% of the ‘‘most invested’’ and 12% of the ‘‘sporadic’’ workers, but also higher than that of the ‘‘occasional’’ and non-workers (33%). Although the high school patterns of work are influenced by prior socioeconomic background and school performance, defined by gender, race, parental education, and early educational promise, even when these and

50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% No Work

Fig. 1.

Steady

Occasional Most Invested Sporadic

Total

Receipt of 4-Year BA/BS degree (2002) by High School Work Investment Categories.

142


other relevant factors are controlled, teenage work patterns are found to have independent influence on postsecondary educational achievement. The ‘‘steady’’ work pattern is found to be particularly salutary. Why do steady workers during high school have the advantage in postsecondary educational attainment? We find evidence that subsequent work patterns are critical. More recent investigation by YDS researchers (Mortimer & Staff, 2004) demonstrates that the work patterns the young people pursue in the years immediately following the high school period appear to mediate the relationship between the ‘‘steady’’ work pattern in high school and early adult educational attainment. In Fig. 2(a) and (b), we show the average months of higher education, full- and part-time work in the years right after high school for the ‘‘steady’’ and ‘‘most invested’’ workers. As shown in Fig. 2(a), the ‘‘steady’’ high school workers averaged nearly identical months of postsecondary school attendance and part-time work (between 7–8 months) in the 4 years immediately following the high school period. Approximately 4 years after the end of high school, months of school attendance and part-time work began to decline, while months of full-time work increased rapidly. Fig. 2(b) highlights a very different pattern of work and school involvement during the transition to young adulthood for the ‘‘most invested’’ high school workers. For these youths, investment in part-time work and school is rapidly replaced with increasing involvement in full-time work. Despite the predictive power of work involvement during high school with respect to educational attainment in young adulthood, we find that the inclusion of subsequent work patterns in this key period of postsecondary education investment renders prior associations insignificant (Mortimer & Staff, 2004). Especially for the ‘‘steady’’ workers, the effective combination of working and studying is a familiar pattern by the time they enter college. Since most college students have to maintain employment to at least partially assume their educational (and/or living) expenses, the time-use strategies these students learned earlier – of balancing school and work – may enable them to continue combining school and work roles without compromising their educational goals. We thus find evidence of agency in the selection to work patterns, in patterns of educational exploration and investment during high school, and in the determination of future educational and work trajectories. Our strategy of assessing domain-specific achievement-relevant indicators (e.g., economic self-efficacy, educational promise) and achievement-relevant behaviors closely following (i.e., patterns of labor force investment during high school, steps taken to go to college) have yielded high returns with respect to the understanding of trajectories of school enrollment and

143

Agency and Structure in Educational Attainment 10 School Pt-work

9 8

Ft-work

7 6 5 4 3 2 1 0 19

20

21

22

(a)

23

24

25

26

23

24

25

26

Age

10 9

School Pt-work

8

Ft-work

7 6 5 4 3 2 1 0 19 (b)

20

21

22 Age

Fig. 2. Average Months of Higher Education, Ft-Work, and Pt-Work in the Years Following the High School Period for the (a)’’Steady’’ High School Workers, and (b) ‘‘Most Invested’’ High School Workers.

144


full-time employment (Grabowski et al., 2001), and eventual BA receipt. While we cannot fully rule out structural explanations, our analyses consistently show that agency-relevant psychological precursors have significant effects on immediate achievement-relevant behaviors, and that these behavioral patterns are associated with educational attainment, net of several indicators of socioeconomic origin.

AGENCY AND PATHWAYS OF TRANSITION TO ADULTHOOD The collaborative work of the YDS team is now moving to a more general, and also more complex objective – to assess how diverse family and achievement-related attitudes and plans, measured in high school, influence successive configurations of roles in the succeeding years that constitute latent pathways of transition to adulthood (see Macmillan & Eliason, 2003). These pathways encompass not only education and work roles, but also markers of adult family formation (leaving the parental home, marriage, and parenthood). Thus, our earlier treatment of attitudes toward school as precursors of work investment is expanding to consider further agentic motivations and expectations related to the family and intimate relations. Our earlier focus on trajectories of educational attainment and employment is now also supplemented by consideration of multifaceted, interrelated trajectories of educational, work, and family behavior, constituting divergent pathways of transition to adulthood. We use a two-staged latent class analysis set forth by Macmillan and Eliason (2003; see also Clogg, 1995) to empirically investigate the simultaneous interplay among these social roles in the transition to young adulthood. This method is especially germane to a life course perspective, in that it enables the identification of role configurations (and how they change or remain stable over time) and the life paths that link these role configurations through stages of the life course.2 Fig. 3(a)–(c) depicts the role trajectories within three general life paths. Life path I (probability in sampled population ¼ 0.165), shown in Fig. 3(a), characterizes a precocious or hastened transition to adulthood. During adolescence, individuals in life path I have a relatively high probability of parenthood (approaching 20%), as well as relatively low probabilities of residing with parents (80%) and attending school (80%) in comparison to those individuals in life paths II and III. An accelerated transition to

Expected Role Probability


1.00

0.90 0.80 0.70 0.60 0.50 0.40 0.30

0.90 0.80 0.70 0.60 0.50 0.40 0.30

0.20

0.20

0.10

0.10

0.00 Ages 14-18

Ages 18-19

Ages 20-21

Ages 23-24

Ages 25-26

0.00 Ages 14-18

Ages 18-19

Age

(a)

In school Children

Ages 20-21

Ages 23-24

Ages 25-26

Age

Married Lives with parents

Steady, fulltime work

(b)

In school Children



1.00


0.90 0.80


1.00

0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Ages 14-18

Ages 18-19

Ages 20-21

Ages 23-24

Ages 25-26

Age

Fig. 3.

In school Children



Latent Life-Path Probabilities for (a) Precocious Transition to Adulthood, (b) Delayed Transition to Adulthood and (c) Multifaceted Transition to Adulthood.

145

(c)

146


adulthood continues in the years immediately following high school, as the probability of school attendance declines sharply, while the probability of steady, full-time work grows rapidly.3 Life path II (probability in sampled population ¼ 0.576), depicted in Fig. 3(b), indicates a delayed transition to adulthood. During adolescence, individuals in this pathway have a very low probability of parenthood. Residence in the parental home (for at least part of the year) and school attendance are prolonged in the years immediately following high school. Family formation is also delayed (the probability of parenthood and marriage are both under 30% at ages 25–26). Interestingly, sporadic, full-time work is favored between the ages of 18 and 21 (not shown), perhaps indicating summer work that is compatible with higher education, followed by a rapid increase in the probability of steady, full-time work. Finally, life path III (probability in sampled population ¼ 0.259), shown in Fig. 3(c), suggests a ‘‘multifaceted’’ transition to young adulthood that is neither rushed nor delayed. Similar to life path II, respondents have a high probability of residing with parents and attending school. The work patterns of individuals in life path III are also similar to those in life path II, as there is a rapid increase in steady, full-time work in the years immediately following high school. However, the probability of having a child rises much faster at ages 23–24 than in life path II (50% at ages 23–24, compared to just 9% for life path II); marriage exhibits a similar pattern. In addition, the probability of residing with parents also declines more rapidly in early young adulthood for this group. To assess the impact of structural placement and agency on the probabilities of following these three life paths, we regress individuals’ life path probabilities on family background factors – race, gender, parents’ highest educational level, family income, and family composition – and future plans and goals for education, work, and family. As shown in Table 2, measures of family background and socioeconomic status have a significant impact on one’s life-path probabilities. In model 1, females and those whose parents have lower levels of education and family income have higher propensities for following a precocious life path. Males and adolescents whose parents have higher educations and income have higher propensities for the slower, delayed entrance into family and work roles, manifested in life path II, than those with lower socioeconomic origins. Parental education is the only significant socioeconomic predictor of the probability of following life path III. Adolescents whose parents have higher levels of education have lower propensities toward this multifaceted path. These findings highlight the significance of structural advantage in the transition to adulthood.

OLS Regression Coefficients of Latent Life Path Probabilities (n ¼ 498). Precocious (Life Path I) Model 1 B

White (vs. non-white) Male (vs. female) Parental education Family income Intact family (vs. non-intact) Marriage importance Parenthood importance Family self-efficacy Career importance Occupational aspiration SEI Occupational designation Economic self-efficacy High educational promise (vs. low) Intercept R2

0.634 0.538 0.418 0.287 0.211 – – – – – – – – 0.752 0.19

(s.e.) (0.34) (0.23) (0.08) (0.06) (0.30) – – – – – – – – (0.46)

Delayed (Life Path II)

Model 2 B 0.752 0.552 0.285 0.241 0.042 0.166 0.247 0.090 0.113 0.020 1.468 0.131 1.100 0.513 0.26

(s.e.) (0.33) (0.23) (0.08) (0.06) (0.29) (0.20) (0.19) (0.16) (0.20) (0.01) (0.49) (0.06) (0.25) 1.14

Model 1 B 0.474 0.946 0.318 0.351 0.009 – – – – – – – – 3.814 0.15

(s.e.) (0.40) (0.28) (0.10) (0.07) (0.35) – – – – – – – – (0.55)

Model 2 B 0.531 0.966 0.207 0.320 0.134 0.166 0.144 0.158 0.098 0.022 1.731 0.102 0.848 2.463 0.19

(s.e.) (0.40) (0.28) (0.10) (0.07) (0.35) (0.25) (0.23) (0.19) (0.25) (0.01) (0.60) (0.07) (0.30) (1.38)

Multi-faceted (Life Path III) Model 1 B 0.086 0.007 0.132 0.045 0.019 – – – – – – – – 0.940 0.03

(s.e.) (0.21) (0.14) (0.05) (0.04) (0.18) – – – – – – – – (0.28)

Model 2 B 0.047 0.011 0.135 0.050 0.028 0.030 0.045 0.087 0.057 0.001 0.086 0.010 0.023 0.792 0.04

(s.e.) (0.21) (0.15) (0.05) (0.04) (0.18) (0.13) (0.12) (0.10) (0.13) (0.00) (0.31) (0.04) (0.16) (0.73)


Table 2.

po0.05. po0.01. po0.001.

147

148


Of particular relevance given our present concerns, the results show that agency and planfulness are also important in predicting subsequent life trajectories. According to our analyses, agentic orientations toward work and education is what matters, not expectations and values with respect to family life. Model 2 shows that the ninth graders with higher educational promise and those with higher occupational aspirations have lower propensities for the precocious path. Economic efficacy also reduces the likelihood of precocious adulthood. Adolescents who specify what their career might be in the future, indicating greater vocational crystallization, have significantly higher probabilities of following this precocious life path. High educational promise and high occupational aspirations upon entry to high school predict higher probabilities of being on a path characterized by ‘‘delayed’’ transitions into adulthood, which is not surprising given the great educational investment and late exit from school that characterize this path. It is interesting to note that adolescents who had greater occupational uncertainty (indicated by failure to answer the question about their future work plans) also have a greater propensity for path II. For those who do know what occupation they hope to have in the future, however, the higher their aspirations, the more likely they will follow this pathway. This pattern of findings suggests that delayed entrance into adulthood can signify vocational indecision, in which youth uncertainty leads to longer time spent in school and later family formation. Alternatively, for those who have formulated high aspirations at an early stage (age 14–15), delayed markers of adulthood may result from a more focused postponement of full-time work and family roles in order to attain the amount of education necessary for high career aspirations. The longer incumbency of preadult roles and uncertainty about future goals are often discussed in tandem when addressing the contemporary transition to adulthood, as both foster extended exploration and enable the postponement of adult role commitments. They do not, however, always coincide. Moreover, delayed timing, when combined with uncertainty, is likely to have a very different socioeconomic outcome than delayed timing without uncertainty. With clear goals, young people can maximally utilize the elongated transition to adulthood to develop their human capital and otherwise groom themselves for future attainments. Without clear goals, the lengthy transition to adulthood can be one of ‘‘floundering,’’ insecurity, and reduced achievement. Attesting to the significance of adolescent agency, the addition of an individual’s plans and expectations for family, career, and school in model 2 explains 7% and 4% more of the variance in the first two life path propensities, respectively, than the measures of family background alone.


149

In addition, we find evidence that while the other socioeconomic predictors remain just as strong, the impact of parents’ education on these two life-path propensities is partially mediated by individual agency. These results suggest that while social position does in fact play a part in ‘‘routing’’ individuals into certain life trajectories and impacts the timing of onset of adult roles, one’s goals and expectations for these roles also have independent effects on his or her early life course trajectories.4 But unlike the other two pathways, Table 2 also suggests that an individual’s propensity for a multifaceted life path is not affected by the importance he or she places on family, education, or career.

DISCUSSION Our assessment of domain-specific indicators of agency and immediate actions taken in the pursuit of goals is shown to have payoff with respect to understanding the process of socioeconomic attainment during adolescence and the transition to adulthood in contemporary urban America. We thereby obtain a glimpse of what may be inside the ‘‘black box’’ intervening between the expression of goals and values in middle adolescence and life course outcomes in early adulthood. Our study is, however, not without limitations. We have confined ourselves to the study of a single panel, located in a particular national setting, in one brief period of historical time. Given its origin in the upper Midwest region of the US, this panel is not well suited to the assessment of racial and ethnic differences. Moreover, we have little information about structural location other than socioeconomic origin. The St. Paul schools, from which the students were selected, do not have visible ‘‘tracks’’ that may constitute institutionally located ‘‘ladders’’ of achievement. In general, the study does not speak to structural placement within the high school that could address the kinds of macrostructural arguments we alluded to at the beginning of this paper. Notwithstanding these limitations, our work does suggest distinct pathways through the early life course, implicating agency, achievement-relevant educational and work behaviors, and early adult attainments. Moreover, our analyses suggest that agency is intricately connected to adolescent employment patterns. A central contribution of our work is the finding that adolescent employment is neither uniform nor randomly distributed during high school. In fact, there are five distinct patterns of work investment that are linked to social background, early educational promise, the quality of work experiences, and early adult socioeconomic attainments.

150


Might the work investment patterns – steady, occasional, sporadic, most invested, and none – be considered a kind of ‘‘emergent’’ structure? They would surely be recognized as structural dimensions or patterns, as ‘‘institutional career lines,’’ if they were assigned by educational or other authorities, or if they were clearly recognized and validated by their connections to the school. However, unlike the German system of apprenticeship, no central authority decides how much students should work while they are attending school. Instead, these work-investment patterns emerge from the youth’s actions in the context of the naturally occurring, ‘‘free’’ labor market (Mortimer & Kru¨ger, 2000). They arise from the multiple decisions and actions of employers, parents, counselors, students themselves, and other ‘‘significant others’’ who persuade, advise or influence them. To the extent that social structures that regulate and channel behavior are emergent from the very agentic behaviors of actors themselves, our attempts to discover distinct influences of agency and structure become increasingly complicated, perhaps futile, as all are part of an interrelated and finely articulated process of life course emergence. However, our analyses show how early goals and values are linked to work behaviors during high school, which in turn have predictive power with respect to subsequent trajectories of work, schooling, and educational attainment. Our most recent, and at this time least fully developed analyses, take this strategy one step further – to consider how a variety of agentic orientations affect life pathways characterized by configurations of school, work, and family roles. We thus explore how early family, school, and work orientations and goals affect life paths of transition to adulthood. The preliminary analyses presented here indicate that expectancies and values surrounding future educational and occupational role enactment must be taken into account to fully understand propensities for earlier or delayed timing of key markers of adulthood. We have come a long way from early models of attainment that simply linked adolescent educational aspirations with adult educational and occupational attainments. In expanding the range of agentic orientations under consideration, and addressing the intervening behaviors that enable goals and values to be actualized, our understanding of agency and the transition to adulthood is enhanced. While structure – here considered by way of social background location – is certainly also important, we find considerable evidence that the psychological constructs exert independent effects. These investigations could fruitfully be extended to other realms – for example, to trajectories of family stability, civic involvement, mental health, and physical well-being. We trust that these reflections and analyses will


151

prod others to explore how structural location, psychological indicators of agency, and agentic behaviors act in concert to produce contemporary life course pathways.

NOTES 1. To assess whether participants (64% of those invited) differed from nonparticipants along common demographic characteristics, a probit analysis of the decision to participate was conducted, including information from the 1980 Census reported at the tract level to characterize the neighborhoods of all eligible families. Boys and older students (some of whom were retained in grade) were less likely to participate than girls and those who were the same age as most of their classmates. (Information about the gender and age of invitees was obtained from school records.) The racial, family structure, and socioeconomic composition of students’ neighborhoods (indicated by median household income, receipt of public assistance, and educational and occupational status) were not related to participation (Finch, Shanahan, Mortimer, & Ryu, 1991). 2. In the first stage, latent class modeling techniques are used to estimate the unobserved latent role configurations of the markers at each age range (14–18, 18–19, 20–21, 23–24, and 25–26), based on the cross-classifications of observed roles at each. The second stage of the analysis generates a latent class model of the agespecific latent class configurations developed in the first stage. Latent life path models are estimated using cross-classifications of unobserved, or latent, role configurations. These cross-classifications are derived from a ‘‘realization’’ of the latent role configuration transition table (Macmillan & Eliason, 2003). 3. The steady ‘‘full-time’’ workers averaged near continuous employment during the sophomore, junior, and senior high school year at more than 20 h per week. In the years immediately following the high school period, steady full-time work indicates those respondents averaging 35 or more hours of paid work per week during 11 or more months of the year. 4. To see if these patterns differ by gender, we included interaction terms between gender and each psychological measure in the model (not shown). These interaction effects were not significant.

REFERENCES Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Brandstaedter, J. (1998). Action perspectives on human development. In: R. M. Lerner (Ed.), Handbook of child psychology, 5th ed., Vol. I: Theoretical models of human development (pp. 807–863). New York: Wiley. Clausen, J. A. (1993). American lives: Looking back at the children of the great depression. New York: Free Press.

152


Clogg, C. (1995). Latent class models. In: G. Arminger, C. Clogg & M. Sobel (Eds), Handbook of statistical modeling for the social and behavioral sciences (pp. 311–359). New York: Plenum Press. Coleman, J. S. (1961). The adolescent society: The social life of the teenagers and its impact on education. New York: Free Press. Committee on the Health and Safety Implications of Child Labor. (1998). Protecting youth at work. Washington, DC: National Academy Press. Entwisle, D. R., Alexander, K. L., & Olson, L. S. (2003). The first grade transition in life course perspective. In: J. T. Mortimer & M. Shanahan (Eds), Handbook of the life course (pp. 229–250). New York: Kluwer Academic/Plenum. Finch, M. D., Shanahan, M. J., Mortimer, J. T., & Ryu, S. (1991). Work experience and control orientation in adolescence. American Sociological Review, 56, 597–611. Gecas, V. (1991). The self-concept as a basis for a theory of motivation. In: J. A. Howard & P. L. Callero (Eds), The self-society dynamic: Cognition, emotion and action (pp. 171–188). New York: Cambridge University Press. Grabowski, L. J., Call, K. T., & Mortimer, J. T. (2001). Global and economic self-efficacy in the educational attainment process. Social Psychology Quarterly, 64, 164–179. Heckhausen, J. (1999). Developmental regulation in adulthood: Age-normative and sociostructural constraints as adaptive challenges. Cambridge and New York: Cambridge University Press. House, J. S., & Mortimer, J. T. (1990). Social structure and the individual: Emerging themes and new directions. Social Psychology Quarterly, 53, 71–80. Lindsay, P., & Knox, W. E. (1984). Continuity and change in work values among young adults: A longitudinal study. American Journal of Sociology, 89, 918–931. Macmillan, R., & Eliason, S. R. (2003). Characterizing the life course as role configurations and pathways: A latent structure approach. In: J. T. Mortimer & M. Shanahan (Eds), Handbook of the life course (pp. 529–554). New York: Kluwer Academic/Plenum. Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969. Mortimer, J. T. (1996). Social psychological aspects of achievement. In: A. C. Kerckhoff (Ed.), Generating social stratification: Toward a new research agenda (pp. 17–36). Boulder: Westview. Mortimer, J. T. (2003). Working and growing up in America. Cambridge, MA: Harvard University Press. Mortimer, J. T., & Lorence, J. (1979). Work experience and occupational value socialization: A longitudinal study. American Journal of Sociology, 84, 1361–1385. Mortimer, J. T., & Kru¨ger, H. (2000). Pathways from school to work in Germany and the United States. In: M. T. Hallinan (Ed.), Handbook of the sociology of education (pp. 475–498). New York: Kluwer Academic/Plenum. Mortimer, J. T., & Shanahan, M. J. (2003). Handbook of the life course. New York: Kluwer Academic/Plenum. Mortimer, J. T., Staff, J., & Oesterle, S. (2003). Strategic patterns of adolescent work and early socioeconomic attainment. In: J. T. Mortimer & M. Shanahan (Eds), Handbook of the life course (pp. 437–460). New York: Kluwer Academic/Plenum. Mortimer, J. T., & Staff, J. (2004). Trajectories of educational and occupational attainment in adolescence and the transition to adulthood. Paper presented at the biennial meeting of the Society for Research on Adolescence, March 11–14, Baltimore. Roberts, K. (1968). The entry into employment: An approach to a general theory. The Sociological Review, 16, 165–184.


153

Rotter, J. B. (1966). Generalized expectancies for internal vs. external control of reinforcement. Psychological Monographs, 80, 1–28. Shanahan, M. J., Elder, G. H., Jr., & Miech, R. A. (1997). History and agency in men’s lives: Pathways to achievement in cohort perspective. Sociology of Education, 70, 54–67. Shanahan, M. J., & Flaherty, B. P. (2001). Dynamic patterns of time use in adolescence. Child Development, 72, 385–401. Wiese, B. S., Freund, A. M., & Baltes, P. B. (2000). Selection, optimization, and compensation: An action-related approach to work and partnership. Journal of Vocational Behavior, 57, 273–300.


154

NON-NORMATIVE LIFE COURSE TRANSITIONS: REFLECTIONS ON THE SIGNIFICANCE OF DEMOGRAPHIC EVENTS ON LIVES Frank F. Furstenberg INTRODUCTION A prominent tenet of life course research holds that ill-timed and unexpected events, non-normative transitions, and disorderly status sequences have profound and lasting consequences for an individual’s success in later life. Simultaneously rooted both in early demographic and sociological studies undertaken in the middle part of the last century, life course theorists have long argued that social timetables regulate status passages, creating social support and expectations, and generating resources and rewards to those who observe culturally mandated schedules (see, for example, Elder, 1998; Neugarten, Moore, & Lowe, 1965; Rindfuss, Swicegood, & Rosenfeld, 1987). In their classic treatise on aging and social stratification, Riley, Johnson, and Foner (1972) described how role transitions were socially orchestrated through age norms that govern movement of populations into and through social roles. In kinship-regulated societies, these rules determine the structure of generations and social reproduction. In modern societies, theorists Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 155–172 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10005-7

155

156

FRANK F. FURSTENBERG

such as Parsons (1954), Merton (1968), and Ryder (1965) explained that social timetables are critical for determining and coordinating the basic institutions of society: the family, educational systems, and the labor market. It is easy to understand how the assumption that unscheduled life events have a singular power to cast a permanent shadow over the life course became pervasive in the sociological and social demographic literature grounded in a life course perspective. Many theoretical arguments appear to support the notion that planning, timing, and orderly sequencing of role transitions increase the probability of successful attainment. First, in the middle of the past century, sociologists who worked in the tradition of symbolic interaction emphasized that preparation for making a status or changes was crucial to an actor’s ability to negotiate transitions. The ideas of role preparation and anticipatory socialization, popular in the sociological literature of the 1950s and 1960s, advanced the proposition that successful acquisition of social roles required training, practice, and selfdefinition (Becker & Strauss, 1956). It logically followed that correct timing was an important feature of successful passage through a series of interrelated status transitions. Social demographers, interested in the transition to adulthood, began to examine the impact of timing of events on long-term success in status attainment and being off-time conferred social risk (Clausen, 1972; Marini, 1984; Winsborough, 1979). With the work of Neugarten et al. (1965), social psychologists and sociologists also began to explore how norms about timing operated to regulate when and under what circumstances individuals would make significant status transitions (Hagestad, 1990; Settersten, 1998). A line of research began to look into how institutional gatekeepers – family, schools, employers, and legal authorities – helped to set and reenforce the social timetable by allocating resources and rewards for staying on schedule and penalties for violating age norms. The costs of either being early or late have been assessed in a host of journal articles. As Elder (1984) ingenuously argued in Children of the Great Depression, historical events such as wars or economic downturns, outside the control of individuals, can drastically revise the opportunities for role acquisition and hence shape the course of lives. It is evident in Elder’s work that he embraced the general assumption that timing – fortuitous or ill – affects the structure of opportunities at both an individual and social level, influencing the life chances of actors and cohorts in a given society. Another related line of theory about the life course argued that the sequencing of life events similarly had important consequences for success in both the labor market and the family. Again, this proposition rests on the

Non-Normative Life Course Transitions

157

assumption that when individuals follow predictable and socially organized pathways, they are more likely to encounter success. Educational completion prior to employment or marriage prior to first birth compared to the opposite order are demographic examples that have received widespread attention by demographers, economists, and sociologists. In a well-known article published in 1961, Harold Wilensky (1961) showed how orderly as opposed to disorderly careers were associated with positive consequences for a range of economic outcomes as well as for psychological adjustment in later life. Thus, there is a long history in the social sciences of believing that nonnormative or unscheduled transitions confer social disadvantage. My own work was deeply grounded in the life course perspective, and on how the timing and sequencing of transitions altered prospects for successful incorporation into adult roles (Furstenberg, 1976; Modell, Furstenberg, & Hershberg, 1976). Of course this set of ideas, as Matilda Riley and her coworkers observed several decades ago, were equally applicable to the passage from adulthood to retirement.

TEENAGE CHILDBEARING AS A NON-NORMATIVE EVENT At the same time that Riley, Elder, Wilensky, and other social scientists were setting out a life course perspective, I was beginning what has turned out to be a life-long study of the social consequences of teenage childbearing. My project in Baltimore, initiated in the mid-1960s, began more by happenstance than a deliberate intention to test a basic proposition of life course theorists. However, I soon came to see the applicability of this theoretical framework for examining the question of how and why premature parenthood occurs and its lasting implications for success in later life for young parents and their children. In my first book on the topic, I grounded my research in the idea that early childbearing was a prime example of a violation of the normative schedule for family formation. My point of departure was the oft-quoted observation by the social demographer, Arthur Campbell (1968), who speculated that: The girl who has an illegitimate child at the age of 16 suddenly has 90 percent of her life’s script written for her.

In the decades that followed a huge literature has accumulated in the U.S., especially, but in Western Europe more generally, on the impact of early

158


childbearing on the later life success of teenage parents and their children (see, for example, Hayes, 1987; Brown & Eisenberg, 1995; Maynard, 1997). It is beyond the scope of this essay to review this literature in any detail or even recount the findings of my own 30-year longitudinal study of a population of teenage mothers who gave birth to a child while in their midteens. Initially, my findings, like the work of others, appear to supply strong support for the conclusion advanced by Campbell without the benefit of much empirical evidence. But over time, with more sophisticated techniques of data analysis, there is growing consensus among researchers in the field that early childbearing did not produce the kinds of effects that life course theorists would have predicted. While policy makers, the public at large, and even teenage mothers themselves, still tend to believe that early childbearing has deleterious, long-term effects on educational and occupational success as well as mental and social health – both for the mother and the offspring – the findings in the literature show that support for this conclusion is, at best, ambiguous (Furstenberg, 2003). Long-term, longitudinal studies reveal, contrary to expectations, that, at most, parenthood during adolescence has only modest long-term impacts on teenage mothers and their children when compared to mothers who delay their first births and their children when the two populations are matched in other respects. Thus, rather than affirming that non-normative demographic transitions like teenage childbearing profoundly shape the course of later life – that demography is destiny – the evidence seems to point in quite the opposite direction. Non-normative events produce diverse consequences, conferring disadvantage on some but seeming to perturb little or not at all the pathways of others. I will return to the reasons why this is so shortly after discussing the other example from another line of research that I have engaged in, the impact of marital dissolution and remarriage for children’s development.

MARITAL DISRUPTION AS A NON-NORMATIVE EVENT Long before the formulation of the life course perspective, social critics and cultural commentators in the early part of the 20th century had reached the conclusion that marital disruption has devastating effects on children’s life chances. Their literature argued that divorce, as a life event or transition, when viewed from the perspective of children, has many of the same properties as early childbearing. It thrusts children into a family situation which


159

is usually unplanned and for which they are typically unprepared. It deprives them of resources in the form of parental time and money, and it restricts their ability to acquire social skills from the absent parent. Moreover, since divorce is frequently followed by remarriage or cohabitation (and sometimes further disruption), it exposes children to a succession of transitions that are often accompanied by geographic moves, changes in household routines, and shifting rules and regulations (Furstenberg & Cherlin, 1991; McLanahan & Sandefur, 1996). In short, there is abundant reason to believe that divorce as a life event fundamentally compromises children’s immediate and long-range welfare. In recent years, the social and personal costs of divorce for children have been the topic of a considerable body of research. As in the case of early childbearing, the popular impression of what this research has shown diverges from what many professional researchers have concluded, though it can hardly be said that consensus exists. A few researchers, most notably in the U.S., Wallerstein and Blakeslee (1989) and Waite and Gallagher (2000), have concluded that divorce is every bit as detrimental to children’s life chances as has been long believed. However, a growing number of researchers in the field have reached the conclusion that, surprisingly, marital disruption is relatively weakly linked to later life outcomes (Moynihan, Smeeding, & Rainwater, 2004). Like many of my peers, I would be inclined to say that the long-term impacts of divorce are modest to moderate in the aggregate, which is to say that divorce confers risk for some but certainly not for most children whose parents’ marriages break up during their childhood years (Furstenberg & Cherlin, 1991; see also Cherlin, 1999). Now, if I am correct in my characterization of the evidence from studies carried out over the past several decades, we might reach the conclusion, at least based on these two prime examples, that the consequences of nonnormative events outcomes in later life are surprisingly minor in light of life course theory. In the remainder of this paper, I will first try to explain why theorists overstated the adverse impact of ill-timed and out-of-sequence events. Then I will try to revise the theory in ways that are more consistent with the evidence. Doing so suggests an agenda for future research, a topic that I develop in the final part of the paper.

WHEN THEORY AND EVIDENCE COLLIDE Looking back at the two case examples that I have mentioned, they share some prominent features. Both could be characterized as social problems

160


where researchers have been working against a backdrop of heated public discourse and have joined the public in treating evidence in a cavalier fashion: the theory seemed so compelling that the evidence collected was uncritically evaluated. Even if Arthur Campbell, a respected social demographer, had engaged in hyperbole in assessing the impact of early childbearing on the subsequent life course of young mothers, surely could any reasonable person doubt that teenage parenthood does not irreparably compromise the life chances of young mothers and their offspring? And, even if Judith Wallerstein was a self-declared advocate for reducing the toll that divorce takes on children, is it reasonable to question her conclusion that divorce produces lasting damage for those who experience the loss of a parent and grow up in a single-parent household or, worse yet, experience a cascade of family change (Ahrons, 2004)? And, indeed the findings in each of these separate lines of research initially seemed to conform to conventional wisdom. In the 1970s and early 1980s, study after study showed that teenage mothers had extraordinarily high rates of school dropout, reliance on welfare, subsequent fertility, unstable marriages, and so on. Similarly, Wallerstein’s conclusions, based on a small clinical study in the 1970s, found widespread support from other research showing a powerful association between divorce and a range of adverse outcomes for children. However, both areas of research – as has been true in many studies of unscheduled and ill-timed events in the life course – have suffered from potentially fatal methodological problems. Economists have long recognized the problem of unobserved differences or endogamy, the inability to identify causality between an event and its consequences. In the 1980s, some sociologists too, began to come to grips with the issue (Lieberson, 1985). Simply put, teenage childbearing, divorce, and other such events, do not occur randomly within a population. To the contrary, these events occurred selectively to individuals or couples who differ in a wide variety of ways that distinguish them from their counterparts who are able to avoid the event. I need not dwell on this point because it has become so widely understood by social scientists that few fail to appreciate the burden it places on reaching causal conclusions about the links between events and their consequences. Over the past couple of decades, a wide variety of methodological strategies have been employed to tease out causality in assessing the influence of timing and sequencing in the life course. Controlling for prior group differences has become standard, but it is widely known that such an approach inevitably over-estimates casual effects by omitting unmeasured differences. Fixed effects models in longitudinal research partially address this problem.


161

Similarly, researchers have used statistical methods for distinguishing between exogenous and endogamous influences. Finally, researchers have devised clever quasi-experimental approaches to identify true casual links. None of these approaches, by itself, has completely solved the problem of selection, but together they have demonstrated that a healthy share of causal influence of ill-timed or out-of-sequence events is social selection rather than true cause. Specifically, much and probably most of the presumed effect of teenage childbearing on later life outcomes appears to be spurious (Maynard, 1997). That is, if we could postpone the births of teenagers by 5 years, it would have little effect on their life chances because most would not end up going to college merely because their first birth occurred later. Or, when they eventually became pregnant, they would not be married or otherwise capable of supporting their child as a single parent. Paraphrasing Campbell’s assertion, we might say that if a pregnant women in the US is poor, black, and has been a low-achiever in school, 70 percent of her life script may be written for her – whether or not she has a child as a teenager or in her early 20 s. Early childbearing usually does not help young women to succeed, but neither does it cause them to fail, having the crushing impact on their lives that Campbell, and many other social scientists, have hypothesized. The story about the consequences of divorce for children appears to be similar but not yet so clearly documented. In the first place, the evidence on the detrimental effects of divorce on children cannot be so easily disregarded in part because researchers have not yet found as many ways to rule out the influence of unobserved differences. Divorce is the outcome of a process that typically involves protracted parental conflict, economic uncertainty, and unresolved sexual and interpersonal problems. Moreover, it frequently occurs in unions formed by couples with prior emotional, social, and economic problems. Sorting out how much the divorce, per se, adds to or complicates the circumstances of children who are already vulnerable and in conflictridden households, is a challenging problem. Still, there is little doubt that many of the presumed effects of divorce are in fact results of differences that occur prior to the event itself such as poor parenting, marital strife, economic and social circumstances, and the like. If my evaluation of the literature on these two examples is correct, it raises a huge issue for life course researchers than has not been adequately addressed: why are the effects of non-normative events not as large or perhaps as long lasting as life course theory has predicted? In the second part of this paper, I will try to answer this question. In doing so, I will suggest modifying, if not abandoning, the overly simple model that assumed that the

162


timing, scheduling, and sequencing of life course transitions invariably produce powerful and persistent consequences for the course of later life. In its place, I argue for a more intricate theory of when and why nonnormative transitions have lasting effects or are of little consequence over the long term.

COMPLICATING THE PICTURE: THE MODERATION OF LIFE EVENTS OVER TIME The problem posed requires an understanding of why non-normative events do not produce the large and lasting impact that theory predicts? Answering this question requires a close examination of how individuals, families, and social systems react to the occurrence of ill-timed and poorly sequenced transitions such as teenage childbearing or marital disruption. Accordingly, I will draw examples from findings of my own research and the work of others to illustrate why and how our theory needs to be revised to take account of the findings of life course research. Several different reasons might explain the seemingly anomalous findings to which I have referred: First, it is possible that these events only affect a small number of the highly vulnerable individuals and that most individuals are relatively robust and impervious to non-normative transitions. This has sometimes been referred to in the literature as sources of resiliency in developmental psychology. A second possibility, which could be considered a variant of the first, is that there is an inevitable distribution of responses to the occurrence of nonnormative events in any given population: some individuals react adversely while others may actually benefit from the unexpected transition; still others may have little or no response. In other words, a series of conditions and circumstances at the individual and social level moderate the impact of nonnormative transitions. Finally, we might imagine that individuals are widely affected by the event but in time these responses fade much as ripples in a pond diminish with distance from the initial splash. Over time, individuals can take actions to repair or recover from the initial response to a non-normative event, offsetting any potential damage done. In fact, these explanations are not entirely mutually exclusive and may all have a part in explaining why the impact of non-normative transitions is not as large as many believed. Let us consider each in turn.


163

Back in the 1960s, when the first interviews were conducted in the Baltimore study, the then pregnant teenagers and their mothers were asked how they initially reacted when they learned that they were pregnant. Recalling the event, the majority of parents replied that they were extremely angry and upset and their daughters, similarly, reported that they were frightened and distressed. A typical response by the pregnant teens went something like this: ‘‘I was shocked and sad. I knew my mother would be furious. I was really scared of what she would do when she found out.’’ However, when then asked how they now felt (because the first interview occurred several months after they learned that they were pregnant), most replied that they had become reconciled to, if not excited by, the prospect of parenthood in the interim. The fact is that many of the parents, who were angry at first, softened their reactions and most of their daughters, fearful of their parents’ reactions, learned that they would be assisted and supported by their families. By this time, a number of parents summed up their change of heart by saying of their daughters’ circumstances, ‘‘Everyone is entitled to make one mistake.’’ The pregnant teens, in turn, had begun to feel as though they could recover from their misstep by making a greater commitment to their future and the future of their child. Young women, who had or were about to enter marriage, were especially positive by the time of the interview because they believed, as many pregnant teens did at the time, that marriage would put them back on course. However, even the teens that did not contemplate matrimony became more sanguine about their prospects of rectifying their situation. Many learned that they could stay in school, receive special services, or would receive help from the baby’s father. On both psychological and sociological levels, most families mobilized their resources to respond to the impending challenge by a redefinition of the situation. I would not want to claim that this redefinition neutralized the potentially adverse effects of early childbearing, but it does suggest that actors reacted neither passively nor uniformly to early childbearing. Many were able to mobilize in response to becoming pregnant in ways that might mitigate some of the impact. And, apart from their ability to redefine the situation, the young mothers and their parents began to explore a range of different possible adaptations to becoming pregnant, including marriage, returning to school, and the realignment of family roles, to mention but a few. True, the unplanned pregnancy forced a series of unanticipated changes in the life course of the teenagers, but the changes were not always negative. Some adolescents claimed at the time as well as in subsequent interviews

164


that the pregnancy was a turning point in their lives. It made them grow up faster, take themselves more seriously, and realize that they had to take seriously their new responsibilities. In many cases, they reported improved relations with their parents, greater engagement in school, and an enhanced sense of self (see Furstenberg, 1976; Furstenberg, Brook-Gunn, & Morgan, 1987). As an indication of these changes, the interviewer who watched the young mothers from one wave to the next, often commented how much and how quickly they had grown up after they became parents. Of course, considerable variation occurred in the years following the first birth of the teen mothers in both their reactions to parenthood and the level of support provided by families and the surrounding community. Among those who hastily married, some entered seemingly viable unions while others found less suitable or compatible partners. Similarly, for those who remained single, some were able to find their way back to school and eventually into the labor force while others were not. Young mothers who found their way to a special school for pregnant teenagers were especially likely to complete high school in the five years following the birth of the first child. The young mothers used public and private support differently. Some, who received public assistance, made good use of it by returning to school; others devoted themselves to caring for their children until they were of school-age at which point they either went back to school or found employment; still others were unable to mobilize to gain economic independence in the years that followed the birth of their first child. I, along with many other researchers, have devoted considerable attention to explaining the sources of these varied life trajectories of teen mothers after the birth of their first children. Clearly, part of the source of variation lies in individual attributes of the young mothers themselves. Physical and mental health, cognitive skills, attractiveness, motivation, flexibility, and a host of other attributes, no doubt, figured into their ability to respond to their changed circumstances. In the language of developmental psychologists, some mothers were more vulnerable and others more resilient in the face of a new challenge. Families too, possess different levels of resources and support to offer to the young mothers in the way of time, money, guidance, social contacts, and the like. They differed in family-based social capital, the ability to mobilize support for their children and cultural capital, and knowledge about the social world. The families, and the young mothers, were also differently embedded in neighborhoods and communities, and thus had differing access to resources to cope with premature parenthood. My collaborators and I have shown what an important part access to services, programs, mentoring,


165

and the like played in determining successful adjustment in the immediate years following the birth of a first child. These conditions all affected the likelihood of other related transitions such as marriage, return to school and graduation, and most notably, the birth of a second child. The probability of a second birth was much greater for those who married than those who remained single and differed among the single women depending on whether they returned to school or not. While information on individual level psychological resources was limited, clearly, some women were more prone to having subsequent unplanned births than were others, a potent predictor of long-term success. In time, some women curtailed their fertility by sterilization and others became infertile, but higher order births proved to delay or deter the ability to gain economic independence and invest in children already born. In some cases, they led to ruptures with families, who were initially willing to offer assistance but whose patience or resources depleted when demands persisted or, worse yet, increased. The longitudinal evidence suggests that there is no single path to recovery nor is there any point from which recovery becomes irretrievable. To make matters even more complicated, new events such as the formation of a new partnership, the death of a parent, or a changing employment situation continually perturb trajectories for better and for worse. At each interview – and there have been seven in all – we discovered that the circumstances of some mothers would change direction. Moreover, sometimes, gains or losses might occur in economic fortunes without affecting other domains of success such as interpersonal relationships, mental and physical health, or selfdefinitions. The correlation of different indicators of success was only modest, complicating our ability to tell a single story about successful and unsuccessful pathways. In short, it is not easy to point to a single set of strategies that invariably paid off in later life success. Women who married, for example, did not necessarily fare better than those who remained single in part because the marriages were so brittle and were often accompanied by higher rates of fertility. Conversely, seeking public assistance did not prove to be a poor strategy if women made use of welfare to get more schooling. To discriminate successful from unsuccessful trajectories, we have to understand how the strands of the life course are interwoven through a series or chain of small decisions, looking at the intentions of the young mothers as they faced certain choices or encountered different possible pathways. An ongoing and continuous feedback took place between life decisions, how they worked out in reality, and how situations were subsequently defined and interpreted by the young mothers, their parents, their partners, and

166


their children. Without seeing the interplay between the intentions, decisions, and immediate consequences, it is difficult to make sense of the chain of strategies that evolved over time to understand how they worked to produce particular outcomes. Many of our analytic techniques are insensitive to this complex process and most of the data that we collect are not sufficient to carry out an analysis that splices together qualitative accounts (motives and interpretations) and quantitative evidence about decisions and their consequences. Doing so properly requires new ways of examining longitudinal information and new techniques for blending qualitative and quantitative information. Even with the crude analytical techniques that I have employed in the Baltimore Study, it is possible to see the emergence of different patterns of success at mid-life. First, women who married early and were able to forge viable and rewarding unions clearly did better than anyone else in the study, but they were only a tiny minority (under 10 percent of the sample) of the young mothers in the study. Case studies of these women showed that they held different attitudes about marriage, came from families with more stable unions, and possessed a greater array of interpersonal skills. They also married men, often the fathers of their first child, who possessed similar attributes. Thus, marriage confers few benefits unless those who marry possess the resources, attitudes, and skills to make it work out. Lacking some or all of these qualities, most entered a succession of unions that did not work out well and indeed left the young mothers and their offspring worse off than many of their never-married counterparts. Some women rapidly took themselves out of the mating game, choosing to pair with men in short-term sexual relationships or they eschewed men entirely. When these women did so to invest in their children, their offspring frequently did well. Alternatively, when they withdrew from men without economic or family support, the children often fared poorly. Second, women who were able to manage to control their fertility did much better in later life, as did their children. Fertility control, like marriage, required motives, resources, and opportunities. For example, some women curtailed their childbearing by staying out of unions or avoiding relationships with men. Others sought sterilization after they had a first or second child. Only a small number were able to use contraception successfully. To understand the implementation of fertility control, we must look at several different possible routes that the young mothers took as well as considering the fact that some were simply more fecund than were others. In sum, women achieve low fertility in many different ways or fail to do so for several different reasons, such as their inability to manage relationships with


167

men, their poor skills at using contraception, or their religion beliefs that prevented them from seeking abortion. Third, work success, not surprisingly, came to those who returned to school and gained credentials, but other women, who were not terribly successful in gaining schooling, managed to find particular economic niches. Again, we concluded that there was a strong link between educational success and economic independence, but the pathways to educational success and economic independence varied quite a bit among the women in the study. Some returned to school immediately, others went to work and then back to school, and still others only went back to school later in life. Although we do not have enough cases to test the efficacy of these different strategies, it appears that they all seemed to produce positive effects. Unquestionably, the women who had the cognitive, psychological, and social support to gain more schooling and develop labor market skills fared reasonably well compared to those who failed to build human capital. Complications only grow when we try to link the success of the mothers to their offspring. Without a doubt, links exist but they are rarely straightforward. Success in employment, for example, was associated with better outcomes for children, but the correlation was relatively weak. Mothers’ actions on their own behalf sometimes led to improvement in their children’s life chances, but very often they did not because the mothers’ investments in themselves came at the cost of time and involvement in their children. Some of the less successful mothers devoted themselves to their children, putting their hopes in their children’s, rather than their own, careers. This strategy occasionally paid off, at least if we believe some of the accounts of children who succeeded despite growing up while on public assistance. Many of the children when they reached early adulthood reported that their mothers had held up their own experiences as teen parents as a negative example to their children. Their mantra became: ‘‘Don’t follow in my footsteps.’’ Working with data from the Baltimore Study, as I have for almost four decades, has taught me to appreciate how much actors construct, interpret, and make meaning of their actions in ways that have powerful consequences for both themselves and other immediate members of their families. Most of the teen mothers that I followed over time felt that they had become pregnant too early and entered parenthood too soon. Almost all wanted their children to take a different pathway. It is no easy matter, however, to explain why some were able to follow their mothers’ wishes while others became second- (or third-) generation teen mothers. The ability of parents to translate their desires to their children differs greatly depending on their parenting skills, their resources to keep their children on track, the support

168


that they receive by others in their household and community, and of course, the children’s individual and social circumstances. It is one thing to say that lives are inextricably linked across the generations, it is quite another to show exactly how such linkages play out within families. One might conjecture that variability in outcome is greater in the populations most susceptible to non-normative events. No doubt, low-income and marginalized populations are more vulnerable to these events for both structural and perhaps psychological reasons. However, I would argue that the broad contours of what I have reported about the difficulty of predicting trajectories of teen mothers and their offspring apply more broadly to less vulnerable populations. John Clausen (1972) used the term loose coupling to refer to the varied pathways that can be taken when events occur in the life course. Not all researchers who have studied divorce have appreciated Clausen’s observation that casual links between powerful events and their consequences can sometimes be quite faint. Many of the same reasons why teen parenthood does not inevitably have adverse consequences for parents and children apply as well to divorce. Divorce, like teenage childbearing, may have positive as well as negative effects. The most obvious of these is that it often curtails chronic conflict between parents or leads to the removal of a dysfunctional parent from the household. Children may bond more strongly with the residential parent (or even the non-residential parent) when the couple is no longer living together. They may have opportunities to acquire new competencies by assuming greater responsibilities in the household. Of course, none of these possibilities is without potential costs, but researchers often ignore them in the literature on divorce, which assumes that the consequences for children are invariably bad (for an exception, see Barber & Eccles, 1992). Quite apart from the potential benefits that divorce might confer to children, researchers have shown that children in the same family react quite different to marital disruption (Hetherington & Kelly, 2002). Depending on their personal interpretations of the reasons for the divorce, their relationships to the residential and non-residential parents, the reactions of siblings, and a host of other conditions that moderate the impact of marital dissolution, children’s reactions to the event vary enormously. Moreover, parents differ in how they manage the divorce, their capabilities for handing lone parenthood, their economic circumstances, and the amount of support that they can garner from others. All of these conditions affect the way that divorce impinges on a child’s life both immediately and in the long term. The immediate reaction of children to the divorce of their parents does not necessarily signal how they may adapt to the divorce over time, as events


169

that follow the divorce in both the parents’ and children’s lives can overshadow the significance of the marital breakup and change its meaning for children. Suppose, for example, that upon the breakup of parental marriage, children encounter a series of new and unstable family circumstances. Compare their circumstances to the opposite situation where the parents are able to work out a collaborative pattern of parenting and are committed to remaining single. Alternatively, we can imagine that one or another remarries and the stepparent proves to be a capable partner and parent for the child. These different scenarios can lead to very different outcomes and, in turn, influence the child’s definition of the impact of the divorce on their lives. Just as in the case of teenage childbearing, a number of different conditions can moderate the impact of divorce by either heightening or attenuating its long-term consequences for parents and children. The initial ‘‘definition of the situation’’ (that is, the way that it is first experienced) sets in place a response trajectory, but evidence seems to suggest that the initial response is not a reliable gauge of the long-term consequences. And, in any event, more than we frequently recognize, early responses are the product of pre-existing individual differences and social conditions often linked to the nature of the transition; these pre-existing differences are partially or largely responsible for the slope of the response trajectory. Thus, we must take great care to take into account the fact that non-normative events often occur selectively to individuals and families that are already vulnerable. Assessing even their initial impact has proved to be a challenge precisely because it requires sorting out endogenous and exogenous casual factors. Subsequent, events may often reinforce the impact of non-normative transitions, but they often have just the opposite impact: they may offset the initial adverse reaction because individuals continue to respond, react, and adjust to their life situations. In this paper, I have attempted to explain why and how this can happen. Let me conclude by summarizing some of the most important conditions that can redirect response trajectories over time. 1. Individuals increase their motivation and commitment in response to a non-normative event. In other words, they mobilize to counteract the predictable negative impact of a non-normative transition. 2. Others in their social networks, especially family members and friends may increase support and resources to help them recover from the event. 3. Social agents intervene providing services and access to opportunities that might not otherwise have occurred. Policies can be designed to reduce the potentially adverse effect of a non-normative transition.

170


4. The experience of the event increases previously undeveloped personal and social capacities. Non-normative transitions can, and often do, promote development. 5. Subsequent life events offset and compensate for initial disadvantages. Over time, other events and transitions may reverse the fortunes of those who encounter non-normative transitions. This can happen merely from good fortune or because of one or more of the other reasons listed above.

CONCLUSION I have suggested that theorists describing the impact of non-normative events – ill-timed, unscheduled and out-of-sequence transitions – have subscribed to a simplistic and overly deterministic model of how the life course evolves over time. They have not given enough attention to the role of actors in defining and moderating the impact of such transitions nor have they taken adequate account of how a wide range of conditions moderate the impact of non-normative events on subsequent life pathways. Drawing from both my own research and the work of others, I have identified a series of important mechanisms that often moderate the impact of non-normative transitions. Consideration of these mechanisms may help to explain why events such as teenage childbearing and divorce do not cast as long or as dark a shadow in the subsequent life course as is often believed. I have also contended that future research on the life course must take fuller account of how circumstances are defined and responded to by actors as well as examine objective markers of success and failure. Failing to do so means that we ignore the process by which individuals experience the event and hence the ways that they seek out and discover possible corrective actions. The management of life course events is an under-investigated topic that brings together personal agency, social support, and opportunity structures in a common theoretical framework that may improve our understanding of how life courses are constructed and how individual development occurs in response to both normative and non-normative transitions.

REFERENCES Ahrons, C. (2004). We’re still family: what grown children have to say about their parents’ divorce. New York: HarperCollins.


171

Barber, B. L., & Eccles, J. S. (1992). Long-term influence of divorce and single parenting on adolescent family and work-related values, behaviors, and aspirations. Psychological Bulletin, 111, 108–126. Becker, H. S., & Strauss, A. L. (1956). Careers, personality, and adult socialization. American Journal of Sociology, 62(3), 253–263. Brown, S. S., & Eisenberg, L. (Eds) (1995). The best intentions: Unintended pregnancy and the well-being of children and families. Washington, DC: National Institute of Medicine. Campbell, A. A. (1968). The role of family planning in the reduction of poverty. Journal of Marriage and the Family, 30(2), 236–245. Cherlin, A. J. (1999). Going to extremes: Family structure, children’s well-being and social science. Presidential address to the Population Association of America, New York, March 26. Clausen, J. A. (1972). The life course of individuals. In: M. W. Riley, M. Johnson & A. Foner (Eds), Aging and society: A sociology of stratification (pp. 457–514). New York: Russell Sage Foundation. Elder, G. H., Jr. (1984). The children of the great depression: Social change in life experience. Chicago: University of Chicago Press. Elder, G. H., Jr. (1998). The life course and human development. In: R. M. Lerner (Ed.), Handbook of child psychology, Volume 1: Theoretical models of human development (5th ed., pp. 939–991). New York: Wiley. Furstenberg, F. F. (1976). Unplanned parenthood: The social consequences of teenage childbearing. New York: The Free Press. Furstenberg, F. F. (2003). Teenage childbearing as a public issue and a private concern. American Review of Sociology, 29, 23–39. Furstenberg, F. F., Brooks-Gunn, J., & Morgan, S. P. (1987). Adolescent mothers in later life. New York: Cambridge University Press. Furstenberg, F. F., & Cherlin, A. J. (1991). Divided families: What happens to children when parents part. Cambridge, MA: Harvard University Press. Hagestad, G. (1990). Social perspectives on the life course. In: R. H. Binstock & L. K. George (Eds), Handbook of aging and the social sciences (3rd ed., pp. 151–168). New York: Academic Press. Hayes, C. D. (1987). Risking the future: Adolescent sexuality, pregnancy, and childbearing (Vol. I). Washington, DC: National Academy Press. Hetherington, E. M., & Kelly, J. (2002). For better or for worse: Divorce reconsidered. New York: W. W. Norton & Co. Lieberson, S. (1985). Making it count. Berkeley: University of California Press. Marini, M. (1984). Age and sequencing norms in the transition to adulthood. Social Forces, 63, 483–507. Maynard, R. A. (Ed.) (1997). Kids having kids: Economic costs and social consequences of teen pregnancy. Washington, DC: The Urban Institute Press. McLanahan, S., & Sandefur, G. (1996). Growing up with a single parent: What hurts, what helps. Cambridge, MA: Harvard University Press. Merton, R. K. (1968). Social theory and social structure. New York: The Free Press. Modell, J., Furstenberg, F., & Hershberg, T. (1976). Social change and transitions to adulthood in historical perspective. Journal of Family History, 1, 7–32. Moynihan, D. P., Smeeding, T. M., & Rainwater, L. (Eds) (2004). The future of the family. New York: Russell Sage Foundation.

172


Neugarten, B., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints and adult socialization. American Journal of Sociology, 70, 710–717. Parsons, T. (1954). Essays in sociological theory. New York: The Free Press. Riley, M., Johnson, M., & Foner, A. (1972). Aging and society: A sociology of stratification (Vol. 3). New York: Russell Sage Foundation. Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How common and does it matter? American Sociological Review, 52, 785–801. Ryder, N. B. (1965). The cohort as a concept in the study of social change. American Sociological Review, 30, 843–861. Settersten, R. A., Jr. (1998). A time to leave home and a time never to return? Age constraints around the living arrangements of young adults. Social Forces, 76(4), 1373–1400. Waite, L., & Gallagher, M. (2000). The case for marriage: Why married people are happier, healthier, and better off financially. New York: Doubleday. Wallerstein, J. S., & Blakeslee, S. (1989). Second chances: Men, women, and children a decade after divorce. New York: Ticknor & Fields. Wilensky, H. (1961). Orderly careers and social participation. American Sociological Review, 26, 521–539. Winsborough, H. (1979). Changes in the transition to adulthood. In: M. W. Riley (Ed.), Aging from birth to death: interdisciplinary perspective (pp. 137–152). Boulder, CO: Westview Press.

THE SECRET OF TRANSITIONS: THE INTERPLAY OF COMPLEXITY AND REDUCTION IN LIFE COURSE ANALYSIS Katherine Bird and Helga Kru¨ger 1. DEFINING THE ISSUE In a sociological perspective, transitions focus our attention on a segment of the life course in which a biographically significant change of social positioning occurs. This apparently simple statement masks the complications involved, which stem from framing the concept within and between disciplines, within and between methodologies, and within and between life course assumptions. Two recent articles (Elder, Johnson, & Crosnoe, 2003; Marshall & Mueller, 2003) outline the development of the life course perspective from its humble beginnings to an established discipline. Analogous to the continual development of overall life course theories, the perception and conceptualisation of transitions has also diversified. The endeavour has certainly not been completed, since complexity arises on many fronts: on the macro-level, status passages refer to institutional resources and guidelines for attaining a new state (Heinz, 1996), while, on the micro-level, status passages are personally conceptualised and modified by biographical actors. Deciphering the Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 173–194 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10006-9

173

174

KATHERINE BIRD AND HELGA KRU¨GER

occurrence, timing and variability of transitions (cf. Hogan, 1981; Rindfuss, Swicegood, & Rosenfeld, 1987) reveals nation-specific diversity, and so does the explanation of how transitions link institutions and actors by defining timetables, and entry or exit markers. And when Elder (1998, p. 958) establishes ‘‘transitions are always embedded in trajectories that give them distinctive form and meaning’’, he draws attention to another level of complexity: the arbitrariness of their interpretation in both the interdependencies of status accumulation and the different perspectives related to internal (biographical and personal) success stories and external evaluations. Given the extent of these problems within an emerging discipline, it is no wonder that more often than not in the literature on transitions we encounter pragmatic strategies associated with specific theoretical objectives. Examples include: 1. To draw attention to the course and prolongation of transition processes before the subsequent status is reached, which calls to mind the controversies about individualisation, such as the loosening of institutional effects on biographies (Beck, Giddens, & Lash, 1996), or just the expansion of periods of personal uncertainty about status passages from one institutional grip to the next (Heinz, 1999; Blossfeld, 1985). 2. To consider the biographical timing of events, an endeavour that has been greatly enriched by approaches to the interlacing of biological and social clocks as internalised external gatekeepers for windows of opportunity, and calls for the investigation of multiple temporal dimensions as frames of reference (Settersten, 2003; Settersten & Mayer, 1997; Mortimer, Oesterle, & Kru¨ger, 2004). 3. To focus on the often neglected socio-personal framing of transitions and stages by the standardising forces of transitions in one’s partner’s life which afflict your own life (Moen, 2003a; O’Rand & Farkas, 2002; Sørensen, 2004). Hagestad’s conceptualisation of durations and transitions (Hagestad, 1992, see also Hagestad & Neugarten, 1985) deciphers even ‘countertransitions’ in order to underline the occurrence of status changes arising from transitions on the part of socially relevant associates, such as being transported into grandmotherhood by a daughter’s giving birth, or being deprived of your marital status by your partner’s death. Reduction of transitions occurs by means of theoretical assumptions and the correspondent definitions of transition markers as well as by means of the postulated empirical abundance of the segment under scrutiny. On the basis of the theoretical approaches outlined above we can identify ‘inline

The Secret of Transitions

175

transitions’; for example, the analysis of transitions with respect to one life course domain (see example 1 above), in contrast to ‘competing transitions’; for example, the analysis of context norms and biographical opportunities (see example 2) – very often relevant in female decisions in favour of or against motherhood – and ‘coupled transitions’; for example, transitions in linked lives (see example 3). Reduction occurs methodologically by means of the data collection and analysis tools employed. In this context, it is important to point out that from the event history analysis perspective – and this applies easily to demographers as well – a transition is minimised by defining it as the occurrence of a precise event, such as leaving school. Thus the terms ‘transition’ and ‘event’ become synonyms. The other extreme is to maximise transitions into changes over a specific duration in life, such as the transition from childhood to adolescence, which opens the door for broader developmental perspectives and associated scholars. Between the two extremes are transitions of much shorter duration, such as from company manager to being unemployed, or from secretary to housewife. And although Elder points out that every transition can be broken down into ‘‘a succession of mini-transitions or choice-points’’ (1998, p. 958), we suggest moving away from analysing the ordering of transitions in a linear fashion (cf. Sackmann & Wingens, 2001) and to reconsidering complexity instead. This endeavour leads us to concentrate on transitions within and between domains of social participation. Given that the individual’s simultaneous participation in other social domains will both affect the transition under study and be affected by it; we wish to advance Levy’s very early suggestion (Levy, 1977, further developed in 1997) to conceive of the life course as a movement through varying participation profiles. The fact that pathways are layered within life domains introduces personal and structural contingency into transitions, which can easily be overlooked. By seeking to reduce complexity, we might accentuate the education, employment and retirement history, and simultaneously neglect transitions within the dynamics of family life, or within the health and illness history of a life. However, each of these histories are subjected to normative and institutional transition programmes that themselves follow a logic specific to that domain of life, and which ‘dissect’ the life course into relatively separate but still interwoven patterns of behaviour. Even in the case of apparently discrete transitions, the analysis might have to take into account the social interplay of change and continuity with respect to other life domains, such as, working while still at school (Mortimer, 2003), or, more formally, undergoing vocational

176


training as dual participation (Kru¨ger, 1999), or being at work while retired (Moen, 2003b). Individuals experience transitions as governed by the different logics of distinct life domains, and it is the interplay of life domains within transitions that attracted our interest. Against this background, we propose that there are ‘interlacing transitions’ that require attention. If we assume that a transition in one life domain is interlaced with simultaneous participation processes in other life domains, some of these abide by socially defined criteria of mutual exclusion producing contradictory forces that push or pull in different directions. We argue that we need to look deeper and further in order to disentangle the various strands of structure and agency that underlie a specific life course arrangement. Introducing the category ‘gender’ facilitates the analysis of interlacing transition processes. We have therefore decided to focus on women’s lives. We will first outline the theoretical basis as their dual integration into society via the labour market and the family to provide a starting point for unravelling the dynamics of two strands of the life course, which, in countries with a strong life course regime like Germany, produce gender inequality (Kru¨ger, 2003). The empirical data presented in the third section offer a new look at the classical transition from employment to homemaking on the birth of a child. The change of perspective reveals mainstream status subsumption processes, which mask the different continuity demands of distinct life domains. The second set of empirical findings focuses on the example of official and actual marital status to examine the dissonance between institutional status definitions and the reality of lone motherhood. In the final section, we draw some conclusions that challenge sociologists to reconsider the social structuration of life as well as the adequacy of our analytical tools for capturing the hierarchical dependency of transitions between different social domains.

2. LAYERED LIFE COURSE PATTERNS: THEORETICAL ADVANCES ON THE FEMALE CASE The simultaneous relevance of more than one life course domain in women’s lives was highlighted 18 years ago by Regina Becker-Schmidt, who coined the phrase ‘dual integration into society’ (doppelte Vergesellschaftung, Becker-Schmidt, 1987). She was referring to the contradictions inherent in women’s socialisation and participation in society as both a future worker


177

and as a caregiver. Much has been said about the objective problems of seeking to combine these two tasks, especially in terms of timetables: for kindergarten, school, paid work and so forth. The qualitative research conducted by Becker-Schmidt and her colleagues also revealed the subjective problems associated with attempting to solve this balancing act: the women studied were ambivalent towards both types of social participation. They could neither say that paid work was the most important thing in their lives nor were their families. To quote the title of the book: ‘‘One is not enough – both are too much’’ (translation of the original title: ‘‘Eines ist zu wenig – beides ist zu viel’’, Becker-Schmidt, Knapp, & Schmidt, 1984). The significance of these findings was to underline how separate fields of social participation are simultaneously relevant in individual life courses, but framed in a contradictory manner. Although Becker-Schmidt and others have made an important contribution to understanding the complexity of women’s lives, their ‘dual integration into society’ and the ensuing ambivalence is, in most life course approaches, ignored in favour of conflictreducing simplicity. Entanglements are eliminated by defining one possible life domain as dominant for each sex and thus blending out all others. The ‘economic theory of the family’ suggests that the primary function of the family is best achieved by a gendered division of labour between a male breadwinner and a female caregiver (for a concise summary of this approach see Blossfeld & Drobnicˇ, 2001).1 Catherine Hakim (2001) has developed the ‘preference theory’, which turned the conflicting demands of paid and family work into an orientational concept of a ‘‘personal work-lifestyle preference’’ (Hakim, 2001, p. 16).2 Following in the early life course tradition of comparing the timing of events across birth cohorts, Karl Ulrich Mayer indirectly supported this view by stating in 1998 that on the basis of indicators such as age at marriage or at first birth, or the family dependent phasing of employment, female life courses were ‘‘highly standardised and relatively homogenous, but – with the exception of marriage – not institutionalised and differentiated’’ (Mayer, 1998, p. 447). The reduction of structural ambivalence in women’s processes of societal integration to ‘rational choice’, ‘preferences’, or the statement of a lack of institutionalisation and differentiation in female life courses per se prompts the question of whether these presumptions primarily result from a theoretical and methodological artefact relating to the underestimation of the interlacing of transitional regulations in the two life domains of the labour market and the family. To examine this question further we will explore some empirical findings concerning two family related transitional segments

178


in female life courses: the transition to motherhood and the transition to lone parenthood.

3. EMPIRICAL FINDINGS 3.1. Interlacing Transition I: A Baby Break The event ‘giving birth’ undoubtedly marks a turning point in patterns of female employment and family life, but is usually reduced to just the occurrence of a series of labour market exits and re-entries.3 If we move away from a construction of the life course as a sequence of discrete transitions towards a conceptualisation of interlacing transitions involving changing participation profiles in the juxtaposed layers of life domains, we uncover societal transition programmes that seek to subsume specific life course domains under their own particular logic. In searching for the strands relevant to transitions around motherhood, we therefore decided to integrate – in addition to labour market exits and re-entries – the instruments developed by the state for shaping labour market ‘timeouts’ on the occurrence of the ‘event’ of giving birth, and thereby incorporated the ‘negative space’ occupied by leave-taking rules into the analysis. This broader view allows us to consider three historical forces with opposing or contradictory objectives that may act out their effects on ‘preferences’. These are (a) The wide-scale expansion of part-time jobs that was intended to pull mothers back into paid employment, whereas (b) occupational career structures push them out. In addition, the introduction and extension of extended periods of (c) maternity and parental leave will pull and push them for sequenced time spaces into one domain or the other. To investigate the effects of the three life course structuring principles we collected life course data on 2,130 West German women from three apprenticeship-completion cohorts: 1960, 1970 and 1980.4 At these dates the women were aged between 18 and 20 years and had just completed training in one of the top 10 occupations for women in Germany.5 During the window of observation for these three cohorts, three types of maternity or parental leave were in force. Initially, new mothers were forbidden from working shortly before and after a birth under the motherhood protection law (Mutterschutz of 8 weeks). In 1979, a motherhood leave (Mutterschaftsurlaub, literally ‘motherhood vacation’) of 6 months was introduced. The extension beyond 8 weeks was optional, as was the longer ‘child-rearing leave’ (Erziehungsurlaub, literally ‘child-rearing vacation’)


179

introduced in 1986. Initially lasting for 10 months, child-rearing leave was extended in a stepwise fashion to 3 years in 1992 and could be taken by mothers or fathers. With the introduction of this last measure, new opportunities for configuring the duration of leave and changing gender-specific participation were established. For the women studied, the median duration between completing training and the birth of the first child varied between 5 and 8.5 years, depending on occupation and cohort. To capture the effects of the different leave regulations, the mothers were regrouped into ‘motherhood cohorts’ according to which type of leave they could take when they gave birth: the motherhood protection cohort, the motherhood leave cohort and the child-rearing leave cohort. During the period covered by the study, nearly no change was observed in the integrational capacity of the labour market segments open to people trained in the 10 occupations investigated. However, the expansion of parttime working did cause considerable change, which was accompanied by shifts in the cultural contexts of women’s lives. For the oldest women in the sample, the dominant norm was that of the stay-at-home wife and mother (as Hradil put it in 1992: The ‘Golden Times of Normality in Family Life’. See also Born, Kru¨ger, & Lorenz-Mayer, 1996). Gradually, this norm transformed into acceptance of employment during motherhood, especially part-time, accompanied by the official recognition in the state’s statistics of the true extent of mothers’ paid work assisting in the family business (Willms-Herget, 1985). The then empirically proven younger cohorts’ visibly higher employment participation fitted well into individualisation theories that highlighted how, as old ties that anchor females firmly into a specific position in family life have dissolved, diversity and discontinuity both within individual life courses and between gendered life courses in general have increased (Beck & Beck-Gernsheim, 1994). In spite of these well-known interpretations, by looking only at women’s actual leave-taking behaviour,6 we found that with the new options for lengthening and sharing parental leave, female life course patterns became more, not less, standardised – and they followed occupational career patterns, not part-time ones. The proportion of mothers who stopped working (that is, took leave or quit their job) within 6 months of giving birth varied according to both motherhood cohort and to occupational career patterns, as shown in Fig. 1. The ‘stopped working’ percentage shows that the motherhood protection cohort (black bars) not only did not stop work as much as normatively expected – normative frames would lead us to expect nearer to 80% at least


180

Proportion of mothers stopping work (%)

100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Nurse

Bank

Office

Ind. office Hairdresser

Dr. asst

Wholesale Spec. sales

Retail

Hotel

Occupation trained for Mothers' protection

Fig. 1.

Motherhood leave

Child-rearing leave

Percentage of Mothers Stopping Work within 6 Months of Giving Birth to their First Child by First Occupation and Motherhood Cohort.

(see Wurzbacher, 1951; Myrdal & Klein, 1956/1968; Pross, 1976) – but in addition, they did so in line with the specific opportunity structures associated with the occupation trained for. The inclusion of the other two motherhood cohorts highlights how – as the length of leave available was extended – an increasing proportion of women stopped working when they had a child, but again in the same occupationally specific manner. It should be noted that in the younger cohorts, leave taking was optional and not compulsory. But we can see that not just the birth itself but the state’s framing of this event has encouraged women to realise this option at this particular time. The frequency of the transition from paid employment to family work within 6 months of giving birth has in fact increased – in spite of the assumption of the younger cohorts’ higher commitment to paid work. Additional event history analyses of our data on the rate of transition from employment to a family break on the birth of a child confirmed that changing leave entitlements have significantly (at the 1% level) increased the likelihood of a labour market exit for mothers in the younger cohort, whereas those not entitled to extended leave were least likely to make this transition (Bird, 2004). Other data also support this finding. Beckmann and Kurtz (2001), for example, report that 84% of West German women took advantage of longer leave regulations.


181

These findings contrast with Mayer’s (1998) assumptions about the lack of institutional effects on transitions in women’s lives, as we clearly discern two equally powerful institutional programmes at work – leave regulations and occupational specifics – which contradict each other, but both also disprove the effects of an assumed lower or higher commitment to paid employment. How can we conceptualise these facts? (a) From status definitions to ambiguity. An event within one strand of the life course (such as giving birth) can be a catalyst for a labour market exit, but is also interwoven with a second strand – the state’s regulation of maternity leave. But the second strand (state regulations) used to remain hidden or its effects were underestimated, since many of these exits are masked if only the event of giving birth is related to the mother’s labour market status. In Germany, the institutionalised participation in the labour market does not formally change while on leave, since leave-takers still enjoy the status of being employed.7 If the resulting status subsumption is analysed with a method that permits only one activity at a time (such as event history analysis), the conclusions drawn will be inaccurate. If, however, we consider the mothers’ factual participation profiles rather than their formal employment status, in the younger cohorts we actually find an increased proportion of stay-at-home mothers dedicated to family work and not to paid employment. (b) From events to echo effects. The state’s regulation of parental leave produces a time frame for the transition back to employment, which impacts on the layered female life trajectory. With the exception of the small percentage of women who did not take advantage of prolonged exit opportunities, our data revealed the high degree of conformity between length of leave available and length of leave taken (Bird, 2004).8 These findings raise questions for both psychologists and sociologists concerning the meaning of a longer break for the early period of motherhood. We can assume that an extended absence from the labour market affects the mothers’ perception and memory of the early period of motherhood. It may provide the feeling of having now shared enough time with the baby, so that the mothers can return to paid work without a bad conscience, or such an absence could lead to increased insecurity about a return to the public sphere in general, or influence their confidence in their ability to perform well at work. Research is also required into how mothers personally anticipate reconciling work and the family as the child grows older (see the timing of debates on rising ‘overprotection’ in mothering, such as Schu¨tze, 1986). Qualitative data collection could provide illuminating answers to these questions if the women in the oldest cohort who took a 3-year break under

182


the motherhood protection law were compared with the youngest mothers who took a 3-year child-rearing leave. More generally, the factual masking of the interlacing transitions between paid and family work has meant that a whole range of important research questions have been neglected. (c) From labour market to occupational structure. In all three cohorts the mothers’ behaviour follows an occupational pattern. Hence, a third strand in this transition is, undoubtedly, the first occupation trained for.9 In the analysis illustrated here, the hairdressers stopped work far less often than, for example, the bank or office employees. Another interesting finding on the life course structuring effect of the first occupation was that the fertility rates also varied between the women who trained for different occupations. The explanation for these phenomena is to be found in the market rules (labour supply) and normality assumptions (including age norms) for the labour force in the German service sector (for more details see Born et al., 1996).10 Occupational specifics are reflected in whether or not women had children and whether they stopped working when they did so. Their occupations negate the rules of cultural change in the acceptance of mothers’ employment. Over the different cohorts, women’s behaviour is dominated by the same occupational patterns, and so are the effects of the motherhood transitions on the long-term female employment outcomes (Bird & Gottschall, 2003). All these findings are undoubtedly evidence of the impact of institutionalisation programmes in female life courses and equally refute the importance of preferences in Hakim’s sense. In summary, our analysis highlights that it is important to expand the depth of focus – in this case to discover the overlapping of the state’s mediating regulations with the peculiarities of the occupation trained for. This implies overcoming or questioning official employment status definitions and identifying the interwoven strands of a ‘setting in change’, also including psychological processes, and to investigate their influence on further transitions around labour market exits and entries. The remarks on the need to question official status definitions that reduce complexity also holds for our second empirical example, although it refers to another transition in a setting in change.

3.2. Interlacing Transition II: The overlapping realities of marital status and lone motherhood The following section focuses on motherhood and marital status, which are the classical transitions for describing change in female life courses across

183


time (see, for examples, Huinink, 1991; To¨lke, 1989; Mayer, 1998). In Germany per official social welfare law (and transportation into everyday life knowledge as well as into life course research), the legal status ‘married’ excludes the co-status of lone parenthood, which being single, divorced or widowed would admit. In order to investigate the category ‘marital status’ in its relevance for exits or entries into lone motherhood, our research project took a two-fold approach. In the standardised questionnaire, we first asked the respondents for dates of marriage, divorce, widowhood, and several pages later we offered a life-history calendar which opened up the chance to mark time slots in simultaneously occupied fields in their lives (see Note 4). We compared the durations of lone motherhood with the data on marital status (Erzberger, 2001a). A total of 261 women provided apparently contradictory information on their marriage and lone motherhood histories. An analysis of these cases identified four combinations of both (illustrated in Fig. 2), all consisting of somewhat confusing participational patterns between the beginning and end of periods of being married and being a lone parent (i.e., solo). The chart is a symbolic representation of the arrangements of layered periods found in the data; the length of the blocks symbolises the overlap of statuses in question rather than the actual durations in each. The vertical lines represent the event of official change in marital status. The example

Type 1

Type 2

Married

Married

Solo

Solo

Type 3

Type 4

Married

Married

Solo

Solo

Fig. 2.

Correspondence of Marital Status and Lone Motherhood.

184


illustrates vividly the lack of synchronicity in interwoven strands of the life course and poses a challenge for the classification of these phenomena: In Type 1, the phase as a lone mother starts before the marriage ended, and ends before a new marriage is recorded. This pattern was found for 114 or 44% of the lone mothers. The average time spent as a lone mother while still married and not married again was 2 years and 2 months. If we follow the traditional logic of transitions, the question arises of which event ‘causes’ lone parenthood and which one ends it. The law would say: from the date of divorce until the date of a new marriage. This clearly does not reflect the empirical reality of women’s lives. If we consider lone parenthood to start with the separation of the partners, then we are still confronted with the question of when does separation start: When the partner moves out, when the decision to move out is made, or when the first irreconcilable differences emerge? And then, the same questions arise with respect to the event of ending lone motherhood. Type 2 is even more confusing. The phase of lone motherhood occurs during the marriage. Such a pattern even occurs fairly often; it was observed for 72 mothers (28% of the sample). How does this situation arise? One of the project members suggested that the husband might be absent for a time because of his work, perhaps as a captain on a ship. Another member suggested that he could have been in prison. A third proposed that a married woman separates from her husband and becomes a lone mother, then later she moves in with a new partner and so ceases to be a lone parent, but is still married to the first man. The interpretation of the two remaining types (marriage as an unsynchronised interlude in lone motherhood or continuing lone motherhood after marriage) also results in unsatisfactory explanations that we do not need to expand on here. We maintain that it is impossible to decide which of the many plausible explanation is in fact the correct one, and whether only one explanation would be valid for all the women in a group. The important lesson to learn from this example is that if we use marital status as a proxy for lone parenthood, then analyses based on this assumption will be inaccurate. Similar to our empirical findings on the interlacing transition to motherhood, other data from our previous qualitative studies within the Bremen Life Course Centre revealed many more important transitions in women’s lives, which again are swept over by the dominance of analytical conventions. Examples include: Women who left paid employment in order to reduce stress factors and so better their chances of becoming pregnant, but it did not happen. In


185

this case, a transition had been made (employment to homemaking) but the anticipated event (birth) had not occurred. Women who returned to employment to be financially independent in anticipation of a divorce, but it did not happen. Again others who left paid employment, because their child was doing badly at school, but no visible event emerges. Evidently, transitions occur in many cases, but they often cannot be associated with a visibly corresponding explanatory event, which influences the perception of transitions in sociological research. If the analysis of transitions is minimised to the study of events, then we end up shooting wide of the mark. The empirical examples presented here have sought to clarify what is meant by the term ‘interlacing transition’. In the first example, status subsumption masked empirical reality. In the second example, using an official definition for an event masked its occurrence or non-occurrence. Both examples show that the reduction of complexity in female life courses can easily seduce researchers into superficial analyses and to working with misleading causal assumptions about the logic of transitions within institutional constraints. Therefore, instead of seeking to reduce complexity we have focused on areas where complexity has been reduced to such an extent that a distorted perception of reality has emerged.

4. DISCUSSION: INTERLACING TRANSITIONS – WEAVING THE STRANDS OF A LIFE COURSE Transitions involve structural interventions in the shaping of the life course and precipitate reorganisation processes affecting the self within layered frames of social order. Many authors have made statements with respect to the timing and sequencing of transitions, to their relation to trajectories, and to gatekeeping practices of institutions that provide standardisations and normality assumptions about which transition is expected, which regulations are to be followed, and how goals are personally and socially combined (for an overview see Mortimer & Shanahan, 2003; Heinz & Marshall, 2003). Although it is an indispensable goal of sociological theory to reduce the complexity of individual lives to a manageable explanation of human life course patterns, simplification should not be taken so far that the distinguishing features of lives lived simultaneously in several social fields are eclipsed. To draw attention to the interweaving of different strands of social participation we introduced the term ‘interlacing transition’. The reintroduction

186


of the necessary intricacy has implications for a multi-dimensional study of transitions. We have to take into account: 1. Theories shape awareness and define timeframes. Transitions are most appropriate for probing the multi-layered facets of the life course. But it depends on our basic assumptions about life course patterns whether we consider it worthwhile to devote sufficient attention to the turning points that we discover and permit a large enough window of observation for sounding out their depths. As the empirical examples showed: Regarding the birth of a child as the core of an interlacing transition reveals the regulations and expectations for behaviour to be found in the working environment (occupational specifics), the legal regulation of paid employment (leave regulations) and social norms (cultural role models for motherhood). These deeper layers of the transition are often overlooked because the theoretical treatment of simultaneously relevant participation domains has been underdeveloped. A foreshortened perception of the transition to lone motherhood that only focuses on official status is not able to reveal the variation in real experience found if we consider this to be an interlacing transition. Enlarging the window of observation beyond the moment of change in official status aids the identification of the topology of the transition and may indicate social change. Other researchers, primarily in North America, are following a similar strategy. Phyllis Moen (2003b) and her associates (Han & Moen, 2001) are studying retirement as an interlacing transition. The strands that are relevant to the timing and form of retirement include career pathways and the opportunities and restrictions presented by living with a partner. As Moen (2003b, p. 269) pointed out, ‘‘[the transition to retirement] is a process embedded in a number of overlapping contexts’’. 2. Interlacing transitions can reframe contemporary discourse. Searching for and following the strands of an interlacing transition is a good guide for sociological life course research to expose the contradictory forces of the different societal programmes that can point the life course in different directions. The ambiguity and contradictions inherent in the female life course offer ample opportunities for identifying the interweaving of different social domains. The lack of a theoretical framework for managing the contradictory forces produced by interlacing transitions has meant that, at least in


187

Germany, the analysis of male life courses might have been oversimplified. The increase in more flexible working practices and the emerging ideal of the ‘‘new father’’ indicate that the contradictory strands in women’s lives are increasingly finding their counterpart in men’s lives. A new framework for analysing the issue of reconciling work and the family, for example, may facilitate its perception as a problem that involves mothers and fathers equally, along with their employers and the legal opportunities and restrictions imposed by the state. 3. The role of the state. The perception of interlacing transitions underlines Mayer and Mu¨ller’s (1986) early proposal that, if we are to fathom the significance of transitions for the shape of the life course, then – at least in countries with a high degree of life course standardisation – we need an analysis that incorporates the role of the state. They elaborated the effects of state-run social insurance introduced to protect the individual worker against the risks of work-related accidents, illness, unemployment and old age. Other researchers have focused on the state’s intervention in the life course by structuring the education, employment and retirement trajectories (Kohli, 1985; Blossfeld, 1985; Heinz, 1999). The state’s power to anchor status definitions in legislation and in formal distinctions (in certificates and documents) should not be underestimated. But this is not the end of the story. We can expand on these previous analyses to conceptualise the state as an important intervening power in shaping interlaced transitions in the family and employment trajectories. On the other hand, the second empirical example on lone motherhood highlighted the inadequacy of relying solely on official status definitions to determine when and whether co-transitions in family life occur. Even though the state can frame a specific transition (both normatively and administratively), we should not, without reason, assume conformity of individual action. As Elder et al. (2003, p. 8) pointed out ‘‘Individuals choose the paths they follow, yet choices are always constrained by the opportunities structured by social institutions and culture’’. Too rigid a focus on the dimensions of state action can mask the processual character of transitions or blend out turning points that should be considered transitions, but have not yet found their appropriate place in theories of the life course. 4. Re-framing the instruments of transition analysis. We can only capture the contingent and multi-layered nature of interlacing transitions if we use empirical tools that are more sensitive to complexity.

188


The collection of quantitative data needs to be reconsidered with respect to the extent to which a questionnaire permits the recording of synchronicity within different layers of the life course. Analogously, the analysis of these data needs to include opportunities for reversing the practice of subsuming simultaneous status positions under one dominant one. In classic event history analysis, using large data sets such as the German Socio-Economic Panel (SOEP), the state space is defined by the researcher as consisting of mutually exclusive statuses. Discrete transitions from one state to another can be analysed, but not the attainment of the target state without leaving the origin state. Although such methods of data reduction certainly facilitate its analysis, we would argue that this approach not only masks multi-layered activities, but also defines partial transitions as of secondary importance for the timing and ordering of events.11 In order to overcome the temptation to reductionism inherent in this method, additional methods of data analysis are necessary to assess the long-term distribution of non-discrete participation profiles. Explorative techniques such as optimal matching of large-scale data sets (Abbott, 1995; Erzberger, 2001b; Aisenbery, 2000) are extremely useful in mapping out the actual sequence of activities during a specific timeframe and thus also offer a means to compare the layeredness of life sequences with simple event patterns (Erzberger & Kluge, 2000; Erzberger, 2001b). Even within an event history analysis, exploratory analyses can facilitate the development of a parametric model (Wu, 2003). In interpreting the results of either explorative analyses or parametric modelling, however, the limitations of a quantitative approach become apparent. The relationship between quantitative and qualitative research is more than a simple either-or (cf. Kluge & Kelle, 2001). Without the contribution of our previous qualitative research in uncovering the variety and steering forces in female life courses we would have missed the importance of interlacing transitions in women’s lives, and the discovery of new questions to be studied. In addition, it would have been much harder to construct an instrument for quantitative research that did not force women to subsume the layers of their lives into just one, but to expand on events and durations in juxtaposed time spans. 5. Transitions: Definitions in progress. We believe it to be self-evident that the conceptualisation of the term ‘transition’ should not lead to a search for a formal definition, but is an ongoing process being pushed forward by the interaction of theoretical and methodological advances. As we learn more

189


about how to study and understand lives, we automatically move away from the simple guillotine-like perception of transitions. Theories of the life course based on a reduction of complexity have certainly been useful in mapping out a rough plan. However, they can lead to the ‘disappearance’ of specific transitions resulting in an oversimplification that could explain the suggestion that the female life course is ‘‘not institutionalised and differentiated’’ (Mayer, 1998, p. 224), or Hakim’s attribution of women’s decision-making to wide-reaching preferences for either employment, or family work (or both just part-time). We can agree with Glen Elder (1985, p. 31) who prefers the term ‘‘transition’’ to ‘‘life event’’ because changes in the life course are not just sudden events but are embedded in interceding, preceding and succeeding processes. However, Elder’s reasoning stops just short of the mark because it does not consider how an interlacing transition implies simultaneous changes of social positioning in other layers of the life course. An event in one strand may have repercussions for the others, which themselves are continuously pushing or pulling the life course in specific directions. The proposal to perceive transitions as the interlacing of transformation processes in participation patterns and thinking in terms of status configurations rather than status changes will, hopefully, lead to more precise empirical and theoretical approaches that will consign oversimplifications to the past.

NOTES 1. In a recent edited volume comparing transitions between paid and family work in an international context, the authors referred to the economic theory of the family as ‘‘the most highly developed and influential theory in this field so far available’’ (Blossfeld & Drobnicˇ, 2001, p. 16). 2. For an extensive critique of this theory see, for example, Crompton and Harris (1999). 3. This can lead to neglecting the partner’s labour market opportunities, an oversight that is being corrected by, for example, studies on couples’ careers (Blossfeld & Drobnicˇ, 2001; Elder, 1996; Marx-Ferree, 1997); on the pathways into retirement (Moen, 2003b; Han & Moen, 2001); or simply the husband’s income which, as our study of older women revealed, provided the best argument in hindering women’s intentions to leave or return to paid employment (Born et al., 1996). 4. The research was conducted within the Special Research Program ‘Status Passages and Risks in the Life Course’ at the Bremen Life Course Centre and was financed by the Deutsche Forschungsgemeinschaft. The survey was restricted to West Germany because the structure and normative context of the labour market for

190


women in East Germany was too different during the historical period investigated to allow a meaningful cohort comparison. The data were gathered at the end of 1997 with a life-history calendar that was designed to find out what the women actually did, rather than their official status at any particular point in time. For this reason we asked them to record all their activities at different times, rather than making them decide which was their principal activity. By asking about different types and hours of employment, motherhood leave and family breaks, and time spent supporting and caring for dependent family members, it was possible to gather information on parallel layers of activity over time. Interestingly, there were many women who stated that their family activity continued long after they had returned to work, confirming the lasting importance of dual social participation, while at the same time highlighting how fallacious it is to ask women to only name their principal activity. 5. The occupations were nurse, qualified bank employee, trained office employee, industrial office employee, hairdresser, doctor’s clerical and medical assistant, office employee in a wholesale and/or export firm, specialist sales assistant (in a baker’s or butcher’s), general retail sales assistant and hotel receptionist. In all three cohorts, between 65% and 76% of all female apprentices trained for one of these occupations. 6. Men who have made real use of the new opportunity to take child-rearing leave do not require further consideration here, because between 1986 and 2001 no more than 2% of eligible fathers took leave (Schneider & Rost, 1998; BMFSFJ, 2002). 7. Women (or men) who take child-rearing leave are now entitled to 3 years of leave, or even longer if another child is born before leave ends. These women still have an employment contract and the guarantee of a job to return to. So, formally, these women are counted as employed in statistics used to calculate the employment rate. Although the Statistisches Bundesamt has since 1996 made a distinction between people who are employed and those who are on temporary leave (including childrearing leave), this distinction is rarely included in calculations of the female employment rate. If no account is taken of temporary leave, then 48% of mothers with at least one child under three are in paid employment. If those on leave are excluded, the employment rate sinks to 30% (all figures taken from Beckmann, 2003, p. 6). 8. Other research (Bu¨chel & SpieX, 2002; Ondrich, SpieX, & Yang, 2002; Beckmann & Engelbrech, 2001) has shown that highly qualified women and those who grew up in East Germany were more likely than other women to return to work before the end of their leave entitlement. 9. Not only did the women train in this occupation, they also worked in it. After completing training, 85% of the women worked full-time in this occupation. 10. The normality assumptions can have particularly dramatic effects. In the 1960s a norm of celibacy for nurses was still prevalent, which at least partially accounts for the high proportion of nurses without children in this cohort (33%) compared to the subsequent one (19%). Similarly, Cremer (1984) showed how the expectation that hairdressers are young and attractive made it increasingly difficult for mature women to remain in the occupation. 11. Analyses of large data sets with sophisticated methods run the danger of degenerating into what Hartmut Esser termed ‘variable sociology’ (Esser, 1996). Although the variation between the variables may be statistically explained, the processes in the life courses that generated them are reduced to possibly misleading causal assumptions.

191


REFERENCES Abbott, A. (1995). Sequence analysis: New methods for old ideas. Annual Review of Sociology, 21, 93–113. Aisenbery, S. (2000). Optimal matching analysis. Opladen: Leske + Budrich. Beck, U., & Beck-Gernsheim, E. (Eds) (1994). Riskante Freiheiten. Individualisierung in modernen Gesellschaften. Frankfurt am Main: Edition Suhrkamp. Beck, U., Giddens, A., & Lash, S. (1996). Reflexive Modernisierung. Eine Kontroverse. Frankfurt: Suhrkamp. Becker-Schmidt, R. (1987). Die doppelte Vergesellschaftung – die doppelte Unterdru¨ckung: Besonderheiten der Frauenforschung in den Sozialwissenschaften. In: L. Unterkircher & I. Wagner (Eds), Die andere Ha¨lfte der Gesellschaf, (Vol. 1, pp. 9–25). Wein: Verlag des O¨sterreichischen Gewerkschaftsbundes. Becker-Schmidt, R., Knapp, G.-A., & Schmidt, B. (1984). Eines ist zuwenig – beides ist zu viel. Erfahrungen von Arbeiterfrauen zwischen Familie und Fabrik. Bonn: Verlag Neue Gesellschaft. Beckmann, P. (2003). Die Bescha¨ftigungsquote: – (k)ein guter Indikator fu¨r die Erwerbsta¨tigkeit von Frauen? IAB-Kurzbericht, 11/2003. Retrieved August 1, 2003 from http:// doku.iab.de/grauepap/2003/kb1103_langfassung.pdf Beckmann, P., & Engelbrech, G. (2001). Die schwierige Balance: Frauen zwischen Beruf und Familie. Personalfu¨hrung, 6, 118–129. Beckmann, P., & Kurtz, B. (2001). Die Betreuung der Kinder ist der Schlu¨ssel. IAB-Kurzbericht 10. Nu¨rnberg: Bundesanstalt fu¨r Arbeit. Bird, K. (2004). Reconciling work and the family: The impact of parental leave policies and occupation on the female life course. Frankfurt: Peter Lang. Bird, K., & Gottschall, K. (2003). Erosion of the male-breadwinner model? Female labormarket participation and family-leave policies in Germany. In: H. Gottfried & L. Reese (Eds), Equity in the workplace: Gendering workplace policy analysis (pp. 281–308). Lanham, MD: Lexington Books. Blossfeld, H.-P. (1985). Bildungsexpansion und Berufschancen. Empirische Analysen zur Lage der Berufsanfa¨nger in der Bundesrepublik. Frankfurt/New York: Campus-Verlag. Blossfeld, H.-P., & Drobnicˇ, S. (2001). Theoretical perspectives on couples’ careers. In: H. P. Blossfeld & S. Drobnicˇ (Eds), Careers of couples in contemporary society. From male breadwinner to dual earner families (pp. 16–50). Oxford: Oxford University Press. Born, C., Kru¨ger, H., & Lorenz-Mayer, D. (1996). Der unentdeckte Wandel. Anna¨herung an das Verha¨ltnis von Struktur und Norm im weiblichen Lebenslauf. Berlin: edition sigma. Bu¨chel, F., & SpieX, K. (2002). Form der Kinderbetreuung und Arbeitsmarktverhalten von Mu¨ttern in West- und Ostdeutschland. Stuttgart: Kohlhammer. Bundesministerium fu¨r Familie, Senioren, Frauen und Jugend (BMFSFJ). (2002). Bundesstatistik Erziehungsgeld 2001. Retrieved April 14, 2004 from www.bmfsfj.de/ RedaktionBMFSFJ/Abteilung2/Pdf-Anlagen/PRM-24020-Statistik-zum-Bundeserziehungs.pdf Cremer, C. (1984). Scho¨nheit wird zur Pflicht. Friseurin – Beruflichkeit auf Zeit. In: C. Mayer, H. Kru¨ger, U. Rabe-Kleberg & I. Schu¨tte (Eds), Ma¨dchen und Frauen. Beruf und Biographie (pp. 85–98). Mu¨nchen: DJI Verlag. Crompton, R., & Harris, F. (1999). Employment, careers and families: The significance of choice and constraint in women’s lives. In: R. Crompton (Ed.), Restructuring gender

192


relations and employment. The decline of the male breadwinner (pp. 128–149). Oxford: Oxford University Press. Elder, G. H. (1985). Perspectives on the life course. In: G. H. Elder (Ed.), Life course dynamics. Trajectories and transitions, 1968–1980 (pp. 23–49). Ithaca: Cornell University Press. Elder, G. H. (1996). The life course paradigm: Social change and individual development. In: K. Lu¨scher (Ed.), Examining lives in context: Perspectives on the ecology of human development (pp. 101–139). Washington, DC: APA Press. Elder, G. H. (1998).The life course and human development. In: R. M. Lerner, (Ed.), Handbook of child psychology. Theoretical models of human development (Vol.1, pp. 939–991). New York: Wiley. Elder, G. H., Johnson, K. M., & Crosnoe, R. (2003). The emergence and development of life course theory. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course (pp. 3–19). New York: Kluwer Academic & Plenum Publishers. Erzberger, C. (2001a). U¨ber die Notwendigkeit qualitativer Forschung: Das Beispiel der Alleinerziehungsdaten in quantitativen Daten. In: S. Kluge & U. Kelle (Eds), Methodeninnovation in der Lebenslaufforschung. Integration qualitativer und quantitativer Verfahren in der Lebenslauf- und Biographieforschung (pp. 169–188). Weinheim/Mu¨nchen: Juventa. Erzberger, C. (2001b). Sequenzmusteranalyse als fallorientierte Analysestrategie. In: R. Sackmann & M. Wingens (Eds), Strukturen des Lebenslaufs: U¨bergang–Sequenz–Verlauf (pp. 135–162). Weinheim: Juventa. Erzberger, C., & Kluge, S. (2000). Repra¨sentativita¨t qualitativer Untersuchungen: Lebensverlaufmuster als Ressource fu¨r Auswahlentscheidungen. In: W.R. Heinz (Ed.), U¨berga¨nge. Individualisierung, Flexibilisierung und Institutionalisierung des Lebensverlaufs. Zeitschrift fu¨r Soziologie der Erziehung und Sozialisation. 3. Beiheft, 298–313. Esser, H. (1996). What is wrong with ‘‘variable sociology’’? European Sociological Review, 12(2), 159–166. Hagestad, G. O. (1992). Assigning rights and duties: Age, duration, and gender in social institutions. In: W. R. Heinz (Ed.), Institutions and gatekeeping in the life course (Vol. 3, pp. 261–279). Weinheim: Deutscher Studien-Verlag. Hagestad, G. O., & Neugarten, B. L. (1985). Age and the life course. In: R. H. Binstock & E. Shanas (Eds), Handbook of aging and the social sciences (pp. 35–61). New York: Von Nostrand Reinhold Company. Hakim, C. (2001). Work-lifestyle choices in the 21st century: Preference theory. Oxford: Oxford University Press. Han, S.-K., & Moen, P. (2001). Coupled careers: Pathways through work and marriage in the United States. In: H.-P. Blossfeld & S. Drobnicˇ (Eds), Careers of couples in contemporary society. From male breadwinner to dual-earner families (pp. 201–231). New York: Oxford University. Heinz, W. R. (1996). Status passages as micro-macro linkages in life course research. In: A. Weymann & W. R. Heinz (Eds), Society and biography. Interrelationships between social structure, institutions and the life course (pp. 51–65). Weinheim: Deutscher Studien Verlag. Heinz, W. R. (1999). Job-entry patterns in a life-course perspective. In: W. R. Heinz (Ed.), From education to work: Cross-national perspectives (pp. 214–234). New York: Cambridge University Press. Heinz, W. R., & Marshall, V. W. (Eds) (2003). Social dynamics of the life course. Transitions, institutions and interrelations. New York: Aldine de Gruyter.


193

Hogan, D. P. (1981). Transitions and social change: The early lives of American men. New York: Academic Press. Hradil, S. (1992). Die ‘‘objektive’’ und die ‘‘subjektive’’ Modernisierung. Der Wandel der westdeutschen Sozialstruktur und die Wiedervereinigung. Aus Politik und Zeitgeschichte. Beilage zur Wochenzeitung Das Parlament, B-29-30/92, 3–14. Huinink, J. (1991). Familienentwicklung in der Bundesrepublik Deutschland. In: K. U. Mayer, J. Allmendinger & J. Huinink (Eds), Vom Regen in die Traufe: Frauen zwischen Beruf und Familie (pp. 289–317). Frankfurt/New York: Campus. Kluge, S., & Kelle, U. (Eds) (2001). Methodeninnovation in der Lebenslaufforschung. Integration qualitativer und quantitativer Verfahren in der Lebenslauf- und Biographieforschung. Weinheim/Mu¨nchen: Juventa. Kohli, M. (1985). Die Institutionalisierung des Lebenslaufs. Ko¨lner Zeitschrift fu¨r Soziologie und Sozialpsychologie, 37, 1–29. Kru¨ger, H. (1999). Gender and skills. Distributive ramifications of the German skill system. In: P. D. Culpepper & D. Finegold (Eds), The German skills machine. Sustaining comparative advantage in a global economy (pp. 189–227). New York, Oxford: Berghahn Books. Kru¨ger, H. (2003). The life-course regime: Ambiguities between interrelatedness and individualization. In: W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course. Transitions, institutions and interrelations (pp. 33–56). New York: Aldine de Gruyter. Levy, R. (1977). Der Lebenslauf als Statusbiographie. Die weibliche Normalbiographie in makrosoziologischer Perspektive. Stuttgart: Ferdinand Enke. Levy, R. (1997). Status passages as critical life course transitions. A theoretical sketch. In: W. R. Walter (Ed.), Theoretical advances in life course research (pp. 74–96). Weinheim: Deutscher Studien Verlag. Marshall, V. W., & Mueller, M. M. (2003). Theoretical roots of the life-course perspective. In: W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course. Transitions, institutions and interrelations (pp. 3–32). New York: Aldine de Gruyter. Marx-Ferree, M. (1997). Gender, conflict and change: Family roles in biographical perspectives. In: W.R. Heinz (Ed.), Theoretical advances in life-course research (2nd ed., pp. 123–137), Weinheim: Deutscher Studien Verlag. Mayer, K. U. (1998). Lebensverlauf. In: B. Scha¨fers & W. Zapf (Eds), Handwo¨rterbuch zur Gesellschaft Deutschlands (pp. 438–451). Opladen: Leske + Budrich. Mayer, K. U., & Mu¨ller, W. (1986). The state and the structure of the life course. In: A. B. Sørensen, F. E. Weinert & L. R. Sherrod (Eds), Human development and the life course: Multidiscliplinary persepectives (pp. 217–245). Hillsdale: Lawrence Erlbaum Associates. Moen, P. (2003a). Linked lives. Dual careers, gender and the contingent life course. In: W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course. Transitions, institutions and interrelations (pp. 237–258). New York: Aldine de Gruyter. Moen, P. (2003b). Midcourse: Navigating retirement and a new life stage. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course (pp. 269–291). New York: Kluwer Academic & Plenum Publishers. Mortimer, J. T. (2003). Working and growing up in America. Cambridge: Harvard University Press. Mortimer, J. T., & Shanahan, M. J. (Eds) (2003). Handbook of the life course. New York: Kluwer Academic & Plenum Publishers. Mortimer, J.T., Oesterle, S., & Kru¨ger, H. (2004). The transition to adulthood: Age norms, institutional structures, and the timing of educational completion and parenthood. In:

194


R. Macmillan (Ed.), Advances in life course research (Vol. 9). The structure of the life course: Standardized? individualized? differentiated?, Amsterdam: Elsevier Science. Myrdal, A., & Klein, V. (1968). Women’s two roles. Home and work. London: Routledge & Kegan Paul Ltd. Ondrich, J., SpieX, K., & Yang, Q. (2002). The effects of maternity leave on women’s pay in Germany 1984–1994. DIW Discussion Paper, 289. Berlin: Deutsches Institut fu¨r Wirtschaftsforschung. O’Rand, A. M., & Farkas, J. I. (2002). Joint retirement among US dual earner couples in the 1990s: Family and market influences on labor exit patterns. International Journal of Sociology, 32, 11–29. Pross, H. (1976). Die Wirklichkeit der Hausfrau. Reinbek bei Hamburg: Rowohlt. Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How common and does it matter? American Sociological Review, 52, 785–801. Sackmann, R., & Wingens, M. (2001). Theoretische Konzepte im Lebenslauf: U¨bergang, Sequenz und Verlauf. In: R. Sackmann & M. Wingens (Eds), Strukturen des Lebenlaufs. U¨bergang – Sequenz – Verlauf (pp. 17–48). Weinheim/Mu¨nchen: Juventa. Schneider, N. F., & Rost, H. (1998). Vom Wandel keine Spur – warum ist Erziehungsurlaub weiblich? In: M. Oechsle & B. Geissler (Eds), Die ungleiche Gleichheit. Junge Frauen und der Wandel im Geschlechterverha¨ltnis (pp. 217–236). Opladen: Leske + Budrich. Schu¨tze, Y. (1986). Die gute Mutter. Zur Geschichte des normativen Musters ‘‘Mutterliebe’’. Bielefeld: Kleine. Settersten, R. A., Jr. (2003). Age structuring and the rhythm of the life course. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course (pp. 81–98). New York: Kluwer Academic & Plenum Publishers. Settersten, R. A., Jr. & Mayer, K. U. (1997). The measurement of age, age structuring and the life course. Annual Review of Sociolology, 23, 233–261 Sørensen, A. (2004). Economic relations between women and men: New realities and the reinterpretation of dependence. In: J. Z. Giele & E. Holst (Eds), Changing life patterns in Western industrial societies (pp. 281–297). Amsterdam: Elsevier. To¨lke, A. (1989). Lebensverlaüfe von Frauen. Familia¨re Ereignisse, Ausbildungs- und Erwerbsverhalten. Mu¨nchen: DJI Verlag. Willms-Herget, A. (1985). Frauenarbeit. Zur Integration von Frauen auf dem Arbeitsmarkt. Frankfurt/New York: Campus. Wu, L. L. (2003). Event history models and life course analysis. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course (pp. 477–502). New York: Kluwer Academic & Plenum Publishers. Wurzbacher, G. (1951). Leitbilder gegenwa¨rtigen deutschen Familienlebens. Stuttgart: Enke Verlag.

PART III: BIOGRAPHICAL RECONSTRUCTION

195


196

LIFE COURSE TRANSITIONS AND SOCIAL IDENTITY CHANGE Nicholas Emler The thesis to be developed in this chapter is that transitions in the life course correspond to a particular kind of category change, namely shifts in social identity. It is therefore built upon a hitherto largely unrepresented discipline in the study of the life course, social psychology. The utility of a perspective that places social identity at the centre of the life course is that it draws attention to the kinds of events and circumstances associated with, and to some degree responsible for, life course transitions; it helps make sense of the consequences or effects of transitions; it directs attention to questions previously relatively neglected in the life course literature and it has the capacity to link the approaches of different disciplines to the analysis of the life course. The chapter will unfold as follows. First, the traditional view of transition within developmental psychology will be briefly set out as a counterpoint to the view recommended here. The concept of social identity will be elaborated, noting the manner in which it is currently constructed in social psychology, particularly in Social Identity Theory (SIT) and SelfCategorisation Theory, and then emphasising what is believed to be the key quality of social identity, the manner in which it connects individuals to their social worlds. Discussion will then turn to the psychological elements and processes in identity formation, taking the example of political identity.


197

198

NICHOLAS EMLER

Finally, the chapter considers transition processes and associated changes in social relations.

PROTOTYPICAL TRANSITIONS: THE LESSONS FROM COGNITIVE DEVELOPMENT The butterfly provides one of the more extreme and clear-cut examples of a life cycle marked by transitions. Periods spent successively as a caterpillar, a chrysalis and a butterfly are separated by quite rapid changes between these states. This life cycle would seem therefore to capture the essential features of transitions, namely that they are rapid, qualitative changes intervening between longer periods of relative stability. Human life cycles contain nothing resembling the dramatic physical changes that result in butterflies. Unquestionably, however, they do involve changes, and at least some of the changes appear to be qualitative. But are these changes also transitions? An influential prototype for this kind of change has been the Piagetian model of cognitive development (e.g., Piaget, 1952). Until Piaget’s work found a wider audience in the Anglophone world (cf. Flavell, 1963), two models dominated thinking about psychological development. One had its roots in psychoanalysis, a perspective that represented personality development as a sequence of qualitatively distinct stages. The other originated in learning theory and regarded development quite differently. From the perspective of this model, change was considered to be quantitative, not qualitative, and was taken to be gradual. Piaget’s ideas were clearly closer to psychoanalysis in one respect; he described developmental change as a sequence of qualitatively distinct stages, not as gradual quantitative change. But in other important respects it was quite distinct. Piaget provided a very detailed formal description of each stage, and his descriptions appeared to be empirically verifiable (Flavell, 1963). The power and persuasiveness of Piaget’s account lent popularity to other stage models of psychological development, including models that addressed the themes previously within the province of psychoanalysis. Examples included Marcia’s (1966) reworking of Erikson’s (1950) stage theory of identity development and Loevinger’s (1976) analysis of ego development. But one of the most influential stage models to follow the route pioneered by Piaget was Kohlberg’s (1976) theory of moral development. It drew closely on Piaget’s cognitive developmental theory in numerous respects. It adopted his proposal that cognitive functioning takes the form of structural wholes,

Life Course Transitions and Social Identity Change

199

integrated systems of thinking or generalised strategies for relating to the world, that each possess a degree of internal equilibrium. It also regarded the general cognitive stages identified by Piaget as the foundations for corresponding general stages of moral reasoning. But of particular interest here is the account Kohlberg provided of stage transitions. Stage transitions, Kohlberg proposed, are qualitative changes in cognitive structure precipitated by cognitive conflict or disequilibrium. The conflict might be largely internal in origin – for example, mutually contradictory conclusions generated by the same reasoning system – or external in origin – for example, arising from difficulties of assimilating new experiences within existing cognitive structures or of solving newly encountered problems within existing frameworks. The critical feature of the transition points or stage changes for Kohlberg was that they were inherently unstable and unable to endure without resolution. The resolution would be the construction of a new equilibrium. Kohlberg’s theory has been criticised on numerous counts. The criticisms have included challenges to most of his central claims about stages in development of moral reasoning, namely the claims that these stages are ordered in a progressive, universal, invariant and irreversible sequence. I am less concerned here with the force of these challenges than with the appropriateness of his neo-Piagetian concept of transition as applied to developmental change. It has turned out that empirically it is difficult if not impossible to pin down a transition point in the sense of a relatively limited period during which one form of cognitive functioning is being replaced by another. Moreover, this proves to be the case for the kinds of cognitive changes first described by Piaget, for example the change from preoperational to concretely operational thinking as well as the developmental shifts in moral reasoning anticipated by Kohlberg, for example the change from pre-conventional to conventional moral reasoning. Finding clear cases of short, unstable transitional phases in the data has proved to be difficult. In contrast, with respect to moral reasoning, Kohlberg’s own longitudinal data (Colby & Kohlberg, 1987) indicate that what should in theory be a transitional condition, one in which forms of moral reasoning corresponding to adjacent stages are used concurrently, is actually both more common and more enduring than the stable states – those using just one form of moral reasoning or the other – between which it is supposed to intervene only briefly. In the case of both Piaget’s and Kohlberg’s stages the changes involved do appear to be qualitative; in each case there is a category change in cognitive functioning. But in the concept of developmental stage both Piaget and

200

NICHOLAS EMLER

Kohlberg combine the proposals that psychological development involves category changes and that category changes have the characteristics of transitions; they are relatively abrupt shifts intervening between periods of relative stability and continuity. This model of stages does not appear to be verifiable in the cognitive domain. Are there other kinds of category change during the life course that do have the properties to qualify them as transitions? A promising class of candidates would appear to be provided by changes in social identity.

SOCIAL IDENTITY Over the past 30 years social psychology’s thinking about social identity has been dominated by a particular point of view, at least within Europe. This is the view first articulated in detail by Henri Tajfel (1978) as Social Identity Theory (SIT). Tajfel’s core idea, now widely accepted in social psychology (see Brown, 1986), is that in essence people’s subjective sense of who they are is to an important degree determined by the manner in which they define themselves socially. More specifically, what matters is which social categories form part of their self-definition. This argument grew out of research on prejudice and its connection to stereotyping. Tajfel saw that stereotyping is a process of social categorisation; it involves assigning other people to categories and regarding members of a common category as if they were qualitatively distinct from people outside the category. In other words, categories have boundaries and at the boundaries there are discontinuities. Tajfel (1969) drew on his work in cognitive psychology that predicted cognitive and perceptual effects of categorisation; the boundaries can be artificially created by perceiving greater discontinuities across imposed category boundaries and greater similarities within categories than exist in reality – processes also referred to as, respectively, contrast and assimilation, or sharpening and levelling. Just as there is no objective discontinuity in the electromagnetic spectrum between the areas conventionally labelled ‘green’ and ‘blue’ so also there is no objective discontinuity between people who are categorised, respectively, as ‘black’ and ‘white’. But social categorisation can only operate as an effective way of organising and simplifying experience of the social world by perceiving it as if this was the case. Thinking about the link between stereotyping as social categorisation on the one hand and prejudice on the other, led inevitably to thinking about self-categorisation. It was recognised that various ways of categorising


201

social objects – other people – tend to correspond, such that those who are categorised together descriptively are also categorised together evaluatively or emotionally – one feels similarly about them – and categorised together behaviourally – one treats them similarly (Doise, 1978). But why should one feel negative about particular categories and positive about others, and why should one discriminate against the former and favour the latter? One very strong reason for making the evaluative/emotional and behavioural distinctions in a particular direction is that any way of categorising the social world that is comprehensive or exhaustive must include a category to which the self belongs. And it would be natural to regard this category as better than others, to feel more positive about it than others, to regard it as more worthy than others and, when the opportunity arises, to treat it more favourably than others. It soon became apparent that matters were not this simple. Among the problems are that people do not invariably feel positive about the social categories they inhabit (Giles & Powesland, 1975). Nor is it clear, as was originally suggested, that in emphasising both the categorical distinctiveness and superiority of their own categories people are primarily driven by the desire to see themselves positively, to defend their own self-esteem (Rubin & Hewstone, 1998). But for the present discussion this is less significant than the fact that self-categorisations and social identifications of the self became of concern in social psychology. In effect, the focus of theorising and research shifted to recognition that social categorisation shapes how people think about themselves and not just how they think about others. The significance of self-categorisations has been explored in detail by one of Tajfel’s students, John Turner, and systematised as Self-Categorisation Theory (SCT; Turner, 1987). SCT incorporates several key ideas. One is that any individual potentially has access to multiple self-categorisations, such that from situation to situation the salience and therefore the influence of any one of these can vary. Another is that self-categorisation has a wide range of consequences. The particular category that is salient at any moment will shape affiliation and communication patterns, sensitivity to sources of influence, choices and judgments, accounts given and opinions expressed. Perhaps the most interesting propositions of SCT, however, are that such categorisations are organised hierarchically and that self-categorisations at different hierarchical levels are mutually exclusive. One may categorise oneself at the highest level of abstraction, perhaps as a living entity and in this respect like all other members of this category, at the lowest level that is to say as a unique individual distinct from all others, or at some level in

202

NICHOLAS EMLER

between. The levels in between are occupied by ‘‘social categories’’ – Italian, catholic, socialist, female, student, chemist, etc. – but Turner’s point is that one cannot simultaneously think and behave in terms of categories at different levels of the hierarchy. And most attention has been given to the personal and social levels of categorisation as, necessarily, alternative ways of thinking about the self. The origins, particularly of SIT, in the study of racism and prejudice, have been reflected in a research focus on ethnic and national identities. However, both SIT and SCT have aspired to be generic accounts of social identity, in principle applicable to social identity as such and therefore to any potential form of social identity. The minimal group paradigm (Tajfel, Flament, Billig, & Bundy, 1971) reflected this general aspiration. This paradigm created conditions in which all meaning is stripped away from social categories except the one quality they require for them to provide social identities, namely that each is a categorisation shared by two or more people (they divide a population into mutually exclusive categories). But SIT, and also SCT have stimulated research in which a wide variety of real social identities have been studied, including identities based on work-groups (Hennessy & West, 1999), occupations (Skevington, 1981; Marson, 2001), enrolment in different university courses (Reicher, 1984), support for different political parties (Abrams, 1994), and allegiance to different football clubs (Platow et al., 1999). Nonetheless, these different kinds of identities have more often been studied as opportunities to validate general principles by demonstrating replication across different cases. They have less often been examined to identify significant differences among kinds of social identities. I shall argue that there are some such differences and that they are important, particularly from a life course perspective. Finally, something should be said here about the relation between the concepts of social identity and social role. In particular, there is a question as to whether the former concept adds anything not already provided by the traditional sociological analysis of roles. According to one view, social categories and roles are actually alternative bases for structuring interactions (Mitchell, 1969). I suggest, however, that they emphasise different facets of the ways people relate to one another. The role concept, with its roots in drama, focuses on the manner in which the responsibility for tasks is distributed among positions within a shared structure. The emphasis therefore is upon integration of different activities. This approach draws our attention to the manner in which roles relate to and complement other roles – role sets being constituted by organisational structures; human organisations have, indeed, been defined as a system of roles (Katz & Kahn, 1978).


203

There are three obvious ways in which a social identity perspective differs. First, it emphasises differentiation rather than integration; its focus, as previously noted, is on the processes that reinforce similarities within categories and differences between categories, on boundaries rather than on links. Second, it emphasises who people are seen to be and feel themselves to be in contrast to the role perspective emphasis on the tasks people perform. In these two respects the two concepts are somewhat complementary. Third, however, social identity appears to be a more inclusive concept. All social roles create categories of people and thus social identities, but many categories do not entail roles in a shared social organisational structure. Think, for example, of social identities based on religion, nationality or ethnicity. As to whether social identity or social role is the more useful concept in studying the life course; the answer I think is that both have a contribution to make. This chapter focuses on the hitherto more neglected contribution of the two. The paradigm case in the social psychology of social identity remains ethnic identity, an identity that is normally relatively fixed over the life course. And perhaps for this reason, analysis of change in social identity – which naturally arises as a central question for a life course approach – has not been extensive within these theoretical traditions. The principle exception has been Tajfel’s own examination of social mobility (Tajfel, 1978). But mobility was proposed as a defensive manoeuvre, undertaken only rarely and then to protect self-worth by escaping from a negative identity. The absence of compelling evidence that membership of low prestige social categories is associated with lower self-esteem (Emler, 2001) argues against this motivation. Within the life course frame social identity change is normative, not exceptional. The challenge is to explain the process of change.

CORE SOCIAL IDENTITIES Self-categorisation theory emphasises the situational specificity of social identification, which is to say categories and category contrasts that are momentarily salient for the individual social actor. Indeed, Turner (1987) notes that social categorisations can be spontaneous or emergent constructions within a particular situation, the so-called minimal group being just such a situation; social identities are therefore not confined to classifications that are culturally provided or built into the social structure of a society. From the perspective of life course analysis, however, social identities that both dominate or characterise entire periods of life and that are widely recognised and rooted within the culture are of particular interest.

204

NICHOLAS EMLER

But is it really the case that some social identities are more dominant or central, chronically more salient? One kind of research that might decide this would sample individuals’ identity salience across time, much in the way that experience sampling methods have been used to determine the prevalence of different mood states (cf. Czikszentmihalyi & Larson, 1987), but such research is yet to be done. Indeed, it is more generally true that to date there has been almost no longitudinal research focused specifically on social identity, so that the evidence available to us is cross-sectional – comparisons of people at different ages or life stages. There are, however, other kinds of indications that some social identities are reliably more central or significant in people’s lives. One such indication comes from research on the strength of different kinds of identification in the world of work. Marson (2001) showed that of the three kinds of identification in this domain, respectively with a work team, with an employing organisation and with an occupation or profession, the third consistently emerged as by far the strongest across a variety of populations. Second, as Fiske (1998) points out, some categorisations are more regularly important because they are visually highly salient and therefore socially functional; they are immediately available to structure social interactions. Fiske’s top three categories on this basis are race, gender and age. However, other categorisations can be and often are made chronically visually salient by manipulating appearance. This can be done through conventionalised styles of grooming and adornment and even through bodily movement, as Mauss (1935) observed many years ago. But these options are usually combined with an even more visually powerful marker, dress. The alternative categories with which young people have identified – hippy, skinhead, goth, punk and so on – are signalled by combinations of dress codes and bodily adornments, as are many other significant social categorisations based on ethnicity, religion and even class and occupation. The point is that such visual markers of social identity lend cross-situational durability, and therefore importance, to these identities because they cannot rapidly be changed. And the rapid changes in appearance that are possible – shaving the head, acquiring a tattoo – are only significant if they cannot be equally rapidly reversed. Another source of evidence comes from research on the manner in which connections are anticipated in networks of acquaintances. Milgram’s (1967) studies of the ‘‘small-world’’ phenomenon led to a research showing that people are surprisingly good at working out how they can efficiently reach others they do not yet know through the intermediary of personal acquaintances (Klineberg, 2000). It appears that this facility is based on the


205

fact that acquaintances have multiple social identities; they can be simultaneously classified in several different ways. However, when people undertake this task they typically use only a small number of these potential classifications, and usually only two of them. Those most commonly used are based, respectively, on occupation and geography (Killworth & Barnard, 1978). This last example underlines another feature of the social identities that is likely to be most important over the life course: these are the principle categories applied to people by their own acquaintances. Social identities are therefore social in two distinct respects. As categories they are socially defined and the most relevant categories will be widely recognised within a culture. As applied to particular individuals, however, these identities are validated by others. A person may aspire to a particular identity but unless this aspiration is acknowledged by at least some significant others it remains no more than a private desire. Social identities are viable bases for relating to others only if at some level they have been negotiated with and accepted by a relevant audience, an argument we have made on the basis of our studies of delinquent and non-delinquent identities (Emler & Reicher, 1995); social identities assume reality and become consequential when they are performed before an audience and when the audience confirms the authenticity of the performance (cf. also Emler, 1990). If identities reflect how others define us, they might nonetheless be relationship-specific. This was the essence of William James’s famous declaration that we have as many selves as people who know us, the implication being that identities are entirely relationship-specific. Though at some level this might be true, the degree to which it applies is probably trivial. It would be very odd if people who know you well did not agree whether you were a mother, a Catholic, of Irish descent, a civil servant and so on. It might be countered that these are not the essence of the self that others know, but the response is that they and other such categories do constitute the social self. There will be, in addition, a history to each relationship that is its own and a judgment by the other on this basis as to one’s personality. But such judgments are also not, as a matter of evidence, relationship-specific; our acquaintances tend to agree about our personality (Kenrick & Stringfield, 1980), and the better they know us the more they agree (Kenny, 1994). Some measure of agreement is additionally generated by the fact that our acquaintances also know each other and share with each other their views of us. Up to this point the importance of social identities to everyday life can be summarised as follows: our social identities shape the character and content

206

NICHOLAS EMLER

of our relations with others. Perhaps therefore an indicator of the centrality of an identity is the scope of such effects – how many details of our relationships are affected, and with how many of the others we know and interact with. Social identities are likely to shape two further details of our connections to the social world: who we know and who we encounter on a daily basis. In the first case, the effect is that we will be personally acquainted with more people who share a social identity with us than will people who do not share the social identity with us. University students will know more members of this category (and mothers will know more mothers, socialists more socialists, Glaswegians more Glaswegians, delinquents more delinquents, married people more married people) than will people who are not university students etc. This, of course is the basis for the ‘‘small-world’’ facility described by Killworth and Bernard (1978). Additionally, for students and for members of some other categories, many and perhaps the majority of their daily encounters are with other members of the same category. But although these two features – who else one knows and who else one encounters most regularly – are sometimes strongly related, they are not the same. Take the extreme case in which a priest or a doctor is one of a kind in a small, rural community. In each case their daily encounters will be strongly shaped by their vocation, many of the priest’s encounters are with other community residents as his parishioners, and the doctor’s with residents as patients. Other members of their own respective vocational categories will not dominate their array of daily social contacts. But both will have personal acquaintances who do belong to their own social category at a higher rate than will other members of their community; the doctor will know more doctors than will his or her patients, the priest will know more priests than will his parishioners. In other words, part of what it means to belong to a social category is that one will be personally acquainted with other members of the category at a higher rate than the base rate for the population as a whole. In a series of studies we have collected data on people’s patterns of routine social contacts (e.g., Emler, 1990; Emler & McNamara, 1996). These data have been derived from a variety of groups, including adolescents in high school, university students, young people in employment, and adults working as managers, in effect groups of people belonging to different social categories. The data indicated that various features of contact patterns are associated with category membership. This point is illustrated in Fig. 1 with data (from Emler, 2000) on the social contact patterns of young people in this case all at the same point in the life course but occupying one of four

207

Life Course Transitions and Social Identity Change 20 Univ

18

Coll

16

Employ

14

Unemp

12 10 8 6 4 2 0 Close Fr

Fig. 1.

Other Fr

Cas Aqu

Formal

Stranger

Contact Patterns: Number of Different People Encountered in Each Relationship Category (Over 7 days).

distinct categories. These data do not give us the social identities of the contacts, but it is a fair guess that many are members of the same social categories as the respondent – university students in the case of the first group, vocational college students in the case of the second, work colleagues in the case of the third. The guess is supported by the common locations for encounters in each group, respectively, on the university campus, at the college and in the workplace. Something else that these data reveal is that alternative social category memberships can be associated with dramatically different rates of social contact;for example, compare the unemployed group with others. To summarise, social identities connect individual to social worlds in the following respects. First, individuals are identified by significant others with particular social categories and on the basis of this others have particular expectations about how an individual will and should behave, about who else that person knows and about how that person should be treated. In other words, social identities structure interactions. Second, social identities are reflected in the categories of people an individual is acquainted with. This has consequences for the resources to which they have access and their value to others. Third, social identities are linked to patterns of routine social contact; often, but not always, people will tend to have more regular contact with other members of the same social category than will those who

208

NICHOLAS EMLER

do not belong to that category. This has consequences for, among other things, the range of ideas, opinions and influences to which individuals are most exposed.

CHANGING SOCIAL IDENTITY If the above summary is appropriate, then it has important implications for transitions in social identity. In effect, changes in social identity require changes in social relationships. Moreover, the foregoing analysis implies two kinds of change in social relationships. The first is change in how you are defined, and therefore treated by the people who already know you – the substance and character of your interactions with them. The second is change in who you know and with whom you regularly associate. Superficially, it might appear that changes of these latter kinds would not always and invariably occur. Think of a child who grows to adulthood, gets married and becomes a parent in a small agricultural community. At each of these transitions there might be small adjustments in the patterns of daily contact, for example more with the new spouse and less with other opposite sex peers – but there will be little change in who this individual knows. Childhood friends and playmates are likely to go through similar transitions at similar ages, and so become adult neighbours and acquaintances. The problem with this scenario is that entirely closed human communities in which the only turnover in membership arises from births and deaths do not and never have existed. On the other hand, there is undoubtedly variation across time and societies in the extent of turnover in community membership and thus variation in the continuity of any individual’s acquaintances (Slater, 1968). Moreover, to the extent that emotional ties and regular contacts persist with the same people across transitions in social identity these would appear to provide obstacles to the process of successful transition. Because social identities derive stability from the continuity of relations with others, this very continuity is also a source of inertia. If the people who know you are in the habit of thinking of you in certain terms, treating you in particular ways, and having specific expectations about you, they will not readily or easily switch to thinking, behaving and evaluating in quite different ways. The mother who has difficulty adjusting to the change in status of her beloved son from child to adult is only a more extreme example of a commonplace phenomenon. It may be that the various rituals of transition – wedding services, baptisms, graduation ceremonies,


209

21st birthday parties, coronations, retirement celebrations – are social devices to counteract such inertia. In these rituals, communities gather and publicly acknowledge a movement of one or more of their number from one social category to another. Later, I will reconsider these transition rituals together with the contemporary significance of changes in acquaintance and contact patterns. But if a change in social identity is a change in the way the individual is connected to the social world, there are nonetheless other, more internal and psychological aspects to the process of social identity change. This can be illustrated with the case of political identity.

THE FORMATION OF POLITICAL IDENTITY The political life of societies tends to be played out as a competition between groups of people who have adopted differing political identities. For a long time political psychology was dominated by questions about the determinants of this choice, interpreted as one of positioning along a left–right ideological dimension: why do some people end up aligned with the political left and others with the political centre or right? However, it gradually emerged that an additional and perhaps more significant question existed: why do people become politically aligned at all? This question was promoted by the observation that many adults appear to have no political leanings (Converse, 1964). Subsequently, evidence for a wide variety of variations in political involvement emerged, including variations in the extent of political participation, interest in politics, opinion stability, political knowledge, attention to political events and so on. Research in which these dimensions of variability have been concurrently assessed has indicated that they are strongly interrelated and are apparently manifestations of a common latent variable. In one such study, Nie, Junn and Stehlik-Barry (1996) labelled this variable ‘political engagement’. Another interpretation is possible, however. Identity formation is a process in which different elements are related in a causal sequence. In the case of political identity, the following sequence might be proposed (cf. Emler, 2002). First, one needs to have some interest in politics, and this is a necessary condition for taking a second step, namely, paying attention to political information and events. Such attentiveness allows the individual to accumulate knowledge of the political world. This individual can then begin to form judgments or opinions on the basis of this knowledge, and

210

NICHOLAS EMLER

subsequently to organise these opinions into coherent political positions. Finally, an established political position provides the foundation for ideologically consistent political action. We have made a preliminary test (Emler, Romney, & Bynner, 2004) of a sequential model of political identity formation by drawing on data from a large sample of 16–20 -year-olds (Bynner, Romney & Emler, 2003). This test supports a slightly modified sequence, which begins with political interest leading to attentiveness. This later, in turn, precedes responsiveness defined as a positive attitude towards politics. Responsiveness in its turn leads to action; in the data set action was assessed in terms of attending political meetings, taking part in demonstrations, etc. The final three steps in the sequence are from action to voting intention, and thence to opinionation, this last step reflecting the extent to which individuals have clear opinions on political questions. This was an imperfect test given that it was not based on data collected specifically to test the model. In particular, there was no measure of political knowledge. However, the relevance of this example is to make the point that a developmental model of identity formation or change is testable, and can provide a basis for assessing progress towards a new identity. I would also argue that the four kinds of psychological elements that appear to be involved in the construction of a political identity, namely, motivational, cognitive, attitudinal and behavioural elements, will be found in all significant social identities, and particularly in those that dominate and define phases in the life course. This tells us that social identity change is a process requiring time because it requires a sequence of interconnected psychological changes. Having considered the psychological elements of identity formation and change, it is important to recognise that the psychological changes are also socially supported. In the case of political identity, motivation or interest will be encouraged or discouraged by acquaintances. Palmonari, Pombeni, & Kirchler (1992) give the example of young people belonging to friendship groups in which interest in politics violates group norms. Friends and acquaintances may be sources of political information but are likely to play an even more important part in opinion formation and attitude organisation, providing opportunities to rehearse positions and check their social meaning (Emler, 1990). Finally, others will be role models for, and mobilisers to, action. Progress in the formation of political identity, therefore, will depend on access to others willing and able to provide these supporting roles. Parallel social processes are likely to be at work in other kinds of social identity change.


211

SOCIAL IDENTITY CHANGE AND THE PROCESS OF TRANSITION There are features of the change in political identity, from unaligned to aligned, that would seem to make it quite an atypical example of the identity changes characterising the life course. In particular, there is no abrupt transition in identity. On the contrary, the change is characteristically gradual. But is it really the case that the life course is periodically punctuated by rapid shifts in social identity? This brings us back to the concept of transition as a change that is abrupt as well as qualitative. Major life course transitions are often represented as specific events – puberty, graduation from university, the first day at work, a wedding (or divorce), birth of a first child, retirement day. But, with the exception of childbirth, there are common examples of more extended transitions corresponding to each of these cases. It is now common for many couples to move gradually towards shared domestic arrangements long before marriage and in many cases the relationship is never formalised as a legal marriage. The move into work is frequently drawn out through work experience and periods of training or apprenticeship. At the other end of their working lives, it is not uncommon for people to go part-time long before they finally cease working for an income. In Finland, a high proportion of students never terminate their studies with a completed qualification; instead they become progressively more part-time as students, spending correspondingly more time working for an income, and so pass gradually rather than abruptly from the status of a student to that of a worker. The most conspicuous case of protracted transition, however, is that from childhood to adulthood, a process to which the very term ‘‘extended transition’’ was first regularly applied. In contemporary societies there is no specific event to mark this transition. Instead, there is a progressive accumulation of legal rights at different ages, together with gradual financial and domestic independence from parents, which may not be completed until the late 20s or even beyond (Jones & Wallace, 1992). These various extended transitions have both advantages and disadvantages. In so far as new social identities require new knowledge, attitudes and behavioural repertoires, the extended time scale provides more opportunity to develop and integrate these. The major disadvantage is in achieving social recognition of the new identity. To see how this disadvantage might be overcome, let us return to the process by which new social identities are established between the individuals who are to adopt them and the social worlds in which they dwell. How is this

212

NICHOLAS EMLER

social world led to accept an individual’s movement into a new identity? As already noted, rituals of transition provide devices for securing collective and public consent to new identities. But, as already noted, both the prevalence and the significance of many of these appear to be on the decline. This decline can be linked to a concurrent rise in resort to two other options, either to change one’s appearance or to change the people one knows. Formal changes in appearance are, like rituals of transition, currently becoming less commonplace. In Britain most children used to go to school in uniforms, and then changed or abandoned uniforms as their age-based identities changed, but these practices are in decline. Manner of dress also signifies certain religious identities, and thus their adoption and abandonment. Practitioners of a few vocations still signal their identities by routinely wearing the appropriate apparel. But occupation-linked uniforms now more often signal a role only temporarily occupied, a time- and place-specific performance. On the other hand, one might speculate that when teenagers are drawn to sub-cultures with highly distinctive dress codes this will be filling the gap left by the transition rituals that would in former times have marked a change in identity from child to adult. What of the other option, changing the people one knows? This is most readily accomplished by a radical alteration in the patterns of daily contact. In effect, we interact with the people whose paths cross ours in the course of our regular daily activities (Emler, 2000). This might seem a trivial point but it is quite fundamental. Social contact is generated primarily by routines that take people into particular places to do particular things at particular times. Contact patterns are therefore altered by changes in these routines. Birth of a first child usually forces a substantial shift in a woman’s routine (but also usually has rather less effect on the father’s routine). The change produced by childbirth in a mother’s routine arises from the fact that she must commit her time in a new way, and spend more of it in a particular setting. This, namely relocation in physical space, is the principle way in which routines, more generally, are altered. There are many examples of this effect – going to school for the first time, moving from elementary to high school (and, still a common pattern in Britain going to a residential or a boarding school), leaving home to go to university, leaving the parental home to set up one’s own, starting a new job, doing military service, moving on retirement to a warmer climate. Each of these physical displacements has the effect of altering patterns of routine social contact, and therefore each potentially supports a shift in social identity. Equally evident is that there are cultural variations in common practice that may reflect the problems a particular economy poses for identity


213

change. Children growing up on a Kibbutz have been expected to work on the Kibbutz as adults but have also been expected at the end of their education to leave for a year and travel (and during this period also find a spouse). And there are changes in practice over time. In Britain, the tradition of leaving home to attend university is in the decline as rising costs force more students to attend a local university so that they can continue living at home. But this trend has been accompanied by another that produces a break in routine and a discontinuity in relationships, the ‘‘gap year’’ in which travelling abroad intervenes between high school and university. It is also evident that physical displacement is not a purely modern solution. Some tribal societies have created discontinuity of association in their otherwise settled communities by physically segregating the sexes for a time at puberty (Goode, 1959). In the 19th century it was the fashion for young Englishmen of aristocratic background to depart on a European tour, and in the 20th century for the females to be sent off to ‘‘finishing schools’’ in Switzerland. More generally, leaving home at the end of childhood to find fame and fortune as well as to find an adult identity, has long been a theme of folklore and fairytales. Indeed, life as a journey is an almost universal metaphor, but perhaps not so metaphorical; real journeys are involved. At the conclusion of this particular journey, I wish to argue that it has shown that transitions in the life course as changes in social identity are more likely to be gradual than abrupt, for two kinds of reasons. First, identity change requires a number of psychological changes that take time to build. Second, social identities derive resistance to change from the relationships with others in which they are embedded and through which they are confirmed. But a common response to this inertia is some displacement in physical space, supporting both the formation of new relationships on a new basis and a return to the old ones from the vantage of a new category. One hope is that the future will bring the kind of longitudinal evidence needed to move these ideas about social identity change in the life course from the realm of speculation to that of empirically grounded observation.

REFERENCES Abrams, D. (1994). Political distinctiveness: An identity optimising approach. European Journal of Social Psychology, 24, 357–365. Brown, R. (1986). Social psychology: The second edition. New York: Free Press.

214

NICHOLAS EMLER

Bynner, J., Romney, D., & Emler, N. (2003). Dimensions of political and related facets of identity in late adolescence. Journal of Youth Studies, 6, 319–335. Colby, A., & Kohlberg, L. (1987). The measurement of moral judgment, Vol. 1. Theoretical foundations and research validation. New York: Cambridge University Press. Converse, P. (1964). The nature of belief systems in mass publics. In: D. Apter (Ed.), Ideology and discontent (pp. 206–261). New York: Free Press. Czikszentmihalyi, M., & Larson, R. (1987). Validity and reliability of the experience sampling method. Journal of Nervous and Mental Disease, 175, 526–536. Doise, W. (1978). Groups and individuals: Explanations in social psychology. New York: Cambridge University Press. Emler, N. (1990). A social psychology of reputation. European Review of Social Psychology, 1, 171–193. Emler, N. (2000). Social structures and individual lives: Effects of participation in the social institutions of family, education and work. In: J. Bynner & R.-K. Silbereisen (Eds), Adversity and challenge in the life course in England and the new Germany (pp. 62–84). London: Macmillan. Emler, N. (2001). Self esteem: The costs and causes of low self worth. New York: York Publishing Services. Emler, N. (2002). Morality and political orientations: An analysis of their relationship. European Review of Social Psychology, 13, 259–291. Emler, N., & McNamara, S. (1996). The social contact patterns of young people: Effects of participation in the social institutions of family, education and work. In: H. Helve & J. Bynner (Eds), Youth and life management: Research perspectives (pp. 121–139). Helsinki: Helsinki University Press. Emler, N., & Reicher, S. (1995). Adolescence and delinquency: The collective management of reputation. Oxford: Blackwell. Emler, N., Romney, D., & Bynner, J. (2004). The development of political engagement. EARA 9th Biennial Conference, Porto, May 2004. Erikson, E. H. (1950). Childhood and society. New York: Norton. Fiske, S. (1998). Stereotyping, prejudice and discrimination. In: D. Gilbert, S. T. Fiske & G. Lindzey (Eds), The handbook of social psychology (4th ed., Vol. 2, pp. 357–411). New York: McGraw-Hill. Flavell, J. H. (1963). The developmental psychology of Jean Piaget. New York: Van Nostrand. Giles, H., & Powesland, P. F. (1975). Speech style and social evaluation. London: Academic Press. Goode, W. (1959). The theoretical importance of love. American Sociological Review, 24, 38–47. Hennessy, J., & West, M. A. (1999). Intergroup behaviour in organisations: A field test of social identity theory. Small Group Research, 30, 361–382. Jones, G., & Wallace, C. (1992). Youth, family and citizenship. Buckingham: Open University Press. Katz, D., & Kahn, R. (1978). The social psychology of organisations (2nd ed.). New York: Wiley. Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford Press. Kenrick, D. T., & Stringfield, D. O. (1980). Personality traits and the eye of the beholder: Crossing some traditional philosophical boundaries in the search for consistency in all of the people. Psychological Review, 87, 88–104.


215

Killworth, P. D., & Bernard, H. R. (1978). The reverse small world experiment. Social Networks, 1, 159–192. Klineberg, J. (2000). Navigation in a small world. Nature, 406, 845. Kohlberg, L. (1976). Moral stages and moralization. In: T. Lickona (Ed.), Moral development and behaviour: Theory, research and social issues. New York: Holt, Rinehart & Winston. Loevinger, J. (1976). Ego development: Conceptions and theories. San Fransisco: Jossey-Bass. Marcia, J. (1966). Development and validation of ego-identity status. Journal of Personality and Social Psychology, 3, 551–558. Marson, K. (2001). Work-based social identities. British Psychological Society Social Psychology Section Annual Conference, Surrey, July 2001. Mauss, M. (1935). Les techniques du corps. Journal de la Psychologie, 32, 271–293. Milgram, S. (1967). The small-world problem. Psychology Today, 1, 60–67. Mitchell, J. C. (1969). Social networks in urban situations. Manchester: Manchester University Press. Nie, N. H., Junn, J., & Stehlik-Barry, K. (1996). Education and democratic citizenship in America. Chicago: University of Chicago Press. Palmonari, A., Pombeni, M.L., & Kirchler, E. (1992). Evolution of the self-concept in adolescence and social categorisation processes. In: W. Stroebe & M. Hewstone (Eds), European review of social psychology (Vol. 3, pp. 285–308). Oxford: Wiley. Piaget, J. (1952). The origins of intelligence in children. London: Routledge. Platow, M., Durante, M., Williams, N., Garrett, M., Walshe, J., Cinotta, S., Lianos, G., & Barutchu, A. (1999). The contribution of sport fan social identity to the production of prosocial behaviour. Group Dynamics, 3, 161–169. Reicher, S. D. (1984). Social influence in the crowd: Attidudinal and behavioural effects of deindividuation in conditions of high and low group salience. British Journal of Social Psychology, 23, 341–350. Rubin, M., & Hewstone, M. (1998). Social identity theory’s self-esteem hypothesis: A review and some suggestions for clarification. Personality and Social Psychology Review, 2, 40–62. Skevington, S. (1981). Intergroup relations and nursing. European Journal of Social Psychology, 11, 43–59. Slater, P. E. (1968). Some social consequences of temporary systems. In: W. Bennis & P. E. Slater (Eds), The temporary society. New York: Harper & Row. Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25, 79–97. Tajfel, H. (1978). Differentiation between social groups: Studies in the social psychology of intergroup relations. London: Academic Press. Tajfel, H., Flament, C., Billig, M. G., & Bundy, R. F. (1971). Social categorization and intergroup behaviour. European Journal of Social Psychology, 1, 149–177. Turner, J. C. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell.


216

THE IMPACT OF PERSONALITY AND LIVING CONTEXT ON REMEMBERING BIOGRAPHICAL TRANSITIONS Pasqualina Perrig-Chiello and Walter J. Perrig 1. BIOGRAPHICAL EXPERIENCES AND THEIR IMPORTANCE FOR THE REGULATION OF WELL-BEING The question, as to what extent past experiences have an impact on current well-being and future anticipations, has provided an important, but controversial, research topic since the very beginning of psychology as a discipline in its own right. Some investigators have emphasized the importance of past biographical experiences, claiming that, at any given time, individuals are very much the product of their own life history quite apart from external, situational demands, opportunities and barriers (Freud, 1917; Erikson, 1959; McAdams, 1993). Others have pointed to a lack of reliable empirical evidence on this matter, owing to the largely retrospective character of most of the data (Rutter, 1996). Recent advances in life-span developmental psychology have meanwhile brought increasing empirical insight and rigor to this crucial research question. It has been suggested that individual lives can be characterized as a Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 217–235 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10008-2

217

218

PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG

series of interrelated transitions. Transitions define points in the human life cycle when roles are transformed, redefined and left behind for new roles (e.g. starting school, experiencing puberty, starting work, leaving home, getting married, having children, retirement, etc.). These changes involve and give shape and direction to various aspects of a person’s life (Mercer, Nichols, & Doyle, 1989; Sugarman, 2001). It has been argued that early transitions can have lasting effects on subsequent transitions, even after many years and decades have passed (Elder, 1998). Some studies have shown a wide range of effects in later life following the experience of specific transitions in childhood (Wadsworth, Maclean, Kuh, & Rodgers, 1990; Wertlieb, 1997) and adolescence. For example, psychological functioning during puberty and adolescence was significantly correlated with that in old age (Vaillant, 1990; Clausen, 1991). Ryff and Heidrich (1997) showed that reported normative early life events and transitions were significant predictors for multiple aspects of present and future well-being across different age groups. Such findings lead to questions about the underlying mechanisms responsible for such a relationship. For example, can a biographical transition directly affect, or even give rise to, a specific developmental outcome in later life course? Could there also be a third, mediating variable behind the observed correlation? Or, is it the case that some kind of backward inference, that is, for example recollected (or even reconstructed) mental projections from the present into the past can account for the data? This chapter is devoted to these questions and mainly elaborates on those memory processes that are related to autobiographical experience and its role on regulation of well-being (see also McAdams contribution in this volume).

2. TRANSITIONS AND LIFE EVENTS: OBJECTIVE FACTS VERSUS SUBJECTIVE RECONSTRUCTIONS Despite the persuasiveness of the relationship between earlier transitions and later developmental outcomes, the theoretical basis of these findings is far beyond clear. Because these findings are mainly based on retrospective reports, the question cannot easily be answered, whether a past transition per se has a long-term effect on later life, or whether it is rather the subjective post-hoc interpretation that matters (Rutter, 1996). From a memory theoretical point of view, subjective reports of the past are, like all autobiographical memories, the products of personal

The Impact of Personality and Living Context

219

reconstruction. For example, if a relationship is found between the remembered past and later life events, it could well be that the memorized emotional quality of the previous events accounts for later well-being and not the event per se (Schacter, 1996; Schwarz & Strack, 1999; Keyes & Ryff, 1999). In memory research there is abundant evidence that reports of the past are not objective records of events but constructions that are biased by multiple sources. Remembering embraces memory traces, amalgamated with actual perceived information and contextual constraints. These are all subject to current decision-making processes that can be very subtle, resulting in statements with varying accuracy and giving rise to subjective (rather than objective) truth values (Tulving, 1983; Johnson, Hashtroudi, & Lindsay, 1993; Rubin & Berntsen, 2003). Furthermore, recalling biographical episodes not only includes re-experiencing the past, but also the factual retrieval of autobiographical knowledge that e.g. might have been reported by others. There is good evidence in the literature for a remember/know distinction in memory (Conway, 1987; Gardiner & Java, 1990; Rubin, 1986) even down at the neuropsychological level (Wheeler, Stuss, & Tulving, 1997). Also, it is generally accepted that autobiographical knowledge is stored and organized within a hierarchical framework (Conway & Rubin, 1993). A lifetime period is the most general, most abstract and most inclusive type of knowledge, and denotes typically units of years. General events (at the next lower hierarchical level) represent common or frequently experienced episodes that are amalgamated and consolidated in memory, giving rise to a kind of basic, but personalized, knowledge store (Bartlett, 1932/1993). Specific, single-event knowledge (at the lowest hierarchical level) is particularly salient, and gives rise to unique memory traces that appear to be stored separately. In subjective reports of past life transitions, it will never be easy to separate out what is actually re-experienced (episodic memory) from what is generally known – and expected (semantic memory) in any of the three types of autobiographical memory. Neither it is clear, whether memory reports represent real facts from the past or just constructed inferences induced by the retrieval setting. Thus, in exploring past-event reports the distinction between the more cognitive-oriented analyses of autobiographical reports and the more experiential-oriented analyses of reminiscence is an important one (Bluck & Alea, 2002). Memorized life transitions can thus be considered as products of a kind of motivated forgetting or remembering serving individual psychodynamic functions (such as equilibration of self-identity). There are ample reports in the literature of work describing phenomena related to constructed, reconstructed and distorted memories in general

220


(Ayers & Reder, 1998; Brainerd, Reyna, Wright, & Mojardin, 2003) and in autobiographical memory research in particular (Brewer, 1988; Conway & Rubin, 1993). There are impressive demonstrations of how misleading information or faint feelings of familiarity can bias memory reports (Whittlesea, 2002). Certainly, the most dramatic demonstrations of false memories describe how participants in certain experimental studies can begin to ‘‘remember’’ painful autobiographical events that were simply suggested to them and which never really happened (Loftus, 2003). Even so, the precise constructive mechanisms underlying false or distorted reports of past events are not yet completely understood and continue to be investigated (Conway, 1996; Bluck, 2003; Bluck & Alea, 2002). Bartlett’s sensible analysis of remembering and his theoretical conclusions (Bartlett, 1932/1993) help to better understand these processes. In his view, the remembering of autobiographical events is influenced by ‘‘schematic’’ knowledge. Bartlett’s ‘‘schema’’ refers to the activation of relevant past experience in such a way that it affects the interpretation of any presently incoming sensory information. Although this usefully ensures the updating of general knowledge held in memory, and helps us to respond in situations where there is inadequate information to hand, it can sometimes lead us astray through faulty expectations. For example, when a subject starts to remember a complex situation, the first thing that comes to mind is often a general impression of the whole, which is an ‘‘attitude’’ toward it. By ‘‘attitude’’, Bartlett referred to a complex psychological state or process, which is very largely a matter of feeling or affect. He characterized this state as a collection of subjective phenomena such as ‘‘yby doubt, hesitation, surprise, astonishment, confidence, dislike, repulsion and so on’’ (Bartlett, 1932/1993, p. 207). Recall of the situation or event is thus a reconstruction, made largely on the basis of this current cognitive–emotional attitude. The general effect is that of a justification of the attitude: ‘‘This and this and this must have occurred, in order that my present state should be what it is’’ (Bartlett, 1932/1993, p. 202). If the topic of remembering is a biographical transition, the relevant ‘‘schema’’ for self-related knowledge will be activated and may control the process of remembering. Here, the actual schema held by a person about her/himself (self-concept) can be considered as his/her hierarchically organized, autobiographical knowledge that has developed over time and has been very largely determined by immediately preceding or actual experiences. Therefore, the current schemata or themes of the self allows for constancy as well as change. Obviously, there is enormous stability in the preservation and continuity of one’s self-identity and of personality, that is,


221

the complex of all the attributes that characterize a unique individual over time. But there is also instability in the construction of one’s autobiographical memories, which are dependent on the situational living contexts of different retrieval episodes. Predictions from this theoretical perspective can be applied to (1) the correlation between remembered life events and actual well-being, (2) the correlation between self-identity/personality and memory performance and (3) the influence of particular living contexts on remembering the past. Based on the arguments set out above, we would expect firstly that any studies describing past transitions and actual life situations by the means of subjective reports should find significant relationships between the two. The reason for this is twofold: On one side autobiographical schematic knowledge contributes to the constancy and stability of self-identity over time (Bluck, 2003), on the other side the constructive processes involved in experiencing the actual situation as well as remembering the past are selfserving in the regulation of the actual well-being. Actual well-being and stability in self-identity is based on conditions of the present, but also on the past, that serves as reference or justification for the actual situation. Such self-serving regulatory processes do operate subtly and smoothly during stable periods of life, but might be extreme in their effects in specific living contexts (e.g. critical life situations or in therapeutic settings). Our further predictions relate to the role of personality and the situational living context in reports of past and present life events and circumstances. By personality, we refer to the complex of attributes that characterize a unique individual over time, which includes autobiographical knowledge, self-perception and self-identity. From the constancy necessary in the accumulation and integration of a stable self, we can predict secondly that individual differences in personality correlate with differences in type or style of cognitive as well as of emotional processes. Because differences of personality can be measured reliably by questionnaires, personality inventories can be used as a direct test of personality. From this it follows that we should also be able to find relationships between life-transition reports and personality measures. Indeed, such are the results of the studies we report below. Thirdly, quite apart from personality variables, we should also be able to demonstrate correlations between the current living context and retrospective life-transition reports. The importance of shedding light into these processes has been pointed out by different authors (Baltes, Lindenberger, & Staudinger, 1998; Bluck, 2003). And again, we are able to present empirical evidence below for this prediction.

222


3. RELATIONSHIP BETWEEN BIOGRAPHICAL RECOLLECTION, PERSONALITY, LIVING CONTEXT AND WELL-BEING: THREE ILLUSTRATIVE STUDIES ON MIDDLE AND OLD AGE In this contribution we want to give illustrative examples on how autobiographical reports of past life transitions or life events and actual well-being are related, and how personality and specific living contexts, such as going through a particular biographical transition like marital separation or dismissal from work may have an impact on autobiographical recollection. The findings reported here are based on data from two research programs: (1) from the ‘‘Basle Interdisciplinary Study on Ageing’’ (IDA-Study1) focusing on old persons (Perrig–Chiello, Perrig, Sta¨helin, Krebs, & Ehrsam, 1996) and (2) from the study ‘‘Transitions and Life Perspectives in Middle Age’’2 focusing persons in middle age (Perrig-Chiello, Ho¨pflinger, Kaiser, & Sturzenegger, 1999). Based on the theoretical rationale presented above, we analyzed the relationships between the emotional valence of autobiographical memories, actual well-being, personality and actual living context. We hypothesize that: (a) There is a positive relationship between the emotional valence of retrospective recollected autobiographical memories and current wellbeing. (b) This relationship is due on the one hand to inherent personality variables and on the other to specific living-context factors, which mediate the effects. 3.1. The Impact of Personality on Biographical Reports and Well-Being in Middle and Old Age (Study 1) 3.1.1. Evidence from Research on Old Age The interview data reported here are original and were collected through the Basel Interdisciplinary Study on Ageing (IDA-Study) (Perrig-Chiello et al., 1996). The aim of the IDA-Study was the examination of the cross-sectional and longitudinal changes of health, well-being and autonomy in old age, and their determinants. The project was specifically designed to document the availability of physical, psychological and social resources, and to investigate their impact on these three outcome variables (health, well-being


223

and autonomy). The project is a follow-up of a longitudinal study that started in 1960 and takes into account the data collected in 1993. Participants in the IDA-study had to be 65 years or older at the time of interview. They were selected by random sampling from the longitudinal pool, which still comprised 3,768 people in 1993. For the IDA-study, 442 healthy persons aged 65–94 (309 males and 133 females, mean age: 74.95 years) agreed to participate. Psychological assessment included: (a) Life event inventory: A list of thematically ordered life events was presented to the participants (own health, housing condition, financial condition, family concerns, activities, etc.) (see the appendix). For each of these domains they were asked to indicate whether and how they had experienced specific events within the last 10 years. (b) Personality: In order to assess personality traits we used the two main subscales ‘‘Extraversion’’ and ‘‘Neuroticism’’ from the Freiburger Perso¨nlichkeits-Inventar (Fahrenberg, Hampel, & Selg, 1970, 1984) (being the most used personality inventory in German-speaking countries). (c) Psychological well-being was assessed by means of a 9-item-scale which taps the dimensions ‘‘satisfaction with own past’’, ‘‘purpose of life’’ and ‘‘mastery’’. This instrument meets all psychometric standards (see Perrig-Chiello, 1996, 1997). In order to answer the specific question about the relationship between the number of the reported positive and negative life events, personality variables and current well-being, we first performed correlational analyses. Results show that the reported number of positive life events is negatively correlated with neuroticism (R ¼ 0:132; po0.05) and positively with psychological well-being (R ¼ 0:118; po0.05), whereas negative life events are positively correlated with neuroticism (R ¼ 0:123; po0.05). In a second step, we calculated multiple-regression analyses in order to determine the predictive power of the reported number of positive and negative life events on actual psychological well-being. Since in our data we had an overrepresentation of men, we controlled also for gender. Results show that the number of reported negative and positive life events is indeed significant predictor of psychological well-being, independently of gender. However, as soon as neuroticism is introduced as additional predictor, the number of reported life events loses its predictive power in favor of neuroticism (see Table 1). Even though these effects are not very large, they are reliable. Furthermore, the same phenomenon was demonstrated in another study with a middle-aged sample, as described in the following section.

224


Table 1.

Life Events and Neuroticism: Which Measure is Better as Predicting Current Well-Being?. Predictors of Well-Being

Standard b (p)

Model 1

Number of negative life events Number of positive life events Gender N ¼ 301

0.11 (0.03) 0.16 (0.001) 0.05 (n.s.) R ¼ 0:17; R2 ¼ 0:03; p ¼ 0:004

Model 2

Neuroticism Number of negative life events Number of positive life events Gender N ¼ 301

0.256 (o0.0001) 0.02 (n.s.) 0.087 (n.s.) 0.02 (n.s.) R ¼ 0:28; R2 ¼ 0:08; po0.0001

3.1.2. Evidence from Research on Middle Age The findings reported here stem from an interdisciplinary longitudinal study entitled ‘‘Transitions and Life Perspectives in Middle Age’’ (Perrig-Chiello et al., 1999). The study investigated the impact of past transitions on current well-being and on anticipation of old age. A total of 268 middle-aged persons (197 women; 71 men) participated in the study (mean age ¼ 47:2 years). This sample is a subsample of a larger survey study (N ¼ 1015) and can be considered as being representative of a healthy middle-aged urban population in Switzerland. Participants in the study completed two questionnaires (psychological well-being and personality) and were given an indepth interview on biographical transitions (timing and emotional valence of transitions) (Perrig-Chiello & Ho¨pflinger, 2001). The crucial variables were assessed as follows: (a) Reported biographical transitions: A list of age-normed and frequent non-age-normed transitions was presented and participants had to indicate the age at which the specific transition occurred, as well as the emotional valence of this event as experienced at that time. The list referred to transitions across the whole life span (from school entry, puberty, first love through anticipation of retirement and transition to old age). (b) Psychological well-being was assessed by means of the same nine-item test as in the IDA-Study mentioned previously. (c) Personality was assessed by means of the NEO-Five Factor Inventory (Costa & McCrae, 1985).


225

Regression analyses were performed to evaluate the predictive power of emotional valence of experienced transitions along with personality variables on present psychological well-being. Age group and gender, emotional valence of past transitions as well as personality served as predictor variables. We performed hierarchical regression analyses; age group and gender were entered as block 1 (R2 ¼ 0:05), emotional valence of transitions as block 2 (R2 ¼ 0:13) and personality variables as block 3 (R2 ¼ 0:42). The analyses revealed that, in particular, emotional valence of one transition, namely puberty, was a significant predictor of actual psychological wellbeing (b ¼ 0:18; po0.05). However, as soon as personality variables were entered, psychological well-being was best predicted by neuroticism (b ¼ 0:53; po0.01) and by conscientiousness (b ¼ 0:18; po0.05). By entering these two variables, the predictive power of the remembered emotional valence of past transitions disappeared: only low scores on neuroticism and high scores on conscientiousness were associated with high psychological well-being (see also Perrig-Chiello & Perren, 2005). On the one hand, our results suggest that recollection of past transitions, especially the transition to adulthood, is significantly related to current wellbeing. In fact, it has been suggested that adolescence can be seen as an anchor point from which we begin the story of our adult lives and that the memories from this period help us define who we are later in life (Fitzgerald, 1988; Fitzgerald & Shifley-Grove, 1999; Josselson, 1987; Rybash, 1999). On the other , however, our results also showed that the predictive power of the remembered emotional valence of this specific transition vanishes as soon as personality variables are considered as predictors for current psychological well-being. These results indicate that the reconstruction of one’s own past, as well as the perception of one’s present state may both be driven by personality factors.

3.2. Personality Predicts not only Biographical Memories but also Episodic Recollection (Study 2) As described in the two precedent studies, personality factors (especially neuroticism) can be significant and robust predictors for biographical recollection. Now, one could conclude that biographical reports are mere products of motivated recollection and are consequently biased and unreliable. Furthermore, a memory bias of this kind might stand in contrast to more cognitive forms of episodic recollection. The latter kind of remembering might be considered to be free of influence of personality factors, and

226


hence, trust worthier. However, this assumption may be erroneous, and it is reasonable to assume that personality as the individual bundle of complex features of an integrated stable self is determining cognitive processes like remembering in general as well. The relationship between objective memory performance and personality factors is, at best, rather obscure. For example, only a few investigations to date have looked at the relation between personality and episodic memory (Perrig-Chiello, Perrig, & Sta¨helin, 2000; Meier, Perrig-Chiello, & Perrig, 2002), and there is a growing conviction that personality variables are interwoven with memory outcomes (Cavanaugh & Green, 1990). Gold and Arbuckle (1990) proposed a model of personality–cognition relations in old age. Recurring to the ‘‘big 5’’ central measures (introversion–extraversion, neuroticism, openness, agreeableness and conscientiousness), they hypothesized that most of these traits have both direct and indirect effects on memory. Furthermore, they argue that, in later life, senescent changes in the nervous system, coupled with reductions in externally driven cognitive demands, can result in increased influences of personality on cognitive functioning. These effects are viewed not as direct determinants of specific skills or processes, but rather as having a more general facilitative or detrimental effect on cognition. Neuroticism, in particular, has been associated with negative outcome expectations and with poorer cognitive functioning (Gold & Arbuckle, 1990). Results from these studies suggest that this approach could be a very promising way to study interindividual differences in episodic memory. Within the IDA-Study we had the opportunity to examine the impact of personality on episodic memory performance in a sample of 287 healthy adults aged 68–95 years. The purpose was to examine the contribution of extraversion and neuroticism to the explanation of episodic memory variability in old age. To assess personality, we used the two main subscales ‘‘Extraversion’’ and ‘‘Neuroticism’’ from the Freiburger Perso¨nlichkeitsInventar (Fahrenberg et al., 1970). For episodic memory assessment we used a computerized test (Perrig et al., 1994). This test allows to assess different memory dimensions (working memory, recognition and free recall). We expected higher extraversion and lower neuroticism to be related to better memory performance. Consistent with our expectations, extraversion was associated with higher (b ¼ 0:16; po0.01), and neuroticism was associated with lower levels of episodic memory performance (b ¼ 0:18; po0.01) (for more details see Meier, Perrig-Chiello, & Perrig, 2002). In conclusion, and in line with previous research, our results indicate that the relationship between memory and personality is not very strong, but consistent (Arbuckle, Gold, Andres, Schwartzman, & Chaikelson, 1992;


227

Hultsch, Hertzog, Small, & Dixon, 1999). Even though the relationship between personality and memory has not been studied extensively, this relationship exists and deserves to be studied more fully. Overall, we conclude, that both autobiographical and episodic memory recollection, in general, can be predicted by personality variables. In other words, we believe that recollection – whether episodic or semantic – are all in some way influenced by personality factors. But this explains only a part of the variance of the phenomenon. In order to gain a better appreciation of the determining factors of biographical recollection, other predictors, such as actual living context, should be taken into account. The influence of such factors is described in the following study.

3.3. The Importance of Living Context on Biographical Recollection (Study 3) With respect to the results gathered so far and considering the fact that the living context is thought to play an important role in the prediction of biographical and episodic memory outcomes, we took the opportunity to test this possibility using extreme group comparisons.3 We compared two groups of middle-aged women living in two different contexts. One group consisted of an ‘‘average’’ sample of middle-aged women (92 women from study 2, mean age: 52 years); the other group was a subclinical sample of women of the same age (45 women, mean age 51.56 years). The latter were all new recruits, receiving a psychotherapeutic treatment at the time of data collection in various outpatient clinics, for problems related to a specific critical life course transition, namely the loss of a partner (separation/ divorce, N ¼ 35; death of spouse, N ¼ 6; combination of loss of partner and dismissal from work, N ¼ 5). Common to all these women were symptoms of grief due to the loss of a partner. Assessment instruments and procedure were the same as described in Section 3.1.2. We expected that the problem-loaded context of the subclinical group (being in a difficult partnership transition) would lead to an overall pessimistic view of past biographical events. In accordance with empirical results from studies of mood-dependent information processing (having persons with negative mood reporting more negative events (Perrig & PerrigChiello, 1988; Eich & Macaulay, 2000). Together with what had been stated in the introduction part about the function of reconstructive memories and motivated remembering, we hypothesized that the subclinical group would report more negative autobiographical events, and that the reported

228


emotional valence of their recollection would be overall more pessimistic than that of the control group. According to our expectations, the subclinical group did indeed report having experienced significantly more negative transitions than the control group (see Table 2). The emotional valence of the reported family transitions, e.g. was significantly more negative in the subclinical group compared to the controls. In particular, transitions concerning the own offspring were rated much more negatively by the subclinical compared to the control group (e.g. birth of the first child and departure of the last child). However, this phenomenon and interpretation of pessimistic biographical recollection by the subclinical group has to be restricted, due to one exception. This exception concerns the remembered partnership translations namely, ‘‘First love’’: Women in the subclinical group undergoing a difficult partnershiprelated transition at the present time reported a significantly higher

Table 2. Reported Emotional Valence of Passed Transitions: Comparison Subclinical Sample versus Average Group. Emotional Valence of Reported Transitions

Subclinical Group (N ¼ 45) Mean/SD

Average Group (N ¼ 92) Mean/SD

Z

Df

t

p

Number of negative transitions Emotion. valence of transitions (total) Emotion valence of family transitions Puberty First job First love Departure from home Marriage Pregnancy 1 Birth 1 Separation Departure of last child Menopause Death of mother Death of father

3.8/2.01

2.91/2.01

0.89

135

2.43

0.017

6.37/1.16

6.68/1.16

0.30

134

1.435

0.154

6.05/1.12

6.77/2.02

0.72

129

2.216

0.028

5.04/2.43 7.18/2.81 8.87/1.71 7.47/2.92 7.82/2.4 7.93/2.74 5.76/3.60 4.23/3.59 3.38/2.68

4.87/2.38 7.8/2.42 8.08/2.34 7.33/2.80 7.87/2.43 8.18/2.71 7.51/2.87 3.40/2.99 5.84/2.79

0.18 0.62 0.79 0.14 0.04 0.25 1.75 0.83 2.47

134 133 132 132 125 111 111 54 33

0.404 1.333 2.001 0.271 0.097 0.480 2.836 0.945 2.651

0.687 0.185 0.047 0.787 0.923 0.632 0.005 0.349 0.012

4.63/2.94 2.69/2.39 2.7/2.38

5.64/2.03 3.04/1.78 3.23/2.02

1.01 0.35 0.54

94 40 87

1.917 0.545 1.130

0.058 0.589 0.261


229

(positive) emotional valence of this romantic biographical event in the past compared to the control group (see Table 2). How can we explain this seemingly contradictory result? The large majority of women in the subclinical group were in the process of mourning due to separation, divorce or death from/of their spouse. According to the interview data, in most of the cases, the lost spouse actually was their first love. Furthermore, this first love was remembered as being highly positive, and compared to other people’s first love, ‘‘much better’’ (social comparison: 6.42 versus 5.32, p o0.01). Could this be a compensating mechanism – a self-serving biased recollection – aimed at reducing self-doubts concerning this new broken long-term relation? Or, for the women to give a plausible socially accepted explanation for their excessively strong, psychological reaction to that loss? In conclusion, we can say that there is a difference between knowing and remembering our past. The current living context not only determines perception of the actual situation but systematically bias reports on earlier life events – for good or bad. Remembering our past goes beyond reporting autobiographical facts, and implies reconstructive recollection and interpretation, motivated by the perception, that is the interpretation or meaning of the actual situation. Again, we would emphasize that these processes need not to be deliberate or accompanied by conscious insight. Especially in periods of change, growth, when new challenges arise, biographical memory seems to be tuned to operate selectively (Fitzgerald & Shifley-Grove, 1999). Distortions and self-serving biases are inevitably associated with this recollective process.

4. DISCUSSION The aim of this contribution was to demonstrate possible psychological sources of biased remembering of autobiographical events and their relation to well-being based on data from two research programs on middle and old age. Our results suggest that the recollection of passed life events is associated with personality variables (especially neuroticism and extraversion). Current well-being is related to the emotional valence of biographical recollection. This latter relationship, however, seems to be mediated by the personality variables, because it disappears as soon as personality variables are introduced into the regression analysis, at least for episodic memory performance, in general. Our finding fits well with the theoretical framework of constructive processes in perception and memory (Bartlett, 1932/1993;

230


Schacter, 1996). Personality ‘‘schemata’’ and personality ‘‘attitudes’’, or, in other words, the organized set of autobiographical knowledge that constitutes our self-identity, relates past and present well-being in a consequent meaningful way. Finally, from a developmental perspective, several authors have also underlined the importance of subjective reconstruction of life for current and future well-being. For example, Erikson (1959) had stated that the ability to really see one’s self requires a continuous perspective, both in retrospect and prospect. It mandates linking the presently understood past and the anticipated future with the experiences present in the individual. In a similar vein, for McAdams (1993) life stories bind together events in time, organizing present reality by connecting past and future. These stories are less about facts and more about meaning. In the subjective telling of the past, the past is constructed. Butler (1963) has postulated the adaptive function of life review: Achievement of integrity in such a life review promotes successful aging. Life review has the function of equilibrating the sense of self-worth, coherence and of reconciliation of one’s past and present. In this sense, the remembered emotional valence of past transitions should not (and probably cannot) be measured exclusively in terms of true/not true, but rather in terms of preserving a sense of coherence, of continuity of the ‘‘self’’. Ryff (1991) has suggested that comparisons of the self in the present with the self in the past constitute an important source for equilibrating actual well-being for adults (see also Suls & Mullen, 1982). And finally, Keyes and Ryff (1999) have pointed out that this is how people construe their experiences, not just experience per se, that matters. The meaning people attach to some events reflects a concern for maintaining a favorable image. Our results also provide additional empirical evidence that not only autobiographical recollection is influenced by personality, but episodic memory in general. From our findings presented here, one might assume that personality predicts better episodic recollection in memory tests, while the actual living context is more predictive in autobiographical reports. Hence, in order to predict subjective biographical recollection more accurately, actual well-being and living context have to be taken into account. In fact, our results on extreme group comparison show that retrospective recollection and evaluation of past transitions reflect not only conscious phenomenological reexperiencing of real-life events, but also a person’s motivation of reconstructing his/her life in order to gain a sense of coherence and control. In this way, autobiographical reports can be considered to be the result – to a substantial degree – of an internal dynamics of the psyche, and as such, they may not be available to conscious introspection. According to Schacter (1996), the


231

complex mixture of personal knowledge that we retain about our past is woven together to form our life stories. These are the biographies of self that provide narrative continuity between past and future – a set of memories that form the core of personal identity. On how exactly personality does influence the emergence of actual living conditions, or on how exactly personality relates to cognitive processes – or vice versa – must remain open here, but should be a most important research topic for the future. To unwrap this complex package of hidden causalities provides a formidable challenge for psychological investigations and promises large steps forward in understanding better the human being.

NOTES 1. National Research Program 32 ‘‘Ageing’’, Swiss National Science Foundation. 2. Swiss Priority Program ‘‘Switzerland toward the Future’’. 3. Original data based on a master thesis (Rusca, 2003), performed within the research program ‘‘Transitions and Life Perspectives in Middle Age’’ under supervision of the first author.

ACKNOWLEDGMENTS This research was supported by grants by the Swiss National Science Foundation (No. 4032-035642 and No. 5004-058457). We wish to express our gratefulness to Dr. Josephine Cock, University of Berne, for her helpful comments on an earlier draft.

REFERENCES Arbuckle, T. Y., Gold, D. P., Andres, D., Schwartzman, A., & Chaikelson, J. (1992). The role of psychosocial context, age, and intelligence in memory performance of older men. Psychology and Aging, 7, 25–36. Ayers, M. S., & Reder, L. M. (1998). A theoretical review of the misinformation effect: Predictions from an activation-based memory model. Psychonomic Bulletin and Review, 5(1), 1–21. Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (1998). Life-span theory in developmental psychology. In: W. Damon & R. M. Lerner (Eds), Handbook of child psychology: Theoretical models of human development (Vol. 1, pp. 321–356). New York: Wiley. Bartlett, F. C. (1932/1993). Remembering. Cambridge: Cambridge University Press.

232


Bluck, S. (2003). Autobiographical memory. Exploring its function in everyday life. Memory, 11, 113–123. Bluck, S., & Alea, N. (2002). Exploring the functions of autobiographical memory: Why do I remember the automn? In: J. D. Webster & B. K. Haight (Eds), Critical advances in reminiscence: From theory to application (pp. 61–75). New York: Springer. Butler, R. N. (1963). The life review: An interpretation of reminiscence in the aged. Psychiatry, 26, 65–76. Brainerd, C. J., Reyna, V. F., Wright, R., & Mojardin, A. H. (2003). Recollection rejection: False-memory editing in children and adults. Psychological Review, 110(4), 762–784. Brewer, W. F. (1988). Memory for randomly sampled autobiographical events. In: U. Neisser & E. Winograd (Eds), Remembering reconsidered: Ecological and traditional approaches to the study of memory. New York: Cambridge University Press. Cavanaugh, J. C., & Green, E. E. (1990). I believe, therefore I can: Self-efficacy beliefs in memory aging. In: E. A. Lovelace (Ed.), Aging and cognition. Mental processes, selfawareness and interventions. Amsterdam: North-Holland. Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American Journal of Sociology, 96, 805–842. Conway, M. A. (1987). Verifying autobiographical facts. Cognition, 26(1), 39–58. Conway, M. A. (1996). Autobiographical knowledge and autobiographical memories. In: D. C. Rubin (Ed.), Remembering our past. Studies in autobiographical memory (pp. 67–94). New York: Cambridge University Press. Conway, M. A., & Rubin, D. C. (1993). The structure of autobiographical memory. In: A. E. Collins, S. E. Gathercole, M. A. Conway & P. E. M. Morris (Eds), Theories of memory (pp. 103–137). Hove, Sussex, UK: Erlbaum. Costa, P. T., & McCrae, R. R. (1985). Manual for the NEO personality inventory. Odessa, FL: Psychological Assessment Resources, Inc. Eich, E., & Macaulay, D. (2000). Are real moods required to reveal mood-congruent and mooddependent memory? Memory & Cognition, 20, 277–290. Elder, G. H. (1998). The life course as developmental theory. Child Development, 69, 1–12. Erikson, E. H. (1959). Identity and the life cycle. New York: International University Press. Fahrenberg, J., Hampel, R., & Selg, H. (1970, 1984). Freiburger Perso¨nlichkeits-Inventar. Go¨ttingen: Hogrefe. Fitzgerald, J. M. (1988). Vivid memories and the reminiscence phenomenon: The role of a self narrative. Human Development, 31, 261–273. Fitzgerald, J. M., & Shifley-Grove, S. S. (1999). Memory and affect: Autobiographical memory distribution and availability in normal and recently detoxified alcoholics. Journal of Adult Development, 6, 11–19. Freud, S. (1917). Vorlesungen zur Einfu¨hrung in die Psychoanalyse. Reprint, Frankfurt: Fischer, 1981. Gardiner, J. M., & Java, R. I. (1990). Recollective experience in word and nonword recognition. Memory & Cognition, 18, 23–30. Gold, D. P., & Arbuckle, T. Y. (1990). Interactions between personality and cognition and their implications for theories of aging. In: E. A. Lovelace (Ed.), Aging and cognition. Amsterdam: North-Holland. Hultsch, D. F., Hertzog, C., Small, B. J., & Dixon, R. A. (1999). Use it or lose it: Engaged lifestyle as a buffer of cognitive decline in aging? Psychology and Aging, 14, 245–263. Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological Bulletin, 114, 3–28.


233

Josselson, R. (1987). Finding herself: Pathways to identity development in women. San Francisco: Jossey-Bass. Keyes, C. L. M., & Ryff, C. D. (1999). Psychological well-being in midlife. In: S. L. Willis & J. D. Reid (Eds), Life in the middle. Psychological and social development in middle age (pp. 161–182). San Diego: Academic Press. Loftus, E. F. (2003). Make-believe memories. American Psychologist, 58(11), 867–873. McAdams, D. P. (1993). The stories we live by: Personal myths and the making of the self. New York: Morrow. Meier, B., Perrig-Chiello, P., & Perrig, W. (2002). The impact of personality on memory performance in elderly. Aging, Neuropsychology, and Cognition, 9(2), 135–144. Mercer, R. T., Nichols, E. G., & Doyle, G. C. (1989). Transitions in a woman’s life. Major life events in developmental context. New York: Springer. Perrig, W. J., & Perrig-Chiello, P. (1988). Mood congruity effects in absence of mood. Memory & Cognition, 16, 102–109. Perrig, W., Kling, V., Meier, B., Hofer, D., Perrig-Chiello, P., Ruch, M., & Serafin, D. (1994). Computerunterstu¨tzter Geda¨chtnis-Funktions-Test (GFT). Version 2.0 Manual und Diskette. Institut fu¨r Psychologie, Universita¨t Basel. Perrig-Chiello, P. (1996). Wie ko¨nnen wir Wohlbefinden erfassen? In: H. W. Heiss, F. Huber, B. Peter & H. B. Sta¨helin (Eds), Wohlbefinden im Alter – geriatrische und gerontologische Strategien (pp. 7–17). Reinach: Roche Pharma. Perrig-Chiello, P. (1997). Ressourcen des Wohlbefindens im Alter. Weinheim: Juventa. Perrig-Chiello, P., & Ho¨pflinger, F. (2001). Zwischen den Generationen – Frauen und Ma¨nner im mittleren Lebensalter. Zu¨rich: Seismo-Verlag. Perrig-Chiello, P., Ho¨pflinger, F., Kaiser, A., & Sturzenegger, M. (1999). Psychosoziale Aspekte der Lebensbedingungen von Frauen und Ma¨nnern im mittleren Lebensalter. Zeitschrift fu¨r Familienforschung, 11, 5–27. Perrig-Chiello, P., & Perren, S. (2005). Biographical transitions from a midlife perspective. Journal of Adult Development, 12(3), 169–181. Perrig-Chiello, P., Perrig, W. J., & Sta¨helin, H. B. (2000). Differential aspects of memory selfevaluation in old and very old people. Ageing & Mental Health, 4(2), 130–135. Perrig-Chiello, P., Perrig, W. J., Sta¨helin, H. B., Krebs, E., & Ehrsam, R. (1996). Autonomie, Wohlbefinden und Gesundheit im Alter: Eine interdisziplina¨re Altersstudie (IDA). Zeitschrift fu¨r Gerontologie und Geriatrie, 29, 95–109. Rubin, D. C. (1986). Autobiographical memory. Cambridge: Cambridge University Press. Rubin, D. C., & Berntsen, D. (2003). Life scripts help to maintain autobiographical memories of highly positive, but not highly negative, events. Memory & Cognition, 31(1), 1–14. Rusca, E. (2003). Soziobiographische Ressourcen und Wohlbefinden von Frauen im mittleren Lebensalter in einer als kritisch erlebten Transition. Lizentiatsarbeit, Institut fu¨r Psychologie, Universita¨t Bern. Rutter, M. (1996). Transitions and turning points in developmental psychopathology: As applied to the age span between childhood and mid-adulthood. International Journal of Behavioral Development, 19, 603–626. Rybash, J. M. (1999). Aging and autobiographical memory: The long and bumpy road. Journal of Adult Development, 6, 1–10. Ryff, C. D. (1991). Possible selves in adulthood and old age: A tale of shifting horizons. Psychology and Aging, 6, 286–295.

234


Ryff, C. D., & Heidrich, S. M. (1997). Experience and well-being: Explorations on domains of life and how they matter. International Journal of Behavioral Development, 20, 193–206. Schacter, D. L. (1996). Searching for memory. The brain, the mind, and the past. New York: Basic Books. Schwarz, N., & Strack, F. (1999). Reports of subjective well-being: Judgmental processes and their methodological implications. In: D. Kahneman, E. Diener & N. Schwarz (Eds), Well-being. The foundations of hedonistic psychology (pp. 61–85). New York: Russell Sage Foundation. Sugarman, L. (2001). Life-span development. Frameworks, accounts and strategies (2nd ed.). East Sussex: Psychology Press. Suls, J., & Mullen, B. (1982). From the cradle to the grave: Comparison and self-evaluation across the life-span. In: J. Suls, (Ed.), Psychological perspectives on the self (Vol. 1, pp. 97–125). Hillsdale, NJ: Erlbaum. Tulving, E. (1983). Elements of episodic memory. Oxford: Oxford University Press. Vaillant, G. E. (1990). Avoiding negative life outcomes: Evidence from a forty-five year study. In: P. B. Baltes & M. M. Baltes (Eds), Successful aging: Perspectives from the behavioural sciences (pp. 323–358). New York: Cambridge University Press. Wadsworth, M. E. J., Maclean, M., Kuh, D., & Rodgers, B. (1990). Children of divorced and separated parents. Family Practice, 7, 104–109. Wertlieb, D. (1997). Children whose parents divorce: Life trajectories and turning points. In: I. H. Gotlib & B. Wheaton (Eds), Stress and adversity of the life course (pp. 179–197). Cambridge: Cambridge University Press. Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The frontal lobes and autonoetic consciousness. Psychological Bulletin, 21(3), 331–354. Whittlesea, B. W. A. (2002). False memory and the discrepancy-attribution hypothesis: The prototype-familiarity illusion. Journal of Experimental Psychology: General, 131(1), 96–115.

APPENDIX. LIFE EVENT INVENTORY

Basle Interdisciplinary Study on Aging (IDA-Study) Life Event Inventory Which important events, positive as well as negative, come to mind if you think back the last ten years? I will provide some topics/themes. Please tell me in each case which event(s) come to mind and how you experience those events at present.

235

The Impact of Personality and Living Context positive ambivalent

Employment, work ..................... (i.e. retirement or retirement of one's ... spouse/partner, change of job)..............

Illness and Health ................................ (i.e. injury, illness concerning oneself, or one's spouse/partner, children).........

Residence .............................................. (move, relocation i.e. move to another . town, renovation) ...................................

Death ...................................................... (spouse/partner, children, other family . members) ................................................

Partnership and Familial situation ... (separation, divorce, .............................. grandparenthood) ...................................

Financial situation ............................... .................................................................. .................................................................. Recreation, entertainment .................. (travel, social activities).........................

negative

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0

2 2 2

1 1 1

0 0 0


236

STUDYING LIVES IN TIME: A NARRATIVE APPROACH Dan P. McAdams Social scientists conceive of the human life course in many different ways. Life-span developmental theorists like Erikson (1963) and Levinson (1978) imagine a sequence of stages through which the individual passes on the journey of life. Each period or season offers its own distinctive challenges, and each individual confronts those challenges in a predetermined and predictable order. More sociologically informed theorists invoke social roles, generational cohorts, and historical events in their conceptions of human lives in time, suggesting that development is rather too contingent and contextual to conform to any predetermined sequence (Dannefer, 1984; Elder, 1995). The person moves through time along a dynamic trajectory that results from the complex interplay between human agency and the manifold forces of social structure. A third approach in the social sciences argues for the primacy of personal traits in shaping the life course (McCrae & Costa, 1990; Roberts & Pomerantz, 2004). Human beings differ with respect to broad psychological tendencies that are surprisingly consistent, stable, and pervasive over developmental time. While contexts may change, personal dispositions continue to exert their steady and predictable influence. Conceptions of the life course that emphasize developmental stages, contingent trajectories, and personal traits all have considerable value in making sense of how individual lives evolve over time. Stage theories spell out common developmental demands that people face in many different cultures


237

238

DAN P. McADAMS

and contexts. Theories emphasizing context and contingency in the life course pay careful attention to economic, social, cultural, and historical forces that impact lives in variable ways, and how individuals react to and sometimes resist those forces. Trait conceptions underscore the importance of human individuality, and how people construct lives that are consistent with their internal dispositions. For all their value, however, these three major approaches to the life course pay only passing attention to the subjective meanings of lives. How do individuals themselves make sense of their lives in time? What meanings do persons construct for themselves as they move through developmental stages, confront the vagaries of off-time and on-time challenges in the life course, and act in accord with their internal dispositions? The current chapter affirms a fourth approach to the life course, one that emphasizes the meanings that people make of their own lives in time. The chapter’s thesis is that those meanings often take the form of stories (Cohler, 1982; McAdams, 1985; Polkinghorne, 1988). People construe their own lives as evolving stories that aim to reconstruct the past and imagine the future in meaningful and coherent ways. Narrative approaches to the life course see persons as storytellers, see lives as stories told, and see the life course as a psychosocial construction reflecting both personal inclinations and the narrative conventions and traditions that prevail in a given society.

PERSONALITY, NARRATIVE, AND THE LIFE COURSE The last two decades have witnessed an upsurge of interest among scholars and social scientists in the role of stories in social life. Narrative approaches have had a major impact in sociology (Holstein & Gubrium, 2000), criminology (Maruna, 2001), gerontology (Birren, Kenyon, Ruth, Shroots, & Svendson, 1996), organizational studies (Gabriel, 2000), educational research (Casey, 1996; Ely, 2003), theology and moral philosophy (MacIntyre, 1984), developmental psychology (Fivush & Haden, 2003), social psychology (Gergen, 1991; Murray & Holmes, 1994), clinical psychology (Angus & McLeod, 2004; Lieblich, McAdams, & Josselson, 2004), and the cognitive sciences (Conway & Pleydell-Pearce, 2000; Schank & Abelson, 1995). Bruner (1990) and Sarbin (1986) deem narrative to be a new and powerful root metaphor for the social sciences. Josselson and Lieblich (1993) describe the narrative study of lives as a loosely organized, interdisciplinary endeavor

Studying Lives in Time: A Narrative Approach

239

aimed at telling and analyzing the stories people construct to make sense of social life in time, with some special emphasis given over to narrative accounts, such as those told by women and people of color, that have historically been suppressed or marginalized (see also Franz & Stewart, 1994; Rosenwald & Ochberg, 1992). The many different approaches today in social science incorporating narrative range from those employing storytelling methodologies to examine social behavior (e.g., Baumeister, 1994; Woike, 1995) to those promoting narrative theories about social life itself, viewing lives as ongoing stories that are told, revised, and retold in culture (e.g., Gregg, 1991; Hermans, 1996). The particular narrative approach to be described in the current chapter evolved within the field of personality psychology (McAdams, 1985), though its origins were strongly influenced by the approaches in life-span developmental psychology (Erikson, 1963), life course sociology (Bertaux, 1981), Sartrean philosophy (Charme, 1984), and studies of narrative and biography in the humanities (Edel, 1978; Langbaum, 1982; Ricoeur, 1984). Personality psychology views itself as the scientific study of the whole person (Allport, 1937; McAdams, 1997). The aim of personality psychology is to provide a scientifically credible account of human individuality – how the individual person is (a) like all other persons, (b) like some other person, and (c) like no other person (Kluckhohn & Murray, 1953). As such, personality psychology considers both species-typical characteristics of human nature and individual differences between people. Research in the field, however, tends to focus mainly on individual differences. What are the most important individual differences in psychological and social life? Current thinking in personality psychology tends to focus on three different classes or levels of individual differences – dispositional traits, characteristic adaptations, and integrative life stories (Hooker, 2002; Hooker & McAdams, 2003; McAdams, 1995, 2006a; Sheldon, 2004). From the standpoint of personality psychology, then, life narrative holds a particularly prominent position within an evolving constellation of characteristics – including traits and adaptations – that specify how a person is similar to and different from other persons. Within the constellation of human individuality, the most recognizable and longitudinally consistent aspect of the person is his or her dispositional traits (Matthews, Deary, & Whiteman, 2003; McCrae & Costa, 1990). Dispositional traits are those broad, stylistic, and linear dimensions of human individuality, bearing names such as ‘‘extraversion/introversion’’ and ‘‘depressiveness,’’ that account for general consistencies in behavior, thought, and feeling across situations and over time. While traits were once

240

DAN P. McADAMS

the object of considerable derision among those psychologists who emphasize the role of situations in shaping behavior (e.g., Mischel, 1968), traits today are the coin of the realm in personality research. The scientific viability of the trait concept comes from many sources. First, a consistent body of research shows that trait attributions based on careful observations reflect real differences in the behavior and personalities of the people about whom the attributions are being made (Funder & Colvin, 1991). Second, research consistently shows that scores on self-report trait scales predict general consistencies in behavior when sampled across different situations and over time (Kenrick & Funder, 1988) and predict important outcomes such as wellbeing, health, and even longevity (Friedman et al., 1993; Matthews et al., 2003). Third, individual differences in self-report trait scores show considerable stability over developmental time, especially during the adult years (Costa & McCrae, 1994; Roberts & Del Vecchio, 2000). Fourth, twin studies suggest that traits are at least moderately heritable, with as much as 50–60% of the variance in trait scores being attributed to genetic differences between people (Bouchard, Lykken, McGue, Segal, & Tellegen, 1990). Fifth, 30 years of factor-analytic studies have resulted in an emerging consensus in the field of personality regarding the organization of the trait universe. Many personality psychologists today subscribe to the Big Five model of traits, grouping the many possible traits that might be invoked in describing human individuality into the five general categories of extraversion/introversion, neuroticism, conscientiousness, agreeableness, and openness to experience (Goldberg, 1993; John & Srivastava, 1999; McCrae & Costa, 1990). Showing considerable stability over time, the Big Five traits provide something akin to a dispositional signature for human lives. A highly extraverted person tends to express high levels of energy and social interest, enjoys being with people, and seeks opportunities for social stimulation to a greater extent, on average, than does a highly introverted (low extraversion) person. If a person scores high on an extraversion scale at, say, age 30, he or she is likely to score fairly high again 20 years down the road, at age 50. If the person is low on extraversion at Time 1, chances are good that he or she will score low again at Time 2. In other words, people tend to hold their relative position in the normal distribution of scores for any given trait. Of course, this differential continuity in traits is not perfect. People can and do move around in the distribution over time (Roberts & Pomerantz, 2004). But the general trend is toward relative stability. Traits are well designed to show the interindividual stability of personality over the long haul. But this stability is of a very general kind, referring to a signature style of behavior, thought, and feeling. The style may be apparent very early on in


241

an interpersonal interaction, even within the first minute or two (Ambady, Hallahan, & Rosenthal, 1995). But as one gets to know a person better, one begins to see aspects of human individuality that appear to go beyond traits, aspects that are also subject to considerable change over the life course (McAdams, 1994). Personality begins with traits, but it does not end there. At a second level of human individuality, personality psychologists speak of personal motives, goals, values, interests, strategies, schemas, stagespecific concerns, and conditional patterns of behavior that are situated in time, place, and/or social roles (Cantor, 1990; Emmons, 1986; Freund & Baltes, 2000; Little, 1999; Mischel & Shoda, 1995; Schwartz, 1994; Winter, John, Stewart, Klohnen, & Duncan, 1998). These many different aspects of human individuality may be provisionally grouped under the label characteristic adaptations. If dispositional traits sketch an outline of human individuality, characteristic adaptations fill in some of the details. Though sometimes loosely related to dispositional traits, people’s goals, plans, schemas, coping strategies, and the like express the highly specific and personalized ways in which they have adapted to the variegated demands of social life. Unlike traits, many characteristic adaptations are expected to change substantially over the life course, as the circumstances of life change and as new developmental issues arise. For example, longitudinal and crosssectional studies show that adults’ concern for the next generation – what Erikson (1963) called the developmental task of generativity – rise and fall over the adult life course, reaching something of a peak in the middle-adult years (McAdams, 2001; McAdams, de St. Aubin, & Logan, 1993; Rossi, 2001; Vaillant & Milofsky, 1980). But traits and adaptations do not provide the full picture. People differ from each other in ways that go beyond their dispositional signature and their characteristic motivational and developmental concerns, and this becomes especially apparent as people move into and through their adult years. At a third level of human individuality, then, reside integrative life narratives (Hooker & McAdams, 2003; McAdams, 1995, 1996). These are the internalized and evolving stories of the self that people construct to make sense of their lives in time. These stories, or narrative identities (Singer, 2004), combine a person’s selective reconstruction of the past with his or her imagined anticipation of the personal future to produce a story with plot and characters to explain who I am, how I came to be, and where my life is going in the future. As Giddens (1991) writes, ‘‘a person’s identity is not to be found in behavior, nor – important though this is – in the reactions of others, but in the capacity to keep a particular narrative going’’ (p. 54). Narrative identities provide life with some degree of unity and purpose.

242

DAN P. McADAMS

As people develop over the life course, they rewrite their own self-defining life stories, albeit in typically implicit and unconscious ways. If, then, dispositional traits sketch the outline and characteristic adaptations fill in some of the details of human individuality, integrative life stories speak to what a person’s life means in the overall, set in subjective time. Personality itself, then, is a unique patterning of traits, adaptations, and stories evolving in a complex interpersonal, social, cultural, and historical context.

THE DEVELOPMENT OF THE STORY Stories are fundamentally about the vicissitudes of human intention organized in time (Bruner, 1990; Ricoeur, 1984). The developmental origins of storytelling, therefore, may be traced back to infancy. By the end of the first year of life, human infants appear to have a rudimentary sense of human intentionality, as indexed by their preference for observing and imitating goal-directed rather than random behaviors (Tomasello, 2000). By age 2, children have developed what Howe and Courage (1997) call an autobiographical self. They begin to construe the world from the standpoint of a subjective, appropriating self (what William James called ‘‘the I’’) who narrates and remembers personal events as episodes that happen to ‘‘me.’’ Parents typically assist children in the encoding, recollection, and telling of personal memories, encouraging their effort to narrate their experiences and acquainting them with the conventions for good storytelling (Fivush & Haden, 2003). By the time the children have reached their fourth birthday, they typically understand that stories contain characters who seek to enact their intentions, characters who want things and try to get them over time. With few exceptions, young children come to understand that people have minds, that minds contain desires and thoughts, and that people act upon those desires and thoughts in goal-directed ways (Wellman, 1993). By age 5, they are able to tell stories about the self that express this implicit understanding of human minds and human intentionality. By the age of 5, they have developed a story schema, or a set of expectations regarding what a story should look and sound like (Mandler, 1984). A story should be set in a particular time and place. It should involve a motivated character who sets out to get or do something. The character should eventually run into obstacles in the goal-directed action, which will lead to reactions and changes and ultimately some kind of a resolution. When stories do not conform to the schema, children find them odd, and they may seek to reformulate noncanonical narratives so that they fit the schema better.


243

Although children can tell stories about their own experiences by age 5, they do not have, nor are they really working on, narrative identities. Children do not have the developmental maturity, the social experiences, nor the cognitive skills to see their own lives as stories evolving over time, stories for which they are both the author and the protagonist. Beginning with Erikson (1963), many developmental theorists have argued that people do not construct full-fledged identities until adolescence or young adulthood (Arnett, 2000). McAdams (1985) suggests that what Erikson viewed as the challenge of identity formation in adolescence and young adulthood is largely about formulating a story for one’s life – what Sartre called a true novel (Charme, 1984) – that selectively reconstructs the past and imagines the future as an integrated temporal whole, to provide life with meaning and purpose and situate the person’s imagined life trajectory within a recognizable societal niche. As Hankiss (1981) suggested, the challenge of identity involves ‘‘mythologically re-arranging’’ one’s life history into a coherent narrative of self (p. 203). Similarly, Kohli (1981) suggested that the creation of a life story presupposes an advanced facility with autobiographical forms: ‘‘The autobiographical form presupposes a developed individuality, a selfconscious ‘I’ being able to grasp itself as the organizer of its own life history’’ (p. 64). A number of theorists contend that the full utilization of the autobiographical form typically awaits late adolescence. For example, Habermas and Bluck (2000) argue that the ability to construe one’s life as an evolving narrative of the self that integrates the reconstructed past and imagined future requires mastery of four different cognitive skills. First, the individual must be able to order events into temporal sequences, what Habermas and Bluck (2000) refer to as temporal coherence. Second, the individual must be able to assimilate his or her life to what society deems to be the typical course of life, showing what Habermas and Bluck (2000) call autobiographical coherence. Third, Habermas and Bluck identify causal coherence as the ability to link different autobiographical events into causal sequences to explain how a particular aspect of the self came to be. A teenager may explain her distrust of authority by telling how she used to trust her father completely, but how she gradually lost confidence in him through a series of events wherein he disappointed her. Finally, life storytelling requires what Habermas and Bluck call thematic coherence – the inductive derivation of themes or principles from a series of personal events. A man may come to conclude that ‘‘I am somebody who follows my deepest dreams’’ by noting how in many different events in the past he defied convention and acted in accord with his most cherished beliefs. Habermas and Bluck (2000) review research, showing that

244

DAN P. McADAMS

facility with causal and thematic coherence tends not to appear in selfnarration until the late teenage years. Equipped with a full complement of cognitive skills necessary for life narration and encouraged by their psychosocial environments to figure out who they are and how they are to fit into the adult world, older adolescents and young adults – what Arnett (2000) terms emerging adults – typically seek to arrange their lives into narrative identities. The move toward narrative identity in emerging adulthood is especially common in modern societies, wherein emerging adults are strongly urged to build a life that both expresses their individualized tastes and proclivities and conforms in some manner to the exigencies of a rapidly changing economic and cultural world (Gergen, 1991; Giddens, 1991; McAdams, 1996). The development of narrative identity in the emerging adult years occurs gradually, unevenly, and in the context of close personal relationships (McAdams, 1993; Thorne, 2000). Emerging adults try out different stories for their lives, imagine and tell past events and future goals in different ways, with different friends and family members, for different purposes and effects, in a social context wherein other emerging adults are doing pretty much the same thing. Stories are shared and refined. Stories are revised to reflect changing priorities, new insights, transformed goals and beliefs. Eventually, many people settle on a few key story lines, accentuate a set of key events from the past and downplay many others, imagine a circumscribed set of alternative and yet moreor-less realistic scenarios for the future, and gradually give their lives a narrative shape. Emerging adulthood is prime time for the development of narrative identity. But the process of constructing one’s life as a narrative does not end in one’s 20s. People continue to work on their life stories across the bulk of the adult life span (Birren et al., 1996; Cohler, 1982; Hooker & McAdams, 2003; McAdams, 1993). The midlife years, for instance, may be occasioned by considerable identity work for many modern adults. Life-stage theorists have written about how the realization that one’s life is now more than half over can bring to the psychosocial fore concerns about loss and mortality and can stimulate the actualization of long-suppressed tendencies, such as traditionally masculine tendencies in women and feminine tendencies among men (Gutmann, 1987; Levinson, 1978). Life course theories emphasize changing social roles and relationships in the midlife years and shifting contingencies in the social ecology of everyday life (Elder, 1995). Theorists of many different stripes tend to agree that midlife can be a period wherein considerable revision of one’s life story is likely to occur. Even as one’s traits continue to express continuity in selfhood, people change their narrative


245

understandings of the past, present, and future in response to a wide range of expected and unexpected life events and transitions, from off-time setbacks like the premature death of a spouse to on-time markers like the children leaving home to go to college, one’s 40th birthday, or the transition from work to retirement. Changes in narrative identity constitute real personality change – just as real as changes in dispositional traits, but different. Changes in narrative identity are typically the kinds of changes that people allude to when they maintain that they have ‘‘matured’’ with ‘‘experience,’’ that they are ‘‘very different people’’ today than they once were (McAdams, 1994). They are the kinds of changes, furthermore, that therapists and counselors often try to effect in their clients (Angus & McLeod, 2004). Conceptions of personality that focus exclusively on traits, therefore, underestimate the nature and extent of development over the span of life. Stories describe change and development in life, and life stories themselves change and develop over the life course.

STUDYING LIFE STORIES Social scientists have developed many different methodologies for studying narrative identity. Some researchers prefer in-depth interviews while others collect short written accounts of important life scenes. Some researchers engage in a deep, hermeneutical process for each case while others employ content-analysis systems to many cases in an effort to quantify results. Within personality psychology, narrative researchers have gathered in-depth case data from interviews and short written accounts through open-ended questionnaires (McAdams, 1999; Singer, 2004). Most personality researchers, furthermore, have tried to quantify their data in order to test hypotheses about life stories in select samples of individuals. For example, research has examined the relations between aspects or features of life stories on the one hand and traits (McAdams et al., 2004), social motives (Woike, 1995), values and personal ideologies (de St. Aubin, 1996), coping with stress (King, Scollon, Ramsey, & Williams, 2000), mental health and well-being (Bauer & McAdams, 2004), drug and alcohol abuse (Singer, 1997), and patterns of interpersonal relationships (Thorne, 2000) on the other. One standard procedure used in a number of studies is McAdams’s (1993) Life Story Interview. The interviewer begins by asking the participant to think about his or her life as if it were a book and to provide a brief outline of that book. Then the interviewer zeroes in on particular scenes or moments that stand out in the story – high points, low points, and turning

246

DAN P. McADAMS

points, for example. For each of these scenes, the participant describes in detail what happened in the event, who was there, what he or she was thinking and feeling in the scene, and what he or she thinks the scene may say about who the person is, was, or will be. Many of these key scenes qualify for what Singer (1995) has termed self-defining memories – vivid, affectively charged events from the past that express key hopes, fears, and conflicts in life. The interviewer then moves to the future and asks the participant to imagine what the next chapter in his or her life story might be like, focusing on dreams and fears regarding the future. Toward the end of the interview, the participant considers more abstract and philosophical questions regarding the life story, articulating those religious and/or social beliefs and values that may be important in the narrative and commenting on what message or moral the story seems to express. Depending on the purpose of the investigation, accounts obtained from the Life Story Interview, and similar procedures, may be analyzed according to many different systems, some of which emphasize structural features such as narrative coherence and complexity and others of which emphasize story content. With respect to content, a number of researchers have examined Bakan’s (1966) distinction between agency and communion in life-narrative accounts (e.g., Hermans, 1996; McAdams, 1985; McAdams, Hoffman, Mansfield, & Day, 1996; Woike, 1995). Agency refers to characters’ strivings to expand, assert, control, or defend the self, and is manifested in themes related to achievement, power, status, and self-mastery. By contrast, communion refers to characters’ strivings to connect the self with others, manifested through themes of friendship and love, interpersonal communication, caring for and helping others, and feeling a sense of community. In their various manifestations, agency and communion appear to be highly salient themes in life stories and common dimensions upon which stories can be said to differ in meaningful ways. Another thematic distinction that has enjoyed currency in recent research is that between contamination sequences and redemption sequences in life narrative (McAdams & Bowman, 2001; McAdams, Reynolds, Lewis, Patten, & Bowman, 2001). In a contamination sequence, an affectively positive scene is rather suddenly transformed into something negative. The main character’s initial state of strong joy or excitement suddenly gives way to fear, sadness, shame, guilt, or some other negative emotional reaction. The scene’s opening goodness is contaminated, ruined by what follows. Most everybody can recall contamination sequence from their lives. But some individuals use this narrative form in making sense of their lives more often than do others. Research suggests that adults who express more


247

examples of contamination in their life narrative accounts tend to score much lower on measures of psychological well-being and higher on measures of depression compared to individuals who show fewer contamination sequences in life stories (McAdams et al., 2001). In a study of 74 midlife adults, self-report measures of self-esteem, life satisfaction, and sense of life coherence were negatively correlated with, and depression positively correlated with, scores on contamination sequences derived from content analysis of the life narrative interviews. Contamination sequence scores were also significantly associated with a narrative measure of depressogenic attributional style – the tendency to attribute negative life events to internal, stable, and global causes (Peterson & Seligman, 1984). But contamination sequence scores from these same life stories proved to be a significantly strong predictor of depression than was depressogenic attributional style (Adler, Kissel, & McAdams, in press). In a redemption sequence, a bad or affectively negative life-story scene turns good – the bad is salvaged or redeemed by the positive outcome that ensues (McAdams et al., 2001). Perhaps not surprisingly, the redemptive form of life storytelling tends to be positively associated with a number of positive psychosocial features – such as life satisfaction among midlife American adults and among college students (McAdams et al., 2001). In one study, researchers coded written accounts of 10 key scenes in life narratives from 125 college students. Redemption imagery in the stories was positively associated with numerous self-report measures of well-being, such as life satisfaction and purpose in life. The researchers also coded the overall emotional quality of the stories, from positive to negative. The results showed that redemptive stories are not exactly the same thing as happy, positive stories about the self. Redemption sequences were only weakly correlated with overall emotional positivity of the life stories. Furthermore, redemption sequences proved to be substantially stronger predictors of wellbeing than did the overall emotional tone of the narrative accounts. The results suggest that it is not so much that happy people tell happy stories about their lives. Instead, happy people tend to tell life stories filled with episodes in which suffering is redeemed by positive outcomes. Redemption sequences are at the center of a collection of narrative themes that tend to characterize the life stories told by highly generative American adults. In a series of studies, McAdams and colleagues (Mansfield & McAdams, 1996; McAdams & Bowman, 2001; McAdams, Diamond, de St. Aubin, & Mansfield, 1997; McAdams, Ruetzel, & Foley, 1986; McAdams et al., 2001; Van de Water & McAdams, 1989) collected life-narrative accounts from midlife adults who score especially high on objective indices of generativity.

248

DAN P. McADAMS

Generativity is an adult’s concern for and commitment to promoting the well-being of future generations. A growing body of research shows that adults who score high on generativity measures tend to be more deeply invested in a wide range of endeavors aimed at improving the world around them – from parenting to volunteer work to voting (McAdams, 2001; McAdams & de St. Aubin, 1992; Rossi, 2001). Compared to their counterparts scoring lower in generativity, moreover, highly generative adults tend to construct life stories that feature a wide range of variations on the theme of redemption. In addition, their stories often feature these related themes: (1) a childhood sense of feeling special or advantaged; (2) an early sensitivity to the suffering or oppression of others; (3) the consolidation of a simple but compelling personal ideology in adolescence and the commitment to that ideology through the adult years; (4) tension between agentic and communal strivings in adulthood; and (5) anticipating growth and fruition for the future. Taken together, these themes comprise what McAdams (2006b) calls the redemptive self. Table 1 summarizes the central themes of the redemptive self. The redemptive self represents a particular kind of narrative identity that serves well the psychosocial needs of many highly generative men and women in their midlife years. The story reinforces their commitments to promoting the well-being of future generations. Believing that they were once the beneficiaries of early advantages in life, they feel some obligation to give back for the blessings they received. Their stories tell them that early on they were chosen or advantaged in some way in a world where many other people suffer. The contrast between their sense of personal blessing and their early awareness of the suffering of others may set up in their story a moral challenge and a felt obligation to be of some good use to others (Colby & Damon, 1992). The beliefs they consolidated in adolescence serve as their steady guides. In that generativity is often hard work, they manage to persevere in part because their story tells them that bad things are often overcome, suffering gives way to redemption. They look to the future with hope, even as they realize that the world is a very dangerous place. The hero in the story is the dogged, if sometimes guileless, protagonist who remains upbeat and focused on making a positive difference in his or her family, neighborhood, church, community, society – even against the odds. McAdams (2006b) also suggests that the redemptive self, for better and for worse, may represent a characteristically American way of narrating a caring and productive life at midlife. As highly generative American adults shape their lives into redemptive narratives, they implicitly draw upon an optimistic and highly individualistic understanding of the life course


Table 1.

249

The Redemptive Self: Six Themes Characterizing Life Stories Constructed by Highly Generative American Adults.

1. Early advantage

2. Suffering of others

3. Ideological steadfastness

4. Redemption sequences 5. Power vs. love

6. Prosocial future

As a young child, the story’s protagonist enjoys a special advantage or blessing that singles him or her out in the family or vis-a-vis peers. From an early age onward, the protagonist feels that he or she is special in a positive way. Early in the story, the protagonist witnesses the suffering or misfortune of other people and feels sympathy or empathy for them. Objects of the protagonist’s concern might include the sick, dying, disabled, mentally ill, economically disadvantaged, or any of a number of other groups or individuals that might require special care or help. By adolescence, the protagonist has established a clear and coherent belief system that governs his or her life. The belief system, often rooted in religion, remains relatively stable and steadfast over time. Once the belief system is established, the protagonist does not experience profound ideological doubt, uncertainty, or crisis. Bad or affectively negative life events are immediately followed by good or affectively positive outcomes. The bad scene is redeemed, salvaged, and made better by what follows. As an adult, the protagonist repeatedly finds that strong agentic desires to distinguish the self by having a positive impact on the world repeatedly conflict with equally strong communal desires to form loving relationships and be accepted by others as an equal. In looking to the future chapters of the life story, the protagonist sets goals that aim to benefit society in general or its institutions.

Source: McAdams (2006b) and McAdams et al. (1997).

that celebrates personal redemption through the discourses of atonement (religion), emancipation (politics), recovery (medicine), self-actualization (psychology), and upward social mobility (economics). These discourses have their origins in such canonical American texts as the spiritual autobiographies of the New England Puritans, Benjamin Franklin’s autobiography, Emerson’s 19th-century lectures and essays on ‘‘self-reliance,’’ Horatio Alger stories, the Gettysburg Address, and the powerful narratives written by escaped slaves before the American Civil War. Variations on the same themes run through 20th-century American autobiography and fiction, American television and movies, and the burgeoning literature of American self-help. Translated into myths of national identity, these themes are reflected in such quintessentially American notions as ‘‘the cho-

250

DAN P. McADAMS

sen people,’’ ‘‘manifest destiny,’’ and ‘‘the American dream.’’ The life stories constructed by highly generative American adults today employ rich metaphors and ways of thinking about identity that Americans have both cherished and contested, found both inspiring and problematic, for over 300 years – from the Puritan landing in New England in 1630 to the most recent episodes of The Oprah Winfrey Show.

NARRATIVE, CULTURE, AND THE LIFE COURSE Research on the redemptive self suggests that life stories may say as much about the culture and society wherein the narrator lives as they do about the narrator’s life itself. Many social scientists see life stories as cultural texts (Rosenwald & Ochberg, 1992; Shotter & Gergen, 1989). Life stories mirror the culture wherein the story is made and told. Stories live in culture. They are born, they grow, they proliferate, and they eventually die according to the norms, rules, and traditions that prevail in a given society, according to the society’s implicit understandings of what counts as a tellable story, a tellable life. As Rosenwald (1992) puts it, ‘‘when people tell life stories, they do so in accordance with the models of intelligibility specific to the culture’’ (p. 265). Habermas and Bluck (2000) contend that before a person can formulate a convincing life story, he or she must become acquainted with the culture’s concept of biography. In modern Western cultures, Denzin (1989) and McAdams (1996) suggest, biographies are expected to begin in the family, to involve growth and expansion in the early years, to trace later problems back to earlier conflicts, to incorporate epiphanies and turning points that mark change in the protagonist’s quest, and to be couched in the discourse of progress versus decline. But other societies tell about lives in different ways and have different views of what constitutes a good story to tell (Gregg, 1991). Even within a given society, furthermore, different stories compete for dominance and acceptance. Feminists such as Heilbrun (1988) argue that in Western societies many women ‘‘have been deprived of the narratives, or the texts, plots, or examples, by which they might assume power over – take control over – their lives’’ (p. 17). It is painfully clear that life stories echo gender and class constructions in society and reflect, in one way or another, prevailing patterns of hegemony in the economic, political, and cultural contexts wherein human lives are situated. Power elites in society privilege certain life stories over others, and therefore a number of narrative researchers and clinicians seek to give voice and expression to forms of life


251

narrative that have traditionally been suppressed or marginalized (Franz & Stewart, 1994; White & Epston, 1990). At the same time, it is clear that people, even those whose lives do not affirm a society’s dominant narratives, do not simply acquiesce to prevailing cultural norms and standards with respect to the narrative construction of lives. Human agency confronts the instrumental and expressive givens in a particular social ecology and seeks to appropriate what it can, picking and choosing among different narrative options, making a life and a life story that may even defy the status quo. Gjerde (2004) argues that the relation between self and culture is complex and often contested. People sometimes resist the norms to construct individual life patterns that defy cultural convention. Gjerde (2004) writes: ‘‘Culture can be said to exist as contested representations situated in public domains or institutions in which power is both exercised and resisted’’ (p. 146). Consequently, life stories represent something of a resultant compromise between human agency and social structure. Narrative identity is a psychosocial construction – a joint product of individual and society. Life stories draw on the stories that people learn as active participants in culture – stories about childhood, adolescence, adulthood, and aging. Stories capture and elaborate metaphors and images that are especially resonant in a given culture. Stories distinguish between what culture glorifies as good characters and vilifies as bad characters, and they present the many varieties who fall in between. Stories depict full and fragmented lives that are exciting, frightening, infuriating, enlightening, admirable, heroic, dignified, ignoble, disgusting, wise, foolish, and boring. Stories teach people how to live and what their lives may mean. Stories also spell out expected stages and trajectories in the life course. From a narrative perspective, human lives do not follow a natural stage-like progression; they do not conform readily to social roles and norms; and they do not follow the predictable trends ascribed to a given dispositional trait. Instead, life stories follow whatever form the individual and society manage to work out. In the best cases, that form meets what the individual feels to be his or her inner needs and aspirations while making a productive contribution within a specified psychosocial niche. In cases that are less ideal but probably more common, individuals struggle to find life-narrative forms that more-or-less see them through a difficult life terrain, amidst personal setbacks, failures, frustrations, and a demanding and stubbornly uncooperative world. People do the best they can in an effort to make sense of their lives in time, even under trying circumstances.

252

DAN P. McADAMS

Culture, then, provides each person with an extensive menu of stories about how to live, and each person chooses from the menu (McAdams, 2006b). Because different people within a given culture have different experiences and opportunities, no two people get exactly the same menu. Furthermore, a person cannot eat everything on the menu, so narrative choices spell out a person’s relationship to culture. When the food comes from the kitchen, people doctor it to their own tastes. They add pepper and salt; they mix things up and throw some things away; they nibble from somebody else’s plate; they may even send the order back and ask to see the menu again. This is to say that individuals select and appropriate in the making of narrative identity. They choose from competing stories, rejecting many others, and they modify the stories they choose to fit their own unique life, guided by the unique circumstances of their social, political, and economic worlds, by their family backgrounds and educational experiences, and by their dispositional traits and characteristic adaptations. A person constructs a narrative identity by appropriating stories from culture. Self and culture come to terms with each other through the narrative. In conclusion, a narrative approach to understanding lives in time complements prevailing models of the human life course that posit either (1) developmental stages (e.g., Levinson, 1978), (2) life trajectories (e.g., Elder, 1995), or (3) personality traits (e.g., McCrae & Costa, 1990), while calling into question the extent to which any model of the life course is itself something of a narrative convention. In describing the life course in terms of predictable passages and transitions, contextualized life trajectories, or stable dispositional traits, social scientists are prioritizing certain stories about life over others. Yet, the imposition of one kind of narrative frame over others is probably an inescapable feature of theory-making in the social sciences. Indeed, some general stories about the life course may have more scientific validity than others. Data can be garnered to support many different models of the life course, including those outlining developmental stages, contextualized life trajectories, and stable dispositional traits. One advantage, however, of narrative models for the life course is that they prioritize the individual’s own construction of lives in time, rather than projecting the construction offered by theorists themselves. Of course, narrative approaches themselves are a projection of theorists’ preferences. In prioritizing the story over other forms of human expression, narrative approaches assume that storytelling is a natural, even universal, human tendency, and that it represents a dominant way in which people make sense of their lives in time (Bruner, 1990). Reasonable people may wish to take exception to that assumption.


253

Narrative approaches focus attention on the meanings that people make as they live their lives over time and try, over time, to make sense of them. And narrative approaches show that those meanings are more than either the idiosyncratic tales of self-contained individuals or the passive recitations of society’s dominant narrative forms. Instead, life stories represent the creative, contested, and constantly evolving interplay between a storytelling agent and a complexly structured and storied world. As lives play out over time, stories ultimately emerge, narrated in fits and starts by a compromised but determined narrator who doubles as the protagonist of the tale. The author may feel that it is his own story to tell. But even in the most heroic tales, authorship is always joint – a project shared by the narrator himself and the world wherein his story is told.

REFERENCES Adler, J., Kissel, E., & McAdams, D. P. (in press). Emerging from the CAVE: Attributions in life narratives and their relations to psychological well-being. Cognitive Therapy and Behavior. Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart & Winston. Ambady, G., Hallahan, M., & Rosenthal, R. (1995). On judging and being judged accurately in zero-acquaintance situations. Journal of Personality and Social Psychology, 69, 518–529. Angus, L. E., & McLeod, J. (Eds) (2004). Handbook of narrative and psychotherapy. London: Sage. Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469–480. Bakan, D. (1966). The duality of human existence. Boston: Beacon Press. Bauer, J. J., & McAdams, D. P. (2004). Growth goals, maturity, and well-being. Developmental Psychology, 40, 114–127. Baumeister, R. F. (1994). Introduction to symposium: Stories as samples. Personality and Social Psychology Bulletin, 20, 649. Bertaux, D. (Ed.) (1981). Biography and society. Beverly Hills, CA: Sage. Birren, J., Kenyon, G., Ruth, J., Shroots, J. J. G., & Svendson, J. (Eds) (1996). Aging and biography. New York: Springer. Bouchard, T. J., Jr., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990). Sources of human psychological differences: The Minnesota study of twins reared apart. Science, 250, 223–228. Bruner, J. S. (1990). Acts of meaning. Cambridge, MA: Harvard University Press. Cantor, N. (1990). From thought to behavior: ‘‘Having’’ and ‘‘doing’’ in the study of personality and cognition. American Psychologist, 45, 735–750. Casey, K. (1996). The new narrative research in education. In: M. W. Apple (Ed.), Review of research in education (Vol. 21, pp. 211–253). Washington, DC: American Educational Research Association.

254

DAN P. McADAMS

Charme, S. (1984). Meaning and myth in the study of lives: A Satrean perspective. Philadelphia, PA: University of Pennsylvania Press. Cohler, B. J. (1982). Personal narrative and the life course. In: P. Baltes, & O. G. Brim (Eds), Lifespan development and behavior (Vol. 4, pp. 205–241). New York: Academic Press. Colby, A., & Damon, W. (1992). Some do care: Contemporary lives of moral commitment. New York: The Free Press. Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self-memory system. Psychological Review, 107, 261–288. Costa, P. T., Jr., & McCrae, R. R. (1994). Set like plaster? Evidence for the stability of adult personality. In: T. F. Heatherton & &J. L. Weinberger (Eds), Can personality change? (pp. 21–40). Washington, DC: APA Books. Dannefer, D. (1984). Adult development and social theory: A paradigmatic reappraisal. American Sociological Review, 49, 100–116. de St. Aubin, E. (1996). Personal ideology polarity: Its emotional foundation and its manifestation in individual value systems, religiosity, political orientation, and assumptions concerning human nature. Journal of Personality and Social Psychology, 71, 152–165. Denzin, N. (1989). Interpretive biography. Newbury Park, CA: Sage. Edel, L. (1978). Biography: A manifesto. Biography, 1, 1–3. Elder, G. H., Jr. (1995). The life course paradigm: Social change and individual development. In: P. Moen, G. H. Elder Jr. & K. Luscher (Eds), Examining lives in context (pp. 101–139). Washington, DC: APA Books. Ely, M. (2003). Braiding essence: Learning what I thought I already knew about teaching qualitative research. In: R. Josselson, A. Lieblich & D. P. McAdams (Eds), Up close and personal: The teaching and learning of narrative research (pp. 215–238). Washington, DC: APA Books. Emmons, R. A. (1986). Personal strivings: An approach to personality and subjective wellbeing. Journal of Personality and Social Psychology, 51, 1058–1068. Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Norton. Fivush, R., & Haden, C. (Eds) (2003). Autobiographical memory and the construction of a narrative self. Mahwah, NJ: Erlbaum. Franz, C., & Stewart, A. J. (Eds) (1994). Women creating lives. Boulder, CO: Westview. Freund, A. M., & Baltes, P. B. (2000). The orchestration of selection, optimization, and compensation: An action-theoretical conceptualization of a theory of developmental regulation. In: W. J. Perrig & A. Grob (Eds), Control of human behavior, mental processes, and consciousness (pp. 35–58). Mahwah, NJ: Erlbaum. Friedman, H. S., Tucker, J. S., Tomlinson-Keasy, C., Schwartz, J. E., Wingard, D. L., & Criqui, M. H. (1993). Does childhood personality predict longevity? Journal of Personality and Social Psychology, 65, 176–185. Funder, D., & Colvin, C. R. (1991). Explorations in behavioral consistency: Properties of persons, situations, and behaviors. Journal of Personality and Social Psychology, 60, 773–794. Gabriel, Y. (2000). Storytelling in organizations: Facts, fictions, and fantasies. New York: Oxford University Press. Gergen, K. J. (1991). The saturated self. New York: Basic Books. Giddens, A. (1991). Modernity and self-identity: Self and society in the late modern age. Stanford, CA: Stanford University Press. Gjerde, P. F. (2004). Culture, power, and experience: Toward a person-centered cultural psychology. Human Development, 47, 138–157.


255

Goldberg, L. R. (1993). The structure of phenotypic personality traits. American Psychologist, 48, 26–34. Gregg, G. (1991). Self-representation: Life-narrative studies in identity and ideology. New York: Greenwood Press. Gutmann, D. (1987). Reclaimed powers. Evanston, IL: Northwestern University Press. Habermas, T., & Bluck, S. (2000). Getting a life: The emergence of the life story in adolescence. Psychological Bulletin, 126, 748–769. Hankiss, A. (1981). On the mythological rearranging of one’s life history. In: D. Bertaux (Ed.), Biography and society (pp. 203–209). Beverly Hills, CA: Sage. Heilbrun, C. (1988). Writing a woman’s life. New York: Norton. Hermans, H. J. M. (1996). Voicing the self: From information processing to dialogical interchange. Psychological Bulletin, 119, 31–50. Holstein, J. A., & Gubrium, J. F. (2000). The self we live by: Narrative identity in a postmodern world. New York: Oxford University Press. Hooker, K. (2002). New directions for research in personality and aging: A comprehensive model linking levels, structures, and processes. Journal of Research in Personality, 36, 318–334. Hooker, K., & McAdams, D. P. (2003). Personality reconsidered: A new agenda for aging research. Journal of Gerontology: Psychological Sciences, 58B, P296–P304. Howe, M. L., & Courage, M. L. (1997). The emergence and early development of autobiographical memory. Psychological Review, 104, 499–523. John, O. P., & Srivastava, S. (1999). The big five trait taxonomy: History, measurement, and theoretical perspectives. In: L. Pervin & O. P. John (Eds), Handbook of personality: Theory and research (2nd ed., pp. 102–138). New York: Guilford. Josselson, R., & Lieblich, A. (Eds) (1993). The narrative study of lives. Thousand Oaks, CA: Sage. Kenrick, D. T., & Funder, D. C. (1988). Profiting from controversy: Lessons from the personsituation debate. American Psychologist, 43, 23–34. King, L., Scollon, C. K., Ramsey, C. M., & Williams, T. (2000). Stories of life transition: Happy endings, subjective well-being, and ego development in parents of children with Down Syndrome. Journal of Research in Personality, 34, 509–536. Kluckhohn, C., & Murray, H. A. (1953). Personality formation: The determinants. In: C. Kluckhohn, H. A. Murray & D. M. Schneider (Eds), Personality in nature, society, and culture (pp. 53–67). New York: Alfred A. Knopf. Kohli, M. (1981). Biography: Account, text, method. In: D. Bertaux (Ed.), Biography and society. Beverly Hills, CA: Sage. Langbaum, J. (1982). The mysteries of identity: A theme in modern literature. Chicago: University of Chicago Press. Levinson, D. J. (1978). The seasons of a man’s life. New York: Alfred A. Knopf. Lieblich, A., McAdams, D. P., & Josselson, R. (Eds) (2004). Healing plots: The narrative basis of psychotherapy. Washington, DC: APA Books. Little, B. R. (1999). Personality and motivation: Personal action and the conative evolution. In: L. Pervin, & O. P. John (Eds), Handbook of personality: Theory and research (2nd ed., pp. 501–524). New York: Guilford. MacIntyre, A. (1984). After virtue. Notre Dame, IN: University of Notre Dame Press. Mandler, J. M. (1984). Stories, scripts, and scenes: Aspects of schema theory. Hillsdale, NJ: Erlbaum.

256

DAN P. McADAMS

Mansfield, E., & McAdams, D. P. (1996). Generativity and themes of agency and communion in adult autobiography. Personality and Social Psychology Bulletin, 22, 721–731. Maruna, S. (2001). Making good: How ex-convicts reform and rebuild their lives. Washington, DC: APA Books. Matthews, G., Deary, I. J., & Whiteman, M. C. (2003). Personality traits (2nd ed.). Cambridge: Cambridge University Press. McAdams, D. P. (1985). Power, intimacy, and the life story: Personological inquiries into identity. New York: Guilford. McAdams, D. P. (1993). The stories we live by. New York: Guilford. McAdams, D. P. (1994). Can personality change? Levels of stability and growth in personality across the life span. In: T. F. Heatherton & J. L. Weinberger (Eds), Can personality change? (pp. 299–313). Washington, DC: APA Books. McAdams, D. P. (1995). What do we know when we know a person? Journal of Personality, 63, 365–396. McAdams, D. P. (1996). Personality, modernity, and the storied self: A contemporary framework for studying persons. Psychological Inquiry, 7, 295–321. McAdams, D. P. (1997). A conceptual history of personality psychology. In: R. Hogan, J. Johnson & S. Briggs (Eds), Handbook of personality psychology (pp. 3–39). San Diego, CA: Academic Press. McAdams, D. P. (1999). Personal narratives and the life story. In: L. Pervin & O. P. John (Eds), Handbook of personality: Theory and research (2nd ed., pp. 478–500). New York: Guilford. McAdams, D. P. (2001). Generativity in midlife. In: M. Lachman (Ed.), Handbook of midlife development (pp. 395–443). New York: Wiley. McAdams, D. P. (2006a). The person: A new introduction to personality psychology (4th ed.). New York: Wiley. McAdams, D. P. (2006b). The redemptive self: Stories Americans live by. New York: Oxford University Press. McAdams, D. P., Anyidoho, N. A., Brown, C., Huang, Y. T., Kaplan, B., & Machado, M. A. (2004). Traits and stories: Links between dispositional and narrative features of personality. Journal of Personality, 72, 761–784. McAdams, D. P., & Bowman, P. J., Jr. (2001). Narrating life’s turning points: Redemption and contamination. In: D. P. McAdams, R. Josselson & A. Lieblich (Eds), Turns in the road: Narrative studies of lives in transition (pp. 3–34). Washington, DC: APA Books. McAdams, D. P., & de St. Aubin, E. (1992). A theory of generativity and its assessment through self-report, behavioral acts, and narrative themes in autobiography. Journal of Personality and Social Psychology, 62, 1003–1015. McAdams, D. P., de St. Aubin, E., & Logan, R. L. (1993). Generativity among young, midlife, and older adults. Psychology and Aging, 8, 221–230. McAdams, D. P., Diamond, A., de St. Aubin, E., & Mansfield, E. D. (1997). Stories of commitment: The psychosocial construction of generative lives. Journal of Personality and Social Psychology, 72, 678–694. McAdams, D. P., Hoffman, B. J., Mansfield, E., & Day, R. (1996). Themes of agency and communion in significant autobiographical scenes. Journal of Personality, 64, 339–378. McAdams, D. P., Reynolds, J., Lewis, M., Patten, A., & Bowman, P. J., Jr. (2001). When bad things turn good and good things turn bad: Sequences of redemption and contamination


257

in life narrative, and their relation to psychosocial adaptation in midlife adults and in students. Personality and Social Psychology Bulletin, 27, 472–483. McAdams, D. P., Ruetzel, K., & Foley, J. M. (1986). Complexity and generativity at midlife: Relations among social motives, ego development, and adults’ plans for the future. Journal of Personality and Social Psychology, 50, 800–807. McCrae, R. R., & Costa, P. T., Jr. (1990). Personality in adulthood. New York: Guilford. Mischel, W. (1968). Personality and assessment. New York: Wiley. Mischel, W., & Shoda, Y. (1995). A cognitive-affective-system theory of personality: Reconceptualizing situations, dispositions, dynamics, and invariance in personality structure. Psychological Review, 102, 246–268. Murray, S. L., & Holmes, J. (1994). Storytelling in close relationships: The construction of confidence. Personality and Social Psychology Bulletin, 20, 650–663. Peterson, C., & Seligman, M. E. P. (1984). Causal explanations as a risk factor for depression: Theory and evidence. Psychological Review, 91, 347–374. Polkinghorne, D. (1988). Narrative knowing and the human sciences. Albany, NY: SUNY Press. Ricoeur, P. (1984). Time and narrative. Chicago: University of Chicago Press. Roberts, B. W., & Del Vecchio, W. (2000). The rank-order consistency of personality from childhood to old age: A quantitative review of longitudinal studies. Psychological Bulletin, 126, 3–25. Roberts, B. W., & Pomerantz, E. (2004). On traits, situations, and their integration: A developmental perspective. Personality and Social Psychology Review, 8, 402–416. Rosenwald, G. (1992). Conclusion: Reflections on narrative self-understanding. In: G. Rosenwald & R. L. Ochberg (Eds), Storied lives: The cultural politics of self-understanding (pp. 265–289). New Haven, CT: Yale University Press. Rosenwald, G. C., & Ochberg, R. L. (Eds) (1992). Storied lives: The cultural politics of selfunderstanding. New Haven, CT: Yale University Press. Rossi, A. (Ed.) (2001). Caring and doing for others. Chicago: University of Chicago Press. Sarbin, T. (1986). The narrative as a root metaphor for psychology. In: T. Sarbin (Ed.), Narrative psychology: The storied nature of human conduct (pp. 3–21). New York: Praeger. Schank, R., & Abelson, R. (1995). Knowledge and memory: The real story. In: R. S. Wyer (Ed.), Advances in social cognition (Vol. 8, pp. 1–86). Hillsdale, NJ: Erlbaum. Schwartz, S. H. (1994). Are there universal aspects in the structure and content of human values? Journal of Social Issues, 50, 19–46. Sheldon, K. M. (2004). The psychology of optimal being: An integrated, multi-level perspective. Mahwah, NJ: Erlbaum. Shotter, J., & Gergen, K. J. (Eds) (1989). Texts of identity. London: Sage. Singer, J. A. (1995). Seeing one’s self: Locating narrative memory in a framework of personality. Journal of Personality, 63, 429–457. Singer, J. A. (1997). Message in a bottle. New York: The Free Press. Singer, J. A. (2004). Narrative identity and meaning making across the adult lifespan: An introduction. Journal of Personality, 72, 437–459. Thorne, A. (2000). Personal memory telling and personality development. Personality and Social Psychology Review, 4, 45–56. Tomasello, M. (2000). Culture and cognitive development. Current Directions in Psychological Science, 9, 37–40.

258

DAN P. McADAMS

Vaillant, G. E., & Milofsky, E. (1980). The natural history of male psychological health: IX. Empirical evidence for Erikson’s model of the life cycle. American Journal of Psychiatry, 137, 1349–1359. Van de Water, D., & McAdams, D. P. (1989). Generativity and Erikson’s ‘‘belief in the species’’. Journal of Research in Personality, 23, 435–449. Wellman, H. M. (1993). Early understanding of mind: The normal case. In: S. Baron-Cohen, H. Tager-Flusberg & D. J. Cohen (Eds), Understanding other minds: Perspectives from autism (pp. 10–39). New York: Oxford University Press. White, M., & Epston, M. (1990). Narrative means to therapeutic ends. New York: Norton. Winter, D. G., John, O. P., Stewart, A. J., Klohnen, E. C., & Duncan, L. E. (1998). Traits and motives: Toward an integration of two traditions in personality research. Psychological Review, 105, 230–250. Woike, B. (1995). Most memorable experiences: Evidence for a link between implicit and explicit motives and social-cognitive processes in everyday life. Journal of Personality and Social Psychology, 68, 1081–1091.

PART IV: METHODOLOGICAL INNOVATIONS

259


260

LIFE COURSE ANALYSIS: TWO (COMPLEMENTARY) CULTURES? SOME REFLECTIONS WITH EXAMPLES FROM THE ANALYSIS OF THE TRANSITION TO ADULTHOOD Francesco C. Billari 1. LIFE COURSE ANALYSIS IN DEMOGRAPHY The life course approach as an interdisciplinary program of study has been under development since the mid-1970s (Mayer & Tuma, 1990). The idea of studying the unfolding of individual lives within their local and broader context has unavoidably brought life course scholars to emphasize complexity rather than simplicity. In their end-of-millennium review, Giele and Elder (1998) identify four chief elements as fundamentally shaping life courses: individual development (human agency), history and culture (location in time and place), social relations (linked lives). The intersection of such elements constitutes the fourth chief element: the timing of lives. In the life course approach, a set of interconnected trajectories lies at the heart of the analysis; trajectories are themselves shaped by events. Towards an Interdisciplinary Perspective on the Life Course Advances in Life Course Research, Volume 10, 261–281 Copyright r 2005 by Elsevier Ltd. All rights of reproduction in any form reserved ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10010-0

261

262

FRANCESCO C. BILLARI

Demographers who are familiar with the life course approach are aware that the key elements outlined by Giele and Elder are naturally linked to notions used to define the coordinates of populations in demography. The human agency concept lies behind the use of age as a privileged time axis. Location in time and, to an extent, the idea of linked lives suggests cohorts or groups of cohorts as basic descriptive units for comparison. History and culture emphasize the importance of period and location in space. As events lie at the very heart of the tradition of demography, it is not a surprise that the life course approach has been particularly influential in demographic research (Dykstra & van Wissen, 1999). Large-scale surveys collecting individual-level retrospective demographic history, starting from the World Fertility Survey (WFS) have given the biggest impetus to collect in most demographic surveys data on the timing of events in relevant trajectories (Hobcraft & Murphy, 1986). Elsewhere (Billari, 2003), life course analysis has been defined as the statistical analysis of data on the timing of events (when do events happen?), their sequencing (in which order do events happen?), and their quantum (how many events happen?). Ideally, life course analysis includes the possibility to analyze the timing, sequencing, and quantum of events as depending on the elements mentioned by Giele and Elder: individual-level human development, social relations, location in time and place. A progressively increasing complexity has shaped – as we shall see in more detail in this paper – methods for the analysis of life courses (Billari, 2003). While scholars who advocate a wider use of qualitative research see their approach as a way to cope with the complexity of life courses, those who, like myself, feel more comfortable with an emphasis on quantitative research have more and more moved to complex methods as well. Complexity comes at some costs, but it also brings some paybacks. We shall try to disentangle costs and payback by looking at a specific field of study: the transition to adulthood. The study of the transition to adulthood has been a primary topic for life course scholars, and it has greatly benefited from advances in the life course approach and in life course analysis. The need for methodological advances in this area comes from the peculiar ‘‘demographic density’’ of early adulthood: in advanced societies several events happen during this period, and diversity between individual lives becomes importantly visible (Rindfuss, 1991). Processes that have usually been studied using a life course approach include leaving the parental home, starting and/or dissolving a union, having a child, migration, job entry and exit, retirement. In this paper I argue that there are two main approaches to life course analysis (and in particular the analysis of demographic life courses), and

Life Course Analysis: Two (Complementary) Cultures?

263

that these two approaches serve complementary, and equally necessary, aims. The first is the event-based approach, based in general on event history analysis; the latter set of techniques has become one of the basic ingredients of advanced methodological courses in population sciences during the last decades of the twentieth century. Connected to this approach, I review developments from the econometric literature on program evaluation, which aims at discovering causal relationships within life courses and appears still to not have strongly influenced life course research. The event-based approach is mainly targeted at the explanation of life courses, more specifically at the decomposition of the life course picture into the raw materials (in particular, events) that constitute it, and in the search for the causes that underlie the timing of these events. The second is the holistic approach, mostly relying on sequence analysis, which has had a less significant impact on life course research with respect to the event-based approach. This second approach is mainly targeted at what we may define complex description of life courses, or at the construction of what sociologists will define ‘‘idealtypes’’ of trajectories. Looking at life courses as conceptual units, analyses based on the holistic approach aim at gazing the whole picture, in a complex way that somehow parallels what can be obtained using qualitative analysis. Both approaches have been supported and have become widespread with the availability of corresponding specific software packages. The paper is structured as follows. In Section 2, I start from analyzing a debate between two cultures in statistical analysis and argue that the main approaches in life course analysis are related to these two cultures. In Section 3, I shortly review the ‘‘raw material’’, event-based approach, while in Section 4, I review the ‘‘whole picture’’, holistic approach. Further reflections on the complementarity between the two approaches, future research directions, and needs for data collection are discussed in Section 5.

2. TWO CULTURES? A DEBATE IN STATISTICS AND ITS IMPLICATIONS FOR LIFE COURSE ANALYSIS Life course analysis inevitably and heavily relies on statistics. It is thus useful to look at the debate within the statistical community to better grasp present and future directions in life course analysis. The idea that there are ‘‘two cultures’’ in statistical modeling has been launched in a controversial paper by Breiman (2001), published in ‘‘Statistical Science’’. The paper is followed by comments of other statisticians and by a rejoinder of the author.

264


The two cultures to which Breiman refers are: (1) a ‘‘data modeling culture’’, which is the mainstream in statistics, which assumes that ‘‘data are generated by a given stochastic data model’’; (2) an ‘‘algorithmic modeling culture’’, clearly favored by Breiman, which treats the data generation mechanism as unknown. According to Breiman, the focus on data models has ‘‘led to irrelevant theory and questionable scientific conclusions’’, ‘‘kept statisticians from using more suitable algorithmic models’’, and ‘‘prevented statisticians from working on exciting new problems’’. On the other hand, algorithmic issues point to the ‘‘multiplicity of good models’’ – illustrated by reference to the Japanese movie ‘‘Rashomon’’ to the conflict between simplicity and (predictive) accuracy – the Occam razor issue, to the dimensionality problem ‘‘curse or blessing?’’ In particular, these points relate to the issue that prediction may be improved when using simultaneously more than one model of the same data, and that using more complex models improves prediction at the expense of interpretability (the latter being in practical applications, according to Breiman, is a kind of secondorder need), and to the importance of adding dimensionality to problems. Breiman’s thought has clearly been shaped by his experience as a statistical consultant for businesses; it is clear that in this type of experience data have to be taken as ‘‘given’’. The most important comment – in terms of ‘‘culture’’ confrontation – on Breiman’s paper is the one by Cox, which is in turn criticized in Breiman’s rejoinder. Cox argues that the starting point is not ‘‘data’’ but ‘‘an issue, a question or a scientific hypothesis’’ and that real scientific applications are targeted at unraveling causal links using statistics, not only in form of the analysis of collected data, but also in terms of data collection design. Modelbased statistical techniques provide, in his view, the best opportunity to illuminate causality. It is clear that Cox’s thoughts have been shaped by the interaction with the scientific community in the medical and social sciences rather than with business or sciences involving very complex data (for example, astronomy or genomics). We can draw some lessons and inspiration from the Breiman–Cox debate for what follows. The event-based approach to life course analysis is strictly connected to what Breiman defines the ‘‘data modeling culture’’ – not by chance is Cox one of the founders of modern event history analysis (Cox, 1972). Also, connected to the arguments of Cox, and differently from statisticians who specialize in data analysis, scientists interested in the study of the life course do not necessarily have to take data as given, and sometimes they can design a study in view of unraveling the cause(s) behind the timing of a certain event. Scientific hypotheses may take the shape of a parametric


265

or semi-parametric statistical model, and statistical techniques may help us in comparing the goodness-of-fit of such models. For instance, in the case of models for the age at first marriage or first union, theory-based parametric formulation of the hazard rate have been described and can possibly be compared. As we shall see more in detail later in this paper, the statistical significance of the impact of covariates on hazard rates can be tested using standard approaches, and this has been extensively used in event history analysis looking for ‘‘causal’’ approaches (see Blossfeld & Rohwer, 2002). Nevertheless, the algorithmic culture defended by Breiman has also something to contribute in the field of life course analysis. First of all, it is not always possible in this field to design a study in view of performing a specific study. An important example is the study of archival data, collected for administrative needs, and used by historical demographers. One can say without doubt that analyses of archival data have contributed greatly to the analysis of life courses. Second, as the life course is a complex set of interrelated trajectories, disentangling causal links may not be the only aim; more specifically, algorithmic models can contribute greatly to the analysis of life courses as whole conceptual units. A very similar argument to the one illustrated here is presented by Ritschard and Oris (in this volume), who distinguish between a ‘‘statistical modeling’’ and a ‘‘data mining’’ approach to the analysis of life course data in demography – it is not by chance that they find this important distinction starting from issues raised in historical demography, a field in which the use of archival data has allowed to reach important scientific knowledge. These considerations justify the view that is taken in the remainder of this paper: both approaches are useful, in a complementary way. In other words, would we only ‘‘adopt’’ one of the two approaches (or the underlying statistical culture), we will know less about life courses. They are of different importance for different problems, and a review and reflection on these situations is necessary. Let us consider them in turn, starting with the eventbased approach.

3. THE EVENT-BASED, OR ‘‘CAUSALITY’’, CULTURE 3.1. Event History Analysis: from the ‘‘Causal Approach’’ to ‘‘Multilevel and Multiprocess Modeling’’ The set of statistical techniques that is now broadly defined as event history analysis has become since the 1980s as one of the principal toolkits of

266


demography and life course research in general (Allison, 1984; Yamaguchi, 1991; Courgeau & Lelie`vre, 1992; Petersen, 1995; Blossfeld, Hamerle, & Mayer, 1989; Blossfeld & Rohwer, 2002; Wu, 2003). More ambitiously, the diffusion of this approach has been linked to a change of paradigm in demography by Courgeau and Lelie`vre (1996), and as a solution to the problem of search for causality in the social sciences (Eerola, 1994; Petersen, 1995; Bocquier, 1996; Blossfeld, 1996; Blossfeld & Rohwer, 2002). Event history analysis generalizes life-table and standardization techniques that have been extensively used in twentieth-century demography. The statistical models of event history analysis usually aim to model individual-level data collected from sample surveys or population registers. Event history analysis focuses on the time-to-event as the dependent variable, and since its early applications in demography it has been applied to several types of events in the study of mortality, migration, family formation, and dissolution (Hobcraft & Murphy, 1986). Besides the possibility to describe and compare life courses, the statistical models of event history analysis have contributed to the explanation of life course dynamics by linking time-to-event with explanatory variables (covariates). Covariates can be external to a trajectory (as is the case for macro-level variables, period effects), or internal to a trajectory (as is the case of other trajectories of the life course that are potentially influencing the time-to-event in a trajectory of primary interest). The difference between external and internal covariates is relevant when one aims at giving a causal interpretation of the impact of changes of the covariates on the hazard rate of a given phenomenon. External covariates can be classified into three main categories. First, some covariates that are fixed during a life course or starting from a particular point of the life course, and are time-constant (gender, race, cohort, age at marriage when studying time-to-divorce). Second, other covariates have a temporal development that cannot be influenced by the process of interest (age of an individual in the case of time-to-divorce). Third, some covariates are located at an aggregate-level (‘‘macro-level’’) or proxy the social dynamics (period, regional economic indicators, policy indicators). In that sense, age-period-cohort models can be reconstructed at the individual level. Multilevel event history models have been developed, taking advantage of the improvement of software packages, at the end of the 1990s. They allow the explicit modeling of the case of individuals who are aggregated in household or regional clusters (see e.g. Barber, Murphy, Axinn, & Maples, 2000). As such covariates are ‘‘exogenous’’ – to use an econometric term – a causal interpretation could be justified. This would in principle allow the


267

identification of causal effects in many types of life courses analyses. It is however questionable that time-constant covariates such as race or gender could be seen as causal factors in the timing of an event. The same is true about age – in fact age is usually recognized as a proxy for other causal factors that are unmeasured (Settersten & Mayer, 1997). Also in the case of macro-level factors, it is often the case that they can be seen as ‘‘causing’’ the timing of an event – nevertheless they can be considered triggering factors in what could be defined a ‘‘situational mechanism’’ (Hedstro¨m & Swedberg, 1998), as they represent the situation (‘‘location in time and place’’ to use Giele and Elder definition) in which a life course is unfolding. Internal covariates usually refer to other trajectories of the same individual or of linked individuals, and their use allows researchers to study the very complex interdependencies between trajectories that go to the very core of the life course approach. Event history analysis may take into account unobserved factors underlying these complex interdependencies, such as unobserved value orientations or attitudes. There have been different approaches in the literature of the 1990s, with respect to the focus on the relevance of time-constant unobserved factors for the analysis of parallel and potentially interdependent trajectories (Wu, 2003). The so-called causal approach (Blossfeld & Rohwer, 2002), assumes that all factors that are relevant to the simultaneous analysis of several trajectories are observed and included in the past history of the trajectories. This assumption, however, is not completely explicit in the illustration of the proponents of the ‘‘causal approach’’ or in papers using that approach. Other modeling approaches, especially in the econometric literature, allow for the inclusion in event history models of the impact of timeconstant unobserved factors. The most general applied approach to this second type of strategy has been developed in the multilevel and multiprocess modeling of life courses (Lillard, 1993; Lillard & Panis, 2000). These models constitute in fact a generalization of the ‘‘causal approach’’, which has been often neglected by scholars looking for the causal impact of changes in one life course trajectory on other life course trajectories. The ‘‘multilevel and multiprocess’’ modeling approach proposed by Lillard and Panis aims at the simultaneous evaluation of the (1) impact of changing covariates (including those related to other life course trajectories and thus to be considered as internal covariates) on the hazard of experiencing an event; and (2) possible common, time-constant, factors that influence a set of hazard rates. The first aim is the same of the ‘‘causal approach’’, but it is reinforced by the fact that in multilevel and multiprocess hazard models it is possible to control for the common factors included in the second aim. In statistical terms, it is always

268


possible to test whether the general multilevel and multiprocess hazard model can be safely restricted to the model following the causal approach – this is done by testing whether the correlation of the unobserved factors that influence a set of processes is zero. In other words, the causal approach hypothesizes that there are no common and unobserved factors that affect the set of processes under study. The importance of taking into account the potential role of common factors affecting life course trajectories can be described by one example. Le Goff (2002) analyzed the transition to first marriage and first birth among cohabiting couples in France and West Germany, by using a Lillardtype model with a system of hazard equations with possibly correlated unobserved heterogeneity. This could be contrasted for instance to Blossfeld and Mills (2001), who analyze the same problem without allowing for potentially correlated unobserved factors affecting life course trajectories, but who explicitly refer in the title of their paper to the ‘‘causal approach to interrelated family events’’. We focus on Le Goff’s results for West Germany, for which the positive correlation between factors affecting entry into marriage and into parenthood is statistically significant (+0.5455). In Table 1, we show the results of the multiprocess model (Model (a)) and by comparison those of the model without unobserved heterogeneity and possible correlation between unobserved factors (Model (b)). We see that not taking the correlation into account leads to important biases in the estimation of the impact of getting married on the transition to first birth (the bias is up to 148% and never lower than 41%). More specifically, the ‘‘causal’’ accelerating impact of marriage is exaggerated if one does not take into

Table 1. Impact of Getting Married on the Log-Hazard of Conception for Cohabitants: Models Taking (a) and Not Taking (b) into Account the Correlation between Factors Leading to Marriage and First Birth. Model (a)

Cohort (1953–1961) Cohort (1962–1974)

Model (b)

At marriage

After 1 year

At marriage

After 1 year

0.2935 0.4836

0.4505 0.866

0.7268 0.9726

0.9208 1.2233

Note: Other covariates are included in the model (see Le Goff, 2002, Appendix). The estimated correlation between unobserved factors simultaneously affecting marriage and first birth (in Model (a)) is 0.5455. Source: Own elaboration on estimates by Le Goff (2002).


269

account that the propensity to marry and to have a child are highly interrelated, and this is due to common factors that are often unobserved.

3.2. The (Causal) Impact of Events: Ideas from the Program Evaluation Literature Unraveling causal links controlling for observed and unobserved spurious factors is the primary focus of another approach, which originates from the – mostly econometric – literature focusing on program evaluation from observational data. In program evaluation, the main task is to estimate the impact of participating in a certain program (usually, a labor market program), the treatment, on a specific outcome. This estimate is used as a support for policy decisions. Two key issues arise. First, to illuminate policies, one wants to isolate the causal impact of a certain program from other effects that link the program with the outcome. Second, because for cost– benefit evaluation the size of the causal impact, and not only its direction or statistical significance, matters, the emphasis on minimizing estimation bias is usually stronger than in traditional life course analysis. The characteristic problem in program evaluation is that participation in a program is voluntary and can be due to factors that are themselves correlated with outcomes. Thus, we face a similar situation when we want to evaluate the impact on some outcome measure of life course events. Life course events are subject to choice, and factors leading to choices may be influencing the outcome as well. With respect to the transition to adulthood, this scenario applies to several types of substantive problems, and some studies have already used approaches originating from the evaluation literature to study the transition to adulthood and its implications. First, we may be interested in evaluating the causal impact of events (i.e. timing or sequencing) in the transition to adulthood on the subsequent pathways to adulthood; a typical question is whether teenage childbearing influence subsequent educational or labor outcomes during early adulthood (see e.g. Hotz, Mullin, & Sanders, 1997), or, as in the traditional program evaluation literature, whether participation in a certain welfare program affects subsequent labor outcomes (see Dehejia & Wahba, 1999). An alternative but related issue is to study the long-term impact of choices during the transition to adulthood on later years. Second, we may be interested in evaluating the causal impact of events involving relevant others, in particular youths’ parents, on the transition to adulthood; a typical question is whether parental divorce has a causal impact on educational outcomes or family choices in

270


the transition to adulthood (see Painter & Levine, 2000). Third, we may want to study the causal impact of pathways to adulthood on relevant others; a typical question is whether out-of-wedlock births cause problems to children; or, whether the leaving home of a child, and in particular the transition to an ‘‘empty nest’’ has a causal impact on parental outcomes, for example happiness (Mazzuco, 2003). Although in some issues (i.e. welfare programs) experimental designs are also feasible, we shall focus only on observational, non-experimental studies. The methodological literature on the issue, although relatively recent (the origins can be traced to Heckman & Robb, 1985), has become vast (for a non-technical review see, for instance, Bryson, Dorsett, & Purdon, 2002). The basic ‘‘evaluation problem’’ or the ‘‘fundamental problem of causal inference’’ (Holland, 1986) is that to truly know the effect of a certain event (i.e. the participation into a program), we must compare the observed outcome of an individual who has experienced the event of interest with the outcome that would have resulted had that person not experienced the event. Equivalently, we must compare the observed outcome of an individual who has not experienced the event of interest with the outcome that would have resulted had that person experienced the event. This ‘‘counterfactual’’ outcome cannot be observed, and all approaches developed in the program evaluation literature aim at providing an estimate of the counterfactual and at using this in order to identify the causal effect of an event. In addition, the impact of an event can be different across different groups of a population. For these reasons one needs to distinguish what is called the average effect of treatment (what impact would the event have on a randomly drawn individual) and the average effect of treatment on the treated (what impact has the event had on individuals who have experienced the event). The average effect of treatment indicates the average benefit were the event to become compulsory (this is for instance interesting in the case of welfare programs). The average effect of treatment on the treated indicates the benefit, or cost of a specific event and is thus the parameter of main interest in our field. For instance, it is the answer to questions like ‘‘how different a teen mother’s subsequent Ys would be if she postponed or forewent the birth’’ (Hotz et al., 1997, p. 578). Hotz and colleagues explain why they focus on the average effect of treatment on the treated in their study of the effects of teenage childbearing by using two reasons. First, a pragmatic reason: ‘‘this causal effect is more readily identified from available data than are the causal effects applicable to the full population of women’’ (p. 579). Second, a substantive, policy-oriented, reason: ‘‘policies that seek to reduce the rate of teenage childbearing will likely target women who, under the


271

status quo, would become teenage mothers. Knowledge of a1 is sufficient to assess the potential consequences of eliminating teenage childbearing for these women’’ (p. 579). We thus want to estimate the effect of treatment by controlling for spurious dependence, possibly including dependence due to unobserved factors. To estimate treatment effects from observational studies, there are three approaches in the literature. First, one can use data on highly related individuals, usually twins, and exploiting the variation between them. This approach has been used for instance by Kohler, Skytthe and Christensen (2001) to estimate the impact of age at first birth on completed fertility. Twin data however are not widely collected and the results are not always easy to generalize. Second, one can use an instrumental variables (IV) approach, looking for variables that are good predictors of the event of interest but are not related to the outcome measure. This approach is hardly feasible when dealing with life courses, and the most promising avenue uses IV to estimate bounds of causal effects, as in the case of the analysis of the effects of teenage childbearing by Hotz et al. (1997). The third approach is the most promising for life course research, propensity scores matching, originally proposed by Rosenbaum and Rubin (1983), with the matching of treated and untreated individuals according to observed covariates summarized in a ‘‘propensity score’’. We examine more in depth propensity score matching. It is basically a two-step approach (although steps are relatively complex and some iteration may be required). In the first ‘‘parametric’’ step, the ‘‘propensity score’’ is estimated from a set of (possibly abundant) covariates that are supposed to affect the probability that the event of interest (treatment) is experienced. These covariates may also influence the outcome of interest (the outcome also possibly being the probability to experience another event) but the outcome is not used in the estimation of the propensity score. For instance, one can use a probit or logit model with the probability of experiencing a parental divorce as a function of a set of youth and family characteristics (Painter & Levine, 2000 use this as a robustness check for their results). In the second ‘‘nonparametric’’ stage, individuals who experienced the event of interest (treated) are matched to individuals who have not experienced the event of interest (untreated), according to the propensity scores estimated in the first step. Matching is ideally similar to what is done with controlled experiments, but different procedures are possible (for a practical review, see Becker & Ichino, 2002). For instance, each treated individual could be matched to the nearest untreated individual (nearest-neighbor matching) but if treated individuals are fewer that untreated it is also possible to match

272


one treated case with more than one untreated case; in addition, untreated cases may be matched with more that one treated case. It is however necessary that treated and untreated have a large enough ‘‘common support’’ in terms of comparability since treatment effects can be estimated only within the common support. Using another approach, Rosenbaum and Rubin (1984) propose to divide the treated and untreated group in a number of groups (i.e. with the same score range), and matching occurs when the groups of treated and untreated have similar covariates. After the matching, outcomes are compared and tests of statistical significance of the difference may be computed. The use of matching on propensity score permits to reduce the bias due to observed variables influencing both the probability of experiencing the event of interest, and the outcome (Rosenbaum & Rubin, 1983). To control for fixed unobserved characteristics influencing both the potential ‘‘cause’’ and the outcome, it is necessary to use other techniques usually relying on longitudinal data. For instance, instead of the difference in outcomes between treated and untreated one can estimate the difference-in-differences between treated and untreated. Let us recall the results of an example. Mazzuco (2003) uses propensity score matching to evaluate the causal impact of the departure of the last coresident child on parental well-being, in terms of satisfaction and of selfrated health. He uses data from the European Community Household Panel and is thus able to use the difference-in-differences estimator: how much does the transition into the empty nest stage affect a 1-year change of wellbeing and self-rated health? Mazzuco compares a country characterized by relatively early home-leaving (France) with a country characterized by exceptionally late home-leaving (Italy). We report his results in Table 2. They show that the departure of the last child does not have a (causal) impact on the well-being of fathers in both countries, and that it has opposite impacts on mothers. French mothers have a higher satisfaction after the departure of their last child, and their perceived health does not change. On the contrary, Italian mothers have a lower satisfaction and their self-rated health worsens. For the substantive interpretation of results we refer to the paper by Mazzuco. The production of easy-to-use packages for estimating treatment effects based on propensity score matching has and will have a crucial role in the diffusion of the methodology. Recently, Becker and Ichino (2002) have produced a set of Stata programs that allow building estimators based on propensity score matching. This becomes also a good opportunity for scholars interested in the life course.

273


Table 2. The Comparative Impact of the Departure of the Last Child (Transition into the ‘‘Empty-Nest’’) on (Co-Resident) Parents’ Satisfaction and Well-Being in France and Italy. France

Satisfaction Self-Rated Health

Italy

Mothers

Fathers

+0.407 0.005

0.066 0.059

Mothers

Fathers

0.568 0.149

0.023 0.013

Note: Estimates based on propensity score matching using data from the European Community Household Panel (Mazzuco, 2003). Statistically significant differences with po0.05. Satisfaction is measured with a sum of satisfaction items ranging from 4 to 24, self-rated health ranges from 1 to 5.

4. THE ALGORITHMIC OR ‘‘HOLISTIC’’ CULTURE Older notions of life course – for instance the concept of family life cycle – were holistic, with more or less explicit reference to biological structures (Settersten & Mayer, 1997). This has no longer been true in the life course literature since the 1990s. Nevertheless, by focusing mainly on specific events, with what Elder (1985) has called the ‘‘short-view in analytical scope’’ researchers may not grasp a unitary, holistic, perspective on life courses. There are two main reasons to complement event-based analysis with a holistic approach (Billari, 2001a). For the sake of simplicity, they may be defined as the strong and the pragmatic perspective. From the strong perspective, present for instance in neo-classical economics, life courses are seen subject to accurate inter-temporal planning, for instance as an outcome of utility maximization. This has also led to study empirically long-term plans in life courses and their consistency, as well as to critique on the realism of a dynamic programming view of lives from experimental economics. For instance, Keane and Wolpin (1997) in a paper on the career decisions of young men estimate empirically a model that involves ‘‘the repeated numerical solution of a discrete-choice, finite-horizon optimization problem, formulated as a dynamic programming problem’’ (Keane & Wolpin, 1997, p. 476). As life courses are assumed to be the (uncertain) outcome of planning, a holistic perspective is thus hypothesized to be present in the behavior of individuals themselves. In the social-demographic literature, the notion of strategy has been emphasized. In the psychological literature, individual life courses are assumed to be marked by internalized timetables. For theoretical

274


approaches that assume that the individuals look holistically at their own lives, it is undoubtedly necessary to have tools that allow to follow the same perspective, and to analyze the life course as a whole. From the pragmatic perspective, the life course as a conceptual unit is thought of as being a contingent result of subsequent events. Following this viewpoint, researchers should focus principally on events when they wish to explain individual behavior (as discussed earlier). A holistic view is still useful as a way to describe and to summarize the timing, sequencing, and quantum of life course events. Comparative research across countries or regions or across cohorts is one of the examples where the description of life courses as whole conceptual units might provide particular insights. Since the 1990s, Andrew Abbott introduced the social science to sequence analysis (see Abbott & Tsay, 2000). In sequence analysis, each life course or trajectory in the life course is represented as a string of characters (also numerical). This representation is identical to the one used to code DNA molecules in the biological sciences. Different analytical strategies to analyze data arranged as sequences have been proposed. Optimal matching analysis (OMA) is a method originally created for the alignment of DNA sequences. The goal of OMA is to compute a matrix of dissimilarities between pairs of sequences (thus, of life courses). This matrix can be used as the input for any kind of statistical analysis requiring proximity data (e.g. cluster analysis or multidimensional scaling). Applications in the sociology of occupations have appeared since the 1990s, and a review of the potential applications in the field of demography is provided in Billari (2001a), while a demographic application on the transition to adulthood is discussed in Billari (2001b). OMA has also specific drawbacks (Wu, 2000). The definition of distance between states has an important influence on results, and thus it has to be based on theoretical grounds. In addition, it can be difficult to understand which variables in the definition of the obtained groups are more relevant than others. Alternative approaches to the analysis of life courses as a conceptual unit have also been discussed, although they have not yet had a large impact on the literature. These include the use of correspondence analysis (van der Heijden, 1987, Chapter 8) and data mining techniques (Billari, Prskawetz, & Fu¨rnkranz, 2000). As an example of the use of the algorithmic culture we discuss the analysis of sequencing in the transition to adulthood among U.S. youth presented by Mouw (2005). Mouw applies a monothetic divisive algorithm (MDA) to create clusters of life courses that are internally as homogeneous as possible and externally as heterogeneous as possible. The study follows the approach suggested by Billari and Piccarreta (2001; 2005) in an analysis of the


275

transition to adulthood in Italy. As usual for sequence analysis, the coding of variables is a key step. Mouw defines a sequence of dummy variables called P (1 if an individual has left the parental home, 0 otherwise), E (1 if an individual has left education, 0 otherwise), L (1 if an individual has started working, 0 otherwise), M (1 if an individual has married, 0 otherwise), B (1 if an individual has already become a parent, 0 otherwise). An example for the coding of an individual life course is reported in Table 3. Sequences are then processed by the MDA by choosing the splitting variable that maximizes within-group homogeneity and between-group heterogeneity, according to a Gini heterogeneity index. It also allows the use of an R2 type of statistic to choose the optimal number of splits. As common in algorithmic approaches, a tree representation is useful. We reproduce (Fig. 1) the tree for U.S. women, which indicates that the most discriminating circumstance is whether they had become mother by age 28 (to the right of the tree, that is about two thirds of the sample). Six groups of women’s life courses are found. Group 1 is characterized by delayed transition to motherhood and marriage. Group 2 is characterized by delayed transition to motherhood, but marriage taking place at least by age 32. Group 3 is characterized by having had birth by age 28, without having been married by that age and having not begun work by age 33. Group 4 differs from Group 3 by the experience of work by age 33. Group 5 is characterized by a birth between age 24 and 28, and marriage by age 29. Group 6 is characterized by earlier fertility (transition to motherhood by age 24) and a transition to marriage by age 29.

5. CONCLUDING REMARKS From the review we have made of the existing literature we can safely conclude that no single approach is the ‘‘solution’’ to life course analysis. On the one side, there are different approaches that are useful in specific circumstances within what we have defined the event-based, or ‘‘causality’’ culture – we have mentioned event history analysis and approaches based on program evaluation, although in some cases the difference between the two approaches is negligible or non-existent. On the other, by definition the ‘‘algorithmic’’ or holistic culture sees a multiplicity of ‘‘good’’ models as a reasonable outcome. The two cultures prove to be more fruitful vis-a`-vis each other when answering to different questions? To put it simply, on one hand, when we would like to ask ‘‘what would happen to the risk of experiencing a certain event ify?’’, we shall probably use models rooted in the

276


Table 3. Coding of the Life Course of an Individual who has Left Home at Age 23, Finished School at 25, Entered the Labor Force at 26, Got Married at 30 and had His or Her First Child at Age 32.

P E L M B

22

23

24

25

26

27

28

29

30

31

32

33

34

35

0 0 0 0 0

1 0 0 0 0

1 0 0 0 0

1 1 0 0 0

1 1 1 0 0

1 1 1 0 0

1 1 1 0 0

1 1 1 0 0

1 1 1 1 0

1 1 1 1 0

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

1 1 1 1 1

Source: Mouw (2005).

All women N=2,939

B28=0 N=971

B28=1 N=1,968

M32=0 N=465

M32=1 N=506

Group 1

Group 2

M29=0 N=721

M29=1 N=1,247

L33=0 N=298

L33=1 N=423

B24=0 N=352

B24=1 N=895

Group 3

Group 4

Group 5

Group 6

Fig. 1. Tree Representation of the Results of Applying a Monothetic Divisive Algorithm to Data on the Transition of Adulthood of U.S. Women. B24 ¼ 1 and B28 ¼ 1 indicate having become a mother respectively by age 24 and 28 (B24 ¼ 0 and B28 ¼ 0 indicate the opposite), M29 ¼ 1 and M32 ¼ 1 indicate having married respectively by age 29 and 32 (M29 ¼ 0 and M32 ¼ 0 indicate the opposite), L33 ¼ 1 indicates having worked by age 33 (L33 ¼ 0 indicates the opposite. Source: Mouw (2005).

277


‘‘event-based’’ culture. On the other, when we would like to ask ‘‘what are the key factors that differentiate life courses in their unity, and what are the consequences of a broadly defined trajectory?’’, we shall probably use models rooted in the algorithmic culture. However, also within the same study the two approaches can fruitfully complement each other. For instance, Mouw (2005) uses the output of the clustering procedure we illustrated in Section 5 as an input for a regression analysis that is much more oriented towards causality, under the heading ‘‘Does the sequence matter?’’ Regression analyses show important difference in the risk of experiencing outcomes such as poverty at age 35, with women in Group 3 with the worst outcome in a logit regression model (see Table 4). Sequences are also found to influence subsequent happiness and depression (women in Group 6 are the happiest and least likely to be depressed). The same spirit of complementarity between the two cultures inspires the work of McVicar and Anyadike-Danes (2001) who use clusters obtained starting from an optimal matching analysis of sequences to predict successful and unsuccessful transitions from school to work. It is necessary to mention some caveats in these concluding remarks. In the ‘‘causality’’ culture we should refrain from the idea of finding the ‘‘true’’ impact of events in the life course, as is sometimes done in the econometric literature – the impact of an event, if we go back to the conceptual framework on the life course by Giele and Elder, depends on the role of relevant others (‘‘linked lives’’) as well as on other contextual factors (‘‘location in time and place’’). The search for causality in the life course is thus a never ending story, and the most promising new analyses based on either Table 4. The Effect of Sequences on Life Course Outcomes: Coefficients Related to Group Membership as Reported in Fig. 1 in a Logit Model for Poverty Status at Age 35 (other coefficients not reported here). Coefficient

Group Group Group Group Group Group

1 2 3 4 5 6

(B28 ¼ 0; (B28 ¼ 0; (B28 ¼ 1; (B28 ¼ 1; (B28 ¼ 1; (B28 ¼ 1;

M32 ¼ 0) M32 ¼ 1) M29 ¼ 0; L33 ¼ 0) M29 ¼ 0; L33 ¼ 1): reference M29 ¼ 1; B24 ¼ 0) M29 ¼ 1; B24 ¼ 1)

Source: Mouw (2005). Statistically significant differences with po0.01.

Standard error

0.810 0.297 1.352

0.549 0.547 0.255

0.411 0.074

0.432 0.239

278


simultaneous hazard models or on other types of program evaluation take what we may define a multiplicity of causal impacts for a multiplicity of contexts as a strong point in the analysis. The caveat for the algorithmic culture is possibly a similar one – not always what has been developed in other fields can be directly translated to the analysis of life courses, and a fecund interaction is fundamental in order to obtain optimal results. No approach can work without appropriate software, and here it is worth mentioning that some scholars have sometime worked hard to help other scholars implementing new approaches. Transition Data Analysis (TDA) (Rohwer & Po¨tter, 2003), applied Maximum Likelihood (aML) (Lillard & Panis, 2000) have been explicitly developed to study life courses using hazard models, with aML allowing for the inclusion of correlated unobserved heterogeneity between processes. Commercial packages have and will be used, with Stata playing a prominent role in the program evaluation approach. For what concerns the algorithmic approach, TDA is also providing a toolkit for sequence analysis, including optimal matching analysis, while a set of non-standard programs written by researchers for the moment is used for other applications. To conclude, data are crucial ingredients for our analyses. Breiman defined what we have put in parallel to the event-based or causality culture as ‘‘data modeling culture’’. The sort of data that have been collected since the 1970s in demography included detailed retrospective reconstruction of life course trajectories. For a while this seemed to be the final answer to the search for causality (see, for instance, the claims of Blossfeld & Rohwer, 2002). Nevertheless, by emphasizing the importance of unobserved heterogeneity and of the bias introduced when not taking it into account we acknowledge that more work has to be done. Prospective panel studies, aiming at observing what is unobservable in a retrospective study (e.g. intentions, value orientations and social capital) will provide to the causal approach a much better equipment. On the other hand, the algorithmic approach is to be exploited when data that are not directly collected for research purposes – such as in the case of administrative registers – are to be analyzed. Much more work is still to be done, in different directions, to progress in life course analysis.

ACKNOWLEDGMENTS The author gratefully acknowledges for comments and suggestions the editors of this volume, the members of the Working Group on Transitions to


279

Adulthood in Developed Societies of the International Union for the Scientific Study of Population (IUSSP), and Stefano Mazzuco. A presentation based on these ideas was given at a lecture organized at the Department of Sociology of the University of Calgary by Prof. Anne H. Gauthier. Financial support is acknowledged to the Department of Sociology of the University of Calgary, Universita` Bocconi and MIUR.

NOTE 1. This is the symbol for the average effect of treatment on the treated.

REFERENCES Abbott, A., & Tsay, A. (2000). Sequence analysis and optimal matching methods in sociology. Sociological Methods & Research, 29, 3–33. Allison, P. D. (1984). Event history analysis. Regression for longitudinal event data. Newbury Park, CA: Sage. Barber, J. S., Murphy, S. A., Axinn, W. G., & Maples, J. (2000). Discrete-time multilevel hazard analysis. Sociological Methodology, 30, 201–235. Becker, S. O., & Ichino, A. (2002). Estimation of average treatment effects based on propensity scores. The Stata Journal, 7, 1–19. Billari, F. C. (2001a). Sequence analysis in demographic research and applications. Canadian Studies in Population, 28, 439–458. Billari, F. C. (2001b). The analysis of early life courses: Complex descriptions of the transition to adulthood. Journal of Population Research, 18, 119–142. Billari, F. C. (2003). Life course analysis. In: P. Demeny & G. McNicoll (Eds), Encyclopedia of population (Revised edition, pp. 588–590). New York: Macmillan. Billari, F. C., & Piccarreta, R. (2001). Life courses as sequences: An experiment in classification via monothetic divisive algorithms. In: S. Borra, R. Rocchi, M. Vichi & M. Schader (Eds), Advances in classification and data analysis (pp. 351–358). Berlin: Springer. Billari, F. C., & Piccarreta, R. (2005). Analysing demographic life courses through sequence analysis. Mathematical Population Studies, 12, 81–106. Billari, F. C., Prskawetz, A., & Fu¨rnkranz, J. (2000). Timing, sequencing and quantum of life course events: A machine learning approach, Working Paper WP-2000-010, Rostock: Max Planck Institute for Demographic Research. Blossfeld, H.-P. (1996). Macro-sociology, rational choice theory, and time. A theoretical perspective on the empirical analysis of social processes. European Sociological Review, 12, 181–206. Blossfeld, H.-P., & Mills, M. (2001). A causal approach to interrelated family events: A crossnational comparison of cohabitation, nonmarital conception, and marriage. Canadian Studies in Population, 28, 409–437. Blossfeld, H.-P., & Rohwer, G. (2002). Techniques of event history modeling (2nd ed.). Mahwah, NJ: Lawrence Erlbaum.

280


Blossfeld, H.-P., Hamerle, A., & Mayer, K. U. (1989). Event history analysis. Statistical theory and application in the social sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Bocquier, P. (1996). L’analyse des enqueˆtes biographiques a` l’aide du logiciel STATA. Paris: CEPED. Breiman, L. (2001). Statistical modeling: The two cultures. (with comments and rejoinder). Statistical Science, 16, 199–231. Bryson, A., Dorsett, R., & Purdon, S. (2002). The use of propensity score matching in the evaluation of active labour market policies. Working Paper no. 4, Department of Work and Pensions, London. Courgeau, D., & Lelie`vre, E. (1992). Event history analysis in demography. Oxford: Clarendon Press. Courgeau, D., & Lelie`vre, E. (1996). Changement de paradigme en de´mographie. Population, 51, 645–653. Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society, 34, 187–220. Dehejia, R. H., & Wahba, S. (1999). Causal effects in nonexperimental studies: Reevaluation of the evaluation of training programs. Journal of the American Statistical Association, 94, 1053–1062. Dykstra, P. A., & van Wissen, L. J. G. (1999). Introduction: The life course approach as an interdisciplinary framework for population studies. In: L. J. G. van Wissen & P. A. Dykstra (Eds), Population issues. An interdisciplinary focus (pp. 1–22). New York, NY: Kluwer Academic/Plenum Publishers. Eerola, M. (1994). Probabilistic causality in longitudinal studies. New York: Springer. Elder, G. H. (1985). Life course dynamics trajectories and transitions, 1968–1980. Ithaca, NY: Cornell University Press. Giele, J. Z., & Elder, G. H., Jr. (Eds) (1998). Methods of life course research. Qualitative and quantitative approaches. Thousand Oaks, CA: Sage. Heckman, J., & Robb, R. (1985). Alternative methods for evaluating the impact of interventions. In: J. Heckman & B. Singer (Eds), Longitudinal analysis of labor market data. New York: Cambridge University Press. Hedstro¨m, P., & Swedberg, R. (Eds) (1998). Social mechanisms. Cambridge: Cambridge University Press. Hobcraft, J., & Murphy, M. (1986). Demographic event history analysis: A selective review. Population Index, 52, 3–27. Holland, P. W. (1986). Statistics and causal inference. Journal of the American Statistical Association, 81, 945–970. Hotz, V. J., Mullin, C. H., & Sanders, S. G. (1997). Bounding causal effects using data from a contaminated natural experiment: Analysing the effect of teenage childbearing. Review of Economic Studies, 64, 575–603. Keane, M. R., & Wolpin, K. I. (1997). The career decisions of young men. Journal of Political Economy, 105, 473–522. Kohler, H.-P., Skytthe, A., & Christensen, K. (2001). The age at first birth and completed fertility reconsidered: Findings from a sample of identical twins. Working Paper WP 2001-006, Max Planck Institute for Demographic Research, Rostock. Le Goff, J.-M. (2002). Cohabiting unions in France and West Germany: Transitions to first birth and first marriage. Demographic Research, 7, 591–624.


281

Lillard, L. A. (1993). Simultaneous equations for hazards: Marriage duration and fertility timing. Journal of Econometrics, 56, 189–217. Lillard, L. A., & Panis, C. W. A. (2000). AML multilevel multiprocess statistical software, release 1.0. Los Angeles, CA: EconWare. Mayer, K. U., & Tuma, N. B. (Eds) (1990). Event history analysis in life course research. Madison, WI: University of Wisconsin Press. Mazzuco, S. (2003). When a child leaves the nest. A comparative analysis of effects of children departures from home on parents’ wellbeing. Mimeo, University of Padova. McVicar, D., & Anyadike-Danes, M. (2001). Predicting successful and unsuccessful transitions from school to work by using sequence methods. Journal of the Royal Statistical Association, Series A, 165, 317–334. Mouw, T. (2005). Sequences of early adult transitions: How variable are they, and does it matter? In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research and public policy. Chicago: University of Chicago Press. Painter, G., & Levine, D. I. (2000). Family structure and youths’ outcomes. Which correlations are causal? The Journal of Human Resources, 35, 524–549. Petersen, T. (1995). Analysis of event histories. In: G. Arminger, C. C. Clogg & M. E. Sobel (Eds), Handbook of statistical modeling for the social and behavioral sciences (pp. 453–517). New York: Plenum Press. Rindfuss, R. R. (1991). The young adult years: Diversity, structural change, and fertility. Demography, 28, 493–512. Rohwer, G., & Po¨tter, U. (2003). TDA user’s manual. Bochum: Ruhr University Bochum. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41–55. Rosenbaum, P. R., & Rubin, D. B. (1984). Reducing bias in observational studies using subclassification on the propensity score. Journal of the American Statistical Association, 79, 516–524. Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring and the life course. Annual Review of Sociology, 23, 233–261. van der Heijden, P. G. M. (1987). Correspondence analysis of longitudinal categorical data. Leiden: DSWO Press. Wu, L. L. (2000). Some comments on ‘‘sequence analysis and optimal matching methods in sociology: Review and prospect’’. Sociological Methods & Research, 29, 41–64. Wu, L. L. (2003). Event history models for life course analysis. In: J. Mortimer & M. Shanahan (Eds), Handbook of the life course (pp. 477–502). New York: Plenum Press. Yamaguchi, K. (1991). Event history analysis. Newbury Park, CA: Sage.


282

LIFE COURSE DATA IN DEMOGRAPHY AND SOCIAL SCIENCES: STATISTICAL AND DATA-MINING APPROACHES$ Gilbert Ritschard and Michel Oris 1. FROM DEMOGRAPHIC ANALYSIS TO LIFE COURSE APPROACH This chapter has essentially a methodological purpose. It discusses recent advances in statistical event history analysis and Markov models and promotes the use of tools from the developing field of data mining, with special attention to the discovering of characteristic sequences and induction trees. Before turning to these methodological aspects, we begin here by explaining why demographers have been relatively reluctant to implement the life course paradigm and methods, while the quantitative focus and the concepts of demographic analysis a priori favored such implementation. A real intellectual crisis has been needed before demographers integrated the necessity to face up the challenge of shifting ‘‘from structure to process, from macro to micro, from analysis to synthesis, from certainty to uncertainty’’ $

The funding for this research is from Swiss National Science Foundation, projects 1114-68113 and 100012-105478.


283

284

GILBERT RITSCHARD AND MICHEL ORIS

(Willekens, 1999, p. 26). This retrospective look also shows impressive progresses to promote a real interdisciplinarity in population studies, knotting the ties between demography and the social sciences. Although demographic analysis has a long history (see Dupaˆquier & Dupaˆquier, 1985), the methods still used today have essentially been elaborated between the mid-nineteenth and the mid-twentieth centuries in Western societies that felt successively threatened by race degeneration, declining birth rates, and ageing. The macro frame was that of the demographic transition, i.e. the evolution from young populations with high fertility and high (especially infant and child) mortality to ageing populations with net reproduction below the threshold of generations’ renewal and a tremendous increase in life expectancy at birth. From a methodological point of view, a starting point in demographic analysis has been the mortality table. It implies a dynamic perception of population with entrances (births or in-migrations) and exits (deaths or outmigrations), and the idea that other events (like migration) can censor a given risk (like mortality). It rapidly constrained the conceptual distinction between a generation – i.e. those who are born in the same year or period – and a cohort – i.e. those who are experiencing an event (marriage for example) in the same year or period. The mortality table also resulted not only in an average age at death but also in a distribution of the risk along the life course, providing a survival curve. Finally, the immediate comparison of these curves among sexes, and even more among matrimonial statuses, revealed selection processes, like those supporting the over-mortality of singles compared to married people. Heterogeneity and differential frailty were not ignored. After the generalization of mortality analysis (from mortality- to life tables), certainly no scientific discipline was better prepared for the life course methods than demography. Nevertheless, the paradigms clearly diverged. First, dealing with structures and flows, demography has been a science of reconstruction and description of patterns and behaviors, through a wellestablished quantitative methodology, and the conviction that higher the number of observations, more accurate – and possibly useful – were the results (for a typical example, see Vallin, 2001). Demography was a science of the masses, growing or stagnating, young or old, but not of the individuals! Second, the engagement of generations of scholars was largely motivated by the central character of population issues and the location of demography at a crossroad between economy, sociology, epidemiological studies, territorial analysis, political sciences, and, more recently, cultural and gender approaches. However, research and collaborations were in reality highly segmented, with a clear tendency to specialization on a geographical

Life Course Data in Demography and Social Sciences

285

and/or thematic basis (typically mortality, fertility, marriage and family formation, or dissolution, migrations, structures, prospective). Among many others, the last edition of the excellent Encyclopedia of Population (Demeny & McNicoll, 2003) illustrates that propensity. Such segmentation was clearly inscribed in the methods. In the estimations of mortality, mobility was statistically treated as a censor but explicitly presented as a bias for a ‘‘pure’’ analysis of mortality. Clearly, the approach consisted in studying a demographic behavior as independently as possible from the other ones, without systemic perspectives. Third, the demographic evolution made apparent the limits of the established methodology. Both mortality and life tables can be calculated longitudinally based on the observation of generations, or on crosssectional data, i.e. the observation of deaths by age classes at a given time point – mixing thus generations with different history together. For the simplest table, that of mortality, since expectation of life at birth now exceeds 80, the longitudinal approach implies a population reconstruction from at least the 1920s, what is quite difficult, especially if we do not accept the hypothesis of a null effect of migration. Inversely, all the statistical offices of the developed countries collect the data for the calculation of cross-sectional measures for a long time. However, can we accept the underlying hypothesis of continuity while the duration of life wins 1 year every 3–4 years and while we know that the – generational – age distribution of those gains has drastically changed during the last decades? Similarly, while in the context of the so-called ‘‘second demographic transition’’ there are so many changes in the fertility calendar, do we have to constrain ourselves to the observation of those generations who have finished their fertile life and renounce to study the present with other indicators than those of the moment? What is today the rationale of detailing the access to marriage while cohabitation is rising? How record an informal event like entrance in cohabitation? Both data collection and analytical tools have been challenged by recent changes in demographic behaviors and family dynamics (for a more in-depth discussion of those issues, see Caselli, Vallin, & Wunsch, 2001). A real intellectual crisis resulted from the hesitation about the status of demography within the social sciences, as well as from the frustration against segmentation and the deficiency of old methods. The conscience that description, especially some quantification with a pretension of objectivity, hid and diffused ideological visions about ‘‘good’’ or ‘‘optimal’’ populations also grew (Greenhalgh, 1996; van de Kaa, 1996; Ve´ron, 1993; Szreter, 1993; Hogdson, 1988, 1991).

286


Among the many reactions, revisions, and re-examinations, new approaches and new methods rapidly emerged. No significant use of the life course statistical tools can be observed before the mid-1980s, while for example Cox’s foundling paper is dated 1972. When they finally have been integrated by demographers, the new methods found many uses. Probably the most obvious progress they supported was to replace demography in its family setting. Something that could seem very strange, but perfectly illustrates this assertion, is indeed the discovery, precisely in the 1980s, of an almost complete absence of dialogue between demography and family sociology. While family is the place where most of the demographic behaviors take place and, to some extent, are decided, ‘‘few textbooks on population contain a chapter devoted to the demography of the family. Where such chapter does exist, it is generally shorter and more superficial than those that deal with fertility, mortality, nuptiality, and migration, or with the dynamics of age structure’’ (Ho¨hn, 1992, p. 3). In 1982, the International Union for the Scientific Study of Population created an ad hoc committee to develop its study, but even in 1992 the animators of this group saw family demography as ‘‘a recent and relatively underdeveloped branch of population studies’’ (Berquo & Xenos, 1992, p. 8). Its development has been extraordinary in the last years. Francesco Billari chapter in this volume provides a nice illustration of such a change, which is part of a shift from macro to micro, from an emphasis on macro-economic evolutions as the essential determinants of demographic ‘‘answers’’, to a multi-causal – multivariate – approach of behaviors, a shift also from average results to a more detailed study of distributions. In a quantitative discipline, major evolutions necessarily imply to take up technical challenges. ‘‘The traditional demographic analysis of such events as births, marriages, divorces, deaths, and migration has the advantage that number of these events can be related to individuals in the same age group and can, therefore, be measured more easily and included in models. The inclusion of other family members in such analyses causes difficulties because they will generally differ in age and sex, and complications are also introduced because they do not generally live together continuously’’ (Ho¨hn, 1992, p. 3). Although several attempts have been made to construct a ‘‘household demography’’ (Van Imhoff, Kuijsten, Hooimeijer, & van Wissen, 1995), the life course paradigm and its methodological individualism clearly imposed themselves. Offering both concepts and statistical methods, it represents a shift toward microanalysis of individual data and causal research that not only deeply renews the discipline, but also provides the vocabulary for a new interdisciplinarity, first within the social sciences, then beyond (Blossfeld &


287

Rohwer, 2002; Dykstra & van Wissen, 1999). The first substantial gain has been the study of multiple events, marriage and first birth, or moving and starting a new job for instance, a kind of investigation that also raises the issue of event sequencing and interactions that is typically treated with event history analysis. If people have several careers that they must make compatible, their life transitions also reflect socioeconomic constraints, cultural norms (about the ‘‘proper’’ age, sex, or behavior), as well as compromises between several individual aspirations within or beyond the domestic unit. Through researches in this huge area, family demography made for sure tremendous progress during the last 20 years. However, the shift has been so sudden that globally the complexity of causalities remains too often underestimated (see Courgeau & Lelie`vre, 1993; Blossfeld & Rohwer, 2002; Bocquier, 1996; Alter, 1998; Billari, 2005), as well as several technical traps. The problem is essentially that when studying a population of individuals observed along the time, since each life, the product of complex and multiple interactions is, as a matter of fact, unique. Hence, interpreting and generalizing from samples require several cautions. In the next section, we recall the main event history regression models and discuss the question of heterogeneity. We cannot consider that the elaboration of indicators at an individual level about household, family, and community contexts is enough to deal with the more and more raised issue of ‘‘linked’’ or ‘‘interdependent’’ lives (Hagestad, 2003). We show the interest of robust estimates and shared frailty in that perspective. In the same section, we also present the Markovian models that are particularly useful for the study of transitions within a set of states (matrimonial or social status, for example) periodically observed. In the interdisciplinary perspective, which is one of the life courses, we consider it important to go beyond the simple transitions typically studied in demography (from single to married, from the first to a possible second child, from life to death, and so on), and to investigate how, from a starting position, a destination is selected among several possible. While family dynamics and life courses are more and more open, such investigations are essential to deal with the characterization of transitions as ‘‘normal’’ or ‘‘non-normal’’ without falling again in the trap of ideological reading (see, for example, the discussion in Oris & Poulain (2003) about the stigmatization of early home leaving). Indeed, we assess more globally that there is a deficit of research on trajectories between aggregate descriptions and causal analysis. Regression models attempt to quantify how a factor, measured by an indicator, affects a risk. However, such results tell us nothing about the calendar and no more about the alternatives to this risk in life courses. It is essential to look

288


carefully at transitions in trajectories to target properly a causal analysis, and this step is clearly too often superficial, if not absent. Several methods, recently developed or recently made available in statistical packages, offer opportunities to fill this gap. Among them, we promote in Section 3 some highly flexible heuristic tools from the developing field of data mining, especially mining event-sequential association rules, and induction trees that seem to us the more promising for life course data analysis.

2. STATISTICAL MODELING OF LIFE EVENTS Life course data are longitudinal in their essence. Here, we focus on events, an event being the change of state of some discrete variable, e.g. the marital status, the number of children, the job, or the place of residence. Such data are collected mainly in two ways: as a collection of time-stamped events or as state sequences. In the former case, each individual is described by a collection of time-stamped events, i.e. the realization of each event of interest (e.g. being married, birth of a child, end of job, moving) is mentioned together with the time at which it occurred. In the latter case, the life events of each individual are represented by the sequence of states of the variables of interest. Panel data are special cases of state sequences where the states are observed at periodic time. The first kind of data is typically analyzed with event history regression methods, while methods for state-sequence analysis like Markov transition models are best suited for the latter. We briefly discuss hereafter the scope and limits of these approaches.

2.1. Event History Regression Models When we have time-stamped events, the question of interest is the duration of the spell between two successive events, or somewhat equivalently, the hazard rate h(t) for the next event to occur precisely after a duration t, i.e. the conditional probability for the event to occur at t knowing that it did not occur before t. Longitudinal-regression models focus on this aspect. They express either the duration or the hazard rate as a function of covariates. It is worth mentioning that these models are also known as survival models, especially in area like biomedicine and engineering where the event of concern is just death or breakdown. There are continuous-time models and discrete-time forms. With continuous time, the main formulations (see Blossfeld, Hamerle, & Mayer, 1989;


289

Courgeau & Lelie`vre, 1993) are as a duration model or as a proportionalhazards model. Duration models consider ln(T), the logarithm of the time to the event, as a linear function of the explanatory factors. Proportionalhazards models suppose that the ratio between the hazard for a given profile (in terms of the covariates) and that for a reference baseline profile remains constant over time and expresses the logarithm of this ratio (or proportion) as a linear function of the covariates. Duration models, also known as accelerated failure time models, assume usually an exponential, Weibull, log-normal, log-logistic, or gamma distribution for the duration T. The proportional-hazards model is compatible with for instance, exponential, Weibull, and Gompertz duration distributions. It includes also the perhaps most widely used Cox (1972) semiparametric model that requires no assumptions on the form of the duration distribution. Most statistical packages (SAS, S-Plus, Stata, R, TDA, etc.) provide procedures for estimating such models. At least until version 13, SPSS, however, offers only support for the Cox model. Discrete-time models (see Allison, 1982; Yamaguchi, 1991) include the proportional hazard-odds model, also owe to Cox (1972), the discrete proportional-hazards model (Aranza–Ordaz, 1983), and the log-rate model (Holford, 1980). In the first model, it is not just the hazard ratio, but the ratio of the odds of the hazards that is supposed to be constant and having a logarithm depending linearly on the covariates. The discrete proportional hazards model expresses the log minus log of the complementary hazard as a linear function of the covariates. The log-rate model on its side expresses the log-hazard in terms of proper and interaction effects of categorical variables and also possibly of their interactions with duration. For the estimation of the proportional hazard-odds model, some assumptions are usually required upon the baseline hazard-odds. Letting b0t be the baseline log hazard-odds after a duration t, the most common assumptions are that it remains constant with t, is linear in t (Gompertz), or is linear in ln(t) (Weibull). With these assumptions, a proportional hazardodds model can, if we organize the data in a person–period form, simply be estimated as a logistic regression. Hence, it can be estimated by any software that proposes logistic regression. Likewise, a log-rate model can be estimated with any log-linear model procedure that allows for weighted cell frequencies. Indeed, the log-rate model is a log-linear model of the weighted number of events occurring in a time interval, the weight being the inverse of the population at risk in this interval. The fitting of a discrete proportional hazards model requires the perhaps less frequently implemented procedures for binary regression with a complementary log-log link.

290


A common issue with the time to event models is the handling of censored data. Censored data occur when the observed start (left) and/or end (right) time of a spell are not its actual start and end time. For instance, if we observe job duration, some jobs may not be terminated at the time of the survey and are hence right-censored. Though no event is recorded at the end of the right-censored spells, these cases are taken into account by entering the population at risk for job length lower or equal to the observed duration. Another issue is the handling of time-varying covariates. The solution is quite straightforward in the discrete-time setting that works on person–time data. For the continuous case, there are two major solutions: an ad hoc extension of the Cox model that allows for discrete-time-varying covariate and the episode-splitting approach (see Blossfeld & Rohwer, 2002 for details). Time-varying covariates offer a way to test and relax the somehow strong proportionality assumption required by most hazard-rate models. Indeed, this assumption implies the time independency of the ratio of hazards of any two individuals, which clearly does not hold when the ratio depends on a time-varying variable. It is common practice to check the significance of the interaction of a supposed time-independent variable with t or ln(t). A significant interaction would provide evidence against time invariance (see Therneau & Grambsch, 2000 for other tests of proportional hazards and more advanced developments of the Cox model). This event-history modeling, especially the Cox proportional-hazards and discrete-time proportional hazard-odds models, has become popular among demographers. Together with other social science scientists, historical demographers have to face issues like competing events (multiple destinations), repeatable events, and interacting events. The first two can easily be handled with a software like TDA (Rohwer & Po¨tter, 2002) that supports episodes defined by four parameters, namely the origin state, the start time, the destination state, and the end time. The interaction between events, marriage end, and first child, for instance, needs a simultaneous equation approach that has been investigated by Lillard (1993), and is discussed more in depth in Billari’s contribution to this volume. 2.1.1. Shared Heterogeneity and Multi-Level Modeling A further issue of importance, shared heterogeneity, is concerned with the sampling nature of the data. These are often clustered, i.e. the individual data come from a selection of groups, parishes, or families for example. In such cases, members of a same group share a same contextual framework and it is then of primary importance to distinguish effects that hold at the

291


group level from those that work at the individual level. A very concrete example is the study of orphans’ survival after father’s death by Beekink, van Poppel, and Liefbroer (1999) for a 19th-century Dutch provincial town. In the event-history file, initially each orphan was considered as a single individual while there were not individuals but groups of siblings that entered in the population at risk because of a shared event – dad’s death – and supported this experience while sharing the same household context. Taking into account the interrelatedness of the observations changed the results! Along the same line, both in contemporary and historical demography, the issue of the death clustering at the family level is a growing concern (Alter, Oris, & Brostro¨m, 2001). All those studies extend the original discussion of ‘‘the impact of heterogeneity in individual frailty on the dynamics of mortality’’ by Vaupel, Manton, and Stallard (1979). To explain this aspect, let us consider the case of a simple linear regression of the number of children on the education level in the presence of three clusters like those depicted in Fig. 1, where the clusters are, let us say, three villages. A simple regression on the whole data set is a straight line with a positive slope, indicating that the number of children increases with education. This effect clearly holds at the aggregated village level, i.e. the higher the average education level in a village, the higher the average number of children. This aggregated effect results despite the regression is fitted on individual data. A separate regression on each cluster exhibits a negative

9 8

y = 15.6 - 0.8 x

y = 12.5 - 0.8 x

7

Children

6 5 4 3

y = 3.2 + 0.2 x

2 y = 6.2 - 0.8 x

1 0 1

Fig. 1.

3

5

7 9 Education

11

13

15

Multi-Level: A Simple Example with Three Clusters.

292


Table 1.

Alternative Linear Models in Presence of G Clusters g.

Model

Constant

Effect of Covariate

Variance of Error Term

Number of Parameters

m1

Average model

Same

Same

Same

1+c+1

m2 m3

Group specific Group specific

Group specific Group specific

Group specific Same

G(1+c+1) G(1+c)+1

m4

Independent Seemingly independent Dummies

Group specific

Same

Same

G+c+1

m5

Random effects

2(1+c)+1

Shared frailty

Random across groups Same

Same

m6

Random across groups Random across groups

Same

2+c+1

slope in each of the three villages, indicating a negative effect of education on the number of children at the individual level. Indeed similar misleading results may appear when event-history regressions are fitted on clustered data as illustrated by the examples discussed by Beekink et al. (1999) and Alter et al. (2001). What are the solutions? Table 1 summarizes alternative formulations that can be adopted when we are in the presence of G groups. For simplicity, we consider here regression models with c covariates, generalization to more complex models like event-history models being straightforward. Model m1 will capture effects at the group level. In models m2–m4, differences between groups are introduced by means of additional parameters, an approach that is suitable as long as G is not too large. Model m2 fits separate models on each clusters, while in m3, the regressions are only seemingly independent, since the variance of the error term is supposed to be the same in each group. Model m4 corresponds to the well-known case where, for each group, a specific effect is introduced as a dummy variable. For a large number of groups, random effect models1 m5 and m6 are best suited. In these models, the regression coefficients are allowed to vary randomly from one group to another. In the shared frailty formulation, only the constant is random, while the other coefficients remain the same for all groups. The main advantage of these random effect formulations is that their number of parameters is, as can be shown from the last column, independent of the number of clusters. Random effect models m5 and m6 may, therefore, have a much lower number of parameters than models m2–m4 when G is large. Even if we are interested in the aggregated effect, estimating them with individual data, as with model m1, for example, requires some caution.


293

Indeed, the standard errors of the aggregated effects are derived from individual residuals, which may either over- or underestimate the betweengroup discrepancy. For instance, in our example of Fig. 1, leaving out in turn each of the three groups leads to great variations in the slope that would be underestimated by the classical standard error. This aspect has been investigated among others by Kish and Frankel (1974) and, for the Cox model, for instance, by Lin and Wei (1989). In such settings, it is good practice to use robust estimates of the variance of the regression coefficients. Such robust estimates are usually obtained as grouped jackknife estimates, i.e. by measuring the discrepancy of estimates obtained by leaving out successively each of the G clusters, and can be expressed as sandwich estimates (see Therneau & Grambsch, 2000, pp. 170–173). Facilities for dealing with clusters are offered by several statistical systems, Stata 8, S-Plus 6.2, and R 2.0 for instance. All the three mentioned programs propose options to get robust standard errors. They also permit the introduction of a shared frailty in parametric-hazard rate and Cox models. Complete random effects are only available with discrete models that can be fitted with logistic regression procedures. Indeed, logistic models are special cases of generalized linear models (GLM).2 Hence, multilevel-logistic regression is available whenever multilevel GLM is implemented. Barber, Murphy, Axinn, and Maples (2000), for instance, show how to estimate a model with several random effects with the HLM (Bryk, Raudenbush, & Congdon, 1996) and MLN (Goldstein et al., 1998) programs. 2.1.2. Illustration To illustrate the scope of robust standard errors and shared frailty, we consider a data set of 5,351 migrants collected from the 19th-century population registers of the Belgian commune of Sart (see Alter & Oris, 2000; Alter et al., 2001, for a detailed description). This data set provides, among others, information about the emigration date, the destination and the date of return after emigration. Table 2 shows results of the fit of a continuoustime Cox model. The hazard modeled is that of return after a time between 0 and 5 years, no return or return after 5 years being censored. We fitted a basic model, i.e. without the cluster or frailty options, the same model but requesting robust standard errors for the coefficients, and the model with a gamma g(1/y,1/y) distributed frailty term shared by members of a same family.3 The hazard ratios reported are just the exponential of the coefficients. They indicate the hazard ratio for two profiles that differ by one unit of the corresponding variable. For the frailty model this interpretation holds,

294


Table 2.

Cox Model for Return within 5 Years after Emigration. Coefficient

Hazard Ratio

p-Value (in %)

Basic Frailty Basic Frailty Basic Robust Frailty Economic ratio Man Single Born in Ardennes Age when leaving To To To To

1.02 0.28 0.40 0.25 0.01

Ardennes rural urban/indust other

Head or spouse of Child of head Other parenthood No parenthood Standard deviation

pffiffi y of family effect

0.30 0.18 0.52 0.17 0.00

2.76 0.76 1.49 1.29 1.01

1.35 0.83 1.68 1.18 1.00

0.2 0.1 1.2 4.1 12.0

3.8 0.2 1.2 15.0 17.0

45.0 5.6 0.3 28.0 62.0

Destination reference category 0.32 0.60 0.73 0.55 5.7 0.07 0.23 0.93 0.79 50.0 1.25 1.25 0.29 0.29 0.0

14.0 68.0 0.0

0.2 6.8 0.0

Parenthood reference category 0.02 0.25 1.02 0.78 89.0 0.12 0.27 1.13 0.76 54.0 0.50 0.54 0.61 0.58 6.7

90.0 56.0 7.3

19.0 26.0 9.0

1.75

0.0

Note: Sart 1812–1900, n ¼ 5,351.

assuming the two profiles have the same frailty. For instance, according to the basic model, the chances to return for a single are about one and a half times the chances to return for a non-single. Likewise, the probability to return is for a man about 3/4 of that for a woman. We checked on the basic model that none of the time-covariate interactions is significant, which comforts the proportionality assumption. The coefficients are indeed the same for the basic and robust standarderrors models. The significance of the coefficients differs, however, as can be seen from the p-values. To be born in the Ardennes is significant at the 5% level when we do not care about the cluster effect, while it is clearly not when we control for it. This indicates that the seemingly significant birth-place effect does not work at the family aggregated level. Likewise, we may notice that, though the effect of the household economic ratio is significant among families, its significance is not as clear as we would expect from the basic model. Let us now look at the results with a family shared frailty. First, we may notice the highly significant variance of the random term, which clearly indicates a between-families discrepancy. Two variables that looked


295

significant become non-significant, namely the gender (man) and the economic ratio. This is not surprising for the latter, which is a typical family contextual factor shared by members of the same family. Gender, on the other hand, is clearly an individual characteristic. Its lack of significance in the frailty model seems to indicate that the effect is not systematic within the families. Its overall significance follows probably from differences among male and female singles. A reverse phenomenon is observed for the rural destination effect that becomes significantly different from the reference Ardennes in the frailty model.

2.2. Markov Transition Models In the presence of state sequences in panel data form, the natural question is what are the transition probabilities from the states at time t 1 to the possible states at time t, and how are these probabilities affected by individual histories or contextual characteristics. Homogenous-Markov models assume that these probabilities are independent of time t. In first-order models, the transitions are supposed to depend only upon the state at t 1, which means that the first lag summarizes the whole history of states at t 1 and before. Models of higher order k consider that the transitions depend on k lags, i.e. on the states at t k,y,t 1. Thus, basic Markov models state that the transition probabilities remain constant over time and depend on a limited, usually small, set of previous states. Markov models of order k generate, when we are in the presence of s states, sk transition distributions, i.e. a huge number of probabilities. They may be approximated by mixture transition distribution (MTD) models (Raftery & Tavare´, 1994; Berchtold, 2001; Berchtold & Raftery, 2002) that involve a much lower number of parameters, which renders the models easier to interpret. Other extensions of the Markov model include the hidden Markov model (HMM) (see Rabiner, 1989; MacDonald & Zucchini, 1997) in which the successive states of the observed variables are only indirectly linked through an unobserved Markov chain and the double chain Markov model (DCMM) (Paliwal, 1993; Berchtold, 1999, 2002), which states that the observed states are outcomes of a Markov process randomly selected by a hidden process. The use of hidden processes is a way to relax the usually strong homogeneity assumption. For example, when studying social mobility with data covering a whole century, it is hardly defendable to assume that the same process works during the whole period (Lynch, 1998, p. 96).

296


Despite their interest, there has been only a limited use of Markov models, especially of non-homogenous ones, by historians and demographers. A search on ‘‘Markov’’ within the famous Population Index database4 results in only 28 hits among thousands of references. Moreover, most of those 28 hits refer to working papers or highly focused articles (with an emphasis on the study of multistate population dynamics). The main reason for such a limited use is that standard statistical packages offer only limited facilities to fit such models. The available tools require a heavy coding task that discourages most potential users. We can expect, however, that Markov modeling will become much more popular with the recent release of March 2 (Berchtold & Berchtold, 2004). This software offers a friendly way to estimate Markov models without writing down any line of code.

2.2.1. Illustration To illustrate the nature of knowledge we can expect from such an analysis, we consider here the Blossfeld and Rohwer (2002) sample of 600 job episodes extracted from the German Life History Study. The episodes have been classified into three job-length categories: (1) p3 years, (2)43 and p10 years, and (3)410 years, and the data reorganized into 162 individual sequences of 2–9 job episodes, dropping the cases with a single episode. The question considered is how the present episode length depends upon those of the preceding jobs. Notice that the job-length sequences considered here are not panel data, which demonstrates that Markovian models are not restricted to panel data. In this setting, the subscript t refers simply to the position in the sequence rather to a specific time period. The first- and second-order homogenous transition matrices are given in Table 3. The same table also gives the distribution of the independence model in which the transition probabilities stay the same irrespective of the previous job length. Let us briefly illustrate how these tables should be read. The independence distribution implies that the overall probability for a new job to be a short one is 50%, while this probability is 35% for a medium job and 15% for a long job. The first-order matrix indicates that the probability that a new job started after a short one has 57% chances to be again a short job. This probability falls to 43% after a job of medium length and to 20% after a long job. From the second-order matrix, it follows, for instance, that this same probability is 55% when the preceding short job was itself preceded by a short one, 60% when the preceding short job followed a medium job and 100% when the preceding short job followed a long job. The last column in the tables gives the half-length of a conservative 95% confidence

297


Table 3.

First and Second-Order Homogenous Markov Matrices. t–2

1

Job Length at t 2

3

Half Confidence Interval

0.50

0.35

0.15

0.07

1 2 3

0.57 0.43 0.20

0.30 0.42 0.53

0.13 0.15 0.27

0.10 0.13 0.29

1 1 1 2 2 2 3 3 3

0.55 0.60 1 0.37 0.50 0.45 0.33 0 1

0.30 0.30 0 0.45 0.41 0.33 0.17 0.87 0

0.15 0.10 0 0.18 0.09 0.22 0.50 0.13 0

0.11 0.20 0.65 0.18 0.20 0.38 0.46 0.40 1

t–1

Independent First Order

Second Order

1 2 3 1 2 3 1 2 3

interval for the probabilities in the concerned row. Hence, probabilities smaller than this half-length should be considered as non-significant. A glance at these tables leads to the following remarks. The first-order matrix exhibits some differences in the transition probabilities after a short (1), medium (2), or long (3) job. After a first job, the probability to start a short job is significantly higher than to start a medium or long job, while this is not the case after a medium or long job. The second-order matrix does not provide evidence on the impact of the second lag job length. The main differences concern the transition probabilities after long jobs (3), which are mostly not statistically significant due to the low number of cases concerned. This was confirmed by fitting an MTD model for which we obtained a weight of 1 for the first lag and, hence, 0 for the second lag. For relaxing the homogeneity assumption, we consider an HMM model with a two-hidden-state process. Fitting this model, we get the distribution of the initial state of the hidden variable, the transition matrix of the hidden process, and the distributions of the transition to the job-length categories associated to each of the two hidden states. These results are given in Table 4. In addition, we get estimates (not shown here) of the most likely sequence of hidden states associated to each observed sequence. Looking at the cross tabulation below of these estimated hidden states with the observed job length we see that the first hidden state is mainly associated to

298


Table 4. Hidden State at t-1

t

Two State Hidden Markov Model. Hidden State at t 1 2

Half Confidence Interval

Initial

0.56

0.44

0.11

1 2

0.78 0.53

0.22 0.47

0.12 0.19

1 1 2

0.75 0.05

Job Length at t 2

3

0.23 0.58

0.02 0.37

0.12 0.18

short jobs and the second hidden state to medium and long jobs. This may suggest considering only two types of jobs: p3 years and 43 years. Hidden

1 2

Observed 1

2

3

118 0

19 65

0 35

Table 5 summarizes goodness-of-fit statistics for our fitted models and for the sake of comparison of the independence model. The shown statistics are the number of independent parameters p, the deviance measured as minus twice the log-likelihood5 ( 2LogLik), the likelihood-ratio w2 statistics that measures the improvement in 2LogLik over independence, its associated degrees of freedom and its significance level, the pseudo R2 that gives the relative improvement in 2LogLik and the Akaike (AIC) and Bayesian (BIC) information criteria.6 These figures show that the fitted models do not make much better than the independence model. We get the smallest 2LogLik value for the second-order homogenous model, but at the cost of 11 additional independent parameters. The first-order homogenous model is the only one that significantly improves the 2LogLik of the independence model. It is also slightly better in terms of the AIC. However, no model outperforms the independence model in terms of the BIC. These relatively bad results are largely attributable here to the insufficient number of data considered. This stresses a limitation of this Markov-modeling approach,

299


Table 5. Global Model Goodness-of-Fit Statistics. Model m

p

Independent Homogenous order 1 Homogenous order 2 HMM 2 states

2 6 13 7

2LogLik 472.8 462.6 460.6 468.6

w2

df

Sig

BIC

AIC

Pseudo R2

0 10.2 12.2 4.2

0 4 11 5

— 0.04 0.35 0.52

483.7 495.4 531.7 506.9

476.8 474.6 486.6 482.6

0 0.022 0.026 0.009

Note: Number of sequences ¼ 107, usable n ¼ 237.

namely the complexity of the models in terms of number of estimated parameters that requires a very large number of data.

3. MINING LONGITUDINAL LIFE COURSE DATA Despite the last decade great boost in the use of data-mining tools for the knowledge discovery from data (KDD) in fields ranging from genetics to finance, from marketing to medical diagnosing, from text analysis to image or speech recognition, such approaches have received only little attention for extracting interesting knowledge from longitudinal data describing life courses. An important exception is Blockeel, Fu¨rnkranz, Prskawetz, and Billari (2001) who showed how mining frequent itemsets may be used to detect temporal changes in event-sequences frequency from the Austrian FFS data. In Billari, Fu¨rnkranz, and Prskawetz (2000), three of the same authors also experienced an induction-tree approach for exploring differences in Austrian and Italian life-event sequences. We initiated ourselves (Oris, Ritschard, & Berchtold, 2003) social-mobility analysis with induction trees. Data mining is mainly concerned with the characterization of interesting pattern, either per se (unsupervised learning) or for a classification or prediction purpose (supervised learning). Unlike the statistical-modeling approach, it makes no assumptions about an underlying process generating the data and proceeds mainly heuristically. Beside their non-parametric or assumption-free characteristic, datamining methods present also the advantage for our social demographic framework to be able to handle sequences of the various family, education, work, health, emotional, and other personal events that define a life course. They seem in that regard promising tools for gaining knowledge about life trajectories and should thus usefully complement the previously discussed

300


statistical methods. Event-history models, for example, focus on the risk of a given transition, but do not provide insights on trajectories. Markov models, on the other hand, attempt to characterize the stochastic process that drives successive transitions between states. They provide in that sense some synthetic information about trajectories. However, only trajectories between states of a generally unique variable, social, or civil status, for example, can be investigated this way. Markov models, even those allowing for covariates, can hardly handle together the various life events. Furthermore, Markov models remain quite rigid by assuming that the transition probabilities do not depend upon the present time but only on a small limited number (the order of the model) of previous states. From a substantial standpoint, the hereafter discussed sequence-mining approach is best suited to discover among the many possible trajectories, for example, from the diversity of formations to the diversity of working lives, those that are typical of real life courses of real persons and by contrast those that are atypical. Since data-mining methods are mainly assumption-free, exploring trajectories with them may answer to the criticisms of the French sociologist Pierre Bourdieu (1986) about the ‘‘biographical delusion’’. Bourdieu, in fact, denounced the concept of ‘‘life cycle’’, and its emphasis on norms, norms supposed to lend to ‘‘normal’’ trajectories. With the assumption-free mining of longitudinal data, we precisely pass the boarder between the ‘‘causality’’ or ‘‘data-modeling culture’’ and what Breiman (2001) calls the ‘‘algorithmic culture’’ (see Billari, this volume). In the rest of this section, we shortly describe the mining of sequential rules and the induction tree approach, focusing on the nature of knowledge we may expect from such tools (for a more general introduction to data mining, see Hand, Mannila, & Smyth, 2001; or Han & Kamber, 2001). These books cover many more methods. The two tools discussed here are, however, in our mind, the two more promising ones for longitudinal data.

3.1. Mining Event-Sequential Association Rules Each life course can be seen as a sequence of life events: birth, important disease, recovering from disease, starting school, ending school, first job, first union, leaving home, first child, death of father, marriage, etc. Mining sequential-association rules aims at determining the most typical sequences or subsequences together with their frequencies, and at deriving association rules like having experienced the subsequence first job, first union, first child,


301

is most likely to be followed by a sequence marriage, second child. By contrast, indeed, mining frequent sequences and rules also reveals atypical life courses. Note that event sequences differ from state sequences as considered by Markov models or optimal matching. Nevertheless, sequence mining could as well be applied to state sequences. Technically, the mining of frequent-event sequences and sequentialassociation rules is a special case of the mining of frequent itemsets and association rules. In data mining, an itemset is a set of items that are selected together and an association rule is just a rule that says that if A occurs then B is very likely to occur too. The basic tuning parameters of the mining process are the support and the confidence thresholds. The support is the minimal frequency in the database for an item set to be selected, while the confidence of the rule is the probability that the consequence occurs when the premise is observed. These basic-selection criteria are complemented by other additional measure of the interestingness of the rule, like the proportion of the rule its counter examples. Most algorithms for seeking frequentitemsets and rules are variants of the well-known Apriori algorithm (Agrawal & Srikant, 1994; Mannila, Toivonen, & Verkamo, 1994). A typical application consists in finding the items that are more often ordered together by customers. Sequences that we consider here differ from general itemsets in that order matters. Multiple algorithms adapted for sequences have been proposed since the pioneering contributions by Agrawal and Srikant (1995) and Mannila, Toivonen, and Verkamo (1997).

3.1.1. Illustration We have not yet ourselves experienced a sequential rule mining analysis on demographic data. For the sake of illustration, we report here the analysis carried out by Blockeel et al. (2001). The data considered originated from the 1995 Austrian Fertility and Family Survey (FFS). The events analyzed are those of the partnership and fertility retrospective histories of 4,581 women and 1,539 men aged between 20 and 54 at the survey time. The observed women and men were partitioned into 5 years cohorts and the objective of the analysis was to discover frequent partnership and birth event patterns that mostly varied among cohorts. The mining was done by means of the Warmr process implemented in the ACE Data-Mining System (Blockeel, Dehaspe, Ramon, & Struyf, 2004). The search was not limited to simple sequences of strictly ordered events but allowed for more complex patterns combining multiple subsequences. An important pattern found was having a child after first union and having

302


both a marriage and a second child after this first birth, the marriage and second child being not ordered. The seeking of such not strictly ordered pattern requires indeed some filtering, namely the elimination of redundant patterns. For example, completing the above mentioned pattern with the additional condition of having a marriage after the first union would not bring any new information and is therefore redundant. Also, the rules generated were restricted to premises refereeing to the cohort. Finally, only patterns that exhibit a great discrepancy in the proportion of individuals satisfying it in each cohort were retained. Fig. 2 is an example of outcome provided by this analysis. It shows the strong declining proportion of individuals who started their first union when they married. The mining process found this pattern, i.e. date of first union equals date of marriage, to be the one that exhibits the strongest changes in frequency among cohorts. Indeed, many other patterns, sometimes more complicated, were also found to have great variability in their frequency.

Fig. 2. Negative Trend in the Proportion of First Unions Starting at Marriage. Source: Reproduced from Blockeel et al. (2001) with permission from the authors.


303

3.2. Social Transition Analysis with Induction Trees Let us now turn to induction trees and the insight they may provide on the understanding of mobility. In mobility analysis, the focus is on how states at previous time t 1, t 2,y, and possibly some additional covariates, influence the present state at t. This setting is very similar to that of Markovian models. In contrast with this parametric-modeling approach, the tree induction is, however, a non-parametric method. It provides a heuristic way to catch how the previous states and covariates jointly influence the state at t. Though we focus here on intergenerational social-mobility analysis, it is worth mentioning that the scope of induction trees for life course analysis is much broader. For instance, De Rose and Pallara (1997) used a tree approach for segmenting time to marriage curves of Italian women; Billari et al. (2000) used trees for analyzing differences in event sequences between Austrians and Italians; and we can easily imagine many other applications. Induction trees, i.e. decision trees induced from data, are basically supervised classification tools (Quinlan, 1986). As pointed out in Ritschard and Zighed (2003), they also convey powerful descriptive information. Their learning principle is quite simple and they produce easily interpretable results. An induced tree defines rules for predicting the value of a response variable from a set of potential predictors. The set of rules indeed characterizes a partition of the cases, each rule defining a class. The prediction inside each class of this partition is simply the modal-observed value when the response is categorical and the mean observed value when it is quantitative. In the quantitative case, the tree is called a regression tree (Breiman, Friedman, Olshen, & Stone, 1984). Extension in this case includes model trees (Malerba, Appice, Ceci, & Monopoli, 2002) and logistic model trees (Landwehr, Hall, & Frank, 2003), which use a linear or logistic regression for the prediction inside the classes of the partition. Tree algorithms have also been proposed for predicting functions instead of values and those that like RECPAM (Ciampi, Hogg, McKinney, & Thiffault, 1988) predict, for instance, survival functions may be of special interest for life course analysis. Here we consider only categorical responses, i.e. classification trees. The easiest way to describe the tree induction principle is by looking at an example. We begin therefore by describing the framework of the illustration we will consider. We use social family history data on intergenerational-social transition in the 19th-century Geneva (Ryczkowska & Ritschard, 2004). The data were collected from the marriage-registration acts that provide the profession of

304


the spouses as well as that of their parents. For 572 acts, it has been possible to find a match with the marriage of the father of one of the spouses. For these cases, we have the profession of the married man, of his father at the son’s marriage, of the matched father at his own marriage, and of the grandfather at the matched father marriage. The professions were grouped into three social statuses, namely low, high, and clock and watchmakers who formed an important specific corporation in the 19th century Geneva. The variable we want to predict is the status of the son at his marriage, which is clearly a categorical response, and we consider four potential predictors. The first three are status variables, namely the status of the father at son’s marriage, the status of the matched father at his own marriage, and the status of the grandfather at father’s marriage. The fourth predictor is the birthplace that can take one of the 12 values: Geneva city (GEcity); Geneva surrounding land (GEland); neighboring France (neighbF); Vaud (VD); which is a neighboring region of Geneva; Neuchatel (NE), a further French-speaking region also specialized in watch and clock making, other French-speaking Switzerland (otherFrCH), German-speaking Switzerland (GermanCH), Italian-speaking Switzerland (TI), France (F), Germany (D), Italy (I), and other. The grown tree is shown in Fig. 3. The tree-growing principle is as follows. First, all cases are grouped together in a root node (at the top of the tree) in which the distribution of the response variable, the status of the married man for our analysis, is its marginal distribution. The goal is to split this group in new nodes such that the distribution of the response variable differs as much as possible from one node to the other. The splitting is done iteratively using the categorical values of the predictor selected at each step. At the first step, we seek the predictor that best splits the root node and split the node according to the values of this predictor. The process is then repeated at each new node until a stopping rule is reached. Stopping rules typically concern the minimal node size, the maximal number of levels or the statistical significance of the improvement in the optimized criterion. In our study, we have retained the CHAID method (Kass, 1980) that selects at each step the predictor that, when it is cross tabulated with the response variable, generates the most significant independence w2 statistics.7 CHAID also seeks the aggregation level of the categories of the predictors that generates the most significant w2 and then splits indeed according to the optimally merged categories. We generated the tree of Fig. 3 with Answer Tree 3.1 (SPSS, 2001) by setting the minimal node size to 15 and requiring a maximal significance level of 5%.


Social Transition Tree with Birth Place Covariate.

305

Fig. 3.

306


Alternative methods, among which CART proposed by Breiman et al. (1984) and C4.5 due to Quinlan (1993) are among the best known, differ mainly by the criteria used for selecting the split variable at each step.8 3.2.1. Knowledge Provided by the Tree Looking at Fig. 3, we see that the first split is done according to the father status at son’s marriage. This tells us that among the four attributes considered, the status of the father is the most discriminating. The status of the married man depends, for instance, more on the father’s status than on his birthplace. The distribution inside the nodes of the first level are just the columns of the cross classification of the statuses of the father and the son. We observe here that the clock makers form a much closed group with a high probability for the son to become a clock maker when the father himself is a clock maker while this probability is much lower for the three other groups. A similar result holds for the high classes, while there are evidences about social ascension possibilities when the father belongs to the lower class. Three of the four first-level nodes are split further. The only one that is not split is that of married men whose father belongs to the clock and watch makers. This node is thus a terminal leaf, which indicates that the status of clock maker father conveys all the significant information for predicting the status of the son. This is a consequence of the strong social reproduction process inside the class of clock makers. The married men whose father was dead are split according to the grandfather’s status, which means that the grandfather’s status is more discriminating for this subgroup than the status of the father at his own marriage. There is a strong tendency for the married man to reproduce the grandfather’s status when the father is deceased. The group defined by a high status of the father as well as that defined by a low father’s status is split according to the birthplace. Both splits are binary. They do not make use, however, of the same binary partition of birthplaces. In both cases, i.e. with a father belonging to the low or high classes, the men born in neighboring France, in German-speaking Switzerland, or in Vaud have a relatively high probability to get only a low status. This is also true for men born in French-speaking Switzerland outside Geneva and Neuchatel when their father belongs to the lower class. The additional levels show that when the high position of the father results from a recent social ascension, i.e. ascension since the father’s marriage (level 3) or from the position of the grandfather (level 4), the reproduction of the father’s status by the married man is less strong. The subtree that concerns the men whose father was dead, shows effects of


307

the grandfather’s status very similar to those of the status of the father when he is alive at the marriage. 3.2.2. Goodness-of-Fit of the Descriptive Tree Classically, the quality of a tree is evaluated in terms of its classification predictive quality, which is measured by the correct classification rate of the tree. Recall that the classification is done by assigning to each case the most frequent value in its leaf. For our tree, the correct classification rate is 57.6%. This corresponds to a 42.4% error rate. At the root node, before introducing any predictor, the correct classification rate is 44.4%, which gives an error rate of 55.6%. Our tree allows thus a 24% ( ¼ (55.6-42.4)/55.6) reduction of the error rate. These figures are, nevertheless, irrelevant in our case, since we are not using the tree for classification purposes. We do not consider the classification results. The descriptive knowledge considered follows directly from the distributions inside the nodes. Hence, we consider the tree as a probability tree rather than a classification tree. In Ritschard and Zighed (2003), we have proposed indicators that better suit this descriptive point of view. We can, for instance, measure with a likelihood-ratio w2 (G2 ) the divergence between the distributions predicted by the tree (those in the leaves) and those of the finest partition that our four predictors may generate.9 We get 312.5 for 300 degrees of freedom, and its p-value is 29.8% indicating apparently a good fit. Note that though the four predictors define theoretically 576 different profiles, only the 163 actually observed are taken into account. When these profiles are cross tabulated with the three statuses, we get 489 cells. For 572 data, this gives an average of a bit more than one per cell, which is insufficient to ensure the w2 distribution of G2. Hence, we should not attach here too much confidence to the p-value (Table 6). For comparison purposes, Table 6 reports the G2 statistic for a set of nested trees, namely the independence tree corresponding to the root node only, the tree expanded respectively one level only, two levels and three levels, the fitted tree and the saturated tree that generates the finest partition. Beside G2, its degrees of freedom and significance level, the table shows the BIC and AIC information criteria and the adjusted pseudo R2. The latter measures the percentage of reduction of the G 2 =df ratio as compared with the independence tree. The BIC and the AIC are G2’s penalized for the complexity. We see that with less than three levels there is a lack of fit, the divergence with the finest partition being significant at the 5% level. The difference in G2s between two nested trees can also be compared with a w2 distribution with degrees of freedom being the difference in these degrees for the two

308


Table 6. Tree Independent Level 1 Level 2 Level 3 Fitted Saturated

Goodness-of-Fit of the Tree and Subtrees.

G2

Df

sig

BIC

AIC

Pseudo R2

482.3 408.2 356.0 327.6 312.5 0

324 318 310 304 300 0

0.000 0.000 0.037 0.168 0.298 1

2319.6 1493.9 1492.5 1502.2 1512.5 3104.7

812.3 750.2 714.0 697.6 690.5 978.0

0 0.14 0.23 0.28 0.30 1

models. Thus, the Level 3 tree differs by DG 2 ¼ 15:1 and Ddf ¼ 4 from the fitted model, which is clearly significant. Hence, the two splits leading to Level 4 look jointly statistically significant. From the BIC point of view, the Level 2 tree provides the best compromise between fit and complexity. Level 3 or 4 trees seem, however, preferable according to the interesting insight brought by the additional levels and the significant divergence of Level 2 with the saturated model. The AIC, which is known, however, to underestimate the impact of complexity, selects here the fitted tree. Trees look really promising thanks mainly to their ease of use and to their visual outcome. When it comes to interpretation, one should be aware, nevertheless, that trees may be instable in the sense that small changes in the data could alter the structure especially splits and variables selected at higher levels. It is then important to avoid growing too complex trees. Relaying on BIC or AIC criteria should help determining a somewhat robust tree. Splits behind the optimal BIC or AIC will be less reliable and their interpretation then requires more caution.

4. CONCLUSION This paper stressed the scope and limits of various methods available for analyzing life course data globally, and especially in demography. Demographers and historical demographers invented their own longitudinalanalytical tools like the life tables or the family reconstitution, almost since the birth of their discipline. However, everywhere but probably more in the French-speaking areas, those sciences of the masses hesitated to take a step further, while they were so close from the life course perspective and methods. Many academics are still living this transitiony. For adepts of highly quantitative social sciences, we wanted to both introduce and illustrate


309

promising methodological perspectives without hiding the complexity of the new approaches. At the same time, we did not elaborate this contribution only for our disciplinary fellows, since one of the most important evolutions is that the analytical techniques obviously lend us to neighboring disciplines that share the same tools and explore similar concepts, giving to the interdisciplinary ambition a growing substance. We have chosen to illustrate different approaches, and especially the emerging data-mining techniques that should be able to provide original additional insights on results provided by more classical statistical methods. The discussion, however, is by no means exhaustive. Among the techniques we did not discuss, optimal matching (Abbot & Forrest, 1986; Malo & Munoz, 2003) deserves special attention. Optimal matching is, like Markovian models, a state-sequence analysis tool. It is merely a data-mining approach, since it proceeds heuristically. Unlike the mining of frequent sequences that does not care about the similarity between sequences, optimal matching is concerned with the discovering of similarities between sequence patterns. Optimal matching evaluates the proximity between two sequences by seeking the minimal number of changes that can transform a sequence a into a sequence b. Survival (Ciampi et al., 1988; Segal, 1988) and risk trees (Leblanc & Crowley, 1992) developed in the field of biomedicine during the first half of the 1990s would also merit further attention from historians and demographers. It is worth mentioning that the statistical and data-mining approaches are not substitutes for one another. They are complementary, each method bringing its own insight. The choice of a method will be dictated by the kind of data available: spell durations, event sequences, state sequences, and indeed the type of results expected: knowledge about probability of transitions, effects on these risks, characteristic trajectories, or life sequences. Another important element for this choice, at least for the end user, is the availability of user-friendly softwares and the level of expertise required to run the method and interpret the results. Many softwares propose duration or hazard models and/or classification trees. It is less obvious to find friendly tools for Markovian models and the mining of sequential rules. March 2 is a promising solution for Markovian models, while specialized softwares like Clementine propose sequence mining tools (see http://www.kdnuggets.com for a list of commercial and free data mining softwares). The use and interpretation of hazard models is very similar to that of other regression-like models, which renders them attractive. The interpretation of induction trees is also very straightforward and looks therefore as a promising tool. Nevertheless, the fine tuning of trees, which may be highly instable, requires

310


generally more care than hazard models. Mining frequent sequential patterns also requires some experience to get interesting patterns. In any case, the new highlights provided by these data-mining approaches are worth the effort.

NOTES 1. Random effect models are also known as multilevel, hierarchical or mixedeffect models. 2. GLM models (McCullagh & Nelder, 1989) cover a large number of parametric models. They assume a distribution of the natural exponential family for the dependent variable and are, in their simpler form, simply characterized by a link function that describes how the mean of the dependent variable is linked to the linear form of the explanatory variables. For example, we get the classical linear model with a Gaussian distribution and the identity link, the logistic model with a Bernoulli distribution and the logit link, and the log-linear model with a Poisson distribution and the log link. 3. The estimations were obtained with S-plus 6.2. We suspect a bug in Stata 8 that was not able to converge within 24 h for the frailty model while S-Plus provided the results within 2 min. Formally, the estimated hazard model is h(t, x1,y, xp) ¼ ng h0(t) exp(b1x1+ y+bpxp), where h0(t) is the baseline hazard function and ng the shared frailty term. We estimated this model assuming a gamma g(1/y,1/y) distribution for the frailty term ng, for which we have E(ng) ¼ 1 and Var(ng) ¼ y. 4. http://popindex.princeton.edu/ 5. The deviance -2LogLik may be seen as the distance between the predictions generated by the model and the observed counts. Hence it is a measure of global fit. However, it cannot be used here to test the fit since we do not know its distribution. 6. The AIC and BIC criteria are penalized forms of the –2LogLik that take account of the complexity, i.e. the number of estimated parameters. Among the two, the BIC is usually preferred since the AIC is known to insufficiently penalize complexity. The model with the minimal BIC offers the best compromise between fit and complexity. 7. Significance is generally evaluated with a Bonferroni correction for taking account of the multiple test sequence that controls each split decision. 8. CART maximizes the reduction in the Gini index also known as the quadratic entropy. It generates only successive binary splits. C4.5 uses the gain ratio defined as the reduction in Shannon’s entropy normalized by the entropy of the distribution among the classes of the generated partition. Unlike the CHAID method, for which the significance of the w2 provides a natural validation for the split, CART and C4.5 do not have such a natural split validation criteria. These methods complete therefore the growing process with a post pruning round that, starting from the leaves, eliminates unreliable splits. Only splits that improve the predictive error rate are retained. There are also graph induction tools like SIPINA (Zighed & Rakotomalala, 1996), which generalize trees by allowing the merge of nodes with similar inside distribution.


311

9. G2 is indeed the deviance -2LogLik. It measures how far the counts predicted by the tree are from those observed for the finest possible partition. When the predicted counts are not too small, it has an approximate w2 distribution and can be used for testing the goodness-of-fit. Note that the w2 reported in Table 5 would correspond here to the difference between the G2 of the tree and that of the root node (independence). We expect w2 to be large while G2 should be small.

REFERENCES Abbot, A., & Forrest, J. (1986). Optimal matching methods for historical sequences. Journal of Interdisciplinary History, 16, 471–494. Agrawal, R., & Srikant, R. (1994). Fast algorithm for mining association rules in large databases, In: Proceedings of the International Conference on Very Large Data Base (VLDB’94), Santiago de Chile (pp. 487–499). San Mateo: Morgan-Kaufman. Agrawal, R., & Srikant, R. (1995). Mining sequential patterns, In: Proceedings of the International Conference on Data Engineering (ICDE), Taipei, Taiwan (pp. 487–499). Taiwan: IEEE Computer Society. Allison, P. D. (1982). Discrete-time methods for the analysis of event histories. In: S. Lienhardt (Ed.), Sociological methodology (pp. 61–98). San Francisco: Jossey-Bass Publishers. Alter, G. (1998). L’event history analysis en de´mographie historique: Difficulte´s et perspectives. Annales de de´mographie historique, 2, 23–35. Alter, G., & Oris, M. (2000). Mortality and economic stress: Individual and household responses in a nineteenth-century Belgian village. In: T. Bengtsson & O. Saito (Eds), Population and economy: From hunger to modern economic growth (pp. 335–370). Oxford: Oxford University Press. Alter, G., Oris, M., & Brostro¨m, G. (2001). The family and mortality: A case study from rural Belgium. Annales de de´mographie historique, 1, 11–31. Aranza-Ordaz, F. J. (1983). An extension of the proportional hazards model for grouped data. Biometrics, 39, 109–117. Barber, J. S., Murphy, S. A., Axinn, W. G., & Maples, J. (2000). Discrete-time multilevel hazard analysis. In: M. E. Sobel & M. P. Becker (Eds), Sociological methodology, (pp. 201–235). New York: The American Sociological Association. Beekink, E., van Poppel, F., & Liefbroer, A. C. (1999). Surviving the loss of the parent in a nineteenth-century Dutch provincial town. Journal of Social History, 32, 614–670. Berchtold, A. (1999). The double chain Markov model. Communications in Statistics: Theory and Methods, 28(11), 2569–2589. Berchtold, A. (2001). Estimation in the mixture transition distribution model. Journal of Time Series Analysis, 22(4), 379–397. Berchtold, A. (2002). High-order extensions of the double chain Markov model. Stochastic Models, 18(2), 193–227. Berchtold, A., & Berchtold, A. (2004). MARCH 2.01: Markovian model computation and analysis. User’s guide, www.andreberchtold.com/march.html. Berchtold, A., & Raftery, A. E. (2002). The mixture transition distribution model for high-order Markov chains and non-Gaussian time series. Statistical Science, 17(3), 328–356. Berquo, E., & Xenos, P. (1992). Editor’s introduction. In: E. Berquo & P. Xenos (Eds), Family systems and cultural change (pp. 8–12). Oxford: Clarendon Press.

312


Billari, F. C., Fu¨rnkranz, J., & Prskawetz, A. (2000). Timing, sequencing, and quantum of life course events: A machine learning approach. Working Paper no. 010, Max-Plank-Institute for Demographic Research, Rostock. Billari, F. C. (2005). Life course analysis. Two cultures? Some reflections with examples from the analysis of the transition to adulthood. In: R. Levy, P. Ghisletta, J.-M, Le Goff, D. Spini & E. Widmer (Eds), Towards an interdisciplinary perspective on the life course (pp. 267–286), Advances in Life Course Research, Vol. 10. Amsterdam: Elsevier. Blockeel, H., Dehaspe, L., Ramon, J., & Struyf, J. (2004). The ACE data mining system. User’s manual. Katholieke Universiteit Leuven, Leuven. Blockeel, H., Fu¨rnkranz, J., Prskawetz, A., & Billari, F. (2001). Detecting temporal change in event sequences: An application to demographic data. In: L. D. Raedt & A. Siebes (Eds), Principles of data mining and knowledge discovery: 5th European conference, PKDD 2001 (Vol. 2168 of LNCS, pp. 29–41). Freiburg in Brisgau: Springer. Blossfeld, H.-P., Hamerle, A., & Mayer, K. U. (1989). Event history analysis, statistical theory and application in the social sciences. Hillsdale, NJ: Lawrence Erlbaum. Blossfeld, H.-P., & Rohwer, G. (2002). Techniques of event history modeling, new approaches to causal analysis (2nd ed.). Mahwah, NJ: Lawrence Erlbaum. Bocquier, P. (1996). L’analyse des enqueˆtes biographiques a` l’aide du logiciel STATA. Paris: Centre franc- ais sur la population et le de´veloppement. Bourdieu, P. (1986). L’illusion biographique. Actes de la Recherche en Sciences sociales, 62–63, 69–72. Breiman, L. (2001). Satistical modeling: The two cultures (with discussion). Statistical Science, 16(3), 199–231. Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and regression trees. New York: Chapman & Hall. Bryk, A., Raudenbush, S. W., & Congdon, R. (1996). HLM: Hierarchical linear and nonlinear modeling with the HLM/2 l and HLM/3 l programs. Chicago: Scientific Software International. Caselli, G., Vallin, J., & Wunsch, G. (Eds), (2001). De´mographie: analyse et synthe`se, vol. I. La dynamique des populations. Paris: Institut national d’e´tudes de´mographiques. Ciampi, A., Hogg, S. A., McKinney, S., & Thiffault, J. (1988). RECPAM: A computer program for recursive partitioning and amalgamation for censored survival data and other situations frequently occurring in biostatistics I. Methods and program features. Computer Methods and Programs in Biomedicine, 26(3), 239–256. Courgeau, D., & Lelie`vre, E. (1993). Event history analysis in demography. Oxford: Clarendon Press. Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society, Series B, 34(2), 187–220. Demeny, P., & McNicoll, G. (Eds) (2003). Encyclopedia of population (Vol. 2). New York: McMillan. De Rose, A., & Pallara, A. (1997). Survival trees: An alternative non-parametric multivariate technique for life history analysis. European Journal of Population, 13, 223–241. Dupaˆquier, J., & Dupaˆquier, M. (1985). Histoire de la de´mographie. Paris: Perrin. Dykstra, P., & van Wissen, L. J. G. (1999). Introduction: The life course approach as an interdisciplinary framework for population studies. In: L. J. G. van Wissen & P. Dykstra (Eds), Population issues: An interdisciplinary focus (pp. 1–22). New York: Plenum Press.


313

Goldstein, H., Rasbash, J., Plewis, I., Draper, D., Browne, W., Yang, M., Woodhouse, G., & Haely, M. (1998). A user guide to MLwiN. Technical Report, Multilevels Models Project, London. Greenhalgh, S. (1996). The social construction of population science: An intellectual, institutional and political history of twentieth-century demography. Comparative Studies in Society and History, 38(1), 26–66. Hagestad, G. O. (2003). Interdependent lives and relationships in changing times: A life course view of families and aging. In: R. A. H. Settersten (Ed.), Toward new understanding of later life (pp. 135–160). Amytiville, New York: Baywood Publishing. Hand, D. J., Mannila, H., & Smyth, P. (2001). Principles of data mining (adaptive computation and machine learning). Cambridge, MA: MIT Press. Han, J., & Kamber, M. (2001). Data mining: Concept and techniques. San Francisco: Morgan Kaufmann. Hogdson, D. (1988). Demography as social science and policy science. Population and Development Review, 9(1), 541–569. Hogdson, D. (1991). The ideological origins of the population association of America. Population and Development Review, 17(1), 1–34. Ho¨hn, C. (1992). The IUSSP programme in family demography. In: E. Berquo & P. Xenos (Eds), Family systems and cultural change, (pp. 3–7). Oxford: Clarendon Press. Holford, T. R. (1980). The analysis of rates and survivorship using log-linear models. Biometrics, 65, 159–165. Kass, G. V. (1980). An exploratory technique for investigating large quantities of categorical data. Applied Statistics, 29(2), 119–127. Kish, L., & Frankel, M. R. (1974). Inference from complex samples (with discussion). Journal of the Royal Statistical Society, Series B, 36, 1–37. Landwehr, N., Hall, M., & Frank, E. (2003). Logistic model trees. In: N. Lavrac, D. Gamberger, L. Todorovski & H. Blockeel (Eds), Machine learning: ECML 2003 (Vol. 2837 of LNAI, pp. 241–252). Berlin: Springer. Leblanc, M., & Crowley, J. (1992). Relative risk trees for censored survival data. Biometrics, 48, 411–425. Lillard, L. A. (1993). Simultaneous equations for hazards: Marriage duration and fertility timing. Journal of Econometrics, 56, 189–217. Lin, D. Y., & Wei, L. J. (1989). The robust inference for the Cox proportional hazards model. Journal of the American Statistical Association, 84, 1074–1078. Lynch, K. A. (1998). Old and new research in historical patterns of social mobility. Historical Methods, 31(3), 93–98. MacDonald, I. L., & Zucchini, W. (1997). Hidden Markov and other models for discrete-valued time series. London: Chapman & Hall. Malerba, D., Appice, A., Ceci, M., & Monopoli, M. (2002). Trading-off local versus global effects of regression nodes in model trees. In: M.-S. Hacid, Z. W. Ras, D. A. Zighed & Y. Kodratoff (Eds), Foundations of intelligent systems, ISMIS 2002 (Vol. 2366 of LNAI, pp. 393–402). Berlin: Springer. Malo, M. A., & Munoz, F. (2003). Employment status mobility from a lifecycle perspective: A sequence analysis of work-histories in the BHPS. Demographic Research, 9(7), 471–494. Mannila, H., Toivonen, H., & Verkamo, A. I. (1994). Efficient algorithms for discovering association rules. In: Proceedings of the AAAI’94 workshop knowledge discovery in databases (KDD’94), Seattle, WA (pp. 181–192).

314


Mannila, H., Toivonen, H., & Verkamo, A. I. (1997). Discovery of frequent episodes in event sequences. Data Mining and Knowledge Discovery, 1(3), 259–289. McCullagh, P., & Nelder, J. A. (1989). Generalized linear models. London: Chapman & Hall. Oris, M., & Poulain, M. (2003). Entre blocages spatiaux et nouvelles dynamiques familiales: Quitter ses parents. In: Populations et de´fis urbains. Chaire Quetelet 1989 (pp. 313–337). Louvain-la-Neuve: Academia-Bruylant/L’Harmattan. Oris, M., Ritschard, G., & Berchtold, A. (2003). The use of Markov process and induction trees for the study of intergenerational social mobility in nineteenth century Geneva. In: Social science history association annual meeting, Baltimore. Paliwal, K. K. (1993). Use of temporal correlation between successive frames in a hidden Markov model based speech recognizer. Proceedings of the ICASSP, 2, 215–218. Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1, 81–106. Quinlan, J. R. (1993). C4.5: Programs for machine learning. San Mateo: Morgan Kaufmann. Rabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech recognition. Proceedings of the IEEE, 77(2), 257–286. Raftery, A. E., & Tavare´, S. (1994). Estimation and modelling repeated patterns in high order Markov chains with the mixture transition distribution model. Applied Statistics, 43, 179–199. Ritschard, G., & Zighed, D. A. (2003). Goodness-of-fit measures for induction trees. In: N. Zhong, Z. Ras, S. Tsumo & E. Suzuki (Eds), Foundations of intelligent systems, ISMIS03 (Vol. 2871 of LNAI, pp. 57–64). Berlin: Springer. Rohwer, G., & Po¨tter, U. (2002). TDA user’s manual. Software, Ruhr-Universita¨t Bochum, Fakulta¨t fu¨r Sozialwissenschaften, Bochum. Ryczkowska, G., & Ritschard, G. (2004). Mobilite´s sociales et spatiales: Parcours intergeńe´rationnels d’apre`s les mariages genevois, 1830–1880. In: Proceedings of the Fifth European social science history conference, Berlin. Segal, M. R. (1988). Regression trees for censored data. Biometrics, 44, 35–47. SPSS (Ed.). (2001). Answer tree 3.0 user’s guide. Chicago: SPSS Inc. Szreter, S. (1993). The idea of demographic transition: A critical intellectual history. Population and Development Review, 19, 659–701. Therneau, T. M., & Grambsch, P. M. (2000). Modeling survival data. New York: Springer. Vallin, J. (2001). Populations et individus. In: G. Caselli, J. Vallin & G. Wunsch (Eds), De´mographie: Analyse et synthe`se, La dynamique des populations (Vol. I, pp. 9–12). Paris: Institut national d’e´tudes de´mographiques. Van de Kaa, D. J. (1996). Anchored narratives: The story and findings of half a century of research into the determinants of fertility. Population Studies, 50(3), 389–432. Van Imhoff, E., Kuijsten, A., Hooimeijer, P., & van Wissen, L. J. G. (1995). Household demography and household modeling. New York: Plenum Press. Vaupel, J. W., Manton, K. G., & Stallard, E. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16, 439–454. Ve´ron, J. (1993). Arithme´tique de l’homme: la de´mographie entre science et politique. Paris: Editions du Seuil. Willekens, F. J. (1999). The life course: Models and analysis. In: L. J. G. van Wissen & P. Dykstra (Eds), Population issues: An interdisciplinary focus (pp. 23–51). New York: Plenum Press. Yamaguchi, K. (1991). Event history analysis. ASRM 28, Newbury Park, London: Sage. Zighed, D. A., & Rakotomalala, R. (1996). SIPINA-W(c) for Windows. User’s guide, Laboratory ERIC – University of Lyon 2, Lyon.

FIVE STEPS IN LATENT CURVE MODELING WITH LONGITUDINAL LIFE-SPAN DATA John J. McArdle INTRODUCTION During the past few decades, scientists from many disciplines have been engaged in the study of human development over what is termed the ‘‘lifespan.’’ Of course, this term has early roots in the physical and biological sciences. Life-span developmental research in psychology continues to promote the combination of (1) a paradigm shift in thinking about human development as a process occurring over all ages in different ways, and (2) the utility of a multivariate perspective on data collection and analysis (e.g., Baltes, Lindenberger, & Staudinger, 1998). The theoretical underpinnings of life-span developmental psychology are outlined by Abeles (1987, p. 3): Development is (1) a life-long process, (2) multi-dimensional, (3) multi-directional, and (4) multi-determined. These kind of meta-theoretical principles have become so embedded in the fiber of modern developmental research that they are now hardly ever debated or even discussed. Instead, the current issues focus on the best ways to formalize these ideas into practical multivariable research. The life-span research presented in this paper attempts to add to this practice using contemporary statistical modeling analyses.


315

316

JOHN J. MCARDLE

In some respects, the life-span view of developmental psychology is relatively new (for overview, see Baltes & Nesselroade, 1979). This area emerged as a separate discipline in the early to mid-1970s with contributions of a few key researchers. The Department of Psychology at West Virginia University was a fertile source of much important work – at one time, Warner Schaie was the head of this department, and Paul Baltes and John Nesselroade were working together in it, first as assistant professors, and then as associate professors. About a decade earlier, Schaie began his longitudinal research by following the people from Paul Horst’s multivariate study at the University of Washington (later termed the ‘‘Seattle Longitudinal Study’’). Baltes, then a promising student of Ernst Boesch (Saarbruecken), a Piaget Ph.D., was well-versed in the elegant theoretical work of Klaus Riegel, the founder of modern dialectic psychology. Nesselroade was then a promising student of Raymond Cattell, the founder of modern multivariate psychology and a major contributor to the fields of cognition and personality. This group stimulated debate and controversy in research on the separation of Age (A) effects, Cohort (C) effects, and Time-Period (T) effects. Schaie (1965) promoted a formal separation of ACT effects in and ANOVA fashion using what he termed the ‘‘general developmental model.’’ Perhaps surprisingly, Baltes was one of the first critics of the ACT approach (Baltes, 1968) – he suggested not to use Schaie’s model for explanatory purposes but descriptive ones only. In this spirit, Baltes shifted the emphasis to analyses of cohort-sequential and longitudinal sequences (see Schaie & Baltes, 1975). Soon afterwards, Schaie’s work was debated vigorously by Cattell and his students: e.g., ‘‘Rarely has such a Tower of Babel sprung up so quickly as in this area of research.’’ (Cattell, 1970, p. 154; also Horn & Donaldson, 1976). Of course, Cattell himself had investigated complex multivariate designs before (e.g., Cattell, 1963, 1969), but it was Nesselroade who advocated the inclusion of multivariate measurements, and he was soon active in the debate as well (e.g., Baltes & Nesselroade, 1970). As it turned out, these controversies on ACT methods were going on in other areas of research, especially Sociology, and stimulated many others and led to some of the most important contributions to life-span research. There were immediate practical applications such as the ‘‘longitudinal and cross-sectional sequences’’ in research by Baltes and Nesselroade (1972) on West Virginia Junior High and High School students. These activities also stimulated an important series of books on ‘‘life-span developmental psychology’’ focusing on the integration of substance and method (e.g., Goulet & Baltes, 1970; Baltes, Reese, & Nesselroade, 1977; Baltes &

Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data

317

Nesselroade, 1979; Baltes, 1983). A large number of psychologists soon followed the active lead of this small group from Morgantown, West Virginia (afterwards at Penn State, now at the University of Virginia), and the result became known as the ‘‘life-span movement in psychology’’ (see Nesselroade, 2001). The ACT controversy that stimulated this life-span research could be considered just a natural extension of previous longitudinal considerations (e.g., Wohlwill, 1973; Bell, 1954; McArdle & Bell, 2000). Also, subsequent statistical studies proved that the general ACT problem was ‘‘not identified’’ and unique solutions were only possible using highly restricted models (for review, see Horn & McArdle, 1980; McArdle & Anderson, 1990; Donaldson & Horn, 1992). But, in retrospect, the survival and popularity of the lifespan movement can now be seen due in large part to a successful integration of developmental theory and multivariate method. Clearly, the substantive thinking was enhanced by the methodological thinking, and vice versa. Most of these methodological debates concluded that the collection of longitudinal data are a necessary ingredient for life-span research, and many researchers have defined these issues in extensive detail, but most have emphasized ‘‘the explanation of inter-individual differences (or similarities) in intra-individual change patterns’’ (e.g., Wohlwill, 1973; Baltes & Nesselroade, 1979). During the last decade, many methodologists have added to the current knowledge base, and classic models for ‘‘growth-curve analysis’’ seems to have been revived as an important research technique (e.g., see Rogosa & Willett, 1985; McArdle & Epstein, 1987). In a general sense, the term growth-curve analysis denotes the processes of describing, testing hypotheses, and making scientific inferences about the growth and change patterns in a wide range of time-related phenomena. More specifically, growth curves are longitudinal data with special properties of repeated measures. Thus, in both theory and practice, the term ‘‘growth curve’’ is not limited to the phases of life, where the organism ‘‘grows,’’ but it can also be used to describe and analyze phases of life, where the organism ‘‘declines.’’ The phenomena that are studied here both ‘‘grow and decline over age,’’ and these comprehensive features lead to an unusual set of opportunities for developmental analyses of data. This paper describes a current application of one class of longitudinal growth-curve analysis models – latent curve modeling (LCM) using structural equation modeling (SEM) techniques. The analytic techniques of this chapter are presented in the next five sections. These are presented in a sequence of difficulty, as ‘‘steps’’ of developmental data analysis. Step 1: Describing the Observed and Unobserved Longitudinal Data. I consider some

318

JOHN J. MCARDLE

useful ways to summarize longitudinal data, including statistical information from both the complete and incomplete cases. Step 2: Characterizing the Developmental Shape of Both Individual and Groups. I try to describe both the group and individual characteristics of the longitudinal data. We demonstrate how the SEM approach is generally easy and flexible. Step 3: Examining the Predictors of Individual and Group Differences in Developmental Shapes. We recognize that individual differences in growth may be the result of combinations of other measured variables. I describe how SEM can be used in a ‘‘multi-level’’ form, can be extended to include concepts from latent path analysis, and can provide empirical evidence for hypotheses about the precursors or correlates of individual longitudinal patterns. Step 4: Examining Group Differences in Developmental Shapes. Individual differences in growth may be the result of combinations of different patterns of change for persons in different groups. To study these possibilities, I describe how SEM can be used in a ‘‘multi-group’’ form, can be extended to include concepts from ‘‘latent mixture’’ models, and can provide empirical evidence for hypotheses about the heterogeneity in longitudinal growth patterns. Step 5: Studying Dynamic Determinants Among Variables Over Time. I show how the time-dependent nature of the latent variables can be represented in SEM and used to study ‘‘lead–lag’’ relations using simple dynamic expressions. As an illustration within all five steps, I present alternative analyses of a longitudinal data collection on intellectual abilities over the life span from the Bradway–McArdle Longitudinal Study (McArdle & Hamagami, 1996, 2004; McArdle, Hamagami, Meredith, & Bradway, 2001). In Table 1, I outline the multiple phases of data collection on intellectual abilities measured over a large part of the life span – at seven occasions between ages 2 and 72. Some of the data collected appear in the plots of Fig. 1 and will be described in the next section. Analyses of these lifespan data have been reported in more detail in other recent papers, so here I mainly summarize these results and highlight the life-span features of prior analyses. These illustrations are used to convey the main presumptions and techniques as well as the benefits and limitations involved in using this approach in developmental research. Mathematical and statistical issues are not presented in detail, and formal equations are not presented. This is an overview of the general data analysis methodology to demonstrate the flexibility of these methods for life-span developmental research.

319


Table 1.

Overview of the Bradway–McArdle Longitudinal Data at Seven Time Points (N ¼ 111).

Date

Sample Size

Age Range

Measures

Investigators

1931 1941 1956 1969 1984 1992 1998

212 138 111 48 54 51 32

2–7 12–17 27–32 40–45 55–60 61–66 67–72

SB SB+ SB, WAIS+ SB, WAIS+ WAIS, WJ-R+ WAIS, WJ-R+ WAIS, WJ-R+

RT+, QM KB KB JK+, KB JM+, KB JM+, KB JM+, KB

Notes: 1. Measurement batteries are the Stanford–Binet (SB), Wechsler Adult Intelligence Scale (WAIS), and Woodcock–Johnson Revised (WJ-R). (+) indicates that additional demographic and/or psychometric measurements were included. 2. Study Investigators were Robert L. Thorndike (RT), Quinn McNemar (QM), Katherine P. Bradway (KB), Jon Kangas (JK), and John J. McArdle (JM).

Fig. 1.

A Plot of Individual Growth Curves for the Bradway–McArdle Longitudinal Data (N ¼ 111).

320

JOHN J. MCARDLE

STEP 1: DESCRIBING THE OBSERVED AND UNOBSERVED LONGITUDINAL DATA The first step in any useful data analysis is an adequate description of the data. The collection of longitudinal data can be extremely difficult, especially over large portions of the life-span, so the unique aspects of these data should be emphasized.

The Bradway–McArdle Longitudinal Study The data examined here come from persons, who were first measured in 1931 when they were aged 2–7 as part of the larger standardization sample of the Stanford–Binet test (N ¼ 212). Many of these persons who could be found (N ¼ 138) were measured again about 10 years later by Katherine P. Bradway as part of her doctoral dissertation in 1944. Many of these same persons were measured twice more by Bradway as adults at average ages of 30 and 42 using the Wechsler Adult Intelligence Scales (WAIS) (N ¼ 111; for further details, see Bradway & Thompson, 1962; Kangas & Bradway, 1971). In 1984 and 1992, at average ages 57 and 65, the current author, working with Bradway, measured as many of these same subjects as we could locate (see McArdle & Hamagami, 1996; McArdle et al., 2001). About half (N ¼ 55) of the adolescents tested in 1944 were measured again in 1984 at ages 55–57 and in 1992 at ages ranging from 64 to 72 (McArdle et al., 2001). The seventh testing of the same individuals in 1998 yielded retests on most of the same people (N ¼ 32), and attrition from the study was complicated by the advancing age of the persons (e.g., attrition due to fatigue, illness, and death). Although many other researchers were involved in data collection (see Table 1), we now term this whole collection the Bradway– McArdle Longitudinal study. Up to now, our published analyses have focused only the first six time points of longitudinal data. Fig. 1 is a display of individual growth curve data for this measure at each age-at-testing for N ¼ 111 individuals. The y-axis indexes the scores on ‘‘Verbal-Knowledge’’ and the x-axis is an index of the age-at-testing. The connected lines in this figure are graphic descriptions of the pattern of Verbal-Knowledge scores for each of the individuals, so each line is termed a growth curves or trajectory. This kind of plot allows us to see some overall trends of rapid rises in early childhood and adolescence to long periods of consistency at older ages. There are also a hint of


321

patterns of growth and change for some individuals, and this interesting possibility requires a rigorous examination. The analysis of growth and change in similar constructs spanning childhood to adulthood still represents a major challenge for research over the life span. It is easy to connect the data for a person over many time points of testing, but it is much harder to insure that these scores meet the measurement requirements of ‘‘exactly repeated measures.’’ Growth-curve analyses typically require: (a) the same entities are repeatedly observed, (b) the same procedures of measurement and scaling of observations are used, and (c) the timing of the observations is known. In prior psychometric research, we have used both scale-level factor analysis and Item Response Theory (IRT) models to create comparable measurements over time. Composite scores from the early Stanford–Binet tests (at ages of 4, 14, 30, and 42), as well as the WAIS tests (at average ages 30, 42, 56, and 64) were used to form a Verbal-Knowledge (or Crystallized Intelligence gc) score. These scores are in a comparable metric, because they were based on a Rasch-scaling of selected items in each test (see Hamagami, 1998; McArdle & Nesselroade, 2003; McArdle et al., 2004).

Describing the Observed Data The sample sizes, means, standard deviations, and correlations of these IRT-scaled measures over six occasions are listed in Table 2. The overall subject participation shows a nearly continual loss of participants over the 60 years. This is not an unusual loss of participants, even for a much smaller period of time. The means and standard deviations show the simple pattern described earlier, with rapid rises in childhood and adolescence together with very little growth or decline in adulthood. The correlations over time, the unique statistical information of the longitudinal data, present a complex pattern of results, some correlations suggesting high stability of individual differences (e.g., r40:9) and others suggesting low long-term stability (ro0.1). Fig. 2 is a scatter-plot matrix of the raw data at all occasions. In this figure, the frequency histograms are placed on the main diagonal and the pair-wise scatter plots are placed on the off-diagonals. These scatter plots of growth data provide another way of describing the relative mobility of cognition in these people over time. The spread of points is rather narrow when the time interval is shortest (i.e., at consecutive measurements), and the spread of points is much wider when the observations are most further

322

JOHN J. MCARDLE

Table 2. Observed and Unobserved Summary Statistics based on the Longitudinal Data at Six Time Points (N ¼ 111; MLE-MAR in brackets; Step 1, see Figs. 1 and 2). (a) Observed means, standard deviations, and ranges

Age Age Age Age Age Age

04 14 29 42 57 65

N

Mean

[MLE]

SD

[MLE]

Min

Max

110 111 110 49 51 51

22.1 72.2 84.4 87.9 88.9 88.1

[22.2] [72.2] [84.4] [87.2] [87.7] [87.1]

9.8 9.3 7.0 6.7 5.8 7.1

[9.6] [9.3] [6.9] [6.8] [7.0] [7.1]

0.0 40.6 58.2 69.2 73.8 68.7

37.6 90.2 95.7 100.0 99.3 98.6

(b)Observed and unobserved correlations (each entry includes pairwise r and [MLE-MAR r]) Age 04 Age 04

1.000

Age 14

0.553 [0.558] 0.233 [0.224] 0.194 [0.143] 0.304 [0.211] 0.079 [0.171]

Age 29 Age 42 Age 57 Age 65

Age 14

Age 29

Age 42

Age 57

Age 65

1.000 0.680 [0.679] 0.377 [0.412] 0.489 [0.574] 0.472 [0.579]

1.000 0.812 [0.787] 0.800 [0.858] 0.759 [0.791]

1.000 0.798 [0.843] 0.663 [0.744]

1.000 0.911 [0.946]

1.000

(c) Patterns of complete (x) and incomplete (o) data

Age 04 Age 14 Age 29 Age 42 Age 57 Age 65 Frequency

1

2

3

4

5

6

7

8

9

x x x o o o 36

x x x x x x 29

x x x o o x 12

x x x x o o 9

x x x o x o 8

x x x x x o 7

x x x o x x 6

x x x x o x 3

x x o x x x 1

apart in time (i.e., from childhood at age 4 to adulthood at ages 42–65). The correlation matrix is a simple summary of these features, but the plots are uniquely informative about the patterns of the people.

323

GC_AGE65

GC_AGE57

GC_AGE42

GC_AGE29

GC_AGE14

GC_AGE04


GC_AGE04

Fig. 2.

GC_AGE14

GC_AGE29

GC_AGE42

GC_AGE57

GC_AGE65

Relationships among the Scores of the Six ‘‘General Knowledge’’ (Gc) Variables Over Time.

Results from Dealing with Incomplete Information The summary information presented in Table 2 is not limited to only those participants with complete data at all six time points of measurement. As is usual, we can represent all available information about the observed means and variances and the ‘‘pair-wise’’ correlations. To deal with this problem, I present a description of the patterns of complete and incomplete data in Table 2c. In these data, there are only nine different patterns of incomplete data (out of a possible set of 70 patterns), and most of the persons are either measured at all six times (n ¼ 29) or only the first three times (n ¼ 36). These incomplete data patterns can be represented as the ‘‘percentage of data’’ or ‘‘coverage’’ for each covariance of these scores – in some cases 100% of the participants are used but in some cases only 29.7% of the information is available (at ages 42 and 65). In Tables 2a and b, I also use brackets to list an ‘‘incomplete data’’ estimate of the sample means, standard deviations, and correlations. These estimates are based on what is termed maximum likelihood estimation under

324

JOHN J. MCARDLE

missing at random assumptions (MLE-MAR; Little, 1995; McArdle, 1994; Cnaan, Laird, & Slasor, 1997). This approach allows us to examine these summary statistics ‘‘as if all persons were measured at all occasions.’’ These newly estimated statistics are fairly close to the pair-wise estimates and this indicates these data meet the minimal conditions of ‘‘missing at random’’ (MAR; Little, 1995). Most importantly, these estimated statistics do not suffer from some common statistical problems (local linear dependency), and we can generally use all available information from every person – that is, we do not need to select a subset of persons with complete data (n ¼ 29) out of all those measured (N ¼ 111). Dealing with incomplete data as unobserved but important scores can alter inferences about growth and change. For example, from Table 2b, the pairwise correlation of General Knowledge from age 4 to 42 is only r ¼ 0:19; and from age 4 to 65 only r ¼ 0:08; and these are certainly not as high as comparable correlations recently reported elsewhere (e.g., r47 in Deary, Whalley, Lemmon, Crawford, & Starr, 2000). It would be reasonable to critique our lower over-time correlations as being underestimated due to the loss of participants (i.e., selective attrition). This concern leads us to try to account for the patterns of incomplete data. It is interesting that when we do use contemporary technique to account for attrition, we obtain similar results – the MLE-MAR-estimated correlation of General Knowledge from age 4 to 42 is only r ¼ 0:14; and from age 4 to 65 only r ¼ 0:17 (from Table 2b). This new result implies these low correlations over time may be due to initial selection but are probably not attributable to subsequent attrition. This description of unobserved statistics is an essential part of all contemporary growth models.

STEP 2: CHARACTERIZING DEVELOPMENTAL SHAPES FOR GROUPS AND INDIVIDUALS The second step in a useful data analysis is the attempt to highlight the key features of the data in terms of a model. In contemporary behavioral science research, one common approach to growth-curve analysis is to write a trajectory equation for each group and individual. Some mathematical and statistical aspects of these kinds of models are described next. Linear Growth Models for Repeated Measures Most forms of growth-curve analyses require longitudinal data with repeated measurements where we observe the variable (Y) at multiple


325

occasions (t ¼ 1 to T) on the same person (n ¼ 1 to N) and we can symbolize the scores (as Y ½tn ). This model includes three unobserved or latent scores representing: (1) the individual’s initial level (y0), (2) the slopes representing the individual linear change over time (y1), and (3) the independent errors of measurements (e[t]). To indicate the form of the systematic change, we use a set of group coefficients or basis weights, which define the timing or shape of the trajectory over time (e.g., Age½t ¼ t; or as A½t ¼ AgeðtÞ). It is typical to estimate the fixed group means for intercept and slopes (m0, m1) and also the implied random variance and covariance terms (s20, s21, s01) describing the distribution of individual deviations (d0n, d1n) around those means. We also usually assume that there is only one random error variance (s2e ), and the error terms are assumed to be normally distributed and presumably uncorrelated with all other components. The path diagram of Fig. 3 is an exact translation of the necessary matrix algebra of these models. These diagrams can be conceptually useful devices for understanding the basic modeling concepts. They are also practically 0s = 0.77 d0

ds

σs=1.5

σ0=9.8

µ 0=17.9

µs=10.7 1

y0

ys

α [1]=0.4 α [6]=6.5

α [2]=5.0

a[3]=6.2 a[4]=6.4 Y[1] Y1

σe=17.4 e1 e[1]

Fig. 3.

Y[2] Y2

17.4 e[2] e2

V3 Y[3]

17.4 e[3]

Y[4] Y4

17.4 e[4]

α [5]=6.5

V Y[5] Y55

17.4 e[5]

Y[6] Y6

17.4 e[6]

A Path Diagram of Numerical Results from a Latent Basis Growth Model from the Bradway–McArdle Longitudinal Data.

326

JOHN J. MCARDLE

useful because they can be used to represent the input and output of any of the SEM computer programs. These diagrams were originally used with variations on multiple regression and factor analysis models, but contemporary work has shown how these diagrams can be used in the context of growth and change (e.g., McArdle, 1986; McArdle & Epstein, 1987; McArdle & Anderson, 1990; McArdle & Woodcock, 1997). In this path diagram, the observed variables are drawn as squares, the unobserved variables are drawn as circles, and the required constant is included as a triangle. Model parameters representing ‘‘fixed’’ or ‘‘group’’ coefficients are drawn as one-headed arrows while ‘‘random’’ or ‘‘individual’’ features are drawn as two-headed arrows. In this case, the initial level and slopes are often assumed to be random variables with ‘‘fixed’’ means (m0, m1) but ‘‘random’’ variances (s20, s21) and correlations (r01).

Basic Linear Growth Models Some initial growth-curve modeling results for the Bradway–McArdle data are presented in Table 3. In these longitudinal models, any change score (y1) is assumed to be constant within an individual but it is not assumed to be the same between individuals. We do not usually estimate the unobserved scores but we do estimate several parameters, which characterize the key features of these unobserved scores. The first model labeled 3a is no-growth model fitted with only three parameters: An initial level mean (m0 ¼ 68:6), an initial standard deviation (s0 o 0.01), and an error variance (s2e ¼ 740). The parameters also yields a model likelihood (L2 ¼ 2276), which shows the no-growth baseline is a poor fit compared to the totally unrestricted model ðw2 ¼ 1510; df ¼ 24). This model is typically just used as a baseline against which to judge the fit of more informative models. The second linear growth model labeled 3b uses a fixed set of basis coefficients formed by taking A½t ¼ ½Age½t=10 or fixed values (e.g., A½t ¼ ½0:4; 1:4; 2:9; 4:2; 5:7; 6:5). This linear scaling is only one of many that could be used, but it was chosen to permit a practical interpretation of the slope parameters in terms of a per-decade change. In contrast to the nogrowth baseline, this linear growth model also has three more free parameters: a slope mean (m1), standard deviation (s1), and correlation (r01). This model fitted yields a new likelihood (L2 ¼ 2092), which is a large distance from the unrestricted model (w2 ¼ 1142 on df ¼ 3) but is an improvement in fit over the previous baseline (M1 vs. M0: Dw2 ¼ 368 on Ddf ¼ 3).

327


Table 3. Selected Results from Five Latent Growth Models Fitted to Longitudinal Data at Six Time Points (N ¼ 111; Step 2, see Figs. 3–5). Parameters

3a (Level)

3b (Linear)

3c (Quadratic)

3d (Latent)

3e (Spline)

Basis a[04] Basis a[14] Basis a[29]

1, 0 1, 0 1, 0

1, 0.4 1, 1.4 1, 2.9

1, 0.4, 120.42 1, 1.4, 121.42 1, 2.9, 122.92

1, 0.4 1, 5.0 (0.6) 1, 6.2 (0.07)

Basis a[42]

1, 0

1, 4.2

1, 4.2, 124.22

1, 6.4 (0.08)

Basis a[57]

1, 0

1, 5.7

1, 5.7, 125.72

1, 6.5 (0.08)

1, 0 68.6 (1.2) — —

1, 6.5 41.8 (1.4) 9.5 (0.40) —

1, 6.5 17.9 (1.0) 10.7 (0.19) —

1, 0, 0 1, 1, 0 1, 1, 0.80 (0.03) 1, 1, 0.98 (0.04) 1, 1, 1.01 (0.04) 1, 1, 1 22.2 (0.94) 50.0 (0.86) 15.1 (0.86)

740 (54) o0.01 (?) — —

345 (28) o0.01 (?) o0.01 (?) 40.99 (?)

41.3 (?) 22.3 (?) 29.3 (?) 0.77 (?)

17.4 (1.5) 9.8 (0.79) 1.5 (0.14) 0.77 (0.05)

8.3 (0.95) 9.4 (0.69) 8.1 (0.69) 0.50 (0.08)

— —

— —

15.9 (?) 0.33 (?), 0.84 (?)

— —

3

6

10

10

13

24

21

17

17

14

2276 1510

2092 41142

1830 4617

1583 122

1507 46

0.75

0.69

0.56

0.23

0.14

Fixed effects

Basis a[65] Level m0 Slope m1 Accelerate ma

1, 6.5, 47.8 0.42 0.12

2 1 26.5

(?) (?) (?)

Random effects Error se2 Level s0 Slope s1 Correlation r01 Accelerate sa Correlation r0a, rsa

7.1 (0.66) 0.60 (0.08), 0.04 (0.12)

Fit indices Numbers of parameters Degree of freedom Log likelihood Likelihood ratio (DLL) RMSEA

Note: ‘?’ freely estimated from data.

The resulting means describe a function that starts low at age 4 (m0 ¼ 41:8) but increases between ages 4 and 65 (by m1 ¼ 9:5 per decade). The variance estimates of the intercept and slope parameters are too small to interpret (sj o 0.02), but the error variance has been reduced (from s2e fM0g ¼ 740 to s2e fM1g ¼ 345). Of course, the main problem is that simple straight-line linear growth model does not fit these data very well.

328

JOHN J. MCARDLE

Including Polynomial Nonlinearity in Growth The data of Fig. 1 suggest some nonlinearity over age in the scores of almost all persons. To deal with this complexity, Wishart (1938) introduced the use of power polynomials to better fit the curvature apparent in growth data. The individual growth curve (consisting of t ¼ 1; T occasions) is summarized into a small set of linear orthogonal polynomial coefficients based on a fixed power-series of time (A[t], A[t]2, A[t]3,y, A[t]p) describing the general nonlinear shape of the growth curve. A second-order (quadratic) polynomial growth model can be introduced using a new set of component scores (e.g., yp) introduced to represent another level of change with a set of powered coefficients (1=pA½tp ) . This leads to an implied change model, where the change is linear with time (i.e., acceleration). Of course, a model of growth data might require this form of a second-order (quadratic), thirdorder (cubic), or even higher-order polynomial model fitted to the data. In all cases, additional variance and covariance terms are added to account for individual differences in latent scores. This polynomial growth curve approach remains popular (e.g., Bryk & Raudenbush, 1992). A quadratic polynomial model 3c was fitted next, including an additional slope variable (acceleration) defined by a fixed basis ð12A½t2 Þ; one new mean, one new deviation, and two additional correlations. This 10 parameter model is still far worse than the unrestricted model (w2 4617 on df ¼ 17), but is an improved fit compared to the linear model (M2 vs. M1: Dw2 ¼ 524 on Ddf ¼ 3). The error or unique variance has also been reduced substantially (from s2e fM2g ¼ 41:3) but problems arose in the variance–covariance estimation (sj o 0.02), numerical convergence was not achieved, and this flexible polynomial model seems to inappropriate for these data. Nonlinearity Using Latent Basis Curves A different alternative of the linear growth model was proposed by Rao (1958) and Tucker (1958, 1966) in the form of summations of ‘‘latent curves’’ (see Meredith & Tisak, 1990). The use of this latent growth curve offers a relatively simple way to investigate the shape of a growth curve – we allow the curve basis to take on a form based on the empirical data. In this approach, we write a model for the same person at different occasions but some of the basis coefficients (e.g., a[2], a[3], a[4], and a[5]) are free to be estimated. That is, the actual ages of the persons are known but the basis parameters are allowed to be freely estimated, and we end up with an optimal shape for the group curve and individual differences. In this model, the


329

A[t] is estimated just like a common factor loading and thus has mathematical and statistical identification problems. There are many alternative ways to estimate these parameters but, in general, there are many free parameters in the two-component growth model (p ¼ 6 þ T 1). The use of an estimated basis has been termed a ‘‘meta-meter’’ or ‘‘latent time’’ scale that can be plotted against the actual age curve for visual interpretation (see McArdle & Epstein, 1987). The fourth model fitted 3d was a latent basis growth model where some of the loadings, A[t], are free to vary. For the purposes of estimation, we fixed A½1 ¼ 0:4 (at age ¼ 4) and A½65 ¼ 6:5 (at age ¼ 65), but the four other coefficients were estimated from the data. This results in a large improvement in the model likelihood (L2 ¼ 1583), which is much closer to the unrestricted model (w2 ¼ 122 on df ¼ 17), and substantially better than the nested baseline model (Dw2 ¼ 1386 on Ddf ¼ 7), the nested linear model (Dw2 ¼ 1018 on Ddf ¼ 4), and better than the non-nested quadratic model with the same numbers of parameters ( 1830 vs. 1583). The error variance has also been reduced (s2e fM3g ¼ 17:4). The interpretation of the model parameters is an important part of the latent growth analysis, and these are displayed in Figs. 3 and 4. The estimated latent means are m0 ¼ 17:9 and m1 ¼ 10:7; their deviations are s0 ¼ 9:8 and s1 ¼ 1:5; and the two latent factors have correlation of r01 ¼ 0:77: These values were included on the path diagram of Fig. 3. The estimated basis coefficients were A½t ¼ ½0:4; 5:0; 6:2; 6:4; 6:5; 6:5; and these are repeated in Fig. 4 also (and discussed later). Nonlinearity Using Linear Spline Models Another way to deal with nonlinearity is by introducing the concepts of connected lines or splines (Seber & Wild, 1989). As a simple example here, we could write a model for the same person at different occasions where we define some critical age (i.e., C ¼ 30 years), and estimate two slope scores – y1 before the age C and y2 after age C. This implies the estimated parameters for the means and covariances of the intercept are now ‘‘re-centered’’ at age C, and the slopes represent changes before and after age C. These models can be combined with the polynomial models, and the critical cutoff ages (Cn) can be estimated as well and these concepts can create a potentially informative look at individual segments of growth (e.g., Cudeck & du Toit, 2001; Hamagami & McArdle, 2001). This leads us naturally to the idea for the final model 3e, which is a spline model composed of two latent slopes – the first representing changes up to age 14 and the second representing changes after age 14. The exact

330

Solid line = Exponential growth model (no decline in later ages) Dashed line = Latent Basis model (allowing decline at any age)

Latent Growth Score Expectations for the General Knowledge Scores. (a) Group Expectations from Two Models. (b) Individual Expectations from the Final Model.

JOHN J. MCARDLE

Fig. 4.


331

parameterization we used yielded a starting level of 22.2 at age 4 with a big average gain jump of 50 points (79.4) up to age 14, followed by a small steady gain of 15.1 points (78.1) from 14 to 65. However, this two latent splines should not be ignored, because it shows much better fit overall (w2 ¼ 46 on df ¼ 14) and when compared to the nested one-slope model (M6 vs. M4: Dw2 ¼ 76 on Ddf ¼ 3) is notable. Model-Based Expectations about Growth In the models just presented, the changes within a person is initially represented by the latent means and variance terms in the growth models. In later interpretations, we can examine the relative size of these parameters and make substantive interpretations about the group and individual differences. These parameters also allow us to form the expected growth curve charts for both the observed and true scores. Details of derivations are not presented here, but it is still useful to consider some key properties of the model (see McArdle, 1986, 1989; McArdle & Woodcock, 1997). These parameters can be combined to create expected means (my½t ¼ m0 þ m1 A½t). Here, we obtain the expected group curve at mean scores of m½t ¼ ½22:2; 72:1; 84:4; 86:6; 87:6; 87:7: These are plotted in Fig. 5a. This is a growth curve with a shape that rises quickly between ages 4 and 14, peaks at age 42, and exhibits no declines by age 65. All coefficients can be interpreted in terms of changes over decades, so the average 10-year change is 10.7 points, but a person who is 14 years old can be said to have the latent age of a 50 year old (i.e., A½14 ¼ 5:0). The individual differences in this model are seen in the variances for the level (s20 ¼ 9:52 ) and the slope (s21 ¼ 1:52 ) parameters. If we consider 100% of the changes between ages 4 and 65 (the observed period), we now say that 76% of all cognitive growth in knowledge occurred by age 14, 95% by age 42, and so on. The highest estimated coefficient never was higher than the fixed value at age 65 – this is important because it means the Verbal-Knowledge ability does not decline within people repeatedly measured over this span of ages. This is not the typical finding with cross-sectional age comparisons (e.g., McArdle et al., 2002). Goodness-of-Fit, Expectations and Residuals Over Time The choice between using a latent basis, or a polynomial, or spline or other nonlinear model can be substantively important. The initial idea from Wishart (1938) was that the basic shape of each individual curve could be captured with a small number of fixed parameters and random variance

332

Fig. 5.

JOHN J. MCARDLE

Individual Level Latent Growth Score Parameters and Residuals. (a) Individual Estimates of Latent Levels and Slopes. (b) Individual Residuals of the Final Growth Model.


333

components. In some cases, a fixed basis polynomial model is more parsimonious than a free basis model. However, the polynomial model also has: (a) a relatively fixed curvature and (b) requires an additional estimation of covariances among the new latent scores (yp; this is true even if orthogonal polynomial coefficients are used). Thus, the polynomial model may add more complexity (via parameters to be estimated) than is actually needed and the latent basis model may prove to be more efficient and interpretable. Although many recent textbooks overlook the latent basis model (e.g., Duncan, T. E., Duncan, S. C., Strycker, Li, & Alpert, 1998; Singer & Willett, 2003), we always treat this as an empirical choice (McArdle & Bell, 2000; McArdle & Nesselroade, 2003). We do not usually need to make any specific decision about these model differences at this point. It is enough to say that the two-part linear slope model fits best, it is far better than the others considered, and it is far more interesting than simply adding unique variances or covariances to improve fit. However, the interpretation of the two-slope model is very similar to what we have already described in the one-slope model, which changes direction. The one latent slope model may need to be reconsidered and the complexity of individual differences should not be overlooked. There are many other models that could fit to these data, and others have done so (e.g., Cudeck & du Toit, 2001; Hamagami & McArdle, 2001). In such an enterprise it may be important to do a more detailed analysis of the curve at the individual level. In Fig. 5a, we plot individual estimates of the latent parameters for the level and slope (These are Bayes-restricted estimates based on the latent model 3d). In this plot, many of the individuals have estimated levels and slopes in large cluster, but other individuals are scattered with negative slopes. The variation apparent here will be examined in the next section. The second plot 5b is a description of the residuals between the individual expected curve (based on the estimated level and slope) and the observed data. Here it appears that at least one individual is poorly represented (i.e., the large positive residual at age 45), and there is much more misfit at age 14 than at any other point in time. These are aspects of model fitting that need to be improved with subsequent analyses.

STEP 3: MODELING PREDICTORS OF INDIVIDUAL DIFFERENCES IN DEVELOPMENTAL SCORES The group mean parameters estimated in the previous analyses allow us to plot the group trajectory over time. Similarly the estimated variance

334

JOHN J. MCARDLE

parameters allow us to consider the size of the between group differences at each age. However, no prior information obtained in model fitting tells us about the sources of this common variance. To further explore the differences between persons we need to expand the basic latent growth model to include impacts on the latent parameters.

The Multilevel Growth Model Let us assume a variable termed X indicates some measurable difference between persons (e.g., sex, educational level, etc.). If we measure this variable at one occasion, we might like to examine its influence in the context of a growth model for Y[t]. One popular model is based on the use of ‘‘adjusted’’ growth parameters as popularly represented in the techniques of the analysis of covariance. In growth curve terms this model is written with fixed (group) coefficients (g) with some effect on the measured Y[t] scores at each occasions, and the X is an independent observed (or assigned) predictor variable. In this case, the growth parameters (m0:x ; m1:x ; s0:x ; s1:x ; s0;1:x ) are conditional on the expected values of the measured X variable. The parameters of the changes may be defined by the resulting difference or differential equation, but the reduction of error variance (sex ) is often considered as a way to understand the overall impacts. The apparent complexity of the covariance model leads to a simpler and increasingly popular way to add an external variable – we can write a mixedor multi-level model, where the X variable has a direct effect on the parameters of the growth curve. This means we have intercepts (v) and regression slopes (g) for the effect of X on the two latent components of Y[t]. This model is drawn as a path diagram in Fig. 6. This diagram is the same as Fig. 3 except that here, we have included several X variables as predictors of the levels and slope components. It may be useful to write this as a reduced form of SEM, so it is now clear that the three unobserved residual terms are not simply separable by standard linear regression or covariance equation (see McArdle & Hamagami, 1996). This path diagram gives the basic idea of external variable models, several other more complex alternatives will be considered in later sections. In this simple linear model, as in more complex models to follow, we can always add predictors X for the intercepts and the slopes. In some areas of research these models have been termed mixed-effects models (Laird & Ware, 1982; Littell, Miliken, Stoup, & Wolfinger, 1996; Singer, 1999). In other areas of research these same models have been termed random-coefficients or multi-level


335

−.53 8.82

e0*

Educ Dad

.08

es*

Educ Person

.02

1.34

2.1

9.3

8.9

63.2

21.9 1

y0

ys

=0

.76 .95

Y[1] Y1

17.4 e1 e[1]

Fig. 6.

Y[2] Y2

17.4 e2 e[2]

VY33 Y[3]

17.4 e[3] e3

Educ Mom

.25

Y[4] Y4

17.4 e4 e[4]

.98 1.0

VY55 Y[5]

17.4 e[5] e5

=1

Y[6] Y6

17.4 e6 e[6]

Results from Latent Growth With Mixed-effects or Multi-level Predictors Latent Path Predictors.

models, or slopes as outcomes or hierarchical linear models (e.g., Bryk & Raudenbush, 1992). In other SEM research these models were considered using factor analysis terminology as latent growth models with incomplete data and extension variables (e.g., McArdle & Epstein, 1987; McArdle & Hamagami, 1992). Using any terminology, these models can be generically represented by the parameters in the path diagram of Fig. 6, and this is a common way to understand the between group differences in within group changes. Once considered in this way, no new model fitting procedure is required. Results for Educational Influences on Latent Scores A variety of additional variables have been measured in the Bradway– McArdle collections, including demographic (e.g., gender, educational attainment by age 30 and 56, etc.), self reported health behaviors (e.g., smoking, drinking, physical exercise, etc.) and other problems (e.g., general

336

JOHN J. MCARDLE

health, illness, medical procedures, etc.), and personality measures (e.g., 16 PF factors). In the analyses presented here we consider four variables: (1) Gender (effect coded as –0.5 for males and +0.5 for females), (2) the person’s own educational level (at age 30 in years), (3) Father’s educational level (in years), and (4) Mother’s educational level (in years). We started with a baseline latent growth model with no predictors of the levels and slopes but then added additional variables as predictors of these levels and slopes (as in Fig. 6). In these mixed-models, we only fit the latent basis curve model (with a 0–1 basis), and we added Gender as an effectcoded variable (i.e., M ¼ 0:5 and F ¼ þ0:5), and the three indices of educational attainment centered at 12 years. Table 4 is a list of results including these four variables and the six intellectual ability measures. The first model 4a used all four variables and has a misfit (w2 ¼ 153 on df ¼ 33), which is a reasonable fit when compared to Table 4. Selected Results from Two Latent Growth Models with Four Extension Variables Fit to the Longitudinal Data at Six Occasions (Step 3, see Fig. 7). Parameters

4a: {Latent} Level

4a: {Latent} Slope

4b: {Latent} Level

4b: {Latent} Slope

Fixed effects Basis A[t] Intercept g0 Regression from Gender gg Regression from Educ_Pers gp Regression from Educ_Dad gd Regression from Educ_Mom gm

[ ¼ 0, 0.76, 0.95, 0.98, 1.00, ¼ 1] 22.8 (1.5) [0] 62.6 (1.7) [0] 0.73 (2.0) [0.04] 1.63 (2.0) [0.09]

[ ¼ 0, 0.76, 0.95, 0.98, 1.00, ¼ 1] 23.3 (1.5) [0] 62.6 (1.6) [0] 0 0

0.20 (0.04) [0.048]

1.32 (0.47) [0.32]

0.95 (0.47) [0.32]

0.68 (0.48) [ 0.23] 0.65 (0.55) [0.19]

0.58 (0.38) [0.19]

8.7 (0.82)

9.1 (0.75)

0.78 (0.54) [ 0.23]

0.13 (0.44) [0.00]

1.44 (0.45) [0.35] 0.24 (0.39) [ 0.08] 0

0

Random effects Residual sd Error se2 Correlation rd0,

9.0 (0.75) ds

17.4 (1.5) 0.81 (0.04)

8.8 (0.82) 17.5 (1.5) 0.81 (0.04)

Fit indices Numbers of parameters Degree of freedom Likelihood ratio RMSEA

32

28

33 153 0.18

37 159 0.17

Note: Latent growth with all four predictors set to zero yields L2 ¼ 195 on df ¼ 41.


337

free latent curve model (M7 vs. M4: Dw2 ¼ 32 on Ddf ¼ 16). This implies that the patterning of the correlations of predictors and outcomes can be estimated using two latent variables (see McArdle & Prescott, 1992). The MLE parameters suggest the following interpretations: 0 – The latent basis coefficients (A[t]) were unaffected by the inclusion of the predictors. 1 – There are no accurate (significant) differences between Males and Females on either levels or slopes. 2 – The Person’s Educational level (at age 30) was not predictive of the initial cognitive level (at age 4) but was a positive indicator of overall changes in adulthood (g ¼ 1:32½0:32). 3 – The educational level of the Father (at age 14) was a positive contributor to the initial cognitive level (g ¼ 0:95½0:47) but not to the adult slope. 4 – The educational level of the Mother (at age 14) was not related to the initial level or slope. The second set of results 4b was based on a model, where all coefficients for the Gender and Mothers’ Education were fixed at zero. This model fits the data just about as well as before (M8 vs. M7: Dw2 ¼ 6 on Ddf ¼ 4), but the results obtained now differs in one respect – the coefficient for the Father’s Education on initial level is no longer accurate. This is probably due to the fact that the Mother’s Education (a) no longer has a direct effect, (b) it is positively correlated (r ¼ 0:62) with Father’s Education, and (c) the previous ‘‘suppression’’ effect has been eliminated and the coefficient is not as strong. In essence, the previous interpretation (4a) needed to consider the negative coefficient of Mother’s Education, and reinterpret the overall impact or do so in terms of the difference between parental educational levels. Upto this point, it seems the only direct educational influence found here is that of the Person’s Educational level on their own slope. However, the model can be expanded to include more complex path representations, and this leads to a comprehensive series of mixed-model path equations. Assuming the three education variables are centered at grade 12, and we view the person’s educational attainment as an outcome of the parent’s educational attainment, we estimated a simultaneous latent path model where the Mother’s Education (0.25) and Father’s Education (0.08) had direct effects on the Person’s Education, but only the Person’s Education had an effect on the latent level (0.20) and slope (1.34). This resulted in a good overall fit (w2 ¼ 162 on df ¼ 39; so Dw2 ¼ 3 on Ddf ¼ 2), and emphasizes the positive but indirect effects of the Mother (0.025 1.34) on the adult slopes of their children. This model is drawn as an SEM path diagram in Fig. 6 to illustrate how the basic principles of multiple regression and path analysis follow directly to the latent variables of a growth model (see McArdle, 2001; McArdle & Hamagami, 2003).

338

JOHN J. MCARDLE

STEP 4: STUDIES OF GROUP DIFFERENCES IN DEVELOPMENTAL SHAPES The next step in longitudinal data analyses deals with the investigation of group differences in developmental shapes or trajectories. While this is a natural question raised in any multiple group SEM context, it is often overlooked in this literature on mixed-effects or multi-level models. This contrast between the two approaches was an issue raised by McArdle and Epstein (1987) and McArdle (1990), and we now explore some alternatives. Group Differences from a Multiple Group Perspective One initial representation of group differences uses a new set of estimated parameters to summarize between groups. This idea is clearly represented by coding a set of variables (X) to characterizing the group differences and then examining the effect of this set (X) on the model parameters (see, e.g., Fig. 6). However, this method is limiting in a number of important ways. For example, some reasonable forms of growth processes are not immediately possible to account for within the standard framework. For example, different groups of people could have different ‘‘amplitude’’ or be in different ‘‘phases’’ in their characteristic growth pattern. These features of growth are not separated within the basic level and shape parameters although these features may be realistic features of development. An SEM treatment of this kind of a model uses concepts derived from multiple-group factor analysis (e.g., Jo¨reskog & So¨rbom, 1979; McArdle & Cattell, 1994). In these kinds of models, each group (g ¼ 1toG) is assumed to follow a latent growth model where the basis parameters (A½tðgÞ ) are defined by the application. Fig. 7 is a path diagram representing these kinds of multiple group growth models (from McArdle & Hamagami, 1992). Since the groups need to be independent (each person can only be in one group), this kind of grouping is most easily done for discrete categorical variables (i.e., sex, but not educational level). In a most fundamental form, the multiple group growth model permits the examination of the presumed invariance of the latent basis functions (i.e., A[t](1) ¼ A[t](2) ¼ y ¼ A[t](g) ¼ y ¼ A[t](G)). The rejection of this model implies that each independent group has a different shape of growth. This parametric model is not one that is easily represented using standard mixed-effects or multi-level models (for details, see McArdle & Hamagami, 1996). If invariance is found, we can also examine the equality of the

Group 1

ρ0,s

y0 * σ0

σs

µs

µ0 1

y0

σe e[1]

Y[2]

α1 α2

σe e[2]

Y[3]

V

σe e[3]

Y[4]

σe

α4

Fig. 7.

σ0

α5

Y[6]

1

σe e[6]

e[1]

Y[2]

σe e[2]

ys

α1 α2 α3

Y[1]

σs

µs

µ0

α6

σe e[5]

ys *

y0

Y[5]

σe e[4]

y0*

ys

α3

Y[1]

Group 2

ys *

Y[3]

V

σe e[3]

Y[4]

σe

α4

A Latent Growth Curve Path Model for Separate Groups.

α6

Y[6]

Y[5]

σe e[4]

α5

σe e[5]

e[6]


ρ0,s

339

340

JOHN J. MCARDLE

ðGÞ ðGÞ variances of the latent levels and slope (sðgÞ and sðgÞ s ¼ . . . ss ). 0 ¼ . . . s0 ðgÞ Further analyses could include the error deviations (se ), the total slope variance and covariances, and functions of all the other parameters. We may bring back the typical mixed-effects group difference parameters when we examine the invariance of the latent means for initial levels and slopes ðGÞ ðgÞ ðGÞ (mðgÞ 0 ¼ m0 and m1 ¼ m1 ). Of course, these group differences in the fixed effects can be coded in the same way as in the typical mixed-effects analyses. However, in important ways, each of these multiple group hypotheses represents a different kind of nonlinearity than was possible to examine using the mixed-effects approach.

Results for Group Differences in Cognitive Growth for Males and Females The group differences due to sex, education, and complete versus incomplete data were also studied using the multiple group growth curves approach. To illustrate this kind of analysis here, Table 5 gives the initial longitudinal data on the six intellectual ability measures separately for each Gender. In these cases, the two groups are created so the unrestricted likelihood for these data is based on two sets of mean and covariance matrices; one for males and one for females. The first model (5a) allows both groups to have completely different latent growth curves. The model now includes 10 parameters for each group, and the 20 MLEs are listed in the first two columns. This results in a fairly reasonable fit to both data sets (w2 ¼ 148 on df ¼ 34). A few small differences in MLEs can be seen between the two groups, but the key difference appears to be the smaller error variance for the females (F fs2e g ¼ 13:9 but Mfs2e g ¼ 21:0). The second model (5b) adds the restriction that the latent basis coefficients, while free to vary, must be identical across males and females. This model is similar in fit to the free model (w2 ¼ 152 on df ¼ 38; Dw2 ¼ 4 on Ddf ¼ 4), and this indicates the shapes of the curves may be considered to be parallel across both groups. The third model (5c) adds the restriction that all latent basis parameters, while free to vary, must be identical across males and females. This model is similar in fit to the free model (w2 ¼ 169 on df ¼ 44) and various comparisons are clear. (a) This is slightly worse in fit than the previous parallel shape model (w2 ¼ 17 on Ddf ¼ 6), indicating some of the latent means or covariances are different – and it is the error variance here. (b) This is not much worse in fit than the free group model (w2 ¼ 21 on Ddf ¼ 10), indicating not much overall differences between groups. (c) This is a bit worse in

341


Table 5. Numerical Results from Multiple Group Latent Growth Models Fitted to Male & Female Longitudinal Data (Step 4, see Fig. 8). Growth Model Parameters

5a: Latent Growth for Males (n ¼ 52)

5a: Latent Growth for Females (n ¼ 59)

5b: Loading Invariance over Both Groups

5c: Total Invariance over Both Groups

Fixed effects Basis a[04] Basis a[14] Basis a[29] Basis a[42] Basis a[57] Basis a[65] Level m0

1, 1, 1, 1,

1, 0 0.75 (0.02) 0.94 (0.02) 0.99 (0.02) 1.01 (0.02) 1, 1 21.6 (1.4)

1, 1, 1, 1,

1, 0 0.77 (0.01) 0.96 (0.01) 0.98 (0.02) 1.00 (0.02) 1, 1 22.7 (1.4)

65.0 (1.9)

65.9 (1.5)

Error se2

21.0 (2.7)

13.9 (1.6)

Level s0 Slope ss

8.7 (1.1) 9.7 (1.3)

9.8 (1.0) 9.3 (1.1)

Correlation r0s

0.64 (0.05)

0.85 (0.05)

Slope ms

1, 0 0.76 (0.01) 0.95 (0.01) 0.98 (0.01) 1.00 (0.01) 1, 1 21.5(1.4), 22.8 (1.4) 64.8 (1.6), 66.0 (1.5) 1, 1, 1, 1,

1, 0 0.76 (0.01) 0.95 (0.01) 0.98 (0.01) 1.00 (0.01) 1, 1 22.2 (0.97)

1, 1, 1, 1,

65.5 (1.2)vsp:0.5

Random effects

Numbers of parameters Log likelihood Likelihood ratio Degree of freedom RMSEA

20

21.4 (2.8), 14.0 (1.7) 8.7 (1.1),9.7 (1.0) 9.6 (1.2), 9.2 (1.1) 0.63 (0.11), 0.85 (0.05) 16

17.4 (1.5) 9.3 (0.75) 9.4 (0.84) 0.74 (0.05)

10

1571. 148 34

1573 152 38

1582 169 44

0.24

0.23

0.23

fit than the one-group model (w2 ¼ 48 on Ddf ¼ 10), even though the MLE parameters are identical, and this indicates there may be some basic differences between groups that are not captured by this growth model. As suggested earlier, these multiple group growth models have been used to compare the complete and incomplete subsets of these data. In all such cases, the key results for (a) the complete data only and for (b) the complete and incomplete data together, and the parameters are much the same (as in Diggle & Kenward, 1994; Little, 1995; McArdle, 1994). As a statistical test for parameter invariance over these groups we calculated the difference in

342

JOHN J. MCARDLE

the likelihoods, and these differences are trivial (w2 ¼ 21 on Ddf ¼ 10). This suggests that selective dropout or subject attrition can be considered random with respect to the General Knowledge abilities measured here. This result allowed us to combine the complete and incomplete data for a more powerful analysis. Mixture Models for Latent Groups Another fundamental problem is the discrimination of models of multiple curves for the same people from models of multiple groups with different curves. It is possible for us to have, say, three clusters of people each with a distinct growth curve but when we aggregate information over all the people we end up with three multiple factors and mixed curves. This is the essence of a latent grouping of people, and parallels the focus on a ‘‘person centered approach’’ to multivariate data analysis (e.g., Block, 1971; Cattell & Birkett, 1980; Magnusson, 1995). The recent series of models termed growth mixture models have been developed for this purpose (Wedel & DeSarbo, 2002; Mutheń & Mutheń, 1995, 2002; Nagin, 1999). In these analyses, the distribution of the latent parameters are assumed to come from a ‘‘mixture’’ of two or more overlapping distributions. Current techniques in mixture models have largely been developed under the assumption of a small number of discrete or probabilistic ‘‘classes of persons’’ based on mixtures of multivariate normals (e.g., two classes). More formally, we can write a model as a probability weighted sum of curves, where the probability of class membership (Prob{cn}) is defined for the person in c ¼ 1 to C classes. In this kind of growth mixture analysis we estimate the most likely threshold parameter for each latent distribution (tp, for the pth parameter), while simultaneously estimate the separate model parameters for the resulting latent groups. The growth mixture models may be seen as a model-restricted fuzzy-set cluster analysis – a multiple group model without exact knowledge of group membership for each individual. The concept of an unknown or latent grouping can be successively based on the logic of multiple group factorial invariance – starting with equality of latent level means and variances, then on the latent slope means and variances, then on both the level and slope, then on the growth loadings, and so on. The resulting MLEs yield a likelihood which can be compared to the results obtained from a model with one less class, so the mixture model distribution can be treated as a hypothesis to be investigated. As in standard discriminant analysis, we can also

343


estimate the probability of assignment of individuals to each class in the mixture. A variety of new computer programs have been developed for this purpose (e.g., Mplus, by Mutheń & Mutheń, 2000). Results from Latent Mixture Models These latent growth mixture models were fit using all the longitudinal data (Fig. 1) and some results are described in Fig. 8. In a first latent mixture model, we estimated a two-class model with free parameters for both groups. This model required 21 parameters, adding the one more than the previous two-group model, and led to another likelihood (L2 ¼ 1544). We need to recognize that the statistical basis of this comparison is still somewhat controversial, but if we consider the threshold as an implied parameter in some previous models, we can get some sense of the gain in fit. The extra parameter fitted was the threshold (t ¼ 0:55) – a z-score that suggests the total group can be considered a mixture of two classes of different sizes, n ¼ 72 and 39, with different growth patterns. By contrast to the one-class model (w2 ¼ 78 on df ¼ 10) or even versus the unrestricted two-group model (w2 ¼ 92 on df ¼ 1), two groups are better than one. By inspection, we can see that the larger group has a growth pattern with parameters that are very similar to the overall pattern described earlier (i.e., 100

100

90

90

80

80

70

70

60

60

Model 1, Class 1

50

Model 2, Class 1

50

Model 1, Class 2

40

Model 2, Class 2

40

30

30

Same latent basis A[t], Fit χ2 =30, df=3, Prob(Class1)=.92

20 10

Free latent basis A[t], Fit χ2 =34, df=12, Prob(Class1)=.67

20 10

0

0 0

4

14

30 42 56 Age (Years)

65

75

0

4

14

30 42 56 Age (Years)

65

75

Fig. 8. Results from SEM Growth-Mixture Model with Two ‘‘Classes’’ of Growth Curves. (Note: Model Includes Two Classes of LGMs and Probability of Group Assignment (Results from Mplus).)

344

JOHN J. MCARDLE

M4). However, the smaller second group seems to have a higher initial basis (a½tð1Þ ¼ 0:73 vs. a½tð2Þ ¼ 0:84), a much higher initial mean (mð1Þ 0 ¼ 16:9 vs. ð1Þ ð2Þ mð2Þ ¼ 31:4), lower slope changes (m ¼ 71:1 vs. m ¼ 55:7), and much low0 1 1 ð2Þ er variances (sð1Þ ¼ 7:1 vs. s ¼ 1:9). This second group shows almost no 0 0 ð2Þ variance in slope (sð1Þ ¼ 7:1 vs. s 40:1), and this could reflect an estimation 1 1 problem, or even a problem of limitation of measurement. Assuming these are not critical problems, the second smaller class appears to be an initially higher functioning group which slowly moves toward the final adult levels. In second latent mixture model, we allowed the possibility of two latent classes (C ¼ 2) with different parameters for the latent means and variance but assuming the same growth basis. This results in a more equal separation of persons (t ¼ 0:07; so n ¼ 56 and 55) with same free basis coefficients (A½t ¼ ½¼ 0; 0:76; 0:95; 0:98; 1:00; ¼ 1), but with the same pattern of means and variances as the first model. This model loses a bit in fit (L2 ¼ 1565; so M14 vs. M13: Dw2 ¼ 36 on Ddf ¼ 6), so it may not be a good idea to force exactly the same shape on groups that start at different points. Either way, this still implies that a small group of persons started much higher than average score and had a smaller change over time but the two classes of curves eventually do converge in adulthood.

STEP 5: STUDYING DYNAMIC DETERMINANTS ACROSS MULTIPLE VARIABLES In recent research, we have considered some ways to improve the clarity of the basic dynamic change interpretations with conventional SEM analytic techniques. This has led to the development of a set of alternative models, based on classical principles of dynamic growth and change (including decline), but represented in the form of latent difference scores (e.g., McArdle, 2001; McArdle & Nesselroade, 1994). This alternative representation makes it relatively easy to represent a dynamic hypothesis about the change within a variable, and about the time-ordered determination of one variable upon another. An overview of this new approach is described here as the fifth step in longitudinal data analysis. Modeling Latent Difference Scores The introduction of multiple variables at each longitudinal occasion of measurement leads naturally to questions about time-dependent relationships


345

among growth. A classical SEM for multiple variables over time is based on a latent variable cross-lagged regression model (see Cook & Campbell, 1979; Rogosa, 1978). This model can be written for latent scores with over-time auto-regressions (fy, fx) and cross-regressions (dyx, dx) for time-lagged predictors, but the standard applications of this model do not include systematic growth components (i.e., individual slopes). For this reasons, recent SEM analyses have examination of parallel growth curves, including the correlation of various components (McArdle, 1988, 1990; Willett & Sayer, 1994). A popular alternative used in multi-level and mixed-effects modeling is based on the analysis of covariance with X[t] as time-varying covariates. In this model, the regressions (d) are fixed (group) coefficients with the same effect on Y[t] scores at all occasions. These last two models are easy to implement using existing computer software (e.g., Sliwinski & Buschke, 1999; Sullivan, Rosenbloom, Lim, & Pfefferman, 2000; Verbeke, Molenberghs, Krickeberg, & Fienberg, 2000), but the typical applications are limited to a few basic forms of dynamic hypotheses. To expand our SEM for other dynamic concepts we start in a different way. First, we assume we have a pair of observed scores (Y ½t and Y ½t 1) measured over a defined interval of time (Dt ¼ 1), and we write a model with latent scores (y½t and y½t 1), and corresponding errors of measurement (e½t and e½t 1), so the new latent variable Dy[t] is directly interpreted as a latent difference score. This latent difference score (Dy½tn ) is not the same as an observed difference score (DY ½tn ), because the latent score is considered separate from the removal of the model-based error component. In this latent difference score approach, we do not directly define the A[t] coefficients, but instead we directly define changes as an accumulation of the first differences among latent variables. This deceptively simple algebraic device allows us to generally define the trajectory equation as an accumulation of the latent changes (Dy½t) up to time t based on any model of change. Using this approach, all of the previous latent growth models can be drawn in terms of first differences, and some new models can more easily emerge (as in McArdle & Nesselroade, 1994; McArdle, 2001; McArdle & Hamagami, 2001). For example, we can write a composite change expression model, where we permit both a systematic constant change (a) and a systematic proportional change (b) over time. This linear difference model lead to a nonlinear mixed-effects model trajectory from a simple accumulation of first differences. It can also be fitted using standard SEM software (e.g., LISREL, Mx, etc.) To deal with multiple variables we can now write a bivariate dynamic change score model, such as the model depicted in the path diagram of Fig. 9.

346

σy2

eey[2] 2

ey[1]

σx0,y0

y0

σy0,ys ys*

Y[6]

Y[3] V3

Y[4]

y[1]

y[2]

y[3]

y[4] e4

βy

µy0

σys,xs

βy ∆ y[2]

σys µys

Y[5]

Y[6]

ys

βy ∆ y[3]

y[6]

y[5] e5

βy

βy ∆ y[5]

∆ y[4]

∆ y[6]

αy γx

γx

γx

γx

γx

σy,x

K

µxs x*s σx0,ys

Y[1]

σy0

y*0

σy0,xs

ey[6]

ey[4]

σx0,xs

xs

σxs µx0

γy

γy

βx σx0

∆x[3]

∆x[2]

x0

σx2

Fig. 9.

γy

γy

αx

βx

∆x[4]

βx

∆x[6]

∆x[5]

βx

βx

x[1]

x[2]

x[3]

x[4] e4

x[5] e5

x[6]

X[1]

X[6]

X[3] V 3

X[4]

X[5]

X[6]

ex[1]

eex[2] 2

ex[4]

ex[6]

A Bivariate Latent Difference Score (BLDS) Path Model for Dynamic Hypotheses.

JOHN J. MCARDLE

x*0

γy


347

We assume a dual change score model within each variable (parameters a and b) but also permit a coupling hypothesis (parameters g) across different variables. This model is used to estimate the time-dependent effect of latent x[t] on Dy[t] (gyx) as well as coupling parameter representing the time-dependent effect of latent y[t] on Dx[t] (gxy). This model subsumes all aspects of the previous cross-lagged, correlated growth, and time-varying covariate models as special cases and this is useful because results from these dynamic alternatives can be quite different (see McArdle & Hamagami, 2001). These latent difference score models can lead to more complex nonlinear trajectory equations (e.g., non-homogeneous equations), but the use of latent difference scores makes it practical to analyze a variety of dynamic models using standard SEM. Results from Fitting Latent Difference Score Models The latent difference score dynamic models were fitted (using Mx and Mplus) to both the General Knowledge variable (now termed Verbal) and also to a second set of Non-Verbal scores. We will only highlight the dynamic results here (but see McArdle & Hamagami, 2004). In order to fit the dual change model the additive slope coefficient was fixed for identification purposes (a ¼ 1) but the mean of the slopes was allowed to be free (m1). This allowed estimation of: (a) auto-proportion effects (b ¼ 0:34; 1:38), (b) initial level means (m0 ¼ 14:8; 32:1) at Age ¼ 5, and (c) linear slope means (m1 ¼ 26:0; 110:1) for each 5-year-period after (Age 4 5). The goodnessof-fit of the dual change model was compared to every other nested alternative and these comparisons show the best fit was achieved using this model (e.g., versus a linear growth, w2 ¼ 584; 439 on df ¼ 1). Several alternative Verbal–Non-Verbal bivariate coupling models based on Fig. 9 were fitted to the data. A first model included six dynamic coefficients (two each for a, b, g), four latent means (m), six latent deviations (s), and six latent correlations (r). This model was fitted with N ¼ 111 individuals with at least one point of data, 498 individual data observations, and yield one overall fit ( 2logL ¼ 7118), which was different from a random baseline (w2 ¼ 379 on df ¼ 16). The model parameters can be used to form the expected values described in Fig. 9. These parameters are specific to the time interval chosen (i.e., Dt ¼ 5) and any calculation of the other information (e.g., explained latent variance) requires a specific interval of age. However, these seemingly small differences can accumulate over longer periods of time so the larger N[t] is expected to account for an increasing variance in DV [t] over age.

348

JOHN J. MCARDLE

The fitting of a further sequence of alternative models is needed to interpret the replicability of the coupling across the V [t] and N [t] variables. Table 6 gives details about two specific models fit to examine whether one or more of the coupling parameters (g) were different from zero. In the first alternative model (6a), the parameter representing the effect of N[t] on DV[t] was fixed to zero (gx ¼ 0), and this led to a notable loss of fit (w2 ¼ 123 on df ¼ 1). The second alternative (6b) assumed no effect from V ½t on DN½tðgy ¼ 0Þ and this is a much smaller loss of fit (w2 ¼ 27 on df ¼ 1). These results were examined in a number of results of additional ways, but these comparisons suggest that the second model 6b was the most reasonable representation of these longitudinal data. The resulting interpretation is a dynamic process, where scores on Non-Verbal abilities have a tendency to impact score changes on the Knowledge-Verbal scores, but there is no notable reverse effect. The estimated model parameters are highly dependent on the scalings used, but the trajectory expectations allow us to interpret the results in a relatively ‘‘scale-free’’ form – Fig. 10 gives a summary of this state-space plot as a vector field plot (for details, see Boker & McArdle, 1995; McArdle et al., 2001). Any pair of coordinates is a starting point (y0, x0) and the directional arrow is a display of the expected pair of 5-year changes (Dy, Dx) from this point. The last two figures show an interesting dynamic property – the change expectations of a dynamic model depend on the starting point. From this perspective, we can also interpret the positive level–level correlation (ry0;x0 ¼ 0:78), which describes the placement of the individuals in the vector field, and the small slope–slope correlation (rys;xs ¼ 0:06), which describes the location of the subsequent change scores for individuals in the vector field. The resulting ‘‘flow’’ shows a dynamic process, where scores on Non-Verbal abilities have a tendency to impact score changes on the Verbal scores, but there is no notable reverse effect.

DISCUSSION This chapter can only serve as an introduction to a general class of procedures that can be classified under the rubric of ‘‘latent growth curve modeling techniques.’’ The brevity of this chapter cannot do justice to the broad applicability of these techniques, and the five steps outlined here represent just one way to organize some of the inherent complexities of this topic. In this last section, I discuss some of the future directions and challenges for this kind of research.

Model Parameters

6a (No Coupling from Non-Verbal) Knowledge

Non-Verbal

6b (No Coupling from Verbal) Knowledge

Non-Verbal

15.5 0.01 ¼1 0.77 — ¼0

31.8 46.7 ¼1 0.58 0.70 —

5.85 1.64 8.82 o0.99 o0.99 0.39 o0.99 0.39

10.8 4.92 6.34 0.52

(a) Fixed effects Initial mean m0 Slope mean ms Loading a Proportion b Coupling from knowledge gnk Coupling from non-verbal gkn

14.8 26.1 ¼1 0.34 — 0.15

32.1 42.4 ¼1 0.40 ¼0 — (b) Random effects

Error deviation se Initial deviation s0 Slope deviation ss Correlation r0s Correlation ry0, x0 Correlation rys, xs Correlation ry0, xs Correlation rys, x0

7.04 2.87 4.29 o0.99 o0.99 0.63 0.44 0.85

10.5 2.84 5.52 0.40

(c) Goodness-of-fit 2log L Parameters

7241 19

7145 19


Table 6. Results of Bivariate Latent Difference Score (BLDS) Dynamics Models Fitted to Two Variables over Six Occasions (for details, see McArdle & Hamagami, 2004a, b; Step 5, see Figs. 9 and 10).

349

JOHN J. MCARDLE

350 100 90 80

Non-Verbal

70 60 50 40 30 20 10 0

0

10

20

30

40

50

60

70

80

90

100

Verbal

Fig. 10. Results from the Vector Field Representation of the BDLS for the Sixoccasion Longitudinal Data. (Note: Each Arrow Represents the Expected Change of Direction in the Scores – the Derivatives – for Pairs of Initial Starting Abilities.)

Formal models for the analysis of these kinds of growth curves have been developed in many different substantive domains. Researchers have found many creative ways to analyze average trends in growth data, including classical Analysis of Variance techniques (e.g., Pothoff & Roy, 1964; Bock, 1975). These classical methods provide powerful and accurate tests of ‘‘group trends.’’ However, the introduction of individual differences in change analyses led to a great deal of statistical controversy in model fitting. Early work on these problems led to the polynomial growth models by Wishart (1934), where an individual regression coefficient was used to describe a growth characteristic of the person (see Rogosa & Willett, 1985). Other current techniques have roots in the important innovations by Meredith & Tisak (1990), who showed how the ‘‘Tuckerized curve’’ models


351

(named in recognition of Tucker’s contributions) could be represented and fitted using structural equation modeling based on restricted common factors. For these reasons, the term latent curve models (LCM) seems appropriate for any technique that describes the underlying growth in terms of latent changes using the classical assumptions (e.g., independence of errors). These innovative techniques were important because this made it possible to represent a wide range of alternative growth and change models by adding the benefits of the structural equation modeling techniques (Meredith & Tisak, 1990; McArdle & Prescott, 1997; McArdle, 1986; McArdle & Epstein, 1987; McArdle & Anderson, 1990; McArdle & Hamagami, 1992, 2001). During the last decade it has become possible to prove that many seemingly different statistical models have identical properties and can yield identical results. They are all based on fitting observed raw-score longitudinal growth data to the same theoretical model using the same likelihoodbased techniques (as in Little & Rubin, 1987; McArdle, 1994; McArdle & Bell, 2000). These latent growth models have since been expanded upon and described by many others (McArdle & Woodcock, 1997; Willett & Sayer, 1994; Mutheń & Curran, 1997; Metha & West, 2000). The contemporary basis of latent curve analyses can also be found in the recent developments of multi-level models (Goldstein, 1995; Bryk & Raudenbush, 1992) or mixedeffects models (Laird & Ware, 1982; Singer, 1998). In important work by Browne & du Toit (1991), classical nonlinear models were added as part of this same framework (see Cudeck & du Toit, 2001; McArdle & Hamagami, 1996, 2001; Pinheiro & Bates, 2000). When these options are added to the latent variable path analysis models of SEM (e.g., McArdle & Prescott, 1992), many limitations apparent in previous longitudinal research can be overcome. In a similar way, most of the models presented here can be estimated using contemporary computer programs including: (1) the SAS and S-Plus packages (Littell et al., 1996; Singer, 1998; Verbeke & Molenberghs, 2000; Pinheiro & Bates, 2000), (2) general SEM programs such as LISREL (McArdle & Epstein, 1987; Cudeck & duToit, 2001), Mx (Neale, Boker, Xie, & Maes, 1999), Mplus (Mutheń & Mutheń, 2002), and AMOS (Arbuckle & Wothke, 1999), and (3) specialized softwares such as MIXOR (Hedecker & Gibbons, 1996) and MLn (Goldstein, 1995). The results for the classical SEM for latent curves will be the same no matter what program is used (for demonstration, see Ferrer, Hamagami, & McArdle, 2004). Some of the more complex models used here (e.g., Table 6) now require specialized SEM software (e.g., Mx or Mplus), but it is likely that these routines will become a part of standard computer packages in the near future.

352

JOHN J. MCARDLE

At the same time, I need to recognize that these SEM developments represent a limited class of longitudinal data analyses (e.g., Nesselroade & Baltes, 1979; Collins & Sayer, 2001), and we should be on the lookout for possible improvements. In the analyses presented here, I have also tried to present some of our most up-to-date interpretations dealing with the developmental-dynamic processes around aspects of human cognition over the life-span. Indeed, some of the most difficult problems for future work on latent curves will not involve SEM statistical analysis or computer programming, but will be focused on the rather elusive meaning of the latent model parameters themselves (Zeger & Harlow, 1987; McArdle & Nesselroade, 2003). These substantive-methodological problems remain among the most difficult dynamic challenges for our future work.

ACKNOWLEDGMENTS The work described here has been supported since 1980 by the National Institute on Aging (Grant#AG-07137). I am especially grateful to the work of my close friend and colleague, Fumiaki Hamagami, and to my close collaborations with Katherine P. Bradway and John L. Horn. This research was also helped by the support of my many friends and colleagues, including Steven Aggen, Paul Baltes, Steven Boker, Emilio Ferrer-Caja, Paolo Ghisletta, Patty Hulick, Bill Meredith, John Nesselroade, Carol Prescott, and Dick Woodcock. I also thank the PaVie team for excellent editorial suggestions and assistance in the presentation and preparation of this manuscript. Reprints can be obtained from the author at the Jefferson Psychometric Laboratory, P.O. Box 400400, Department of Psychology, University of Virginia, Charlottesville, VA 22904, USA.

REFERENCES Abeles, R. (Ed.) (1987). Life-span perspectives and social psychology. Hillsdale, NJ: Erlbaum. Arbuckle, J. L., & Wothke, W. (1999). AMOS 4.0 user’s guide. Chicago: Smallwaters. Baltes, P. B. (1968). Longitudinal and cross-sectional sequences in the study of age and generational effects. Human Development, 11, 145–171. Baltes, P. B. (1983). Life-span developmental psychology: Observations on history and theory revisited. In: R. M. Lerner (Ed.), Developmental psychology: Historical and philosophical perspectives (pp. 79–111). Hillsdale, NJ: Erlbaum. Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (1998). Life-span theory in developmental psychology. In: R. M. Lerner (Ed.), Handbook of child psychology: Vol. 1. Theoretical models of human development (5th ed., pp. 1029–1143). New York: Wiley.


353

Baltes, P. B., & Nesselroade, J. R. (1970). Multivariate longitudinal and cross-sectional sequences for analyzing ontogenetic and generational change: A methodological note. Developmental Psychology, 2, 63–168. Baltes, P. B., & Nesselroade, J. R. (1972). Cultural change and adolescent personality development: An application of longitudinal sequences. Developmental Psychology, 7, 244–256. Baltes, P. B., & Nesselroade, J. R. (1979). History and rationale of longitudinal research. In: J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in the study of behavior and development. New York: Academic Press. Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (Eds) (1977). Life-span development methodology: Introduction to research methods. Monterey, CA: Brooks-Cole. Bell, R. Q. (1954). An experimental test of the accelerated longitudinal approach. Child Development, 25, 281–286. Block, J. (1971). Lives through time. Berkeley, CA: Bancroft. Bock, R. D. (1975). Multivariate statistical methods in behavioral research. New York: McGrawHill. Boker, S. M., & McArdle, J. J. (1995). Statistical vector field analysis applied to mixed crosssectional and longitudinal data. Experimental Aging Research, 21, 77–93. Bradway, K. P., & Thompson, C. W. (1962). Intelligence at adulthood: A 25 year follow-up. Journal of Educational Psychology, 53(1), 1–14. Browne, M., & du Toit, S. H. C. (1991). Models for learning data. In: L. Collins & J. L. Horn (Eds), Best methods for the analysis of change (pp. 47–68). Washington, DC: APA Press. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage. Cattell, R. B. (1963). The structuring of change by P-technique and incremental R-technique. In: C. W. Harris (Ed.), Problems in measuring change (pp. 167–198). Madison, WI: The University of Wisconsin Press. Cattell, R. B. (1969). Comparing factor trait and state scores across ages and cultures. Journal of Gerontology, 24, 348–360. Cattell, R. B. (1970). Separating endogenous, exogenous, ecogenic, and epogenic component curves in developmental data. Developmental Psychology, 3(2), 151–162. Cattell, R. B., & Birkett, H. (1980). Can P-technique diagnosis be practicably shortened? Some proposals and a test of a 50 day abridgement. Multivariate Experimental Clinical Research, 5, 1–16. Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design & analysis issues for field settings. Boston, MS: Houghton Mifflin. Cnaan, A., Laird, N. M., & Slasor, P. (1997). Using the general linear mixed model to analyse unbalanced repeated measures and longitudinal data. Statistics in Medicine, 16, 2349–2380. Collins, L., & Sayer, A. (Eds) (2001). New Methods for the Analysis of Change (pp. 137–176). Washington, DC: APA Press. Cudeck, R., & du Toit, S. H. C. (2001). Mixed-effects models in the study of individual differences with repeated measures data. Multivariate Behavioral Research, 31, 371–403. Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of individual differences in mental ability from childhood to old age: Follow-up of the 1932 Scottish Mental Survey. Intelligence, 28, 49–55. Diggle, P. J., & Kenward, M. G. (1994). Informative drop-out in longitudinal data analysis. Applied Statistics, 43, 49–93.

354

JOHN J. MCARDLE

Donaldson, G., & Horn, J. L. (1992). Age, cohort, and time developmental muddles: Easy in practice, hard in theory. Experimental Aging Research, 18, 213–222. Duncan, T. E., Duncan, S. C., Strycker, L. A., Li, F., & Alpert, A. (1998). An introduction to latent variable growth curve modeling: Concepts, issues, and applications. Mahwah, NJ: Erlbaum. Ferrer, E., Hamagami, F., & McArdle, J. J. (2004). Modeling latent growth curves with incomplete data using different types of structural equation modeling and multilevel software. Structural Equation Modeling, 11, 452–483. Goldstein, H. (1995). Multilevel statistical models (2nd Ed.). New York: Oxford Press. Goulet, L. R., & Baltes, P. B. (Eds) (1970). Life-span psychology: Research and theory. New York: Academic Press. Hamagami, F. (1998). A developmental-based item factor analysis. In: J. J. McArdle & R. W. Woodcock (Eds), Human abilities in theory and practice (pp. 231–246). Mahwah, NJ: Erlbaum. ‘Hamagami, F., & McArdle, J. J. (2001). Advanced studies of individual differences linear dynamic models for longitudinal data analysis. In: G. A. Marcoulides & R. E. Schumacker (Eds), New developments and techniques in structural equation modeling (pp. 203–246). Mahwah, NJ: Lawrence Erlbaum Associates. Hedecker, D., & Gibbons, R. (1996). MIXOR: A computer program for mixed-effects ordinal regression analysis. Computer Methods and Programs in Biomedicine, 49, 157–176. Horn, J. L., & Donaldson, G. (1976). On the myth of intellectual decline in adulthood. American Psychologist, 31, 701–709. Horn, J. L., & McArdle, J. J. (1980). Perspectives on mathematical/statistical model building (MASMOB) in research on aging. In: L. W. Poon (Ed.), Aging in the 1980’s: Selected contemporary issues in the psychology of aging (pp. 503–541). Washington, DC: American Psychological Association. Jo¨reskog, K. G., & So¨rbom, D. (1979). Advances in factor analysis and structural equation models. Cambridge, UK: Abt Books. Kangas, J., & Bradway, K. P. (1971). Intelligence at middle age: A thirty-eight year follow-up. Developmental Psychology, 5(2), 333–337. Laird, N. M., & Ware, J. H. (1982). Random effects models for longitudinal data. Biometrics, 38, 963–974. Littell, R. C., Miliken, G. A., Stoup, W. W., & Wolfinger, R. D. (1996). SAS system for mixed models. Cary, NC: SAS institute. Little, R. J. A. (1995). Modeling the dropout mechanism in repeated-measures studies. Journal of the American Statistical Association, 90, 1112–1121. Little, R. T. A., & Rubin, D. B. (1987). Statistical analysis with missing data. New York: Wiley. Magnusson, D. (1995). Individual development: A holistic, integrated model. In: P. Moen, G. H. J. Elder & K. Luescher (Eds), Examining lives in context (pp. 19–60). Washington, DC: American Psychological Associations. McArdle, J. J. (1986). Latent variable growth within behavior genetic models. Behavior Genetics, 16(1), 163–200. McArdle, J. J. (1988). Dynamic but structural equation modeling of repeated measures data. In: J. R. Nesselroade & R. B. Cattell (Eds), Handbook of multivariate experimental psychology (pp. 561–614). New York: Plenum Press. McArdle, J. J. (1989). Structural modeling experiments using multiple growth functions. In: P. Ackerman, R. Kanfer & R. Cudeck (Eds), Learning and individual differences: Abilities, motivation, and methodology (pp. 71–117). Hillsdale, NJ: Erlbaum.


355

McArdle, J. J. (1990). Principle versus principals of structural factor analyses. Multivariate Behavioral Research, 25, 81–87. McArdle, J. J. (1994). Structural factor analysis experiments with incomplete data. Multivariate Behavioral Research, 29(4), 409–454. McArdle, J. J. (2001). A latent difference score approach to longitudinal dynamic structural analyses. In: R. Cudeck, S. du Toit & D. Sorbom (Eds), Structural equation modeling: Present and future (pp. 342–380). Lincolnwood, IL: Scientific Software International. McArdle, J. J., & Anderson, E. (1990). Latent variable growth models for research on aging. In: J. E. Birren & K. W. Schaie (Eds), The handbook of the psychology of aging (pp. 21–43). New York: Plenum Press. McArdle, J. J., & Bell, R. Q. (2000). Recent trends in modeling longitudinal data by latent growth curve methods. In: T. D. Little, K. U. Schnabel & J. Baumert (Eds), Modeling longitudinal and multiple-group data: Practical issues, applied approaches, and scientific examples (pp. 69–107). Mahwah, NJ: Erlbaum. McArdle, J. J., & Cattell, R. B. (1994). Structural equation models of factorial invariance in parallel proportional profiles and oblique confactor problems. Multivariate Behavioral Research, 29, 63–113. McArdle, J. J., & Epstein, D. B. (1987). Latent growth curves within developmental structural equation models. Child Development, 58(1), 110–133. McArdle, J. J., Ferrer-Caja, E., Hamagami, F., & Woodcock, R. W. (2002). Comparative longitudinal structural analyses of the growth and decline of multiple intellectual abilities over the life span. Developmental Psychology, 38, 115–142. McArdle, J. J., & Hamagami, E. (1992). Modeling incomplete longitudinal and cross-sectional data using latent growth structural models. Experimental Aging Research, 18(3), 145–166. McArdle, J. J., & Hamagami, F. (1996). Multilevel models from a multiple group structural equation perspective. In: G. Marcoulides & R. Schumacker (Eds), Advanced structural equation modeling techniques (pp. 89–124). Hillsdale, NJ: Erlbaum. McArdle, J. J., & Hamagami, F. (2001). Linear dynamic analyses of incomplete longitudinal data. In: L. Collins & A. Sayer (Eds), New methods for the analysis of change (pp. 137–176). Washington, DC: APA Press. McArdle, J. J., & Hamagami, F. (2003). Structural equation models for evaluating dynamic concepts within longitudinal twin analyses. Behavior Genetics, 33(3), 137–159. McArdle, J. J., & Hamagami, F. (2004). Methods for dynamic change hypotheses. In: K. van Montfort, J. Oud & A. Satorra (Eds), Recent developments in structural equation models (pp. 295–335). Dordrecht, The Netherlands: Kluwer Academic Publisher. McArdle, J. J., Hamagami, F., Jones, K., Jolesz, F., Kikinis, R., Spiro, A., et al. (2004). Structural modeling of dynamic changes in memory and brain structure using longitudinal data from the Normative Aging Study. Journals of Gerontology: Psychological Sciences, 59B, P294–P304. McArdle, J. J., Hamagami, F., Meredith, W., & Bradway, K. P. (2001). Modeling the dynamic hypotheses of Gf-Gc theory using longitudinal life-span data. Learning and Individual Differences, 12(2000), 53–79. McArdle, J. J., & Nesselroade, J. R. (1994). Using multivariate data to structure developmental change. In: S. H. Cohen & H. W. Reese (Eds), Life-span developmental psychology: Methodological contributions (pp. 223–267). Hillsdale, NJ: Lawrence Erlbaum Associates.

356

JOHN J. MCARDLE

McArdle, J. J., & Nesselroade, J. R. (2003). Growth curve analyses in contemporary psychological research. In: J. Schinka & W. Velicer (Eds), Comprehensive handbook of psychology, volume two: Research methods in psychology (pp. 447–480). New York: Pergamon Press. McArdle, J. J., & Prescott, C. A. (1992). Age-based construct validation using structural equation models. Experimental Aging Research, 18(3), 87–115. McArdle, J. J., & Prescott, C. A. (1997). Contemporary models for the biometric genetic analysis of intellectual abilities. In: D. P. Flanagan, J. L. Genshaft & P. L. Harrison (Eds), Contemporary intellectual assessment: Theories, tests, and issues (pp. 403–436). New York, NY: Guilford Publications. McArdle, J. J., & Woodcock, J. R. (1997). Expanding test-rest designs to include developmental time-lag components. Psychological Methods, 2(4), 403–435. Meredith, W., & Tisak, J. (1990). Latent curve analysis. Psychometrika, 55, 107–122. Metha, P. D., & West, S. G. (2000). Putting the individual back into individual growth curves. Psychological Methods, 5(1), 23–43. Mutheń, B. O., & Curran, P. (1997). General longitudinal modeling of individual differences in experimental designs: A latent variable framework for analysis and power estimation. Psychological Methods, 2, 371–402. Mutheń, B. O., & Mutheń, L. K. (1995). Tailoring psychometric techniques for epidemiological and clinical applications. Technical Report. Mutheń, B. O., & Mutheń, L. K. (2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882–891. Mutheń, L. K., & Mutheń, B. O. (2002). Mplus, the comprehensive modeling program for applied researchers users guide. Los Angeles, CA: Muthen & Muthen. Nagin, D. (1999). Analyzing developmental trajectories: Semi-parametric. Group-based approach. Psychological Methods, 4, 139–177. Neale, M. C., Boker, S. M., Xie, G., & Maes, H. H. (1999). Mx statistical modeling (5th ed.). Unpublished program manual, Virginia Institute of Psychiatric and Behavioral Genetics, Medical College of Virginia, Virginia Commonwealth University, Richmond, VA. Nesselroade, J. R. (2001). Paul B. Baltes: Gerontology with a life-span perspective. Contemporary Gerontology, 7(3), 75–77. Nesselroade, J. R., & Baltes, P. B. (Eds) (1979). Longitudinal research in the study of behavior and development. New York: Academic Press. Pinheiro, J. C., & Bates, D. M. (2000). Mixed-effects models in S and S-PLUS. New York: Springer. Pothoff, R. F., & Roy, S. N. (1964). A generalized multivariate analysis model useful especially for growth curve problems. Biometrics, 51, 313–326. Rao, C. R. (1958). Some statistical methods for the comparison of growth curves. Biometrics, 14, 1–17. Rogosa, D. (1978). Causal models in longitudinal research: Rationale, formulation, and interpretation. In: J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in human development: Design and analysis (pp. 263–302). New York, NY: Academic Press. Rogosa, D., & Willett, J. B. (1985). Understanding correlates of change by modeling individual differences in growth. Psychometrika, 50, 203–228. Schaie, K. W. (1965). A general model for the study of developmental problems. Psychological Bulletin, 64, 92–107.


357

Schaie, K. W., & Baltes, P. B. (1975). On sequential strategies in developmental research: Description or explanation? Human Development, 18, 384–390. Seber, G. A. F., & Wild, C. J. (1989). Nonlinear models. New York: Wiley. Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics, 24(4), 323–355. Singer, J. D., & Willett, J. (2003). Applied longitudinal data analysis. New York: Oxford University Press. Sliwinski, M. J., & Buschke, H. (1999). Cross-sectional and longitudinal relationships among age, cognition, and processing speed. Psychology and Aging, 14, 18–33. Sullivan, E. V., Rosenbloom, M. J., Lim, K. O., & Pfefferman, A. (2000). Longitudinal changes in cognition, gait, balance in abstinent and relapsed alcoholic men: Relationships to changes in brain structure. Neuropsychology, 14(2), 178–188. Tucker, L. R. (1958). Determination of parameters of a functional relation by factor analysis. Psychometrika, 23, 19–23. Tucker, L. R. (1966). Learning theory and multivariate experiment: Illustration by determination of generalized learning curves. In: R. B. Cattell (Ed.), Handbook of multivariate experimental psychology. Chicago, IL: Rand McNally. Verbeke, G., & Molenberghs, G. (2000). Linear mixed models for longitudinal data. New York, NY: Springer. Verbeke, G., Molenberghs, G., Krickeberg, K., & Fienberg, S. (Eds) (2000). Linear mixed models for longitudinal data. New York: Springer Verlag. Wohlwill, J. F. (1973). The study of behavioral development. New York, NY: Academic Press. Willett, J. B., & Sayer, A. G. (1994). Using covariance structure analysis to detect correlates and predictors of individual change over time. Psychological Bulletin, 116, 363–381. Wishart, J. (1934). Bibliography of agricultural statistics 1931–1933. Journal of the Royal Statistical Society, 1, 94–106. Wishart, J. (1938). Growth rate determinations in nutrition studies with the bacon pig, and their analyses. Biometrika, 30, 16–28. Zeger, S. L., & Harlow, S. D. (1987). Mathematical models from laws of growth to tools for biologic analysis: Fifty years of growth. Growth, 51, 1–21.


358

AFTERTHOUGHTS

359


360

INCITATIONS FOR INTERDISCIPLINARITY IN LIFE COURSE RESEARCH Rene´ Levy, Paolo Ghisletta, Jean-Marie Le Goff, Dario Spini and Eric Widmer HOUSING THE HARVEST Having gone through this volume, a critical reader might come to the conclusion that interdisciplinarity can be found more easily between the contributions than within them (even though several of them address it directly, e.g., Settersten & Gannon; Mortimer et al.).1 However, the contributors share the common belief that studying humans’ unfolding lives in a web of complicated interactions within their changing contexts requires the adoption of an interdisciplinary research paradigm. To be sure, the life-span/life course research traditions stemming from disciplines such as sociology, psychology, social psychology and demography certainly have allowed scholars to answer some key questions germane to this field (Baltes, Lindenberger & Staudinger, in press; Elder, 1998). The empirical evidence accumulated over the years contributed heavily to the validity of the enterprise represented by life course research (Baltes, Reese, & Nesselroade, 1977). The development toward interdisciplinarity needs, however, not only solid disciplinary foundations and the shared wish to cooperate, but also


361

362

RENE´ LEVY ET AL.

hard and time-consuming work in interdisciplinary groups to progress concretely in this direction, possibly along the three lines sketched out in our introduction: constructing theoretical bridges between disciplinary approaches, building on common concepts that help describe and analyze life courses, and working on transversal substantive themes. In this final chapter, the editors take up the four thematic groups of contributions, agency and structure, transitions, biographical reconstruction and methodological innovations, in order to scan them for elements that seem instrumental for further building up interdisciplinarity in life course research.

INDIVIDUAL AGENCY WITHIN SOCIAL STRUCTURE, AND STRUCTURAL AGENCY Defining agency as the capacity to act intentionally, planfully and reflexively in a temporal and biographical mode (Marshall, Settersten & Gannon) points to the fact that the enactment of agency is, among other things, a cognitive and emotional process, informed by the social environment, and unfolding through time.2 Following Marshall’s argument, agency should be distinguished from various related concepts, such as useful resources for action, social action itself, intentions that motivate behavior, the social and physical structuring of choices and unexplained variance. Structure, on the other hand, is usually defined as the set of social constraints and opportunities within which individual agency plays out. Contributions to this volume show that structures and agency form a system of interrelated elements rather than a chain of distinct factors with a clear causal ordering, for instance, from the macrosocial down to psychobiological levels, or inversely, from cognitive skills upward to social structures. We conclude from these contributions that social structures have agency of their own, that agency has structures of its own, and that both should be jointly looked at when studying life course issues. To start with the first of these two combinations: social structures, in some sense, have agency, too. States, as well as firms and other institutions whose functioning has direct implications for individual life courses, are shaped by human beings, be they simple members, clients or institutional decision-makers, with purposes, goals, and agency. The latter, institutional entrepreneurs in the sense of Eisenstadt (1968), put up programs to instantiate their views about the life course of their citizens, patients or employees.

Incitations for Interdisciplinarity in Life Course Research

363

The agency of institutional entrepreneurs is transformed into structures that both bound and orient the agency of other actors. A promising line of inquiry in life course research concerns the way in which various institutional actors participate in the structuring of the life course in democratic societies. European sociology, especially in Germany (for instance, Mayer & Mu¨ller, 1986; Mayer & Schoepflin, 1989) has much emphasized the importance of the modern State (with all its agencies, including the education system and the law) and the development of the market economy in this regard, leading to trends of standardization, individualization and sexual typification of individual trajectories (Kohli, 1985, 1986; Kru¨ger & Levy, 2001). Social historians show that these changes responded in many cases to instrumental thinking and action of political leaders, such as Bismarck in Germany, Lord Beveridge in the United Kingdom, or Roosevelt with the New Deal (Elder, 1974). Structural changes therefore are often led by institutional agency, a fairly complex and constrained process that does not only depend, in democratic societies, on a few institutional entrepreneurs, but also on various political forces, including social movements – that can themselves be analyzed in a life course perspective. Following Berger and Luckmann’s (1966) line of thought, institutional agency may then be objectivated, i.e., become a constraining social fact, which may later be interiorized by individuals who incorporate it when making life course plans. On the other hand, agency has structures of its own, both psychological and social. Narratives usually have a temporal, causal and thematic coherence. Goals and motives are correlated with psychological traits (McAdams). Life insight of individuals, that is, how to act for one’s own good and the good of others, is deeply intertwined with personality and cognition abilities. Agency as a cognitive process is therefore strongly shaped by psychological characteristics, such as resources of various kinds, as studies of the effects of personality on memories show (Perrig-Chiello & Perrig). This capacity to act or react, which developmental psychologists sometimes refer to as resilience or coping whereas sociologists may rather speak of planful competence (as Settersten and Gannon remind us), displays various complementary strategies, such as the selection–optimization–compensation or SOC strategies found in relation with the aging process (Baltes & Baltes, 1990). Agency seen as a cognitive process matters a lot for social behavior and social structures throughout the life course. For instance, as Mortimer et al. show, adolescent goals are transformed into part-time work habits, which later unfold in distinct ways of entering the job market. Work on the effect of early childbearing provides similar results. Contrary to social deterministic approaches, like the statement that ‘‘The girl who has an illegitimate

364

RENE´ LEVY ET AL.

child at the age of 16 suddenly has 90 percent of her life’s script written for her’’ (Campbell, 1968), Furstenberg points to the fact that non-normative events produce diverse consequences over the life course depending on mental health, cognitive skills, motivation and flexibility of individuals. Settersten and Gannon provide examples from various phases of the life course (childhood and adolescence, early adulthood, old age, etc.) in which social constraints are dealt with in creative ways by individuals. Thus agency, as a cognitive process, makes a difference for individual lives. Agency, however, is not only constrained or enhanced by psychological and macrosocial factors, but also by microsocial factors unfolding through individual time. Rather than considering agency as a black box, not to be opened in empirical analyses, some sociologists propose to see it as path dependent in a specific sense: narratives, goals, motivations and cognitive abilities are shaped to a large extent by the actual trajectories themselves (Abbott, 1992). In this regard, the long-time experience of social psychologists in dealing with identity formation in social contexts is particularly useful. In a second perspective (Emler), narratives, life wisdom and goals are not constructed by isolated actors: they emerge in connection with narratives, life wisdom and goals of significant others, in social networks in which various agencies shape each other. Agency is not only an individual phenomenon, but also a collective one (Settersten and Gannon): studying coagency between connected people (especially in gendered relationships) is addressing the central issue of linked lives (Elder, 1996) in a novel and interdisciplinary way. More generally, a fruitful option for dealing with the agency–structure debate suggested by this volume is to consider both of them as variables in interaction rather than as main effects to be statistically controlled for. The concept of ambivalence (Lu¨scher) is especially helpful in this respect. Lu¨scher suggests four basic ways of experiencing and dealing with intergenerational ambivalence within family relationships, with specific couplings of agency (the subjective dimension of the model) and structure (its institutional dimension): solidarity, emancipation, captivation and atomization. At one end of the spectrum, solidarity is defined as a situation in which family members feel subjectively committed to the maintenance of institutionalized patterns of help and relationships, while, at the other end of the spectrum, atomization corresponds to cases in which family cohesiveness is neither subjectively nor institutionally assured. According to Lu¨scher, we may expect the various types of interaction between agency and structure to be useful for the understanding of turning points in the life course, such as


365

the transition to parenthood, which are associated with both new constraints and new roles. Specific life stages, such as old age with its increasing tension between autonomy and dependency, may be especially conducive to ambivalence. This may be particularly true if we agree with Settersten and Gannon’s statement that the life course has become at a time more standardized and more destandardized in recent decades, a situation that may lead to increased ambivalence throughout the life course. Because of the emerging acknowledgement of the intricacy of the relationship between agency and social structure as variables, there is a growing need for interdisciplinary approaches that may help to capture the interplay between agency as a cognitive process with its own set of psychological and microsocial constraints, and social structures, which are partly the result of institutional agency. This implies a stronger partnership between life-span psychology, social psychology, social demography and life course sociology. As suggested by several contributions, sociologists and social demographers often overlook the role of personality traits and identity narratives when explaining individual trajectories, while psychologists, developmental or social, often disregard contextual factors, in particular those stemming from the historical context in which individuals are embedded. The contributions to this volume suggest some ways in which sociologists may unfold the richness and intricacies of the ‘‘homo psychologicus’’ and include its elements in their explanatory models of the life course. Likewise, it suggests how life-span psychologists may include more sophisticated and refined conceptualizations and measures of the social contexts in their research.

TRANSITIONS AS SOCIAL AND PSYCHOLOGICAL ‘‘ANALYZERS’’ Life courses go through and are marked by transitions. Transitions and the stages between them define each other mutually, not only in the formal sense of stages being bounded by transitions and transitions being inserted between stages, but also in a more substantive sense: most transitions are being prepared, anticipated and influenced in many ways during the phases preceding them, and influence in turn the phases that follow. Transitions may be ‘‘normative’’ in psychological language or, as sociologists would prefer to say, modal or statistically predominant, they may also be socially considered to be desirable, possibly indispensable, or they may be non-normative or even deviant with respect to social norms, as in the case of the teenage

366

RENE´ LEVY ET AL.

mothers and divorced women discussed by Furstenberg. Some transitions belong to the ‘‘normal’’ course of peoples’ lives and contribute to their regular ‘‘progress’’, while others may throw a life off its rails and become turning points, biographical accidents provoking life course bifurcations. Some transitions are relatively smooth, making little ‘‘unrest’’ for the persons directly or indirectly concerned, others, even if ‘‘normative’’ in the above sense, may have considerable and partly non-anticipated consequences (such as the transition from a cohabiting couple to a family with a first child; Widmer, Levy, Pollien, Hammer, & Gauthier, 2003b; Widmer, Kellerhals, & Levy, 2005; Moen & Han, 2001). Even if it is somewhat reductionist, life course transitions may be seen as moments of important change, in contrast to the periods between them that are rather marked by stability, especially if we think of social–psychological and psychological correlates. These should not be underestimated because, as we know, individuals are not only socialized actors with ‘‘internal’’ cultural competencies, they are also field- or context dependent in many respects (identity, self-esteem, etc.), even if these dependencies may themselves vary within (see, for example, Kernis & Goldman, 2003; Nesselroade, 1991) and between individuals, and also across life phases. For this reason, transitions are likely to be in many respects more critical for the person than stable periods.3 As a person’s situation changes, he or she finds him- or herself in new conditions that are different from the preceding ones, not only in terms of being more or less advantageous, more or less restrictive for spontaneous initiatives, etc., but also in terms of their quality or nature. In a sociological perspective, this is reason enough to postulate a series of consequences that are not only of a sociological order, but concern also the levels of social-psychological and psychological analysis. The following four paragraphs are an attempt at formulating interdisciplinary hypotheses along this line. A first, highly probable corollary concerns identity, for oneself as well as for others, as several contributors remind us (especially Marshall and Emler): after a transition, a person will typically assume other roles, respond to other expectancies, have other rights and duties and interact with other people on other terms than before. According to the number and importance of the role shifts implied in a transition, this change will be more unsettling or even dramatic, or less. It will affect the person’s self-image, the representations others have of him or her and also her or his actively performed self-presentation to others. Identity may appear as less settled during a transition and more put into question, open to various directions and influences, also to voluntaristic self-influence.


367

Second, taking up one of Kohli’s (1985, 1986) theses, we may ask ourselves whether on the individual level transitions are also moments of heightened ‘‘biographization’’, i.e., moments in which the transiting lifepassengers have an increased sense of their being themselves the constructors of their life. It is, of course, also easy to figure out the contrary: transitions that rather restrict opportunities of action are more apt at eliciting sentiments of being passively channeled rather than actively piloting one’s life. This means at any rate the question of agency will be of particular relevance in transitions, as particularly well highlighted by Furstenberg’s comments on the variability of the outcomes he finds for non-normative transitions, but also by the analyses of Mortimer et al. Do we know enough about variations of persons’ control beliefs across major and minor biographical transitions, and about their consequences? A third, more general hypothetical corollary can be seen as an extension of the first one: it is more than plausible that throughout the life span, lifelong socialization is not continuous but rather rhythmical, being boosted by life course transitions and slowing down during the phases in-between, because in transitions familiar contexts of reference loose relevance and are replaced by new, less well-known ones. One may extend this argument to integrate the apparently contradicting ideas about when major phases of socialization take place in a lifetime: birth and the very first years of life represent the first biographical transition of a human being, the person’s first entry into a social field,4 and this field’s active and passive exploration while being in a particularly fragile – and therefore subjectively highly significant, affectively ‘‘mobilizing’’ – situation. Consequently, the socialization taking place in these circumstances is bound to be particularly impressive and rich in consequences. In comparison, later spells of socialization, e.g., in adolescence, concern an already (at least partly) structured, i.e., socialized person and need considerable social and psychological weight to supersede or basically relativize the elements already acquired in prior socialization processes; moreover, it concerns an individual that is already ‘‘biographically constructed’’ and becomes, at least potentially, an active and critical participant in her/his own socialization. We may see these elements as a background to some of Furstenberg’s arguments, based on the idea that earlier socialization may be part of selection processes (e.g., problematic socialization outcomes increase peoples’ chances of ending up in out-of-schedule situations) that create some of the unobserved variation in results about negative consequences of non-normative transitions, leading to simplistic or at least hasty conclusions about the negative consequences of such non-normative transitions in women’s or children’s life courses.

368

RENE´ LEVY ET AL.

A fourth corollary of transitions is related to the management of the change they imply. Transitions will be envisaged by the concerned persons with more apprehension than the stable periods between them, possibly also with more ambivalence (Lu¨scher) because they represent – to varying degrees, of course – potential risks in a person’s biography. The ways actors themselves and their social environment handle these risks and the subjective insecurity they entail is an obviously important theme for life course research; this theme can be conceptualized in terms of individual and collective adaptation or coping. There may be a vast array of forms and resources for coping related to diverse transitions, some of them specialized, some of them of a general nature: rituals, other forms of dramatization (including private ones), emergence of professionalized transition specialists (gate keepers as well as transition helpers), problem-solving literature for both professionals and lay persons, repair institutions (such as hospitals, rehabilitation clinics, training institutions for vocational reorientation, etc.), but also general resources and styles of coping that are not situation-specific. All these arguments point to the fact that life course transitions are not so much instantaneous moments of switching from one situation to another – Bird and Kru¨ger warn us against a ‘‘guillotine-like perception of transitions’’ – but rather processes that may be of considerable duration and complexity. Transitions, if they are not triggered by unexpected events, are mostly prepared for and subject to anticipation. They may be composed of several differentiated changes, and can imprint their reality on the person’s everyday life progressively. All of this does not happen in a tick, but takes biographical time. This reason is enough in itself to justify Elder’s (1998) insistence on the necessity to consider each transition as a series of minitransitions or decisional moments; Levinson’s (1990) concept of transitional periods may often be more appropriate than the simple term of transition; the same holds a fortiori for the term of event. Another important aspect is the way in which transitions are embedded in the whole trajectory of a person: what are the factors that trigger and modulate a transition to begin with, and what consequences do transitions and their outcomes have for the subsequent trajectories? How can we theorize the oft-cited cumulativity of life course developments? Let us mention just one example. According to research in several – especially European – countries, the transition from the situation of a cohabiting couple without children (pre-child phase) to the one of a family with a pre-school child is quite systematically accompanied by a switch of the couple to a more traditionally gendered task organization, relatively egalitarian convictions of the partners notwithstanding (Li & Currie, 1992; Born, Kru¨ger, & Lorenz-Meyer, 1996;


369

Kalicki, Fthenakis, Peitz, & Engfer, 1998; Kalicki, Fthenakis, & Peitz, 1999; Widmer, Kellerhals, & Levy, 2003a; Widmer et al., 2003b). This ‘‘retraditionalization’’ of the families’ internal structure seems to be quite resilient in later family phases and is not strongly influenced by the female partner’s degree of resuming paid work. It may well be that it is precisely in analyzing closely how transitions are initiated and produced, and how they produce in turn their consequences, that we will be able to better understand how the agency-in-structure vision advocated by Settersten and Gannon can be more finely modeled theoretically. Several contributions rightly underline the variability and complexity of life courses and specific transitions (especially Furstenberg and Mortimer et al.) and of diverse factors to be uncovered behind this variability (Mortimer et al., Marshall). Bird and Kru¨ger remind us with a powerful argumentation that matters are more complex than a merely sequential conceptualization might suggest, because the complexity of transitions is not restricted to this ‘‘linear’’ aspect. This is what they call ‘‘inline’’ transitions, which have to be completed by adding to the overall picture also ‘‘competing’’ and ‘‘coupled’’ transitions. They warn us usefully against several kinds of substantively inadequate complexity reductions that lie in wait for researchers. We have not only to take into account the fact that most transitions are not just events and that all three types of transitions may occur together, but we should also resist the simplifying ‘‘offers’’ suggested by some easygoing technical terms, such as formal status definitions that may not fully coincide with the practical reality of persons’ lives (see their examples of being married while living alone), or technical terms like event history analysis. An aspect that has long been at the core of life course analysis is what has been generally called ‘‘the timing of events’’, including the precise timing of important events and transitions as well as the duration of the phases between them, and often also the normative schedules concerning them (age norms). Furstenberg shows convincingly that we should be much more circumspect about the social and psychological meaning of transitions being out-of-time or out-of-order. The fact that a life course does not replicate the ‘‘normative’’ pattern, when such a pattern exists, is ostensibly not enough in itself to diagnose a problem for the future of the non-respectful life course passengers. Furstenberg considers a series of additional conditions that are likely to play a crucial role for such a situation to be socially and subjectively problematic – or on the contrary enhancing, by way of mobilizing latent potentials and resources. This is certainly a research area in need of further conceptual refinement as well as interdisciplinary treatment, and touching directly on agency, for that matter.

370

RENE´ LEVY ET AL.

What about transitions in a psychological or social–psychological perspective? Stage models, which would seem to be prima vista candidates for psychological equivalents of social transitions, however prominent they have been in the developmental psychologies of authors like Piaget, Kohlberg or Erikson, are not supported by research on cognitive functioning. The topic may be more promising with respect to identity changes, in line with our above hypotheses. The weak or non-existent empirical basis of psychological stage models is also one of the main reason for the absence of stages in Baltes’ SOC model of life-span development (see next section). This situation on the side of life-span psychology contrasts strangely with many life course sociologists’ stressing the timing of life events. The idea of more or less abrupt endogenous changes is put into question in a significant part of lifespan psychology in favor of a gradual view of age-related changes. In some respects similarly, life course sociology shows a strong commitment to considering age as the basic dimension for the study of life-long development, even if it seems to have some difficulties to conceptualize its meaning (Settersten & Mayer, 1997); however, more than psychology, sociological approaches insist strongly on events or transitions. Can these views be brought together? Maybe a renewed attention to transitions and their varying (and not necessarily close) relationship with age is in order from both the psychological and the sociological side, developing a stand that is less ‘‘naturally’’ and immediately oriented toward age (without neglecting it) and theoretically more outspoken about what can provoke change. We shall push forward on this tread in the section about interdisciplinarity.

PERSONALITY, BIOGRAPHICAL RECONSTRUCTIONS AND THE LIFE COURSE: TOWARD A SYSTEMIC AND DYNAMIC APPROACH Personality and Identity Across the Life Course As already mentioned, the relationship between agency and structure is a central theme of the sociological approach to the life course. However, as often underlined by sociologists, psychologists tend to neglect the importance of the structural components of the life course. The inverse is also true; psychologists often consider the theories of sociologists, which usually insist on the motivation and goal orientation of the individual across the life course, as oversimplifications with respect to their disciplinary knowledge.


371

Recent advances in the psychological and psychosocial study of the individual across the life span are illustrated by several chapters of this volume and have important implications for the development of a truly interdisciplinary approach to the life course. Putting into perspective different aspects of such developments in personality and identity theory will enable us to better articulate the different contributions of this volume.

Personality as a Framework for Studying Individuals across Time In the framework of personality theory, there is some consensus about the important dimensions of personality that must be distinguished and related with each other. However, there are also fundamental debates, especially around two questions that have to be tackled seriously by life course theorists. The first concerns the interplay of biology and personality and the second the question of the degree to which personality interacts with the social environment and with trajectories in different life domains. Since the seminal work of Allport (1937), personality psychology has been broadly defined as the scientific study of the individual person, and personality development across the life span has been a central theme. Theories of personality development across the life span have been proposed on the basis of the psychodynamic tradition (Sigmund Freud & Carl Gustav Jung, for example) and developed in an offspring known as life-cycle psychology (Erikson, 1963, 1968; Havighurst, 1972; Levinson, 1978, 1996; Neugarten, 1977). However, these theoretical frameworks have been repeatedly criticized for lacking empirical support to their theoretical claims. More recently, alternative systemic models (Hooker & McAdams, 2003; McAdams, 1996; McCrae & Costa, 2003; Mischel & Shoda, 1998) have emerged and offer a synthesis of empirical results and theoretical developments. The model of McCrae and Costa (2003) is known as the Five-Factor Theory (FFT). This model makes the assumption that our personality is founded on basic tendencies, which are composed mainly by the Big Five dimensions of personality (traits of neuroticism, extraversion, openness, agreeableness and conscientiousness), but also by sexual orientations, some cognitive abilities and artistic talents. These basic traits appear to be fairly stable across the life course, influenced mainly by biological factors (some changes appear but are described as predictably linked to age; see Srivatava, John, Gosling, & Potter, 2003), and to influence the other two components of personality across the life course, namely, the characteristic adaptations – personal strivings, attitudes, worldviews, strategies or processes of coping

372

RENE´ LEVY ET AL.

and adaptation (Baltes & Baltes, 1990; Brandtsta¨ter, Krampen, & Heil, 1993; Heckhausen, 1999) – and the self-concept system (self-schemas, identity, personal myths, life narratives). Contrary to the basic tendencies, these two elements are based on learning and are subject to external influences like cultural norms and life events. Finally, we find the objective biography, which represents the main observable outcome of basic personality tendencies and characteristic adaptations. McAdams (1996 and this volume) proposes a comprehensive model of the person distinguishing three levels (traits, personal action constructs or characteristic adaptations, and the life story) that has been further elaborated in a more recent version by Hooker (2002) and Hooker and McAdams (2003). These authors’ new model, named the Six-Foci Model of personality (SFM), is based on three structural components and three process-related components (processes producing corresponding structure components). The three structural components correspond to the three levels of personality already described by McAdams (1996). The corresponding processes are respectively states (emotions, moods, hunger, fatigue, anxiousness), self-regulatory processes (especially primary and secondary control; see Schulz & Heckhausen, 1996), and processes of self-narrating (remembering, reminiscence and storytelling). This new systemic model of personality, in complement with the FFT model, allows for a heuristic description of the aging person and focuses researchers’ attention on the self and its relations to contents (mainly present-oriented), goals (mainly future-oriented) and reconstructions of biography (mainly past-oriented).

Personality, Identity and the Social Environment Some of the main issues tackled in this volume concern, on the one hand, relationships between different areas of personality and identity, and on the other, how dynamics in personality and identity are related to dynamics in the social environment. The relations between the different components of the personality and identity systems are very complex; FFT and SFM models have different views in this respect. FFT postulates a causal chain of influence starting from biology and traits, mediated by the experiences and learning of the individual across the life span, which finally results in its biographical trajectory. Clearly, here, differences in traits between individuals are thought to influence the other dimensions of personality (identity and characteristic adaptations). Perrig-Chiello and Perrig exemplify this kind of model; they argue that the relationships between well-being and


373

autobiographical recollection or episodic memory are indeed influenced by personality traits like extraversion, conscientiousness and neuroticism, as predicted by FFT. However, as in other studies, the direction and strength of this influence of traits, as well as its variability across domains of the personality structure, remains an open question (Mischel, 2004). Moreover, in order to evaluate more precisely the relationships between life trajectories and personality or identity structures and process, two additional issues need to be developed in future research. The first is the necessity to combine different methodological strategies for life course research (nomothetic– idiographic; qualitative–quantitative, individual versus aggregated hierarchical data, etc.), with special attention to longitudinal designs. The second is the need to articulate different levels of analysis in order to better understand the relationships between the ‘‘objective’’ and ‘‘subjective’’ biography and between the individual, relational and collective levels. In Perrig-Chiello and Perrig, for example, transitions and life events are recorded on the basis of memory, and the social context is simplified to a small number of general variables (mostly gender and education). Sociologists and social demographers would wish to include more detailed information on the social position of individuals and on the institutionally monitored life transitions. Life trajectories cannot be assessed only with individuals’ memory-based recollection as autobiographical memory, even in such areas as job history or parents’ occupations (see Scott & Alwin, 1998), because they are subject to memory biases and to well-known processes of reconstruction that Perrig-Chiello and Perrig interestingly also develop in their chapter. The chapter by Emler shows that the relationships between identity changes and life transitions are indeed complex and should be modeled more explicitly. On the basis of social-psychological research, he shows that identity development does not correspond to a general pattern of qualitatively different hierarchical stages across the life span as postulated in the Piagetian and in particular in the life-cycle tradition. His results indicate that self-categorization processes are related to the social environment in which they take place. Consequently, the content and processes related to the self are primarily associated with time and location, two central dimensions of the life course approach (Settersten, 1999). Emler shows also that social identity is related to social relationships and locations, and that changes in social identity (related to life transitions or turning points) are often followed or anticipated by geographical moves leading to changes in the social participation of individuals. This systemic relationship between social identity (defined and constructed through interaction with relevant others in different settings and types of

374

RENE´ LEVY ET AL.

relationships) and personality on the one hand and the social and cultural context on the other is also at the heart of McAdams and colleagues’ work, as illustrated in their research about generativity in midlife. Generativity is a central theme of identity, which is in turn related to a wide range of engagements within society. It is in direct interaction with the social environment, in which individuals continuously reformulate their identity stories. The life-stories orientation reminds us of an essential element in the way individuals and life stories are related: meanings must not be ignored when considering how the individuals interact with their social environment. It is through meanings that individuals are related to their social networks, to their engagements in different life spheres and to their life trajectories (past, present and future). It is also through shared meanings or social representations that individuals are related to ideologies and their sociocultural context (see, for example, the observations of Marshall on how different institutions may create different stories and ideologies about death and dying; see also Chryssochoou, 2003). As such, the two socialized components of personality (characteristic adaptations and identity) are in constant interaction with the social environment and biography. If individuals are indeed active in the construction of the meaning they attribute to their life, one conclusion of the chapters discussed here is that the social context participates directly in the definition of the personality development of individuals by way of communication processes (socialization, social influence, conformity) and through the effects of non-normative events that induce specific social–psychological processes of adaptation and coping (e.g., in the sphere of health, Taylor & Brown, 1988; Taylor, Kemeny, Reed, Bower, & Gruenewald, 2000). In this regard, social psychologists in some way depict a more complex individual life course than the motivated actor often seemingly referred to by sociologists. To sum up, personality and social psychologists have developed models like FFT and SFM that share the idea that three components should be considered in any global comprehension of personal development. These include, at the structural level, personality traits (relatively stable and structured early in the life course), psychosocial regulations (goals, secondary control, coping processes, etc.) and identity (life story, self-categorization processes, etc.). The relationships among these three components (and their associated processes), and especially their relationships with the social structure and the structure of the life course, are still in question. However, personality psychology and social psychology have now developed a comprehensive model that should be useful for the interdisciplinary life course perspective.


375

METHODOLOGICAL AND DATA-ANALYTICAL APPROACHES Collection of Life Course Data Following Scott and Alwin (1998), we may distinguish two important issues regarding the question of life-history data. The first is about what should be measured, i.e., what kind of data to collect, while the second is about how to do these measurements; it refers to data design. Concerning the first, Scott and Alwin distinguish between three kinds of measures in life course research: events, experiences and meanings. The term of event refers to the collection of event data in different life domains (e.g., family, professional career) with the aim to analyze chronologies, sequences or interactions between life events. This corresponds to the definition of a life history given by Elder (1992): ‘‘a lifetime chronology of events and activities that typically and variably combine data records on education, work-life, family and residence’’ (p. 1122). Events are typically investigated by demographers (Billari; Oris & Ritschard), to some extent also by sociologists. The second approach measures cumulated experiences during an individual’s past in order to analyze his/her present situation or his/her expectations about the future. For example, the work experience of an adolescent is considered as a predictor of the transition to the labor market or to high school, as shown by Mortimer et al. The third kind of measure focuses on the meaning or the evaluation a person attributes to her/his past trajectory. This past can be contrasted with the present situation or future expectations. The approach to narrative identity proposed by McAdams exemplifies research collecting meanings of life courses. Some contributions to this volume, and more generally most examples in the literature, suggest that a fourth type of measurement in the life course approach has to be considered, referring to the context or the institutionalization of life courses. Data on context can be succinct information used as a complement to life-event data and helps to depict the structure of constraints and opportunities in persons’ social environment. This kind of information is illustrated by Bird and Kru¨ger by their taking into account the legislation on motherhood leaves in Germany that influences the duration of career interruptions. Context data include also interviews of life course agents, i.e., persons who are present or give support at a specific stage or transition in the life course, as developed by Marshall’s analysis of the transition to death. One methodological difficulty is the integration of these contextual informations and life course histories.

376

RENE´ LEVY ET AL.

Concerning the second issue, two types of data collection design are currently used in life course research. The first one is the cross-sectional collection of retrospective data. The reconstruction of individual life courses with archive or administrative data forms a first subtype of retrospective data. For historians interested in somewhat remote periods, this is practically the only possible strategy; its main disadvantage is that these data were not originally collected for research purposes (Oris & Ritschard); researchers using them have no other choice than to accept them – and their limitations – as ‘‘given’’ (Billari). The second subtype of retrospective data is generated by surveys in which persons are interviewed about events or experiences in their past. The quality of data collected in retrospective surveys depends strongly on the memory of respondents. Tools like life event history calendars minimize the potential biases (Freedman, Thornton, Camburn, Alwin, & Young-DeMarco, 1988; Belli, 1998). Retrospective data are often interesting to collect in order to analyze the influence of a historical event or of a specific historical period (economic crisis, war, etc.) on life courses. However, various studies on the impact of a historical event on trajectories show generally that this impact varies according to the situation or the stage in the life course where persons are interviewed. A range of samples allowing to compare the impact of such a historical event on trajectories in different generations, i.e., of people born at different dates can be used in order to analyze this impact in terms of cohort effects (Ryder, 1965). This strategy of combining the choice of design and of sample characteristics is also appealing to analyze the effects of contextual changes (Bird & Kru¨ger). However, retrospective data present several limitations, the main one being that only factual data can be collected. This implies that only information on events and experiences can be analyzed. The other longitudinal design is the repeated collection of prospective data. In psychology, this type of methodology is usually referred to as longitudinal design, in other disciplines of the social sciences as panel design. In this case, the same measures are applied to the same sample of persons at various points in time. This kind of data collection allows taking into account current experiences as well as intentions about the future. This enables life course researchers to confront intentions with their realization in subsequent waves of the survey, or to investigate how expectations evolve across the life course (Nurmi & Salmela-Aro, 2000). Panel surveys are especially interesting for research centered on agency. Several defaults of the panel design have also been underlined, especially the attrition of cases across succesive vawes of the panel or the risk to break time series if questions or experimental modalities are changed underway (MacArdle).


377

What might then be the best strategy for data collection in interdisciplinary research? As mentioned above, events are the research topic favored by demographers, historians, and, to a lesser extent, sociologists. Experience is a topic common to all social science disciplines, but it is probably most prominent in sociology and psychology. Meanings are a research topic of particular interest to social psychology, but also to sociology. The necessity to collect context data is more often invoked by demographers or sociologists than by psychologists. A preliminary condition of interdisciplinarity is then to combine measures of events, experiences, meanings and context. Moreover, a panel design allows to take into account each topic, especially when it is completed by a qualitative survey. This type of panel design is also especially interesting to analyze micromechanisms during a transition or a change in the life course. It should be noted that retrospective surveys allow also doing interdisciplinary work.

ANALYSIS OF LIFE COURSE DATA Recently, certain data-analytical tools have come to play a critical role in advancing knowledge about the life course. If integrated and combined, these tools promise further interesting applications. We can identify four general families of such tools that we consider especially helpful to study the life course in an interdisciplinary perspective. While some of these techniques have developed in parallel in different disciplines, others did so mainly within a given discipline and may hence be less known by scholars of other research fields. Life course scientists have a vital interest to get familiarized with these methods even though some of them may seem rather exotic at first sight. Some of them are directly treated by the contributions in this volume, others are only hinted at, so it may be helpful to give a structured overview of this rapidly evolving array. The first set of data-analytical tools we would like to mention falls under the label of structural equation modeling (SEM) and has been adopted especially in psychological research. This set of techniques aims at explaining the interrelationships observed among a set of chosen variables, usually by the means of a correlation matrix. One of the major advantages of SEM is that the researcher has complete freedom as to how to represent the structure of the data. The researcher adopting SEM must define what the underlying structure accounting for the interrelationships within the data might be and translate that hypothesis (or series of hypotheses if several alternative specifications are formulated) into a testable and rejectable

378

RENE´ LEVY ET AL.

model, to be tested against the data at hand. This set of techniques further has the desirable property of partitioning the variance of single variables or indicators into a portion said to be common among the chosen variables, hence representing latent constructs that cannot be observed nor measured directly, but are thought to influence the measurable properties of the variables, and a portion said to be unique to each variable. Hence, measurement problems may often be circumvented with the application of SEM. Some promising advanced applications of SEM to the study of the life course are illustrated by McArdle’s contribution. There, changes between adjacent repeated measures of a set of variables assessed on the same individuals are defined as latent differences, so that occasion-specific measurement threats may be isolated and eliminated from the important information gathered on the individuals. One may then test the existence of variance in latent difference scores, which translates into interindividual differences in change. Once differential change is established, correlates and antecedents of change may be tested, one of the major goals of longitudinal research (Baltes & Nesselroade, 1979). A second set of techniques is that of event history or survival analysis (EHA), adopted most heavily in demography (Billari and Ritschard & Oris). The basic question motivating the application of EHA is what affects the probability of the occurrence at a given time of a specific event (e.g., marriage, birth of the first child, onset of a disease, death). Unlike SEM, a socalled survival model does not have to be specified by the researcher. The most popular example of EHA indeed is Cox’ proportional hazards model, in which a semi-parametric hazard function provides very reasonable estimates of the influences exerted by chosen covariates on the occurrence of the event scrutinized. Particularly appealing is the test of time constancy of each predictor (in practice unfortunately often overlooked). By introducing in the model not only chosen predictors, but also their interactions with the underlying time dimension, it is possible to test whether the potential effect of a predictor holds across time or is manifested only at certain time periods. Hence, what is believed to be an important predictor of the occurrence of an event can be tested for its temporal relevance, much in the vein of what Bird and Kru¨ger remind us, namely that life course scholars ought to pay close attention to the time dimension. Moreover, the regularity of time-ordered events may also be assessed with this set of techniques. A third set of techniques is that of multilevel models (MLM), also known as random effects, mixed effects, and hierarchical linear models. This set of techniques more than any other recognizes the possibility that the data under inspection are structured according to either pre-defined (and usually


379

hierarchical) organizations (such as households within neighborhoods and family members within households) or to configurations that were not planned, but nevertheless resulted because of empirical research contingencies (e.g., multistage sampling). Such situations do not meet the basic assumption validating results from ordinary linear regression that the units of observation are independent of each other, and ignoring this structure of the data usually results in biased standard errors of the parameters (Goldstein, 1987). MLM can hence be conceived as regression equations with not one source of variance (usually referred to as the errors, or the residuals), but several sources of variance, the sources themselves being organized according to the structure of the data (e.g., a first error may be associated with members within a household and a second with households within a neighborhood). Hence, variables at the individual and at the contextual levels can be modeled together, making sure their distinctions are properly respected and not artificially removed.5 This approach is particularly promising for enriching the ‘‘agency and structure’’ discussion previously presented. Indeed, variables measured at the individual (i.e., agentic) level can be analyzed in concomitance with variables assessed at the contextual (i.e., structural) level. Moreover, the so called cross-level interactions may be defined that allow for the estimation not only of main effects at the individual and contextual levels, but also of their interaction. It is this interaction between agentic and structural variables that most often motivates life course scholars. A fourth, promising analytical approach is represented by exploratory analyses of repeated measures data. These techniques are newer but are quickly gaining popularity in life course research, thanks to their capacity to address not only quantitatively but also qualitatively motivated questions. Ritschard and Oris discuss Markov transition models and longitudinal data mining. Both techniques are concerned with sequences of events that do not need to follow specific assumptions. Mathematical rules can be established to explain the probability of switching from one event to another and the effect exerted by chosen covariates on these switches. While Markov models are parametric, data mining is non-parametric and aims at deriving association rules among the most typical sequences and their frequencies. A typical practical application is that of online bookstores, where customers purchasing item A are notified of similar items bought by preceding customers who also purchased item A. Another promising longitudinal exploratory technique is that of optimal matching (Abbott, 1995), of which Bird and Kru¨ger remind us in their contribution. This technique, unlike EHA, is not focused on a specific event, but aims instead at producing typical sequences, i.e., in the case of life courses, longitudinal constellations of states,

RENE´ LEVY ET AL.

380

allowing to study whole trajectories. The assumptions of the states are minimal, so that their complexity may drive the synthesis (Widmer et al., 2003b). Through pairwise comparisons this inductive method computes the minimal distance separating each pair of individual trajectories available in the sample according to pre-defined ‘‘costs’’ needed to transform one trajectory into another. The resulting distance matrix may then be fed into a clustering procedure to obtain groups as a function of their shared types of trajectories. These four families of techniques6 have developed separately, with little or no links between them. In recent years, however, methodological advances have allowed to combine appealing features of these methods to benefit further from their applications (Billari, Ritschard & Oris, and McArdle). Examples include the combination of EHA and MLM (Ritschard & Oris) that allow for conditioning the probability of the occurrence of an event on information at different levels of the data organization. Similarly, the recent addition of SEM latent variables in EHA refines the measurement properties of predictors, so that their effects are less attenuated by unrelated variance such as error. SEM and MLM have also recently been combined to provide for another methodological synergy (Ghisletta & Lindenberger, 2004; Rovine & Molenaar, 2000). Here, the advantages of SEM with respect to measurement properties are joined with the power of MLM to disentangle effects stemming from different, hierarchically organized sources of variance. Methodological refinements have contributed much to the advancement of not only empirical but also theoretical knowledge about the life course. At the same time, the methodological and data-analytical techniques have advanced themselves, geared as they have become to address further theoretical questions raised by life course scholars. The combination of now wellestablished techniques and the quickly evolving field of linear and non-linear dynamical systems will further contribute to paving the road of life course research. We believe that a fundamental ingredient for successful life course research is the presence of continued synergetic communication between the different disciplines concerned. Good data-analytical tools have emerged in each discipline, and we are confident that yet better tools will be developed by combining discipline-specific existing techniques. These ameliorations are instrumental to acquire deeper knowledge of life course phenomena.

INTERDISCIPLINARITY After having sifted through the four sections of this volume with a view to common or interacting themes between its contributions, let us take up the


381

three axes we propose for developing interdisciplinarity in life course research, common concepts, theoretical bridges, and transversal themes – what elements can this volume contribute to each of these? We have to realize that common concepts and transversal themes are closer to each other than both are to theoretical bridges, since we almost inevitably use concepts to refer to themes. This allows us to treat both aspects together. Common concepts have to develop from interdisciplinary work on transversal themes – they need not be identical from the outset, otherwise interdisciplinarity would be restricted to the rare instances where the same concept is used in more than one discipline (examples could be socialization or coping), and even if the words were not the same, interdisciplinary connections would boil down in such cases to simple terminological translation (as in the case of ‘‘lifespan’’ and ‘‘life course,’’ or of the different names used for some data-analytical methods). Conceptual comparison and elaboration become interesting if concepts are not the same between two or more disciplines, but substantively close enough to favor interdisciplinary exchange and elaboration, which is most likely to be fruitful when starting from common themes. Among the candidates for becoming common or transversal themes that have emerged in this volume, let us single out socialization, age, identity change and life course transitions, agency, coping, and gender. In varying configurations, each of these substantive areas promises for interdisciplinary collaboration to bring about mutual enrichment, greater strength of analytical grip, and more complete understanding of life course phenomena, especially if we start from the principle to look first at interdependencies rather than causal chains because the latter would fix a priori hierarchies of causation between the disciplines. Socialization is perhaps the most traditionally common area, at least between psychology, social psychology and sociology, because the three disciplines consider it to be one of the most basic processes of personal development; the particular perspectives developed by each of them is a crucial source of complementarity. The situation is different for age, the analytical status of which is clearly less obvious and more controversial. But then, interdisciplinary discussion of this situation should lead to spell out more explicitly the disciplinespecific perspectives and assumptions, facilitating to arrive at a more encompassing and articulate conceptualization of the phenomena concerning age and the related processes. As an illustration, let us elaborate somewhat further on this theme. In psychology, the influential article by Wohlwill (1970) pinpointed this problem. This author characterized the status of age as a convenient, descriptive and data-organizing tool that lacked however

382

RENE´ LEVY ET AL.

theoretical meaning. He urged scholars to investigate the aspects of behavior that might be lawfully related to age. Among the few examples embracing this deeper analysis of the variable of age we can cite the search for markers of biological age (or biomarkers; e.g., Anstey, Lord, & Smith, 1996), the opposition of chronological age to other time definitions that are theoretically better justified in the light of the phenomenon under analysis (e.g., time left to the onset of pre-clinical dementia when studying memory performance in older adults; Sliwinski, Hofer, Hall, & Buschke, 2003), and, from a more analytical point of view, transformations of age in relation to the variable considered in order to better understand the age-related mechanisms (e.g., McArdle, 1986). In much the same vein, sociologists Settersten and Mayer (1997) have advised, ‘‘chronological age itself is an ‘empty’ variable...it is whatever age presumably indexes that is thought to be important.’’ The major difference between psychological and sociological uses of age in the life course/life-span field is probably that in developmental psychology, it indicates changes implied in forms of physical and cognitive maturation and physical or physiological aging, i.e., biological age, whereas sociologists think in terms of social age, again with several specific meanings that are not always clearly distinguished, especially in the sense of age norms versus the more structural sense of specific roles or role sets that define the way individuals are embedded in the social world. In sum, life course scholars more than others are urged to move the age variable from the right side to the left side of the equation: age should not be used to explain behavior, but should itself become subject to analysis. A further hint at a sociological contribution to the substantive interpretation of age can be seen in Kohli’s (1985) discovery of biographical chronologization as a rather recent historical process, related to what Weber already analyzed as the growing bureaucratization and rationalization of modern societies. According to Kohli, the sequencing and temporal ordering of modern life courses that has taken place in the last 2–3 centuries is explained mainly by the structural changes brought about by modernization, reliance on age for the legal and procedural attribution of a series of rights and duties corresponding to what Weber called bureaucratical rationality (among such institutional innovations, let us mention the implementation of compulsory scolarization of all children along with the prohibition of children’s employment, the fixation of a series of legal age thresholds, and various welfare-state regulations with life course incidences, be they age related or not). Seen from this vantage point, the more or less regular timing of some crucial events in modern life courses, especially ‘‘normative transitions’’, no longer appears as a definitional element of life courses as such, but as one of


383

the various ways in which they may be socially institutionalized and standardized. Facing this empirical situation, we turn out to be rather poor in theoretical tools permitting to conceptualize such findings – take the pure timing dimension away and see what remains in terms of life course analytical tools! Not much, at least in the conceptual traditions we mentioned up to now; we shall come back to this question with respect to conceptual bridges. For identity change and life course transitions, we have proposed some hypotheses in an interdisciplinary perspective; this may be somewhat of a test area for our postulate of a priori symmetry between the disciplines – sociologists as well as demographers will certainly have a tendency to assume that social regulations trigger identity changes rather than the other way round, but a more agentic and also a (social-) psychological perspective will want to consider with equal interest the possibilities of active individual construction of transitions. Again, the various disciplines can only gain at working together in this area because they have developed different and potentially complementary conceptions of an issue they share, but that goes beyond their specific horizon. Let us pass more summarily on the remaining examples of transversal themes. Agency as a theme is almost per definition a meeting place for psychological, social–psychological and sociological aspects, including cognitive, affective, cultural, behavioral and interactional dimensions; coping may be less common a topic for demography and sociology than for the other two disciplines considered here, but its importance has long been acknowledged there, too. Finally, gender may need less insistence on the fruitfulness of an interdisciplinary approach than any of the other themes, but it should be underscored that probably few other perspectives than the study of life courses are equally apt at highlighting processes of gendering, and notably not only on the level of interindividual doing gender, but also on an institutional level – provided, of course, research is done in a gendersensitive way (Eichler, 1988). What about theoretical bridges? We may distinguish between two levels of theorizing, a Merton-like level of middle-range theorizing that corresponds to the concepts used in subject areas like those we have just mentioned, and a more metatheoretical or abstract level of general conceptual thinking. One track of interdisciplinary development on this second level is indicated by the recent FFT and SFM models in social psychology mentioned above. They provide explicit entry points for interactions with the person’s social environment as conceptualized, e.g., by sociological approaches – an aspect they share with Bronfenbrenner’s principle of ecological psychology or

384

RENE´ LEVY ET AL.

Erikson’s opening for social aspects entering into the stage-defining dilemmas of epigenesis he postulates, but their advantage can be seen in their firmer grounding in empirical research. This theoretical ‘‘offer’’ from personality psychology may be seen as a major anchor point for building conceptual bridges between psychology and sociology – let us try to develop, as a counterpart, a sociological anchor point, taking up the question of how and in which ways the link between specific role sets and their corresponding ages is socially constructed, especially on the level of macro- and mesosocial institutions. There is a promising track in sociological thinking, indicated, for example, by Kerckhoff (1993, p. 13) in a passage where he thinks about concepts for analyzing professional careers: ‘‘...we need to chart the movementyof individuals over time as they pass through a number of stages in the life course and occupy positions within hierarchically structured social organizations. At each stage, we need to be able to identify a set of locations that are hierarchically ordered, and we need to measure the personal characteristics of the individuals occupying those locations. yCharting the flow of individuals between structural locations across stages in the life course constitutes describing the ‘careers’ of those individuals, (i.e.,)ythe pathway (they follow)ybetween positions in the social structure occupied at different points in the life course.’’7 In a more general stance, we can define this structural part of the life course as a movement through social space. Taking into account the basic understanding that social space is generally organized in relatively welldefined and well-delimited social fields (Lewin, 1935; Bourdieu, 1980) and that – due to multiple participation – our social participation is generally definable by status/role sets or profiles rather than by single statuses and roles, we can reformulate this heuristic definition of the life course as a sequence of participation profiles (Levy, 1977, 1991).8 The linking to age of specific features of such sequences, especially of specific transitions between subsequent participation profiles, for instance, on the basis of cultural age norms (Neugarten, Moore, & Lowe, 1965; Settersten, 1997) some of which may be more officially institutionalized (Kohli’s chronologization), is but one of the multiple ways of life course standardization. The sequential ordering of a series of institutional participations (Kru¨ger, 2001, Bird & Kru¨ger) is another one. It is, however, important to stress that life course standardization is not a necessary ingredient of this analytical vision, it is one of its variable dimensions. Analyzing life courses as sequences of participation profiles, e.g., in a social structural perspective, enables us to more fully characterize the mechanisms that relate biographically changing social


385

participation to ongoing time without allowing the social structural aspects to go unobserved or to remain poorly defined (Settersten, 2005). Turning back to articulating the sociological and psychological perspectives, we may then ask to what extent psychological development and aging (cognitive, emotional, moral, identity), besides its ontogenetic ‘‘push factors’’, can also be shown to develop according to a threshold rather than gradual model that would, however, not express endogenous stages, but materialize in relation with transitions in participation profiles, and more specifically to what extent some changes of participation profiles, rarely studied in sociology, could have a psychological impact, e.g., profile extensions or restrictions in the sense that the number of simultaneous participations before and after a transition grows larger or smaller.9 Another substantive theoretical axis that may help articulate or integrate conceptual contributions of different disciplines can be seen in the distinction of various system levels, once different usages of some terms are spelled out and possibly agreed upon (e.g., what constitutes micro- and macrolevel phenomena differs dramatically between psychologists and sociologists). It would certainly be simplistic to assign to each discipline its proper system level, thus restraining it to the level in question (as Devereux, 1967, once proposed with his ‘‘complementaristic’’ conception of ethnopsychoanalysis). One discipline can legitimately study several such levels, some of which are also focused by other disciplines. This is especially evident for the subject matters shared by social psychology and psychology on the one hand, and sociology and social psychology on the other, but also and probably even more so of sociology and social demography. In this perspective, Settersten and Gannon’s formulation of agency within structure is not only a formula to overcome the artificial opposition of these complementary aspects of social reality, but to integrate more generally individual- and social-level explanations. Even if the classical distinction between structure and agency may have a singular ideological and political interest, it can be seen as a special case of this more general dimension of system-level differentiation. Spelling out and studying the mechanisms that relate different system levels with each other remains one of the lesser studied and poorly conceptualized areas in all concerned disciplines, and calls intrinsically for interdisciplinarity (Diewald, 2001). Saying this, we should also note that this volume (and probably most similar ones) does not cover all the possible system types and levels that are relevant for life course research. The social sciences certainly have to develop more explicit and integrative theoretical and methodological interfaces toward the biological system and its processes of ontogenesis on one hand (Shanahan, Hofer, & Shanahan, 2003) and, on

RENE´ LEVY ET AL.

386

the other, toward historical and cultural processes as well as actual processes of institutional framing that intervene heavily on the life course (see Elder, 1974; Heinz, 1992; Kru¨ger & Levy, 2001). Let us end this volume with a very general statement about interdisciplinarity. Once we are convinced of interdisciplinary work as a superior way to study life courses, we have to define the ways we want to ‘‘mix’’ the conceptual tools our various disciplines can bring to bear on the topic. One basic principle has already been stated, it is what we could call the principle of a priori disciplinary symmetry. A second basic principle should inform our attempts, we could call it an articulation principle: the various disciplines cannot and should not be simply melted into some unifying theoretical mold, given that they focus on complementary aspects of the same phenomena as well as on different and complementary levels of the systemic organization of reality. For that reason, to develop its potential, their scientific synergy needs explicit articulations, not fusion, and even less conceptual vagueness. The way leading to this goal is certainly difficult and laborious; we hope that this volume is more dynamic than the proverbial signpost that points to a direction without going there itself, i.e., that we have not only been able to point out some promising tracks, but also to undertake some real steps in that direction.

NOTES 1. Throughout this final chapter, all references to contributing authors concern their contributions to this volume. 2. Let us recall Marshall’s distinction between the principle of agency as an ingredient of human nature and the practical competence and possibility to act as a variable characteristic. Freire’s (1972) concept of conscientization can be cited as aiming at empowering oppressed individuals in order to give them the practical capacity to actualize this latent part of their being humans. 3. ‘‘Critical’’ does not necessarily mean threatening and even less unexpected – the following considerations are not limited to ‘‘non-normative’’ transitions or events. What is implied here is simply the idea that compared to the more or less stable periods between transitions, the latter tend to entail to a larger extent processes of change and adaptation on various levels. It goes without saying that this postulate is not meant to exclude the possibility of important processes and consequences attributable to what goes on during the phases between transitions. 4. We use this term in a rather general sense, much as it was already used by Lewin (1951) in his field theory approach where he defined a field as the totality of simultaneous and interdependent features that form a situation; however, we would insist more than he did on the systemic character of social fields (see also Bourdieu, 1984).


387

5. Two methodological problems that may be avoided by properly applying MLMs are the atomistic and the ecological fallacies, which incorrectly assign group effects at the individual level or, vice versa, individual effects at the group level, respectively. 6. A fifth, particularly promising set of techniques that life course scholars have an interest to adopt is that of linear and non-linear dynamical systems (e.g., Kaplan & Glass, 1995; in this volume, it has only been alluded to by McArdle). In this large set of techniques, the outcome of interest is not the value of a variable at a given time point, but the change in that variable over time. These methods presuppose that the human organism can be conceived as an open living system, continuously fluctuating, achieving a dynamic equilibrium through self-organizing structures tending towards homeostasis (not limited to its physiological meaning). These techniques provide a way to characterize many of the basic phenomena of development, including change, stability, variability, stages, continuity, and the combination of quantitative and qualitative change as well as the emergence of new forms of structure and function. 7. Similar ideas can be found in the thinking of Rosow (1976). 8. Without neglecting the fact that participation in a social field implies also holding a position in the field’s internal structure and assuming the correspond role. All these aspects have multiple implications that may be of importance when analyzing life course sequences, but cannot be spelled out here. 9. Among the rare studies of this aspect, let us cite Thoits (1986) and Moen, Dempster-McCain, & Williams (1989, 1992) who show beneficial effects of multiple role occupancy on women’s health. In this perspective, retirement corresponds to a major profile-narrowing.

REFERENCES Abbott, A. (1992). From causes to events: Notes on narrative positivism. Sociological Methods & Research, 20, 428–455. Abbott, A. (1995). Sequence analysis: New methods for old ideas. Annual Review of Sociology, 21, 93–113. Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart & Winston. Anstey, K. J., Lord, S. R., & Smith, G. A. (1996). Measuring human functional age: A review of empirical findings. Experimental Aging Research, 22, 245–266. Baltes, P. B., & Baltes, M. M. (Eds) (1990). Successful aging: Perspectives from the behavorial sciences. New York: Cambridge University Press. Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (in press). Lifespan theory in developmental psychology. In: R. M. Lerner (Ed.), Theoretical models of human development (Vol. 1 of the Handbook of child psychology). New York, NY: Wiley. Baltes, P. B., & Nesselroade, J. R. (1979). History and rationale of longitudinal research. In: J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in the study of behavior and development. New York: Academic Press. Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (Eds) (1977). Life-span development methodology: Introduction to research methods. Monterey, CA: Brooks-Cole.

388

RENE´ LEVY ET AL.

Belli, R. (1998). The structure of autobiographical memory and the event history calendar: Potential improvements in the quality of retrospective reports in surveys. Memory, 6(4), 383–406. Berger, P., & Luckmann, T. (1966). The social construction of reality: A treatise in the sociology of knowledge. Garden City, New York: Doubleday. Born, C., Kru¨ger, H., & Lorenz-Meyer, D. (1996). Der unentdeckte Wandel. Anna¨herung an das Verha¨ltnis von Struktur und Norm im weiblichen Lebenslauf. Berlin: Sigma. Bourdieu, P. (1980). Questions de sociologie. Paris: Minuit. Bourdieu, P. (1984). Distinction, a social critique of the judgement of taste. Cambridge: Harvard University Press. Brandtsta¨ter, J., Krampen, G., & Heil, F. E. (1993). Adaptative resources of the aging self: Outlines of an emergent perspective. International Journal of Behavioral Development, 16(2), 323–349. Campbell, A. A. (1968). The role of family planning in the reduction of poverty. Journal of Marriage and the Family, 30(2), 236–245. Chryssochoou, X. (2003). Studying identity in social psychology: Some thoughts on the definition of identity and its relation to action. Language and Politics, 2(2), 225–242. Devereux, G. (1967). From anxiety to method in the behavioral sciences. Den Haag-Paris: Mouton. Diewald, M. (2001). Unitary social science for causal understanding: Experiences and prospects of life course research. Canadian studies in population. Special Issue on Longitudinal Methodology, 28(2), 219–248. Eichler, M. (1988). Nonsexist research methods: A practical guide. London: Allen & Unwin. Eisenstadt, S. (1968). Social institutions. In: D. L. Sills (Ed.), Encyclopedia of the social sciences (pp. 409–429). New York: Macmillan and Free Press. Elder, G. H. (1974, 1999). Children of the great depression. Chicago: University of Chicago Press. Elder, G. H. (1992). Life course. In: E. F. Bogatta & M. L. Bogatta (Eds), Encyclopedia of sociology, (Vol. 3, pp. 1120–1130). New York: Macmillan. Elder, G. H. (1996). The life course paradigm: Social change and individual development. In: K. Lu¨scher (Ed.), Examining lives in context: Perspectives on the ecology of human development (pp. 101–139). Washington, DC: APA Press. Elder, G. H. J. (1998). The life course and human development. In: R. M. Lerner (Ed.), Handbook of child psychology (5th ed., Vol. 1, pp. 939–991). New York, NY: Wiley. Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Norton. Erikson, E. H. (1968). Identity: Youth and crisis. New York: W. W. Norton & Company, Inc. Freedman, D., Thornton, A., Camburn, D., Alwin, D., & Young-DeMarco, L. (1988). The life history calendar: A technique for collecting retrospective data. Sociological Methodology, 18, 37–68. Freire, P. (1972). Pedagogy of the oppressed. New York: Herder & Herder. Ghisletta, P., & Lindenberger, U. (2004). Static and dynamic longitudinal structural analyses of cognitive changes in old age. Gerontology, 50, 12–16. Goldstein, H. I. (1987). Multilevel models in educational and social research. London, UK: Oxford University Press. Havighurst, R. H. (1972). Developmental tasks and education (3rd ed.). New York: David McKay. Heckhausen, J. (1999). Developmental regulation in adulthood: Age-normative and sociostructural constraints as adaptative challenges. New York: Cambridge University Press.


389

Heinz, W. (Ed.) (1992). Institutions and gatekeeping in the life course. Weinheim: Deutscher Studienverlag. Hooker, K. (2002). New directions for research in personality and aging: A comprehensive model for linking levels, structures, and processes. Journal of Research in Personality, 36, 318–334. Hooker, K., & McAdams, D. P. (2003). Personality reconsidered: A new agenda for aging research. Journal of Gerontology: Psychological Sciences, 58B(6), 296–304. Kalicki, B., Fthenakis, W. E., & Peitz, G. (1999). The emergence of traditional gender-roles at the transition to parenthood. Poster, SRCD Biennial Meeting, 15–18 April 1999, Albuquerque, NM. Kalicki, B., Fthenakis, W. E., Peitz, G., & Engfer, A. (1998). Gender-roles at the transition to parenthood. Poster, XVth Biennial ISSBD Meeting, July 1st–4th, Berne. Kaplan, D., & Glass, L. (1995). Understanding nonlinear dynamics. New York, NY: Springer. Kerckhoff, A. C. (1993). Diverging pathways. Social structure and career deflections. Cambridge: Cambridge University Press. Kernis, M. H., & Goldman, B. M. (2003). Stability and variability inself-concept and self-esteem. In: M. R. Leary & J. P. Tangney (Eds), Handbook of self and identity (pp. 106–127). New York: The Guildford Press. Kohli, M. (1985). Die institutionalisierung des lebenslaufs. Ko¨lner Zeitschrift fu¨r Soziologie und Sozialpsychologie, 37, 1–29. Kohli, M. (1986). The world we forgot: A historical review of the life course. In: V. W. Marshall (Ed.), Later life: The social psychology of aging (pp. 271–303). Beverly Hills, CA: Sage. Kru¨ger, H. (2001). Geschlecht, territorien, institutionen. Beitrag zu einer soziologie der lebenslauf-relationalita¨t. In: C. Born & H. Kru¨ger (Eds), Individualisierung und verflechtung. Geschlecht und generation im deutschen lebenslaufregime (pp. 257–299). Mu¨nchen: Juventa. Kru¨ger, H., & Levy, R. (2001). Linking life courses, work, and the family: Theorising a not so visible ne xus between women and men. Canadian Journal of Sociology, 26(2), 145–166. Levinson, D. J. (1978). The seasons of man’s life. New York: Ballantine. Levinson, D. J. (1990). A theory of life structure development in adulthood. In: C. N. Alexander & E. J. Langer (Eds), Higher stages of human development. Perspectives on adult growth (pp. 35–53). Oxford: Oxford University Press. Levinson, D. J. (1996). The seasons of a woman’s life. New York: Ballantine. Levy, R. (1977). Der Lebenslauf als Statusbiographie. Die weibliche Normalbiographie in makrosoziologischer Perspektive. Stuttgart: Enke. Levy, R. (1991). Status passages as critical life course transitions. A theoretical sketch. In: W. R. Heinz (Ed.), Theoretical advances in life course research (pp. 87–114). Weinheim: Deutscher Studienverlag. Lewin, K. (1935/1951). Field theory in social science. New York: Harper & Bros. Li, P. S., & Currie, D. (1992). Gender differences in work interruptions as unequal effects of marriage and childrearing: Findings from a Canadian national survey. Journal of Comparative Family Studies, 23(2), 217–229. Mayer, K. U., & Mu¨ller, W. (1986). The state and the structure of the life course. In: A. B. Sørensen, et al. (Eds), Human development and the life course (pp. 217–245). Hillsdale: Lawrence Erlbaum Associates. Mayer, K. U., & Schoepflin, U. (1989). The state and the life course. Annual Review of Sociology, 15, 187–209.

390

RENE´ LEVY ET AL.

McAdams, D. P. (1996). Personality, modernity, and the storied self: A contemporary framework for studying persons. Psychological Inquiry, 7(4), 295–321. McArdle, J. J. (1986). Latent growth within behavior genetic models. Behavior Genetics, 16, 163–200. McCrae, R. R., & Costa, P. T. (2003). Personality in adulthood: A five-factor theory perspective (2nd ed.). New York: The Guildford Press. Mischel, W. (2004). Toward an integrative science of the person. Annual Review of Psychology, 55, 1–22. Mischel, W., & Shoda, Y. (1998). Reconciling processing dynamics and personality dispositions. Annual Review of Psychology, 49, 229–258. Moen, P., Dempster-McClain, D., & Williams, R. M. (1989). Social integration and longevity: An event history analysis of women’s roles and resilience. American Sociological Review, 54(4), 635–647. Moen, P., Dempster-McCain, D., & Williams, R. M. (1992). Successful aging: A life-course perspective on women’s multiples roles and health. American Journal of Sociology, 97(6), 1612–1638. Moen, P., & Han, S.-K. (2001). Reframing careers: Work, family, and gender. In: V. W. Marshall, W. R. Heinz, H. Kru¨ger & A. Verma (Eds), Restructuring work and the life course (pp. 424–445). Toronto: University of Toronto Press. Nesselroade, J. R. (1991). The warp and woof of the developmental fabric. In: R. M. Downs, L. S. Liben & D. S. Palermo (Eds), Visions of aesthetics, the environment, & development: The legacy of Joachim F. Wohlwill (pp. 213–240). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Neugarten, B. (1977). Personality and aging. In: J. E. Birren & K. W. Schaie (Eds), Handbook of the psychology of aging (pp. 626–649). New York: Van Nostrand Reinhold. Neugarten, B., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints, and adult socialization. American Journal of Sociology, 70(6), 710–717. Nurmi, J.-E., & Salmela-Aro, K. (2000). Goal construction, reconstruction and depressive symptomatology in a life span context: The transition from school to work. Journal of Personality, 70(3), 385–420. Rosow, I. (1976). Status and role change through the life span. In: R. H. Binstock & E. Shanas (Eds), Handbook of aging and the social sciences. New York: Van Nostrand Reinhold. Rovine, M. J., & Molenaar, P. C. M. (2000). A structural modeling approach to a multilevel random coefficients model. Multivariate Behavioral Research, 35, 51–88. Ryder, N. B. (1965). The cohort as a concept in the study of social change. American Sociological Review, 30, 29–46. Schulz, R., & Heckhausen, J. (1996). A life span model of successful aging. American Psychologist, 51, 702–714. Scott, J., & Alwin, D. F. (1998). Retrospective versus prospective measurement of life histories in longitudinal research. In: J. Z. Giele & G. H. Elder, Jr. (Eds), Methods of life course research: Qualitative and quantitative approaches. Thousand Oaks, CA: Sage. Settersten, R. A. (1997). The salience of age in the life course. Human Development, 40, 257–281. Settersten, R. A. (2005). Toward a stronger partnership between life-course sociology and lifespan psychology. Research in Human Development, 2(1+2), 25–41. Settersten, R. A., Jr. (1999). Lives in time and place. Amityville, NY: Baywood Publishing Company, Inc.


391

Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring, and the life course. Annual Review of Sociology, 23, 233–261. Shanahan, M. J., Hofer, S. M., & Shanahan, L. (2003). Biological models of behavior and the life course. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course. New York: Kluwer Academic/Plenum Publishers. Sliwinski, M. J., Hofer, S. M., Hall, C., & Buschke, H. (2003). Modeling memory decline in older adults: The importance of preclinical dementia, attrition, and chronological age. Psychology and Aging, 18, 658–671. Srivatava, S., John, O. P., Gosling, S. D., & Potter, J. (2003). Development of personality in early and middle adulthood: St like plaster or persistent change? Journal of Personality and Social Psychology, 84(5), 1041–1053. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103(2), 193–210. Taylor, S. E., Kemeny, M. E., Reed, G. M., Bower, J. E., & Gruenewald, T. L. (2000). Psychological resources, positive illusions, and health. American Psychologist, 55 (99–109). Thoits, P. A. (1986). Multiple identities: Examining gender and marital status differences in distress. American Sociological Review, 1(2), 259–272. Widmer, E., Kellerhals, J., & Levy, R. (2005). What pluralization of the life course? An analysis of personal trajectories and conjugal interactions in contemporary Switzerland. In: H. Kriesi, P. Farago, M. Kohli & M. Zarin-Nejadan (Eds), Contemporary Switzerland: revisiting the special case. Houndmills: Palgrave, Macmillan. Widmer, E., Kellerhals, J., & Levy, R., with the collaboration of M. Ernst Sta¨hli & R. Hammer (2003a). Couples contemporains – cohe´sion, re´gulation et conflits. Une enqueˆte sociologique. Zurich: Seismo. Widmer, E., Levy, R., Pollien, A., Hammer, R., & Gauthier, J.-A. (2003b). Entre standardisation, individualisation et sexuation: Une analyse des trajectoires personnelles en Suisse [Between standardization, individualization, and sexuation: An analysis of personal trajectories in Switzerland]. Swiss Journal of Sociology, 29, 35–67. Wohlwill, J. F. (1970). The age variable in psychological research. Psychological Review, 77, 49–64.


392

AUTHOR INDEX Abbott, A. 13, 188, 274, 309, 364, 379 Abelson, R. 238 Abrams, D. 202 Ackerman, S. 62 Adler, J. 247 Agrawal, R. 301 Ahrons, C. 160 Aisenbery, S. 188 Alea, N. 219–220 Alexander, K.L. 132 Allison, P.D. 266, 289 Allport, G.W. 239, 371 Alpert, A. 333 Alter, G. 287, 291–293 Alwin, D.F. 37, 69, 85, 373, 375–376 Ambady, G. 241 Anderson, E. 317, 326, 351 Andres, D. 226 Anstey, K.J. 382 Anyadike-Danes, M. 277 Anyidoho, N.A. 245 Appice, A. 303 Aranza-Ordaz, F.J. 289 Arbuckle, J.L. 351 Arbuckle, T.Y. 226 Archer, M.S. 70, 85–86 Arnett, J.J. 243–244 Attanucci, J. 14 Attias-Donfut, C. 95 Axinn, W.G. 266, 293 Ayers, M.S. 220 Bakan, D. 62, 246 Baltes, P.B. 11–12, 14, 28, 43, 48, 132, 221, 231, 241, 315–317, 361, 378

Bandura, A. 132 Barber, B.L. 168 Barber, J.S. 266, 293 Bartlett, F.C. 219–220, 229 Barutchu, A. 202 Bates, D.M. 351 Bauer, J.J. 245 Bauman, Z. 99 Baumeister, R.F. 239 Bawin-Legros, B. 95 Bu¨chel, F. 190 Beck, U. 42, 174, 179 Becker, H.S. 156 Becker, S.O. 271–272 Becker-Schmidt, R. 176–177 Beck-Gernsheim, E. 179 Beckmann, P. 180, 190 Beekink, E. 291–292 Begin, J. 107 Bell, R.Q. 317, 333, 351 Belli, R. 376 Belsky, J. 123 Bengtson, V.L. 95 Berchtold, A. 295–296, 299 Berger, P.L. 69–70, 73, 363 Bernard, H.R. 206 Berntsen, D. 219 Berquo, E. 286 Bickel, J.-F. 96 Billari, F.C. 261–262, 273–274, 287, 299, 301–303 Billig, M.G. 202 Bird, K. 180–182 Birkett, H. 342 Blakeslee, S. 159

393

394 Blechler, M. 46 Bleuler, E. 97 Block, J. 342 Blockeel, H. 299, 301–302 Blossfeld, H.-P. 174, 177, 187, 189, 265–268, 278, 287–288, 290, 296 Bluck, S. 219–221, 243, 250 Bock, R.D. 350 Bocquier, P. 266, 287 Bogen, S. 122 Boker, S.M. 348, 351 Born, C. 179, 182, 189, 368 Bouchard, T.J. 240 Bourdieu, P. 300, 384, 386 Bower, J.E. 374 Bowman, P.J. 246–247 Bradway, K.P. 318, 320, 348 Brainerd, C.J. 220 Brand, C. 109 Brandtsta¨dter, J. 40, 43, 132, 372 Brannen, J. 102, 108 Breiman, L. 26, 263, 300, 303, 306 Brewer, W.F. 220 Bronfenbrenner, U. 28, 38 Brooks-Gunn, J. 164 Brostro¨m, G. 291–293 Brown, C. 245 Brown, J.D. 374 Brown, R. 200 Browne, M. 351 Browne, W. 293 Bruner, J.S. 238, 242, 252 Bryk, A.S. 293, 328, 335, 351 Bryson, A. 270 Buckholdt, D.R. 85 Bundy, R.F. 202 Burkhardt, A. 109 Buschke, H. 382 Busch-Rossnagel, N.A. 60, 84 Butler, R.N. 82, 230 Bynner, J. 210

AUTHOR INDEX Cain, L.D. 72 Call, K.T. 135, 144 Camburn, D. 376 Campbell, A.A. 157, 364 Campbell, C. 68–69 Campbell, D.T. 345 Campbell, R.T. 85 Cantor, N. 241 Carstensen, L.L. 47 Caselli, G. 285 Casey, K. 238 Cattell, R.B. 316, 338, 342 Cavanaugh, J.C. 226 Ceci, M. 303 Chaikelson, J. 226 Charme, S. 239, 243 Chartrand, E. 107 Cherlin, A.J. 110, 159 Christensen, K. 271 Chryssochoou, X. 374 Ciampi, A. 303, 309 Cinnirella, M. 17 Cinotta, S. 202 Clarke, P. 69 Clausen, J.A. 40, 132, 135, 156, 168 Clausen, J.S. 68, 80, 218 Clogg, C. 144 Cnaan, A. 324 Coenen-Huther, J. 106, 111 Cohler, B.J. 99, 109, 114, 238, 244 Colby, A. 199, 248 Coleman, J.S. 138 Collins, R. 85 Colvin, C.R. 240 Congdon, R. 293 Conger, R.D. 123 Connidis, I.A. 106, 121 Converse, P. 209 Conway, M.A. 219–220, 238 Cook, T.D. 44, 345 Costa, P.T. 224, 237, 239–240, 252, 371 Counts, D. 73 Courage, M.L. 242

395

Author Index Courgeau, D. 12, 19, 266, 287, 289 Cox, D.R. 264, 289 Crawford, J.R. 324 Cremer, C. 190 Criqui, M.H. 240 Crompton, R. 189 Crosnoe, R. 173, 187 Crowley, J. 309 Cudeck, R. 329, 333, 351 Cumming, E. 73 Curran, P. 351 Currie, D. 368 Czikszentmihalyi, M. 204 Dahrendorf, R. 69 Damon, W. 248 Dannefer, D. 41, 47, 59, 64, 85, 237 Day, R. 246 De Bruijn, B.J. 14 de Coninck, F. 19 De Rose, A. 303 de St. Aubin, E. 241, 245, 247–249 Deary, I.J. 239–240, 324 Dehaspe, L. 301 Dehejia, R.H. 269 Del Vecchio, W. 240 Demeny, P. 285 Dempster-McClain, D. 387 Denzin, N. 250 Devereux, G. 385 Diamond, A. 247, 249 Diehl, M. 60 Diewald, M. 40, 42, 385 Diggle, P.J. 341 Dilthey, W. 81 Dixon, R.A. 227 Doise, W. 17, 201 Donaldson, G. 316–317 Donati, P. 95 Dorsett, R. 270 Doyle, G.C. 218 Draper, D. 293 Drobnicˇ, S. 177, 189

du Toit, S.H.C. 329, 333, 351 Duncan, L.E. 241 Duncan, S.C. 333 Duncan, T.E. 333 Dupaˆquier, J. 284 Dupaˆquier, M. 284 Durante, M. 202 Durkheim, E. 42 Durkin, K. 16 Dykstra, P.A. 262, 287 Eccles, J.S. 44, 168 Edel, L. 239 Edelstein, W. 14 Edmunds, J. 100 Eerola, M. 266 Ehmer, J. 118 Ehrsam, R. 222 Eich, E. 227 Eichler, M. 383 Eisenstadt, S. 362 Elder, G.H.Jr. 6, 12, 17, 44, 60, 63, 65–66, 68, 72, 106, 123, 127–128, 134, 155–156, 173–174, 187, 189, 218, 232, 237, 244, 252, 273, 361, 363–364, 368, 375, 386 Eliason, S.R. 144, 151 Ely, M. 238 Emler, N. 14, 197, 203, 205–206, 209–210, 212 Emmons, R.A. 241 Engelbrech, G. 190 Engfer, A. 369 Engstler, H. 111, 124 Entwisle, D.R. 132 Epstein, D.B. 317, 326, 329, 335, 338, 351 Epston, M. 251 Erikson, E.H. 14, 16, 82, 198, 217, 230, 237, 239, 241, 243, 371 Erzberger, C. 23, 183, 188 Esping-Andersen, G. 49

396 Esser, H. 190 Essex, M.J. 16 Fahrenberg, J. 223, 226 Farkas, J.I. 174 Ferrer, E. 351 Ferrer-Caja, E. 331 Filipp, S.-H. 106–107 Finch, J. 95 Finch, M.D. 151 Fingerman, K.L. 108 Fitzgerald, J.M. 225, 229 Flaherty, B.P. 139 Flament, C. 202 Flanagan, C.A. 46 Flavell, J.H. 198 Foley, J.M. 247 Foner, A. 155 Forrest, J.D. 29, 309 Foster, E.M. 46 Frank, E. 303 Frankel, M.R. 293 Freedman, D. 376 Freire, P. 386 Freud, S. 98, 217 Freund, A.M. 40, 132, 241 Friedman, H.S. 240 Friedman, J.H. 303, 306 Friis, H. 95 Fu¨rnkranz, J. 274, 299, 301–303 Fthenakis, W.E. 369 Funder, D.C. 240 Furstenberg, F.F. 44, 46, 155, 157–159, 164 Gabriel, Y. 238 Gallagher, M. 159 Gannon, L. 35 Gardiner, J.M. 219 Garrett, M. 202 Gauthier, A. 95 Gauthier, J.-A. 366, 369, 380 Gecas, V. 47, 132

AUTHOR INDEX George, L.K. 11, 62, 69, 86 Gergen, K.J. 238, 244 Ghisletta, P. 361, 380 Giarusso, R. 95 Gibbons, R. 351 Giddens, A. 61, 65, 70, 85, 174, 241, 244 Giele, J. 40 Gifford, E.J. 46 Gigerenzer, G. 41 Giles, H. 201 Gilligan, C. 14 Gjerde, P.F. 251 Glass, L. 387 Godard, F. 19 Goffman, I. 75, 81, 112 Gold, D.P. 226 Goldberg, L.R. 240 Goldman, B.M. 366 Goldstein, H. 293, 351, 379 Goode, W. 213 Goody, J. 86 Gosling, S.D. 371 Gottschall, K. 182 Grabowski, L.J. 135, 144 Grambsch, P.M. 290, 293 Green, E.E. 226 Greenhalgh, S. 285 Gregg, G. 239, 250 Griffin, D.W. 105 Gruenewald, T.L. 374 Grunebaum, H.U. 99, 114 Gubrium, J.F. 85, 238 Gutmann, D. 61, 244 Gutschner, P. 118 Habermas, T. 243, 250 Haely, M. 293 Hagestad, G.O. 16, 47, 156, 174, 287 Hakim, C. 177 Hall, C. 382 Hall, M. 303 Hallahan, M. 241

397

Author Index Hamagami, F. 318, 320–321, 329, 331, 333–335, 337–338, 345, 347–349, 351 Hamerle, A. 266, 288 Hammer, R. 366, 369, 380 Hampel, R. 223, 226 Han, J. 300 Han, S.-K. 186, 189, 366 Hand, D.J. 300 Hankiss, A. 243 Harlow, S.D. 315, 352 Harootyan, R.A. 95 Harris, F. 189 Hashtroudi, S. 219 Havighurst, R.H. 16, 371 Hay, E. 108 Hayes, C. D. 158 Heckhausen, J. 43, 132, 372 Heckman, J. 270 Hedecker, D. 351 Heidrich, S.M. 218 Heil, F.E. 372 Heilbrun, C. 250 Heinz, W.R. 64, 85, 173–174, 185, 187 Hennessy, J. 202 Henry, W.E. 73 Herman, M.R. 44 Hermans, H.J.M. 239, 246 Hershberg, T. 157 Hertzog, C. 227 Hetherington, E.M. 168 Hewstone, M. 201 Ho¨hn, C. 286 Hobcraft, J. 262, 266 Hoch, H. 110, 125 Hofer, D. 226 Hofer, S.M. 382, 385 Hoffman, B.J. 246 Hogan, D.P. 16, 174 Hogdson, D. 285 Hogg, S.A. 303, 309 Holford, T.R. 289 Holland, P.W. 270

Holman, T.B. 107 Holmes, J. 238 Holstein, J.A. 85, 238 Hooimeijer, P. 286 Hooker, K. 239, 241, 244, 371–372 Horn, J.L. 316–317 Hotz, V.J. 269–271 House, J.S. 133 Howe, M.L. 242 Ho¨pflinger, F. 222, 224 Hradil, S. 179 Huang, Y.T. 245 Hughes, E.C. 72 Huinink, J. 183 Hultsch, D.F. 227 Ichino, A. 271–272 James, A. 40, 45 Java, R.I. 219 Jekeli, I. 106 Jenks, C. 40, 45 Jepperson, R.L. 59, 65–66 John, O.P. 240–241, 371 Johnson, K.M. 60, 63, 173, 187 Johnson, M.J. 107, 155 Johnson, M.K. 219 Jolesz, F. 321 Jones, G. 211 Jones, K. 321 Jones, R.L. 86 Josselson, R. 225 Jo¨reskog, K.G. 338 Julien, D. 107 Junn, J. 209 Kahana, B. 61 Kahana, E. 61 Kahn, R. 48, 202 Kaiser, A. 222, 224 Kalicki, B. 369 Kalish, R.A. 75–76

398 Kamber, M. 300 Kangas, J. 320 Kaplan, B. 245 Kaplan, D. 387 Kass, G.V. 304 Kastenbaum, R. 85 Katz, D. 202 Keane, M.R. 273 Kellerhals, J. 106, 366, 369 Kelly, J. 168 Kemeny, M.E. 374 Kennedy, S. 46 Kenny, D.A. 205 Kenrick, D.T. 205, 240 Kenward, M.G. 341 Kerckhoff, A.C. 384 Kernis, M.H. 366 Keyes, C.L.M. 219, 230 Kikinis, R. 321 Killworth, P.D. 206 King, L. 245 Kiser, E. 65 Kish, L. 293 Kissel, E. 247 Klein, D. 104 Klein, V. 180 Klineberg, J. 204 Kling, K.C. 16 Kling, V. 226 Klohnen, E.C. 241 Kluckhohn, C. 239 Kluge, S. 188 Knapp, G.-A. 177 Knellessen, O. 122 Knox, W.E. 135 Kohlberg, L. 198–199 Kohler, H.-P. 271 Kohli, M. 12, 187, 243, 363, 367, 382 Krampen, G. 372 Krebs, E. 222 Kru¨ger, H. 50, 134, 150, 173–174, 176, 179, 182, 189, 363, 368, 384, 386

AUTHOR INDEX Kruglanski, A.W. 14 Ko¨ttig, M. 29 Kuh, D. 218 Kuijsten, A. 286 Kurtz, B. 180 Labouvie-Vief, G. 14 Laird, N.M. 324, 334, 351 Lalive d’Epinay, C. 96, 125 Landwehr, N. 303 Lang, F.R. 47, 102, 115 Langbaum, J. 239 Larson, R. 204 Lash, S. 174 Lautmann, R. 110 Lawton, M.P. 61 Layder, D. 69 Le Goff, J.-M. 111, 268, 361 Leary, T. 101 Leblanc, M. 309 Lee, J.C. 131 Lee, S. 97, 125 Lelie`vre, E. 12, 19, 266, 287, 289 Lemmon, H. 324 Lerner, R.M. 60, 84 Lettke, F. 102, 104, 106–107, 122 Leveille, S. 107 Levine, D.I. 270–271 Levinson, D.J. 237, 244, 252, 368, 371 Levy, R. 3, 11, 50, 175, 361, 363, 366, 369, 380, 384, 386 Lewin, K. 384, 386 Lewis, M. 246–247 Li, A. 50 Li, F. 333 Li, P.S. 368 Li, S.-C. 48 Lianos, G. 202 Lieberson, S. 160 Liefbroer, A.C. 291–292 Lillard, L.A. 267, 278, 290 Lim, K.O. 345 Lin, D.Y. 293

Author Index Lindenberger, U. 14, 28, 43, 221, 315, 361, 380 Lindsay, D.S. 219 Lindsay, P. 135 Littell, R.C. 351 Little, B.R. 241 Little, R.J.A. 324, 341 Little, R.T.A. 351 Littwak, E. 95, 125 Loevinger, J. 198 Loftus, E.F. 220 Logan, R.L. 121, 241 Lord, S.R. 382 Lorence, J. 135 Lorenz-Mayer, D. 179, 182, 189 Lorenz-Meyer, D. 107, 116, 368 Lowe, J.C. 17, 155–156, 384 Lowenstein, A. 117 Lu¨scher, K. 93, 95–96, 102, 106–107, 110–111, 115, 118, 122 Luckmann, T. 69–70, 73, 363 Ludewig-Kedmi, R. 113 Lykken, D.T. 240 Lynch, K.A. 295 Mabry, B. 95, 123 Macaulay, D. 227 MacDonald, I.L. 295 Machado, M.A. 245 MacIntyre, A. 238 Maclean, M. 218 Macmillan, R. 144, 151 Maes, H.H. 351 Magnusson, D. 342 Malerba, D. 303 Malo, M.A. 309 Mandler, J.M. 242 Mannila, H. 300–301 Mansfield, E.D. 246–247, 249 Manting, D. 18 Manton, K.G. 291 Maples, J. 266, 293 Marcia, J. 198

399 Marini, M. 156 Markman, H.J. 107 Markus, H.R. 80, 132 Marshall, V.W. 57, 66, 69–70, 72–74, 78, 81–83, 85–86, 95, 173, 185 Maruna, S. 238 Marx-Ferree, M. 189 Mason, J. 95 Matthews, G. 239–240 Matthews, S.H. 95 Mayer, A.-K. 106–107 Mayer, K.U. 7, 12, 16, 42, 49, 174, 177, 181, 183, 187, 189, 266–267, 273, 288, 363, 370, 382 Mazzuco, S. 273 McAdams, D.P. 14, 17, 121, 126, 217, 230, 233, 237–239, 241, 243–250, 252, 371–372 McArdle, J.J. 315, 317–318, 320–321, 324, 326, 329, 331, 333–335, 337–338, 341, 344–345, 347–349, 351–352, 382 McCrae, R.R. 224, 237, 239–240, 252, 371 McCullagh, P. 310 McGue, M. 240 McKinney, S. 303, 309 McLanahan, S. 159 McLoyd, V. 46 McMullin, J.A. 69–70, 121, 124 McNamara, S. 206 McNicoll, G. 285 McVicar, D. 277 Mead, G.H. 100, 119 Meier, B. 226 Mercer, R.T. 218 Meredith, W. 318, 320, 328, 348, 350–351 Merton, R.K. 156 Metha, P.D. 351 Meyer, J.W. 59, 65–66 Meyerhoff, B.G. 86

400 Miech, R.A. 68, 134 Milardo, R. 107 Milgram, S. 204 Milhoj, P. 95 Miliken, G.A. 334, 351 Mills, C.W. 49 Mills, M. 268 Milofsky, E. 241 Mischel, W. 240–241, 371, 373 Mitchell, J.C. 202 Mitteness, L.S. 114 Mu¨ller, W. 187, 363 Modell, J. 157 Moen, P. 174, 176, 186, 189, 366, 387 Mojardin, A.H. 220 Molenaar, P.C.M. 16, 380 Monopoli, M. 303 Moore, J.W. 17, 155–156, 384 Morgan, S.P. 164 Mortimer, J.T. 131, 133–136, 138–140, 144, 150–151, 174–175 Moskowitz, D.S. 62 Mounoud, P. 16 Mouw, T. 274, 276–277 Mueller, M.M. 66, 83, 173 Mugny, G. 17 Mullen, B. 230 Mullin, C.H. 269–271 Munoz, F. 309 Murphy, M. 262, 266 Murphy, S.A. 266, 293 Murray, H.A. 239 Murray, S.L. 238 Mutheń, B.O. 342, 351 Mutheń, L.K. 342, 351 Myrdal, A. 180 Nagin, D. 342 Neale, M.C. 351 Nelder, J.A. 310 Nesselroade, J.R. 316–317, 321, 333, 344–345, 352, 366, 378

AUTHOR INDEX Neugarten, B.L. 16–17, 155–156, 174, 371, 384 Nichols, E.G. 218, 233 Nie, N.H. 209 Noam, G. 14 Nurius, P. 80, 132 Nurmi, J.-E. 376 Nydegger, C.N. 114 Oakes, J.M. 39 Oesterle, S. 138, 140, 174 Ogg, J. 117 Olshen, R.A. 303, 306 Olson, L.S. 132 Ondrich, J. 190 O’Rand, A. 63, 65–66, 85 O’Rand, A.M. 174 Oris, M. 283, 287, 291–293, 299 Otscheret, E. 122 Painter, G. 270–271 Pajung-Bilger, B. 96, 102 Paliwal, K.K. 295 Pallara, A. 303 Panis, C.W.A. 267, 278 Parker, R. 111 Parsons, T. 95, 156 Pascual-Leone, J. 14 Passel, J.S. 39 Patten, A. 246–247 Peitz, G. 369 Perren, S. 225 Perrig, W.J. 217, 222, 226–227 Perrig-Chiello, P. 217, 222–227, 233 Petersen, T. 266 Peterson, C. 247 Pfefferman, A. 345 Phillips, M. 44 Philo, C. 45 Piaget, J. 198 Piccarreta, R. 274 Pillemer, K. 95–96, 106, 112, 118

401

Author Index Pinheiro, J.C. 351 Plakans, A. 118 Platow, M. 202 Plewis, I. 293 Pleydell-Pearce, C.W. 238 Polkinghorne, D. 238 Pollien, A. 366, 369, 380 Pomerantz, E. 237, 240 Pothoff, R.F. 350 Potter, J. 371 Poulain, M. 287 Powesland, P.F. 201 Prein, G. 23 Prescott, C.A. 337, 351 Pressat, R. 17 Pross, H. 180 Prout, A. 40, 45, 49–50 Prskawetz, A. 274, 299, 301–303 Po¨tter, U. 278, 290 Purdon, S. 270 Pyke, K.D. 123 Quinlan, J.R. 303, 306 Rabiner, L.R. 295 Raftery, A.E. 295 Rakotomalala, R. 310 Ramon, J. 301 Ramsey, C.M. 245 Rao, C.R. 328 Rasbash, J. 293 Raudenbush, S.W. 293, 328, 335, 351 Reder, L.M. 220, 231 Reed, G.M. 374 Reicher, S.D. 202, 205 Rein, M. 95 Reinharz, S. 96 Reyna, V.F. 220 Reynolds, J. 246–247 Richards, L.N. 95 Ricoeur, P. 239, 242 Riegel, K.R. 14, 84

Riley, J.W. 37 Riley, M.W. 37, 72, 155 Rindfuss, R.R. 16, 87, 155, 174, 262 Ritschard, G. 283, 299, 303, 307 Robb, R. 270 Roberts, B.W. 237, 240 Roberts, K. 134 Roberts, R.E.L. 95, 127 Robertson, E.B. 123, 127 Rodgers, B. 218 Rogosa, D. 317, 350 Rohwer, G. 265–267, 278, 287, 290, 296 Romney, D. 210 Rosenbaum, P.R. 271–272 Rosenbloom, M.J. 345 Rosenfeld, R.A. 16, 87, 155, 174 Rosenthal, C.J. 95 Rosenthal, G. 29 Rosenthal, R. 241 Rosenwald, G. 250 Rosow, I. 387 Ross, K. 46 Rossi, A.S. 62, 95 Rossi, P.H. 39, 95 Rossier, C. 111 Rost, H. 190 Rothermund, K. 43 Rotter, J.B. 132 Rovine, M.J. 380 Rowe, J. 48 Roy, S.N. 350 Rubin, D.B. 271–272, 351 Rubin, D.C. 219–220 Rubin, M. 201 Ruch, M. 226 Rudorf, S. 109 Ruetzel, K. 247 Rumbaut, R. 46 Rusca, E. 231 Rutter, M. 217–218 Rybash, J.M. 225 Ryczkowska, G. 303

402 Ryder, N.B. 156, 376 Ryff, C.D. 16, 69, 80, 218–219, 230 Ryu, S. 151 Sackmann, R. 175 Salmela-Aro, K. 376 Sameroff, A. 44 Sandefur, G. 159 Sanders, S.G. 269–271 Sarbin, T. 238 Sauvain-Dugerdil, C. 111 Sayer, A.G. 345, 351 Schacter, D.L. 219, 230 Schaie, K.W. 12 Schank, R. 238 Schmidt, B. 177 Schneider, N.F. 190 Schoeni, R. 46 Schoepflin, U. 12, 363 Schu¨tze, Y. 181 Schulz, R. 372 Schutz, A. 70 Schwartz, J.E. 240 Schwartz, S.H. 241 Schwartzman, A. 226 Schwarz, N. 219 Scollon, C.K. 245 Scott, J. 373, 375 Seber, G.A.F. 329 Segal, M.R. 309 Segal, N.L. 240 Selg, H. 223, 226 Seligman, M.E.P. 247 Serafin, D. 226 Settersten, R.A. 4–5, 16, 28, 35, 39–40, 42–44, 46, 48, 52, 60, 69, 156, 174, 267, 273, 370, 373, 382, 384–385 Sewell, W.H. 37, 70, 85 Shanahan, L. 385 Shanahan, M.J. 50, 68, 133–134, 139, 151, 385 Shanas, E. 95 Sheldon, K.M. 239

AUTHOR INDEX Shifley-Grove, S.S. 225, 229 Shoda, Y. 241, 371 Shuey, K.M. 106 Sibeon, R. 43 Silverstein, M. 95 Singer, J.A. 241, 245–246 Singer, J.D. 351 Skevington, S. 202 Skinner, M.L. 123 Skytthe, A. 271 Slasor, P. 324 Slater, P.E. 208 Sliwinski, M.J. 382 Small, B.J. 227 Smelser, N.J. 70, 99, 113, 122 Smith, G.A. 382 Smith, J. 40 Smyth, P. 300 Spangler, D. 106 SpieX, K. 190 Spini, D. 361 Spiro, A. 321 So¨rbom, D. 338 Sørensen, A. 174 Srikant, R. 301 Srivastava, S. 240, 371 Staff, J. 131, 138, 140 Stallard, E. 291 Starr, J.M. 324 Staudinger, U. M. 14, 28, 43, 100, 120, 221, 315, 361 Stehlik-Barry, K. 209 Stehouwer, J. 95 Stewart, A.J. 241 Sta¨helin, H.B. 222, 226 Stone, C.J. 303, 306 Stoup, W.W. 334, 351 Strack, F. 219, 234 Strassen, J.-F. 95 Strauss, A.L. 156 Stringfield, D.O. 205 Struyf, J. 301

403

Author Index Strycker, L.A. 333 Sturzenegger, M. 222, 224 Stuss, D.T. 219 Sugarman, L. 218 Suitor, J.J. 106, 112 Sullivan, E.V. 345 Suls, J. 230, 234 Sussman, M.B. 95 Swicegood, C.G. 16, 87, 155, 174 Szreter, S. 285 Szydlik, M. 95 Tajfel, H. 200, 202–203 Tashakkori, A. 23 Tavare´, S. 295 Taylor, S.E. 374 Teddlie, C. 23 Tellegen, A. 240 Therneau, T.M. 290, 293 Thiffault, J. 303, 309 Thoits, P.A. 387 Thom, R. 16 Thomas, W.I. 63 Thompson, C.W. 320 Thompson, M.M. 105 Thorne, A. 244–245 Thornton, A. 376 Thu¨rlemann, F. 122 Tisak, J. 328, 350–351 To¨lke, A. 183 Toivonen, H. 301 Tomasello, M. 242 Tomlinson-Keasy, C. 240 Tostain, M. 14 Townsend, P. 95 Tsay, A. 274 Tucker, J.S. 240 Tucker, L.R. 328 Tulving, E. 219 Turiel, E. 14 Turner, B.S. 100 Turner, J.C. 201, 203

Uhlenberg, P. 59 Vaillant, G.E. 218, 241 Vallin, J. 284–285 Van de Kaa, D.J. 285 Van de Water, D. 247 van der Heijden, P.G.M. 274 van der Maas, H.L. 16 Van Imhoff, E. 286 van Poppel, F. 291–292 van Wissen, L.J.G. 262, 286–287 Vaupel, J.W. 291 Verkamo, A.I. 301 Viaud, J. 11, 16 von Allmen, M. 106 von Matt, P. 97 Ve´ron, J. 285 Wadsworth, M.E.J. 218 Wahba, S. 269 Waite, L. 159 Wallace, C. 211 Wallerstein, J.S. 159 Walshe, J. 202 Ware, J.H. 334, 351 Weber, M. 64, 69 Weddeburn, D. 95 Wei, L.J. 293 Wellman, H.M. 242 Wertlieb, D. 218 West, M.A. 202 West, S.G. 351 Whalley, L.J. 324 Wheeler, M.A. 219 White, M. 251 Whiteman, M.C. 239–240 Whittlesea, B.W.A. 220 Widmer, E. 361, 366, 369, 380 Wiese, B.S. 132 Wild, C.J. 329 Wilensky, H. 157 Willekens, F.J. 284 Willett, J.B. 317, 333, 345, 350–351

404 Williams, N. 202 Williams, R.M. 387 Williams, T. 245 Willms-Herget, A. 179 Willson, A.E. 106 Wingard, D.L. 240 Wingens, M. 175 Winsborough, H. 156 Winter, D.G. 241 Wishart, J. 328, 331 Wohlwill, J.F. 5, 381 Woike, B. 239, 245–246 Wolfinger, R.D. 334, 351 Wolpin, K.I. 273 Woodcock, J.R. 326, 331, 351 Woodcock, R.W. 331 Woodhouse, G. 293 Wothke, W. 351 Wright, R. 220 Wu, L. 50

AUTHOR INDEX Wu, L.L. 188, 266–267, 274 Wunsch, G. 285 Wurzbacher, G. 180 Xenos, P. 286 Xie, G. 351 Yamaguchi, K. 266, 289 Yang, M. 293 Yang, Q. 190 Young-DeMarco, L. 376 Zanna, M.P. 105 Zeger, S.L. 352 Zighed, D.A. 303, 307, 310 Zima, P.V. 105 Znaniecki, F. 63 Zucchini, W. 295 Zuroff, D.C. 62

SUBJECT INDEX accelerated failure time model 289 ACT effects 316–317 action-structure problem 69–70 adaptation 61 adjacent institutionalization 51 adolescence 45, 144, 146, 248 affection 107 agency 39–41, 44, 49, 58–69, 83, 144–149, 246, 362–365 agency and structure 21–22, 35, 71–83, 131, 364–365, 370, 379 agency within structure 41–43, 51–52, 385 agentic decision-making and behavior 133 agentic strategies 133 aging and dying 58, 71–72 AIC criteria 298, 307, 308 algorithmic culture 274–278 ambivalence 93, 95, 97–105, 110–119 ambivalence diversification 108–110 ambivalence understanding elements 98 analysis of covariance 334–335 assessment and differentiation of ambivalence 105–108 assessment of relationships 105–108, 112 attitude 220, 230 autobiographical coherence 243 autobiographical memory 24, 219–220 autobiographical reflection selectivity 81 autobiographical self 242 average effect of treatment 270–271 awareness and timeframes 186

baby break 178–182 Bartlett’s schema 220 basic tendencies 371–372 BIC criteria 298, 307–308 Big Five traits 240, 371 biographical experiences 217–218 biographical memories 225–227 biographical recollection 227–230 biographical reconstruction 23–25, 195, 370–374 biographical transition 217–218, 222, 224 Bradway–McArdle collection 335 Bradway–McArdle Longitudinal Study 318, 320–321 Captivation 104 Caring 113–117 causal approach 265–269 causal coherence 243 causal impacts of events 269–273 causal inference 270 ‘causality’ culture 265–273 challenges and contradictions of life course 35 changes in identity 81–83 changing social identity 208–209 characteristic adaptations 241, 371–372 child-rearing leave 178–180 childhood and adolescence 44–45 childhood friends and playmates 208 class-theoretical models 38 classification tree 303, 309 cognitive functioning 198–199, 226

405

406 cognitive growth for males and females 340–342 cognitive skills 243 communion 62, 246 conceptualizing ambivalence 97–104 constructivist views 40 contamination sequences 246–247 contemporary discourse reframing 186–187 continuous-time model 288–289 conventional approach of structure 37 coping 368 core social identities 203–208 covariates 265–267, 271–272, 288–290 Cox’s model 289–290, 293–295 cultural variations 76, 212–213 culture 76, 250–253 data-analytical approach 375–377 data-analytical tool 377, 380 data collection 376 data-mining approaches 283, 309 data-mining method 300 data modeling culture 26, 264, 278, 300 decisional ambivalence 116–117 demographic analysis 283–288 demographic density 262 demographic events significance 155 demographic evolution 285 demography 261–263, 283 developmental data analysis steps 317 developmental psychology 315, 382 developmental regulation model 43 developmental scores 333–337 developmental shapes 324–333, 338–344 dialectical processes 69–70 discrete-time model 289 dispositional traits 239–242 diversification context of research status 108–110 divorce 110, 158–161, 168–169

SUBJECT INDEX double chain Markov model (DCMM) 295–296 dynamic determinants 344–348 dynamic structure 37–38 early adulthood through midlife 45–47 early childbearing 157–159 early marriage 166 early work investment 136–140 educational attainment 131, 134, 142, 144 educational influence 335–337 educational orientations 136–140 ego strength 61–62 emancipation 102 emerging adults 244 emotional valence 222, 225, 228–229 end of the life course 57, 60, 83–84 Note: See also last chapters of life environmental proactivity 61 episodic memory recollection 227 episodic recollection 225–227 Erikson’s theory 121 event-based approach 263–264 event-based culture 265–273 event history analysis (EHA) 263, 265–269, 378 event-history model 300 event history regression models 288–295 event-sequential association rules mining 300–302 events to echo effects 181–182 evidence in life transitions 64–65 expectations 331–333 extended transition 211 external causality 18 extravert person 240 factor-analytic study 240 family relationship 45, 364 fertility project types 111 first-order homogenous transition matrices 296–298

Subject Index fitting latent difference score models 347–352 Five-Factor Theory (FFT) 371–373 formal aspects of life course 11–20 friends and acquaintances 210 gender 176 gender difference 5–6, 138 general developmental model 316 generalized linear model 293 generativity 61–62, 107, 121, 241, 248, 374 gerontology 20, 47–48 Gini heterogeneity index 275 good life course 72–77 group differences 160, 318, 338–344 growth curve 317, 320, 340 growth-curve analyses 321, 324–325 growth mixture model 342–343 hazard model 267–268, 309 hidden Markov model (HMM) 295, 297–298 hierarchical linear model 335, 378 highly generative American adults 247–250 historical force types 178 holistic approaches 26, 263 holistic culture 273–275 homogenous-Markov model 295 human agency 35, 131, 251 human individuality 238–242, 247 human infants 242 identity 24, 81–83, 100, 119, 197, 208–209, 211, 366, 370–374 identity change 24, 83, 197, 211, 383 identity formation process 209–210, 243 incomplete information 323–324 individual agency 362–365 individual differences 61, 221, 240

407 individual differences predictors modelling 333–337 individual life course 273–274, 362 individual orientations and actions 133 individuals across time 371–372 individuals and life stories 374 induction tree 303–308 inheritance 118–119 inline transitions 174–175, 369 innovative methods of analysis 25–27 institutional framing 50–51 integrative life narrative 241 interdependent lives see linked lives interdisciplinarity 361, 380–386 interdisciplinarity pathways 9–10 interdisciplinary perspectives 3 intergenerational ambivalence 102 intergenerational relations 93, 95–97 intergenerational relations research module 100–104 intergenerational-social transition 303 interlacing transition 176, 178, 182, 185 Interlacing transitions and contemporary discourse 186 internal causality 18 International Union for the Scientific Study of Population 279, 286 introversion/extraversion 239–240 Keane and Wolpin model 273 knowledge provided by induction tree 306–307 Knowledge-Verbal score 348 Kohlberg’s theory 198–199 Konstanz Inheritance Survey 118 labor force 136–137 labor market 36, 111, 134, 150, 167, 176–178, 182 last chapters of life 71, 81 latent basis curves 328–329 latent curve model (LCM) 315, 317, 351

408 latent difference scores modeling 344–345 latent mixture model 318, 343–344 latent scores 325, 335–337 layered life course patterns 176–178 learning theory 198 leave taking 178–180 Liberal Market States 49–50 life course analysis 173, 261, 303, 369 life course context 375 life course data 265, 283, 375, 377 life course events 11, 17–20 life course pattern 176–178 life course perspective 83, 119, 173, 308 life course research 3–5, 361 life course shaping elements 261 life course sociology 36, 38 life course stages 11, 13–15 life course terminology 58 life course theories 63, 244 life course trajectories 76, 133, 267, 278 life course transition 155, 185, 197, 227, 383 life event inventory 223, 234 life event research 19 life events moderation 162–170 life narrative 239, 246–247 life review 120, 230 life-span developmental psychology 223, 315 life-span psychology 42 life stories 245–250 Life Story Interview 245 life table 285 life trajectory 373 lifetime period 219 linear growth models 324–327 linear spline models 324–331 linked lives 6, 12, 22 living context 217, 222 lone motherhood 182–187 long-suppressed tendencies 244 longitudinal-analytical tools 308

SUBJECT INDEX longitudinal data 26, 199, 299–300, 317, 320, 324, 338, 379 longitudinal life-span data 315 longitudinal-regression model 288 loose coupling 63–64, 66, 168 macro–micro links 5 Note: See also micro–macro links macrostructural perspective of life course 131–132 making sense of agency 67–69 management of change 368 marital disruption 158–159 marital status 182–185 Markov transition model 288, 295–296, 379 Marriage 18–19 masculine trait 61–63 maternity and parental leave 178 memorized life transition 219 memory and personality relationship 226–227 memory research 219 methodological approach 374 methodological preliminaries of research 104–105 methodological strategy 160 methods of analysis 25–27 micro–macro links 133 Note: See also macro–micro links microfication 47 middle age 224–225 mixed-effects model 27, 310, 334 mixture model 342–343 mixture transition distribution (MTD) 295 MLE parameter 337 mobility 285, 303 model-based expectations about growth 331 model-restricted fuzzy-set cluster analysis 342 modeling approaches 267

Subject Index monothetic divisive algorithm (MDA) 274 moral reasoning 199 mortality tables 283–284 mother–adult daughter relationship 114 motherhood 179–184 movement through social space 384 MTD model 295 multi-level model 26, 378–379 multi-level modeling 290–293 multilevel and multiprocess modeling 265–269 multilevel growth model 334–335 multiple ambivalence 117 multiple group perspective 338–340 multiple-regression analyses 223 multiprocess model results 268 narrative 250–253 narrative approaches 13, 237–239, 252 narrative identities 241–244 national identity 202, 249–250 negative transition 228 neuroticism 223, 225–226 new father 187 new–old view of intergenerational relations 93 nonlinearity 328–331 non-normative event 157–159, 364 non-normative life course 155 nonparametric stage 271 Non-Verbal abilities 348 normative transition 382–383 objective facts versus subjective reconstructions 218–221 observed data 321 occupational career structures 178 off-time transitions 112 old age 47–49, 222–229 ontogenesis 13 ontogenetic push factors 385

409 operationalizing the life course 77–81 optimal matching 26, 274, 277–278, 309, 379 overcoming resistance 64 parent–child relationship 93, 96, 106–107, 109 part-time work 142, 179, 363 partnership translations 228 past life transition 219, 222 Person’s Education 337 personal ambivalence 116–117 personal attribute models 38 personal constructs 132 personal relationship 96, 116 personality 101, 133, 198, 205, 217, 221–222, 238, 245, 363, 370 personality and identity 370–371 personality–cognition relation 226 personality dimensions 371–372 personality psychology 239, 245, 371, 384 physical displacement 212–213 planful competence 21, 40, 68, 132 political engagement 209 political identity 209–210 polynomial model 328, 333 polynomial nonlinearity in growth 328 Population Index database 296 post-modernity 98 potential predictor 303–304 predictors of early work 136 pre-life course perspective 84 principle of life course 63–64 production of life 60–61 program evaluation literature 269–272 propensity scores matching 271 proportional-hazard model 289 proportional hazard-odd model 289 prototypical transitions 198–200 psychological analyzer 365–370 psychological well-being 223–225

410 random-coefficients 334–335 random effect model 292, 310 rational choice theory 65 reconstructive recollection 229 redemption sequences 246–247 redemptive self themes 248–249 regression analyses 112, 225, 277 regression tree 303 regulation of well-being 217–218 residuals 331 response trajectories 169–170 responsibility 65–67 retraditionalization 369 retrospective data 376 rituals of transition 208, 212 role of the state 187 SCT and SIT 202 second demographic transition 285 second-order homogenous transition 296 self-categorisation 200–201, 373 Self-Categorisation Theory (SCT) 24, 197, 201, 203 self-concept 372 self-defining memories 246 sequence analysis 263, 274 sequence-mining approach 300 sequential institutionalization 51 shaping life courses 261 shared conception 72 shared heterogeneity 290–293 simple linear regression 291 simultaneous institutionalization 51 Six-Foci Model (SFM) 372, 374, 383 slopes as outcomes 335 small-world phenomenon 204 social address model 38 social analyzer 365–370 social behavior 363 social categorisation 200, 203–204 social demographers 9 social environment 372

SUBJECT INDEX social identity 24, 197, 200, 208, 211, 373 social identity change 197, 211–213 Social Identity Theory (SIT) 200 social-mobility analysis 299 social niche model 38 social psychological perspective 131, 370, 383 social psychology 8, 14, 16, 18, 197, 200, 385 social sciences 157, 237, 283 social structure 35, 37–39, 41, 64, 69, 362 social theory 59 social transition analysis with induction trees 303 socialization 9, 70, 367, 381 socioeconomic achievement trajectories 140–144 sociology 67 socio-personal framing of transition 174 solidarity 102 solidarity perspective 95 specific life periods 44–49 stage theories 237–238 stage transition 199 statistical modeling of life events 288 statistical models 351 status attainment 135 status changes 81–83 status definitions to ambiguity 181 stopped working 179–180 stories 250–251 story development 242–245 storytelling 239, 242, 247, 252 strands of life course 185–189 stratification algorithmic modeling culture 264 structural agency 363–365 structural ambivalence 117 structural equation modeling (SEM) techniques 317 structure and agency see agency and structure

411

Subject Index structure 36–52 studying lives in time 237 subjective meanings of lives 238 survival analysis 378 teenage childbearing 157–158, 168–169, 269–270 temporal coherence 243 the good death 72–73 thematic coherence 243–244 theoretical bridges 383–384 theoretical linkage of interdisciplinarity 10–11 theory-making in social sciences 252 timeframes 186 time-stamped event 288 timing of events 156, 174, 369 traditional demographic analysis 286 trait conception 238 transition 11–13, 21–24, 188–189, 197, 365–370, 381–385 transition analysis instruments 187–188 transition characteristics 15 transition process 174, 211–213 transition reduction 175–176 transition to adulthood 131, 144–149, 156, 225, 262, 269 transitions and life events 218–221 Transitions and Life Perspectives in Middle Age 224

transversal substantive themes 20–27 transversal themes 381–383 tree-growing principle 304 true novel 243 turning points 110–113, 178 two-staged latent class analysis 144 U.N. Convention on the Rights of the Child 45, 48 understanding lives 252 unexplained variance 63–64, 83–84 United Nations Principles for Older Persons 48 unplanned pregnancy 163–164 Verbal-Knowledge score 320–321, 331 Verbal–Non-Verbal bivariate coupling model 347 vignettes use 105 Warmr process 301–302 welfare-state regime 49–50 well-being 222 work investment 136–144, 149–150 work pattern 137–138, 140–142, 146 World Fertility Survey (WFS) 262 Youth Development Study (YDS), the 135–136 youth 46, 94, 100, 134, 138, 139


412

Towards an Interdisciplinary Perspective on the Life Course, Volume 10 (Advances in Life Course Research)

Children at the Millennium (Advances in Life Course Research) (Advances in Life Course Research)

New Frontiers in Socialization (Advances in Life Course Research) (Advances in Life Course Research)

Changing Life Patterns in Western Industrial Societies, Volume 8 (Advances in Life Course Research)

Steps towards Life: A Perspective on Evolution

Steps towards Life: A Perspective on Evolution

A Life-Course Perspective on Migration and Integration

Disability and the Life Course: Global Perspectives

Social Problems across the Life Course

Aging, Society, and the Life Course:

Teamwork: An Interdisciplinary Perspective

On Course

Phraseology: An interdisciplinary perspective

Institutional Abuse: Perspectives Across the Life Course

Social Identities Across the Life Course

Advances in Accounting Behavioral Research, Volume 10

Advances in Ecological Research, Volume 10

Advances in Cancer Research, Volume 10

ADVANCES IN CATALYSIS VOLUME 10, Volume 10

Paths to Successful Development: Personality in the Life Course

Psyche and Eros: Mind and Gender in the Life Course

Encyclopedia of the Life Course and Human Development, Volume 3: Later Life, Appendices, Index

Greek: An Intensive Course

Latin: an intensive course

AVR: An Introductory Course

Sinhalese : An Introductory Course

Mechanics: An Intensive Course

Mechanics: An Intensive Course

Logic: An Introductory Course

Towards an Interdisciplinary Perspective on the Life Course, Volume 10 (Advances in Life Course Research)

Children at the Millennium (Advances in Life Course Research) (Advances in Life Course Research)

New Frontiers in Socialization (Advances in Life Course Research) (Advances in Life Course Research)

Changing Life Patterns in Western Industrial Societies, Volume 8 (Advances in Life Course Research)

Steps towards Life: A Perspective on Evolution

Steps towards Life: A Perspective on Evolution

A Life-Course Perspective on Migration and Integration

Disability and the Life Course: Global Perspectives

Social Problems across the Life Course

Aging, Society, and the Life Course:

Teamwork: An Interdisciplinary Perspective

On Course

Phraseology: An interdisciplinary perspective

Institutional Abuse: Perspectives Across the Life Course

Social Identities Across the Life Course

Advances in Accounting Behavioral Research, Volume 10

Advances in Ecological Research, Volume 10

Advances in Cancer Research, Volume 10

ADVANCES IN CATALYSIS VOLUME 10, Volume 10

Paths to Successful Development: Personality in the Life Course

Psyche and Eros: Mind and Gender in the Life Course

Encyclopedia of the Life Course and Human Development, Volume 3: Later Life, Appendices, Index

Greek: An Intensive Course

Latin: an intensive course

AVR: An Introductory Course

Sinhalese : An Introductory Course

Mechanics: An Intensive Course

Mechanics: An Intensive Course

Logic: An Introductory Course

Recommend Documents