Advances in Experimental Social Psychology, Volume 25

ADVANCES IN

Experimental Social Psychology

VOLUME 25

This Page Intentionally Left Blank

ADVANCES IN

Experimental Social Psychology

EDITED BY

Mark P. Zanna DEPARTMENT OF PSYCHOLOGY UNIVERSITY OF WATERLOO WATERLOO. ONTARIO, CANADA

VOLUME 25

ACADEMIC PRESS, INC. Harcourt Brace Jmnwich, Publishers San Diego New York Boston London Sydney Tokyo Toronto

This book is printed on acid-free paper. @ Copyright 0 1992 by ACADEMIC PRESS,INC. All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy. recording, or any information storage and retrieval system, without permission in writing from the publisher.

Academic Press, Inc. 1250 Sixth Avenue, San Diego, California 92101-4311 United Kingdom Edition published by

ACADEMIC PRESS LIMITED 24-28 Oval Road. London NW 1 7DX

Library of Congress Catalog Number: 64-23452 International Standard Book Number: 0-12-015225-8

PRINTED

m THE u r n STATES OF AMERICA

929394959697

BB

9 8 7 6 5 4 3 2 1

CONTENTS

Contributors

.....................................................

ix

Universals in the Content and Structure of Values: Theoretical Advances and Empirical Tests in 20 Countries Shalom H. Schwartz 1. 11. 111.

IV. V. VI. VII.

Introduction .... Theory Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Empirical Studies . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . Theory Assessment . . .................. .......... Studying the Antecedents and Consequences Concluding Remarks . . . . . . . . . . . . . . . . . Appendix: Values Included in the Survey Ins ...................... References ........................... ...............

1

3 16 37 51 59

60 62

Motivational Foundations of Behavioral Confirmation Mark Snyder I. 11. 111.

IV. V.

v1. v11. VIII.

IX. X.

........... Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Phenomenon of Behavioral Confirmation in Social Interaction . . . . . . . . . . ........... The Mechanisms of Behavioral Confirmation ........... The Strategy of Functional Analysis . . . . . . . The Functions Engaged in Behavioral Confirmation Scenarios . . . . . . . . . . . . . . Different Functions for Perceiver and Target? . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Matter of Behavioral Disconfirmation . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . Theoretical and Practical Implications of the Functional Analysis . . . . . . . . . . . The Functional Analysis in Context ........................ . . . . . . . . . . . Conclusion ...................... ................ ........... ........... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V

67 67 69 70 73 77 83 91 95 105 107

vi

CONTENTS

A Relational Model of Authority in Groups Tom R. 'Ifrlerand E. Allan Lind I. 11.

111. IV.

The Social Psychology of Authority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Social Psychology of Procedural Justice . . . . . . . . . . . . . . . . . . . .. , . . . . . . Conclusions ....................................................... Appendix: Measures of Outcome Favorability, Distributive Justice, Pmcedural Justice, Legitimacy, and Related Variables . . . . . . . . . . . . . . . . . . . . References ........................................................

I15 137 162 167 186

You Can't Always Think What You Want: Problems in the Suppression of Unwanted Thoughts Daniel M . Wegner 1. 11.

111. IV. V. VI.

Introduction . . .. .. . .. .. .. . . . .. .................... Suppression in the Sh Instigation of Suppression ......................... Thought Suppression Pmces Postsuppression Processes . . Conclusion . . . . . . . . . . . . . . . . . . . ......................................... References . . . . . . . .

193 194 196 202 21 1 217 220

Affect in Social Judgments and Decisions: A Multiprocess Model Joseph Paul Forgas I. 11. 111.

IV. V.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , . . . . . . .. . . . . . . . . . ... The Multipmess Model of Affect in Social Judgments . . . . . . . . . . . . . . . . . . . Mechanisms of Mood Congruence in Judgments . . . . . . . . . . . . . . . . . . . . . . . . . The Research Program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References ........................................................

227 232 242 249 267 269

The Social Psychology of Stanley Milgram Thomas Blass I. 11.

Introduction ....................................................... Underlying Characteristics of His Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.

277 278

111. IV. V.

CONTENTS

vii

Shortcomings and Flaws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Enduring Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . , . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

290 293 317 320

The Impact of Accountability on Judgment and Choice: Toward a Social Contingency Model Philip E. Tetlock 1. 11.

111. IV. V. VI.

Introduction . _............_... .......... The Metaphorical Roots of the Cognitive Research Program . . . . . . . . . . . . . . . The Political Research Program . . . . . . . . , . . . . . . . . . . . . . . . . . Normative Assessments of Coping Strategie ..........._.._... Another Look at Normative Issues . . . . . . . . . . . . . Concluding Comments . . . ..... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33 I 333 335 347 359 368 37 1

........................................................... ........... .............................

377 391

Index Contents of Other Volumes

This Page Intentionally Left Blank

CONTRIBUTORS

Numbers in parentheses indicate the pages on which the authors' contributions begin.

THOMAS BLASS (277), Department of Psychology, University of Maryland, Baltimore County, Catonsville, Maryland 2 1228 JOSEPHPAULFORGAS (227), School of Psychology, University of New South Wales, Sydney, New South Wales 2033, Australia

E. ALLANLIND (1 15), American Bar Foundation, Chicago, Illinois 6061 1 SHALOM H. SCHWARTZ(l), Department of Psychology, The Hebrew University of Jerusalem, Jerusalem, Israel 9 1905 MARKSNYDER(67), Department of Psychology, University of Minnesota, Minneapolis, Minnesota 55455 PHILIPE. TETLOCK(33 I), Department of Psychology, University of California, Berkeley, California 94720 TOM R. TYLER (115), Department of Psychology, University of California, Berkeley, California 94720

M. WEGNER (193), Department of Psychology, University of Virginia, DANIEL Charlottesville, Virginia 22903

ix

This Page Intentionally Left Blank

UNIVERSALS IN THE CONTENT AND STRUCTURE OF VALUES: THEORETICAL ADVANCES AND EMPIRICAL TESTS IN 20 COUNTRIES Shalom H. Schwartz*

I. Introduction “The value concept, more than any other, should occupy a central position . . . able to unify the apparently diverse interests of all the sciences concerned with human behavior.” These words, proclaiming the centrality of the value concept, were written by a psychologist (Rokeach, 1973, p. 3), but similar stands have been taken by sociologists (e.g., Williams, 1968) and anthropologists (e.g., Kluckhohn, 1951). These theorists view values as the criteria people use to select and justify actions and to evaluate people (including the self) and events. We, too, adopt this view of values as criteria rather than as qualities inherent in objects. This article discusses work that is part of a larger project intended to explore the importance of values in a wide variety of contexts. The project addresses three broad questions about values. First, how are the value priorities of individuals affected by their social experiences? That is, how do the common experiences people have, because of their shared locations in the social structure (their education, age, gender, occupation, etc.), influence their value priorities? And, how do individuals’ unique experiences (trauma, relations with parents, immigration, etc.) affect their value priorities? Second, how do the value priorities held by individuals affect their behavioral orientations and choices? That is, how do value priorities influence ideologies, attitudes, and actions in the political, religious, environmental, and other domains? *This article was written with the collaboration of Sonia Roccas and Lilach Sagiv of the Department of Psychology of The Hebrew University. They have contributed to the development of the research reported here, performed most of the analyses, and critiqued successive versions of the manuscript. ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY. VOL. 25

1

Copyright 0 1992 by Academic R s s . Inc. All rights of reproduction in any form R X N C ~ .

2

SHALOM H. SCHWARTZ

To date, aspects of these first two questions have most often been studied within single cultures (e.g., Feather, 1975; Homer & Kahle, 1988; Levy, 1990; Rokeach, 1973). However, the current project is also concerned with the extent to which systematic associations among values, social experience, and behavioral orientations hold across cultures. Associations that are universal would point to fundamental processes relating value priorities to their antecedents and consequences. Associations that are culture specific would direct attention to important culture-linked moderators. The identification of such moderators would enrich our understanding of how value priorities function. Third, the project addresses the question of cross-cultural or cross-national differences in value priorities and seeks to identify some of their causes and effects. For example, how do the value priorities of matched groups from countries as diverse as China, Poland, Spain, the United States, Venezuela, and Zimbabwe differ? What aspects of the ecology, history, technology, and social and political structures of the societies from which these groups are drawn might account for these differences in value priorities? How might the differences in value priorities account for differences among the societies in other domains, such as educational policies, political involvement, health, law, etc? In tackling this broad question, we have taken whole cultural groups as our unit of analysis, following in the footsteps of Hofstede’s (1980) monumental work. Furthermore, in our search for causal links, we have adopted the ecocultural framework that guides much contemporary cross-cultural psychology research (Berry, 1975, 1986; Georgas, 1988; Segall, Dasen, Berry, & Poortinga, 1990). Work on this third question raises the problem of whether dimensions of values that are meaningful in the study of individuals might also apply at the cultural level. In the current discussion, we limit our analyses to dimensions of values at the individual level, which is the appropriate level for the study of the first two questions. Preliminary analyses at the cultural level indicate that somewhat different value dimensions emerge when cultures are taken as the unit of analysis. Nonetheless, in contrast to Hofstede’s (1980) findings, the dimensions derived at the two levels in our research appear to be closely related (Schwartz, in press). In this article, we deal with a set of basic issues that must be resolved before the three broad questions elaborated above can be researched effectively: 1. Value content: It is necessary to identify first the substantive content of

human values. What types of values are likely to be recognized and used to form priorities within and across all cultures? What might determine the nature of the content of values? Do values form some universal set of types? 2. Comprehensiveness: Have we identified a comprehensive set of values types? That is, does the set include all the types of values to which individuals are likely to attribute at least moderate importance as criteria of evaluation? If the value set is not comprehensive, studies of the correlates

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

3

of value priorities will be compromised: Influential values that might counterbalance or outweigh the values that were measured would necessarily be overlooked, so the assessed priorities would be distorted. 3. Equivalence of meaning: Do the values have the same or similar meanings among the differing groups of persons under study? Minimal equivalence of meaning is a sine qua non for effective cross-cultural comparison. Even within societies, equivalence of meaning cannot be taken for granted in comparisons of groups differentiated by age, gender, education, etc. 4. Value structure: Consistent conflicts and compatibilities among values (e.g., conflict between independence and conformity; compatibility between equality and helpfulness) may point to a meaningful structure that underlies relations among single values. Does such a value structure exist? Is it universal? Identification of a universal structure would permit the derivation of basic value dimensions that could be used for purposes of comparison. Evidence that groups have similar value structures would justify assumptions about equivalence of meaning of the values comprising the structure. Variations in value structures across groups would suggest interesting differences in the ways these groups organize their understanding of the world. Note that srrucrure refers to the relations of conflict and compatibility among values, not to their relative importance to a group or individual. The latter is referred to as value priorities or hierarchies. The first steps in tackling these basic issues were taken by Schwartz and Bilsky (1987, 1990). They proposed a tentative theory of the universal content and structure of human values, which they tested with data from seven countries. In this article, we revise that theory with numerous modifications and extensions. We also introduce a new values instrument developed on the basis of the revised theory. We then assess the viability of the revised version of the theory with data gathered from 40 samples in 20 countries. Finally, we propose ways to use the theory and instrument to generate and test hypotheses about the antecedents and consequences of value priorities.

11. Theory Development

A. THE EARLY VERSION OF THE THEORY'

Schwartz and Bilsky (1987, 1990) generated a conceptual definition of values that incorporates the five formal features of values recurrently mentioned in the 'This overview is necessarily brief. For a full presentation, see Schwartz and Bilsky (1987, 1990).

4

SHALOM H. SCHWARTZ

literature. Values (1) are concepts or beliefs, (2) pertain to desirable end states or behaviors, (3) transcend specific situations, (4) guide selection or evaluation of behavior and events, and (5) are ordered by relative importance. Values, understood this way, differ from attitudes primarily in their generality or abstractness (feature 3) and in their hierarchical ordering by importance (feature 5) (cf. Bem, 1970, and Rokeach, 1973, but contrast Levy and Guttman, 1974). In addition to the formal features of values, Schwartz and Bilsky (1987, 1990) proposed that the primary content aspect of a value is the type of goal or motivational concern that it expresses. They derived a universal typology of the different contents of values by reasoning that values represent, in the form of conscious goals, three universal requirements of human existence to which all individuals and societies must be responsive: needs of individuals as biological organisms, requisites of coordinated social interaction, and survival and welfare needs of groups. From an evolutionary viewpoint (Buss, 1986), these goals have crucial survival significance. Eight distinct motivational types2 of values were derived from the three universal human requirements. For example, the motivational type resfricfiveconformity was derived from the prerequisite of smooth interaction and group survival, which prescribes that individuals restrain impulses and inhibit actions that might hurt others. The eight types derived were prosocial, restrictive conformity, enjoyment, achievement, maturity, self-direction, security, and power. Responses to the Rokeach Value Survey (Rokeach, 1973), or to variants of it, supported the assumption that individuals in seven different countries experienced the first seven values types as distinct. Additional values presumed to measure power were included for one sample and emerged as distinct. Interrelations among all the Rokeach values were accounted for by the hypothesized motivational types; with these 36 values, no further motivational types emerged empirically (Schwartz & Bilsky, 1990). In addition to propositions regarding the universal content of values, the theory specified a set of dynamic relations among the motivational types of values. Actions taken in the pursuit of each value type have psychological, practical, and social consequences that may be compatible or may conflict with the pursuit of other value types. For example, actions intended to express obedience (restrictive conformity) are likely to conflict with actions in pursuit of independence (self-direction), but they are compatible with actions promoting social order (security). Based on an analysis of the likelihood of conflict or compatibility between each pair of value types, an overall structure of relations among the seven value types was postulated. Findings for the samples studied suggested that the dynamics of conflict and compatibility among the motivational types of values had much in common across the seven countries. For example, there was strong evidence for the 2We have relabeled what were formerly called morivarional domains as motivational types.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

5

universality of compatibility among value types that support smooth social relations (security, prosocial, restrictive conformity), among those concerned with self-enhancement (achievement, enjoyment), and among those expressing comfort with or reliance on one’s uniqueness (maturity, self-direction). There was also strong evidence for the universality of value conflict between an emphasis on independence of thought and action versus conforming self-restraint (selfdirection vs. restrictive conformity) and between concern for others versus pursuit of personal success ( prosocial vs. achievement).

B. MODIFICATIONS OF THE THEORY: VALUE CONTENT With regard to value content, we have modified the earlier version of the theory in several ways. First, we define and derive three potentially universal motivational types of values that were mentioned only briefly by Schwartz and Bilsky (1990), and we specify methods for measuring them. To date, the tradition and stimulation value types have neither been derived fully nor studied empirically, and the power type has been examined only tentatively. Second, we develop the possibility that spirituality may constitute another universal type with unusual characteristics, and we propose ways to examine this type. Third, based on empirical findings and on conceptual considerations, we suggest modifications of the definitions and contents of four of the earlier types (enjoyment, maturity, prosocial, security). These modifications are intended to sharpen the meanings of the types, thereby clarifying their origins in universal requirements and their relations to other value types. We relabel three of these earlier types to reflect their changed meanings. We now present the hypothesized motivational types of values according to the revised theory, elaborating only on modifications. Following the description of each type, we list in parentheses the specific values included in our new survey instrument, primarily in order to measure that type. The meaning of each type is thus made more concrete and explicit. For some types, other values were judged a priori to express the goal of that type as a secondary meaning. Such values are endowed with multiple meanings that link them to more than one value type. Table I summarizes the postulated links of each of the single values in our survey with the motivational types of values. It indicates the primary motivational type associated with each value, and, where appropriate, a secondary type as well.

I . Self-Direction The defining goal of this value type is independent thought and actionchoosing, creating, exploring. Self-direction was derived from organismic needs for control and mastery (e.g., Bandura, 1977; Deci, 1975; White, 1959) and

6

SHALOM H. SCHWARTZ TABLE I OF SINGLE VALUES POSTULATED ASSOCIATIONS WITH MOTIVATIONAL TYPESOF VALUES Primary motivational type and value Self-direction Freedom Creativity Independent Choosing own goals curious Self-respect Stimulation An exciting life A varied life Daring Hedonism Pleasure Enjoying life Achievement Ambitious Influential Capable Successful Intelligent Self-respect Power Social power Wealth Authority Preserving my public image Social recognition Security National security Reciprocation of favors Family security Sense of belonging Social order Healthy Clean Conformity Obedient Self-discipline Politeness Honoring of parents and elders Tradition Respect for tradition

Secondary motivational type

Power -

Self-direct, universalism -

Security Achievement Conformity -

Hedonism Conformity -

-

Tradition

(continued)

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

7

TABLE I (Continued) ~

Primary motivational type and value Devout Accepting my portion in life Humble Moderate Spirituality A spiritual life Meaning in life Inner harmony Detachment Benevolence Helpful Responsible Forgiving Honest Loyal Mature love True friendship Universalism Equality Unity with nature Wisdom A world of beauty Social justice Broad-minded Protecting the environment A world at peace

Secondary motivational type Spirituality Spirituality Conformity Security

Universalism Universalism, security -

Conformity

Conformity Universalism Spirituality -

-

interactional requirements of autonomy and independence (e.g., Kluckhohn, 1951; Kohn & Schooler, 1983; Morris, 1956). Although the goal of this type remains unchanged, we have changed the set of values intended to measure it so as to fit this goal more closely (creativity, freedom, choosing own goals, curious, independent).

2 . Stimulation Stimulation values derive from the presumed organismic need for variety and stimulation in order to maintain an optimal level of activation (Berlyne, 1960; Houston & Mednick, 1963; Maddi, 1961). This need is probably related to the needs underlying self-direction values (cf. Deci, 1975). Biologically based variations in the need for stimulation and arousal, conditioned by social experience, may produce individual differences in the importance of stimulation values. Such

8

SHALOM H. SCHWARTZ

a biological base has been claimed for thrill-seeking (Farley, 1986), a personality variable related to stimulation values. The motivational goal of stimulation values is excitement, novelty, and challenge in life (a varied life, an exciting life, daring).

3 . Hedonism This value type, derived from organismic needs and the pleasure associated with satisfying them, is referred to by scholars from many disciplines (e.g., Bentham, 1938/1948; Freud, 1933; Moms, 1956; Williams, 1968). It was formerly called “enjoyment” in order to justify including two values from the Rokeach list, happiness and cheerful. Neither of these values is convincingly derived from organismic needs, however. By dropping them we can define the motivational goal of this type more sharply as pleasure or sensuous gratification for oneselfj (pleasure, enjoying life).

4 . Achievement The defining goal of this value type is personal success through demonstrating competence according to social standards. Competent performance is a requirement if individuals are to obtain resources for survival and if social interaction and institutional functioning are to succeed. Achievement values are mentioned in many sources (e.g., Maslow, 1959; Rokeach, 1973; Scott, 1965). As defined here, achievement values emphasize demonstrating competence in terms of prevailing cultural standards, thereby obtaining social approval. This differs from McClelland’s (McClelland, Atkinson, Clark, & Lowell, 1953) definition of achievement motivation to meet internal standards of excellence. The latter is more closely related to self-direction values (ambitious, successful, capable, influential). 5. Power Power values are probably grounded in more than one type of universal requirement. The functioning of social institutions apparently requires some degree of status differentiation (Durkheim, 189311964; Parsons, 1957), and a dominance/submission dimension emerges in most empirical analyses of interperson3Happiness is assuredly an important value for many people. However, its empirical location in past research, very near the center of the multidimensional projection of the values space in each sample, indicated that its meaning is complex and ambiguous. Happiness can apparently be attained through the successful pursuit of any of the different types of values and it is positively correlated with them all.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

9

a1 relations both within and across cultures (Lonner, 1980). To Justify this fact of social life, and to motivate group members to accept it, groups must treat power as a value. Power values may also be transformations of the individual needs for dominance and control identified by analysts of social motives (e.g., Korman, 1974; Schutz, 1958). Power values have been mentioned by value analysts as well (Allport, 1961; Gordon, 1960). We view the central goal of power values as attainment of social status and prestige, and control or dominance over people and resources (authority, wealth, social power, preserving my public image, social recognition). Both power values and achievement values focus on social esteem. However, achievement values (e.g., successful, ambitious) emphasize the active demonstration of competence in concrete interaction, whereas power values (e.g., authority, wealth) emphasize the attainment or preservation of a dominant position within the more general social system. 6 . Security

The motivational goal of this value type is safety, harmony, and stability of society, of relationships, and of self. It derives from basic individual and group requirements (cf. Kluckhohn, 195 1; Maslow, 1959; Williams, 1968). One might postulate the existence of a single security value type or of two separate types, individual and group. This depends upon whether we assume that some security values serve primarily individual interests (e.g., healthy) and others serve primarily collective interests (e.g., national security), or that even values that refer to collectivities express, to a significant degree, the goal of security for self (or those with whom one identifies). In order to address this question empirically, we included individual security values, absent in past research, in our new value survey (social order, family security, national security, reciprocation of favors, clean, sense of belonging, healthy). 7. Conformiry4

The defining goal of this value type is restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms. It is derived from the requirement that individuals inhibit inclinations that might be socially disruptive if interaction and group functioning are to run smoothly. Conformity type values are mentioned in virtually all value analyses (e.g., Freud, 1930; Kohn & Schooler, 1983; Moms, 1956; Parsons, 1957). As we define them, conformity values emphasize self-restraint in everyday interaction, 4Formerly called resrricrive conformiry.

10

SHALOM H. SCHWARTZ

usually with close others (obedient, self-discipline, politeness, honoring parents and elders). 8. Tradition

Groups everywhere develop symbols and practices that represent their shared experience and fate. These eventually become sanctioned as traditions and customs that are valued by the group’s members (Sumner, 1906). Traditional modes of behavior become symbols of the group’s solidarity, expressions of its unique worth, and presumed guarantors of its survival (Durkheim, 1912/1954; Parsons, 1957). Traditions most often take the form of religious rites, beliefs, and norms of behavior (Radcliffe-Brown, 1952). The motivational goal of tradition values is respect, commitment, and acceptance of the customs and ideas that one’s culture or religion impose on the individual (respect for tradition, humble, devout, accepting my portion in life, moderate).

9. Spirituality Theologians, philosophers, and sociologists of religion emphasize that the basic rationale for traditional creeds and customs is to endow life with meaning and coherence in the face of the seeming meaninglessness of everyday existence (e.g., Buber, 1958; Niebuhr, 1935; Tillich, 1956; Yinger, 1957). Most religions supply answers to the question of the ultimate meaning of reality by referring to some supernatural being or force; nonreligious perspectives, such as humanism, locate sources of meaning in the natural world (Glock & Stark, 1965; Kaplan, 1961). These answers refer to what is commonly known as spiritual concerns. Hence, we label the relevant value type spirituality. If answering the question of ultimate meaning is a basic human need (Coles, 1990; Niebuhr, 1935; King, 1954), then spirituality values may constitute another universal type. The motivational goal of such values would be meaning and inner harmony through the transcendence of everyday reality. The idea that spiritual values are a universal type raises two problems. First, the pursuit of meaning and coherence as described by theologians and philosophers may entail a level of sophisticated, effortful thought that is beyond that in which most people typically engage. Instead, most people may satisfy their need for coherence through pursuing tradition, security, and conformity values. Thus, spirituality values may not serve as guiding principles for all people. Second, it is most likely that spirituality is represented by different values for different groups. Examination of literature on the major Eastern and Western religions and philosophies reveals that spiritual values take widely varying forms. Meaning or coherence may be found, for example, through contact with the supernatural, unity with nature, social action on behalf of one’s group,

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

11

detachment from material cares and personal desires, or discovering one’s “true” self (Heschel, 1955; Kaplan, 1961; King, 1954). Hence, no particular set of spiritual values may be universally distinctive, even if a general spirituality type of value is universal. The set of potential spirituality values studied here was gleaned from widely varied sources. The values include a spiritual life, meaning in life, inner harmony, detachment, unity with nature, accepting my portion in life, and devout. Spirituality was viewed as the secondary meaning for the last three values. A consistent grouping of any subset of these values in the analyses from different cultures would constitute evidence both for the universal existence of a spirituality type and for its common meaning. The emergence of different consistent groupings in different culture areas would suggest that there are several types of spirituality values rather than a single, universal spirituality type. The absence of any consistent groupings would support two interpretations: Either the theological and philosophical goals of spirituality do not serve as guiding principles for most people, or these goals are subsumed within and expressed through other value types. Including presumed spirituality values, a motivational type whose universality we doubt, puts to the test our ability to discriminate empirically between universal and nonuniversal value types. 10. Benevolence This is a more narrowly defined version of the earlier prosocial value type. Whereas prosocial referred to concern for the welfare of all people in all settings, benevolence focuses on concern for the welfare of close others in everyday interaction. The narrow focus is more in keeping with Schwartz and Bilsky’s (1987) derivation of this type from the need for positive interaction in order to promote the flourishing of groups (cf. Kluckhohn, 1951; Williams, 1968) and from the organismic need for affiliation (cf. Korman, 1974; Maslow, 1959). The motivational goal of benevolence values is preservation and enhancement of the welfare of people with whom one is in frequent personal contact (helpful, loyal, forgiving, honest, responsible, true friendship, mature love).

I I . Universalism This value type includes the former maturiry value type and part of the former prosocial value type. The maturity type had not been derived a priori from universal human requirements, but it emerged empirically in all seven countries studied. Although we have linked it post hoc to Maslow’s (1959) notion of the self-actualized person, it did not seem related to any clear, universal motivation. We therefore examined the results from the first few samples studied here in

12

SHALOM H. SCHWARTZ

search of a partitioning of values that would be conceptually stronger. This search suggested the new value type we have labeled ~niversalism.~ The motivational goal of universalism is understanding, appreciation, tolerance, and protection for the welfare of all people and for nature. This contrasts with the narrower focus of benevolence values. The motivational goal of universalism values can be derived from those survival needs of groups and individuals that become apparent when people come into contact with those outside the extended primary group and become aware of the scarcity of natural resources. People may then realize that failure to accept others who are different and treat them justly will lead to life-threatening strife, and failure to protect the natural environment will lead to the destruction of the resources on which life depends. Although such values may be absent in the few remaining small, isolated, homogeneous cultures, they are likely to be recognized to some degree in virtually all others. Research on collectivist versus individualist cultures demonstrates the importance of distinguishing between the universalism and benevolence types of prosocial concern (Schwartz, 1990a, in press). Members of collectivist cultures tend to show great concern for the welfare of members of their own ingroup but relative indifference to the needs of outsiders. Members of individualist cultures tend to distinguish less sharply between ingroups and outgroups when responding to their needs (Triandis, 1990; Triandis, McCusker, & Hui, 1990). This suggests a pattern of much greater emphasis on benevolence than on universalism values in collectivist cultures and more equal emphasis on both value types in individualist cultures (broad-minded, social justice, equality, world at peace, world of beauty, unity with nature, wisdom, protecting the environment). We address three basic theoretical questions regarding the content of the 11 motivational types of values: Are all of the value types present as distinctive organizing principles in all samples? Do the same specific values constitute each motivational type in each culture? Are any other value types necessary to account for the organization of single values? Operationally, evidence for or against the existence of value types and the consistency of the single values that constitute them is provided by examination of two-dimensional projections of the multidimensional space that represents the correlations among all single values within each sample (see Section III,B ,3). Some variation in the distinctiveness of types and in the location of single values across samples seems likely. In that case, we can seek broader categories that 5We also reexamined the smallest space projections of the value intercorrelations for the seven countries analyzed in the past (Schwartz & Bilsky, 1987, 1990). applying the new distinction between universalism and benevolence value types. This new distinction fit the earlier data better than the former distinctionbetween maturity and prosocial value types had done. It yielded a clearer partitioning of values into two regions, with 30% fewer errors in the placement of single values.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

13

exhibit greater universality and can suggest shared organizing principles at a more abstract level.

C. MODIFICATIONS OF THE THEORY: DYNAMIC STRUCTURE OF VALUE RELATIONS The additions and revisions to the motivational types of values require a new set of hypotheses that specify the theorized dynamic relations among all the types. The first principle used to generate these hypotheses is the interests facet in the theoretical definition of values (Schwartz & Bilsky, 1987). If values are viewed as goals, then their attainment must serve the interests of the individual and/or of some collectivity. Values that serve individual interests are postulated to be opposed to those that serve collective interests. The interests facet organized relations among the values types in all seven countries studied (Schwartz & Bilsky, 1990).The three value types postulated to serve individual interests (enjoyment, achievement, self-direction) formed a contiguous region in the two-dimensional projection of the values space in each country. This region was opposed to another contiguous region formed by the three value types postulated to serve collective interests (prosocial, restrictive conformity, security). Maturity values, postulated to serve both types of interests, were located on the boundary between these regions. Using the same reasoning, we hypothesize that the five value types that serve primarily individual interests (power, achievement, hedonism, stimulation, selfdirection) form a contiguous region opposed to another contiguous region formed by the three value types that serve primarily collective interests (benevolence, tradition, conformity). Universalism and security, as now defined, serve both types of interests and are therefore hypothesized to be located on the boundaries between these regions. The specific location of spirituality may vary depending on the particular values, if any, that constitute it, but almost all its potential values point to a general location in the collective region.6 interests facet should not be confused with the individualism-collectivism contrast in culture-level research (Hofstede, 1980; Schwartz, in press; Triandis, 1990). Many values that serve individual interests may indeed tend to receive greater priority in the way institutions function in individualist cultures, and many values that serve collective interests receive greater institutional priority in collectivist cultures. However, when we move from the level of individual values, on which the theory of this article focuses, to the level of cultural (institutionally organized) values, exceptions abound (Schwartz, in press). For example, the values wealth, social power, and authority, which primarily serve the interests of the individual person in our current analysis, tend to receive greater cultural priority in collectivist cultures in a culture-level analysis. Moreover, the values loyal, equality, and responsible, which primarily serve the interests of the collectivity in the current analysis, receive greater cultural priority in individualist cultures in a culture-level analysis.

14

SHALOM H. SCHWARTZ

More fine-tuned hypotheses are based on analyses of the underlying similarity of the goals defining each value type and of the psychological, social, and practical consequences of the simultaneous pursuit of each pair of value types. Each of the value compatibilities and conflicts predicted in past research includes types whose content has been modified in the revised theory. Our hypotheses are therefore new (though often related), and they are more detailed. The overall dynamic structure of relations among value types resulting from the specific, partly redundant hypotheses is presented in Fig. 1. Adjacent value types are postulated to be most compatible. Increasing distance around the circular order indicates decreasing compatibility and greater conflict. Value types that emerge in opposing directions from the origin are postulated to be in greatest conflict.

I , Compatibilities We reason that simultaneous pursuit of values from the following nine sets of types is compatible: (1) power and achievement-both emphasize social superiority and esteem; (2) achievement and hedonism-both are concerned with selfindulgence; (3) hedonism and stimulation-both entail a desire for affectively pleasant arousal; (4)stimulation and self-direction-both involve intrinsic motivation for mastery and openness to change; (5) self-direction and universalism-

\

UNIVERSALISM

SELF-DXRECTXON

L BENEVOLENCE

TRADITION

CONFORMITY

STIMULATION

1

SECURITY

HEDONISM

ACHIEVEMENT

POWER

Fig. 1. Theoretical structure of relations among motivational types of values.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

15

both express reliance on one’s own judgment and comfort with the diversity of existence; (6) universalism and benevolence-both are concerned with enhancement of others and transcendence of selfish interests (most spirituality values also share this concern); (7) tradition and conformity-both stress self-restraint and submission (some spirituality values also share this stress); (8) conformity and security-both emphasize protection of order and harmony in relations; (9) security and power--both stress avoiding or overcoming the threat of uncertainties by controlling relationships and resources. 2 . Conflicts We further reason that simultaneous pursuit of values from the following sets of types gives rise to strong psychological and/or social conflict: (1) selfdirection and stimulation versus conformity, tradition, and security-emphasizing own independent thought and action and favoring change conflict with submissive self-restriction, preservation of traditional practices, and protection of stability; (2) universalism and benevolence versus achievement and poweracceptance of others as equals and concern for their welfare interferes with the pursuit of one’s own relative success and dominance over others; (3) hedonism versus conformity and tradition-indulgence of one’s own desires contradicts restraint of one’s own impulses and acceptance of externally imposed limits; (4) spirituality versus hedonism, power, and achievement-the search for meaning through the transcendence of everyday reality contradicts the pursuit of sensual and material rewards. The full set of structural hypotheses overdetermines the order of value types shown in Fig. 1. The hypotheses place benevolence and tradition in adjacent positions, although no rationale was developed for their compatibility. Spirituality does not appear in Fig. 1 because of our doubts about the universality of a single spirituality type. The most likely location for a spirituality type is between benevolence and tradition, but a location between benevolence and universalism is also possible. This is based on considering the probable conflicts and compatibilities of the potential spirituality values. It also follows from the fact that the single values included to operationalize different conceptions of spirituality are closest to the definitions of one of these three value types.

D. TERMINAL VERSUS INSTRUMENTAL VALUES The first facet in Schwartz and Bilsky’s (1987) definition of values classified values as representing either terminal goals (end states, phrased as nouns, e.g., obedience) or instrumental goals (modes of behavior, phrased as adjectives, e.g., obedient). This distinction has been proposed both by philosophers (e.g.,

16

SHALOM H.SCHWARTZ

Lovejoy, 1950; Rescher, 1969) and by psychologists (e.g., Braithwaite & Law, 1985; Feather, 1975; Rokeach, 1973), although some have doubted its validity (e.g., Dewey, 1957). Schwartz and Bilsky (1990) found distinct regions occupied almost exclusively by terminal or by instrumental values in the multidimensional value space in all seven samples they studied. This provided support for the meaningfulness of the instrumental-terminal facet for the organization of people’s values. There were, however, puzzling elements in the relevant past findings. Distinct regions emerged for the values usually phrased as terminal and instrumental, even where all the values were phrased in terminal form (Finland) and where the language (Chinese) precluded discrimination between terminal and instrumental values (Hong Kong). This suggested that something other than the terminalinstrumental distinction was producing the empirical discrimination observed. As an alternative interpretation, Schwartz and Bilsky (1990) proposed that the empirical discrimination might be an artifact of the serial order in which the values were rated. They did not explain how an order effect might produce distinct regions. One process, however, seems clear: As respondents go through the questionnaire, they may shift their subjective scale of importance as they encounter values of greater or lesser importance than those encountered previously. Consequently, shifts in scale use would reduce correlations among values located at a distance from each other because they would be likely to be rated on different subjective scales. If the empirical distinction between terminal and instrumental values is due to shifting scale use, it should be reduced or disappear when respondents first anchor their ratings for the whole scale before rating the values. In the new survey form, as in previous research, a list of terminal values preceded a list of instrumental values. Here, however, an anchoring technique was introduced prior to rating each list in order to minimize shifts in scale use. If distinct terminal and instrumental regions emerge in the multidimensional value space, despite the probable reduction in scale use shifts, the conceptual significance of this facet will be reinforced. If, however, distinct regions do not emerge, further doubt will be shed on the significance of this facet for the organization of people’s values.

111. Empirical Studies

A. THE THEORY-BASED VALUE SURVEY

In order to test the hypotheses derived from the revised theory, it was necessary to generate a new, theory-based survey to measure people’s value priorities. A brief overview of the rationale for and construction of the new value survey is provided here.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

17

Based on the motivational goal that defines each of the eleven potential universal value types, we selected specific values to represent each type from among all values that express that goal. Values from the Rokeach (1973) survey and from instruments developed in other cultures (e.g., Braithwaite & Law, 1985; Chinese Culture Connection, 1987; Hofstede, 1980; Levy & Guttman, 1974; MUNO, 1985) were considered. Additional values were obtained from examination of texts on comparative religion and from consultations with Muslim and Druze scholars. To represent the 11 types, 56 values were selected. We adopted Rokeach’s (1973) procedure for further specifying the meaning of each value by providing an additional explanatory phrase in parentheses. The 56 values are listed in Section VII, ordered as they appeared in the questionnaire. Of the values, 21 are identical to those in the Rokeach lists. Both instrumental and terminal values were selected for each value type. In selecting values, preference was given to those judged to have a clear motivational goal. However, values judged to be related to multiple goals were also included if they were presumed to be very important in many cultures (e.g., intelligent, self-respect). A few values were included even though they were not expected to have universally shared meanings (e.g., detachment), in order to assess our ability to detect in the empirical data when values are culture specific or are not well understood. As a direct expression of the definition of values as guiding principles in the individual’s life, the new survey asks respondents to rate each value “AS A GUIDING PRINCIPLE IN MY LIFE,” using the following nine-point scale: of supreme importance (7), very important (61, (unlabeled; 5,4), importanr (3), (unlabeled; 2 , l), not important (O), opposed to my values (- 1). Rating is employed rather than ranking to overcome some of the disadvantages of the latter for cross-cultural work (cf. critiques by Ng et al., 1982, and Rankin & Grube, 1980). In particular, the large number of values that respondents would be required to rank in order to cover the universal value types and to permit local researchers to add values from their own country would swamp reasonable information-processing capacities. Also, rating enables us to measure “negative” values-values people seek to avoid expressing or promoting through their choices and behavior. This phenomenon is especially likely in cross-cultural studies, wherein people may reject values from other cultures.

B. METHODS

I . Samples In order to test the universality of the hypothesized value types and the dynamic relations among them, a set of samples maximally diverse in culture, language, and geographical region was desirable. The samples, briefly

18

SHALOM H. SCHWARTZ

characterized in Table 11, were drawn from 20 countries. They are from cultures on every inhabited continent, representing 13 different languages, and include adherents of eight major religions as well as atheists. Most samples are from two occupational groups (school teachers and university students), but four samples of adults with widely varied occupations are also i n c l ~ d e d . ~ For 13 countries, at least two samples with very similar cultural backgrounds (education, residential area, religious identity, etc.) were included. We anticipated that differences in age, gender, and occupation would lead to differences between the samples from the same culture in the importance of particular values and value types. We assumed, however, that, for the most part, the meanings of values should be quite similar for samples from the same culture. For these countries, comparing within-culture to between-culture variations in the empirical content of value types and their structure of relations can help separate culture-specific from universal aspects of value meanings and structure.* In each country, researchers were asked to collect data from a sample of approximately 200 teachers in grades 4 to 10 in the type of school system that teaches the largest proportion of children. No single occupational group can represent a culture, but grade school teachers may be the best single group: They play an explicit role in value socialization, they are presumably key carriers of culture, and they are probably close to the broad value consensus in societies rather than at the leading edge of change. Teachers are also more numerous, literate, accessible, and receptive to research than most other groups in virtually all societies, qualities that render them especially suitable for cross-cultural research. University undergraduate students were chosen most often as a second sample due to their accessibility and to the availability of student data from other studies of values with which comparisons might be made. In most cases, the samples included teachers of the whole range of grade school subjects or students with widely varying majors. Respondents were dropped before the analyses if they used response 7 (of supreme importance) more than 21 times, or used any other response more than 7We focused on two occupational groups due to the other goals of the wider project, of which the research reported here is a first step-to investigate cultural differences in value priorities as well as antecedents and consequences of individual differences in value priorities both within and across cultures. This is best done with samples from the same occupational groups, in order to minimize confounding of socioeconomic influences with cultural influences. 8For example, if self-respect emerges with achievement values in all samples from one country, but with self-direction values in samples from most other countries, one could conclude that selfrespect has a unique, culture-specific meaning in the particular country. If the location of a value is unique in one sample from a country but not in the other(s), one might attribute this uniqueness either to sample characteristics or to chance variation. Similarly. if two value types (e.g., benevolence and achievement) are in empirical proximity in all samples from one culture but opposed in samples from other cultures, one could conclude that they are uniquely related (compatible rather than in conflict) in that culture.

19

UNIVERSALS IN VALUE CONTENT AND STRUCTURE TABLE II CHARACTERISTICS OF SAMPLES

Country and group Australia Students (Adelaide) General (Adelaide) Brazil Teachers (Brasilia) Students (Brasilia) People’s Republic of China Teachers (Guangzhou) Teachers (Hebei) Teachers (Shanghai) Students (Shanghai) Factory workers (Shanghai) Estonia Teachers (small town) General (rural) Finland Teachers (Helsinki) Students (Helsinki) Germany Students (Trier) Greece Teachers (Athens) Students (Athens) Holland Teachers (Amsterdam) Students (Amsterdam) General (Countrywide) Hong Kong Teachers (Hong Kong) Students (Hong Kong) Israel Students (Jerusalem) General (Jerusalem) Italy Teachers (Rome) Japan Teachers (Hyogo) Students (Osaka) General (Osaka) New Zealand Teachers (South Island) Students (Christchurch)

N

Language

Religionsa > 20%

Percentage female

M (SD)

387 199

English English

PR, RC, N PR, RC

61 49

22 (7.2) 33 (11.9)

154 244

Portuguese Portuguese

RC RC

59 49

23 (3.2) 32 (8.3)

I94 199 21 1 205 208

Chinese Chinese Chinese Chinese Chinese

BU, BU, BU, BU, BU,

N N N N N

37 38 39 61 11

26 (4.2) 28 (9.4) 34 (10.8) 20 (1.3) 26 (8.9)

23 I 210

Estonian Estonian

PR, EO, N PR, EO, N

82 54

40 (12.5) 37 (12.6)

205 295

Finnish Finnish

PR PR

76 80

44 (9.0) 24 (4.9)

377

German

PR, RC

58

22 (3.4)

I95 234

Greek Greek

GO, N GO

55 72

34 (5.6) 20 (1.5)

187 277 240

Dutch Dutch Dutch

PR, RC, N PR, RC, N PR, RC, N

22 54 0

43 (7.4) 23 (3.5) 49 (7.6)

20 I 21 I

Chinese Chinese

CF, PR, N CF, PR, N

55 63

28 (4.2) 22 (1.8)

I99 207

Hebrew Hebrew

JE, N JE, N

50 52

24 (3.1) 32 (12.2)

200

Italian

RC

70

41 (9.3)

229 542 207

Japanese Japanese Japanese

SH, BU SH, BU SH, BU

17 39 54

34 (6.1) 22 (1.4) 45 (11.2)

I99 202

English English

PR, N PR, N

60 62

39 (9.0) 23 (6.5)

CF, CF, CF, CF, CF,

Age

(continued)

20

SHALOM H . SCHWARTZ

TABLE I1 (Continued) Religions" Country and group Poland Teachers (Warsaw) Students (Warsaw) Portugal Teachers (Porto) Students (Porto) Spain Teachers (Madrid) Graduate students (Madrid) Taiwan Teachers (Taipei) United States Students (Illinois) Venezuela Students (Caracas) Zimbabwe Teachers (Harare) Students (Harare)

N

Language

> 20%

Percentage female

Age M (SD)

195 195

Polish Polish

RC RC

75 76

39 (9.7) 24 (4.4)

192 198

Portuguese Portuguese

RC RC

63 73

36 (7.5) 24 (5.9)

186 308

Spanish Spanish

RC RC, N

58 77

31 (8.3) 26 (3.7)

202

Chinese

BU, Folk

69

48 (5.9)

240

English

PR, RC

50

20 (1.3)

185

Spanish

RC

74

22 (4.5)

185 205

English English

Tribal Tribal

28 48

28 (5.8) 19 (.9)

"BU, Buddhism; CF, Confucianism; EO, Eastern Orthodoxy; GO. Greek Orthodoxy; JE, Judaism; N, None; PR, Protestantism; RC, Roman Catholicism; SH, Shintoism.

35 times. Those who concentrated their responses to that degree were assumed to have failed to make a serious effort to differentiate among their values. Those who responded to fewer than 41 values were also excluded. The proportions of respondents dropped from the samples on these bases ranged from 0 to lo%, with a mean of 2%. 2. Procedures

The value survey was prepared in decentered Hebrew and English versions. The English version, together with versions in relevant cognate languages, if available, was provided to researchers in each country. In most non-Englishspeaking countries, the local researchers prepared a native language version, using backtranslation procedures, and sent that version to the author. He then obtained backtranslations into English, and sometimes into Hebrew as well, from native speakers of the relevant language in Israel. Discrepancies from the original versions were noted, and recommended revisions were sent with explanations to the local researchers. The latter made the final decisions regarding modifications. In two instances (Greece and Estonia), local researchers prepared their final versions without consulting the author. Two principles guided the ordering of values in the survey. First, values

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

21

assumed a priori to represent the same value type were separated from each other by at least two other values. Second, values were separated by at least three other values from those in the same quintile on importance, based on pretest measures of importance from Finland and Israel. The order of the values was the same in all language versions. In Brazil, China, Finland, New Zealand, Poland, Spain, and Venezuela, local researchers added up to 6 values to the core 56. When values were added, they were dispersed at equal intervals through the survey, with one restriction: Values that appeared to fit a given value type were separated by at least two other values from other values of that type. If only one instrumental or one terminal value was added, it was placed at the end of the respective list. Values were presented in two lists. The first list contained terminal values (1 to 30) and the second list contained instrumental values (3 1 to 56; see Section VII). Prior to rating the values on each list, respondents read the whole list, chose and rated the value most important to them, and then chose and rated the value they most opposed or the one least important to them. This anchored the response scale for them. In about half the samples, the survey questionnaire was administered in groups of 5-100 respondents. In most others, it was administered to individuals, and the Japanese and Dutch adult samples were obtained in a mail survey. In every case, responses were anonymous. Demographic, attitude, and other supplementary questions were placed at the end of the questionnaire. Raw data were forwarded to Jerusalem, where all analyses reported in this article were performed.

3. Smallest Space Analysis In each sample, the intercorrelation matrix of Pearson correlations between the importance ratings of the values was analyzed with the Guttman-Lingoes Smallest Space Analysis (SSA) (Guttman, 1968; cf. Canter, 1985). This is one of a variety of nonmetric multidimensional scaling (MDS) techniques for structural analysis of similarity data (Davison, 1983; Dillon & Goldstein, 1984). This technique represented the values as points in multidimensional space such that the distances between the points reflected the empirical relations among values as measured by the correlations between their importance ratings. The greater the conceptual similarity between two values, the more related they should be empirically, and hence the closer their locations should be in the multidimensional space. We interpret the SSA by using a conjigurational ver$cation approach (Davison, 1983).9That is, we interpret the configurationsof substantively related 9F0r comparisons of SSA with other multidimensional scaling techniques and with factor and cluster analysis, see Canter (1985). Davison (1983). Dillon and Goldstein (1984). Guttman (1982), Shye (1988), and Zeidner and Feitelson (1989). For a discussion of its relative usefulness for testing hypotheses of the type proposed here, see Schwartz and Bilsky (1990, p. 890).

22

SHALOM H. SCHWARTZ

points that emerge to form regions and the arrangement of these regions in space relative to one another. The MDS axes are not assumed to have substantive meaning because they are arbitrary. The content universe is conceived as a geometrical space in which the specific values are but a sample of all conceivable values comprising the total space with points everywhere. This means that some values at the edge of one region may correlate less with other values of the same region than they do with certain values on the edge of neighboring regions. Partition lines may be straight or curved, as long as they yield regions having continuous boundaries that do not intersect with the boundaries of other regions (Lingoes, 1977, 1981). We have postulated that the motivational content of values is the most powerful principle in the organization of people’s value preferences. If this is so, the organizing effects of motivational content may be evident even in the projection of value points obtained with a two-dimensional solution. The value content hypotheses were tested by examining whether it was possible to partition the points that filled the two-dimensional space into distinct regions that reflected the 10- 11 a priori value types. Because we had specified, in advance, which set of values was expected to constitute the contents of the region for each value type, we were able to seek a region for each set and then to draw boundaries between the regions. The value structure hypotheses were tested by examining whether the regions obtained formed a pattern similar to the theoretical structure of Fig. 1 that represents these hypotheses. Wedgelike regions emerging from a common origin are predicted when two or more of the several elements in a qualitative facet (i.e., the value types in the motivational concern facet here) are in conceptual opposition to one another (Levy, 1985; Shye, 1985). To determine where to place the partition lines between regions, we first drew boundary lines that connected the values that were at the outer edges of each region, avoiding any overlap of region boundaries (Lingoes, 1977, 1981). We then placed the partition lines between these boundaries. Three criteria were used to decide whether a set of value points formed a bounded region confirming the existence of a given value type: The region must include (1) at least 60% of the values postulated a priori to constitute that type and (2) no more than 33% of the values postulated to constitute any other single type. Further, (3) at least 70% of all values in the region had to have been judged a priori as potentially reflecting the goals of the appropriate value type as one of their meanings. If the above criteria were not met, a region combining two value types was formed using the following criteria: The region (1) contained at least 50% of the values postulated to constitute each type, and (2) at least 70% of the values in the region potentially reflected the goals of these two value types. If neither set of criteria were met, the existence of the value type was taken as disconfirmed. We also examined the configuration of value points to assess whether people

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

23

respond differently to terminal versus instrumental values. If the terminalinstrumental distinction affects people’s responses, the value points should be partitionable into two distinct regions, one containing the 30 terminal values and one containing the 26 instrumental values. Evidence for distinct terminal and instrumental value regions was sought not only in the two-dimensional solution of the SSA but also in the two-dimensional projections of higher dimensional solutions (up to four dimensions). The latter projections were also examined for evidence of the motivational concern facet. C. RESULTS

In Fig. 2, we present an example of SSA results for 56 values. The bases and justification for computing this SSA, which averages the results of analyses within each of 36 samples, will be discussed below. For now, we use it as a convenient prototype to aid in elucidating and assessing the value content hypotheses and the structure hypotheses. 1. Distinctiveness of Value Types

Are all 11 types of values present across cultures, forming distinct regions in the SSA? Applying the criteria listed above for identifying a region as confirming the existence of a value type, we sought regions indicative of each value type in the two-dimensional SSA of each sample. l o Results of the 40 within-sample SSAs are summarized in Table Ill. For the vast majority of samples, it was possible to locate regions arrayed much like the prototype. Column one lists the number of distinct regions identified in each sample. The numbers of samples out of 40 in which a distinct region was identified for each value type were as follows: Universalism (34), Benevolence (28), Tradition (36), Conformity (29), Security (32), Power (37), Achievement (40), Hedonism (32), Stimulation (3 I), and Self-Direction (36). Spirituality is not included in the above list because the criteria for confirming ‘There are no decisive criteria for selecting the number of dimensions to be analyzed in an SSA (Coxon, 1982). Suggested conventions include adding dimensions until a stress measure (coefficient of alienation) of <.I5 is attained or until the goodness of fit (measured by the stress) is not substantially improved by adding another dimension (“elbow test”). More important than such conventions is the interpretability of the solution (Borg & Lingoes, 1987). In the c a n t analyses, two-dimensionalsolutions were easily interpretable by applying the theory of motivational types of values. We therefore used the two-dimensional solution even though the coefficients of alienation were >.15 in all samples (range .21 to .32). The emergence of a clear motivational structure of values on the first two dimensions suggests that motivational goal is the most powerful principle organizing value preferences. For the general Japanese sample, the two-dimensional projection did not yield clear evidence of the motivational types. We therefore examined the three-dimensional analysis. The data reported for this sample are based on the projection of dimensions I X 3 in the latter analysis.

SECURITY

ACHIEVEMENT

BENEVOLENCE STIMULATION

frnrdom

UNIVERSALISM

aq@

b r a d

SELF-DIRECTION

I

Fig. 2. Individual-level value structure averaged across 20 countries (36 samples): Two-dimensional smallest space analysis.

25

UNIVERSALS IN VALUE CONTENT AND STRUCTURE TABLE I11 VALUE TYPESA N D STRUCTURE OBSERVED IN 40 SAMPLES

Country and groups Australia Students General Brazil Teachers Students People's Republic of China Teachers (Guangzhou) Teachers (Hebei) Teachers (Shanghai) Students (Shanghai) Factory workers (Shanghai) Estonia Teachers General Finland Teachers Students Germany Students Greece Teachers Students Holland Teachers Students General Hong Kong Teachers Students Israel Students General Italy Teachers Japan Teachers Students General

Distinct types

Distinct + joint types"

Number of correct locations in 52 values

Number of moves to fit ideal structure

Number of moves to fit revised ideal structure

8 10

8 10

50 46

1.5 1

1.5 0.5

8 8

10 10

42 47

3

2.5 0

8 8 6

10 8 8

25

7

I

6

6

41 39 38 38 39

8 6

8

42

8

10 10

I

4.5

1.5 6 5 4 3.5

44

6.5 3.5

6 2.5

10 10

48 46

2.5 0

0

10

10

50

0.5

0.5

8 10

10 10

46 51

I .5 I .5

0.5 0.5

10 10 8

10

10 10

48 51 46

0.5 0.5 1

0.5 0 0.5

8 8

8 8

46 48

2.5 3

2 2

10 10

10 10

47 50

0.5 0.5

0.5 0

8

10

46

0.5

0

8 8 10

LO

45 48 44

I

1 0.5 0.5

8 10

I 6 5.5

1.5

I .5

I .5

(continued)

26

SHALOM H. SCHWARTZ

TABLE I11 (Continued)

Country and groups New Zealand Teachers Students Poland Teachers Students Portugal Teachers Students Spain Teachers Students Taiwan Teachers United States Students Venezuela Students Zimbabwe Teachers Students ~~~

~~~~~~

Distinct typs

Distinct + joint types0

Number of correct locations in 52 values

Number of moves to fit ideal structure

10 8

10 10

48 51

I .5 2

I 2

8 8

10 10

46 49

2 3

2 2

6 10

10 10

50 48

2.5 0.5

2 0.5

10 10

10 10

46 49

0.5 0.5

0.5 0

8

8

45

7.5

7

8

10

48

1

0.5

8

10

46

I .5

I .5

4 8

6 8

43 43

6 4.5

5 4

Number of moves to fit revised ideal structure

~~~

"Distinct types plus instances of joint types hypothesized to be adjacent.

a distinct region were met for this type in only 8 samples. In an attempt to investigate whether different spirituality types emerged in different cultures, we relaxed our criteria: We sought the joint emergence of at least three of the six a priori spirituality values in a distinct bounded region. In meeting the 70% criterion for the whole region, we treated the a priori tradition values as potentially fitting the spirituality definition. Using these criteria, a distinct spirituality region was found in 17 additional samples. The most frequently found combination of three spirituality values (10 samples) included a spiritual life, accepting my portion in life, and detachment. The second most common combination was a spiritual life, inner harmony, and detachment (8 samples). No other combination was found in more than 6 samples. Generally, when no distinct region emerged for a value type, its values were intermixed with those of a type postulated to be adjacent. Column two of Table 111 lists the number of types identifiable in each sample when we also included those instances wherein two types postulated to be adjacent in the ideal structure (Fig. 1) formed a joint region. In 67.5% of the samples, all 10 value types were

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

27

found, and in 92.5% at least 8 or more types were identifiable. The numbers of samples out of 40 in which a value type emerged in a distinct region or in a joint region with one of its two postulated neighbors were as follows: Universalism (38), Benevolence (33), Tradition (37), Conformity (33), Security (36), Power (37), Achievement (40), Hedonism (371, Stimulation (40), and Self-Direction

(40). In many samples, the distinct regions of tradition and of power values that emerged did not extend into the common origin. Rather, these tradition and power regions were located toward the periphery of the two-dimensional space. In 29 of the 36 samples that revealed a distinct region of tradition values, this region was outside the conformity values region, as in the prototype, rather than between the benevolence and conformity regions, as in Fig. 1. This suggests a need for a possible modification in the theory. In 28 of the 37 samples in which a distinct power values region appeared, this region could be connected with the common origin as postulated. However, in 26 samples the boundaries of the power values region could also be drawn so as to place it outside the achievement values region. The more straightforward location of the power region was as connected to the origin in 21 samples and outside achievement in 16 samples. The prototype reflects this juxtaposition of the power and achievement values regions: A boundary drawn horizontally above ambitious in Fig. 2 would form a single broad wedge including both achievement and power values, with the latter located outside. Benevolence and conformity showed a relatively high number (seven) of nondistinct regions. In every case, this was due to the joining of benevolence with conformity into a joint region. These two value types were originally postulated to be separated by tradition. However, as just noted, the tradition region was frequently located outside of conformity, making benevolence and conformity values adjacent.

2. Contents of Value Types For each sample, the number of single values that emerged in regions corresponding to their postulated value type(s) is listed in column three of Table 111. Excluded from this analysis were the spirituality value type and the four values originally included primarily to represent it (spiritual life, meaning in life, inner harmony, detachment). Values were counted as misplaced if they emerged empirically in a region that represented a value type@)other than those with which they were associated a priori on the basis of the motivational goal(s) they express. The number of correct locations for the 52 values ranged from 51 (Greece, students; Holland, students; New Zealand, students) to 38 (Shanghai, teachers and students). The median number of correct locations per sample was 46 (88%).

28

SHALOM H. SCHWARTZ

For four of the Chinese mainland samples (Hebei and all three Shanghai), the individual interest value types emerged as predicted, but the values from the collective and mixed interest value types could not be separated into regions for the postulated types without violating our criteria. Interestingly, these values did form a consistent set of three meaningful regions, but these regions differed form our theory. Rather than forcibly imposing our structure on these samples, it seems more reasonable to treat them as fitting an alternative theory, which deserves speciai study (Zhi-gang, 1990, see below). In order to reveal the values that constitute the contents of each value type, we counted the number of times that each value emerged empirically in each region across samples. This indicates both the most common meaning of each value and any common alternative meanings the value may have in different cultures. Table IV presents the relevant results. In this assessment, we excluded the Shanghai and Hebei samples for the reasons noted above. Each value is listed under the value type in which it emerged most frequently. All other empirical locations for each value are given in parentheses. For 87% (45 of 52) of the values, a single region associated with one motivational type of value accounted for their empirical location in at least 70% of the TABLE IV NUMBER OF TIMESVALUES WEREEMPIRICALLY LOCATEDIN A REGIONOF EACHMOTIVATIONAL TYPE:BASEDON SSA TWO-DIMENSIONAL PROJECTIONS FOR 36 SAMPLESa ~-

Number of locationsb

Value Power (W) Social power Authority Wealth Preserving my public image Social recognition Achievement (AC) Successful Capable Ambitious Influential Intelligent Hedonism (HE) Pleasure Enjoying life Stimulation (ST) Dating A varied life An exciting life Self-direction (SD) curious

36 35 (AC-I) 35 (AC-I) 27 (SE-8)(TR-I) 23 (AC-8)(SE-5) 34 (UN-l)(HE-I) 33 (PO-I)(UN-I)(SE-I ) 32 (PO-l)(SD-I)(ST-l)(UN-I) 28 (PO-6)(SD-I)(ST-I ) 23 (SD-7)(UN-4)(CO-2) 36 36 36 34 (SD-l)(UN-I) 33 (AC-2)(HE-I)

35 (Po-1) (continued)

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

29

TABLE IV (Continued) Value Creativity Freedom Choosing own goals Independent Self-respect Universalism (UN) Protecting the environment Unity with nature A world of beauty Broad-minded Social justice Wisdom Equality A world at peace Inner harmony Benevolence (BE) Helpful Honest Forgiving Loyal Responsible A spiritual life True friendship Mature love Meaning in life Tradition (TR) Accepting my portion in life Devout Humble Respect for tradition Moderate Detachment Conformity (CO) Obedient Honoring of parents and elders Politeness Self-discipline Security (SE) Clean National security Recipmation of favors Social order Family security Sense-of belonging Healthy

Number of locationsb 34 32 29 29 15

(UN-I)(BE-I) (UN-3)(AC-I) (UN-3)(AC-3)(PO-l)

(ST-3)(HE-2)(UN-I)(Po-l) (AC-13)(UN-4)(BE-3)(SE-I)

35 (SE-I) 35 (SE-I) 34 (SD-I)(TR-I) 33 (BE-l)(SD-I)(ST-I) 32 (BE-2)(SE- I )(CO- 1 ) 32 (BE-3)(SD-I) 29 (SD-4)(SE-2)(BE-I) 29 (BE-3)(SE-3)(CO-I) 20 (BE-8)(SE-4)(TR-2)(SD- I )(ST-I ) 36 35 31 29 27 24 23 22

(CO-1)

(UN-4)(CO-I) (CO-6)(UN-I) (CO-7)(UN-2) (UN-S)(TR-4)(SE-2)(SD- I )

(UN-7)(SE-3)(TR-I)(SD-I)(ACI) (UN-9)(SE-3)(CO-I)(TR-I)

15 (UN-9)(SE-S)(TR-4)(CO-3)

34 (CO-I)(SE-I) 31 (CO-2)(UN-2)(BE-I) 30 (BE-3)(CO-2)(UN-I) 29 (CO-S)(SE-I)(UN- I ) 27 (SE-6)(BE-2)(CO-I) 18 (UN-7)(BE-4)(SE-3)(SD-2)(PO-2) 34 (SE-2) 34 (TR-2) 32 (SE-3)(BE-I) 32 (SE-2)(BE- I)(UN- I ) 33 (CO-3) 32 (UN-2)(TR- I)(BE- I) 29 (CO-4)(SD-2)(UN-1) 29 (CO-5)(UN-I)(SD- 1 ) 28 (CO-4)(BE-2)(AC-2) 20 (BE-6)(AC-4)(TR-2)(UN-2)(SD1)(CO- I ) 20 (HE-6)(AC-s)(UN-2)(BE-2)(SD-I)

“In the Japan adult sample, the I X 3 projection of the three-dimensionalsolution was used bLocationsother than those postulated are indicated in parentheses.

30

SHALOM H. SCHWARTZ

samples. Counting the four values originally included primarily to measure spirituality, this level of consistency was reached by 80% of the total 56 values. The presumed spirituality values emerged most often in the regions representing the benevolence, universalism, and tradition types. For every one of the 56 values, its predominant empirical location fit quite well with the goal of the value type with which it was most often associated. Our confidence in the meaningfulness of the findings was strengthened by the fact that the most frequent alternative locations of values also made sense (e.g., intelligent in self-direction, self-respect in achievement, moderate in security). Variations in the value type with which a value was associated may reflect random error, differences due to translation, or real cultural differences in meaning. Detailed examination of these alternatives for each value is feasible and revealing, but it is beyond the scope of this article." 3 . Structure of Dynamic Relations among Values

The total set of hypotheses regarding the dynamic relations of conflict and compatibility among the value types was captured in the circular arrangement of the types presented in Fig. 1. Therefore, for each sample, comparing the arrangement of value types observed in the SSA with this ideal arrangement can provide an overall assessment of the fit between data and theory. As a measure of the fit between the observed structure of value relations and the hypothesized structure, we counted the number of single inversions of the order of adjacent value types (= moves) required to rearrange the observed order to match the ideal order. In cases in which value types formed a joint region, types were split into separate regions to the sides of the joint region where the values that constituted them were predominantly concentrated. Splitting joint regions this way was counted as half a move. If the values constituting the types were not concentrated on one or the other side of the joint region, a full move was counted to split any pair of types into separate regions in the hypothesized order. For example, compare the observed structure of Fig. 2 with the hypothesized structure of Fig. 1. The structures are identical with the exception of the wedge containing the conformity and tradition types of values. Although these two I 'Forgiving, for example, emerged in the universalism region, near broad-minded, in all three Japanese samples reported on here and in two recently analyzed Japanese student samples form Hokaido and Tokyo. It emerged in the benevolence region in practically all the samples from other cultures (see Table IV). Furthermore, no translation problem was evident in backtranslations of this value by four independent translators. Hence, forgiving almost certainly has a unique meaning in Japanese culture. Its consistent location suggests that, in Japan, forgiving primarily has the meaning of tolerance, acceptance, and understanding of actions that are disturbing. This contrasts with its more common meaning of actively forgoing feelings of anger, resentment. and/or a desire to punish, which are initially experienced in response to the disturbing act. Sumiko Iwao (personal communication, April 1991) has confirmed this interpretation.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

31

types are not intermixed in Fig. 2, they are, from the viewpoint of the hypothesized circular arrangement, in a joint region requiring separation. Neither the conformity nor the tradition values are concentrated on one or the other side of the joint region. Hence, it would count one move to separate the two types and place them in the hypothesized order of conformity between security and tradition. No further moves would be required to match the ideal structure. The number of moves required in each sample to match the observed structure to the hypothesized structure of value types is reported in column four of Table 111. The empirical fit was quite impressive. The modal number of moves needed was 0.5 and the median was 1.5. Only 8 of the 40 samples required more than four moves to match the hypothesized structure and only 3 samples required more than six moves.'* The more poorly matched samples included four from China and one from Taiwan. Only one sample matched the ideal perfectly (students from Finland), but in 14 other samples no more than a single move was required to attain a match, indicating that only one pair of adjacent value types was mixed or reversed. The substantial match in most samples between the observed and the hypothesized contents and structure of the value types enabled us to undertake a further examination of universal aspects of value content and structure. This match suggested that a single analysis that combines the samples may provide a picture of the relationships among values across cultures that can serve as a reasonable overall heuristic. Excluded from this analysis, for the reasons noted above, were the one Hebei and three Shanghai samples. The analysis included 36 samples, each with its unique structure of value relations, from 20 countries (Guangzhou remained as a Chinese sample). In combining samples, equal weight was given to each of the countries. For countries providing more than one sample, the correlations among the 56 values within each sample were averaged across samples to generate a single intercorrelation matrix for the country. The resulting 20 country matrixes were, in turn, averaged to generate one overall intercorrelation matrix. Figure 2 presents the two-dimensional projection of the SSA on the overall matrix. Distinct regions emerged for each of the 10 value types. Moreover, every one of the 56 single values emerged in a region appropriate to its primary motivational goal. The overall value structure differed from the hypothesized structure only in the positioning of the distinct tradition and conformity regions in a way that formed a joint wedgelike region emerging from the common origin. These findings strongly supported all the hypotheses regarding compatibilities and conflicts between value types. '*For the 10 value types, 181,440 different arrangements were possible. Any arrangement requiring fewer than seven moves differed significantly ( p < .05) from a random arrangement in the direction of the postulated Structure (fewer than five moves, p < .01). I thank Gennady Levin for calculating these probabilities.

32

SHAIDM H. SCHWARTZ

Next, we performed separate SSA analyses on the set of teacher samples and on the set of student samples in order to examine the robustness of this representation of a universal content and structure of values. The separate SSAs on these sets of samples tested whether the content and structure of value types were similar despite differences in the importance attributed to values by teachers and students found in previous research. In a comparison of teachers with students in each of 14 countries, for example, we had found that teachers attributed more importance than students to conformity, security, tradition, and benevolence values in almost all countries, whereas students attributed more importance than teachers to hedonism, stimulation, and self-direction values (Schwartz, 1990b). As the basis for the teachers' SSA, an intercorrelation matrix was formed by averaging the intercorrelation matrixes for the 15 teacher samples from 15 countries. A parallel matrix was formed in the same way for the 16 student samples from 16 countries. Figure 3 presents the two-dimensional projection of the SSA on the value ratings of the averaged student samples. This projection was easily partitioned into regions representing all 10 value types. Moreover, the structure of relations among the types was virtually identical to the overall structure. All single values emerged in the same appropriate regions in this analysis as in the overall analysis, with the exception of healthy. Here, too, the tradition and conformity regions formed a single wedgelike region emerging from the common origin. Figure 4 presents the two-dimensional projection of the SSA on the value ratings of the averaged teacher samples. This projection was also easily partitioned into regions representing all 10 value types. All single values emerged in the same appropriate regions in this analysis as in the overall analysis, with the exception of healthyI3 and self-respect. l4 The structure of relations among the '3The location of healthy was apparently a compromise, reflecting its relatively high positive correlations with security, hedonism, and achievement values. This suggests that healthy has multiple meanings. For students, correlations with security values were highest, closely followed by correlations with values of both other types. To them, healthy apparently expressed not only a goal of personal safety (security) but also of enjoyment of one's body (hedonism) and of health maintenance (e.g., through exercise-an achievement task). For teachers, correlations were considerably weaker with hedonism and achievement values than with security values. To them, perhaps because they are older, the security connotation may have predominated. Healthy was virtually uncorrelated with power values. A similar variation in the meaning of health was found in separate MDS analyses run by Kristiansen (1990) on the values of British men and women. Health was related to security among men and to hedonism among women. '4The location of self-respect in the SSAs is revealing. It was near the center in many samples, suggesting that self-respect is associated with and based on the attainment of different types of values. The SSAs for the teacher samples suggested that for them self-respect was more related to their socially recognized achievements. Self-respect emerged in the achievement region in 8 of 15 teacher samples and in the self-direction region only twice (cf. Fig. 4). For students, in contrast, the SSAs suggested that self-respect was more related to the assertion of uniqueness and independence. To them, self-respect emerged in the self-direction region in 11 of 16 samples and in the achievement region only three times (cf. Fig. 3).

acceDtinaDortion

.

\

/ S ECURlTY

POWER S

devout

/

d DOWW

: . P a

ACHIEVEMENT

CONFORMITY

UNIVERSALISM Fig. 3.

btoadltded

freeclom

SELF-DIRECTION

\

Student samples. Individual-level value structure averaged across 16 countries: Two-dimensional smallest space analysis.

BENEVOLENCE

UNIVERSALISM

TRADITION

SELF-DIRECTION

SECURITY

STIMULATION

POWER ACHIEVEMENT Fig. 4. Teacher samples. Individual-level value structure averaged across 15 countries: Two-dimensional smallest space analysis.

35

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

value types was identical to the overall structure in Fig. 2. The apparent rotation of regions for the types is a technical matter that carries no conceptual significance. In this structure, too, the tradition and conformity regions formed a single wedgelike region emerging from the common origin. The recurrent location of the tradition region outside the conformity region in the combined analyses, as well as in 29 of the single samples, strongly suggests that reality is better represented by this structural arrangement. This arrangement also confirms all the hypotheses regarding compatibilities and conflicts among value types proposed in the introduction. We therefore took this order as our corrected ideal and counted the number of moves required to rearrange the observed order of value types in each sample to match this corrected ideal. The structure in 6 of the samples was identical to the revised ideal order, and in 12 other samples only 0.5 of a move was required to match it. In 29 of the samples, the number of moves required to match this order was less than the number required to match the original ideal order, and in the other samples there was no difference. (See Table 111, column five.) a. Comparibilities wirhin Cultures. All the hypothesized compatibilities appeared in the SSAs for the combined samples, but this does not indicate how frequently each compatibility emerged across samples. For this purpose, we examined whether each pair of value types was found in adjacent regions in the SSAs for each sample. We also inferred that a pair of types was compatible if the values from the pair were intermixed. The number of samples in which each compatibility hypothesis was confirmed is shown in Table V. None of the comTABLE V OF SAMPLES OUTOF 40 IN WHICHHYPOTHESIZED NUMBER A N D CONFLICTS AMONG VALUE COMPATIBILITIES TYPESWEREC O N ~ ~ R M E D ~~~

Compatibilities Power-achievement Achievement- hedonism Hedonism-stimulation Stimulation-self-dirction Self-direction-universalism Universalism-benevolence Tradition-conformity Conformity-security Security-power

35 33 30 31 38 38 31 36 28

Conflicts Self-directionlstimulation vs. conformity/tradition/s~uri~ Universalism/benewlence vs. power/achievement Hedonism vs. conformity/tradtion Spiritual vs. hedonism/power/achievement

35 31 33 32

36

SHALOM H. SCHWARTZ

patibilities emerged in every one of the 40 samples. However, six compatibilities were found in at least 35 samples (88%). Given the likelihood of some random error, this suggests near universality. The other three compatibilities also appeared with substantial frequency (in at least 70% of samples). Although no consistent spirituality type was found, all four single values selected to represent one or another form of spirituality emerged in a region compatible with spirituality (benevolence, tradition, universalism) in at least 78% of samples. b. ConjZicts within Cultures. Results of examining the SSAs for all 40 samples for the occurrence of the hypothesized conflicts between sets of value types are presented in Table V. A conflict was counted as present when the regions representing the two opposing sets of value types were separated in both directions around the circle by at least two other value types. In the absence of evidence for a cross-culturally consistent spirituality type of values, we represented this type with the single value a spiritual life, because this value was found in all the combinations of spirituality values that had emerged empirically with some frequency. Table V shows that each of the conflict hypotheses was confirmed in at least 80% of the samples. Moreover, in no sample was the region for a value type surrounded by regions for value types with which it was postulated to be in conflict. All conflict hypotheses were confirmed in the SSAs for the combined samples (Figs. 2, 3, and 4). 4 . Terminal versus Instrumental Values

In past research (Schwartz & Bilsky, 1987, 1990), the 18 terminal values of the Rokeach Value Survey were clearly separated from the 18 instrumental values into two distinct regions in the SSA for each of seven countries. Most frequently, the instrumental values formed a central region surrounded by the terminal values. The maximum number of errors of placement observed was four (1 1%). In light of these findings, we set a criterion for rejecting the hypothesis that distinct regions were present here of at least eight errors (14.3%of the 56 values studied). This is a generous criterion intended to minimize the chance of mistakenly accepting our suspicion that the separation was due to an artifact that we had sought to overcome by having respondents anchor their scale use. Only if the regions that represent a conceptual distinction show a consistent shape across samples is it possible to give them a clear interpretation (Levy, 1985). We therefore sought to partition the space into central (instrumental value) and peripheral (terminal value) regions as previously found and, if that was not possible, into distinct regions of any other shape. In each of the 40 samples, we first examined the projection on which the motivational types of values had appeared. This yielded not even one instance of clear separation between terminal and instrumental values. We then examined the three two-dimensional projections produced by the three-dimensional SSA in

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

37

each sample. Of these 120 projections, 2 met the criterion of fewer than eight errors. Finally, we examined the six two-dimensional projections produced by the four-dimensional SSA. Of these 240 projections, 3 met the criterion. The few separations that were found took no consistent shape. We would expect some successful separations of terminal from instrumental values by chance alone, given that 10 projections were searched for each sample and given some flexibility in drawing partitions when up to seven errors are permitted. The five separations observed in 400 projections therefore provide little support for the idea that the terminal-instrumental distinction is a meaningful basis on which people organize their values.

IV. Theory Assessment A. THE NATURE OF THE VALUE TYPES How many different motivational types of values is it possible and worthwhile to distinguish? The data confirmed that people in a large number of cultures implicitly distinguish 10 types of values when assessing the importance of specific values as guiding principles in their lives. The stimulation, power, and tradition types, added to the theory and first studied in this research, emerged in distinct regions in more than 75% of samples. The value types studied earlier, some redefined here, were also consistently found in the analyses. Only the spirituality value type failed to evidence universality to a substantial degree. The data give no support to the idea that there are additional, universal, motivational types of values still missing from the theory. Specific additional values that collaborators from different countries suggested as necessary to cover concepts important in their cultures pointed to no new, potentially universal types. Moreover, when these values were included in the SSAs for the samples in which they were added, regions suggesting new value types did not emerge. Rather, the added values typically emerged in regions appropriate to their meanings (e.g., chastity in conformity or tradition; national identity in security or tradition). Hence, pending evidence to the contrary, the 10 value types here may be taken as tentatively exhausting the distinctive, near universal, motivational types of values. Of course, future theorizing might suggest additional types, just as we have added types here. How near to universality are these value types? This question cannot be answered definitively. Logistics prevent anyone from studying all cultures, as required for a decisive conclusion of universality. Moreover, various sources of error (measurement, sampling, etc.) doubtless produce mixing of adjacent types in some analyses. The proportion of countries in which each value type appeared

38

SHALOM H. SCHWARTZ

in a distinct region in at least one sample is probably the best estimate we can derive of the universality of that type from this research. Viewed this way, the power, achievement, and tradition types were universal, as they emerged in all countries. The hedonism, self-direction, universalism, and security types were found in 95% of countries, and the stimulation, benevolence, and conformity types were found in 90% of countries. There were only 10 instances in which a value type failed to appear in a distinct region in at least one sample for a country. Six of these instances occurred in countries in which only one sample was studied. Given the diversity of countries studied, these findings suggest that all 10 motivational types of values may be quite close to universals. Next, we consider the implications of the findings for understanding particular value types. 1. Spirituality

Spirituality, at least as operationalized here, does not appear to be a universal motivational type of value. Firm evidence for a distinct spirituality region was obtained in only 8 samples and weaker evidence was detected in another 17 samples. It may be that answering the question of ultimate meaning in life is a basic human need that finds expression in a set of values. However, the data suggest two alternative ways in which this need may be expressed other than through a universal spirituality type. First, people may find meaning through the pursuit of other types of values. A spiritual life, meaning in life, unity with nature, and inner harmony emerged frequently in the regions of benevolence and universalism values. This supports the idea that self-transcendence through concern for others, broadly defined, is one common way to pursue meaning. Detachment and acceptance of my portion in life appeared frequently in tradition regions, suggesting that tradition provides an alternative source of meaning. The empirical evidence that these two contrasting answers to questions of meaning appear across cultures fits well with analyses by sociologists of religion (e.g., Yinger, 1957). Second, rather than a single, universal spirituality type, there may be a number of distinct types of spirituality, each consisting of a different subset of specific values. Each type might be expected to appear in a set of cultures that is similar in religion, history, development, or some other variable likely to influence the way people pursue meaning. Surprisingly, no particular subset of spirituality values appeared together in more than 10 samples. Moreover, we discerned no obvious similarities that united the sets of samples in which the various subsets of spirituality values appeared. For example, the most frequent combination-a spiritual life, detachment, accepting my portion in life-was found in samples from Brazil, Holland, People’s Republic of China, Poland, Spain, Taiwan, and Zimbabwe. The sets of samples exhibiting the other combinations of spirituality values were no less heterogeneous.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

39

To make progress on this second approach to understanding spirituality values cross-culturally, it is necessary to generate clearer conceptual definitions of possible subtypes of spirituality. These definitions would suggest the sets of values that should be sampled in future research to measure each subtype of spirituality. They might also suggest different subsets of values in the current survey that could be used as indexes of spirituality subtypes in a reexamination of the available SSA projections for the 40 samples. 2 . Benevolence and Universalism The current version of the theory narrowed the definition of benevolence to a focus on the welfare of people with whom one is in close contact. Values that referred primarily to the welfare of entities outside the ingroup (world at peace, equality, social justice) were grouped with the former “maturity” values to form the new universalism value type. The data strongly supported this new division. Benevolence and universalism values were located in different, separable regions in 34 of 40 samples. Moreover, the values referring primarily to the welfare of ingroup entities (helpful, forgiving, honest, loyal, responsible) were rarely intermixed with those referring to wider social entities (cf. Table IV). The benevolence and universalism regions were almost always adjacent, pointing to the compatibility of their motivations. Nevertheless, the fact that people distinguished implicitly between these two types makes it possible to investigate the extent to which their correlates are different. We hypothesize, for example, that secularism, individualism, and education are correlated with giving priority to universalism but not necessarily to benevolence, whereas conventional religiosity and collectivism are correlated with giving priority to benevolence but not necessarily to universalism (for related ideas and data, see Huismans, 1990; Schwartz, 1990a; Triandis, 1990). The three values related to nature (unity with nature, protecting the environment, a world of beauty) emerged together in the universalism region (cf. Fig. 2) with great consistency. This confirms the idea that concern for nature is closely linked to concern for the welfare of all humankind. The joint emergence of nature, universal welfare, and understanding (broad-minded, wisdom) values in a single region supports the derivation of the motivational goal of universalism that was suggested in the introduction. This goal is presumed to arise with the realization that failure to protect the natural environment or to understand people who are different, and to treat them justly, will lead to strife and to destruction of the resources on which life depends.

3. Tradition and Conformity These two motivational types of values appeared in distinct regions, but tradition was usually found toward the periphery of the circle, outside conformity,

40

SHALOM H. SCHWARTZ

rather than between conformity and benevolence. The order of regions around the circle represents variation in the motivational goals of the value types. Hence, the location of tradition and conformity in the same wedge of the circle suggests that these two value types share the same motivational goal. The motivational goal of conformity values proposed in the theory was “restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms.” The goal of tradition values was “respect, commitment, and acceptance of the customs and ideas that one’s culture or religion impose on the self.” The broader shared goal might be stated as “subordination of self in favor of socially imposed expectations.” Despite their probable common motivational goal, tradition and conformity value types were empirically distinguishable. A difference in the objects to which one subordinates the self may account for their distinctness. Conformity values entail subordination to persons with whom one is in frequent interaction-parents, teachers, and bosses. Tradition values entail subordination to more abstract objects-religious and cultural customs and ideas. As a corollary, conformity values exhort responsivenessto current, possibly changing expectations, whereas tradition values demand responsiveness to immutable expectations set down in the past. Whether it is worthwhile to retain conformity and tradition as distinct value types in the theory will depend upon whether they relate differently to other variables due to their different objects. This is a question for future research. 4 . Power and Achievement

The analyses supported the distinction between power and achievement as motivational types of values because both usually connected with the origin of the circle. However, power values tended to be located more toward the outside of the circle (cf. Fig. 2), and it was possible to draw the boundary of the power region outside of the achievement region in 26 samples. This too may reflect a difference in the objects to which one relates when expressing or pursuing these types of values. As noted in the introduction, both types focus on social esteem. However, achievement values refer more to striving to demonstrate competence in everyday interaction (e.g., ambitious), whereas power values refer more to the abstract outcomes of action in the form of status in the social structure (e.g., wealth). Moreover, achievement values refer to the striving of the individual alone, whereas power values also refer to the hierarchical organization of relations in society. The location of the tradition and power regions toward the periphery of the SSA map reflects a pattern of correlations with substantive significance. A peripheral location indicates that the correlations of a value type range from quite positive, with the types nearby, to negative or much less positive, with the types opposite to it on the map. Central locations signify a narrower range of correla-

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

41

tions with the other value types, all fairly positive. Hence we can infer that tradition values generally conflict more strongly with hedonism, stimulation, and self-direction values than do conformity values. Similarly, power values conflict more strongly with universalism and benevolence values than do achievement values. 5. Security

The inclusion of both individual and group security values in the new survey was intended to address the question of whether there is a single security type or two separate types, individual and group. The analyses within each sample as well as the overall analyses supported the existence of a single security type. This type was consistently located in a region on the boundary between the value types that serve individual interests and those that serve collective interests. The distribution of the single security values within their region (see Figs. 2, 3, and 4), however, suggests that people may distinguish somewhat between the security of self and that of the collectivities of which they are members. Those values most directly concerned with collective interests (social order, family security, national security) were typically located on the side of the region adjacent to conformity and tradition values. The two values most directly concerned with individual interests (healthy, sense of belonging) were located on the side of the region adjacent to the power types. The latter two values sometimes intermixed with preserving my public image and social recognition, two other values whose attainment promotes individual security. However, healthy and sense of belonging were inconsistent in their locations across samples (see Table IV), providing only a weak basis for inferring an individual security subtype. The findings imply that, if there are two subtypes of security values, they are closely related. Indeed, their closeness may indicate that even values that refer to collectivities (e.g., national security) may express the goal of security for self to a significantdegree-the collectivity is viewed as an extension of self. Research into possible differences between the correlates of the individual and collective subtypes will determine whether it is worthwhile to distinguish them in theory.

B. DYNAMIC STRUCTURE OF VALUE TYPES Comparisons of the structural arrangement of value types in each sample with the ideal arrangement based on the theorized compatibilities and conflicts among value types indicated that the ideal was a good, though far from perfect, approximation to a universal structure. In more than half the samples, 1.5 or fewer moves were needed to rearrange the observed structure to match the ideal, but only one sample showed a perfect match. Joining the tradition and conformity

42

S H A M M H. SCHWARTZ

types into one region improved the fit between the observed and ideal structures in 29 samples. This is apparently the best approximation to a universal motivational structure of value types based on our data. No other alternative structural arrangement appeared empirically more than a few times. A perfect match with the ideal structure required that every one of the hypothesized conflicts and compatibilities from which it was constructed be confirmed simultaneously. Therefore, conclusions about the universality of each of the specific, hypothesized value conflicts and compatibilities must be drawn from separate assessments.

1 . Individual, Collective, and Mixed Interests The first principle used to generate hypotheses about dynamic relations among pairs of value types was the interests presumably served by their attainment. We postulated that the set of value types whose attainment serves individual interests (power, achievement, hedonism, stimulation, self-direction) would emerge as one set of adjacent regions, those that serve collective interests (benevolence, tradition, conformity) would emerge as a second set of adjacent regions opposed to the first set, and those that serve mixed interests (universalism, security) would emerge in regions on the boundary between the individual and collective interests regions. The overall SSA analysis (Fig. 2) showed exactly this arrangement. Examination of the analyses in all 40 samples also revealed that, in each and every sample, the value types that serve individual interests and those that serve collective interests formed separate bounded regions that were not intermixed. Thus, the interests that values serve are apparently one universal principle that influences people’s experiences of value conflict and compatibility.

2. Compatibilities We had reasoned that if the simultaneous pursuit of pairs of value types is compatible, then these pairs of types would emerge in adjacent regions in all cultures. We specified nine emphases, assumed to be shared by pairs of value types, that were likely to enable people to pursue these types of values simultaneously. Examination of the adjacency of regions for the pairs of value types we postulated to be compatible supported the idea that these shared emphases organized value priorities in most cultures. Emphases that appeared to organize priorities in at least 88% of samples were (1) concern with enhancement of others and transcendence of selfish interests (uniting benevolence and universalism), (2) reliance upon one’s own judgment and comfort with diversity (uniting self-direction and universalism), (3) intrinsic motivation for mastery and novelty (uniting self-direction and stimulation), (4)

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

43

self-restraint and submission (uniting tradition and conformity), (5) protecting order and harmony in relations (uniting conformity and security), and (6) desire for social superiority and esteem (uniting power and achievement). Not universal but still quite common organizing emphases (supported in at least 70% of samples) were (7) egoistic self-indulgence (uniting hedonism and achievement), (8) desire for affectivelypleasant arousal (uniting hedonism and stimulation), and (9) control of uncertainty (uniting security and power).

3. Conflicts: Basic Conceptual Dimensions Examination of the conflicts observed suggests a simpler way to view value structures. The relationships among the motivational types of values and among the single values can be summarized in terms of a two-dimensional structure. The total value structure can be viewed as composed of four higher order value types that form two basic, bipolar, conceptual dimensions. The first basic dimension places a higher order type combining stimulation and self-direction values in opposition to one combining security, conformity, and tradition values. We call this dimension openness to change versus conservation. It arrays values in terms of the extent to which they motivate people to follow their own intellectual and emotional interests in unpredictable and uncertain directions versus to preserve the status quo and the certainty it provides in relationships with close others, institutions, and traditions. Examination of the SSAs for all 40 samples revealed the presence of this basic dimension in the opposition between the two higher order value types in all samples. Values were counted as representing a higher order type if they were present in its constitutive lower order types in at least 75% of samples (see Table IV). Higher order types were considered in conflict if the regions representing them were separated from each other in both directions around the circle by the regions for the other higher order types. The strength of the opposition can be gauged by counting the frequency with which values representing one higher order type appeared in the region representing the opposing type. The eight specific values that represent the openness to change type (combining self-direction and stimulation) were found empirically in less than 1% of instances in regions of the opposing conservation type (security, conformity, or tradition) across all samples. Similarly, the 14 specific values that represent the higher order conservation type emerged in the opposing higher order openness to change type in less than 1% of instances. The second basic dimension places a higher order type combining power, achievement, and hedonism values in opposition to one combining universalism and benevolence values (including a spiritual life). We call this dimension selfenhancement versus self-rranscendence. It arrays values in terms of the extent to which they motivate people to enhance their own personal interests (even at the

44

SHALOM H. SCHWARTZ

expense of others) versus the extent to which they motivate people to transcend selfish concerns and promote the welfare of others, close and distant, and of nature. This dimension was also revealed in the opposition between the two higher order value types in all 40 samples. The 11 specific values that represent the higher order self-enhancement type were found empirically in regions of the opposing higher order self-transcendence type in less than 1% of the instances across all samples. Similarly, the 17 specific values that represent the higher order self-transcendence type emerged in the opposing higher order self-enhancement type in less than I % of instances. These two basic dimensions encompass three of the four hypothesized and empirically supported conflicts between sets of value types (cf. Table V). The fourth conflict, hedonism versus conformity and tradition, adds no further dimension, but it does point to a duality of meaning in hedonism. This conflict which was observed in 81% of samples, suggests a difference between hedonism and the two other self-enhancement value types (achievement and power). Although similar to them in its focus on self, hedonism is not characterized by the same competitive motivation that achievement and power values express. Moreover, hedonism is virtually free of the motivation to master uncertainty that is apparently present to some degree in achievement and power, as implied by the latter’s frequent proximity to the security-conformity-tradition combination. Instead, hedonism values probably express some degree of the motivation for arousal and challenge that characterizes the higher order openness to change type to which they are most often adjacent. Figure 5 represents the theoretical model of relations among motivational types of values revised to reflect what has been learned from the empirical research. It locates tradition outside conformity and conformity adjacent to benevolence. It also indicates the grouping of the original value types into four broader, higher order types and the two bipolar dimensions that organize relations among these higher order types. Finally, it links hedonism to both the openness to change and the self-enhancement higher order types with which it shares some elements but not others. 4 . A Motivational Continuum

Thus far, we have treated the motivational types of values as discrete categories. This is necessary in order to examine differences in value priorities and to relate specific priorities to other variables. It is also justified both by a priori theory and by the results obtained here. However, the consistent adjacencies and occasional intermixing among pairs of compatible value types, and the emergence of higher order types and basic dimensions, suggest an additional way of thinking about the motivational facet of values.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

45

Self-Transcendence

Se if-Enhancement

u Fig 5. Revised theoretical model of relations among motivational types of values, higher order value types, and bipolar value dimensions.

Single values can also be conceived as arrayed on a continuum of related motivations. This follows directly from Guttman’s (1968) view that, if a sample of items adequately represents all aspects of a content domain, then those items will fill quite evenly the geometrical space formed to represent the intercorrelations among them. Because we have tried to sample all aspects of the values domain, we expect the two-dimensional values space to be filled, with no major gaps. This precludes the appearance of discrete clusters of values that are discernible by empty space around them. Consequently, there should be no clustering that can “reveal” the value types. Rather, it is necessary to partition the space into meaningful regions based on an a priori theory of the conceptual relations among the values. What this means is that the partition lines in the SSAs represent conceptually convenient decisions about where one type of motivation ends and another begins. Because the array of values represents a continuum of motivations, the precise locations of the partition lines are arbitrary. Values found near a partition line express a combination of the related motivational goals associated with the value types on both sides of that partition line. The idea of a continuum of related motivations can be illustrated by examining

46

SHALOM H. SCHWARTZ

the locations of a number of values in the SSA of Fig. 2. For example, within the universalism region, social justice and world at peace are located on the boundary with the benevolence region, whereas wisdom and broad-minded are located nearer the boundary with the self-direction region. This probably signifies that social justice and world at peace express concern for the welfare ncit only of all humankind but also of ingroup members (benevolence). In contrast, the locations of wisdom and broad-minded probably signify that these two values express those aspects of the motivation shared by self-direction and universalismreliance on one’s own judgment and understanding of diversity. Loyal and responsible are located near the boundary of the benevolence and conformity regions. This probably reflects the fact that these two values express two related motives that might induce people to maintain solidarity with close others-the wish to enhance their welfare (benevolence) and the desire to avoid violating others’ expectations (conformity). The borderline location of an exciting life probably reflects the desire for arousal expressed by attributing importance to this value, an element common to hedonism and to stimulation. By moving the partition line, preserving my public image and social recognition could both have been placed in the security rather than in the power region. This borderline location may reflect the goal common to the security and power value types that may underlie both these values-controlling uncertainty in relationships. In sum, the locations of many values support the view that motivational differences between value types can be seen as continuous rather than as discrete. The fuzziness of their boundaries notwithstanding, however, the differences between value types are meaningful. We therefore continue to treat them as discrete when beneficial for research. We recognize, nonetheless, that our theorybased partitioning of the space is arbitrary. It may eventually be superseded by a partitioning, based on a revised theory, that points to discrete value types with greater universal heuristic and predictive power. Such a revised theory must, however, partition the same continuous m a y of values into types, so it is likely to have much in common with the current theory.

5 . Context of Measurement We have operationalized values at the broadest possible contextual level (“guiding principles in MY life”). This followed logically from our definition of values as transsituational, and it permitted the study of all the types of values together. However, because values are expressed in specific contexts, much will be gained from alternative methods that embed values in concrete and varied everyday situations (e.g., school, family, work). Such operations are less likely to reveal basic universals. They are important, however, for clarifying the individual and cultural differences that arise when values are expressed in specific

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

47

judgments and behavior. Studies combining our abstract level of measurement with contextually specific measures would increase our understanding of how values enter into concrete decision-making.

C. GENERALITY OF THE “UNIVERSAL” VALUE STRUCTURE The consistent patterns of compatibility and conflict among value types across cultures support the basic assumption underlying the hypotheses of dynamic relations among values: When people pursue the various types of values, the universal aspects of the human social condition lead to cross-culturally consistent psychological, practical, and social consequences. These consequences, in turn, give rise to the experience of nearly universal conflicts and compatibilities in value priorities. Note that this is a statement about universality in the structure of values, not about the universality of their relative importance. Importance differences are ubiquitous (see below). The generality of the common structure observed so far is supported by the fact that it was quite accurate as a prototype for samples from very diverse cultural, linguistic, geographic, religious, and racial groups. Essentially the same structure was found for samples of university students and of school teachers, and the five occupationally heterogeneous samples of adults also yielded structures of values similar to the ideal structure. Nonetheless, we do not believe that any single value structure is likely to be truly universal, so one must not generalize indiscriminately to new samples. Value structures probably evolve over time as social conditions are transformed. Value structures may even change rapidly in response to major technological, economic, political, and security upheavals. There is a hint in our data, for example, that the usual opposition between universalism values and two security values (national security and social order) is not found in nations undergoing popular revolutions in which totalitarian or colonial regimes have crumbled. Insofar as the basic human condition in which values are grounded remains fairly constant, however, we anticipate that major variations in value structure will be rare. Moreover, changes brought on by sudden social upheavals may only be temporary. For example, the usual opposition between security and universalism values may reemerge after several years, once sociopolitical stability has been restored. Most likely, even when value structures vary due to sample characteristics, they still retain the basic structure anchored in the two dimensions identified here. Consider, for example, how education might effect value structures: The value systems of people with less education may be less well-defined, and value types may therefore not be as distinct in less educated samples. Because most

48

SHALOM H. SCHWARTZ

respondents in all our samples had at least a high school education, we could not study this possible effect. Our samples did differ widely, however, on numerous potential moderators of value structure such as age, gender, and religiosity. Hence, the observed consistency of value structures diminishes the probability that these sample characteristics dramatically alter value structures. Future systematic comparisons of samples may, nonetheless, reveal differences in how people construe the relations among values. In current analyses in Israel, for example, we find that secular people view tradition values as more opposed to self-direction and universalism values and as closer to power values than the religious do. Here again, exceptions uncovered in the search for universality in value structures could be very enlightening. Why did some specific compatibilities and conflicts fail to appear in some of the samples studied here? Before trying to explain anomalous findings, it is critical to discover whether they replicate in multiple samples from the same culture. We plan to address this question by adding samples from new cultures as well as from those studied thus far, with the goal of identifying systematic characteristics of the samples that may explain these exceptions. Certain hints in the data (e.g., in China) suggest intriguing possibilities. As noted above, the three Shanghai samples and the (Chinese) Hebei teachers sample deviated substantially from the ideal structure. The power, achievement, hedonism, stimulation, and self-direction value types emerged clearly in all four samples, and their order relative to each other and in opposition to the remaining values conformed with our theory. However, the values that constitute the other five types could not be partitioned into regions representing each type. Instead, we were able to partition them into three interpretable regions that were clear in the three Shanghai samples and somewhat less clear in Hebei. We tentatively propose that these regions represent three uniquely Chinese value types. The labels we have given to these three types highlight the central motivational goal shared by the values that appeared in each region in at least three of the four samples: societal harmony (social order, national security, social justice, world peace, protecting the environment, world of beauty); virtuous interpersonal behavior (honest, obedient, responsible, loyal, politeness, humble, self-discipline, forgiving, helpful); personal and interpersonal hrmony (family security, honoring parents and elders, accepting my portion in life, inner harmony, healthy, devout). Zhi-gang (1990) linked each of these types to one of the major religious influences in Chinese culture: Societal harmony was linked to Taoism, virtuous interpersonal behavior to Confucianism, and personal and interpersonal harmony to Buddhism. The SSA results from other Oriental samples (Hong Kong, Japan, Taiwan, and Guangzhou) were closer to the ideal structure and did not show this alternative pattern. Data from other cultures with related religious traditions (e.g., Korea and Thailand) should help to clarify the incidence of this alternative value structure.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

49

Our current sampling of cultures was very broad. Still, data from an even wider range of cultures might suggest the existence of one or more variants of the alternative value structure. The samples that deviated most from the ideal structure (requiring 3 or more moves to fit the ideal, see Table 111) were from China, Estonia, Hong Kong, Poland, Taiwan, and Zimbabwe. One might suggest that these samples are from countries that are somewhat less exposed to Western influence than most of the rest. If so, analyses of additional non-Western samples may reveal other alternative structures. We have not detected any alternative structure that is common to several samples, other than the Chinese alternative. However, one may yet emerge. We would be surprised should an alternative materialize in which the two basic oppositions present in the ideal structure are absent.

D. TERMINAL VERSUS INSTRUMENTAL VALUES Separate regions of terminal and of instrumental values failed to appear in the SSA projections in the vast majority of samples. This outcome casts stronidoubt on the significance of the terminal-instrumental facet in organizing people’s values. Above, we have described the relevant procedure that differentiated the current research from earlier studies in which the terminal-instrumental facet emerged consistently in the SSA projections (Schwartz & Bilsky, 1987, 1990). By having respondents anchor their use of the response scale prior to rating the values, we sought to eliminate or reduce shifts of scale use that could have produced the distinct regions that were previously observed. Results of our structural analyses imply that the terminal-instrumental distinction does not affect the way people relate to values. However, Rokeach (1973) and Feather (1975) have asserted that these two forms of values have different impacts on judgment and behavior. This argument can only be settled by studying whether the correlates of the two forms differ. Weishut (1989) undertook such a study, exploiting the fact that many instrumental values can be transformed into terminal phrasing (into nouns: e.g., independent to independence) and terminal values can be transformed into instrumental phrasing (into adjectives: e.g., politeness to polite). He found that, for most values, both phrasings showed quite similar correlations with background and attitudinal variables and that both phrasings received similar importance ratings. Weishut’s findings combine with ours to suggest that only a single form, either terminal or instrumental, is needed to measure values. Use of one form has the advantage of permitting random ordering of all values in a single list. The terminal phrasing seems preferable to the instrumental for two reasons-people usually think of values in noun form (wisdom rather than wise), and the instrumental phrasing is sometimes misunderstood as though it were asking for personality descriptions rather than for value priorities. Moreover, virtually all values

50

SHALOM H.SCHWARTZ

can be phrased as terminal, but some terminal values (e.g., world at peace, pleasure, authority) cannot easily be phrased as instrumental. We view this lack of complete interchangeabilityas a basis for caution, however. Some values may undergo subtle changes of meaning when transformed from instrumental to terminal phrasing-changes that might even move them from one motivational type to another.

E. NORMATIVE IDEALS OR PERSONAL PRIORITIES? When individuals respond to the values survey, do they report their own personal value priorities, or do they report the normatively approved ideals of their group or culture? Respondents are instructed to ask themselves: “What values are important to ME as guiding principles in MY life?” Nonetheless, because values represent cultural ideals, people may be inclined to provide the value priorities they perceive as desirable. Even if they intend to report their personal priorities, people’s responses will largely reflect normative hierarchies if these priorities are determined by cultural norms. If that were so, then the value contents and structure we have identified might better be understood as reflective of cultural conventions rather than as expressive of the psychological characteristics of individuals. We propose two empirical arguments for concluding that the responses to the values survey primarily reflect personal value priorities. First, if responses were determined by cultural ideals, then we would expect high group consensus with regard to the importance of each value. However, in every sample studied, there was substantial individual variance in response to every single value. Indeed, most values elicited ratings over the full scale from - 1 to 7 in most samples, and no value elicited responses over fewer than four scale points in any sample. Second, if responses were fully determined by perceived cultural ideals, then this individual variance might best be interpreted as reflective of error in measurement or error in respondents’ perception of the cultural ideals. In either case, individual variance in value priorities would not be associated systematically with individual differences in background characteristics, in attitudes, or in behavior. However, virtually every association we have examined within samples between individual value priorities and other variables (e.g., age, occupation, gender, education, voting behavior, religiosity, attitudes toward ecology, national pride) has yielded numerous reliable and meaningful findings.l5 Wonsider a few examples of findings, relevant to different value types, that replicated across several samples. Age correlated positively with giving priority to security values and negatively with giving priority to hedonism values. Education correlated positively with emphasizing self-direction values and negatively with emphasizing tradition values. Religiosity conelated positively with giving

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

51

We therefore conclude that responses to the survey primarily reflect personal value priorities. This means that one cannot derive the normative ideals of a culture from the average of individual responses. Perceived normative ideals could be measured with the survey if the referent use to anchor the question were modified, For example, respondents could be asked: “How important is value X as a guiding principle in people’s lives, in the eyes of (culture group Y)?” In some societies, this type of question might cause difficulty and provoke annoyance. This would probably signify a low consensus regarding value priorities and could possibly be overcome by more narrowly specifying the cultural group in question. In other societies, people find it easy to report on consensual, cultural ideals (e.g., India; Jai B. P. Sinha, personal communication, July 20, 1990). In order to characterize actual rather than ideal cultural value priorities, other approaches are required. The average individual priorities of a sample of culture members reflect the value pressures to which they are all exposed through the shared socialization, laws, media, etc., of the culture. Hence the average individual value profile is one way to characterize cultural value priorities. Alternative approaches include content analyses of cultural products (newspapers, child-rearing manuals, popular novels, television shows, films, etc.) and indirect inferences from societal-level functions or statistics (form of legal systems; educational systems; allocations of resources and power among health, political, and economic institutions; demographic distributions; etc. ). The measurement of both individual and cultural value priorities would permit the study of interesting questions concerning the fit between personal and cultural priorities (cf. Feather, 1975).

V. Studying the Antecedents and Consequences of Value Priorities Research on values in the social sciences is primarily concerned with relating the value priorities of individuals or groups to their antecedents in socially structured experience and cultural background on the one hand, and to general attitudes and patterns of behavior on the other. For these purposes, it is necessary to construct indexes of the importance of the values to be studied. Our data strongly support the assumption that members of almost all cultures, when they relate to values as guiding principles, implicitly distinguish 10 basic types of priority to conformity values and negatively with giving priority to stimulation values. Voting for parties on the economic left was associated with emphasizing benevolence values and deemphasizing power values. Voting for parties favorable to classical liberalism was associated with preferring selfdirection values over conformity values. These and other findings will be reported in future papers.

52

SHALOM H. SCHWARTZ

values that express different motivational goals. It is therefore desirable to propose empirical indexes to measure the importance of each of these types for use in future research.

A. INDEXES OF VALUE TYPES Ideally, a structural analysis such as those reported here should be undertaken with the data from any new sample before constructing indexes for each value type. This would reveal the actual value types implicitly distinguished in the sample and the specific values that constitute them,. However, the consistency of our results permits us to propose core cross-cultural indexes for use when new structural analyses are not feasible. These core indexes are built from the data in Table IV. We recommend including in the cross-cultural index for each value type those values that emerged empirically in a region representing that value type in at least 75% of the samples. That is, the 45 values that emerged in one single region in at least 27 of 36 samples should be used (see Table IV). The meanings of these 45 values did not vary much even when they did not emerge in their usual region: They appeared in one of the postulated adjacent regions or in the region of their a priori type in at least 88% of the samples (83% for loyal and 81% for responsible). Because values whose meanings showed less consistency across cultures are excluded, these core indexes can be employed for cross-cultural comparisons. To examine the internal consistency of the core cross-cultural indexes, we computed a coefficient a for the index of each value type in the heterogeneous adult samples from Australia, Holland, Israel, and Japan (Schwartz, Sagiv, 8c Antonovsky, 1991). The stimulation index showed the highest reliability across samples (mean .75, range .70 to .79), and the tradition index showed the lowest reliability (mean .55, range .49 to .69). All reliabilities in all samples were greater than .45, averaging .67 in Australia, .68 in Holland, .71 in Israel, and .60in Japan. Considering the small number of items in each index, these reliabilities are quite reasonable. The availability of multiple indicators for each value type also makes it possible to perform latent variable analyses for causal modeling. For comparisons between groups within a culture, the results of SSAs for that culture may suggest additional values that could be added to the index of a value type. In Hong Kong, for example, intelligent could be added to the index of achievement values because it appeared in that region in both samples. Conversely, SSA results within a culture may suggest that greater accuracy will be attained if some values included in the cross-cultural indexes are excluded. For

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

53

example, forgiving might be dropped from the benevolence index in Japan because, as noted, it appeared in the universalism region in all three Japanese samples. The simplest index of the importance of each value type for an individual is the mean importance the individual attributed to the set of values that constitutes that type. Individuals differ, however, in the ways in which they distribute their importance ratings across the rating scale, and groups may also display such scale use differences. It is therefore desirable to control statistically for differences in scale use when comparing value priorities of various cultural and other groups or when correlating value priorities with other variables within groups. The most common practice is to standardize individuals’ ratings of each value around their own mean or to standardize groups’ ratings around the means for all the groups compared (e.g., Hofstede, 1980; Leung & Bond, 1989). However, standardizing changes the pattern of intercorrelations among values within groups. As a result, SSAs based on standardized ratings rarely yield the same clear regions for each value type that are obtained with analyses of the raw scores. This blumng of the consistent, distinct regions for each motivational type of values, in turn, undermines the rationale for constructing the indexes of value types. To avoid this problem and still control for scale use differences, we recommend a procedure that has little effect on the structure of value relations observed within samples. We propose to use each individual’s mean importance rating for the 56 core values as a covariate in comparisons of group means, or as a third variable whose effect on the correlations between value priorities and other variables is controlled through partial correlation. SSA analyses based on matrixes of intercorrelations among value ratings, partialled for mean importance ratings, yield regional partitionings of value types that are very similar to those based on the raw scores. Hence, this approach does not undermine the rationale for using mean importance scores for the values located in the regions as indexes of the importance of value types. Indexes of the importance of the four higher order value types can be constructed in a similar manner-computing the mean importance of all values that constitute the higher order type. We recommend including in these indexes the values that emerged empirically in the region of the higher order type in at least 75% of samples. Using this criterion, several values that were excluded from the indexes of the specific value types could be included (social recognition in the self-enhancement index; inner harmony, a spiritual life, true friendship, and mature love in the self-transcendence index). Analyses using these indexes should also be controlled for scale use differences by partialing out each individual’s mean importance rating for the 56 single values.

54

SHALOM H. SCHWARTZ

B. GENERATING HYPOTHESES BASED ON THE VALUE STRUCTURE When generating hypotheses about the relations of value priorities with other variables, it is advantageous to conceive people’s value systems as integrated structures of motivational types. Every hypothesis that specifies the association of one value type with an outside variable has clear implications for the associations of the other value types as well. Say, for example, that our theory predicts that the importance attributed to conformity values increases with age. It follows from the opposition in the motivational structure of values between the conformity and stimulation value types (see Figs. 1-4) that the importance of stimulation values would decrease with age. It also follows that the importance of tradition and security values is likely to increase with age, and the importance of hedonism and self-direction values is likely to decrease. Two statements summarize the implications of the interrelatedness of value priorities for generating hypotheses: (1) Any outside variable tends to be similarly associated with value types that are adjacent in the value structure. (2) Associations with any outside variable decrease monotonically as one goes around the circular structure of value types in both directions from the most positively associated value type to the least positively associated value type. Once theory specifies the value type with which an outside variable has its most positive (or least negative) association and its least positive (or most negative) association, the order of the value structure enables us to fill in the expected pattern of associations with all value types. When the whole pattern of associations has been predicted, even nonsignificant associations provide meaningful information. The statistical significance of single correlations or mean differences is no longer the critical test of theory; rather, the whole pattern of associations-both significant and nonsignificant-reflects on the viability of the theory. 1. The Sinusoid Curve of Value Associations

The logic of the organization of the value structure means that predicted associations between value priorities and any outside variable can be represented graphically with a sinusoid curve.16 To draw such a curve, the value types are arrayed on the horizontal axis according to their order around the circular value structure. The strength of association with the outside variable is then plotted on the vertical axis. Figure 6 displays various sinusoid curves that reflect possible patterns of association. Curve A , for example, depicts the hypothetical comlations of value priorities with age as described above. From this curve we can W p k e Huismans suggested this mode of presentation.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

55

+I

-I

SD ST HE AC PO SE CO TR BE UN Fig. 6 . Hypothetical curves representing associations between value priorities and outside variables. SD, self-direction; ST, stimulation; HE, hedonism; AC, achievement; PO, power; SE, security; CO, conformity; TR, tradition; BE, benevolence; UN, universalism. Curves: (A) age; (B) contractual minus communal societies; (C) nationalistic political orientation.

derive predictions that were not generated originally from theory. For example, the curve implies that age is correlated more positively with giving priority to benevolence and power values than to universalism and achievement values, and that these correlations are intermediate in relation to the correlations with the other value types. As this example demonstrates, conceiving value systems as integrated structures of motivational types suggests ways to enrich and expand theory. The sinusoid curve does not replace theory building, of course. Rather, the curve should stimulate thought about why one might expect the particular empirical associations it implies. Theoretical analyses of the processes that might link outside variables with the priority given each specific value type must provide the rationales for these expectations. Theoretical analyses may also suggest deviations from the sinusoid pattern of associations. Deviations, whether grounded in a priori theory or discovered in empirical data, are especially interesting because they focus attention on special circumstances. Consider, for example, a sample in which attributing importance to tradition values decreases with age, although the remaining associations are as predicted. This would suggest that, in this sample, a cohort effect related particularly to tradition is at work (e.g., a return to religion among youth). The hypothesized associations of continuous variables with value priorities are best represented as correlations. Thus, curve A depicts the hypothesized correla-

56

SHALOM H. SCHWARTZ

tions between age and the priority attributed to each value type. The sinusoid curves depict the correlations for each value type relative to the other value types. Neither the absolute levels nor the directions of the correlations are apparent from the curves. Theory should specify whether correlations are expected to be positive or negative. However, when the outside variable is correlated with scale use in the value ratings, the signs of zero-order correlations will be misleading. For example, age may well be correlated with a tendency to give higher importance to values, that is, older people may use the response scale differently from younger people. This problem is eliminated by partialing out the mean rating each individual gives to all 56 values. The resulting partial correlations should then exhibit the signs predicted by theory. Many antecedents of value priorities are not continuous variables. Cultural background or unique experiences, for example, are usually nominal variables. To test hypotheses about the effects of such antecedents on value priorities, mean differences in the importance attributed to each value type are ordinarily examined. Hypotheses about the pattern of mean differences for the whole set of value types can also be summarized graphically, with the value types arrayed along the horizontal axis according to their theorized structural order. In this case, the vertical axis represents the mean importance of a value type in group X minus its mean importance in group Y. The theory postulates, for example, that, if one cultural group gives greater importance to power values than a second group, the second will tend to give greater importance to benevolence values than the first. A plot of the mean differences between groups for each value type should take a sinusoid form according to our structural theory. The mean rating each individual gives to all 56 values should be used as a covariate to eliminate scale use differences between groups. The plot then yields an accurate comparison of the relative priorities for each value type in the two groups.

2 . Exemplary Applications We now provide two illustrations of how consideration of the whole value structure can inform the study of the relations between value priorities and their antecedents or consequences. Because our aim is only to clarify the approaches that can be used and how they operate, the theories underlying the examples chosen will only be developed partially. The mean differences approach is exemplified with hypotheses regarding differences in value priorities between persons from communal societies versus contractual societies (Schwartz, 1990a).l7 Communal societies are characterized "This ideal type distinction is related to the common distinction between collectivist and individualist cultures (Hofstede, 1980; Triandis, 1990). However, this common distinction confounds ideological (value) and social structural elements, making value predictions partly tautological. The current ideal type distinction was made on purely social structural grounds.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

57

by extended primary groups in which people have diffuse mutual obligations and expectations based largely on their enduring ascribed statuses. Contractual societies are characterized by narrow primary groups and by secondary social relations in which people develop specific obligations and expectations largely through negotiation in the process of achieving and modifying statuses. We hypothesized that tradition, conformity, and benevolence values are more important in communal societies, whereas self-direction, stimulation, and universalism values are more important in contractual societies. Reasons for differences in the importance of security, power, achievement, and hedonism values were considered, but no clear hypotheses could be derived. Curve B in Fig. 6 portrays the predicted differences between the mean importance of each value type in contractual versus communal societies (e.g., the United States or New Zealand versus Korea or Taiwan). We assumed that the largest differences occur for conformity and stimulation. The complete sinusoid curve suggests hypotheses for the four types of values whose relations to societal structure were viewed as equivocal in earlier theorizing. Hedonism should be more important in contractual societies and security should be more important in communal societies. The differences regarding achievement and power should be small, although communal societies may attribute more importance to the former and contractual societies may attribute more importance to the latter. The emergence of these additional hypotheses, for which rationales can easily be generated, testifies to the usefulness of the structural approach. Data exemplifying a test of these hypotheses by comparing our teachers samples from New Zealand (more contractual) and Taiwan (more communal) are presented in Fig. 7. The average importance of each motivational type of value in the Taiwan sample has been subtracted from the average importance of that type in the New Zealand sample, after correcting for scale use. The curve based on these data approximates a sinusoid shape except for the more positive than expected differencefor the benevolence value type. It differs from curve B in Fig. 6 in one minor and one major respect. The minor deviation is that the largest difference in favor of the more communal sample was in the importance of the security rather than the adjacent conformity values. The major deviation is that the contractual sample attributed considerably more rather than less importance to benevolence values than the communal sample did. Thus, although the overall pattern of mean differences is supportive of the underlying theory, one deviation raises a problem requiring further investigation. 18This problem is probably due to the use of value types appropriate for discriminating among individuals rather than among cultures. We now argue (Schwartz, in press) that comparisons of the value priorities of cultures should be made on value types derived from analyses at the culture level. We have recently derived seven such culture-level value types. They have much in common with the 10 individual-level types, but their structure reveals that the priorities of cultures and of individuals are organized somewhat differently. In particular, benevolence values are part of a broader culturelevel type (labeled social concern) that is more strongly emphasized in samples we classify as from

SHALOM H. SCHWARTZ 0.7

I

0

-1.2

SD

ST

HE

AC

SE

PO

CO

TR

BE

UN

Fig. 7. Differences in the mean importance attributed to motivational types of values by teachers in a sample from a more contractual society (New Zealandf minus a sample from a more communal society (Taiwan). Abbreviations as in Fig. 6.

Finally, consider an example of applying the structural approach to predict a possible consequence of value priorities-nationalistic political views. On theoretical grounds, we would assume that nationalistic political views are primarily anchored in concerns for personal and collective safety. Nationalist policies emphasize the protection of one’s own group against dangers from without or within. Groups with ideas and aims different from one’s own are seen as threatening; they are to be controlled or suppressed rather than understood and appreciated. An authoritative organization of social relations is desirable in order to mobilize and control the human and material resources needed to overcome threats. Individuals with particular value priorities will be attracted to nationalism or repelled by it, depending upon their needs and goals. Those who give high priority to security and power values are likely to find nationalism a compatible political expression of their guiding principles. In contrast, individuals who give high priority to universalism and self-direction values are likely to view nationalism as the antithesis of their guiding principles. Nationalistic policies also seem compatible with emphases on conformity and tradition but contrary to the ~

~

~

~

~~

~

~

contractual societies than in those we classify as from communal societies. Reasons for this, related mainly to differences in the conception of the person prevalent in these societies, are discussed in Schwartz (in press).

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

59

openness to change implied by stimulation values. Curve C in Fig. 6, representing correlations between value priorities and nationalist political views, summarizes this hypothetical set of associations. Note that curve C is not symmetrical: The hypothesized drop in correlations from power to stimulation values is much steeper than the drop from power to tradition values. This illustrates the fact that the value structure implies only the order of associations for the value types. It does not specify a metric measure of their strength relative to one another. To specify relative strength of associations, theoretical analyses of the links between the goal that a value type expresses and the nature of the outside variable are required. The asymmetry of curve C, for example, would reflect an analysis that finds nationalistic political views quite expressive of tradition values but not especially expressive of stimulation values. These examples have illustrated some of the advantages of conceiving value systems as integrated motivational structures for the generation of hypotheses about the relations of value priorities with other variables. We have begun to test the types of hypotheses we have suggested with data from the samples reported here as well as from other samples. Substantial support is emerging for the idea that associations between value priorities and variables such as age, education, political orientations, religiosity (Schwartz & Huismans, 199I), and social desirability (Schwartz et al., 1991) reflect the integrated motivational structure: These associations are depicted quite accurately by the sinusoid curve.

VI. Concluding Remarks At the outset, we posed four basic issues that the present research has addressed: What are the substantive contents of human values? Can we identify a comprehensive set of values? To what extent is the meaning of particular vaIues equivalent for different groups of people? How are the relations among different values structured? These issues required resolution before the antecedents and consequences of value priorities, or cross-cultural differences in such priorities, could be studied effectively. Substantial progress has been made toward resolving each of these issues. First, we identified 10 motivationally distinct value types that are likely to be recognized within and across cultures and used to form value priorities. Second, we demonstrated that this set of value types is relatively comprehensive, encompassing virtually all the types of values to which individuals attribute at least moderate importance as criteria of evaluation. Third, we assembled evidence from 20 countries showing that the meaning of the value types and of most of the single values that constitute them is reasonably equivalent across most groups. Finally, we discovered two basic dimensions that organize value systems into an

60

SHALOM H . SCHWARTZ

integrated motivational structure with consistent value conflicts and compatibilities. We have pointed to directions for possible refinement in the ways we have resolved each issue. Nonetheless, the main conclusions regarding the content, measurement, and structure of values seem sufficiently well established to justify their adoption as the basis for future research into questions of the type we have posed. By identifying universal aspects of value content and structure, this article has laid the foundations for investigating culture-specific aspects in the future. Against the background of common meanings and structure, it is now possible to compare the value priorities of cultures and groups and to detect genuine variation. l9 Observed differences between groups with regard to the priority they attribute to values will be interpretable in light of equivalences or differences in the meanings of the values revealed here. Unique, culture-specific understandings and applications of values will stand out against the universal patterns we have elucidated. The instrument developed to measure all the value types can be used to test hypotheses that relate value priorities to their antecedents and consequences. The theory of value structures can stimulate the generation of hypotheses about how the whole integrated system of value priorities relates to background, attitude, and behavior variables.

VII. Appendix: Values Included in the Survey Instrument 1 -EQUALITY (equal opportunity for all) 2 -INNER HARMONY (at peace with myself) 3 SOCIAL POWER (control over others, dominance) 4 -PLEASURE (gratification of desires) 5 -FREEDOM (freedom of action and thought) 6 -A SPIRITUAL LIFE (emphasis on spiritual not material matters) 7 -SENSE OF BELONGING (feeling that others care about me) 8 -SOCIAL ORDER (stability of society) 9 -AN EXCITING LIFE (stimulating experiences) 10 -MEANING IN LIFE (a purpose in life)

-

*gSchwartz(in press) reports value priority profiles of 60 samples from 28 cultures on culturelevel value types that were based on the research reported here. For example, students and teachers from the United States are conspicuous for the high importance they attribute to values expressing a desire to get ahead personally in the social hierarchy (e.g., wealth, authority, ambitious, successful) and the low importance they attribute to values expressing social concern (e.g., social justice, equality, loyal, responsible). Spanish and Italian students and teachers show the opposite pattern.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

61

1 I -POLITENESS (courtesy, good manners) 12 -WEALTH (material possessions, money) I3 -NATIONAL SECURITY (protection of my nation from enemies) 14 -SELF-RESPECT (belief in one’s own worth) 15 -RECIPROCATION OF FAVORS (avoidance of indebtedness) 16 -CREATIVITY (uniqueness, imagination) 17 -A WORLD AT PEACE (free of war and conflict) 18 -RESPECT FOR TRADITION (preservation of time-honored customs) 19 -MATURE LOVE (deep emotional and spiritual intimacy) 20 -SELF-DISCIPLINE (self-restraint, resistance to temptation) 21 -DETACHMENT (from worldly concerns) 22 -FAMILY SECURITY (safety for loved ones) 23 -SOCIAL RECOGNITION (respect, approval by others) 24 -UNITY WITH NATURE (fitting into nature) 25 -A VARIED LIFE (filled with challenge, novelty, and change) 26 -WISDOM (a mature understanding of life) 27 -AUTHORITY (the right to lead or command) 28 -TRUE FRIENDSHIP (close, supportive friends) 29 -A WORLD OF BEAUTY (beauty of nature and the arts) 30 -SOCIAL JUSTICE (correcting injustice, care for the weak) 3 1 -INDEPENDENT (self-reliant, self-sufficient) 32 __MODERATE (avoiding extremes of feeling and action) 33 -LOYAL (faithful to my friends, group) 34 -AMBITIOUS (hardworking, aspiring) 35 -BROAD-MINDED (tolerant of different ideas and beliefs) 36 -HUMBLE (modest, self-effacing) 37 -DARING (seeking adventure, risk) 38 -PROTECTING THE ENVIRONMENT (preserving nature) 39 -INFLUENTIAL (having an impact on people and events) 40 -HONORING OF PARENTS AND ELDERS (showing respect) 41 -CHOOSING OWN GOALS (selecting own purposes) 42 -HEALTHY (not being sick physically or mentally) 43 -CAPABLE (competent, effective, efficient) 44 -ACCEPTING MY PORTION IN LIFE (submitting to life’s circumstances) 45 -HONEST (genuine, sincere) 46 -PRESERVING MY PUBLIC IMAGE (protecting my “face”) 47 -OBEDIENT (dutiful, meeting obligations) 48 -INTELLIGENT (logical, thinking) 49 -HELPFUL (working for the welfare of others) 50 -ENJOYING LIFE (enjoying food, sex, leisure, etc.) 5 1 -DEVOUT (holding to religious faith and belief)

62

SHALOM H. SCHWARTZ

52 -RESPONSIBLE (dependable, reliable) 53 -CURIOUS (interested in everything, exploring) 54 -FORGIVING (willing to pardon others) 55 -SUCCESSFUL (achieving goals) 56 -CLEAN (neat, tidy)

Acknowledgments This reskarch was supported by Grant No. 88-OOO85 from the United States-Israel Binational Science Foundation (BSF), Jerusalem, Israel. The contributions of the following persons in gathering the data reported here are gratefully acknowledged: Michael Bond (Hong Kong), Bram Buunk (Holland), Bartolo Campos and Isabel Menezes (Portugal), Norman T. Feather (Australia), James Georgas (Greece), Sipke Huismans (Holland), Saburo Iwawaki (Japan), Sumiko Iwao (Japan), Maria Jarymowicz (Poland), Leo Montada (Germany), Kathleen Myambo and Patrick Chiroro (Zimbabwe), Toomas Niit (Estonia), Wu Peiguan (People’s Republic of China), Mark Radford (Japan), Maria Ros and Hector Grad (Spain), Jose Miguel Salazar (Venezuela), Osamu Takagi (Japan), Alvaro Tamayo (Brazil), Giancarlo Tanucci (Italy), Harry Triandis (United States), Antti Uutela and Markku Verkasalo (Finland), Colleen Ward (New Zealand), Louis Young (Taiwan), and Wei Zhi-Gang (People’s Republic of China).

References Allport, G. W. (1961). Pattern and growth in personaliry. New York: Holt, Rinehart, & Winston. Bandura, A. (1977). Self efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215. Bem, D. J. (1970). Beliefs, attitudes and human afairs. Belmont, CA: BrooksKole. Bentham, J. ( 1948). An introduction to the principles of morals and legislation. New York: Harper. (Original work published 1838) Berlyne, D. E. (1960). Conflict, arousal and curiosiry. New York: McGraw-Hill. Berry, J. W. (1 975). An ecological approach to cross-cultural psychology. Nederlands Ttjdschrifi voor de Psychologie. 30, 5 1-84. Berry, J. W. (1986). The comparative study of cognitive abilities: A summary. In S . E. Newstead, S . H. Irvine, & P. L. Dann (Eds.),Human assessment: Cognition and morivation (pp. 57-74). Dordrecht, Netherlands: Martinus Nijhoff. Borg, I . , & Lingoes, J. C. (1987). Multidimensional similarity structure analysis. New York Springer-Verlag. Braithwaite, V. A., & Law, H. G. (1985). Structure of human values: Testing the adequacy of the Rokeach Value Survey. Journal of Personaliry and Social Psychology, 49, 250-263. Buber, M. (1958). I and thou. New York Charles Scribner’s Sons. Buss, D. M. (1986). Can a social science be anchored in evolutionary biology? Four problems and a strategic solution. Revue Europeenne des Sciences Sociales, 24, 41-50. Canter, D. (Ed.).(1985). Facer theory: Approaches to social research. New York: Springer-Verlag.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

63

Chinese Culture Connection. (1987). Chinese values and the search for culture-free dimensions of culture. Journal of Cross-Cultural Psychology, 18, 143- 164. Coles, R. (1990). The spiritual life of children. Boston: Houghton Mifflin. Coxon, A. P. M. (1982). The users guide to multidimensional scaling with special reference to the MDS (x) library of computer programs. London: Heinemann Educational. Davison, M. (1983). Multidimensional scaling. New York: Wiley. Deci, E. L. (1975). Intrinsic motivation. New York: Plenum. Dewey, J. (1957). Human nature and conduct. New York: Modern Library. Dillon, W. R., & Goldstein, M. (1984). Multivariate analysis. New York: Wiley. Durkheim, E. (1954). The elementaryforms of religious life (J. W. Swain, Trans.). Glencoe, 11: Free Press. (Original work published 1912) Durkheim, E. (1964). The division of labor in society ( G . Simpson, Trans.). Glencoe, 1L: Free Press. (Original work published 1893) Farley, F. (1986). The big T in personality. Psychology Today, 20, 44-52. Feather, N. T. (1975). Values in education and society. New York: Free Press. Freud, S. (1930). Civilization and its disconrenrs. London: Hogarth Press. Freud, S. (1933). New introductory lectures in psychoanalysis. New York: Norton. Georgas, J. (1988). An ecological and social cross-cultural model: The case of Greece. In 1. W. Beny, S. H. Irvine, & E. B. Hunt (Eds.), Indigenous cognition: Functioning in cultural context (pp. 105-123). Dordrecht, Netherlands: Martinus Nijhoff. Glock, G. Y., & Stark, R. (1965). Religion and society in tension. Chicago: Rand McNally. Gordon, L. (1960). Survey of inrerpersonal values. Chicago: Science Research Associates. Guttman, L. (1968). A general nonmetric technique for finding the smallest coordinate space for a configuration of points. Psychometrica, 33, 469-506. Guttman, L. (1982). Facet theory, smallest space analysis and factor analysis. Perceptual and Motor Skills, 54, 491-493. Heschel, A. J. (1955). God in search of man. New York: World. Hofstede, G. (1980). Culture’s consequences: International dlferences in work-related values. Beverly Hills, CA: Sage. Homer, P. M., & Kahle, L. R. (1988). A structural equation test of the value-attitude-behavior hierarchy. Journal of Personality and Social Psychology, 54, 638-646. Houston, J. P., & Mednick. S. A. (1963). Creativity and the need for novelty. Journal of Abnormal and Social Psychology, 66, 137-141. Huisrnans, S. (1990, July). Value priorities and religiosity among Carholics, Protestants and Jews. Paper presented at the 10th International Association of Cross-Cultural Psychology Congress, Nara, Japan. Kaplan, A . (1961). The new world of philosophy. New York: Random House. King, W. L. (1954). Introduction to religion. New York: Harper. Kluckhohn, C. (1951). Values and value-orientations in the theory of action: An exploration in definition and classification. In T. Parsons & E. Shils (Eds.), Toward a general theory ofacrion (pp. 388-433). Cambridge. MA: Harvard University Press. Kohn, M. L., & Schooler, C. (1983). Work and personality. Norwood, NJ: Ablex. Korman, A. (1974). The psychology of motivation. Englewood Cliffs, NJ: Prentice-Hall. Kristiansen, C. M. (1990). The role of values in the relation between gender and health behaviour. Social Behaviour, 5, 127- 134. Leung. K.,& Bond, M. H. (1989). On the empirical identification of dimensions for cross-cultural comparison. Journal of Cross-Cultural Psychology, 20, 133- 15 1 . Levy, S. (1985). Lawful roles of facets in social theories. In D. Canter (Ed.), The facet approach to social research ( pp. 59-96). New York: Springer-Verlag. Levy, S . (1990). Values and deeds. Applied Psychology: An International Review, 39, 379-400.

64

SHALOM H. SCHWARTZ

Levy, S., & Guttman, L.(1974). Values and attitudes of Israeli high school youth. Jerusalem: Israel Institute of Applied Social Research. Lingoes, J. C. (1977). Geometric representations of relational data. Ann Arbor, MI: Mathesis. Lingoes, J. C. (1981). Testing regional hypotheses in multidimensional scaling. In 1. Borg (Ed.), Multidimensional data representations: When and why (pp. 280-310). Ann Arbor, MI: Mathesis. Lonner, W. J. (1980). The search for psychological universals. In H. C. Triandis & W. W. Lambert (Eds.), Handbook of cross-cultural psychology: Vol. I . Perspectives. Boston: Allyn & Bacon. Lovejoy, A. 0. (1950). Terminal and adjectival values. Journal of Philosophy, 47, 593-608. Maddi, S. R. (1961). Exploratory behavior and variation-seeking in man. In D. W. Fiske & S. R. Maddi (Eds.), Functions of varied experience (pp. 253-277). Homewood, I t : Doney. Maslow, A. H. (Ed.). (1959). New knowledge in human values. New York: Harper. McClelland, D. C., Atkinson, J. W., Clark, R. A., & Lowell, E. L. (1953). The achievement motive. New York: Appleton-Century-Crofts. Moms, C. W. (1956). Varieties of human value. Chicago: University of Chicago Press. Munro, D. (1985). A free-format values inventory: Explorations with Zimbabwean student teachers. South African Journal of Psychology, 15, 33-41. Ng, S. H., Akhtar-Hossain, A. B. M., Ball, P., Bond, M. H., Hayashi, K.,Lim, S. P., O’Driscoll, M. P., Sinha, D., & Yang, K.S. (1982). Values in nine countries. In R. Rath, H. S. Asthma, & J. B. H. Sinha (Eds.), Diversity and unity in cross-cultural psychology (pp. 196-205). Lisse, The Netherlands: Swets & Zeitlinger. Niebuhr, R. (1935). An interpretation of Christian ethics. New York: Harper. Parsons, T. (1957). The social system. New York: Free Press. Radcliffe-Brown, A. R. (1952). Structure and funcrion in primitive society. Glencoe, IL: Free Press. Rankin, W. L., & G ~ b e J., W. (1980). A comparison of ranking and rating procedures for value system measurement. European Journal of Social Psychology, 10, 233-246. Rescher, N. (1969). Introduction to value theory. Englewood Cliffs, NJ: F’rentice-Hall. Rokeach, M. (1973). The nature of human values. New York: Free Press. Schutz, W. C. (1958). FIRO: A three-dimensional theory of interpersonal behavior. New York: Rinehart. Schwartz, S. H. (1990a). Individualism-collectivism: Critique and proposed refinements. Journal of Cross-Cultural Psychology, 21, 139- 157. Schwartz, S. H. (1990b. July). Thoughts in response to cross-cultural applicarions and critiques. Paper presented at the 10th International Association of Cross-Cultural Psychology Congress, Nara, Japan. Schwartz, S. H. (in press). Cultural dimensions of values: Toward an understanding of national differences. In U. Kim, H. C. Triandis, & G. Yoon (Eds.), Individualism and collectivism. London: Sage. Schwartz, S. H., & Bilsky, W. (1987). Toward a psychological structure of human values. Journal of Personality and Social Psychology, 53, 550-562. Schwartz, S. H., & Bilsky, W. (1990). Toward a theory of the universal content and structure of values: Extensions and cross-cultural replications. Journal of Personality and Social Psychology, 58, 878-891. Schwartz, S. H., & Huismans, S. (1991). Religiosity and value priorities: A study of Protestants, Catholics and Jews. Israel: Hebrew University of Jerusalem. Submitted for publication. Schwartz, S. H., Sagiv, L., & Antonovsky, A. (1991). The stability and social desirability of value priorities. Israel: Hebrew University of Jerusalem. Submitted for publication. Scott, W. A. (1965). Values and organizations. Chicago: Rand McNally. Segall, M. H., Dasen, P. R., Berry, J. W., & Poortinga. Y. H. (1990). H u m n behavior in global perspective: An inrroducrion to cross-cultural psychology. New York: Pergamon.

UNIVERSALS IN VALUE CONTENT AND STRUCTURE

65

Shye, S. (1985). Nonmetric multivariate models for behavioral action systems. In D. Canter (Ed.), The facer approach to social research (pp. 97- 148). New York: Springer-Verlag. Shye, S. (1988). Inductive and deductive reasoning: A structural analysis of ability tests. Journal of Applied Psychology, 73, 308-3 1 I . Sumner, W. G. (1906). Folkways. Boston: Ginn. Tillich, P. (1956). Systematic theology. Chicago: University of Chicago Press. Triandis, H. C. (1990). Cross-cultural studies of individualism and collectivism. In J. Berman (Ed.), Nebraska Symposium on Motivation, 1989 (pp. 41-133). Lincoln: University of Nebraska Press. Triandis, H. C., McCusker, C., & Hui, C. H. (1990). Multimethod probes of individualism and collectivism. Journal of Personality and Social Psychology, 59, 1006- 1020. Weishut, D. J. N. (1989). The meaningfulness of the distinction berween instrument and terminul values. Unpublished master’s thesis, Hebrew University of Jerusalem. Israel. White, R. W. (1959). Motivation reconsidered: The concept of competence. Psychological Review, 66, 297-333. Williams, R. M., Jr. (1968). Values. In E. Sills (Ed.),International encyclopedia of the social sciences. New York: Macmillan. Yinger, J. M. (1957). Religion, sociery and the individual. New York: Macmillan. Zeidner, M., & Feitelson, D. (1989). Probing the validity of intelligence tests for preschool children: A smallest space analysis. Journal of Psychoeducational Assessment, 7, 175- 193. Zhi-gang, W. (1 990, August). Value structures in China: Within- and cross-cultural comparisons. Paper presented at the 10th International Association of Cross-Cultural Psychology Congress, Nara, Japan.

This Page Intentionally Left Blank

MOTIVATIONAL FOUNDATIONS OF BEHAVIORAL CONFIRMATION

Mark Snyder

I. Introduction When human beings interact with others, they often use their preconceived beliefs and expectations about them as guides to action. Their actions, in turn, may prompt their interaction partners to behave in ways that confirm these initial beliefs. The notion that beliefs and expectations can and do create their own reality is the essence of the self-fulfillingprophecy. As defined by Merton, “The self-fulfilling prophecy is, in the beginning, a false definition of the situation evoking a new behavior which makes the originally false conception come true. The specious validity of the self-fulfilling prophecy perpetuates a reign of error. For the prophet will cite the actual course of events as proof that he was right from the very beginning. Such are the perversities of social logic” (1948, p. 195).

11. The Phenomenon of Behavioral Confirmation

in Social Interaction Behavioral scientists have worked long and hard to document the self-fulfilling nature of beliefs and expectations. Of particular concern to personality and social psychologists are the ways in which preconceptions influence the dynamics and outcomes of social interaction and interpersonal relationships. A series of empirical demonstrations has revealed that a person (typically referred to as the perceiver), having adopted initially erroneous beliefs about another person (referred to as the larger), will act on these beliefs in ways that cause the behavior of the target to confirm the perceiver’s beliefs. For reviews of the literature on ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLIXiY. VOL. 25

67 Copyright 0 1992 by Academic Press, Inc. All rights of repmduction in any form reserved.

68

MARK SNYDER

interactions between people in the roles of perceiver and target, see Darley and Fazio (1980), Harris and Rosenthal (1985), E. E. Jones (1986), R. A. Jones (1977), Jussim (1986), D. T. Miller and h n b u l l (1986), and Snyder (1984). In one investigation of the effects of perceivers’ beliefs on targets’ behavior during the course of social interaction, Snyder, Tanke, and Berscheid (1977) examined the effects on interactions between pairs of college-aged men and women of the commonly held, but stereotyped, assumption that physically attractive people have more socially appealing personalities than the physically unattractive. Before participating in a “getting-acquainted” telephone conversation with a female partner, each man was randomly assigned a snapshot (ostensibly of his partner) that portrayed a physically attractive or a physically unattractive woman. Men who believed their partners to be attractive treated them with more warmth and friendliness than did men who believed their partners to be unattractive. As a result, women thought to be attractive (regardless of their actual looks) reciprocated these overtures and actually came to behave in a friendly and sociable manner; in contrast, women assumed by their partners to be physically unattractive became cool and aloof during the conversation (see also Andersen & Bem, 1981; Thomas & Malone, 1979). The phenomenon of belief creating reality is known by several names in the literature: the selfifu@llingprophecy (e.g., Rosenthal, 1974), expectancy confirmation (e.g., Darley & Fazio, 1980), and behavioral conjirmation (my own preferred terminology, because it emphasizes that it is the target’s actual behavior that confirms the perceiver’s beliefs; e.g., Snyder, 1984). Behavioral confirmation scenarios have been demonstrated in a wide variety of laboratory and nonlaboratory contexts and for a wide range of beliefs and expectations. There have been demonstrations of the behavioral confirmation of stereotypes about the typical personalities of women and men (Skrypnek & Snyder, 1982; von Baeyer, Sherk, & Zanna, 1981; Zanna & Pack, 1975), assumptions about racial differences (Word, Zanna, & Cooper, 1974), beliefs about age differences (Musser & Graziano, 1983), self-images (e.g., Swann & Hill, 1982; Swann & Read, 1981), anticipations of the likely personalities of other people (Fazio, Effrein, & Falender, 1981; Herr, 1986; Ickes, Patterson, Rajecki, & Tanford, 1982; Kelley & Stahelski, 1970; D. T. Miller & Holmes, 1975; Riggs, Monach, Ogburn, & Pahides, 1983; Snyder & Swann, 1978a), expectations of being liked or disliked (Curtis & Miller, 1986; S. C. Jones & Panitch, 1971; Rabiner & Coie, 1989), hypotheses about the personalities of other people (Fong & Markus, 1982; Snyder & Swann, 1978b; Wilson Dallas & Baron, 1985), imputations of stigmatizing conditions to other people (Harris, Milich, Johnston, & Hoover, 1990; Sibicky & Dovidio, 1986), mothers’ stereotyped expectations about infants (Stem & Hildebrandt, 1986), and arbitrary designations of differences in ability and performance competence (Eden & Shani, 1982; King, 1971; C. T. Miller, Malcarne, Clarke, Loboto, & Fitzgerald, 1986; Rosenthal8z Fode, 1963; Rosenthal & Jacobson, 1968; Zanna, Sheras, Cooper, & Shaw, 1975).

BEHAVIORAL CONFIRMATION

69

It would seem, then, that interpersonal beliefs are actively involved in guiding and directing the course and outcome of social interaction, with the effect that they can and do have self-fulfilling consequences. Perceivers operate as if their beliefs were true, and targets come to behave as if these beliefs were in fact true. The subjective reality of perceivers’ beliefs comes to be corroborated by the objective reality of targets’ actions. As such, the behavioral confirmation sequence represents a clear and compelling instance of the influential role that individuals play in shaping and constructing the social worlds in which they operate (for elaboration of the implications of such influences for understanding the interplay of individuals and their social worlds and for conceptualizing the reciprocal linkages between personality and social behavior, see Snyder, 1987).

111. The Mechanisms of Behavioral Confirmation

As even this brief sketch of the literature reveals, the number of experimental demonstrations of behavioral confirmation has grown “sufficiently numerous that the existence of the phenomenon need not be questioned” (E. E. Jones, 1986, p. 43). Increasingly, therefore, investigators have turned to questions of the mechanisms that generate behavioral confirmation. Thus, analyses and discussions of mechanisms have tended to focus on articulating steps in behavioral confirmation sequences (e.g., Brophy & Good, 1974; Cooper, 1979; Darley & Fazio, 1980; Deaux & Major, 1987; E. E. Jones, 1986; Jussim, 1986; D. T. Miller & Turnbull, 1986; Rosenthal, 1974; Snyder, 1984). Integrating and synthesizing these several analyses yield these steps: ( I ) the perceiver adopts beliefs about the target, (2) the perceiver acts as if these beliefs were true and treats the target accordingly, (3) the target assimilates his or her behavior to the perceiver’s overtures, and (4)the perceiver interprets the target’s behavior as confirmation of his or her original beliefs. Such stepwise analyses of the constituents of behavioral confirmation are of considerable utility. At the very least, they provide a useful heuristic for organizing and systematizing the research literature. Moreover, by segmenting behavioral Confirmation sequences into components that are themselves the familiar building blocks of everyday social interaction (i.e., the thoughts and behaviors of perceiver and target), we can more easily comprehend the frequent Occurrence of behavioral confirmation. As one theorist has observed, “while self-fulfilling prophecies are by no means inevitable, there are a number of reasons why they might be widely expected in a variety of settings with different kinds of behavioral expectancies” (E. E. Jones, 1986, p. 234; cf. Jussim, 1991). Nevertheless, as useful as analyses of the sequential steps in behavioral confirmation may be, their limitations should be recognized. Most important, these essentially descriptive analyses are not accounts of the psychological processes

70

MARK SNYDER

that underlie and motivate the events that produce behavioral confirmation. They may specify the how, but not the why. Why, we should ask, do perceivers do the things they do that initiate behavioral confirmation sequences? And why, we should also ask, do targets do the things they do that confirm preconceptions? These questions define the agenda for considering the psychological and rnotivational foundations of behavioral Confirmation in social interaction. In the context of the well-documented effects of beliefs on social interaction, answers to these questions become particularly important, if not critically essential, for understanding the nature of behavioral confirmation. Matters of motivation, I should acknowledge, have not gone totally unrecognized. The literature on self-fulfillingprophecies in the classroom includes some models of the roles of teachers’ and students’ motivations (e.g., Cooper & Good, 1983; Jussim, 1986). Experimental studies of behavioral confirmation increasingly are attending to some of the goals that perceivers and targets bring to their interactions (e.g., Darley, Fleming, Hilton, & Swann, 1988; Neuberg, 1989). And, among the newer looks at person perception are theoretical perspectives that accord central status to interaction goals and related motivational constructs (e.g., Hilton & Darley, 1991; E. E. Jones, 1990). Nevertheless, relatively little is known about the psychological and motivational underpinnings of behavioral confirmation. The present theoretical analysis is addressed to precisely these issues. I hope that such theorizing will be revealing not only about behavioral confirmation, but also (and especially when considered in the larger context of related theorizing on personality, motivation, and social behavior) about the intertwining of cognitive, motivational, and behavioral processes in social interaction and interpersonal relationships.

IV. The Strategy of Functional Analysis One approach to considering systematically the motivational foundations of behavioral confirmation is to perform afunctional analysis. In general terms, a functional analysis is concerned with the reasons and purposes, needs and goals, and plans and motives that underlie and generate psychological phenomena (see Snyder, 1988). In the specific context of behavioral confirmation, a functional analysis is concerned with the needs being met and the motives being served in behavioral confirmation sequences, and with the ways in which the activities of perceivers and targets contribute to the execution of plans and the fulfilling of purposes, both personal and social ones. That is, in functional terms, the question of motivation is framed as “What are the psychological functions being served by those activities of perceiver and target that generate behavioral confirmation?” Functional analyses have a long and distinguished history in psychology. Ry-

BEHAVIORAL CONFIRMATION

71

chology as we know it today owes much to a turn-of-the-century shift away from structural attempts (such as those of Wundt and Tichener) to describe and classify mental events. In contrast, the functionalists (e.g., Angell, 1907; Dewey, 1896; James, 1890) argued that psychology should concern itself with mental and behavioral functions, and, in so doing, understand the adaptation of organisms to their environments. If structuralism was the descriptive psychology of “is,” then functionalism was the purposive psychology of “is for” (Tichener, 1899). From its beginnings in concerns with the functions associated with the gratifications of basic physiological needs, functional psychology spread rapidly in American psychology, leading to the emergence and development of many areas of basic and applied psychology (for reviews, see Boring, 1950; Hilgard, 1987). In spite of its historical importance, functionalism is seldom discussed explicitly in contemporary psychology, even though its purposive and adaptive themes are widely shared (see Buxton, 1985). Within personality and social psychology, the themes of functionalism are most evident in longstanding concerns with the active and purposeful strivings of human beings toward personal and social ends (see Pervin, 1989). Consider, specifically, these aspects of functional thinking in personality and social psychology that set the stage for the present theoretical analysis of behavioral confirmation. In his classic writings on the nature of personality, G. W. Allport emphasized that “personality is something and does something” (1937, p. 48). For Allport, what personality does had a “functional . . . significance” (p. 49), for he believed that the purpose of personality was to determine each person’s unique adjustments to the environment. This concern with the purposes of personality is further elaborated in F. H. Allport’s “teleonomic” theory, in which he argued for the importance of viewing personality in terms of what a person is “trying to do” (1937, p. 204). He offered this example: “Let us imagine three individuals rated or measured as equally honest. One of them might be seeking justice, another might be trying to help others, while the third might be trying to maintain his self-esteem or reputation” (p. 204). Described this way, in reference to their teleonomic trends, the same traits flow from distinctly different purposes that the three people are trying to carry out. Stated in functional terms, Allport’s example and theorizing instruct us that the same events may, for different people, serve quite different psychological functions. In social psychology, the same theme emerges in the functional theories of attitudes and persuasion (e.g., Katz, 1960; Smith, Bruner, & White, 1956). To answer the question “Of what use to people are their attitudes?,” the functional theorists argued that attitudes help people meet needs, execute plans, and achieve ends. Moreover, they offered taxonomies of distinct functions served by attitudes. Functional theorists further proposed that people may hold the same attitudes for different motivational reasons and to serve different psychological functions. As a result, they claimed, attempts to change an attitude would suc-

72

MARK SNYDER

ceed to the extent that they considered the function being served by that attitude. And, although researchers have faced difficulties in testing functional hypotheses, new strategies for identifying and studying attitude functions are demonstrating the usefulness of the functional approach (e.g., Herek, 1987; Shavitt, 1989; Snyder & DeBono, 1989). The functional approach to attitudes is but one example of the general utility and the broad applicability of the strategy of functional analysis for addressing problems of social motivation (on this point, see also Snyder, 1988). In this regard, several features of the functional theories of attitudes illustrate the general appeal of the functional approach in diverse domains and are specifically instructive about the possibility of a functional interpretation of behavioral confirmation and those aspects of social perception and interpersonal behavior that come together in behavioral confirmation scenarios. First, the functional attitude theorists’ concern with the motivational infrastructure of attitudes translates readily into a concern with the motivational foundations of a broad range of phenomena of individual and social functioning. For our present purposes, the logic of functional analysis implies that, just as it is possible to propose and investigate specific functions of attitudes, so too is it possible to identify the personal and social functions served by the events that constitute behavioral confirmation scenarios in social interaction. Second, the assertion that the same attitude can serve different functions for different people can be broadened to apply to any patterns of thoughts, feelings, and actions. Thus, in the specific case of behavioral confirmation, interaction sequences that share the same surface features (i.e.. perceivers acting on their beliefs and targets coming to behave in accord with those beliefs) may reflect differing underlying motivational processes (i.e., they may be performing differing personal and social psychological functions). Third, the claim that attempts to change attitudes will succeed if and when they address the functions of attitudes suggests that the key to much social influence is to be found in the functions that support the thoughts, feelings, or actions to be influenced. If so, then with respect to behavioral confirmation, interventions designed to influence the functions that are served by the activities of perceivers and targets will determine the outcomes of their interactions. It thereby may be possible to specify not only when and why behavioral confirmation occurs but also when and why other theoretically and practically meaningful outcomes (especially behavioral disconformation effects, in which the perceiver’s actions lead the target to behave contrary to the perceiver’s anticipations) will occur. The foregoing discussion attests, in the general case, to the explanatory utility of inquiries about personal and social motivational functions. Functionalist thinking, it should be emphasized, is not something limited to any one domain of psychological inquiry; rather, its theoretical and practical benefits accrue in many and varied domains of personal and social functioning. In the theoretical analysis

BEHAVIORAL CONFIRMATION

73

that follows, the focus is on the motivational foundations of behavioral confirmation. In this analysis, a set of motivational functions is derived from considerations of the structure and dynamics of ongoing social interaction. Specifically, the analysis first considers the motivational properties of the circumstances in which behavioral confirmation typically occurs, and offers propositions about the functions served by the activities of perceivers and targets when they interact in such circumstances. Next, the analysis considers features that are typically absent in such interactions and, based on these considerations, offers propositions about motivational functions that, if engaged by perceivers and targets, would prevent behavioral confirmation from occurring or even generate behavioral disconfirmation. In this analysis, 1 will devote particular attention to a set of four motivational functions: the acquisition of social knowledge, the facilitation of social interaction, the expression of personal attributes, and the defense of threatened identities. For each of these proposed functions, I will draw from the empirical literature on the effects of beliefs on social interaction to illustrate the utility of this functional analysis for understanding why behavioral confirmation and behavioral disconfirmation occur. In addition, I will examine parallels between the set of functions derived in this analysis of behavioral confirmation and those emerging from motivational analyses in other domains (with particular reference to theories of attitude, needs, and goals). Finally, I will discuss the potential for extending the domain of this analysis from the specific case of behavioral confirmation to the general domain of social perception and interpersonal behavior, as well as for applying the logic of the strategy of functional analysis more broadly in diverse arenas of individual and social functioning.

V. The Functions Engaged in Behavioral Confirmation Scenarios

What, then, are the motivational foundations of behavioral confirmation? In investigations of behavioral confirmation, as in social life itself, this phenomenon often occurs when people meet each other and get acquainted through social interaction. In fact, participants in such investigations are often told that they are taking part in studies of “acquaintance processes in social relationships” (e.g., Snyder et al., 1977, p. 659). That the interaction between perceiver and target occurs in a getting-acquainted situation may be a significant source of clues as we search for the functions served by the activities of perceivers and targets in behavioral confirmation sequences. In focusing on getting-acquainted situations and the particular motivational functions engaged in the social interactions that occur in such situations, the present analysis will not examine interactions in

74

MARK SNYDER

other contexts, such as those between teachers and students in classrooms or those between researchers and subjects in laboratories. The self-fulfilling consequences of these forms of interactions have been the focus of theorizing in their own right (e.g., Cooper & Good, 1983; Jussim, 1986, 1990; Rosenthal, 1974), and their motivational foundations are certainly deserving of their own explicit theoretical and empirical scrutiny. At least two sets of motivations, each relevant to a functional analysis, may be engaged in getting-acquainted situations. First, when people get acquainted, they may seek to gel f o know each other, that is, to become familiar with each other’s beliefs, feelings, and behaviors. But, in such circumstances, they may also seek to get along with each other, that is, to discover ways in which to have smoothly flowing and pleasing interactions. Often, efforts to get to know and to get along are intimately intertwined in the acquaintance process; in fact, the terms “get to know” and “get along’’ are often used in everyday discourse as interchangeable synonyms for the activities of getting acquainted. However, getting to know and getting along are (in principle, at least) separate aspects of social interaction. At times, the two purposes may even be at odds with each other. Thus, for example, probing about another person’s deepest and darkest secrets may yield quite a bit of information, but at the cost of a tense and anxious interaction. By the same token, light and pleasant chit-chat, of the variety frequently heard at cocktail parties, may be most conducive to smoothly flowing interactions, although ones so superficial that they hardly yield any sense of understanding of one’s interaction partners. These considerations suggest that the activities of getting acquainted may be undertaken in the pursuit of quite different motivations. The acquaintance process may be motivated by desires to get to know the other person, or it may be guided by considerations of getting along with one’s interaction partner. Let us, therefore, examine the motivations that may be engaged in getting-acquainted interactions from a more explicitly functional perspective. The “getting to know each other” aspects of getting-acquainted situations suggest that the functions associated with the acquisition and use of social knowledge may be relevant to understanding the contributions of perceivers and targets to behavioral confirmation outcomes. The “getting along with each other” aspects of the acquaintance process suggest that the situations in which behavioral confirmation occurs may also engage functions associated with the regulation and facilitation of social interaction.

A. ACQUIRING AND USING SOCIAL KNOWLEDGE

How might motivations associated with social knowledge be involved in behavioral confirmation? With the widespread impact of the theories of Heider

BEHAVIORAL CONFIRMATION

75

(1958), E. E. Jones and Davis (1965), and Kelley (1967), it has become practically a truism of social psychology that people want to understand the motives, intentions, and dispositions of those with whom they interact. Kelley, for one, has commented that such theories describe “processes that operate as if the individual were motivated to attain a cognitive mastery of the causal structure of [the] environment” (1967, p. 193). Such knowledge, it is widely believed, brings a sense of stability and predictability (and perhaps even controllability) to people’s dealings with others. From this perspective, perceivers may act on their preconceived beliefs about other people precisely because of the sense of stability and predictability that doing so provides. When, as a consequence, perceivers induce targets to behave in accord with these beliefs, the sense of stability and predictability associated with their preconceived beliefs is reinforced by the partner’s behavior. When, for example, in division-of-labor negotiations, a perceiver with traditional beliefs about sex roles maneuvers a supposedly female target into “choosing” stereotypically feminine tasks (as happened in an experiment reported by Skrypnek and Snyder, 1982), the perceiver’s beliefs about a stable and predictable world in which men act like men and women act like women are reinforced by this behavioral confirmation outcome. The activities of targets may also be interpreted in light of motivations associated with social knowledge. The target’s reciprocation of the perceiver’s overtures may reflect the target’s inferences about the perceiver’s personality. Having no reason to suspect otherwise (typically targets are not aware that perceivers harbor preconceptions about them), the target may take the perceiver’s behavior at face value and, in line with the well-documented “correspondence bias” phenomenon (e.g., E. E. Jones, 1990) infer that the perceiver’s behavior reflects the perceiver’s own personality. In behavioral confirmation scenarios, the target is unlikely to discover any indications to the contrary; after all, the perceiver is likely to maintain his or her interaction strategy because it has been validated by the target’s confirmatory behavior. Thus, the target may gain a sense of a stable and predictable dispositional understanding of the perceiver. For example, in the aforementioned division-of-labor negotiations, the target who provides behavioral confirmation will undoubtedly be treated to further sex-typed treatment, thus reinforcing her earlier inference that these overtures reflect corresponding stable attributes of the perceiver. Thus, according to this line of reasoning based on the “getting to know” aspects of the acquaintance process, behavioral confirmation may function to promote a sense of a stable and a predictable social world in which people seem to act as they are expected to act. Whether behavioral confirmation does serve this knowledge function, of course, remains to be seen, and will be seen after we examine an alternative functional perspective on the interactions of perceivers and targets.

76

MARK SNYDER

B. REGULATING AND FACILITATING SOCIAL INTERACTION The “getting along” aspects of the acquaintance process are the basis for a quite different interpretation of the functions served by behavioral confirmation, an interpretation in terms of motivations associated with the regulation and facilitation of social interaction. A recurring theme in the social and behavioral sciences is the considerable extent to which the events of social interaction and interpersonal relationships reflect purposive efforts to regulate the flow of interaction, its dynamics, and its outcomes (e.g., Goffman, 1959; E. E. Jones, 1990; Schlenker, 1980; Snyder, 1987). From this perspective, when perceivers treat targets as if they are the types of people they are presumed to be, they may do so to facilitate their social interaction. For example, such motivations may underlie the actions of a perceiver who treats a supposedly sociable target to a display of warmth and friendliness; such a perceiver may be acting on the assumption that a smooth and coordinated interaction will result from such overtures. By the same token, another perceiver who adopts a more reserved posture may do so to protect a supposedly shy target from the demands of an overly animated and socially demanding interaction; that perceiver too may be operating on assumptions about how to produce a smoothly functioning interaction by being sensitive to and accommodatingof the target’s social style. That is, by enacting “scripts” for being responsive to the anticipated behavior of their partners, and whether their anticipations are based on positive or negative expectations about the target’s personality, perceivers may seek to regulate and facilitate the course and outcome of their interaction (for a related perspective on the concept of responsiveness in dyadic interaction, see Davis, 1982; Davis & Perkowitz, 1979). Similar underlying motivations and strategic self-presentational activities may account for why targets act to confirm the preconceived beliefs of perceivers. ?Lpically, the interaction context as presented to the target is rather impoverished. Thus, in many experiments in which perceivers are provided with information about the likely personalities of their interaction partners, targets learn only that they will be participating in a brief interaction (e.g., Andersen & Bem, 1981). Thus, targets may naively assume that perceivers’ behaviors reflect their typical interactional styles. If so, strategic considerations of impression management may lead targets to engage in “face work” (to use Goffrnan’s, 1955, term for it) and take their cues from their interaction partners and, as a result, reciprocate the perceivers’ overtures, and thus present themselves as being more similar to their perceivers than they really are. Such impression management activities, theory and research tell us, may occur in the service of several motivations, including being responsive to and accommodative of their partners in hopes of achieving a smooth flow of interaction, as well as securing the

BEHAVIORAL CONFIRMATION

77

acceptance or approval of the perceiver (e.g., Davis, 1982; E. E. Jones, 1964; Schlenker, 1980). Although it might seem that the dynamics of the “getting along” interpretation apply more readily to targets presenting themselves positively in interactions with perceivers who hold favorable expectations about them, it can be argued that the same dynamics are at work when targets confirm negative beliefs. Thus, in the Snyder and Swann (1978a) experiment, when targets reciprocated the hostile overtures of perceivers who believed them to be of hostile character, they may have done so on the assumption that the perceivers’ actions constituted a meaningful source of cues for how they ought to behave in this unfamiliar situation; moreover, by behaving similarly to their interaction partners, they may have been seeking common ties and bonds with those sharing that situation. Therefore, whether they are interacting with perceivers who hold positive or negative expectation about them, targets may be motivated to take their cues from their interaction partners, with the net effect that they present themselves as similar to their perceivers and thus effectively confirm the beliefs that actually guided their perceivers’ treatment of them. Thus, according to this line of functional reasoning, perceiver and target may be striving to facilitate their interaction by regulating their behavior to be responsive to and accommodative of each other’s perceived personal attributes and interpersonal manners. Whether behavioral confirmation serves this function of facilitating smoothly coordinated and mutually responsive social interaction, or whether it serves the alternative function of acquiring stable and predictable social knowledge, we shall now see-with the aid of relevant analytic considerations and empirical evidence germane to functional hypotheses about the motivational foundations of behavioral confirmation.

VI. Different Functions for Perceiver and Target? To recapitulate the argument thus far, when people interact for the purpose of getting acquainted, they may be concerned both with getting to know each other and with getting along with each other. That is, functions associated with the acquisition of social knowledge and with the regulation of social interaction may be served in the course of interaction sequences. These two functions may constitute alternative (if not actually competing) theoretical accounts of the motivational foundations of behavioral confirmation. It is always tempting, in the face of competing explanations, to pit them against each other and ask which one is the victor in a critical confrontation. Which is it, then acquiring knowledge or regulating interaction, that generates behavioral confirmation? As so often happens when questions of this form are asked, the answer is that

78

MARK SNYDER

neither one is the undisputed champion, but that each has its own domain of applicability. In the case of behavioral confirmation, the fact that social interaction potentially serves multiple functions suggests the intriguing possibility that the activities of one party to the interaction may be supported by motivational foundations different from those of his or her interaction partner. More specifically, there are at least three reasons to argue that the activities of the perceiver serve the function of acquiring social knowledge and the activities of the target serve the function of facilitating social interaction. A. ANALYTIC CONSIDERATIONS The first reason is analytic in nature. In conceptualizing behavioral confirmation, the role of the perceiver is defined in relation to the guiding influence of his or her beliefs about the target. After all, it is the perceiver and not the target who holds preconceived beliefs, stereotyped expectations, or tentative hypotheses about his or her interaction partner. And it is the perceiver who often has more explicit opportunities than the target to use their interaction as an opportunity to evaluate those beliefs. Thus, for example, in interview formats for studying behavioral confirmation, the perceiver asks the questions and the target answers them (e.g., Fazio et al., 1981). For these reasons, the role of the perceiver seems particularly congruent with the function of acquiring and using social knowledge. By contrast, the role of the target is defined in relation to behavioral reactions to the perceiver’s overtures. Indeed, these overtures may provide ready guidelines for targets who find themselves pressed to interact with unfamiliar partners in unfamiliar situations. Thus, in the absence of competing considerations, the role of the target seems to be defined and constrained so as virtually to mandate taking cues from others and fitting one’s presentations of self into the outlines laid down by one’s interaction partner. That is, analytically at least, the target role seems to be defined in ways that make it particularly congruent with the function of regulating and facilitating social interaction. In effect, these analytic considerations point to power differences inherent in the roles of perceiver and target. Structural differences in the information available to perceivers and targets combined with differing potential to guide and direct the interaction combine to make the role of perceiver one of relatively greater power and that of target one of relatively lesser power. In addition to these differences in power inherent in the roles of perceiver and targets, perceivers and targets often interact in circumstances that are marked by differential power (e.g., teachers and students, employers and employees, therapists and clients). Of what consequence are these differences in power? Persons in positions of relatively greater power may be disposed to act on their power to influence and control the dynamics and outcomes of their dealings with those of

BEHAVIORAL CONFIRMATION

79

lesser power. Moreover, the exercise of power may be facilitated by images of one’s interaction partners as stable and predictable, and hence controllable creatures. In this sense, “knowledge is power” and this line of argument would suggest that the relatively powerful perceiver in the behavioral confirmation sequence ought to be motivated by the function of acquiring and using stable and predictable knowledge about the target. Now, persons in positions of lesser power often will act to forestall undesirable outcomes, to ensure favorable outcomes, and to otherwise lessen their dependence on those who hold power over them. At least two routes to such outcomes, each relevant to confirmatory outcomes, suggest themselves. First, the tendency of persons in positions of lesser power to “ambiguate” their behavior for self-protective reasons (e.g., Brown & Levinson, 1987) may make the behavior of powerless targets especially susceptible to confirmatory interpretive cognitive processes on the part of powerful perceivers who harbor preconceptions about them. Second, a behavioral route to lesser dependence and increased power may be available through the strategic use by targets of the tactics of impression management, ingratiation, and other means of ensuring a smooth and satisfying flow of interaction with those of greater power. This line of reasoning leads to the suggestion that the relatively powerless target in the behavioral confirmation sequence should operate in the service of the accommodative function of regulating and facilitating interaction with the perceiver. For a more detailed analysis of the motivational implications of social power for behavioral confirmation, see Copeland (1991).

B. CORRELATIONAL EVIDENCE The second reason is provided by empirical studies of the personality correlates of behavioral confirmation (e.g., Babad, 1979; Babad & Inbar, 1981; Babad, Inbar, & Rosenthal, 1982a, 1982b; Bootzin, 1971; Cooper & Hazelrigg, 1988; Harris, 1989; Harris & Rosenthal, 1986). If people with particular personalities are especially likely to contribute to behavioral confirmation, and if these personalities are in turn linked to specific psychological and motivational functions, then inferences can be made about the functions served by behavioral confirmation. Are there, then, personality correlates of behavioral Confirmation? Using an extensive battery of personality measures, Harris and Rosenthal (1986) attempted to predict indices of the magnitude of behavioral confirmation in interactions between perceivers and targets. They noted associations between the extent to which perceivers elicited behavioral confirmation and high scores on measures of dogmatism and order (attributes that may be typical of people motivated to seek the cognitive stability and predictability associated with the function of acquiring social knowledge). These findings have since been replicated by Harris (1989). In addition, other researchers have noted correlations

80

MARK SNYDER

between the extent to which perceivers view their interactions as opportunities to gain knowledge about the target (e.g., knowledge that would inform them about the target’s personality and that would allow them to make predictions about the target’s behavior) and the extent to which they behave in ways likely to elicit behavioral confirmation (Snyder, Manning, & Ridge, 1990). Finally, there is evidence that, in some interactions between perceivers and targets, perceivers construe their task more in terms of acquiring knowledge about their targets than in terms of facilitating their interaction (Copeland & Snyder, 1989). For targets, by contrast, the available correlational evidence suggests that their activities serve a distinctly different function. Hams and Rosenthal (1986) and Grinder and Swim (1991) have reported associations between the extent to which targets provided behavioral confirmation and high scores on self-monitoring (Snyder, 1987) and other measures of responsiveness to social and interpersonal indicators of appropriate behavior (but, see Hams, 1989, for an exception to this pattern of findings). These personal attributes are ones that ought to be typical of people whose actions serve the function of regulating social interaction; in the specific case of self-monitoring, considerable evidence suggests the habitual involvement of this function in their activities (e.g., Snyder & DeBono, 1987). In addition, some associations have been noted between targets’ responsiveness to perceivers’ expectations and targets’ perceptions of perceivers’ status, associations that may be indicative of targets’ attempts to regulate their interactions in ways designed to facilitate favorable interaction outcomes with those they perceive to be of greater status (Virdin & Neuberg, 1990).

C. EXPERIMENTAL EVIDENCE The third source of evidence comes from experimental studies that employ interventions directed at the purported functional underpinnings of behavioral confirmation. If the activities of perceiver or target serve a purported function, then it should be possible to influence the likelihood of behavioral confirmation by experimentally influencing the likelihood that the function will be engaged. In one series of studies, Snyder and Haugen (1990, 1991) created a basic, minimally defined interaction situation in which behavioral confirmation would not occur and then added features to it designed to engage specific motivational functions. In so doing, they sought to identify those motivational features, the addition of which would result in behavioral confirmation. Critical to this research endeavor was a task unfamiliar to most researchers concerned with behavioral confirmation, namely creating a basic situation in which behavioral confirmation would nor occur in interactions between perceivers and targets. To do so, Snyder and Haugen brought perceivers and targets together in circumstances containing only those elements absolutely and minimally necessary to produce an interaction. This minimally defined situation was

BEHAVIORAL CONFIRMATION

81

designed so that it provided perceivers and targets with no explicitly defined objectives, purposes, or other motivational bases for their interaction. That is, the interaction situation provided nothing more than an opportunity for two people, one of whom held a preconceived belief about the other, to engage in conversation. In their studies, Snyder and Haugen had pairs of college students, a male assigned to the role of perceiver and a female assigned to the role of target, engage in a telephone conversation. Before these conversations, male perceivers were given a Polaroid picture (ostensibly of the female target) that depicted either a woman who was decidedly obese or one of normal weight. Among college students, at least, stereotypes about the linkages between weight and personality hold that the obese are unfriendly, reserved, lazy, boring, sexually cold, and sad. When perceivers and targets conversed in this basic situation, which provided no explicit motivational foundation for their interaction, behavioral confirmation did nor occur. Targets believed to be obese and targets believed to be of normal weight did not behave any differently. Snyder and Haugen then systematically modified this basic situation to influence the functions served by the activities of perceivers and targets. For perceivers, when the function of acquiring knowledge was engaged, behavioral confirmation occurred (Snyder & Haugen, 1990). Some perceivers were induced (by means of an experimental manipulation) to define the function of the acquaintance process as one of using one’s interactions to get to know one’s partners and to get a stable and predictable “picture” of them by finding out the ways that other people are and are not like what they are expected to be. As a result of interacting with such perceivers, women believed to be obese responded in more unfriendly, reserved, and unenthusiastic ways than women believed to be of normal weight. By contrast, engaging the function of regulating social interaction in perceivers did not produce behavioral confirmation. Some perceivers were encouraged (again, as a result of an experimental manipulation) to construe the function of getting acquainted as using first impressions to find ways to get along well with their partners and to facilitate their interactions by being responsive to their partners’ expected personal dispositions and anticipated interpersonal styles. These getting-acquainted conversations did not evoke systematically different behavioral patterns from supposedly “fat” and “normal” targets. The particular involvement of knowledge-oriented functions in the activities of perceivers is also evident in a study of peer counseling processes conducted by Copeland and Snyder (1990). In this study, pairs of college students interacted in a simulation of peer counseling in which one student (the peer counselor) held preconceived expectations about the personality of the other student (the client). These interactions, with counselors serving as perceivers and clients as targets, occurred under circumstances designed to influence functions served by the activities of the peer counselors. Some counselors were instructed merely to

82

M A R K SNYDER

converse with their clients. For other counselors, the peer counseling task was defined primarily in terms of “diagnosis,” or getting to know their clients. And, for still other counselors, the peer counseling task was defined primarily in terms of “rapport,” or getting along well with their clients. Only when peer counselors were motivated by the knowledge-oriented function of getting to know their clients did behavioral confirmation occur; here, during the course of the peer counseling interaction, clients came to behave in accord with the expectations of their peer counselors. The consequences of systematically engaging motivational functions are considerably different for targets. Here, Snyder and Haugen (1991) have found indications that behavioral confirmation occurs when the function of regulating and facilitating social interaction is activated. When targets (who, unknown to them were thought to be obese or normal weight) believed that the acquaintance process was best served by being responsive to their partners and maintaining a smooth conversation, one that flowed without awkward points, they were most likely to provide behavioral confirmation. Thus, targets so motivated presented themselves as relatively more cold, unenthusiastic, and unresponsive when interacting with perceivers who regarded them as obese, and as relatively more flirtatious, exciting, and energetic when conversing with men who believed them to be of normal weight. By contrast, when the function of acquiring social knowledge was engaged for targets (that is, when they believed that getting acquainted was best served by using their interaction to find out what the partner was like, what his personality traits were, and how that type of person can be expected to behave), behavioral confirmation tended not to occur. These targets did not present themselves differently when interacting with perceivers who regarded them as obese or of normal weight. It is instructive, in the context of these findings about the motivations of perceivers and targets, to reexamine previous demonstrations of behavioral confirmation. With the wisdom of hindsight, it is possible to construe the instructions to perceivers in previous studies in terms of knowledge-oriented functions. For example, in Snyder and Swann’s (1978b) study concerning confirmatory hypothesis testing, the experimenter told participants that “they were in an investigation of how people come to understand each other” and that they would “attempt to find out about another person . . . by asking questions” (p. 1203). It may be these knowledge-oriented instructions that motivated perceivers to ask hypothesis-confirming questions that later led to actual behavioral confirmation in the course of subsequent interactions between perceivers and targets. Moreover, and also with the wisdom of hindsight, the fact that targets are often placed in positions of relatively lesser power than perceivers (for example, as interviewees or clients in simulations of employment and counseling interactions) may motivate targets to be particularly responsive to the overtures of their interaction partners. Taken together, the results of this series of experimental investigations provide

BEHAVIORAL CONFIRMATION

83

further converging support for the proposition that, in getting-acquainted situations, behavioral confirmation is the product of an interaction between a perceiver operating in the service of the function of acquiring social knowledge and a target operating in the service of the function of regulating social interaction. Thus, although perceivers and targets may share the common goal of getting acquainted, their manner of pursuing this goal may serve quite different functions. Indeed, it is precisely the fact that what appears to be the very same end goal of getting acquainted can be pursued in the service of quite differing underlying motivational functions that is a defining feature of the functional approach to understanding the links between social perception and interpersonal behavior. Further, it is important to note that some of the “news” in these experiments is to be found in what did not happen. Specifically, the fact that Snyder and Haugen succeeded in creating the minimally defined basic situation in which behavioral confirmation did not occur suggests that behavioral confirmation is by no means the inevitable consequence of interactions between perceivers and targets who meet and interact with each other (see also Jussim, 1991). To the contrary, it was only when specific motivationally relevant functions were explicitly engaged that behavioral confirmation did occur. To be sure, the fact is that the discovery of such motivational boundary conditions may have the effect of “demystifying” the phenomenon of behavioral confirmation, making clear that it is not something that magically occurs when perceivers interact with the targets of their expectations. Nevertheless, such demystification may be a small price to pay for the increment to the theoretical state of affairs in this domain of psychological inquiry. And, in this regard, I should point out that considerations of motivational functions reveal some intriguing psychological ironies associated with behavioral confirmation scenarios. For perceivers, the sense that the world is stable and predictable may be provided by targets who are behaving in ways that run counter to their actual personal dispositions. And, for targets, the quest for smoothly coordinated interactions may facilitate current interactions at the same time as it jeopardizes future ones, especially in the case of targets who, by being responsive to their interaction partners, confirm negative beliefs and expectations. These ironies are intriguing ones, and their implications will be pursued in due course. But first, let us turn to the challenging matter of behavioral disconfirmation, as seen from the functional perspective.

VII. The Matter of Behavioral Disconfirmation This theoretical analysis may have implications not only for when behavioral confirmation will occur but also for when it will not occur. One important criterion for evaluating explanations of behavioral confirmation is their ability to specify when and why interaction culminates in the target behaving contrary to

84

MARK SNYDER

the perceiver’s anticipations (i.e., a behavioral disconfirmation outcome). In this regard, there have been several indications of points in the interaction sequence at which actions by perceivers and targets might produce behavioral disconfirmation (e.g., Andersen & Bern, 1981; Bond, 1972; Darley et al., 1988; Hilton & Darley, 1985; Ickes et al., 1982; D. T. Miller & ”bmbull, 1986; Rubovitz & Maehr, 1971; Snyder, 1984; Swann & Ely, 1984; Swann & Snyder, 1980; it should be noted that, in some of these studies, perceivers clung to their initial expectations, thus demonstrating perceptual confirmation in the face of behavioral disconfirmation). Although these demonstrations say much about when behavioral disconfirmation occurs, they say conspicuously little about why it occurs. A functional analysis, however, provides systematic specificationsof the motivational foundations of disconfirmation outcomes. Specifically, it suggests three routes to behavioral disconfirmation. Consider first what happens when perceivers and targets trade motivational roles, that is, when circumstances create perceivers motivated to facilitate social interaction or targets motivated to acquire social knowledge.

A. TRADING MOTIVATIONAL PLACES Some instances of behavioral disconfirmation may be the product of perceivers, rather than acting in the service of their characteristic function, being motivated instead to regulate and facilitate interaction. For example, a perceiver who treats a target reputed to be cool and aloof to a display of warmth and friendliness designed to compensate for that target’s expected deficits in social skills may elicit pleasantries from the target and, not incidentally, create a behavioral disconfirmation outcome. Such disconfirmation outcomes, interpretable as a result of compensatory interaction strategies on the part of perceivers, have been documented by Bond (1972), Ickes et al. (1982), Major, Cozzarelli, Testa, and McFarlin (1 988), and Swann and Snyder (1980). Moreover, and as we have already seen, even though the outcome was not outright disconfirmation, targets perceived to be obese or of normal weight did not behave differently when interacting with perceivers motivated to get along well with their interaction partners (Snyder & Haugen, 1990). In addition, the lesser likelihood of negative expectations generating behavioral confirmation (e.g., Hilton & Darley, 1985; Major et al., 1988) may reflect “politeness” considerations that sensitize perceivers to the disruptive social consequences of expressing and communicating negative beliefs (see Brown & Levinson, 1987; Deaux & Major, 1987). In a related vein, considerations that prevent targets from acting in the service of their characteristic function of facilitating interaction also seem to preclude behavioral confirmation. Recall that the essence of this functional interpretation of the target is his or her assumption that the perceiver’s behavior reflects the

BEHAVIORAL CONFIRMATION

85

perceiver’s typical interactional style and that reciprocating it will facilitate interaction. The research literature suggests that, when targets are prevented from making this critical assumption, they do not provide behavioral confirmation. Thus, when targets have reason to believe that the perceiver’s behavior reflects preconceived beliefs and expectations (and, presumably, not the person’s natural social styles), they often will behave in ways that contradict the perceiver’s expectations (Hilton & Darley, 1985). Moreover, there are indications that, when targets are encouraged to treat their interactions as opportunities to get to know their interaction partners, behavioral confirmation is attenuated (Snyder & Haugen , I99 1). As a direct test, and a particularly compelling one I think, of the consequences of perceivers and targets reversing their motivational functions, Copeland ( 1991) arranged interactions between participants, one of whom (the perceiver) held an experimentally created expectation about the personality of the other (the target), and one of whom (either the perceiver or target, determined by random assignment to experimental conditions) had the power to choose whether the other would be a participant in a subsequent reward-laden phase of the study. When perceivers had the power to influence the target’s outcomes, perceivers reported their motivations to be primarily knowledge oriented and targets reported their primary motivations to be facilitating favorable interaction outcomes; these interactions resulted in behavioral confirmation of perceivers’ expectations about the targets. However, when it was the perceivers’ subsequent outcomes that were under the control of the target, perceivers reported that their motivations were primarily to facilitate desired outcomes of their interactions and targets reported that their motivations were primarily knowledge oriented; these interactions produced no signs of behavioral confirmation. Thus, when circumstances conspire to reverse the roles-that is, when perceivers become motivated to regulate and facilitate interaction and when targets become motivated to acquire and use knowledge-behavioral Confirmation may be inhibited and actual behavioral disconfirmation may even occur. A functional interpretation therefore provides common motivational foundations for both confirmation and disconfirmation outcomes. Let us pursue the matter of disconfirmation a bit further. A functional approach generates specific prescriptions for interventions designed to short-circuit the process of behavioral confirmation. The way in which it does so is revealed by considering functions that typically are absent from behavioral confirmation scenarios.

B. EXPRESSING PERSONAL IDENTITIES In most behavioral confirmation scenarios, the expectations of concern originate either in beliefs triggered by aspects of the target (e.g., sex, appearance, race) or in reputational information conveyed by third parties (e.g., information

86

MARK SNYDER

from the experimenter about the target). Thus, for example, an interviewer might treat an applicant as if he were well suited for a job if she holds stereotypes about his race (e.g., Word et al., 1974). Or, a teacher might treat a student as if she were intelligent after receiving information from the experimenter about her aptitude (e.g., Rosenthal & Jacobson, 1968). As a result, when perceivers act on such beliefs and expectations, their behavior may not necessarily express their own personality dispositions. For instance, a dispositionally friendly perceiver may act reserved toward a target reputed to be unfriendly. By the same token, when targets take their cues from perceivers and confirm preconceived beliefs, their behaviors may not reflect their own personal dispositions. Thus, a typically sociable target may reciprocate the standoffish treatment of a perceiver who anticipates a cold target. That some behavioral confirmation sequences involve perceivers and targets acting in ways that do not express their own personalities invokes considerations of functions associated with the expression of personal attributes. Generally speaking, social interaction and interpersonal relationships provide opportunities for people who are so motivated to express a wide range of personal attributes, including features of identities, self-conceptions, and dispositions of personality. That is, social interactions have the potential to serve an identity-expressive function for their participants. As we shall see, the motivational forces of this expressive function can be harnessed to intervene and transform behavioral confirmation into behavioral disconfirmation. Consider first the perceiver, and how interventions based on the expressive function can induce the perceiver not to instigate the chain of events that culminates in behavioral confirmation. In one study (reported by Snyder, Campbell, & Preston, 1982), perceivers were sensitized to the fact that impression formation in social interaction is a two-way street. They were reminded that, just as they would be forming an impression of the other person, so also would he or she be forming an impression of what they were like. Moreover, these perceivers were also encouraged to consider which of their values they might be expressing with their interactional strategies, specifically to think about how their styles of getting to know the other person might make them appear to be closed-minded or open-minded. This intervention had considerable impact on perceivers. They did not formulate confirmatory interaction strategies, an outcome readily interpretable as the result of identity-expressive motivations. Having been sensitized to the reciprocal nature of impression formation, perceivers may have been more responsive to their own personal attributes, and correspondingly less responsive to preconceptions of their interaction partners. And, as a result of engaging the widely shared values of open-mindedness, perceivers may have expressed these values by planning interaction strategies that did not prejudge the personalities of their partners. In a related demonstration of how to inhibit behavioral confirmation, Neuberg

BEHAVIORAL CONFIRMATION

87

(1989) had perceivers interview two job applicants, about one of whom they were given negative expectations. Some perceivers learned of the extreme importance of being accurate and selecting only the most qualified applicants. They refrained from prejudicial treatment of applicants about whom they held negative expectations. Hence, they did not elicit the behavioral confirmation that occurred in interviews conducted by other perceivers who had not received the same admonition. One interpretation of the inhibitory effect of this intervention, which is in keeping with Neuberg’s, is that it sensitized perceivers to their values of fairness and equality of opportunity and to the possible prejudicial and discriminatory consequences of differential treatment of job applicants. (This interpretation, it should be noted, implies that the perceiver’s quest for accuracy, which seems to inhibit behavioral confirmation, ought not be equated with the strivings for stability and predictability, which, in the functional account of behavioral confirmation, are hypothesized to motivate the perceiver’s contributions to behavioral confirmation-an implication to which I will return in elaborating the functional analysis.) For another demonstration of the inhibitory effects of accuracy sets on perceivers’ treatment of the targets of their expectations, see Shapiro and Hilton (1988). Central to understanding the role of identity-expressive motivations in the behavior of perceivers is knowing those attributes of the perceivers whose expression can and will have an impact on the dynamics and outcomes of their interactions with targets. Interpreting previous research in terms of functional considerations, to be sure, requires some assumptions about identity attributes of perceivers not actually measured, as reasonable as it is to assume that most perceivers in the Snyder et al. (1982) study regarded themselves as open-minded and that most perceivers in the Neuberg study valued fairness and equality. More direct tests of hypotheses about identity-expressive motivations, however, would be provided by studies in which relevant features of the perceivers’ identities were actually assessed as an explicit part of the research procedures. One such investigation of the power of functional interventions to influence the outcomes of interactions between perceivers and the targets of their expectations has been conducted by Miene, Gresham, and Snyder (1991). In their experiment, Miene et al. (1991) arranged for pairs of previously unacquainted students to get acquainted with each other in telephone conversations. As often happens in such research, prior to these interactions, students assigned the role of perceiver received information about their interaction partners, in this case, that they would be interacting with extraverts or introverts. Of particular concern to understanding the role of the expressive function are those dyads in which perceivers hold expectations that contrast with their own personal identities, that is, dispositionally extraverted perceivers interacting with supposedly introverted targets and dispositionally introverted perceivers interacting with supposedly extraverted targets. When these interactions occurred in the

88

MARK SNYDER

circumstances typical of behavioral confirmation experiments, namely with perceivers motivated to get to know and to get along with their interaction partners, behavioral confirmation occurred. Targets who interacted with perceivers who believed them to be extraverts were seen as behaving in more extraverted fashion than targets interacting with perceivers who thought them to be introverts, even though the perceivers themselves were introverts or extraverts, respectively. However, when perceivers were explicitly encouraged to construe the acquaintance process as one of expressing and conveying their own personalities to their interaction partners, the outcome was behavioral disconfirmation. Under these Circumstances, it was the supposedly extraverted targets who were actually judged to have behaved in more introverted fashion during their conversations with introverted perceivers than targets believed by extraverted perceivers to be introverted. Moreover, these outcomes occurred independently of the actual personalities of the targets themselves. Thus, it would seem that, when the function served by interaction becomes one of expressing the perceiver’s own personal identities, behavioral disconfirmation can and will occur, as it did in these interactions in which the personal identity of the perceiver exerted a greater impact on the course of social interaction than did the perceiver’s expectations about the target. But what about the target? By definition, behavioral confirmation involves targets taking their cues from perceivers’ overtures toward them. Of special concern to behavioral confirmation researchers are those circumstances when perceivers’ initial beliefs are erroneous. Because the usual procedural paradigm involves the random assignment of beliefs to perceivers, behavioral confirmation is typically studied in circumstances in which there is no systematic relation between perceivers’ beliefs and targets’ personalities, thereby setting the stage for the frequent occurrence of interactions in which perceivers hold erroneous beliefs about targets. When behavioral confirmation occurs in such interactions, targets are behaving in ways that, also by definition, do not serve an identityexpressive function. In fact, were targets to use such interactions as opportunities to express their own personal identities, they would provide behavioral evidence that would discon8rm their perceivers’ beliefs. When and why might identity-expressive considerations become prominent on the motivational agendas of targets, thereby increasing the likelihood of behavioral disconfirmation? Targets who are dispositionally inclined to express their identities ought to treat interactions with perceivers as opportunities to display their own personal attitudes, values, and dispositions. Hence, such targets should be less responsive to actions of &heperceiver that might otherwise elicit behavioral confirmation. And, when their own personal attributes directly contradict the perceiver’s beliefs, their identity-expressive orientation will produce behavioral disconfirmation. Relevant to this line of reasoning, Deaux and Major (1987) have argued that targets low in self-monitoring (Snyder, 1987) and high in

BEHAVIORAL CONFIRMATION

89

private self-consciousness (Fenigstein, Scheier, & Buss, 1975) should be unlikely to confirm the beliefs and expectations of the perceivers with whom they interact; for some empirical evidence relevant to this proposition, see Grinder and Swim (1991). And, in fact, there is evidence that at least one of these categories of people (namely, low self-monitors) is characteristically responsive to identity-expressive considerations (Snyder & DeBono, 1987, 1989). Moreover, in some circumstances, targets high in private self-consciousness are particularly successful in getting perceivers to doubt their erroneous preinteraction expectations, suggesting a behavioral disconfirmation outcome (Major et al., 1988). No doubt, there are also situations that engage an identity-expressive function in targets, situations with inherent dividends accruing to those who accurately express and communicate personal attributes. Thus, for example, job interviews may engage an expressive function in candidates highly qualified by virtue of possessing job-relevant attitudes, values, dispositions and other personal attributes. Such candidates (as targets) may provide behavioral disconfirmatioh for erroneous preconceived beliefs of their prospective employers (as perceivers) in personnel-selection interviews. Unfortunately, although many behavioral confirmation experiments have employed job interview formats (e.g., von Baeyer et a]., 1981; Word et al., 1974), none has systematically manipulated the qualifications and motivations of targets. These considerations of the role of identity-expressive motivations in social interaction suggest, in summary, that when perceivers and targets become motivated to express their personal attribntes in social interaction, not only may they be serving the function of expressing their identities, but they may be inhibiting behavioral confirmation and possibly producing behavioral disconfirmation.

C. DEFENDING THREATENED IDENTITIES A further perspective on the when and the why of behavioral disconfirmation versus behavioral confirmation is provided by considering another motivational function typically absent from interactions between perceivers and targets. Usually, the beliefs involved in behavioral confirmation scenarios do not threaten the identity either of the perceiver or of the target. Since the beliefs typically concern the target’s personality, they by definition do not concern the perceiver’s identity and hence can pose no direct threats to it. And, although the beliefs at issue concern the target’s personality, the target after all is kept ignorant of the existence of the perceiver’s beliefs. The target can hardly be threatened by what he or she does not know. What would happen if the behavioral confirmation scenario were rewritten so that the beliefs involved posed some threat to the identity of those involved in it?

90

MARK SNYDER

Imagine a scenario involving a perceiver who believes that someone else does not share his or her self-perceptions. Specifically, this perceiver comes to believe that a target regards the perceiver as weak and submissive. This belief may constitute a threat to the perceiver’s conception that he or she is the dominant and powerful type. In response to this threat to self-conception, the perceiver may maneuver the target into weak and submissive postures that effectively disconfirm the perceiver’s belief that this target, and others as well, may see the perceiver as lacking in power and dominance. Scenarios of this form have been observed in experimental investigations of self-verification processes (e.g., Swann, 1983; Swann & Hill, 1982; Swann & Read, 1981). In one such study, people who saw themselves as rather likable learned that others might regard them as dislikable (Swann & Read, 1981). Effectively, these messages posed threats to these people’s conceptions of self. In subsequent social interaction, they behaved in ways clearly designed to eliminate these threats. Those who had their likable self-conceptions threatened elicited particularly positive reactions from their interaction partners. Effectively, these outcomes provided behavioral disconfirmation for perceivers’ beliefs about how they were regarded by targets. Consider now the involvement of identity-defending motivations in the activities of the target. Whenever targets are confronted by perceivers who expect them to behave in ways that are discrepant with their own self-conceptions, the stage is set for identity-defensiveefforts. Thus, a self-conceived extravert questioned by an interviewer probing for signs of introversion (as occurred in a study reported by Fazio et al., 1981) may experience some threat to self-conception. Similarly, the friendly and outgoing target who receives the cool and aloof treatment of a perceiver who believes her to be highly unsociable (as happened in a study by Christiansen & Rosenthal, 1982) may also experience some threat to self-conception. Yet, targets in such situations typically behave in accord with perceivers’ erroneous expectations rather than with their own self-conceptions and dispositions. Such actions, since they perpetuate the threat to self-conceptions provided by the perceiver’s erroneous beliefs about the targets, can hardly be said to serve identity-defensive functions. Yet, at least two features of behavioral confirmation situations may inhibit identity-defensiveactions by targets. First, because targets are typically unaware of the perceivers’ beliefs about them, they may not recognize identity threats posed by erroneous expectations held by perceivers. For this reason, identitydefensive motivations may not be engaged in such circumstances. In this regard, it is quite telling that the simple procedure of making the target aware of the existence of the perceiver’s preconceptions can be sufficient to short-circuit behavioral confirmation (e.g., Hilton & Darley, 1985). Second, even when identitydefensive motivations are engaged, structural constraints of the interactional setting may limit targets’ opportunities to act in ways that dispel perceivers’ false images and thus defend their threatened identities. In this regard, too, it is

BEHAVIORAL CONFIRMATION

91

instructive to learn that giving targets such a “defensive” response option sharply attenuates the likelihood of behavioral confirmation (D. T. Miller & Holmes, 1975). These considerations, not incidentally, highlight the importance of recognizing that the response options and behavioral opportunities available to targets may often be limited and hence place constraints on their likelihood of confirming or disconfirming the expectations of their perceivers. Moreover, when identity-defensive concerns are aroused for targets, they adopt modes of coping with the overtures of perceivers that produce behavioral disconfirmation outcomes. Swann and Ely (1984) arranged interactions in which perceivers always held beliefs that contradicted the targets’ self-conceptions about extraversion. Of these targets, those who were highly certain of their selfconceived extraversion provided behavioral disconfirmation for the perceivers’ erroneous beliefs. Eventually, by behaving in accord with their chronic and certain self-conceptions, they brought their perceivers’ beliefs into line with their own self-conceptions (see also Major et al., 1988). One interpretation of this outcome, in keeping with the one offered by Swann and Ely, is that the more certain the targets were of their self-conceptions, the more they should have been threatened by encountering perceivers who did not share their views (assuming, of course, that these perceivers possess sufficient credibility that their views are to be taken seriously). Hence, these targets should have been particularly motivated to defend against these threats by behaving in ways that would actively and decisively disconfirm erroneous beliefs about them. In line with this interpretation are observations that disabled people, who are often motivated to correct others’ beliefs and expectations about them, are all the more motivated to do so the more unflattering they consider others’ beliefs to be (for an elaborated discussion of the management of stigmatizing conditions, see especially E. E. Jones et al., 1984). To summarize this line of argument, then, when people experience others’ views of them as threatening to their identities, they can and do take actions that, in their roles as perceivers and targets, generate behavioral disconfirmation outcomes. Further, according to this argument, these disconfirmatory outcomes can be interpreted as serving an identity-defensive function of protecting perceivers and targets from accepting potentially unpleasant, unflattering, threatening, or otherwise unwelcome beliefs about the self.

VIII. Theoretical and Practical Implications of the Functional Analysis For social scientists, behavioral confirmation sequences are of special concern. They represent a particularly complex intertwining of cognitive activities and behavioral processes in the context of ongoing social interaction and interperson-

92

MARK SNYDER

a1 relationships. Behavioral confirmation phenomena also have intriguing implications for the reciprocal influences of “subjective” reality (the perceiver’s beliefs) and “objective” reality (the target’s behavior) and for the mutual interplay of individuals and their social worlds. And, not inconsequentially, they may be involved in a wide range of interaction contexts, including not only the diverse array of interpersonal situations modeled in laboratory investigations of behavioral confirmation, but also nonlaboratory settings involving interactions between teachers and students (e.g., Rosenthal & Jacobson, 1968), interactions between instructors and trainees in industrial settings (e.g., King, 1971), interactions between therapists and clients in counseling and therapeutic settings (e.g., Sibicky & Dovidio, 1986; Snyder & Thomsen, 1988; Vrugt, 1990), interactions between mothers and infants (e.g., Stem & Hildebrandt, 1986), and the interpersonal dynamics of some forms of psychopathology and problems of adjustment such as depression (e.g., Coyne, 1976a, 1976b; Hammen & Peters, 1978; Horowitz et al., 1991; Horowitz & Vitkus, 1986; Howes & Hokanson, 1979; Snyder & White, 1982; Strack & Coyne, 1983). Thus, the study of behavioral confirmation in social interaction, and the resulting understanding of how individuals are in some measure creators of their social worlds, may yield both theoretical and applied benefits. I have argued here that, as much as is known about the sequential events of behavioral confirmation, comparatively little is known about the psychological and motivational foundations of the contributions of perceivers and targets to such sequences. One approach to uncovering these foundations-the approach taken here-is to delve into the reasons and purposes underlying the activities of perceiver and target in behavioral confirmation sequences, that is, to analyze the psychological functions of behavioral confirmation. By definition, all behavioral Confirmation sequences share the same surface features, that is, the sequential events of perceivers acting on beliefs about targets and targets coming to behave in accord with perceivers’ beliefs. Yet, a functional analysis suggests that apparently similar surface features may reflect different underlying motivational processes and may be serving different personal, social, and psychological functions.

A , UNDERSTANDING BOTH CONFIRMATION AND DISCONFIRMATION Specifically, this analysis suggests that diverse functions associated with the acquisition of social knowledge, the facilitation of social interaction, the expression of personal attributes, and the defense of threatened identities may be served by the activities of perceivers and targets in their interactions. Using illustrative evidence drawn from the research literature, I have argued that this set

BEHAVIORAL CONFIRMATION

93

of functions can account for the motivational bases of interactional scenarios that culminate either in behavioral confirmation or in behavioral disconfirmation. According to this theoretical analysis, when people meet, interact, and get acquainted with each other (the prototypical behavioral confirmation situation), functions associated with the acquisition of social knowledge and the regulation of social interaction may be served by the behavioral confirmation that can and does occur in these circumstances. But, although perceiver and target may share the common goal of getting acquainted in such interaction contexts, analytical and empirical considerations converge to support the proposition that the perceiver’s contributions to behavioral confirmation serve primarily the function of acquiring knowledge at the same time as those of the target serve primarily the function of facilitating interaction. In addition to its utility in identifying the motivational foundations of behavioral confirmation in social interaction, the present analysis also helps remove some of the mystery surrounding the heretofore elusive phenomenon of behavioral disconfirmation. A functional analysis provides systematic specifications of some of the motivational foundations of such disconfirmation outcomes, and provides a common framework for understanding how the motivational agendas of perceivers and targets generate both confirmation and disconfirmation. When the activities of the perceiver serve motivational functions other than the acquisition of knowledge, the behavioral confirmation process is inhibited. Three such scenarios have been identified and documented. Behavioral confirmation tends not to occur and behavioral disconfirmation often does occur when perceivers operate in the service of regulating social interaction, expressing personal attributes, and defending threatened identities. By the same token, when the activities of the target serve motivational functions other than the regulation of interaction, the behavioral confirmation process is also inhibited. Again, three such scenarios have been identified and documented. Behavioral confirmation tends not to occur and behavioral disconfirmation often does occur when targets operate in the service of acquiring social knowledge, expressing personal attributes, and defending threatened identities.

B. FUNCTIONAL APPLICATIONS AND PRACTICAL PROBLEMS Carried further, this functional analysis carries considerable practical implications. Previous research has implicated behavioral confirmation in life situations of considerable consequence and significance, including counseling situations (e.g., Harris & Rosenthal, 1986; Sibicki & Dovidio, 1986), job interview settings (e.g., Word et al., 1974), and teacher-student relationships (e.g., Rosenthal & Jacobson, 1968). Thus, therapists and counselors need to know what to

94

MARK SNYDER

expect from their clients, typically having to make assessments of their clients’ well-being and their prognosis for improvement in treatment. Employers also seek a predictable view of job candidates, oftentimes trying to decide during the job interview what they are like and how they will perform on the job. Similarly, teachers often make judgments of their students and get an idea of what kinds of learning curves they can expect from them. That is, counselors, employers, and teachers often function essentially in the role of knowledge-oriented perceivers dealing with clients, candidates, and students about whom they hold expectations. And, they often function as perceivers under the constraints of heavy workloads, time pressures, and looming deadlines, which singly and together may heighten the motivation to seek the perceptual stability and cognitive clarity associated with the knowledge function (a point argued with considerable persuasiveness by Jamieson & Zanna, 1989), which in turn may make them especially likely to instigate the processes of behavioral confirmation. Moreover, as the targets of expectations, clients, candidates, and students are typically in positions of relatively low power and great dependence. And, more often than not, they will ascribe high status to their teachers, employers, and counselors, which may enhance the prospects for behavioral confirmation to occur (see Copeland, 1991, and Virdin & Neuberg, 1990), for related arguments about the responsiveness of targets to status and power). For these reasons, clients, candidates, and students may be especially motivated to facilitate the dynamics and outcomes of their interaction by being responsive to the expectations of counselors, employers, and teachers in the quest for favorable interaction outcomes (see also Snyder & Skrypnek, 1981, for evidence of cognitive processes that may facilitate the behavioral responsiveness of targets in such interactions). From a functional perspective, then, the stage very well may be set for behavioral confirmation of erroneous preconceived expectations to occur in such “professional” interactions (for elaboration, see Snyder & Thomsen, 1988). Moreover, and of added social significance, much concern with behavioral confirmation stems from its implications for the phenomena of stereotyping, prejudice, and discrimination. To some extent, it may be argued, people may cling to social stereotypes that at best have minimal validity because they elicit behavioral confirmation from the targets of their stereotypes. If, using stereotypes, people treat others in ways that bring out behaviors that confirm stereotyped expectations, they may get little opportunity to discover the ways in which their stereotypes are wrong. If, for example, in accord with prevailing stereotypes, people give women more opportunities than men to display sensitivity and empathy, then the behavior of men and women may come to confirm stereotyped beliefs, not necessarily because of any inherent accuracy of the stereotypes, but because of behavioral confirmation. The functional approach suggests that interventions directed at the motivational foundations of behavioral confirmation (perhaps ones specificallydesigned to engage the expressive and defensive func-

BEHAVIORAL CONFIRMATION

95

tions) can contribute to the breaking down of erroneous stereotypes. Such functional interventions can be directed either at perceivers who hold false stereotypes (inducing them not to act as if their stereotypes were true) or at the targets of false stereotypes (inducing them not to assimilate their behavior to the overtures of stereotype holders). For a related discussion of motivational approaches to understanding and addressing the self-fulfilling aspects of stereotypes, see Neuberg ( 1991).

IX. The Functional Analysis in Context Earlier, I noted that the present theoretical analysis has its precedents in earlier functional perspectives on personality, such as those of G. W. Allport (1937) and F. H. Allport (1937), and in the functional theories of attitudes and persuasion, such as those of Katz (1960) and Smith et al. (1956). However, the linkages between the functions thought to be served by behavioral confirmation and those thought to be served by attitudes should be explicitly acknowledged and critically evaluated. Although attitude theorists have tended to emphasize their own preferred sets of functions, certain ones occur with some regularity across diverse functional approaches (e.g., Katz, 1960; Smith et al., 1956). Each of these, it will be seen, has its counterpart in one of the functions employed in analyzing the motivational foundations of behavioral confirmation. Behavioral confirmation in the service of acquiring and using social knowledge clearly shares something with attitudes thought to serve a “knowledge” function by providing frames of reference for structuring, organizing, and bringing a sense of stability to perceptions of the world. Behavioral confirmation serving the function of regulating and facilitating interaction has its counterpart in attitudes serving the “social-adjustive” function of helping people to fit into social situations and to interact smoothly with members of their important reference groups. The role of identity-expressive motivations in preventing behavioral confirmation and in generating behavioral disconfirmation has its parallel in the attitude domain, where some attitudes are thought to serve the “valueexpressive” function of allowing people to express their true selves, their deeply held values, and their enduring personality dispositions. And, finally, the identity-defensive function thought to be served in some behavioral disconfirmation scenarios has its attitudinal counterpart; functional attitude theorists posit that some attitudes serve the “ego-defensive” function of protecting people from accepting unpleasant truths about the self. These close parallels between the functions of behavioral confirmation and the functions of attitudes prompt a question: To what extent is this analysis simply a translation of earlier theoretical statements about the functions of attitudes into

96

MARK SNYDER

the current theory of the functions of behavioral confirmation? The answer to this question is both “some” and “not much.” To be sure, that which makes a functional approach to attitudes appealing also makes a functional interpretation of behavioral confirmation appealing. That is, prior functional analyses, such as those of attitudes, certainly provide precedents for the present analysis of the functions served by behavioral confirmation; hence, the “some” part of the answer. But, and speaking to the “not much” in the answer, it is not simply a matter of reasoning by analogy, of proposing that what worked once with attitudes will work now with behavioral confirmation. The fact of the matter is that the functions used in this analysis emerge independently from considerations of the behavioral confirmation sequence and the context in which it occurs. They are derived, first, from inquiring about the functions these contexts engage (namely, the acquisition of knowledge and the regulation of interaction) and how these could be served by behavioral confirmation. Second, they are derived from inquiring about the functions typically precluded by structural constraints of the interactional setting (namely the expression of personal attributes and the defense of threatened identity) and how the motivational power of these functions could be harnessed to produce outcomes other than behavioral confirmation, including behavioral disconfirmation. That is, analyses of the structural features of social interaction sequences are generative in their own right, and are informative about the functional bases of confirmatory and disconfirmatory outcomes of those interactions. On the other hand, the similarity between the two sets of functions probably should not be dismissed as mere coincidence or curiosity either. That similar sets of functions may underlie phenomena as diverse as attitudes and behavioral confirmation may testify to the generality of these functions as the motivational foundations of a wide range of aspects of individual and social functioning (see Snyder, 1988). In this regard, it is worth noting that these sets of functions embody recurring themes in psychology. In fact, early functional attitude theorists claimed to be attempting to capture the themes of four of the “big picture” grand theories of human nature (psychodynamic theory for the defensive function, behaviorism for the adjustive function, ego psychology for the expressive function, and Gestalt psychology for the knowledge function; see Katz, 1960; Sarnoff & Katz, 1954). Related linkages are evident in the set of functions emerging from this analysis of behavioral confirmation. The function of regulating interaction, with its emphasis on the effects of social evaluations, very much captures the spirit of the social learning theories (e.g., Bandura, 1977). The function of acquiring knowledge, with its focus on the search for meaning, is readily linked to some of the central tenets of the attribution theories (e.g., klley, 1967). The function of

BEHAVIORAL CONFIRMATION

97

expressing personal attributes incorporates many of the notions of the humanistically oriented theories of the development and actualization of the self (e.g., Rogers, 1951). Finally, the function of defending threatened identities, with its emphasis on preserving the integrity of one’s self-image, is very much in keeping with the logic of psychodynamic thinking on the defense mechanisms (Freud, 1946). Moreover, it is not difficult to map linkages between psychological functions of the type being discussed here and theoretical attempts to construct catalogs of the basic human needs. For example, in Maslow’s need hierarchy (1968), the needs for safety and order may correspond to the function of acquiring stable and predictable knowledge, belongingness to regulating and promoting smoothly coordinated interaction, esteem to protecting threatened identities, and selfactualization to expressing personal attributes. Similar parallels may exist with the fruits of empirical labors to identify the fundamental dimensions of personality. For instance, in the “big five” basic traits proposed by, among others, McCrae and Costa (1987), neuroticism has some overlap with the function of defending threatened identities, extraversion and agreeableness with regulating interaction, openness to experience with acquiring knowledge, and conscientiousness with expressing personal attributes. Finally, in the taxonomy of goals mediated by social interaction offered by E. E. Jones and Thibaut (1958), there are resemblances between the goal of gaining cognitive clarity about the environment and the function of acquiring knowledge, between securing motivational and value support and elements of expressing and defending personal identities, and between maximizing beneficent social responses and facilitating social interaction.

A. JUST HOW MANY FUNCTIONS? AT WHAT LEVEL OF ANALYSIS?

Nevertheless, the facts that similar sets of functions emerge independently from analyses of phenomena as diverse as attitudes and behavioral confirmation and that these sets of functions embody recumng themes in psychological inquiries into the nature of needs, traits, and goals should probably not be taken as a sign that either set of functions is necessarily exhaustive. Indeed, in the specific case of behavioral confirmation, it is conceivable that further theoretical and empirical inquiry will dictate additions or modifications to this set of functions. Moreover, the present analysis has focused on social interaction in the context of acquaintance processes. Possibly, a functional analysis of the self-fulfilling consequences of interactions in other contexts (e.g., between teachers and students, between therapists and clients, between managers and workers) will reveal other

98

MARK SNYDER

functions served by the activities of perceivers and targets. Prime candidates for such functions are those served by efforts to control the actions of others and to otherwise determine the outcomes of one’s dealings with them. Needless to say, the work of the functional analysis is not yet complete. Further work will be needed to delineate the boundary conditions of the operation of psychological functions, providing specifications of the ways in which the contexts in which perceivers and targets interact may enhance or constrain their behavioral opportunities to act in the service of particular functions. Moreover, the critical question of the origins of these functions and their linkages to the roles of perceiver and target, as well as to the circumstances in which perceiver and target interact with each other, remains to be addressed. One can only hope that theoretical and empirical attention will be directed to these matters. In the meantime, it is important to make explicit some features of the functional account of behavioral confirmation that may help define its larger theoretical scope and context. When considering the functions implicated in the present analysis, one must keep in mind that these functions have been conceptualized in terms of their linkages to the roles of perceiver and target. In experiments on behavioral confirmation, it is possible to cast one member of the dyad in the role of perceiver (by providing that person with preinteraction expectations) and the other in the role of target (by withholding any preinteraction expectations). In much of social life as it occurs outside of the laboratory, people may be simultaneously holders of expectations about their interaction partners and targets of the expectations of these very same interaction partners. That is, they may be functioning as perceivers and as targets at one and the same time. That social interaction may ask people to play multiple roles poses no necessary problems for the functional analysis; rather, it suggests that multiple functions may need to be considered in attempts to understand the intertwining of these roles and the motivations underlying them. Moreover, when considering this functional analysis, one must recognize that these functions have been conceptualized in terms of their specific relevance to behavioral confirmation. Thus, for this reason, the particular functions served by behavioral confirmation may not be the only functions relevant to the interaction sequences in which confirmation occurs. Consider the knowledge-oriented motivation implicated in the activities of the perceiver. Behavioral confirmation, it has been argued, serves the function of providing perceivers with stable and predictable impressions of their interaction partners, thus creating a sense of an orderly world in which other people behave as they are expected to behave. However, the motivation to achieve stable and predictable impressions is not the only motivation associated with acquiring and using social knowledge. There is also, to be sure, the motivation to acquire accurate social knowledge. Indeed, in behavioral confirmation scenarios, the sense of a stable and predictable social

BEHAVIORAL CONFIRMATION

99

world may be purchased as the cost of inaccurate impressions. That is, for the perceiver, the sense of stability and predictability that is reinforced in the behavioral confirmation scenario may be provided by a target who, although behaving in accord with the perceiver’s expectations, may actually be behaving in ways that do not reflect his or her true personal attributes. Thus, the target who replies in kind to the standoffish treatment of a perceiver who is expecting an introvert may actually be someone with a rather extraverted disposition who can be counted on to be quite friendly and outgoing in most other life situations that do not involve perceivers with erroneous preconceptions. The perceiver does not know it, but the behavioral data that once again prove that this is a stable and predictable world are, in point of act, an unrepresentative and hence inaccurate sampling of the target’s behavior. This trade off between stability and accuracy is made especially evident by the fact that procedures designed to increase accuracy motivations can and do inhibit behavioral confirmation. This point is well illustrated in a study reported by Neuberg (1989). In this study, half of the perceivers were given a negative belief about the person they were interviewing, and half were not given any belief. Orthogonal to this belief, half of the perceivers were instructed to form accurate impressions of their partner, and half were not given these instructions. Neuberg found that perceivers working with an “accuracy” set did nor elicit behavioral confirmation from their targets, whereas those not explicitly encouraged to seek accurate impressions did elicit confirming behaviors from their targets. Taken together, Neuberg’s study and other experimental investigations of the functions of behavioral Confirmation suggest the following characterizations of the several motivations associated with perceivers’ beliefs about their social worlds. When perceivers are motivated by accuracy concerns (as in Neuberg’s study), they do not elicit behavioral confirmation, but when perceivers strive for stability and predictability (as in studies by Snyder & Haugen, 1990, and Copeland & Snyder, 1990), their dealings with other people will lead to behavioral confirmation for their beliefs and expectations. Thus, although the motivational functions associated with acquiring and using social knowledge may include those associated with stability and those associated with accuracy, only the former motivate perceivers to generate behavioral confirmation. I should point out, though, that although perceivers’ sense of stability and predictability at times may be gained at the cost of accuracy, these two facets of beliefs are not necessarily mutually exclusive. One can achieve social knowledge that is stable and that is also accurate. Indeed, the fact that the behavioral confirmation scenario so often involves initially erroneous beliefs makes it possible to tease apart these two aspects of social knowledge and the functions that behavioral confirmation serves for perceivers who act on their beliefs about targets. I should also acknowledge that the distinction that I am drawing between the stability and predictability provided by beliefs and their accuracy has paral-

100

MARK SNYDER

lels with Swann’s ( 1984) important distinction between circumscribed accuracy and global accuracy. Specifically, circumscribed accuracy involves specific beliefs about another person that concern specifically bounded situations (not unlike the beliefs that perceivers form about targets, beliefs that may serve them well in predicting other interactions in which they can elicit further behavioral confirmation from the same targets), whereas global accuracy involves widely generalizable beliefs about another person (not unlike the knowledge that perceivers would form if they correctly inferred the stable and enduring underlying dispositions of the targets, knowledge that would permit accurate predictions about the targets’ behavior in diverse settings and with different interaction partners). In addition, these distinctions have their parallels with some of the epistemic needs and motivations proposed by Kruglanski in his theory of lay epistemology (1987), especially his “need for specific structure” (which includes the motivation to attain knowledge that fits with prior expectations) and his “need for ambiguity” (which is often associated with high fear of invalidity and strong need for accuracy). Moreover, it is also possible to propose differentiations within the general category of functions associated with the facilitation of social interaction. I have argued that the activities of targets in behavioral confirmation sequences serve primarily the function of assuring a smoothly coordinated and responsive flow of social interaction. However, the behavioral confirmation occurring under these motivational circumstances may facilitate current interactions at the possible cost of longer term, continuing relationships. For example, consider a target who reciprocates the formal and reserved interactional stance of a perceiver who falsely believes the target to have a socially awkward personal style. That target may be adopting a coping strategy well suited to the immediate situation, but one that may minimize the chances of future contact and continuing interaction. After all, as responsive as the target has been to the perceiver and as well coordinated as their interaction has been, the perceiver may just as soon not pursue any further involvement with someone who seems so clearly shy and withdrawn. Thus, the motivation to facilitate current interaction and the motivation to ensure future interaction may both be members of the general category of functions associated with regulating social interaction. However, there may be important cases, such as those associated with perceivers holding negative expectations about targets, in which behavioral confirmation that facilitates current interaction may not serve the function of ensuring future interaction. Similar arguments can be made about the functions of expressing personal attributes and of defending threatened identities. The range of personal attributes that can be expressed in social interaction is a broad one indeed, perhaps as broad as the terrain of the psychology of human personality itself. Moreover, these personal attributes can be expressed in a multitude of ways, as a sprawling literature on the presentation of self testifies. Threatened identities can be pro-

BEHAVIORAL CONFIRMATION

101

tected with a wide range of defense mechanisms, as many a practicing clinician quickly learns. However, the present analysis has focused specifically on those expressive and defensive functions that can and do contribute to confirmation and disconfirmation outcomes in social interaction and interpersonal relationships. At the very least, though, these considerations should make clear that hypotheses based on the expressive and defensive functions cannot be fully specified, and tests of these hypotheses cannot be actually executed, without articulation of precisely which personal attributes of perceivers and targets are to be expressed or defended in the course of social interaction. Thus, it should be clear that when the analysis speaks of the functions of acquiring knowledge, regulating social interaction, expressing personal identities, and defending threatened identities, it is really speaking of particular instances of larger categories of functions, instances of particular relevance to behavioral confirmation and disconfirmation. To put it in slightly different terms, this analysis of functions has been conducted at a level of analysis specifically chosen to be appropriate to explicating the phenomena of behavioral confirmation and disconfirmation in social interaction. It is a level of analysis, I believe, appropriate not only to recognizing the motivational nature of these phenomena, but also to explicating the mediating processes by which the motivational agendas of perceivers and targets are defined and enacted in social interaction. In this choice of a level of analysis conducive to theorizing about motivational functions is, needless to say, a recognition of the importance of considerations of other levels of analysis and other psychological constructs of relevance to the interplay of belief and behavior in interaction. Thus, at a somewhat different level of analysis, discussions about the role of interaction goals and task objectives in behavioral confirmation (as well as in social perception and motivated cognition more generally) have been offered (e.g., Hilton & Darley, 1991; Kunda, 1990; Neuberg, 1991). Although the precise linkages among and between these levels of analysis remain to be fully specified, it is worth reiterating one of the core messages of the functional account of behavioral confirmation. From the functionalist perspective, in social interactions, individuals may be pursuing the same specific interaction goals or particular task objectives in the service of quite different underlying psychological functions; moreover, in the specific case of behavioral confirmation, it appears that both parties to the interaction share the common goal or objective of getting acquainted, but that perceivers pursue this goal in the service of the function of acquiring knowledge and that targets pursue this objective in the service of the function of facilitating social interaction. Moreover, and of further relevance to matters of the level of the analysis, the functional analysis, having been conducted in the specific context of acquaintance processes, has of necessity been focused on the early stages of social interaction and interpersonal relationships. The level of analysis, however, can be readily shifted to link the functions of behavioral confirmation with the moti-

102

MARK SNYDER

vational agendas associated with stages in the evolution of interpersonal relationships. Thus, in the early phases of relationships, learning about each other and assessing social compatibility may be primary concerns. If so, functions associated with acquiring knowledge and regulating interaction (and their associated behavioral confirmation and disconfirmation scenarios) should be involved in the interplay of beliefs and behavior in relatively young relationships. Then, as relationships progress and develop, subsequent events may engage other motivations (e.g., shifting values may bring expressive considerations to the fore, and events that threaten one or both partners’ security in the relationship may arouse defensive motivations). The behavioral disconfirmation scenarios associated with these functions may be more prevalent in relatively mature relationships. Moreover, as I hope to demonstrate next, the level of the analysis can also be shifted to move beyond the events of behavioral confirmation sequences.

B. THE SCOPE OF THE FUNCTIONAL ANALYSIS Can the scope of the analysis be extended to encompass other phenomena and processes of individual and social functioning? To be sure, much of the scientific and practical importance of behavioral confirmation sequences can be traced to their intertwining of subjective beliefs and objective behaviors in ongoing social interaction. Yet, the very same questions that a functional analysis asks about behavioral confirmation sequences can be asked about their constituent elements. Thus, one can ask with respect to social perception: What functions are served by the beliefs, assumptions, expectations, stereotypes, and hypotheses that people hold, whether or not they are involved in behavioral confirmation sequences? And, one can ask with respect to interpersonal behavior: What functions are served by social interactions and interpersonal relationships, whether or not they provide the contexts for behavioral confirmation? With respect to social perception, there have been some discussions of the functions of attributions (e.g., Forsyth, 1980; Tetlock, 1981), of stigmatization (e.g., Pryor, Reeder, & McManus, 1991; Pryor, Reeder, Vinacco, & Kott, 1989), of stereotypes (e.g., Miene & Snyder, 1989; Snyder & Miene, 1991), and of personal theories of reality (e.g., Epstein, 1990). Consider specifically the case of stereotyping. A functional analysis of stereotyping would suggest that what may appear to be the very same stereotype may serve quite different functions for different people who hold that stereotype and in different contexts in which that stereotype is engaged. Thus, for instance, negative stereotypes about those of another race may serve knowledge-oriented functions (providing, as most cognitive accounts of stereotyping propose, a ready framework for categorization, inference, and judgment). But, the same racial stereotypes may also serve functions tied to social interaction (promoting, for instance, rela-

BEHAVIORAL CONFIRMATION

103

tionships with others who hold like stereotypes). By the same token, such negative stereotypes may serve an identity-expressivefunction (perhaps being a manifestation of a value system opposed to equality). And finally, these stereotypes may serve an identity-defensive function (serving to shore up a weak and vulnerable sense of self-worth). The challenge of a functional analysis of stereotyping is to determine whether there exist (1) categories of people who tend to hold stereotypes serving particular functions (for example, are there people who tend to hold stereotypes formed on the basis of social considerations and others whose stereotypes flow from value considerations?), (2) types of stereotypes that themselves reflect particular functional motivations (for example, are there some stereotypes that typically serve defensive considerations and others that are most likely to serve knowledge considerations?), and (3) some kinds of situations that promote the formation of stereotypes that serve particular functions (thus, do evaluative and judgmental situations generate stereotypes serving knowledge functions and do situations that pose threats to perceivers’ self-esteem and sense of self-worth lead them to form stereotypes that serve defensive functions?). Moreover, a functional analysis of stereotyping may provide the key to successful attempts to modify erroneous stereotypes. Interventions to alter stereotypes may be effective to the extent that they address the function supporting that stereotype. For example, when stereotypes serve knowledge functions, there may be particular utility in educating people about the errors of categorization and inference that may contribute to stereotyping. But, when the very same stereotypes serve other than knowledge functions, the same educational interventions may have little or no impact at all. Thus, for instance, with stereotypes that serve as expressions of identities and values, it may be necessary to convince people that their stereotypes really do not support their personal values or that, as much as their stereotypes may support some of their values, they may contradict others of their values, perhaps even higher order values of greater importance and centrality in their value systems. Stereotypes that serve to facilitate social interaction patterns may require changes in patterns of social interaction (perhaps even changes in social milieus) to reduce the probability of contact with those who provide social support for the stereotypes to be changed. And, stereotypes serving an identity-defensive function may require concerted attacks on the defense mechanisms thought to be the mechanisms by which that function is served. Some of these implications for functionally based interventions to inhibit stereotyping have been addressed empirically by Snyder and Miene ( 1991). With respect to interpersonal behavior, functional inquiries may contribute to an understanding of the dynamics of social interaction, the processes of friendships and other personal relationships, and the motivational bases of participation in groups and organizations. Consider, for example, the determinants of friendship. Some friendships may be contracted and maintained in the service of

104

MARK SNYDER

functions associated with knowledge acquisition and utilization (included in this category are those friends who provide an understanding of the way the world is and what the people who make up the world are like). Other friendships may serve to facilitate the expression of personal identities and values (as in the case of people seeking out others of like attitudes, values, and personality, thus providing affirmation for these attributes as well as receptive audiences for expressing them). Other friendships may be founded on considerations of facilitating social interaction (such as those friends chosen on the basis of their strategic value in helping one fit in and get along with important peers and other influential reference groups). Still other friendships may serve identity-defensive considerations (as in the case of men insecure in their sex role identity associating with hypermasculine types to shore up their own masculine self-images, or in the case of insecure people in positions of authority surrounding themselves with deferential types who flatter their sense of power and control). Just as types of friendships differ in their functional underpinnings, so too may people differ in their general propensities to form friendships that serve particular functions. Thus, some people may be particularly likely to form friendships that serve knowledge-oriented functions (perhaps those high in need for cognition may seek friends who contribute to knowledge acquisition). Other people may be motivated to seek friends who serve social purposes (perhaps high self-monitors, considering the involvement of social adjustive considerations in other arenas of their lives; Snyder & DeBono, 1989). Still other people may be disposed to form friendships that serve expressive functions (perhaps low self-monitors, who seem to be guided by expressive considerations in other life contexts; Snyder & DeBono, 1989). Finally, other people may gravitate toward friends who fulfill identity-defensivefunctions (perhaps those of chronically low and uncertain selfesteem, who may use friends to shore up their fragile sense of self, or those high in authoritarianism, who may have substantial amounts of the kinds of inner conflict that defensive actions might resolve). In either case, considerations of function may be relevant not only to understanding the origins of friendship, but also the kinds of disruptive events to which each type of friendship is vulnerable and what types of occurrences may contribute to the dissolution of each type of friendship. Moreover, just as friendships can be addressed from a functional perspective, so too can other forms of social relationships. To focus on one example of particular concern to me, there have been considerations of the functions that motivate and sustain ongoing relationships that involve volunteers in helping roles (e.g., Clary & Snyder, 1991a; Clary, Snyder, & Ridge, in press; Omoto & Snyder, 1990; Omoto, Snyder, & Berghuis, in press; Snyder & Omoto, 1991). Every year, millions of people in this country volunteer to help others, providing, among other services, health care to the sick, companionship to the elderly, tutoring to the illiterate, and social support to persons living with AIDS. What

BEHAVIORAL CONFIRMATION

105

motivates people to make continuing commitments to the well-being of others, to devote substantial amounts of time and energy to helping others, often doing so over extended periods of time? A functional approach to the motivational foundations of volunteerism would suggest that acts of volunteerism that appear to be quite similar on the surface may reflect markedly different underlying motivational processes. That is, the same acts of volunteerism may be serving quite different psychological functions for different people. Thus, to illustrate the logic of this functional approach to voluntary helping relationships, the act of volunteering may for one person may be motivated by desires to solidify social ties with friends and significant others who are themselves volunteers, thus serving the function of facilitating social interaction. For another person, the same act of volunteering may flow from personal humanitarian values that prescribe altruistic contributions to society, thus serving the function of expressing personal identity. For yet another person, volunteering may serve the knowledge-oriented function of providing an understanding of the world and how to serve those in need. In still other cases, being a volunteer may serve to protect and defend aspects of identity, perhaps helping convince those in doubt that they are competent and worthwhile, valued, and valuable people. Empirical support for this functional approach to volunteerism has been provided in both laboratory research (e.g.. Clary & Snyder, 1990, 1991b; Ridge et al., 1990, 1991) and field studies (Omoto & Snyder, 1990; Omoto et al., in press; Snyder & Omoto, 1991). Similarly, a diversity of functions (including social comparison, social exchange, social control, social esteem, social identity, and social learning) have been found to underlie people’s descriptions of the needs met by membership in long-term, community-based groups (Forsyth, Elliott, & Welsh, 1991). More generally, the same logic of functional analysis has been applied to understanding membership and participation in advocacy groups, social movements, and political activism (e.g., Snyder & Clary, 1990). In each case, the theme is a common one-what may appear on the surface to be similar forms of participatory activism may reflect quite different underlying functional motivations. Moreover, I would venture that the motivational power of psychological functions can be harnessed to promote individual and collective action in response to social issues and societal problems (for elaboration, see Snyder & Clay, 1990).

X. Conclusion Let me conclude in the most general of terms. Functional analyses, I believe, have the potential to advance theory and research on social perception and interpersonal behavior. They constitute an explicit recognition of the moti-

106

MARK SNYDER

vational and purposive agendas that guide and direct human thoughts, feelings, and actions, and they speak directly to the mediational mechanisms and guiding processes involved in the enactment of these motivational agendas. For these reasons, I believe that the functionalist perspective possesses considerable explanatory power in diverse domains of human functioning. Nevertheless, I should emphasize that I see functional analyses as sharing much in common with related treatments of the roles of needs, motives, plans, and goals in understanding cognitive and behavioral processes in the social realm (e.g., Bruner & Goodman, 1947; Bruner & Postman, 1948; Dodge, Asher, & Parkhurst, 1989; Dweck & Leggett, 1988; Emmons, 1989; Fiske & Neuberg, 1990; Hilton & Darley, 1991; Hilton, Darley, & Fleming, 1989; E. E. Jones & Thibaut, 1958; Kelley & Thibaut, 1978; Kruglanski, 1990; Kunda, 1990; Pervin, 1983; Read & Miller, 1989; Renshaw & Asher, 1982; Showers & Cantor, 1985). I can only hope that the functional approach will derive some “gilt by association” by mingling in the good company of such worthy motivational partners. To be sure, these diverse approaches to human motivation have their points of individuation; in fact, I have gone to some length to articulate the particular appeals of the functional approach and the sources of its integrative explanatory powers. Yet, I do believe that it is the commonalities of these approaches that may provide the most compelling testimony to the power of motivational approaches to understanding individual and social functioning. In fact, it is precisely for this reason that I have gone far beyond the immediate terrain of behavioral confirmation in explicating the particular theoretical and empirical allure of analyses of psychological functions. Looking to the future, I would hope that, just as an analysis of psychological functions may help us to understand the motivational foundations of the chain of events in behavioral confirmation scenarios, it may also be possible for a functional analysis to provide a motivational perspective on the dynamic interplay of cognitive, behavioral, and interpersonal processes in social interaction and interpersonal relationships. Of course, as a longtime student of the power of beliefs and expectations to define and create their own eventual reality, I am mindful of the possibility that this statement just might turn out to be a self-fulfilling prophecy. Would that it would!

Acknowledgments The author’s research on behavioral confirmation in social interaction has been supported over the years by National Science Foundation Grants SOC 75-13872, BNS 77-1 1346. BNS 82-07632, BNS 87-18558, and DBC 91-20973. Preparation of this manuscript has been supported by a fellowship

BEHAVIORAL CONFIRMATION

107

from the James McKeen Cattell Fund and by the Center for Advanced Study in the Behavioral Sciences, Stanford, California, where much of it was written. For their comments and suggestions on earlier versions of this essay, sincere thanks go to Robert Abelson, Ellen Berscheid, Eugene Borgida, Gil Clary, John Copeland, Jennifer Crocker, John Darley, Kenneth Dodge, April Gresham, Monica Harris, Julie Haugen, Leonard Horowitz, William Ickes, E. E. Jones, Peter Miene, Kathleen Much, Steven Neuberg, Allen Omoto, Cynthia Thomsen, and Mark Zanna.

References Allport, F. H. (1937). Teleonomic description in the study of personality. Character and Personaliry, 5, 202-214. Allport, G . W. (1937). Personaliry: A psychological interpretation. New York: Holt. Andersen, S., & Bem, S. L. (1981). Sex typing and androgyny in dyadic interaction. Journal of Personaliry and Social Psychology. 41, 74-86. Angell, J. R. (1907). The province of functional psychology. Psychological Review, 14, 61-91. Babad, E. Y. (1979). Personality correlates of susceptibility to biasing information. Journal of Personality and Social Psychology, 37, 195-202. Babad, E. Y., & Inbar, J. (1981). Performance and personality correlates of teachers’ susceptibility to biasing information. Journal of Personaliry and Social Psychology. 40,553-561 . Babad, E. Y., Inbar, I., & Rosenthal, R. (l982a). Pygmalion, Galatea, and the Golem: Investigations of biased and unbiased teachers. Journal of Educational Psychology. 74, 459-474. Babad, E. Y., Inbar, J.. & Rosenthal. R. (1982b). Teachers’ judgment of students’ potential as a function of teachers’ susceptibility to biasing information. Journal of Personality and Social Psycholog.v, 42, 541-547. Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall. Bond, M. H. (1972). Effect of an impression set on subsequent behavior. Journal of Personaliry and Social Psychology, 24, 301-305. Bootzin, R. R. (1971). Expectancy and individual differences in experimenter bias. Journal of General Psychology, 84, 303-312. Boring, E. G. (1950). A history of experimentalpsychology (2nd ed.). New York: Appleton-CenturyCrofts. Brophy, J., & Good, T. (1974). Teacher-student relationships: Causes and consequences. New York: Holt, Rinehart, & Winston. Brown, P., & Levinson, S . C. (1987). Politeness: Some universals in language usage. Cambridge: Cambridge University Press. Bruner, J. S., & Goodman, C. C. (1947). Value and need as organizing factors in perception. Journal of Abnormal Social Psychology, 42, 33-44. Bruner, J. S . , & Postman, L. (1948). Symbolic value as an organizing factor in perception. Journal of Social Psychology, 27, 203-208. Buxton, C. E. (1985). American functionalism. In C. E. Buxton (Ed.), Points of view in the modern history of psychology (Chap. 5, pp. 113-140). Orlando, FL: Academic Press. Christiansen, D., & Rosenthal. R. (1982). Gender and nonverbal decoding skill as determinants of interpersonal expectancy effects. Journal of Personaliry and Social Psychology, 42, 75-87. Clary, E. G., & Snyder, M. (1990, March). A functional analysis of volunteers’ motivations. Paper presented at the Independent Sector Research Forum. Boston.

108

MARK SNYDER

Clary, E. G., & Snyder, M. (1991a). A functional analysis of altrusim and volunteerism. In M. S . Clark (Ed.), Review of personality and social psychology (Vol. 12). Beverly Hills, CA: Sage. Clary. E. G., & Snyder, M. (1991b. March). The recruitment of volunteers: Persuasive communications. Paper presented at the Independent Sector Research Forum, Cleveland, OH. Clary, E. G., Snyder, M., & Ridge, R. (in press). Volunteers' motivations: A functional strategy for thqrecruitment. placement, and retention of volunteers. Nonprofit Management and Leadership, Cooper, H. M. (1979). Pygmalion grows up: A model for teacher expectation communication and performance influence. Review of Educational Research, 49, 389-410. Cooper, H. M.,& Good, T. (1983). Pygmalion grows up: Studies in the expectation communicarion process. New York: Longman. Cooper, H.M., & Hazelrigg, P. (1988). Personality moderators of interpersonal expectancy effects: An integrative research review. Journal of Personality and Social Psychology, 55, 937-949. Copeland, J. T. (1991). Motivational implications of social power for behavioral confirmation. Doctoral dissertation, University of Minnesota, Minneapolis. Copeland, J. T, & Snyder, M. (1989, August). Efecrs of expectation certainty on counselors' hypothesis-resting strategies. Paper presented at the annual meetings of the American Rychological Association, Atlanta, GA. Copeland, 1. T., & Snyder, M. (1990). Unpublished research on the functions of the perceiver in peer counseling interactions, University of Minnesota, Minneapolis. Coyne, J. (1976a). Depression and the response of others. Journal of Abnormal Psychology, 85, 186- 193. Coyne, J. (1976b). Toward an interactional description of depression. Psychiatry, 39, 28-39. Curtis, R. C., & Miller, K. (1986). Believing another likes or dislikes you: Behaviors making the beliefs come true. Journal of Personalify and Social Psychology, 51, 284-290. Darley, J. M., & Fazio, R. H. (1980). Expectancy-confirmation processes arising in the social interaction sequence. American Psychologist. 35, 867-881. Darley, J. M., Fleming, J. H., Hilton, J. L., & Swann, W. B., Jr. (1988). Dispelling negative expectancies: The impact of interaction goals and target characteristics on the expectancy confirmation process. Journal of Experimental Social Psychology, 24, 19-36. Davis, D. (1982). Determinants of responsiveness in dyadic interaction. In W.Ickes & E. Knowles (Eds.), Personality, roles and social behavior. New York: Springer-Verlag. Davis, D., & Perkowitz, W. T. (1979). Consequences of responsiveness in dyadic interaction: Effects of probability of response and proportion of content-related responses on interpersonal attraction. Journal of Personaliry and Social Psychology, 37, 534-551. Deaux, K., & Major, B. (1987). Putting gender into context: An interactive model of gender-related behavior. Psychological Review. 94, 369-389. Dewey, J. (1896). The reflex an: concept in psychology. Psychological Review, 3, 357-370. Dodge, K. A., Asher, S. R., & Parkhurst, J. T. (1989). Social life as a goal coordination task. Research on Motivation and Education, 3, 107-135. Dweck, C. S.,& Leggett, E. L. (1988). A social-cognitive approach to motivation and personality. Psychological Review, 95, 256-273. Eden, D., & Shani, A. B. (1982). Pygmalion goes to boot camp: Expectancy, leadership, and trainee performance. Journal of Applied Psychology, 67, 194- 199. Emmons, R. A. (1989). The personal striving approach to personality. In L. Pervin (Ed.), Goal concepts in Personality and social psychology (Chap. 3, pp. 87- 126). Hillsdale, NJ: Erlbaum. Epstein, S. (1990). Cognitive-experiential self-theory. In L. Pervin (Ed.), Handbook of personality theory and research (Chap. 7, pp. 165-192). New York Guilford. Fazio, R. H., E f i i n , E. A., & Falender, V. J. (1981). Self-perceptions following social interaction. Journal of Personality and Social Psychology, 41, 232-242.

BEHAVIORAL CONFIRMATION

109

Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private self-consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology. 43, 522-527. Fiske, S. T., & Neuberg, S. L. (1990). A continuum model of impression formation from categorybased to individuating processes: Influences of information and motivation on attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1-74). San Diego: Academic Press. Fong, G. T., & Markus, H. (1982). Self-schemas and judgments about others. Social Cognition, 1, 19 I -204. Forsyth. D. R. (1980). The functions of attributions. Social Psychology Quarterly, 43, 184189. Forsyth, D. R., Elliott, T. R., & Welsh, J. A. (1991, June). Multidimensional model of the functions of groups. Paper presented at the annual meetings of the American Psychological Society, Washington, DC. Freud, A. (1946). The ego and the mechanisms of defense. New York: International Universities Press. Goffman, E. (1955). On face work: An analysis of ritual elements in social interaction. Psychiatry. 18, 213-231. Goffman, E. (1959). The presentation of selfin everyday life. Garden City, NY: Doubleday Anchor Books. Grinder, E. L., & Swim, J. K . (1991). Am I who you think I am?: The role ofbehavioral conjrmation and self-verification in gender identity negotiation. Unpublished manuscript, Pennsylvania State University, University Park. Hammen, C., & Peters, S. D. (1978). Interpersonal consequences of depression: Responses to men and women enacting a depressed role. Journal of Abnormal Psychology, 87, 322-332. Hams, M. J. (1989). Personality moderators of interpersonal expectancy effects. Replication of Harris and Rosenthal (1986). Journal of Research in Personality. 23, 381-397. Harris, M. J., Milich, R., Johnston, E. M., & Hoover, D. W. (1990). Effects of expectancies on children’s social interactions. Journal of Experimental Social Psychology. 26, I - 12. Harris, M. J., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects: 31 metaanalyses. Psychological Bulletin, 97, 363-386. Harris, M. 1.. & Rosenthal, R. (1986). Counselor and client personality as determinants of counselor expectancy effects. Journal of Personality and Social Psychology, 50, 362-369. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Herek, G. M. (1987). Can functions be measured? A new perspective on the functional approach to attitudes. Social Psychology Quarter/-v, 50, 285-303. Herr, P. M. (1986). Consequences of priming: Judgment and behavior. Journal of Personality and Social Ps.vchology. 51, 1106-1 115. Hilgard, E. R. (1987). Psychology in America: A historical survey. Orlando, FL: Harcourt Brace Jovanovich. Hilton, J. L., & Darley, J. M. (1985). Constructing other persons: A limit on the effect. Journal of Experimental Social Psychology, 21, I- 18. Hilton, J. L., & Darley, J. M. (1991). The effects of interaction goals on person perception. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 236-267). San Diego: Academic Press. Hilton, J. L., Darley, J. M., & Fleming, 1. H. (1989). Self-defeating behaviors and self-fulfilling prophecies. In R. C. Curtis (Ed.), Selfdefeating behaviors: Experimental research, clinical impressions, and practical implications (Chap. 3, pp. 41-65). New York: Plenum. Horowitz, L. M., Locke, D. D., Morse, M. B., Waikar, S. V., Dryer, D. C., Tamow, E., & Ghannam, J. ( 1991). Self-derogations and the interpersonal theory. Journal of Personality and Social Psychology, 61, 68-79.

110

MARK SNYDER

Horowitz, L. M., & Vitkus, J. (1986). The interpersonal basis of psychiatric symptoms. Clinical Psychology Review, 6 , 443-469. Howes, M. J., & Hokanson, J. E. (1979). Conversational and social responses to depressive interpersonal behavior. Journal of Abnormal Psychology, 88, 625-634. Ickes, W., Patterson, M. L., Rajecki, D. W., & Tanford, S. (1982). Behavioral and cognitive consequences of reciprocal versus compensatory responses to pre-interaction expectancies. Social Cognition, l, 160-190. James, W. (1890). Principles ofpsychology (Vols. 1-2). New York: Holt. Jamieson, D. W., & Zanna, M. P. (1989). Need for structure in attitude formation and expression. In Pratkanis, A. R., Breckler, S. J., & Greenwald, A. G. (Eds.), Attitude structure and function (pp. 383-406). Hillsdale. NJ: Erlbaum. Jones, E. E. (1964). Ingratiation. New York: Appleton-Century-Crofts. Jones, E. E. (1986). Interpreting interpersonal behavior: The effects of expectancies. Science. 234, 41-46.

Jones, E. E. (1990). Interpersonal perception. New York: Freeman. Jones, E. E., & Davis, K. E. (1965). From acts to dispositions: The attribution process in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2). New York: Academic Press. Jones, E. E . , Farina, A., Hastorf, A., Markus, H., Miller, D. T , & Scott, R. A. (1984). Social stigma: The psychology of marked relationships. New York: Freeman. Jones, E. E., & Thibaut, J. (1958). Interaction goals as bases of inference in interpersonal perception. In R. Tagiuri & L. Petrullo (Eds.),Person perception and interpersonal behavior. Stanford, CA: Stanford University Press. Jones, R. A. ( 1977). SeIf-ju&lling prophecies: Social, psychological and physiological effects of expectancies. Hillsdale, NJ: Erlbaum. Jones, S. C.. & Panitch, D. (1971). The self-fulfilling prophecy and interpersonal attraction. Journal of Experimental Social Psychology, 7 , 356-366. Jussim, L. ( 1986). Self-fulfilling prophecies: A theoretical and integrative review. Psychological Review, 93, 429-445. Jussim, L. (1990). Social reality and social problems: The role of expectancies. Journal of Social Issues, 46, 9-34. Jussim, L. (1991). Social perception and social reality: A reflection-construction model. Psychological Review, 98, 54-73. Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly. 24, 163-204.

Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska Symposium on Motivation. Lincoln: University of Nebraska. Kelley, H. H., & Stahelski, A. J. (1970). The social interaction basis of cooperators’ and competitors’ beliefs about others. Journal of Personality and Social Psychology, 16, 66-91. Kelley, H. H., & Thibaut, 1. W. (1978). Inferpersonalrelations: A theory of interdependence. New York: Wiley (Interscience). King, A. S. (1971). Self-fulfilling prophecies in training the hard-core: Supervisors’ expectations and the underprivileged workers’ performance. Social Science Quarterly. 52, 369-378. Kruglanski, A. W. (1987). Knowledge as a social psychological construct. In D. Bar-Tal & A. W. Kruglanski (Eds.), The social psychology of knowledge. Cambridge: Cambridge University Press. Kruglanski, A. W.( I 990). Motivation for judging and knowing: Implications for causal attribution. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2). New York: Guilford. Kunda, A. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480-498.

BEHAVIORAL CONFIRMATION

111

Major, B., Cozzarelli, C., Testa, M., & McFarlin, D. B. (1988). Self-verification vs. expectancy confirmation in social interaction: The impact of self-focus. Personality and Social Psychology Bulletin. 14, 346-359. Maslow, A. H. (1968). Toward a psychology of being (2nd ed.). New York: Van Nostrand. McCrae, R. R., & Costa, P. T., Jr. (1987). Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 8 1-90. Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8, 193-210. Miene, P. K., Gresham, A. W., & Snyder, M. (1991, June). Motivational functions influence behavioral confirmation. Paper presented at the annual meetings of the American Psychological Society, Washington, DC. Miene, P. K., & Snyder, M. (1989, August). Stereotypes of the elderly: A functional approach. Paper presented at the annual meetings of the American Psychological Association, New Orleans. Miller, C. T., Malcarne, V., Clarke, R., Loboto, D., & Fitzgerald, M. ( 1 986). SelfIfulfIling prophecies and children's expectations about mentally retarded children. Paper presented at the annual meetings of the American Psychological Association, Washington, DC. Miller, D. T., & Holmes, J. G. (1975). The role of situational restrictiveness on self-fulfilling prophecies: A theoretical and empirical extension of Kelley and Stahelski's triangle hypothesis. Journal of Personality and Social Psychology. 31, 66 1-673. Miller, D. T., & Turnbull, W. (1986). Expectancies and interpersonal processes. In M. R. Rosenzweig & L. W. Porter (Eds.), Annual review of psychology (Vol. 37). Palo Alto, CA: Annual Reviews, Inc. Musser, L. M., & Graziano, W. G. (1983). Self-fulflling prophecies in children's interactions with peers. Paper presented at the biennial meetings of the Society for Research in Child Development, Detroit, MI. Neuberg, S. L. (1989). The goal of forming accurate impressions during social interactions: Attenuating the impact of negative expectancies. Journal of Personality and Social Psychology. 56, 374-380. Neuberg, S. L. (1991, June). Stereotypes, prejudice. and expectancy-confirmation processes. Paper presented at the Ontario Symposium on Personality and Social Psychology, Waterloo, Ontario, Canada. Omoto, A. M., & Snyder, M. (1990). Basic research in action: Volunteerism and society's response to AIDS. Personality and Social Psychology Bulletin, 16, 152-166. Omoto, A. M., Snyder, M., & Berghuis, J. P. (in press). The psychology of volunteerism: A conceptual analysis and a program of action research. In J. B. Pryor & G. D. Reeder (Eds.), The social psychology of HIV infection. Hillsdale, NJ: Erlbaum. Pervin, L. A. (1983). The stasis and flow of behavior: Toward a theory of goals. In M. Page (Ed.), Personality: Current theory and research. Lincoln: University of Nebraska Press. Pervin, L. A. (1989). Goal concepts in personality and social psychology. Hillsdale, NJ: Erlbaum. Pryor, I. B., Reeder, G. D., & McManus, J. A. (1991). Fear and loathing in the workplace: Reactions to AIDS-infected co-workers. Personality and Social Psychology Bulletin. 17, 133139.

Pryor, J. B., Reeder, G. D., Vinacco, R., & Kott, T. L. (1989). The instrumental and symbolic functions of attitudes toward persons with AIDS. Journal of Applied Social Psychology, 19, 377-404.

Rabiner, D., & Coie, J. (1989). Effects of expectancy inductions on rejected children's acceptance by unfamiliar peers. Developmental Psychology, 25, 450-457. Read, S. J., & Miller, L. C. (1989). Inter-personalism: Toward a goal-based theory of persons in relationships. In L. Pervin (Ed.), Goal concepts in personality and socialpsychology (Chap. 11, pp. 413-472). Hillsdale, NJ: Erlbaum.

112

MARK SNYDER

Renshaw, P. D., & Asher, S. R. (1982). Social competence and peer status: The distinction between goals and strategies. In K. H. Rubin & H. S. Ross (Eds.), Peer relationships and social skills in childhood (pp. 375-395). New York: Springer-Verlag. Ridge, R. D., Snyder, M., Clary, E. G., French, S.. Miene, P. K.. Haugen, J. A., & Copeland, J. (1990, August). The Volunteer Functions Inventory: Toward an understanding of the motivations to volunteer. Paper presented at the annual meetings of the American Psychological Association, Boston. Ridge, R. D., Snyder, M., Clary, E. G., Haugen, I. A . , Copeland, I. T., & Miene, P. K. (1991, June). Matching messages to motives in persuasion: A functional approach to promoting volunteerisrn. Paper presented at the annual meetings of the American Psychological Society, Washington, DC. Riggs, J. M., Monach, E. M. Ogburn, J. A., & Pahides, S. (1983). Inducing self-perceptions: The role of social interaction. Personality and Social Psychology Bulletin, 9, 253-260. Rogers, C. R. (195 I). Clientcentered therapy. Boston: Houghton-Mifflin. Rosenthal, R. (1974). On the social psychology of the self-fuulfilling prophecy: Further evidence for Pygmalion effects and their mediating mechanisms. New York M. S. S. Inf. C o p . Modular Publications. Rosenthal, R., & Fode, K. L. (1963). Three experiments in experimenter bias. Psychological Reports, 12, 491-51 1. Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom. New York: Holt, Rinehart, & Winston. Rubovitz, R., & Maehr, M. (1971). Pygmalion analyzed: Toward an explanation of the RosenthalJacobson findings. Journal of Personality and Social Psychology, 19, 197-203. Samoff, I., & Katz, D. (1954). The motivational bases of attitude change. Journal of Abnormal and Social Psychology, 49, 115- 124. Schlenker, B. R. (1980). Impression management: The self-concept, social identity, and interpersonal relations. Monterey, CA: Brooks/Cole. Shapiro, J., & Hilton, J. L. (1988. May). The impact of interaction goals on expectancy confirmation. Paper presented at the annual meetings of the Midwestern Psychological Association, Chicago, IL. Shavitt, S. (1989). Functional imperative theory. In A. R. Pratkanis, S . 1. Breckler, & A. G. Greenwald (Eds.), Attitude structure and function. Hillsdale, NJ:Erlbaum. Showers, C., & Cantor. N. (1985). Social cognition: A look at motivated strategies. In M. R. Rosenzweig & L. W. Porter (Eds.), Annual review of psychology (Vol. 36). Palo Alto, CA: Annual Reviews, Inc. Sibicky, M., & Dovidio, J. F. (1986). Stigma of psychological therapy: Stereotypes, interpersonal reactions, and the self-fulfilling prophecy. Journal of Counseling Psychology, 33, 148- 154. Skrypnek, B. J., & Snyder, M. (1982). On the self-perpetuating nature of stereotypes about women and men. Journal of Experimental Social Psychology, 18, 277-291. Smith, M. B., Bruner, J. S . , & White, R. W. (1956). Opinions and personality. New Yo*. Wiley. Snyder, M. (1984). When belief creates reality. In L. Berkowitz (Ed.),Advances in experimental social psychology (Vol. 18, pp. 248-305). Orlando, FL: Academic Press. Snyder, M. (1987). Public appearanceslPrivate realities: The psychology of self-monitoring. New York: Freeman. Snyder, M. (1988, August). Needs andgoals. plans andmotives: The new "new look" in personality and social psychology. Invited state of the art address, American Psychological Association, Atlanta, GA. Snyder, M.. Campbell, B., & Preston, E. (1982). Testing hypotheses about human n a m : Assessing the accuracy of social stereotypes. Social Cognition. 1, 256-272. Snyder, M., & Clary, E. G. (1990, August). Social activism and the problem of inaction: Afunc-

BEHAVIORAL CONFIRMATION

113

tional perspective. Paper presented at the annual meetings of the American Psychological Association, Boston. Snyder, M., & DeBono, K. G. (1987). A functional approach to attitudes and persuasion. In M. P. Zanna, J. M., Olson, & C. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5). Hillsdale, NJ: Erlbaum. Snyder, M., & DeBono, K. G. (1989). Understanding the functions of attitudes: Lessons from personality and social behavior. In A. R. Pratkanis. S. J. Breckler, & A. G. Greenwald (Eds.), Attitude structure and functions. Hillsdale, NJ: Erlbaum. Snyder, M., & Haugen, J. A. (1990, August). Why does behavioral conjrmation occur?A functional perspective. Paper presented at the annual meetings of the American Psychological Association, Boston. Snyder, M., & Haugen, J. A. (1991). Unpublished research on the functions of the target in behavioral confirmation sequences, University of Minnesota, Minneapolis. Snyder, M., Manning, D. I., & Ridge, R. (1990). Unpublished research on the functions of the perceiver in behavioral confirmation, University of Minnesota, Minneapolis. Snyder, M., & Miene, P. K. (1991, June). On the functions of stereotyping and prejudice. Paper presented at the Ontario Symposium on Personality and Social Psychology, Waterloo, Ontario, Canada. Snyder, M.,& Omoto, A. M. (1991). Who helps and why? The psychology of AIDS volunteerism. In S. Oskamp & S. Spacapan (Eds.), Helping and being helped in the real world: The Claremont symposium on applied social psychology. Newbury Park, CA: Sage. Snyder, M., & Skrypnek, B. J. (1981). Testing hypotheses about the self: Assessments of job suitability. Journal of Personalify, 49, 193-21 1. Snyder, M., & Swann, W. B., Jr. (1978a). Behavioral confirmation in social interaction: From social perception to social reality. Journal of Experimental Social Psychology, 14, 148- 162. Snyder, M., & Swann, W. B., Jr. (1978b). Hypothesis-testing processes in social interaction. Journal of Personality and Social Psychology, 36, 1202- I2 12. Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and interpersonal behavior: On the self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666. Snyder, M., & Thomsen, C. J. (1988). Interactions between therapists and clients: Hypothesis-testing and behavioral confirmation. In D. C. Turk & P. Salovey (Eds.), Reasoning, inference. and judgment in clinical psychology: Theory, assessment, and treatment. New York: Free Press. Snyder, M., & White, P. (1982). Moods and memories: Elation, depression, and the remembering of the events of one’s life. Journal of Personality. 50, 149-167. Stem, M., & Hildebrandt, K. A. (1986). Prematurity stereotypic: Effects on mother-infant interaction. Child Development, 57, 308-315. Strack, S., & Coyne, J. (1983). Social confirmation of dysphoria: Shared and private reactions to depression. Journal of Personality and Social Psychology, 44,798-806. Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into harmony with the self. In J. Suls & A. Greenwald (Eds.), Psychological perspectives on the sev. Hillsdale, NJ: Erlbaum. Swann, W. B., Jr. (1984). Quest for accuracy in person perception: A matter of pragmatics. Psychological Review. 91, 457-477. Swann, W. B., Jr., & Ely, R. J. (1984). A battle of wills: Self-verification versus behavioral confirmation. Journal of Personality and Social Psychology, 46, 1287- 1302. Swann, W. B., Jr., & Hill, C. A. (1982). When our identities are mistaken: Reaffirming selfconceptions through social interaction. Journal of Personality and Social Psychology, 43, 5966. Swann, W. B., Jr., & Read, S. J. (1981). Self-verification processes: How we sustain our selfconceptions. Journal of Experimental Social Psychology, 17, 35 1-372.

114

MARK SNYDER

Swann, W. B., Jr., & Snyder, M. (1980). On translating beliefs into action: Theories of ability and their application in an instructional setting. Journal of Personalify and Social Psychology, 38, 879-888.

Tetlock, P. E. (198 1). Influence of self-presentational goals on attributional reports. Social Psychology Quarterly, 44, 300-31 1. Thomas, E. A. E., & Malone, T.W. (1979). On the dynamics of two-person interactions. Psychobgical Review, 86, 331-360. Tichener, E. B. (1899). Structural and functional psychology. Philosophical Review, 8, 290-299. Virdin, L. M., & Neuberg, S. L. (1990, August). Is perceived status a moderator of expectancy confirmation? Paper presented at the annual meetings of the American Psychological Association, Boston. von Baeyer, C. L., Sherk, D. L., & Zanna, M. P. (1981). Impression management in the job interview: When the female applicant meets the male (chauvinist) interviewer. Personality and Social Psychology Bulletin. 7 , 45-52. Vrugt, A. (1990). Negative attitudes, nonverbal behavior and self-fulfilling prophecy in simulated therapy interviews. Journal ofNonver6al Behavior, 14, 77-86. Wilson Dallas, M. E., & Baron, R. S . (1985). Do psychotherapists use a confirmatory strategy during interviewing? Journal of Social and Clinical Psychology, 3, 106- 122. Word, C. 0.. Zanna, M. P., & Cooper, J. (1974). The nonverbal mediation of self-fulfilling prophecies in interracial interaction. Journal ofExperimenta1 Social Psychology, 10, 109-120. Zanna, M. P., & Pack, S . J. (1975). On the self-fulfilling nature of apparent sex differences in behavior. Journal of Experimental Social Psychology, 11, 584-591. Zanna, M. P., Sheras, P. L., Cooper, J., & Shaw, C. (1975). Pygmalion and Galatea: The interactive effects of teacher and student expectancies. Journal of Experimental Social Psychology. 11, 275-287.

A RELATIONAL MODEL OF AUTHORITY IN GROUPS

Tom R. Tyler E. Allan Lind

I. The Social Psychology of Authority A popular image in novels, films, and biographies is that of a group of people, typically strangers, thrown together by shipwreck, war, or other calamity, who must struggle together to survive. William Golding’s novel Lord of the Flies (1962) is one example of such a scenario. The novel describes the conflicts within a group of boys stranded together on an island. Under these circumstances the efforts of groups to survive usually lead them to create rules regulating interactions or to designate authorities to be responsible for making decisions. It is not necessary to study groups on a desert island to witness the creation of rules and authority structures. When everyday groups or organizations-legal, political, managerial, or familial-are faced with problems, a common response is to create rules and empower authorities. This propensity toward structure and authority has been demonstrated in laboratory experiments on rule creation and social dilemmas, as well as in field studies of how people deal with real-world conflicts and problems of scarcity. For example, the experimental research of Thibaut and his colleagues shows that people create rules to preserve profitable, but potentially unstable, relationships (Murdoch, 1967; Murdoch & Rosen, 1970; Thibaut, 1968; Thibaut & Faucheux, 1965; Thibaut & Gruder, 1969). In the seminal Thibaut and Faucheux experiment, bargaining situations were created that involved two parties whose power varied along different dimensions: One party had either high or low influence over the other’s outcomes within the relationship, and the other party had either good or poor alternatives available outside the group. Thibaut and Faucheux hypothesized that when one party has high internal power and the other

ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY. VOL. 25

115

Copyright 0 1992 by Academic h s s . Inc. All nghls of reproduclion in any lorn reserved.

116

TOM R. TYLER AND E. ALLAN LIND

has good alternatives, a clear threat to the viability of the relationship exists. The party with higher internal power is tempted to be exploitative, and the lower power party is then tempted to exit the group. Thibaut and Faucheux found that people were especially likely to agree on rules regulating behavior in this highthreat situation, because both parties were motivated to preserve their mutually beneficial relationship and both had some power with which to bargain. Another example of the creation of authority structures is found in the research literature on social dilemmas. Using an experimentally created renewable resource pool, Messick and colleagues studied how groups respond to the possibility that some group members will deplete a resource that benefits everyone in the group. A frequently observed response is for the group to create an authority structure in which one person regulates the behavior of all members of the group (Messick et al., 1983; Samuelson, Messick, Rutte, & Wilke, 1984). A real-world example of rule creation in response to potential conflict is provided by public responses to the 1974 energy crisis. A survey conducted at the time showed that the public supported the creation of laws, such as a law mandating car pooling, that would require all members of the public to engage in conservation behavior (Sears, Tyler, Citrin, & Kinder, 1978). The survey respondents preferred mandatory conservation rules to calls for voluntary conservation. A real-world example of the empowerment of authorities in response to perceived threat is seen in polls on fear of crime. Polls show that when the public feels that crime is a serious problem, people are especially likely to support increasing the discretionary authority of the police and courts. People are likely to support allowing the police to stop people on the street for identification, to hold people in jail without charges, and to conduct wiretaps (Block, 1971; Thomas & Hyman, 1977). At such times, public endorsement of severe punishments such as the death penalty is also greater ('Qler & Weber, 1982).' These examples show two distinct responses to crisis and conflict in groups: the creation of rules and laws and the empowerment of authorities. As demonstrated by the Thibaut and Faucheux experiment and the Sears et al. survey, groups sometimes create impersonal rules and norms to govern the behavior of group members. On the other hand, as the social dilemma studies and the surveys on crime show, another response is to empower one or more members of the group to make decisions and issue directives to govern group actions. Although similar in purpose, creating rules and empowering authorities are distinctly different group responses. The creation of a rule restricts the actions of all group members, whereas the empowerment of an authority restricts the actions of 'We do not mean to suggest that perceived threat is the major force driving the empowerment of authorities. The literature in symbolic politics shows that self-interest effects of this sort do occur, but public reactions are not dominated by self-interest. As will be seen later in this chapter, the literature on reactions to authorities shows that the impact of self-interest concerns in that context is also rather limited.

AUTHORITY IN GROUPS

117

ordinary members and simultaneously increases the discretion of the leader. Thus, enacting rules moves power from individuals to an impersonal norm, whereas constituting authorities conveys power from some individuals to another individual or to an institution.* This article is devoted to the latter response-the creation of authorities-and to issues surrounding the legitimacy of leaders. As seen in the second half of this article, the personal aspect of the authority process is a key element in the psychology of legitimacy and obedience. Of course, social psychologists recognize that creating authority structures is not a magical solution to problems within groups. To be of value, authorities must function effectively, actually helping the group to solve its problems. Ineffective authority structures will not necessarily facilitate group functioning, and authorities, once created, may make decisions that prove to be counterproductive. Hence a key issue in the study of authority is understanding the conditions under which authorities in fact enhance group functioning (Dawes, Orbell, Simmons, & van de Kragt, 1986).

A. WHAT IS NEEDED FOR AUTHORITIES

TO FUNCTION EFFECTIVELY? The first precondition for the effective functioning of authorities is authoritativeness. That is to say, the directives of authorities can only be effective when, for whatever reason, people obey them. But compliance is not something that authorities can simply assume will occur: Research suggests that people do not automatically accept authorities’ decisions. Studies of legal authorities find that judges and police officers have difficulty securing compliance with a wide variety of orders and legal decisions (Tyler, 1990). Research on small claims courts, for example, shows that judicial orders are accepted and obeyed only about 50% of the time (McEwen & Maiman, 1984). In work settings, studies show that workers often avoid, resist, or ignore the policies and decisions of management (Lissak, 1983). In political settings, citizens sometimes disobey authorities’ directives on issues ranging from paying taxes and fighting wars to recycling waste and conserving resources. Hence, a problem common to all types of organized groups and their leaders is the creation of conditions under which *The distinction between rule-making and authority-empowerment is captured in an item on the F-scale: “What this country needs most, more than laws and political programs, is a few courageous, tireless, devoted leaders in whom the people can put their faith” (see Robinson & Shaver, 1973). Kelman and Hamilton (1989) use an interview item that reads “The government must sometimes set constitutional principles aside in order to deal with problems of the modern world”. In some circumstances these two responses to group crises might be viewed as alternative choices. In other circumstances both might be attempted.

118

TOM R. TYLER AND E. ALLAN LIND

leaders will be accepted as authoritative and their decisions obeyed. Understanding these conditions is essential to understanding the psychology of authority. In this article we focus on one particular aspect of authoritativeness: voluntary compliance with the decisions of authorities. Social psychologists have long distinguished between obedience that is the result of coercion and obedience that is the result of internal attitudes and opinions (i.e., “voluntary” compliance). Raven and French (1958; French & Raven, 1959; see also Collins & Raven, 1969) describe reward power and coercive power, in which obedience is contingent on positive and negative outcomes, and distinguish both of these types of power from legitimate power, in which obedience flows from judgments about the legitimacy of the authority. Legitimate power depends on people taking upon themselves the obligation to obey and voluntarily follow the decisions made by authorities (Kelman & Hamilton, 1989; Qler, 1990). Kelman (1958) distinguishes between compliance, obedience that occurs in response to the provision of material outcomes; identijicarion, obedience that occurs because the authority is held in high esteem; and internalization, obedience that occurs because the values of the authority are accepted (see also Kelman & Hamilton, 1989). In Kelman’s typology either identification or internalization can imbue an authority with legitimacy, leading to voluntary compliance with the authorities’ directives. In limiting our present analysis to internal, legitimacy-related, authority processes we do not mean to suggest that these are the only reasons people obey authorities, but we do think that these processes represent the most potent and most interesting reasons for compliance. We have no doubt that behavioral compliance can be obtained, at least in the short term, by external threats and rewards, but there is evidence in the social psychological literature (beginning with Raven & French, 1958) that obtaining compliance through the use of reward and coercive power is unwieldy, costly, and time-consuming. In addition, coercive power ultimately fails to control behavior because it can neither eliminate private disobedience nor assure continued obedience when the mechanism of coercion (or reward) is removed. Legitimacy, in contrast, nearly always facilitates and is often crucial for the effective exercise of authority. Legitimacy, once established, functions to enhance acceptance of decisions as long as the authority is viewed as legitimate (Qler, 1990). Evidence of the varying effectiveness of different authority structures and different types of power is seen in a classic study by Lewin, Lippitt, and White (1939) on democratic and autocratic leadership in groups. Lewin et al. showed that not all authority structures have the same effect on groups. In particular, their study demonstrated that authority structures differ in their capacity to generate voluntary compliance with decisions. Lewin et al. found that groups of boys obeyed the decisions enacted by democratic leaders following group discussions whether the leaders were present or absent. Because obedience was observed even when the leaders were absent,

AUTHORITY IN GROUPS

119

it is clear that the democratic leadership structure produced voluntary compliance. In contrast, groups with autocratic leaders obeyed the decisions enacted by the leaders when the leaders were present, but not when the leaders were absent. Thus, for the autocratic groups there was little evidence of voluntary compliance. Although Lewin et al. did not measure legitimacy per se, it is reasonable to suppose that differences in the legitimacy of the democratic and autocratic leaders mediated the compliance differences seen in the study. This article focuses on legitimacy because it is the social attitude most central to the effectiveness of authorities. It is important to recognize, however, that legitimacy is not the only attitudinal factor influencing effectiveness. Effectiveness is also influenced by other cognitions about the authority, most notably judgments of his or her expertise with respect to the problem at hand.3 The willingness of group members to accept a leader’s directives is only helpful when the leader knows what directives to issue.

B. WHAT IS THE PSYCHOLOGICAL BASIS OF LEGITIMACY? We begin our analysis of the psychology of legitimacy by considering three theoretical perspectives that have been used to explain how people evaluate their experiences with organizations and groups: social exchange theories, distributive justice theories, and procedural justice theories. Theoretical analyses of authority processes in other social sciences, especially in political science, make frequent reference to what is termed “public choice theory” (Laver, 1981), a utility-based, outcome-oriented theory that shares many of the assumptions of social exchange theories in social psychology. According to public choice theory-and social exchange theory-people obey authorities because they view this as the best way to maximize their outcomes. We contrast these predictions to the predictions of distributive and procedural justice theories, two alternative explanations of how people react to social situations. C. THEORETICAL CONSIDERATIONS I . Social Exchange Theories

Social exchange theories offer one theoretical perspective that speaks clearly to what it is that governs evaluations of group^.^ Social exchange theories and related models of behavior view people as motivated by self-interest in their 3In the Raven and French taxonomy of types of power, expert power (and an associated source of influence based on control over information) is distinguished from legitimate power. 4We use the term social exchange to characterize theories based on self-interest and the perception of outcome contingencies. It is important, though, that we not oversimplify the case. There are

120

TOM R. TYLER AND E. ALLAN LIND

interactions with others (Blau, 1964; Homans, 1961; Kelley & Thibaut, 1978; Thibaut & Kelley, 1959). People enter and leave relationships depending on their judgments concerning the benefits and costs of remaining in the relationship relative to taking other possible actions; people evaluate interactions in which they are involved in terms of rewards and costs, and, by extension, people evaluate and react to authorities and their decisions in terms of what they gain and lose from the authorities' decisions. Some psychological theories of leadership evaluation have adopted the social exchange perspective in positing that the extent to which leaders will be regarded as authoritative depends on the ability to provide favorable outcomes to followers (Hollander, 1980; Hollander & Julian, 1978). A generic social exchange model would suggest that reactions to authorities are, in the final analysis, instrumentally based, developing out of expected or achieved gains and losses. Social exchange models explain the authority-creating behavior outlined at the beginning of this article with reference to the outcomes associated with group membership: People empower leaders in an effort to develop and preserve beneficial group interactions. In social dilemma situations, for example, people do not wish to lose the renewable resource on which they rely, so it is in their self-interest to create authorities. The authorities, in turn, are evaluated against a criterion of self-interest: i.e., how well they solve problems in the group and maintain the flow of resources to group members. Social exchange theories suggest some variables that might be important to the perception of legitimacy. In particular, legitimacy should be tied to the extent to which the authority's decisions and the procedures used to generate those decisions are personally fa~orable.~ The idea that outcomes, presumably assessed relative to the perceiver's expectations (Thibaut & Kelley, 1959), influence evaluations of authorities is a straightforward application of the basic assumption of self-interest, which underlies all social exchange theories.6 Another factor affecting evaluations of procedures, according to social exchange theories, is the desirability of the procedures themselves, by which we mean the procedure's transaction costs, the cost, delay, and frustration experienced in using the pro-

outcome-orientedmodels of social behavior, such as the Kelley md Thibaut (1978) theory of interdependence, that are quite sophisticated about the types of outcomes people seek and the way in which actual outcome contingencies are transformed by norms and by cognitive processes. Lind and Earley (1992) have pointed out that many of the relational effects and psychological processes we describe herein are in fact consistent with the Kelley and Thibaut theory. SRasinski, bier, and Fridkin (1985) have shown that evaluations of authorities are based on performance when the authority's legitimacy is based on personal, rather than institutional, loyalty. 6Sophisticated interdependence theories allow for the possibility that "raw" outcome patterns are transformed psychologically prior to affecting evaluations and behavior (Kelley & Thibaut, 1978; see also Lind & Earley, 1992).

AUTHORITY IN GROUPS

121

cedure (Lissak & Sheppard, 1983). The idea that the desirability of decisionmaking procedures affects leader evaluations arises because social exchange theories view people as willing to sacrifice short-term gain for better long-term outcomes. That is, people may be willing to accept substantial transaction costs, as long as they think that the decisions and decision-making procedures of the authority will benefit them in the long run.

2 . Justice Models In contrast to social exchange models, justice models suggest that people react to authorities not by judging whether the authorities’ decisions benefit their interests, but rather by assessing whether the authority is acting fairly.’ There are two general classes of justice judgments that have received attention in social psychology: distributivejustice judgments, which emphasize the fairness of outcomes and of allocation patterns, and procedural justice judgments, which emphasize the fairness of decision-making procedures and other aspects of social process. Justice theories suggest that authorities will be evaluated positively and their decisions will be obeyed to the extent that the procedure and the authority’s decision satisfy justice norms. a. Distributive Justice. There are a number of theories that deal with people’s concerns about outcome fairness. This view is represented in psychology by relative deprivation theories (Crosby, 1976; Stouffer, Suchman, DeVinney, Star, & Williams, 1949) and equity theories (Adams, 1965; Walster, Walster, & Berscheid, 1978). Theories of distributive fairness suggest that people evaluate authorities by comparing the outcomes they receive to the outcomes others receive and using this comparison to determine whether the outcome distribution accords with accepted principles of fairness. Distributive justice theories differ in terms of what is seen as the preeminent principle of justice. Equity theories (Adams, 1965; Greenberg, 1990a; Walster, Berscheid, & Walster, 1973; Walster & Walster, 1975; Walster et al., 1978) focus on proportionality of inputs and outcomes as the sine qua non of justice. Later distributive justice work has expanded the list of comparativejustice principles to include such criteria as the equality of outcomes across the group or the distribution of outcomes according to need. Another class of distributive justice theories includes those that focus on feelings of relative deprivation. Relative deprivation theorists (Crosby, 1976, 7Demonstrating that people evaluate outcomes, rules, or authorities in justice terms does not preclude the possibility that justice judgments are influenced by a concern with self-interest. In fact, distributive justice theories originally developed within the context of social exchange theory. Justice theories are important here because they suggest factors-distributive and procedural justice-that have some substantial variation beyond that which can be accounted for by simple self-interest.

122

TOM R. TYLER AND E. ALLAN LIND

1982; Folger, 1986; Masters & Smith, 1987; Olson, Herman, & Zanna, 1986) argue that the fundamental source of feelings of injustice is the realization that one’s outcomes fall short of expectations. Expectations are derived, in turn, both from imagined outcomes under different circumstances and from the outcomes of salient others. For relative deprivation theorists, the key questions concern the generation of expectations and the cognitive process by which people explain why their actual outcomes fall short of expectations. The end product of these cognitive processes is the same as that which is of interest to equity theorists, however: the feeling that outcomes are unfair. We consider distributive justice judgments as possible factors in legitimacy because distributivejustice theories of all types view outcome fairness judgments as a critical factor in determining people’s satisfaction with the groups and organizations to which they belong. In addition, there is a substantial body of research linking distributive justice judgments to variables closely related to legitimacy. For example, relative deprivation studies have linked the perceived injustice of outcomes to political unrest, and equity studies have shown that inequity affects satisfaction with work settings and organizations. b. Procedural Justice. People’s reactions to authorities’ decisions have also been linked to judgments about the fairness of the procedures used to make decisions. Procedural justice theories suggest that people focus on how decisions are made, as well as the decisions themselves, in making justice evaluations (Leventhal, 1976, 1980; Lind & Tyler, 1988; Thibaut & Walker, 1975, 1978). Since the publication of Procedural Justice by John Thibaut and Laurens Walker (1973, a substantial body of research has addressed issues of procedural justice (see Lind & Tyler, 1988, for a recent review). As research has accumulated, it has become evident that fairness judgments concerning informal social process are so similar in cause and consequence to fairness judgments concerning formal decision-making procedures that they can be considered simply to be different manifestations of the same type of justice judgment (cf. Bies & Moag, 1986; Lind & Tyler, 1988, p. 214; Tyler & Bies, 1990). The findings of the initial Thibaut and Walker research-that the procedures used to arrive at a decision can have profound effects on fairness judgments-have been widely confirmed in subsequent studies of legal trial procedures (e.g., Lind, Kurtz, Musante, Walker, & Thibaut, 1980), in studies of nontrial procedures used in resolving legal disputes, such as plea bargaining (Casper, Tyler, & Fisher, 1988; Houlden, 1980), mediation, and arbitration (Adler, Hensler, & Nelson, 1983; Lind, MacCoun et a]., 1989, 1990; MacCoun, Lind, Hensler, Bryant, & Ebener, 1988), and in studies of police officer dealings with citizens (Tyler, 1988, 1990; Tyler & Folger, 1980). In addition, researchers have found that concerns about procedural justice extend to organizational (Folger & Greenberg, 1985; Greenberg & Folger, 1983; Sheppard, 1984), political (Qler & Caine, 1981), interpersonal (BarrettHoward & Qler, 1986), and educational (Qler & Caine, 1981) settings. In fact,

AUTHORITY IN GROUPS

123

wherever procedural justice issues have been studied they have emerged as an important concern. * Procedural justice has been shown to be an important factor in the evaluation of groups and institutions. Studies of legal authorities (Tyler, 1984, 1990), political authorities (Tyler & Caine, 1981; Tyler, Rasinski, & McGraw, 1985), and managerial authorities (Alexander & Ruderman, 1987; Folger & Konovsky, 1989) all provide evidence that judgments about the fairness of procedures shape evaluations of authorities. At first glance, this may seem unremarkable, since procedures structure the behavior of authorities, but the link between the evaluations of procedures and the evaluations of those who enact the procedures has been seen even in circumstances, such as formal legal trials, where the authority has relatively little leeway in deciding which procedure to use. 3. Summary: Factors Affecting Legitimacy

The three sets of theories outlined above suggest a number of factors that people might use when they evaluate the authorities with whom they deal. These factors can be differentiated in terms of the extent to which they concern outcome versus procedures and the extent to which they involve assessments of the benefit or cost of outcomes or procedures versus assessments of fairness. Figure 1 illustrates the dimensions and factors that serve as the focus of our review of research on legitimacy. We turn in the next section to a review of research on legitimacy and the evaluation of leaders. The key empirical question is how four factors-outcome favorability, procedural desirability, distributive fairness, and procedural fairness-affect reactions to authorities. A second question, which is essential to understanding the implications of the findings for the psychology of authority, takes us one step back the causal chain. Once we have determined which of these four factors are most important to legitimacy, we will explore the psychological processes underlying that factor. We will examine two general types of studies: those that address specific experiences with authorities (e.g., “When I was stopped by a police officer last month and given a traffic ticket”) and those that explore overall judgments based on multiple encounters with the authority in question (e.g., “My supervisor generally makes decisions following fair procedures”). Because our interest is in cross-situational determinants of legitimacy, we examine research conducted in a variety of settings, looking for patterns of findings that appear to be general and not restricted to a particular setting. *See Qler and Lind (1990) for a discussion of some potential limits to the procedural justice effect. It is an empirical question whether procedural justice concerns affect the treatment of individuals outside the boundary of the group.

124

TOM R. TYLER AND E. ALLAN LIND

Outcomes

Fairness

fairness

Proc4Ursl Desirablllty

Procedural Falrneas

What types of reactions are of concern? As already noted, attitudes about the legitimacy of authorities appear to be a key antecedent of compliance. Hence, we are interested in legitimacy and related judgments of leaders and other authoritie~.~ Of course, data showing compliant behavior per se are of interest. We also report findings on attitudinal and belief variables that seem closely linked to legitimacy and on behaviors that seem likely to be associated with either legitimacy or compliance.

D. EMPIRICAL STUDIES Studies of authority, legitimacy, and compliance have used nonexperimental research designs more frequently than is the case in many other areas of social psychology. The reliance on nonexperimental designs is logical: Authority by its very nature implies a certain level of group structure, and the short-lived and undifferentiated groups used in most laboratory research on group processes are not well-suited for the study of such a phenomenon (Sears, 1986). Researchers have generally found it easier and more fruitful to study existing authorities in real-world settings, even though these settings often preclude the use of experimental designs. But this aspect of authority research raises some special problems. Social psychologists have traditionally viewed the findings of nonexperi%us, we do not examine the large literature on the antecedents of personal reactions to decisions, i.e., personal satisfaction and affective reactions to experiences. We are only concerned with such feelings to the extent they affect evaluations of authorities. A negative outcome might leave one unhappy, but this unhappiness only has implications for authority processes if the authority is held responsible.

AUTHORITY IN GROUPS

125

mental studies with skepticism. In particular, they have been loath to draw causal inferences from correlational data. Fortunately, the literature on the psychology of authority offers a solution: Although a majority of the studies are single-shot survey studies, wherein causal inference is admittedly problematic, most of the key findings have been confirmed in studies that offer much more “leverage” on causal issues. We will point out, as we review the literature, these critical studies-laboratory experiments and panel studies designed to resolve causal ambiguity-and we will attempt to give the reader a clear picture of which findings can lay claim to clear causality and which cannot.

1 . Attitudes toward Authorities a . Legal Authority. 51er (1984) examined the relationship between people’s judgments of outcome favorability, distributive fairness, and procedural fairness in small claims court experiences and their attitudes about the legitimacy of the legal authorities they encountered. He found that procedural fairness had a significant relationship to legitimacy (p = .19, p < .001), as did distributive fairness (p = .24, p < .001). Outcome favorability had no significant relationship (p = .03, nonsignificant [n.~.]). This finding was replicated in another study of experiences with legal authorities, which included experiences with both judges and police officers (Tyler, 1990). The study used a cross-sectional design that involved interviewing a random sample of Chicago residents (n = 1575). Tyler analyzed the attitudes of those respondents in the sample who had had recent personal experience with legal authorities (n = 652). He found that the respondents’ views about the legitimacy of the legal authority were correlated with judgments of the procedural justice of their recent experience ( r = .19, p < .001) and the distributive justice of the outcome of the experience (r = .15, p < .001), but were not strongly related to the favorability of the outcome ( r = .08, n.s.). Tyler, Casper, and Fisher (1989) examined the relationship between evaluations of criminal justice system experiences and general attitudes about the legitimacy of law and government. Their study used a panel sample of defendants on trial for felonies in criminal court ( n = 329). The defendants were interviewed both before and after their encounter with the courts. The results, including a causal model diagrammed in Fig. 2,’O indicate that procedural justice judgments show the strongest relationship to attitudes about the legitimacy of legal authorities, law, and government, and that neither distributive justice judgments nor the favorability of sentences had much direct impact on attitudes about “JFigure2 presents only a portion of the Qler, Casper, and Fisher model. The full path model can be found in the original article.

TOM R. TYLER AND E. ALLAN LlND

Sentence

Fig. 2. Effects of sentence on attitudes toward law and legal authorities. From Sler, Casper, and Fisher (1989).

authorities. The panel design used in this study allows for strong causal inferences because it can control statistically for prior attitudes about authorities. Although these findings suggest that procedural justice is an important factor in legitimacy, it is important not to overstate the case. In both the Vler (1990) and the Tyler et al. (1989) studies, outcome factors were found to influence procedural fairness judgments (i.e., those who won were more likely to think that the process was fair). Through their influence on fairness judgments, outcome factors had an indirect influence on attitudes about authorities and rulerelated behavior. We will return to this issue later in this article. 6 . Managerial Aurhorify. Alexander and Ruderman (1987) examined the relationship between distributive and procedural fairness and trust in management among 2800 federal employees. The employees were interviewed about various aspects of the job environment, including the fairness of decision-making procedures and outcomes. Their judgments about the job environment were then used to predict their overall evaluations of management. Alexander and Ruderman (1987) found that evaluations of management were influenced both by employee assessments of the fairness of outcomes (unique variance explained = 5%) and by their assessments of the fairness of decision-making procedures within their work setting (unique variance explained = 11%). Several studies have examined the factors that influence commitment to organizations, a positive attitude toward the organization that is similar in concept to legitimacy (see O’Reilly & Chatman, 1986). Studying manufacturing workers,

127

AUTHORITY IN GROUPS

Folger and Konovsky (1989) examined the relationship between judgments of the distributive and procedural fairness of pay raises and workers’ organizational commitment. They found that procedural justice explained significant variance in organizational commitment ( r = .43, p < .001), as did distributive justice (r = .33, p < .001). The actual dollar amount of raises had no impact on organizational commitment ( r = .11, n.s.1. Tyler and Schuller (1990) studied organizational commitment and evaluations of supervisors by employees in a variety of work settings. They interviewed a random sample of 409 workers in the Chicago area about recent personal experiences with a supervisor. Attitudinal reactions to supervisors were found to be more closely related to procedural justice judgments than to favorability or distributive justice judgments (see Table I). All of these studies suggest that procedural justice judgments play a central role in determining how authorities are viewed in organizational settings. An experimental study by Kanfer, Sawyer, Earley, and Lind (1987) confirms the direction of causality. Kanfer et al. had student subjects work on a task that was evaluated using either fair or unfair procedures. The subjects rated their supervisor higher when the procedure was fair than when it was unfair. c . Political Authority. Tyler and Caine (1981) used both experimental and correlational methods to study the factors affecting evaluations of political leaders. In an experimental scenario study they manipulated the fairness of the procedure that a political leader used to make a decision. The scenarios also manipulated the favorability of the decision. The results showed that both procedural justice and outcome favorability influenced evaluations of the leader. In a second, correlational, study Tyler and Caine asked respondents about the fairness of the political decision-making process and about the favorability and fairness of political decisions. The survey results showed that procedural fairness judgments

TABLE I EXPERJENCES WITH SUPERVISOR: ATTITUDINAL REACTIONP Supervisor competence Outcome favorability Relative outcome favorability Relative procedural desirability Distributive justice Procedural justice Total R2

-.Ol

-.I1 .06 .06 .73** .56

Organizational commitment

Supervisor trust

Management evaluation

.04

.oo

.I3

.00

.04

.77**

.05 .I0 .39**

.08 .21*

.60

.22

.I4

- .06

.02

.oo

.oo

“N = 409. Entries are standardized regression coefficients for an equation including all five predictors. From Tyler and Schuller (1990). *p < .01. * * p < ,001.

I28

TOM R. TYLER AND E. ALLAN LIND

were significantly correlated with evaluations of political leaders, whereas outcome fairness and favorability showed smaller, and often nonsignificant, correlations with leader evaluations. vier, Rasinski, and McGraw (1985) studied the relationship between outcome favorability, distributive fairness, and procedural fairness and trust in the national government. They found that procedural justice showed a strong independent relationship to the trust measure (p =.61, p < .001; see Table II), with lesser relationships for measures of absolute outcomes (p =. 1 1 , n.s.) and distributive fairness (p = -.01, n.s.). TABLE I1 JUSTICEA N D THE ENDORSEMENT OF POLITICAL AUTHORITIES" Evaluations of Reagan

P Outcome level Absolute Relative Total Fairness Distributive Procedural Total Demographics Party Liberalism Age Education Race Sex Total Total

Trust in national government

P

R2

.09 - .04

R2

.I1

- .04

.07** *

.08*** -.01 .61***

.lI*

.47*** .45***

.34*** -.I0 - .03

.20*** .16***

-. O l

.oo

.oo

- .02

.06 .13**

-.05

.oo .30***

.oo

.54***

.34***

Usefulness analyses

Outcome beyond fairness and demographics Fairness beyond outcome and demographics Demographics beyond outcome and fairness

.oo

.01***

.19***

.28***

.09***

.oo

"Entries in p columns are standardized regression coefficients; entries in R2 columns are squared adjusted multiple correlation coefficients. From Tyler, Rasinski, and McGraw (1985). *p < .05. **p < . O l . ***p < ,001.

AUTHORITY IN GROUPS

129

2 . Legitimacy and Behavior Several studies show that the effects of procedural justice extend beyond attitudes to include compliance and related behaviors. a. Behavior in the Legal Arena. MacCoun et al. (1988) studied litigants in cases involving automobile claims in the New Jersey courts and found that judgments of procedural justice correlated with intentions to accept or reject the arbitrator's award (r = .40, p < .001), as did ratings of the favorability of the award (r = .30, p < .001). Lind, Kulik, Ambrose, and Park (1991), in a study that examined dispute resolution by authoritative decision in federal court, found that procedural justice judgments have strong effects on the acceptance of legal decisions. Lind et al. studied a court program in which arbitrators issued awards that the parties to the lawsuit were free to either accept or reject. Figure 3 shows the result of structural equation analysis of the influence of outcome favorability, relational process concerns (of the sort described later in this article), and procedural justice judgments on acceptance of the arbitration award. The values in the figure are standardized path coefficients. Two features of the final path model are important here. First, there was no direct path from outcome favorability to acceptance of the award; the only outcome effects were those mediated by procedural justice

Judamaar

Fig. 3. Effects of procedural justice on acceptance of arbitration awards. n.s., not significant. From Lind, Kulik, Ambrose, and Park (1991).

130

TOM R. TYLER AND E. ALLAN LIND

judgments. Second, the link between procedural justice judgments and acceptance of the award was very strong indeed, as shown by the large path coefficient. When procedures were viewed as unfair, 54% of the awards were accepted, but when procedures were viewed as fair 77% of the awards were accepted. The study is especially noteworthy because the cases in question involved amounts ranging from $5000 to $5,000,000 and because the decision to accept an arbitration award resulted in the transfer of tens, or even hundreds, of thousands of dollars. Tyler (1990) studied compliance with laws using a panel design in which respondents were interviewed at two points in time, 1 year apart. In one part of his study, Tyler examined the views of the respondents in the panel sample who reported having some experience with legal authorities during the interval between the two interviews (n = 291). The results of a causal analysis of these data are shown in Fig. 4. There are no significant paths from outcome favorability to evaluations of authorities or compliant behavior. This indicates that the outcome of experience did not directly affect evaluations of the authority or obedience to laws. The significant causal paths run from judgments of fairness to attitudes and behaviors. Procedural justice judgments affected views about the legitimacy of legal authorities. And legitimacy judgments, in turn, affected compliance with laws. In addition, distributive justice judgments directly affected compliance with laws.

compliance &lam,

Fig. 4. Panel analysis of experience, views about legal authority, and compliance. From S l e r ( 1990).

AUTHORITY IN GROUPS

131

6 . Behavior in Organizational Settings. Several studies show procedural justice effects on behaviors related to legitimacy and compliance in work settings. An experimental demonstration of the behavioral effects of unfair procedures in work settings is found in Greenberg (1989). In that study college students in a work experiment experienced either fair or unfair work evaluation procedures. Later, the participants were placed in a room that had a poster with telephone numbers for reporting unfair treatment to an “ethical responsibility board.” Greenberg found that experiencing an unfair procedure led participants to take slips of paper with the telephone number attached, suggesting an intention to report the unfair procedure to the ethics board. The study also showed a procedure by outcome interaction, with both an unfair procedure and an unfairly negative outcome necessary for protest behavior to occur. I ’ The behavior in question, a type of protest against the authority in question, is not compliance per se, but rather a behavioral indicator of perceptions of legitimacy, or more precisely perceptions of the absence of legitimacy. Tyler and Schuller ( 1990) also examined influences on organizational behaviors related to legitimacy (decision acceptance, rule following, turnover intention, and grievance filing). The behaviors in question were measured using selfreports in a general population survey of working people. Table 111 reports the results of regression analyses predicting the behaviors from ratings of outcome and procedural desirability and distributive and procedural justice in recent experiences with supervisors. As can be seen from the table, some of the behaviors showed significant links to outcome favorability and distributive justice, but procedural justice was the strongest and most consistent predictor of all four behaviors. Earley and Lind (1987) examined the influence of the fairness of task assignment procedures on people’s acceptance of task assignments in both a laboratory and a field setting; experimental methods were employed in both settings. Causal analyses of the results of the study showed that ratings of the acceptance of the task assignment decision were affected by procedural justice judgments in both settings. In both studies, either the procedural justice manipulation or measures of perceived procedural fairness were found to cause differences in performance that mirrored the findings for acceptance of the decision. Lind, Kanfer, and Earley (1990) report similar findings in a laboratory experiment on the acceptance of goal assignments in a simulated work setting: The manipulation of a variable-the opportunity to voice one’s views about the assigned goal-that is closely linked to procedural justice judgments resulted in higher rated acceptance of the goal and better performance. Unfortunately, in neither the Earley and Lind nor the Lind, Kanfer, and Earley study was the perceived legitimacy of the supervisor measured. Thus, although all three of the ”For a more detailed discussion of such interaction effects, see Lind and Qler (1988, pp. 67-68).

132

TOM R. TYLER AND E. ALLAN LIND TABLE I11 EXPERIENCES WITH SUPERVISOR: BEHAVIORAL REACTIONS~

Outcome favorability Relative outcome favorability Relative procedural desirability Distributive justice Procedural justice Total R2

Intention to sue organization

Follow rules

Intention to quit

.22** .12* .09

.I2 .07

- .02

.oo

.07

.01

.08 .37**

.23* .26* *

.03 - .09 -.28**

-.I1 - .02 - .20*

.48

.05

.I1

.03

Accept decision

-.13

"N = 409.Entries are standardized regresssion coefficients for an equation including all five predictors. From Sler and Schuller (1990). *p

< . O l . * * p < .WI.

studies establish rather clearly that there is a causal link between procedural justice and ratings of decision acceptance, they do not establish that perceived legitimacy mediated acceptance of the decision. However, taken together with the 51er and Schuller survey findings and the findings reported in the previous sections showing that judgments of procedural justice enhance the legitimacy of leaders, these three studies point to procedural fairness as a major factor in compliance with authorities. c. Political Behavior. Procedural justice judgments have also been found to influence behaviors associated with the acceptance of political authorities. Rasinski and Tyler ( 1 987) conducted two studies exploring the relationship between procedural justice judgments and political behavior in the 1986 presidential election. In both studies citizens' vote choices were correlated with judgments about the relative procedural fairness of the two presidential candidates. Judgments about past benefits and costs associated with government policies also correlated with vote choice. Rasinski (1988) examined the relationship between justice concerns and political behaviors, using a sample of 398 residents of Chicago. He studied both general political activity (e.g., working on a political campaign, displaying campaign buttons) and activism related to government benefits and services (e.g., contacting a political official about benefits or services, demonstrating for benefits or services). Both types of behavior showed substantial links to procedural justice judgments. The strongest relationship was found for political activity related to specific benefits government benefits and services (logistic coefficient = 0.42, p < .05). A weaker relationship was found for general political activity (logistic coefficient = 0.32, p < .lo). Political activity relating to specific benefits and services was also linked to outcome favorability concerns

AUTHORITY IN GROUPS

133

(coefficient= .025, p < . lo), but not to distributive fairness (coefficient = 0.22, n.s.). General political activity did not show substantial links to either outcome favorability (coefficient = 0.15, n.s.) or distributive fairness (coefficient = 0.43, n.s.). d. Disputing Behavior. The studies on litigants’ reactions to court procedures, which we reviewed above, show that procedural justice plays a considerable role in the decision to accept an arbitration judgment. There is evidence that procedural justice is also important for compliance in less formal dispute resolution settings. Pruitt, Peirce, McGillicuddy, Welton, and Castrianno (1990) also found that long-term compliance with mediation outcomes was linked to judgments of the fairness of the mediation procedure ( r = .32, p < .05). In fact, procedural justice was a more potent determinant of compliance than was the use of joint problem solving, a factor that has long been thought to be the cornerstone of successful mediation.

3. Summary and Implications The research described above shows that a key factor affecting legitimacy across a variety of settings is the person’s evaluation of the fairness of the procedures used by the authority in question. The studies that have included behavioral measures show much the same pattern of results with respect to measures of obedience and other behaviors linked to authority. Evaluations of both the favorability of the authority’s decision and its distributive fairness have occasionally been found to exert an effect on legitimacy independent of that exerted by procedural fairness. It is clear, though, that in terms of both the strength of effects and the ubiquity of effects, neither outcome factor is nearly as important as procedural justice in determining whether an authority is viewed as legitimate. l 2 Why is procedural justice so central to legitimacy? One explanation can be found in some of the original theorizing on the psychology of procedural justice. In their discussion of the problems of dispute resolution in groups, Thibaut and Walker (1975) noted the potential harm that the process of resolving disputes could do to social relationships within a group or society: poorly resolved disputes can threaten enduring relationships. They argued that the use of procedures regarded as fair by all parties facilitates the maintenance of positive relations among group members, preserving the “fabric of society,” even in the face of the ‘*As noted above, it is important to distinguish personal satisfaction from attitudes such as legitimacy. We are not saying that people are happy if they receive unfavorable outcomes through a fair procedure. We are saying, however, that they ( I ) are more likely to accept the decision and (2) are less likely to blame the authorities and/or institutions they have dealt with. As a consequence, they are more likely to follow the authority’s directives in the future. See Lind and Earley (1992) and Messick and Sentis (1983) for discussions of the distinction between satisfaction and fairness.

134

TOM R. TYLER AND E. ALLAN LIND

conff ict of interest that exists in any group whose members have different preference structures and different beliefs concerning how the group should manage its affairs (Thibaut & Walker, 1975, p. 67). A second reason for the preeminence of procedural justice concerns in judgments of the legitimacy of authority is found in a later analysis by Thibaut and Walker (1978). In many social situations it is not at all clear what decision or action is correct in an objective sense. Indeed, it could be argued that most group decisions concern questions for which, at least at the time the decision is made, there is no way of knowing what course of action will work out best. In these circumstances, Thibaut and Walker argue, what is critical to good decisionmaking is adherence to norms of fairness, and fairness is most evident when procedures that are accepted as just are used to generate the decision. In other words, absent objective indicators of the correctness of a decision, the best guarantee of decision quality is the use of good-which is to say fair-procedures. In terms used by Lind and Earley (1992; Lind et al., 1991), procedural fairness serves as a social heuristic that allows people to determine whether an action (e.g., obeying the decision of an authority) is correct without really weighing all the benefits and costs associated with the action. For an example of the difficulty of clearly establishing the “correct” solution to allocation problems in an organizational setting, consider the problems involved in using equity criteria to determine whether one is fairly compensated. According to equity theory, relevant inputs must be compared to relevant outcomes, and the resulting ratio must be compared across other members of the workplace group. But how can one compare and weight outcomes such as interestingness of work, office size and amenities, vacation options, parking spaces, pay, and the myriad other aspects of work life? Similar problems exist for comparing inputs across individuals. Imagine the difficulty of managing a psychological “equation” of this sort to determine the extent to which one’s outcomes are fair. A much simpler approach is to assume that fair outcomes result from fair procedures, and to look for signs that the procedure used to allocate a given outcome is fair or unfair. A related reason for the importance accorded procedural justice concerns in judgments of legitimacy has to do with links between perceptions of procedures and other cognitions about groups. In an earlier analysis (Lind & ’Qler, 1988) we have argued that procedures are widely viewed as essential elements of any lasting group and that perceptions of procedures are therefore key cognitions about groups. It is no accident, for example, that the drafting of a constitution is generally one of the first enterprises of any group that intends to function on a continuing basis. Through the design of its procedures a group gives itself form and makes manifest its values and goals. Because procedures are widely viewed as manifestations of group values, we argue, they take on enormous symbolic significance for other cognitions about groups. According to this line of thought,

AUTHORITY IN GROUPS

135

perceptions of procedures have greater impact on evaluations of groups than do perceptions of outcomes, because outcomes are generally viewed as one-time responses to particular situations whereas procedures have an enduring quality; this makes an unfair procedure much more threatening than a single unfair outcome. In much the same way, the procedures used by an authority to reach a decision might be seen as an expression of the authority’s values, and the judgment that an authority uses unfair procedures might be viewed with greater concern than the judgment that a particular decision is unfair. This explanation of procedural justice processes can be combined with the Thibaut and Walker and Lind and Earley arguments to explain why it is procedural fairness, rather than distributive fairness that is the critical element of the fairness heuristic and, thus, central to the decision to obey authority. Whatever the explanation, the results reviewed above suggest that the use of fair procedures is an important element, perhaps the key element, for the effective exercise of legitimate authority. If we want to understand the dynamics of authority, we must explore the psychology of procedural justice as it relates to authority. We turn to this task after a consideration of some of the societal implications of the findings described above. 4 . Authority from a Societal Perspective

Effectively exercising authority is a core issue in any organized group. Hence, the questions that we have been examining are always important. They become especially important in situations of resource scarcity, when social conflict is more likely and when allocation is more problematic. As we noted above, the empowerment of authorities seems to be a common response of groups to social conflict and the need to allocate scarce or diminishing resources, and yet it is precisely in such situations that the actions of authorities are most likely to be controversial. It is certainly the case that periods of social conflict and resource scarcity have prompted social psychologists to devote particular attention to problems of authority. In the period around World War 11, a central social concern in America was understanding how to manage ethnic, racial, and industrial conflict. Research at that time addressed the nature of rules and authority structures and the relation of authority structures to the interactions occurring among group members (Lewin et al., 1939) and between groups (Sherif, Harvey, White, Hood, & Sherif, 1961; Sherif & Sherif, 1953). A common theme in much of this research was that effectively designed social structures could dampen destructive social conflict, an argument that encouraged the development of substantial literatures on leadership and group dynamics (Cartwright & Zander, 1953). Another period during which concerns about scarcity emerged as important social issues was the 1970s. During this period, issues of scarcity and natural

136

TOM R. TYLER AND E. ALLAN LIND

resource management were highly salient, with many authors suggesting that the United States was moving into an new era of social conflict stimulated by resource scarcity and economic decline (Blumberg, 1980; decarufel, 1981; Deutsch, 1981; Greenberg, 1981; Lerner, 1981; Ophuls, 1977). Growing social conflict seemed likely to increase political and social instability and unrest. Examples of heightened conflict during this era include disputes concerning the allocation of scarce energy resources and debates over the use of nuclear power. The possibility of heightened social conflict was viewed as especially problematic for American government authorities during the 1970s because the United States was in a historic period during which it seemed particularly vulnerable to social unrest. Alienation from and distrust of legal, political, and industrial authorities was quite high (Caddell, 1979; Citrin, 1974; Lipset and Schneider, 1983; Miller, 1974, 1979, 1983; Wright, 1981). Because the legitimacy of authorities is often viewed as a “cushion of support” that helps societies to survive difficult periods in history (Easton, 1965), the weakness of the support underlying the American legal, political, and economic systems seemed to point to a potentially dangerous vulnerability to destructive social unrest. Responses in the social and behavioral sciences to the concerns about scarcity and conflict during the 1970s included the expansion of research on social dilemmas and procedural justice. The research on social dilemmas flowed directly from a concern with developing ways to manage exhaustible supplies of natural resources. Similarly, psychologists’ initial interest in procedural justice developed from a concern with conflict over resowes. As Thibaut and Walker noted at the very beginning of their book on procedural justice: One prediction that can be advanced with sure confidence is that human life on this planet faces a steady increase in the potential for interpersonal and intergroup conflict. The rising expectations of a continuously more numerous population in competition for control over rapidly diminishing resources create the conditions for an increasingly dangerous existence. It seems clear that the quality of future human life is likely to be importantly determined by the effectiveness with which disputes can be managed, moderated, or resolved. Procedures or methods that may be put to this task of conflict resolution therefore claim our attention. (Thibaut & Walker, 1975, p. 1)

We believe that the social issues that are likely to emerge as central concerns of the 1990s also demonstrate the importance of legitimacy and authority processes. The enormous changes occurring in the governments of Eastern Europe, the former Soviet Union, and East Germany illustrate dramatically the fragility of authority without legitimacy. The efforts to reconstruct the governments of these nations as well as the creation of new authorities in the European Economic Community illustrate the importance of understanding what makes a political authority legitimate. Efforts to discover how a nation-state can command the allegiance and obedience of various ethnic and regional constituencies, a prob-

AUTHORITY IN GROUPS

137

lem illustrated by ethnic conflicts in the United States, the former Soviet Union, the Middle East, Eastern Europe, and Canada, also point out how important it is to understand the psychological foundations of authority processes. In the United States many commentators expect that conflict over resource allocation and social values will be intensified as we become a more multiethnic, multicultural society. There is little doubt that the years ahead will see heightened conflicts over the balance between a “melting pot” model of American society, in which immigrants are expected to assimilate into existing culture, and a model in which immigrant groups preserve their unique cultures. Conflict of this sort has already emerged in both the United States and Canada in discussions of bilingualism. Conflicts over resource allocation will also be intensified by the efforts of traditionally disadvantaged groups-Blacks, Native Americans, and women-to secure policies that correct historical injustices. Perhaps the most dangerous conflicts of the 1990s are those that turn on the definition of legal rights, because rights cannot be easily compromised and therefore solutions to rights conflicts are particularly difficult to find. One of the most visible of these conflicts is that concerning abortion rights, but similar struggles are occurring over issues such as gay rights and the death penalty. As these conflicts are played out in the courts and in the electoral process, authorities must struggle to govern in the face of a substantial diversity of attitudes, beliefs, and values. Legitimacy becomes an especially important, but perhaps also an especially elusive, quality of government. Authority and legitimacy are issues of great importance in organizational settings as well. Supervisors often find themselves trapped between a desire not to “micromanage” and concerns that subordinates will not do what they are told. Close monitoring of subordinates’ actions entails substantial costs and, paradoxically, leads to even less confidence on the part of supervisors that their orders will be obeyed (Strickland, 1958). In addition, if obedience is obtained by the use of coercive power, resentment and private disobedience are likely to pose new problems for the organization. A better solution is the establishment of legitimacy on the part of supervisors in order to obtain voluntary compliance.

11. The Social Psychology of Procedural Justice

The research described above suggests that people evaluate legitimacy for the most part in procedural justice terms. Thus, to understand what drives legitimacy, we must understand what drives procedural justice judgments. In the following sections we review briefly existing theory and research on what makes procedures appear to be fair, devoting particular attention to those features that seem likely to affect procedural justice judgments in the context of interactions with authorities.

138

TOM R. TYLER AND E. ALLAN LIND

A. THEORIES OF PROCEDURAL JUSTICE In the 20 years since social psychologists began to study procedural justice judgments, a number of theories and models have been advanced to explain procedural justice phenomena. In reviewing these explanations, we divide them into two general categories. One category includes theories that view procedural justice judgments as determined largely by outcomes and by concerns external to the procedure and the interaction with authority. These theories, which we term instrumental models, view procedural justice judgments as founded ultimately on how well the procedure in question serves interests external to the experience of the procedure. The second category includes what we term relational models, which focus on relationship issues, especially perceptions of the relationship between the authority and those subject to his or her decision.

I . Instrumental Models of Procedural Justice Thibaut and Walker (1975, 1978) advanced a control-oriented model of what drives judgments about the fairness of dispute resolution procedures. The Thibaut and Walker analysis argues that the key procedural characteristic shaping people’s views about the fairness of the procedures is the distribution of control between disputants and the third-party decision-maker. Thibaut and Walker distinguish between two types of control: process control and decision control. Process control refers to the extent and nature of a disputant’s control over the presentation of evidence. Decision control refers to the extent and nature of a disputant’s control over the actual decisions made. Thibaut and Walker assume that disputants are primarily concerned with the problem or dispute that brings them to third parties. Judgments of the fairness of various dispute resolution procedures are based on instrumental concerns in the sense that disputants are thought to view procedures as means to the end of improving either their own outcomes or their relationship with the person with whom they have a dispute. Thibaut and Walker do not devote much attention to disputants’ concerns about their long-term relationship with the authority who serves as a third party in the dispute resolution procedure. Implicit in the Thibaut and Walker analysis is the assumption that by and large disputants view their experience with the judge and the court system as a one-shot encounter. These assumptions manifest themselves in the view that disputants are most concerned with control as they assess the fairness of procedures. According to Thibaut and Walker, disputants want control-either process control or decision control-because they see control as instrumental to attaining the fair outcomes they desire. Process control is important to disputants, according to this line of thought, because it assures that each party to the dispute can see to it that the third party receives information on contributions and outcomes, and this in turn

AUTHORITY IN GROUPS

139

allows the third party to use equity rules to resolve the dispute fairly. Decision control allows the disputant to accept only those decisions that he or she regards as desirable. (Folger [I9771 introduced the term voice to describe procedural qualities very similar to those termed process control by Thibaut and Walker.) More recently, other procedural justice theorists have offered additional control-oriented models. Brett (1986), for example, argues that decision control is one of the most important elements of a fair dispute resolution procedure because people wish to maintain a feeling of control over what will happen to them once the dispute is resolved. Vidmar (1990) and Leventhal (1976) view people as motivated primarily by the desire to obtain favorable outcomes. According to both of these theorists, decision control is valued because it allows unfavorable outcomes to be rejected, and process control is valued because it allows one to persuade the third party to rule in one's favor. Some theories of procedural justice go beyond perceptions of control and discuss other factors that might drive procedural justice judgments but that are nonetheless instrumental, because procedures are viewed as means to achieve goals outside the encounter with the procedure. For example, Leventhal (1980) identified six potentially important procedural justice rules in outcome allocation contexts: consistency, bias suppression, accuracy, correctability, representativeness, and ethicality. The consistency rule says that allocation procedures should be applied consistently across people and over time. The bias suppression rule says that personal self-interest and blind allegiance to narrow preconceptions should be prevented. The accuracy rule says that decisions must be based as much as possible on good information and informed opinion. The correctability rule says that opportunities must exist to modify and reverse decisions. The representativeness rule says that the allocation process must represent the concerns of all important subgroups and individuals. Finally, the ethicality rule says that the allocation process must be compatible with prevailing moral and ethical standards. These six rules are important, according to Leventhal, because each increases the likelihood that the final allocation will advance group goals. The one common theme in all instrumental analyses is that procedures are ultimately judged not in terms of their own qualities but rather in terms of their implications for outcomes or for external relationships. In the final analysis, according to instrumental models of procedural justice, it is not anything about the social process mandated by a procedure that makes it fair. What matters is whether the procedure serves some external goal.

2 . A Relational Model of Procedural Justice Not all psychologists have accepted the proposition that concerns about justice flow from instrumental motives. We have proposed a group-value model of procedural justice (Lind & vier, 1988) that suggests that procedures are evalu-

140

TOM R. TYLER AND E. ALLAN LIND

ated in terms of their implications for group values and for what they seem to say about how one is viewed by the group using the procedures. The group-value model hypothesizes that in reacting to procedures people are primarily concerned about their long-term social relationship to the authorities or institutions that employ the procedures (Lind & Earley, in press; Lind & Qler, 1988; Qler, 1989; Qler & Schuller, 1990). Procedural justice depends, we argue, on the implications of the procedure for feelings of self-worth and for the belief that the group is functioning properly and fairly. l 3 The group-value model asserts that people are sensitive to procedural nuances because procedures are viewed as manifestations of basic process values in the group, organization, or institution using the procedure. Because procedures acquire substantial significance as symbols of group values, individuals are concerned as much or more with what happens within the procedure as with whether the procedure promotes the attainment of some extraprocedural goal. Of key importance, according to this analysis, are the implications of the procedure for one's relationship with the group or authority that enacts the procedure. To the extent that a procedure is seen as indicating a positive, full-status relationship, it is judged to be fair, and to the extent that a procedure appears to imply that one's relationship with the authority or institution is negative or that one occupies a low-status position, the procedure is viewed as unfair. Because the group-value model predicts that people consider a procedure's implications for their relationship with authorities as they form procedural justice judgments, it is reasonable to ask which particular aspects of the authorityperceiver relationship might be most important for perceptions of procedural justice. Basing his analyses on group-value theory (Lind & Qler, 1988, pp. 221242), Qler (1989) has identified three factors-standing, neutrality, and trustthat are especially important for the judgment that procedures are fair. All three factors have to do with the relationship of the perceiver to the authorities using the procedure. l4 A basic tenet of the group-value model is that people are predisposed to belong to social groups and that they are very attentive to signs and symbols that W ~ e r is e an alternative to either an instrumental or a relational model of procedural justice. One might posit that concerns about justice are intrinsic. Lemer (1982), for example, suggests that "deservingness and justice" are independent motives underlying social behavior. Lerner (1980) has demonstrated the existence of such basic motives by showing that they lead people to distort their judgments about the world. Lerner (1980, 1981) talks about justice as an intrinsic feature of our reactions to others. 14The three relational factors-trust. standing, and neutrality-are theoretically distinct, but we make no claim that they are phenomenologically distinct. That is, it makes sense to distinguish among the three factors in theoretical discussions, but it may well be the case that most people simply lump all of these concerns together as they assess the overall quality of their relationship with an authority.

AUTHORITY IN GROUPS

141

communicate information about their status within their groups. 15 People want to understand, maintain, and improve the social bonds that exist within groups. Research shows that people establish social connections if given even the most tenuous basis for group identification (Brewer & Kramer, 1986; Kramer & Brewer, 1984; Messick & Brewer, 1983; Tajfel, 1981; Tajfel & Turner, 1979; Turner, 1987). One reason that people seek group membership is that groups provide a source of self-validation (Festinger, 1954), giving their members information about the propriety of their behavior, attitudes, and values. People are concerned about their status in groups because high status within a valued group validates their own self-identity. Hence, an important prediction of the groupvalue model is that people value procedural indications that they have positive stunding within valued groups. Information about standing-i.e., status recognition-is often communicated to people by the interpersonal quality of their treatment by those in a position of authority. In particular, when one is treated politely and with dignity and when respect is shown for one’s rights and opinions, feelings of positive social standing are enhanced (Bies & Moag, 1986; Lind, MacCoun et al., 1990; Lind & Qler, 1988; Tyler & Bies, 1990). On the other hand, undignified, disrespectful, or impolite treatment by an authority carries the implication that one is not a full member of the group, and this is very threatening indeed (MacCoun et a]., 1988, p. 54). Lind, MacCoun et al. (1990) refer to these issues as dignitary concerns and to the quality of procedures that meet these concerns as dignitary process. Another feature-respect for one’s rights-is also hypothesized to be a particularly potent indicator of one’s standing in a group or society. Group-value theory assumes that people base their self-identity on their membership in valued groups. People find evidence that they are accepted members of social groups rewarding and are troubled by evidence that they may be rejected or excluded from the group (Cartwright & Zander, 1953; Schachter, 1951). With these concerns in mind, vier (1989) suggests that two other types of relational concerns affect procedural justice judgments: the neurrulify of the decisionmaker and trust in the third party. When considering whether an authority is displaying neutrality, people focus on whether the third party creates a “level playing field” by engaging in evenhanded treatment of all involved. According to Tyler, neutrality involves honesty and lack of bias. Neutral decision-making also uses facts, not opinions, in an effort to produce decisions of objectively high quality. Our analysis of the importance of judgments of neutrality for procedural justice differs from more instrumental analyses, such as that offered by Leventhal (1980), that also emphasize ‘5By srurus we mean the person’s perception that he or she is valued by the group. We do not mean simply wealth or prestige, but rather something more like approbation and acceptance by the group as a full member with all rights associated with membership.

142

TOM R. TYLER AND E. ALLAN LIND

neutrality. We view the perception of neutrality of the authority as a concern in and of itself, because neutrality is a critical quality of the person's relationship with the authority, whereas instrumental analyses view neutrality as a quality that is desired only because it leads to other goals.16 Perhaps the most potent manifestation of neutrality concerns is seen in the belief that one has been discriminated against, that the authority is prejudiced against members of one's subgroup. Quite apart from the suggestion that one will receive less benefit from group membership, prejudice implies that one is somehow less worthy than those who receive more favored treatment, and this implication has extremely powerful consequences for feelings of self-worth. Tyler's third relational factor is the extent to which the authority is seen as trustworthy, i.e., whether the person believes that the authority can be trusted to try to behave fairly. Trust involves beliefs about the intentions of the authority. As is the case with standing and neutrality, the key element in concerns about trust has to do with qualities of the authority and the perceiver-authority relationship, not with the effects of the authority's decision on factors external to the procedural experience. Measures of trustworthiness generally include ratings of the extent to which the authority is trying to be fair and ratings of whether the authority is ethical. Tyler argues that inferences of trustworthiness are especially important because they have implications for the future behavior of that authority and the behavior of other, similar, authorities, and these implications in turn suggest what the perceiver can expect across the whole range and duration of his or her interactions with the group. The relational model shares some elements with other social psychological theories. Like the social identity and self-categorization models of Tajfel and 16This difference in emphasis leads to different predictions. As explained below, if the instrumental theorists are correct, neutrality should be a less powerful predictor of procedural justice judgments than are external goals of the encounter with the authority, such as a fair distribution or maintenance of good relations with other recipients of the distribution. According to our analysis, on the other hand, external considerations should play a less important role than neutrality in determining whether a procedure seems fair. "The distinction between the behavior of an authority (e.g., that the behavior is neutral or ethical) and inferences about the authority's motives (e.g., that the authority is trying to be fair) is important. This distinction underlies, of course, attribution theories (Heider, 1958), which recognize that people often try to use behavior to infer hidden, unobservable character dispositions. On a practical level, recognition of the importance of inferences about motives can be seen in people's comments about their experiences when dealing with police officers and judges. In preliminary interviews concerning encounters with police and judges, Qler (1990) found that people often commented that the legal authority in question had been wrong or dishonest, had shown bias, or had behaved improperly. Nonetheless, the interviewees sometimes expressed satisfaction with the encounter and said they had been treated fairly. Why? Because they believed that the authorities were motivated by a desire to treat them fairly. It appeared that the inferred motivation could outweigh the behavior itself in determining how the encounter was evaluated.

AUTHORITY IN GROUPS

143

Turner (Tajfel, 1981; Tajfel & Timer, 1979; Turner, 1987), our model begins with the premise that people are interested in learning about their own social identity through their groups. Unlike Tajfel and lbrner, though, we focus on the search for information about one’s position within one’s group, rather than on the position of one’s ingroup vis-A-vis other groups. The reason that people are so interested in trust, standing, and neutrality, we would argue, is that they believe that these factors tell them something about whether they will be treated fairly and whether they are valued by the group using the procedure. This difference in focus between our model and the Tajfel and Turner model is perhaps a natural result of the types of groups each of us has studied. Tajfel and Turner have used minimal laboratory groups to study the relationship between self-identity and group membership, whereas we have focused our research on hierarchical organizations. 3. Implications of the Models

Because both types of models outlined above, instrumental and relational, are attempting to explain the basic processes that underlie procedural justice phenomena, they often overlap with respect to predicted relationships among variables, although they differ in how they explain these relationships and in their hypotheses concerning the relative strength of various effects. Consider, for example, five variables that have often been found to affect procedural justice judgments: outcome favorability, control, neutrality, trust, and standing. Instrumental theories predict that outcome favorability and control are powerful determinants of whether a procedure is seen as fair, because favorable decisions, or at least control over decisions, are precisely what people want from procedures. Instrumental theories predict weaker effects for neutrality, trust, and standing because these three variables have to do with relationships within the procedure and are therefore important only insofar as they suggest that the procedure can be used to achieve external goals. But instrumental theories can account for indirect causal links between relational factors and procedural justice: Symbols of positive standing can imply high status within a group, and high-status members typically have better outcomes; trust implies that one might at least receive some fair outcomes; and neutrality implies that bias will not prevent one from receiving favorable outcomes. In contrast, relational models predict that factors that have to do with relationships within the procedure-judgments of neutrality, trust, and dignityhave stronger effects than do factors that have to do with concerns outside the procedure. Relational models can explain control and outcome favorability effects, however. Feelings of control can both result from and contribute to perceptions that one has standing within the group (Lind & Qler, 1988, pp. 236-237; see also Lind et al., 1990). Similarly, favorable outcomes have informational, as

144

TOM R. TYLER AND E. ALLAN LIND

well as hedonic, value: A favorable decision can be taken as a sign of personal worth and favor on the part of the powers that be and therefore can lead to greater trust and greater perceived standing (Conlon, Lind & Lissak, 1989). Factors that seem at first glance to be purely instrumental might nonetheless affect procedural justice judgments by virtue of their relational implications. Thus neither class of models predicts many effects that cannot be explained at all by the other. In addition, it may well be the case that procedural justice depends on both instrumental and relational considerations (see, e.g., Lind et al., 1990). The critical question is really how much each set of considerations contributes to procedural justice and through procedural justice to compliance with authority. To the extent, then, that the research literature supports one model of procedural justice over the other, the support must turn on questions of the relative magnitude of effects. If relational concerns account for the bulk of the variation in procedural justice judgments, and control and outcome concerns account for less variance, this suggests that the field should give less weight to instrumental models and more weight to relational models of justice and authority.

B. RESEARCH ON PROCEDURAL JUSTICE We turn now to a consideration of some research on the factors that affect judgments of procedural justice.

I . Research on Control Issues Much of the procedural justice research conducted since the publication of the Thibaut and Walker book (1975) has focused on the effects of various procedures on the distribution of process control and the effects of process control on judgments of procedural fairness (see Lind & Qler, 1988, for a review). A large number of studies support the Thibaut and Walker finding that the distribution of process control influences assessments of procedural justice: Vesting process control (or voice, to use Folger’s, 1977, term) in the hands of those affected by a decision leads to higher procedural fairness judgments (see, e.g., Folger, 1977; Kanfer et al., 1987; LaTour, 1978; Lind et al., 1980; Lind, Lissak, & Conlon, 1983; Vler, 1987; Tyler, Rasinski, & Spodick, 1985; Walker, LaTour, Lind, & Thibaut, 1974). Indeed, the process control effect is probably the most widely replicated finding in procedural justice research. In that respect, Thibaut and Walker’s control theory has been widely confirmed. A description of only one of the many studies confirming the process control effect is sufficient to convey the nature of the effect. Kanfer et al. (1987) had subjects in a laboratory experiment perform a task that involved generating a list

145

AUTHORITY IN GROUPS

of innovative names for some common household products. The subjects were told that the product names would be evaluated by a supervisor and that their payment for the experiment would depend on how high a rating they received from the supervisor. In the high-process-control condition, the subjects submitted to the supervisor not only the product names, but also a written explanation of why they felt the names were good. In the low-process-control condition, only the names were given to the supervisor. The process control manipulation was crossed with a manipulation of the outcome of the evaluation process: The subjects were told that they had received either a high or a low evaluation. Table IV shows the mean ratings of procedural justice and the mean ratings of the supervisor, averaging across the two outcome conditions. The high-processcontrol procedure resulted in higher ratings of procedural justice and in higher ratings of the supervisor. As research on process control has accumulated, a number of findings have emerged that cast some doubt on whether process control effects can be explained in the sort of instrumental terms used by Thibaut and Walker (1 975) and other instrumental theorists. In their original analysis of control, Thibaut and Walker placed their primary emphasis on decision control, viewing process control as an indirect means of exerting decision control. According to this analysis, process control should exert less powerful effects on procedural justice judgments than does decision control, because decision control allows one to act directly to assure fair outcomes whereas process control allows only indirect attempts to persuade the decision-maker to arrive at a fair decision. Contrary to this prediction, subsequent research has shown that process control is frequently more important to people than decision control (Lind et al., 1983; Tyler, 1987; vier, Rasinski, & Spodick, 1985). In addition, a number of studies have shown process control effects that defy instrumental explanations of the effect. Lind et al. (1983) tested the relative effects of process and decision control on procedural fairness judgments in a laboratory experiment on dispute resolution procedures. The decision control manipulation involved the presence or absence of an opportunity to reject the third-party decision on how the dispute should be resolved; the process control manipulation involved having or not having an advocate to argue one's position to the third party prior to the decision. Thus, TABLE IV PROCESSCONTROL, PROCEDURAL JUSTICE, A N D EVALUATIONS OF SUPERVISORSO Measure Procedural justice judgments Evaluation of suoervisor

High process control

Low process control

6.30 6.10

5.00

~

'Values are mean ratings on %point scales. From Kanfer. Sawyer, Earley, and Lind (1987)

5.20

146

TOM R. TYLER AND E. ALLAN LIND

decision control was a more direct form of control over outcomes than was process control, and if the instrumental view of the process control effect were correct, decision control should have had greater effects on judgments of procedural fairness than did process control. Indeed, one might expect that process control would have little effect on procedural fairness judgments when disputants have decision control. l 8 The results of the Lind et al. (1983) study show that the decision control manipulation had little effect on procedural fairness judgments (see Table V). Process control enhanced disputants’ perceptions of the fairness of the procedure, and it did so whether the disputant had decision control or not. These findings are contrary to what one would expect if outcome instrumentality were the major force underlying the process control effect; instead, the results suggest that there is something about process control in and of itself that disputants like. The findings run contrary to the instrumental idea that process control is simply a poor surrogate for decision control. q l e r , Rasinski, and Spodick (1985) used both survey and laboratory data to test whether process control is related to procedural justice judgments in a fashion that cannot be explained by its indirect effect on decision control. Data from two survey studies, one involving defendant reactions to trial experiences and the other involving student reactions to college courses, were subjected to hierarchical regression analyses to determine whether judgments about process control correlated with procedural fairness judgments over and above any effect of judgments about decision control. In both surveys an independent effect was found for process control. In a third study, Tyler et al. manipulated the level of process control, the level of decision control, and the allocation setting described in scenarios presented to their subjects. Again process control was found to have independent effects on ratings of procedural fairness. In yet another set of survey studies, Tyler (1987) replicated these findings. Even stronger evidence that process control effects might involve something other than instrumental considerations is provided by an experiment by Musante, Gilbert, and Thibaut (1983). Musante et al. found that procedural justice judgments are enhanced by an opportunity to express one’s views and opinions, even when the expression of views is clearly not instrumental to obtaining favorable outcomes. Musante et al. examined the effects of various types of participation in the design of procedures on judgments of the procedures’ fairness. Participants were assembled into six-person groups. Some groups were asked to discuss and decide on the rules to be used in a later adjudication that would affect the participants’ outcomes. Other groups were told to discuss their preference for the ‘*In addition, the study included a manipulation of the outcome of the hearing: Participants were told either that they had been awarded two thirds of the outcome in controversy or that they had been awarded only one third of the outcome in controversy. The results presented in the table are averaged across the two outcome conditions.

147

AUTHORITY IN GROUPS TABLE V PROCESS CONTROL, DECISION CONTROL, AND PROCEDURAL JUSTICE^

High decision control Low decision control

High process control

Low process control

6.32 6.17

5.09 5.35

“Values are mean ratings on 9-poinI scales. From Lind, Lissak, and Conlon (1983).

various rules, but these groups were led to believe that their preferences would have no effect on the rules actually used-participants in these groups were told that the rules to be used had already been selected. In half of these discussiononly groups the rules actually used matched the preferences expressed in the group discussion; in the other groups in the discussion-only condition the rules actually used did not match the group’s preferences. Finally, a set of control groups were neither allowed to discuss nor to decide the rules to be used. Table VI shows the results of the Musante et al. study. Participation in both group discussion and decision-making led to the greatest enhancement of the procedural fairness judgments, but group discussion alone also enhanced procedural fairness relative to the condition that involved neither discussion nor decision-making. Lind et al. (1990) provide an even stronger demonstration of noninstrumental process control effects. Lind et al. (1990) allowed participants in laboratory work groups to voice their opinions about work goals either before or after the goal had been assigned. In a “postdecision voice” condition, subjects were told that the experimenter was interested in hearing what they had to say, but that the goal had already been set and would not be changed. To emphasize the noninstrumental nature of the process control, subjects in the postdecision voice condition were told what the goal would be before they offered their comments and the same goal was announced again after they had made their comments. The results of the experiment, shown in Table VII, show that even postdecision voice enhances judgments of the fairness of the procedures, as well as acceptance of the goal and performance, relative to a no-discussion control group. These findings support the hypothesis advanced by Tyler, Rasinski, and Spodick (1985) that some process control effects may be due to favorable reactions to the opportunity for expression rather than to any instrumentality of process control in achieving good outcomes. In both the Musante et al. study and the Lind et al. study an enhancement of fairness was observed in conditions that are purely expressive, conditions in which there was no suggestion whatsoever of control or influence over anything external to the procedure. Other research, which has investigated the conditions in which the process control effect does not occur, shows that at least part of the effect may turn on perceptions of one’s relationship with the decision-making authority involved in the procedure. Spe-

148

TOM

R. TYLER AND E. ALLAN LIND

TABLE VI CONTROL,DISCUSSION, AND PROCEDURAL JUSTICE^

Procedural justice ratings

No controll no discussion

No control/ discussion ( preference mismatched)

No controll discussion ( preference matched)

Control and discussion

-0.99

-0.56

0.28

1.28

PHigher values indicate greater procedural justice. From Musante. Gilbert. and Thibaut (1983).

cifically, several studies have shown that it is important that the decision-maker be seen as giving due consideration to the views expressed, even if he or she is not influenced by those views (Conlon et al., 1989; Folger, 1977; Folger, Rosenfield, Grove, & Corkran, 1979; Qler, 1987). In two experimental studies, Folger and his colleagues have shown limitations to the process control effect that suggest that the effect does not occur when the basic fairness of the authority is very clear, or when the basic unfairness of the authority is clear. Folger (1977) found that when the authority's interests are opposed to those of the subject, and when the authority nonetheless makes allocations that seem to be moving toward favoring the subject, the process control effect disappears. In this experiment the subjects in the improving allocation conditions may well have decided that the allocator was trying to be fair, and they may have been willing to trust the allocator even when they had no process control: Folger et al. (1979) gave subjects independent information supporting the inference that the authority was acting unfairly, and found that this too diminished or obliterated the process control effect. In the Folger et al. study the subjects may have concluded that the allocator could not be trusted to consider their views under either process control condition and this belief could have undermined the process control effect. TABLE VII PROCEDURAL JUSTICEA N D TIMINGOF INPUP Measure Procedural justice Acceptance of goal Task performance

Mute (no voice)

Postdecision voice

Predecision voice

2.43 4.33 4.67

3.15 5.67 5.85

3.70 5.93 6.58

"Values for procedural justice are means on a 5-point scale; values for goal acceptance m means on IGpoint scale; values for performance are number of units completed. From Lind, Kanfer. and Earley (1990).

AUTHORITY IN GROUPS

149

Tyler (1987) used cross-sectional survey data to test four possible limitations on the process control enhancement of procedural justice judgments. Tyler tested whether the process control effect disappears when ( I ) the decision-maker is seen as biased, (2) when the decision maker is seen as not acting in “good faith,” (3) when the decision maker is seen as not giving the disputant’s views due consideration, and (4) when the outcomes involved were important. Each limitation was tested by selecting individuals who did not believe they had any decision control and who believed that the limitation in question was operative in their case. Within each subset of individuals, the procedural fairness judgments of those who reported high and low process control were compared. For example, to examine the first limitation, Tyler tested the process control effect using only survey respondents who believed the decision-maker was biased. The survey showed that only the third limitation, the lack of due consideration, diminished the process control effect. Perceptions of process control were associated with enhanced procedural justice even when the decision-maker was seen as biased, when he or she was seen as acting in bad faith, and when the outcomes were important, but not when the decision-maker was seen as not giving consideration to the respondent’s views and arguments. A laboratory experiment by Conlon et al. (1989) also demonstrated the importance of belief that the authority is really considering one’s views. Conlon et al. found that subjects who received unfavorable outcomes but who believed their position had been considered give higher procedural justice ratings than do subjects who received better outcomes but who believed that their position had not been considered. In summary, then, there is considerable evidence that the process control effect involves something beyond instrumental concerns. It appears that the opportunity to express one’s opinions and arguments, the chance to tell one’s own side of the story, is a potent factor in enhancing the experience of procedural justice, even when the opportunity for expression really accomplishes nothing outside the procedural relationship. Indeed, it is now clear that the opportunity to state one’s case may be valued as much or more for its own sake as for its instrumental influence on decisions. In addition, where limitations on the process control effect have been found, the limitations occur when there are reasons to suspect that the exercise of process control is thwarted by some more potent feature of the perceiver’s relationship with the authority.

2 . Research on Noncontrol Issues The group-value model of procedural justice suggests that people are concerned about their long-term social relationship to groups and group authorities (Lind & Tyler, 1988). As already noted, the group-value model and Tyler’s later analyses of the potential procedural justice effects of relational concerns predict that people will be concerned about issues of standing, neutrality, and trust.

150

TOM R. TYLER AND E. ALLAN LIND

51er has conducted two studies exploring relational concerns and their effects on procedural justice in legal and managerial settings. In one study, Qler (1988, 1989, 1990) examined reactions to encounters with legal authorities, using interviews with people who had had recent personal experiences with police officers or judges (n = 652). In another study, Tyler and Schuller (1990) examined people’s evaluations of their personal experiences with managers in their workplace, using a sample of 409 workers who were asked about recent experiences with their supervisors. When interviewed about their experience, respondents in both studies were asked to make a series of judgments about various elements of the procedures used and various characteristics of the authority they had encountered. The procedural elements studied corresponded to factors identified by Thibaut and Walker (1975; process control, decision control), Lind and vier (1988; neutrality, trust, standing), and Leventhal (1980; outcome favorability, consistency, correctability). Judgments on these factors were then used as independent variables in regression equations predicting judgments of procedural justice. In addition, each of the factors was used to predict either ratings of legitimacy, in the case of legal authorities, or a surrogate measure-organizational commitmentin the case of managerial authorities. Table VIII shows the results of the regression analyses. Several aspects of procedures make substantial independent contributions to judgments of procedural fairness. Significant regression weights are seen for outcome favorability, control (process and decision control), neutrality, trust, and standing. In both settings, however, the strongest determinants appear to be the relational measures. In the legal arena standing and trust are most important, while in the managerial arena neutrality and trust are the most important. In both settings TABLE VIII ATTRIBUTES ASSOCIATED WITH PROCEDURAL FAIRNESSU Attribute Control Outcome consistent with expectations Correctability Neutrality Trust Standing Outcome favorability A*

Legal authorities

Managerial authorities

.20**

.I o*

.oo .02 .13** .34** .29** .14**

.68

.03 - .01 .21* .41** .11**

.lI**

.I5

‘‘Entries are standardized regression coefficients for an equation with all seven predictors. Control = process and decision control; neutrality = honesty, competence. and lack of bias. From Qler (1988);Tyler and Schuller (1990). ‘ p < .01. *‘p < ,001.

151

AUTHORITY IN GROUPS

outcome favorability and control show some independent relation to procedural justice, but neither appears to account for a very large proportion of the variance in procedural fairness. It is possible to use the data from the Tyler (1988) and Tyler and Schuller (1990) studies to assess more directly the explanatory power of the instrumental and relational models. A series of regression equations were used to compare the capacity of the factors specified by a given model to explain variance in procedural justice judgments that cannot be explained by the factors in the other model, or by other judgments; we refer to this as the unique variance explained by each set of variables. Table IX shows the degree to which each set of factors can explain unique variance in procedural justice and legitimacy. The equations compared relational factors to outcome favorability-the most instrumental of instrumental factors-and to perceived control-a factor that previous research and theory suggests might have both instrumental and relational connotations (e.g., Lind et al., 1990; Lind & Tyler, 1988, pp. 236-237). The results of this analysis indicate that the relational variables dominate judgments about the fairness of procedures. As would be expected from research showing how important procedural justice is to legitimacy, these same relational factors also dominate judgments of the legitimacy of the authorities. In the case of procedural justice judgments concerning experiences with legal authorities, the relational factors explain 24% of the variance that cannot be explained by judgments of outcome favorability and control, whereas outcome favorability or control explains only

TABLE IX PREDICTORS OF PROCEDURAL JUSTICE

AND

LEGITIMACY~

Percentage of variance Procedural justice

Legal authorities Outcome Control Relational factors Total Managerial authorities Outcome Control Relational factors Total

Legitimacy

Alone

Unique

Atone

Unique

23 38

I I 24

0 3 6 6

0 I 3

1 1

I1 9 16

2 0 6

67 69

31

46 73 75

25

18

'Entries are percentages of variance accounted for by each set of factors. From Sler (1988); Sler and Schuller (1990).

152

TOM R. TYLER AND E. ALLAN LIND

about 1% of the variance not explained by relational factors. Findings in the managerial arena are similar: Relational factors explain 25% of unique variance in judgments of procedural justice, and outcome favorability and control explain 0-2%.

An issue that arises with respect to findings like those of the n l e r (1988) and 51er and Schuller (1990) studies concerns the effects of outcome favorability on procedural justice judgments, what Messick and Sentis (1983) refer to as egocentric biases in fairness (see also Vidmar, 1990). It could be argued that at least part of what drives procedural fairness judgments is simply an attempt to legitimize the desire for good outcomes. Thus, the link between compliance and proceduraljustice, which we documented in the first section of this article, might in reality simply reflect outcome effects on both procedural justice and compliance. According to this line of argument, people want to conform to favorable decisions, but they also want to appear to be acting fairly, rather than hedonically. They therefore accentuate the fairness of the procedure that gave them a good outcome and use the enhanced fairness to justify accepting an outcome that they want. The Tyler (1988) and vier and Schuller (1990) findings offer some information on this possibility; their analyses suggest that the link between outcome and procedural justice exists but that it is not very strong. The same conclusions are suggested by the Lind et a]. (1991) study mentioned in our discussion above of the effects of procedural justice on the acceptance of arbitration awards. Figure 3 shows the results of a structural equation analysis of a causal model with paths linking relational concerns and outcome favorability to procedural justice. The relational concerns measured in the study tapped whether there was adequate opportunity to have the case heard by an impartial third party and whether factbased, rather then biased, procedures were used-perceptions relating to neutrality, trust, and standing. As can be seen from the figure, outcome favorability did affect procedural justice judgments, and through its effect on procedural justice outcome favorability had an indirect effect on the acceptance of arbitration awards. Note, however, that the effect of relational concerns was more than twice that of outcome favorability. The Lind et al. study also tested whether the stakes in the case or the litigant’s involvement as an individual or as a corporate actor affected the magnitude of the effects of relational concerns and outcome favorability on procedural justice judgments. The results showed that the impact of relational concerns on procedural justice was as strong for high-stakes cases as for lower stakes cases and as strong for corporate executives as for people in court as private individuals. The Lind et al. study also showed that the outcome and relational effects were unaltered by the stakes in the case or by the litigant’s role in case: there was no evidence that outcome was a more potent consideration in high-stakes cases or for corporate litigants. The experimental literature also offers some conclusions with respect to ego-

AUTHORITY IN GROUPS

153

centric biases in procedural justice (see Lind & Tyler, 1988, pp. 110-112), although it is difficult to compare the strength of outcome and relational effects in experimental studies. In many experiments, outcomes have been found to affect procedural justice judgments in ways that suggest an egocentric bias, but these effects seldom obscure the effects of nonoutcome variables. A reasonable conclusion from all the studies comparing outcome and relational effects is that, although outcomes and relational factors both affect procedural justice judgments, relational factors are the more powerful determinants of procedural justice judgments. We turn now to additional research on each of the three relational factors. a. Srunding. Standing-concern about recognition of one’s status and membership in a valued group, organization, or society-leads people to examine their encounters with authorities for evidence that they are being treated with politeness and dignity and that their person and their rights are being respected. A number of studies confirm the importance attached to issues of politeness, dignity, and respect in assessments of procedural justice. Qler and Folger (1980) examined citizen-police contacts and found that one key concern of citizens in such contacts was the “recognition of citizen rights’’ (p. 292). In two studies, Bies (1985; Bies & Moag, 1986) asked MBA job candidates about what made a job interview fair or unfair. In both studies he found that being dealt with respectfully was one of the most frequently cited criteria of fairness. Another example of the positive effect of rights-affirming behavior is provided by a study of people’s reactions to police responses to home burglaries (Parks, 1976). In most burglaries there is little likelihood that the police will recover the stolen property. Nonetheless, if the police come to the home, take the problem seriously, and write out a report, evaluations of the police do not decline following a burglary. Apparently, by reaffirming people’s right to call on the authorities, the police create a positive experience even when they cannot provide the desired outcome (recovering the stolen goods). Further evidence of the importance attached to issues of standing is provided by the results of two recent studies of the civil justice system (Lind, MacCoun et al., 1989, 1990; MacCoun et al., 1988). Tables X and XI show some of the findings of these studies. The Lind, MacCoun et al. study examined the reactions of litigants in personal injury cases to their litigation experiences. The litigants’ ratings of procedural justice were found to be rather strongly correlated with their judgments of the dignity of a procedure (r = .44).Measures of procedural desirability-the cost and delay associated with the procedure-showed little correlation with procedural justice judgments, nor was there any significant correlation between procedural justice and the amount of money won or lost in the case. In addition, Lind, MacCoun et al. found that litigants thought that adjudicatory procedures such as trial or arbitration were fairer than settlementoriented procedures such as negotiation or mediation. Structural equation analy-

154

TOM R. TYLER AND E. ALLAN LIND

CORRELATES OF

TABLE X JUSTICEJUDGMENTS~

PROCEDURAL

P

r Objective measures Outcome of case Cost of litigation Case duration Subjective measures Outcome relative to expectation Cost relative to expectation Delay relative to expectation Decision and process control Procedural dignity Procedure careful

- .04 - .07 - .07

.01 -.01

- .08

.47* * - .08

-.19** .38** .44** .52**

.32** - .03 - .08 .21* .22* .25* .48

“Values are Pearson and polychoric correlations and standardized regression coefficients. From Lind, MacCoun et al. (1990). * p < .Ol. * * p < ,005.

CORRELATES OF

TABLE XI JUSTICE I N ARBITRATION HEARINGS~

PROCEDURAL

Correlation with procedural justice Outcome: $ won/lost Perceived control Decision control Process control Relational concerns Standing Dignity Respect shown? Trust Ethical Considered my evidence Neutrality Bias

.09

.31** .29**

.30** .47**

.62** .62** .17*

aEntries are Pearson correlations. Data from MacCoun, Lind, Hensler. Bryant, and Ebener (1988). * p < . O l . ‘ * p < ,001.

AUTHORITY IN GROUPS

155

ses showed that this preference was due to the perception that the adjudicatory procedures were more dignified than the settlement-oriented procedures. In the MacCoun et al. study procedural justice ratings showed substantial correlations with two measures linked to standing concerns: judgments that the disputants were treated with respect (r = .47) and with dignity (r = .30). We can use the Tyler (1988) data to examine the relative strength of two components of standing: politeness and respect for rights. If we do so, we find that both aspects of standing are important. Ratings of respect for rights show slightly stronger correlations with procedural justice than do ratings of politeness in both legal (r = .54 vs. r = .49) and organizational settings (r = .71 vs. r = .63), but the correlations are not significantly different. The data from the MacCoun et al. study also suggest that respect might be a more important factor in standing than is the judgment that the procedure treats people with dignity. b. Trust. Concern about authorities’ intentions leads people to examine their encounters with authorities for evidence of trustworthiness, i.e., to look for signs of ethical behavior, an intention to act fairly, and benevolence. These concerns reflect a desire by group members to understand the stable, underlying motivations of authorities, motivations that allow the authority’s future behavior to be predicted. Trust depends on people’s efforts to predict how the authorities will act in the future. Survey research on beliefs about authorities gives some clues about the psychology of trust in authorities. When asked in the abstract how legal and political authorities would treat them if they had dealings with the authority, people almost always say that if they were to deal with a given authority in the future, they would receive fair treatment (Tyler, 1990). This expectation exists despite a widespread belief that injustice often occurs when citizens deal with the police and courts. When asked what generally happens when people in general deal with authority figures, survey respondents say that many people are treated unfairly. This discrepancy between the belief that injustice occurs and the belief that one is oneself invulnerable to the same injustice suggests the presence of an “illusion of personal justice” analogous to illusions of personal control. l9 Other research also demonstrates the existence of an illusion of personal justice by showing that there is a gap between beliefs about justice on personal and societal levels. Crosby (1984a, 1984b) has found that women workers often recognize the existence of sex discrimination but nonetheless believe that they IgBecause of this illusion of personal justice, people approach interactions with authorities with the expectation that they will be treated fairly. Ironically, this may place the authority in a no-win situation vis-8-vis those they govern. If the authority acts fairly, this might not have much influence on the attitudes of those affected by the decision, since they expect fairness. Only disconfirming, negative, experiences shape attitudes (Katz, Gutek, Kahn, & Barton, 1975). Hence, over the course of repeated encounters with authorities, people’s views of authorities should become more negative.

156

TOM R. TYLER AND E. ALLAN LIND

are not themselves the victims of discrimination. Similar findings have emerged from studies of racial and ethnic discrimination (Crosby & Nagata, 1990; Taylor, Wright, Moghaddam, & Lalonde, 1990). Crosby and Nagata argue that this gap reflects a desire for “comfortable ignorance” or distancing. By feeling removed from discrimination, people need not feel personally hurt by it. There are two potentially distinct issues in the illusion of personal justice. First, there is the possibility that the illusion grows out of the interpretation of one’s present situation. A person might know that discrimination exists but believe that he or she is not currently a victim (Crosby, 1984a, 1984b). Careful selection of comparison to others might allow one to think that one has escaped discrimination. Second, there is the possibility that what is happening is an overly optimistic assessment of future risks of discrimination (Qler, 1984). Whatever current injustice exists for oneself and others, the illusion of personal justice might lead one to believe that in the future one will be treated fairly. Some early social psychological research shows that people place inordinate trust in the benevolence of authorities. Pepitone (1950) demonstrated that people exaggerate the benevolence of authorities who have power over them and exaggerate the power of authorities they believe to be benevolently disposed toward them. Pepitone coined the term facilitative distortion to describe the phenomenon. What factors are most closely related to the judgment that an authority is trustworthy? Table XI1 shows some correlations based on the Qler (1990) and Tyler and Schuller (1990) studies. Judgments of the authority’s ethicality appear

PROCEDURAL JUSTICE

Competence How hard did authority try to. . . Solve problem? Find satisfactory solution? Prevent future problems? Reduce anger? Ethicality How hard did authority try to. . . Consider your views? Explain decision? Be fair to you? Take account of your needs?

TABLE XI1 FACTORS RELATEDTO TRUW

AND

Management

Law

.54 .62 .35 .I6

,666

.31

.71 .48 .76 .73

.68 .71 -

.52

“Entries in table are Pearson correlations between factor and procedural justice. From QIer (1990); Qler and Schuler (1990). bAsked only of those who called the police.

157

AUTHORITY IN GROUPS

to be more strongly related to procedural justice judgments than do judgments of his or her competence. In the mediation study by Pruitt et al. (1990), which we described earlier, the procedural fairness of mediation sessions was correlated with perceptions of the mediator’s empathy (r = .26). To the extent that people need the security of believing that they will receive fair treatment, they are likely to be particularly sensitive to any apparent disconfirmation of this belief in the actions of an authority. When people encounter authorities their belief in personal justice is being put to the test. A laboratory experiment by Lind and Lissak (1985) shows how unethical behavior by an authority can violate expectations and change procedural justice judgments. Lind and Lissak exposed subjects to a high-process-control trial procedure. In one condition the judge acted in an unethical fashion, showing that he had a prior friendship with the lawyer representing the subject’s opponent and joking with the lawyer prior to the trial; in another condition the judge showed no sign of knowing any of the parties or their lawyers and treated everyone in a purely professional fashion. This manipulation was crossed with a manipulation of the outcome of the trial: some subjects were told they had won the dispute and some were told they had lost. Table XI11 shows the procedural justice ratings that subjects gave in each of the four conditions. When there was no suggestion that the authority was acting improperly, the outcome of the case had little effect on judgments of procedural justice. On the other hand, when there was a suggestion of impropriety, a negative outcome diminished perceived fairness and a positive outcome enhanced perceived fairness. Lind and Lissak explain their findings by suggesting that people expect procedures to be enacted ethically and with neutrality, and the unexpected appearance of impropriety led to a more intensive cognitive analysis of the procedure than occurred without the impropriety. The more intensive cognitive analysis led the subjects in the impropriety-present conditions to Look to the outcome for information on the fairness of the procedure. c. Neutrality. People are concerned with the neutrality of authorities because evidence of bias or discrimination can constitute a powerful threat to one’s selfimage. Evidence of fundamental dishonesty or incompetence can threaten the

TABLE XI11 UNETHICAL BEHAVIOR A N D PROCEDURAL JUSTICE JUDGMENTSO

Favorable outcome Unfavorable outcome

Judge ethical

Judge unethical

6.28 5.89

6.85 5.26

“Higher values indicate higer procedural justice ratings. From Lind and Lissak (1985).

158

TOM R. TYLER AND E. ALLAN LIND TABLE XIV

PROCEDURAL JUSTICE AND FACTORS RELATED TO NEUTRALITP

Consistency of outcomes With expectations With others Lack of bias Honesty Competence

Management

Law

.31 .21

.54

.I9 .61 .68

.36 .43 .53 .37

"Entries in table are Peanon correlations between factor and procedural justice measure. From S l e r (1990); Tyler and Schuller (1990).

basic assumptions of the authority social system, negating the assumption that those in power will behave in a reasonable, predictable fashion. Table XIV compares several factors relating to neutrality in terms of their correlation with procedural justice. Interestingly, in legal settings, consistency, honesty, and bias appear to be most important, while in work settings, honesty and competence seem to be most important. 3 . Relational Concerns and Value to the Group

The assumption underlying this discussion has been that standing, trust, and neutrality affect people because these factors communicate information about the fundamental fairness of social processes in the group and because fairness of treatment, in turn, is taken to indicate the person's value to the group (see Fig. 520). Indeed, the group value that we were most concerned with in our original group-value theory (Lind & Tyler, 1988, chap. 10) is the perception that one is valued by the group. We have argued here that a good relationship with authorities promotes feelings of procedural fairness and that this in turn leads one to feel valued by the group. This group-value belief is, in turn, a potent determinant of various attitudes and behaviors, including judgments of legitimacy and obedience to authority. Tyler and Schuller (1990)gathered data relevant to this assumption by examining the relationship between judgments about control, standing, trust, and neutrality during a recent personal experience with an authority and ratings of how much the respondent felt the group valued him or her. The results of this analysis, shown in Table XV, indicate that all three relational measures influence 2OFigure 5 does not contain an exhaustive listing of the components of trust, standing, and neutrality. The figure is intended simply to illustrate the causal ordering we propose and to point out the importance of fairness and value to the group as mediating variables in our model.

159

AUTHORITY IN GROUPS

\

Respect

/

for Rights

Absence of Bias or Prejudice

Honesty

-Nartrdity

Fact--

Decision Making

Fig. 5. A relational model of authority.

perceptions of value to the group. Interestingly, measures of decision and process control do not appear to account for much variance in feelings of status beyond that explained by standing, trust, and neutrality. If process control affects feelings of value to the group, as we have suggested (Lind & Qler, 1988), it appears to do so by virtue of its effects on judgments of standing, trust, and neutrality. TABLE XV PREDICTORS OF PROCEDURAL JUSTICE A N D PERCEIVED VALUETO GROUP'

Process control Decision control Neutrality Trust Standing

R=

Procedural justice

Value to group

.02

- .03

.15*

.05 .32* .32* .28*

.78

.63

.14*

.31* .42*

"Entries are standardized regression weights and squared multiple correlation for an equation containing all five predictors. From Tyler and Schuller (1990). * p < ,001.

160

TOM R. TYLER AND E. ALLAN LIND

4. lrnplications

The research reviewed above makes it clear that in arriving at judgments of procedural fairness people are not simply concerned with the settlement or decision that results from their encounter with authorities. People also care about the implications of their experience with authorities for their understanding of the group and their position in it. In fact, it is striking how much issues concerning relationships to authority color the perception of justice. Relational concerns seem to overwhelm concerns about the specific problem or dispute involved. In the end, it appears, people do not approach an authority solely to resolve their problem or dispute. Rather they view encounters with authorities as opportunities to understand from the authority’s actions where they stand in the group and what they can expect in future encounters with authorities. Political scientists often regard the average person as having limited knowledge about political events and a superficial and incoherent ideology about political issues and problems. This image of the person emerges from the work of Converse (1964) on beliefs among the mass public. In contrast, the findings we have reviewed here suggest that people have complex and sophisticated criteria that they use when evaluating decision-making procedures and the authorities who use the procedures (Tyler, 1988; Qler et al., 1989). The complexity of people’s criteria for justice is demonstrated by the multiple concerns affecting fairness judgments. The research described above shows that in both legal and managerial settings people respond to a variety of different aspects of a procedure as they arrive at an assessment of its fairness. Other studies, which have used open-ended response formats, have discovered still other factors that are weighed in generating procedural fairness judgments (see Sheppard & Lewicki, 1987). It is interesting to compare the criteria discovered in psychological studies of fairness judgments to those that underlie legal and philosophical codes of ethics and fairness. Consider, for example, the United States government’s code of ethics for the protection of human participants in behavioral and biomedical research (Belmont Report, 1979). The Belmont code identifies three major aspects of ethical treatment: respect for persons, beneficence, and justice. Respect for persons involves issues of standing and control: “giving weight to autonomous persons’ considered opinions and choices” (p. 23193) and not treating people solely as objects. Beneficence is concerned with issues of trust in the motives of authorities, with the scientist being evaluated on the extent to which he or she is motivated to secure the well-being of those being studied. Finally, justice, as used in the Belmont Report, mandates that studies be neutral and nondiscriminatory and that they treat different classes of people consistently. Legal treatments of procedural rights have focused directly on issues that we

AUTHORITY IN GROUPS

161

would label stunding.*’ An example is the work of Mashaw (1985), which articulates a dignitary theory of legal due process. Mashaw argues that legal procedures should enact the basic values of American society. In particular, he argues that procedures should include several characteristics linked to our most basic conceptions of how the relationship between the individual and society should be structured. Two of these characteristics reflect concerns that seem closely linked to issues of standing in society: procedures should “contribute to a participant’s sense of self-respect” (p. 175) and they should protect people’s privacy. Procedures must also be neutral and treat every person equally. Finally, procedures must share control by providing opportunities for participation. Thus in Mashaw’s legal analysis, as in biomedical ethics, one finds many of the same concerns that social psychologists have discovered in everyday justice judgments. Examination of these normative codes suggests that popular conceptions of procedural justice in fact include the major elements of more formal ethical codes. There may well be instances in which a lay application or counterbalancing of various concerns might fall short of that seen in legal or philosophical analyses, but the picture that emerges from procedural justice research shows lay judgments of procedural fairness to be rather more sophisticated than previously thought. The research we have described, and the relational model we propose, have some interesting practical implications for leaders and followers. For leaders the clear message is that approbation-official approval-is a much more powerful tool than is generally thought. In organizational, legal, and political settings it is common practice to use outcome incentives to secure obedience and compliance. Our work suggests that this approach is likely to meet with limited success. A much more effective approach to leadership would involve making each follower feel that he or she is important to the group. Stylistic considerations are as important as substantive issues: politeness and expressions of concern for followers’ needs will secure greater compliance than will cold outcome incentives. Above all, the leader must be concerned with the appearance of fairness, with convincing followers that he or she is willing to consider their point of view, and that he or she will be even-handed and nondiscriminatory in decision-making. Greenberg ( 1990b) offers an excellent analysis of the management of fairness impressions. His conclusions and suggestions are quite similar to those offered here. This advice runs counter to some popular notions of leadership. Many in positions of authority feel that to allow subordinates to express their views might 2llt is important not to confuse what we have chosen to label srunding (concern about one’s membership and status in a group, organization, or society) with what the laws calls srunding (the right to be involved in the suit in question). As is the case with social psychological and legal uses of the term eyuiry, the same term can mean quite different things in the two disciplines.

162

TOM R. TYLER AND E. ALLAN LIND

be seen as a sign of weakness or might build unreasonable expectations. Our work and that of other theorists and researchers suggest that nothing could be further from the truth. It is precisely by showing concern for followers that a leader can gain the greatest effective power in shaping their behavior. Bies and colleagues (Bies, 1985; Bies & Shapiro, 1987; Bies, Shapiro, & Cummings, 1988; Tyler & Bies, 1990) point out the importance of accounts or explanations in situations in which leaders act in ways that appear to violate norms of procedural fairness. Such accounts serve, we would argue, to ameliorate feelings of injustice by working on subordinates’ judgments of standing, neutrality, and trust. By attending to these essential elements of procedural justice, either directly by enacting behaviors that promote them or indirectly by explaining away procedural improprieties, leaders can enhance their legitimacy and encourage compliance with their directives. For followers, the principal message of our work is a caution. The relational model argues that authority turns on factors that are subjective and stylistic in nature. This raises a problem of what is often termedfulse consciousness (Cohen, 1985; Lind & Tyler, 1988, pp. 201-202) or hollow justice (Greenberg, 1990b). Followers have to be on guard to protect against unscrupulous leaders who offer a false image of concern for their welfare. Concern with relational issues in reactions to authority is not illogical, as we note below, but it does leave one open to exploitation by leaders who manipulate the effects we have described above.

111. Conclusions

The research and theory we have reviewed here have important implications for our understanding of the psychology of authority and legitimate power, but they also have implications that go well beyond the immediate topic of this article. The findings we have outlined suggest a picture of social life that is at variance with much recent research and theory on the psychology of groups. One of the most intriguing aspects of this new picture of social life is the way in which it appears to intertwine research issues that have become separated as social psychologists have allowed the field to be divided into social cognition and group process camps. In this section we will consider briefly each of these implications.

A. A RELATIONAL MODEL OF AUTHORITY It is becoming increasingly clear that the key to authoritativeness and legitimacy lies not in judgments about the decisions of an authority, but rather in

AUTHORITY IN GROUPS

163

judgments about the procedure, the process, and the quality of interaction that characterize encounters with authority. In study after study, our research and that of other psychologists has shown that the favorability of a decision or even the fairness of the decision plays only a minor role in determining legitimacy. Much more important, it appears, are judgments of the fairness of procedures. And judgments of procedural fairness are based, in turn, on process-based inferences about one’s relationship with the authority. The belief that the authority views one as a full member of society, trust in the authority’s ethicality and benevolence, and belief in the authority’s neutrality-these appear to be the crucial factors that lead to voluntary compliance with the directives of authority. In a previous discussion of the finding that procedural justice is a powerful determinant of evaluations of groups, organizations, and societies (Lind & Tyler, 1988, pp. 227-228), we argued that procedures are more important than individual outcomes because procedures endure across repeated encounters with the group, whereas outcomes vary from encounter to encounter. For example, because students can expect to encounter much the same procedure every time they take a test, it is really quite reasonable that they would be more sensitive to unfairness in a testing procedure than to the outcome of any individual test. This “repeated-exposure” explanation does not really work for reactions to authorities, however. Some of the authority encounters studied in the research cited above involve repeated interactions with the same authority, but many others do not, and there is no evidence that the procedural justice-legitimacy effect is restricted to the former. Consider, for example, the Lind et al. (1991) study that found procedural justice to be a powerful determinant of the acceptance of arbitration awards. It is very doubtful indeed that any of the litigants in that study expected to ever submit another case to the same arbitrator or, for that matter, to the same arbitration program. Why then should procedure matter so much if, like outcomes, encounters with procedures are generally one-shot interactions? One might argue that, even if one expects to never again confront the present authority, one might nonetheless believe that other authorities would use the same procedures, and that the emphasis on procedural issues might turn on this belief. But this explanation runs afoul of the research on the factors that drive judgments of procedural fairness. In the final analysis, qualities such as politeness, even-handedness, and ethicality account for most of the variance in procedural justice judgments, and these qualities have a distinctly personal quality. Indeed, as procedural justice research has advanced, the emphasis of the field has moved from fairly objective qualities that are constant across many enactments of the procedure-such as whether one has a lawyer to present one’s case or whether one has the right to reject the decision-to qualities that are much more personal and subjective. Part of the explanation of why people are so concerned with their relations with authorities can be found, we believe, in the ambivalence with which people approach those in positions of authority. We noted at the beginning of this article

164

TOM R. TYLER AND E. ALLAN LIND

that the empowerment of an authority differs from the institution of a law or rule because the very nature of authority is the transfer of power from many to one. In addition to the outcome dependence that accompanies such a transfer of power (Thibaut & Kelley, 1959), authorities are generally viewed as speaking for the group: The authority is often the person who is the symbol of the group. Thus authorities have power over other members both because they regulate important outcomes and because their apparent assessment of group members is taken as indicative of the assessment of that member by the entire group. Having given the authority power over both their outcomes and their feelings of self-worth, people are at once fearful that the authority will harm them and hopeful that the authority will act to improve their situation or their self-image. More clues to the puzzle of what makes authority relations so important lie in the process by which people arrive at cognitions about social events and social structures. We humans are social creatures, very much involved in and dependent on the groups, organizations, and society to which we belong. But in many instances, we never really see or interact with the group as a group. Instead, we have occasional encounters with specific other members of the group, and we are left to draw what information we can from these encounters. Nonetheless we need to understand how these nebulous, but extremely important, entities function, and we therefore engage in a rough inductive process, infemng what we can about the group from the information we have. To this end, authorities are treated as though they were indeed representative, and we look to our treatment by one authority to try to discover how other authorities, and the group as a whole, view us. This inductive inference process might also account for the importance of relational factors in procedural justice judgments. Let us begin with the assumption that there are no qualitatively distinctive authority-assessing processes. Instead, familiar social cognition processes come into play when people attempt to decide whether to obey an authority. The only differences in reacting to authority lie in the increased motivation to obtain information about how one is viewed and a propensity to distort information resulting from the desire to have a positive relationship with the authority. When we meet another individual, we immediately begin making inferences about what that person thinks of us, whether he or she is sincere, and whether he or she harbors any bias against us. We do not wait to see whether the person will make decisions that favor us. Instead, we look to the interaction process to arrive at impressions about the person on various dimensions. If the person is rude, if he or she appears to break the conventions that govern the interaction, or if he or she shows some bias or prejudice against us, we respond negatively. If, on the other hand, the person seems well-disposed toward us, if he or she seems to respect our views, and if we find no evidence of prejudice or bias, we respond favorably.

AUTHORITY IN GROUPS

165

The research literature on legitimacy suggests that these same impression formation processes are brought to bear on people’s reactions to authorities. There may well be instances wherein the standards that one applies to an authority are different from those one would apply to an individual one met on purely social grounds, but the data seem to suggest that the way impressions are formed-by looking at the social process for information relevant to one’s relationship to the person-is similar. For interactions with authorities (and perhaps for many other sorts of interactions), the impression dimensions of concern are standing, trust, and neutrality, and the source of information about these factors is the social process and the procedures that the authority uses. The rather personal conclusions that are drawn with respect to the particular authority one has encountered are then generalized to other authorities in the group. The only differences lie in the particular motivational distortions that people bring to bear on experiences with authorities. Because it is so important to people that they have positive relationships with authorities, they approach encounters with authorities with an illusion of personal justice and they interpret the authority’s actions “through rose-colored glasses.”22 Outcomes and other instrumental concerns have little effect on impressions of authorities, according to this line of analysis, because by the time information on these factors are available, relatively strong impressions of the authority have already been generated. When outcome information does arrive, it is interpreted in terms of already-existing beliefs about the authority, and thus it does not exert as much influence as would otherwise be the case.

B. IMPLICATIONS FOR SOCIAL PSYCHOLOGY Whether or not the analysis just presented is an accurate explanation of the process by which authorities come to be viewed as authoritative, the research makes it clear that people are very much concerned with inferences about relational issues when they interact with authorities. Simple as this finding seems, it reveals some interesting limitations in current thinking about both social cognition and group process. We began this article with a question that seemed to belong very much to the group process side of social psychology: How can a leader obtain voluntary compliance from subordinates? Most of the references cited in the first pages of this article were to classic group process research. But by the time we had followed the question through the research literature, we were led to answers that 2% might be argued, however, that in this respect, too, ordinary social cognition processes are at work. Positivity biases are well-documented, and we would predict that the illusion of personal justice exists in interactions with nonauthorities as well as in interactions with authorities.

166

TOM R. TYLER AND E. ALLAN LIND

are essentially cognitive. Legitimacy, it seems, turns on cognitions concerning how the authority has made his or her decision and these cognitions in turn depend on attributions about the reasons for the authority’s behavior and inferences about the meaning of the social process of the encounter. As we reviewed the research literature it became clear that any attempt to explain legitimacy that is not cognitive runs into serious problems. Attempts to explain legitimacy solely by reference to patterns of outcomes, for example, are simply not supported by the data.23The psychology of authority rests upon cognitions about the authority, cognitions about the authority’s beliefs and motivations, and most especially on cognitions about the relationship between the person and the authority. At the same time, however, the cognitive questions (and answers) we have discovered are rather different from the questions that occupy the attention of social cognition researchers. Specifically, the cognitions that underlie perceived legitimacy are in a sense more social: They concern not just inferences about another person but also, and indeed primarily, inferences about a relationship. What matters most in legitimacy, it seems, are not what might be termedfirstorder inferences, i.e., inferences about personal qualities of the authority such as his or her competence or intelligence, but rather second-order inferences, i.e., inferences about the authority’s beliefs, biases, and cognitions concerning the perceiver. These questions of “what he thinks of me” have certainly received attention in previous work in social psychology-research on impression management focuses on just this issue-but they do not receive much attention in current research and theory on social cognition. An intriguing and almost contradictory picture of social behavior emerges from the literature on legitimacy. Our analysis shows the person to be a very social creature indeed, interested in various aspects of his or her relationship with the authority in question-indeed, interested in these relational issues even more than in his or her own outcomes from the authority’s decision. We see the person also as an information processor: a very attentive and sophisticated assessor of the tone and quality of social process, sifting through the authority’s behavior in search of clues for relational inferences. But at the same time, this preoccupation with relational inference, this sophisticated application of multiple criteria to assessing first procedural justice and then from procedural justice determining legitimacy, is subject to the same limitations and inconsistent heuristics that characterize other social cognition processes. Above all, though, we see the person as an entity that is at once social and individual: searching constantly for information concerning his or her individual place in a social structure and trying to understand the personal implications of his or her social world. 2% the terminology of interdependence theory (Kelley & Thibaut, 1978), this is to say that very strong transformation processes are at work in the exercise of legitimate power.

AUTHORITY IN GROUPS

167

IV. Appendix: Measures of Outcome Favorability, Distributive Justice, Procedural Justice, Legitimacy, and Related Variables" ALEXANDER AND RUDERMAN, 1987 Outcome fairness Pay fairness: All in all, my pay is about what it ought to be. My pay is fair considering what other places in this area pay. My pay is fair compared to the pay of others in this agency. Performance appraisal fairness: My last annual performance appraisal rating was about what it ought to have been.

Fairness of decision-making procedures Appeals procedure fairness: The appeals procedures protect me from unfair treatment if a personnel action is brought against me. Appealing personnel actions is a waste of time. The negotiated grievance procedures protect employees if they have a formal personnel action against them for poor performance. When people perform poorly here they're given a chance to improve their work. Performance appraisal fairness: Generally speaking, annual performance appraisals are done fairly here.

FOLGER AND KONOVSKY, 1989 Outcome favorability Raise as a percentage of salary.

Distributive justice How fair do you consider the size of your raise to be? To what extent did your raise give you the full amount you deserved? How did your raise compare to your expectations? To what extent was the size of your raise related to your performance? 24This appendix contains the measures used in studies for which we report specific values and statistics. In some instances it has been necessary to infer the wording of questions from published reports.

TOM R. TYLER AND E. ALLAN LIND

Procedural justice The supervisor . . . Was honest and ethical in dealing with you. Gave you an opportunity to express your side. Used consistent standards in evaluating your performance. Considered your views regarding your performance. Gave you feedback that helped you learn how well you were doing. Was completely candid and frank with you. Showed a real interest in trying to be fair. Became thoroughly familiar with your performance. Took into account factors beyond your control. Got input from you before a recommendation. Made clear what was expected of you. Discussed plans or objectives to improve your performance. Reviewed with you objectives for improvement. Resolved difficulties about your duties and responsibilities. Obtained accurate information about your performance. Found out how well you thought you were doing your job. Asked for your ideas on what you could do to improve company performance. Helped you find out why you got the size of raise you did. Let you make an appeal about the size of your raise. Allowed you to discuss how your performance was evaluated. Helped you develop an action plan for future performance. Frequently observed your performance.

Organizational commitment Cfrom Mowday, Steers, & Porter, 1979) I am willing to put in a great deal of effort beyond that normally expected in order to help this organization be successful. I talk up this organization to my friends as a great organization to work for. 1 feel very little loyalty to this organization. I would accept almost any type of job assignment in order to keep working for this organization. I find that my values and the organization’s values are very similar. I am proud to tell others that I am part of this organization. 1 could just as well be working for a different organization as long as the type of work was similar. This organization really inspires the very best in me in the way of job performance. It would take very little change in my present circumstances to cause me to leave this organization. I am extremely glad that I chose this organization to work for over others I was considering at the time I joined. There’s not too much to be gained by sticking with this organization indefinitely. Often, 1 find it difficult to agree with this organization’s policies on important matters relating to its employees.

AUTHORITY IN GROUPS

169

I really care about the fate of this organization. For me this is the best of all possible organizations for which to work. Deciding to work for this organization was a definite mistake on my part.

KANFER, SAWYER, EARLEY, AND LIND, 1987 Procedural justice How fair is the procedure used to determine which company will receive the contract? How satisfied are you with the procedure used to evaluate your company’s performance?

Evaluation of supervisor To what extent do you consider the representative to be an ally in your effort to win the contract? How satisfied are you that the best arguments will be presented?

LIND, KANFER, AND EARLEY, 1990 Acceptance of goal How committed are you to attaining the goal assigned to you by the experimenter?

Procedural justice How fair was the way the goal was set?

Task performance [Task units completed in time allotted.]

Control How much control did you have over the goal that was set?

LIND, KULIK, AMBROSE, AND PARK, 1991 Relational process judgments HOWmuch opportunity was there in your case to have the facts heard and decided by an impartial third party?

170

TOM R. TYLER AND E. ALLAN LIND Would you say the procedures and rules that applied to your case were very effective, moderately effective, or not effective at all in getting at the facts? Would you say the procedures and rules that applied to your case were very effective, moderately effective, or not effective at all in assuring that your case was decided by a neutral party?

Outcome favorability judgments Now I’d like to ask you a few questions about the final result of the lawsuit. Was the final result of the case very favorable, somewhat favorable, somewhat unfavorable, or very unfavorable to you (your organization)? Compared to what you expected when you entered the lawsuit, is the final result of the case more favorable to you (your organization), less favorable, or about what you expected?

Procedural justice judgments Overall, fair were the procedures and rules that applied to your case? Would you say they were very fair, somewhat fair, somewhat unfair, or very unfair? Would you say the procedures and rules that applied to your case were very effective, moderately effective, or not effective at all in ensuring that both parties were treated fairly?

Acceptance of awards [Coded from court documents. J

LIND AND LISSAK, 1985 Procedural justice How satisfied were you with the dispute resolution procedure in which you participated? How fair was the dispute resolution procedure in which you participated? How satisfied are you with the procedure used in the dispute resolution? How fair was the procedure used in the dispute resolution?

LIND, LISSAK, AND CONIDN, 1983 Procedural justice How satisfied are you with the procedure used in the dispute resolution? How fair was the procedure used in the dispute resolution?

AUTHORITY IN GROUPS

171

LIND, MAcCOUN ET AL., 1989, 1990 Procedural justice [Arbitration] At the time you went through it, how fair did you think the arbitration hearing was? [Judicial settlement conference] At the time the conference took place, how fair did you think it was? [Trial] At the time you went through it, how fair did you think the trial was? [Settlement] At the time, how fair did you think the settle process was? [All procedures] Finally, looking back, do you think the procedure that was used for resolving your dispute was very fair, somewhat fair, somewhat unfair, or very unfair?

Dignity of procedure Which word best describes your (hearing/settlement conference/trial/expenence [settlement])? Response options: Dignified, Undignified.

Case outcome How did you feel about the final outcome-were you very satisfied, somewhat satisfied, somewhat dissatisfied, very dissatisfied?

Cost of litigation [Dollars spent on attorney fees and litigation expenses.]

Case duration [Days from incident causing litigation to case termination. 1

Outcome relative to expectation Compared to what you expected when the case was first filed, was the final outcome: better than you expected, worse than you expected, just about what you expected?

Decision and process control Outcome control. How much control did you feel you had over the outcome of your case? Response options: A great deal, Some, A little, Not much. Process control. Overall, thinking about your dealings with the court and with your lawyer, how much control would you say you had over the way your case was handled? Response options: A lot, Some, A little, Not much.

172

TOM R. TYLER AND E. ALLAN LIND

Procedure careful Which word best describes your (hearinglsettlement conference/trial/experience [settlement condition])? Question 1 response options: Careful, Casual. Question 2 response options: Superficial, Thorough.

Cost relative to expectation All things considered, do not think you got your money’s worth from your attorney?

Delay relative to expectation Considering what had to be done, do you think the time it took for your case to get resolved was very reasonable, somewhat reasonable, somewhat unreasonable, or very unreasonable? Did it take too much time to resolve the case or was there not enough time? [asked if response to above was not “Very reasonable.”]

MAcCOUN, LIND, HENSLER, BRYANT, AND EBENER, 1988 Procedural justice The rules and procedures used in the hearing were fair. In general, the arbitration hearing was fair. All things considered, what would have been the fairest way to decide this kind of case? (Trial by judge; trial by jury; arbitration.)

Relational concerns Standing The arbitrator(s) treated me with respect. The arbitration hearing was dignified. Trust The arbitrator@) were very ethical in their handling of this case. The arbitrator(s) gave thorough consideration to all the evidence. Neutrality The arbitrator(s) showed favoritism toward one side.

Outcome favorability Was the verdict in [your]/[the other side’s] favor? According to the award designated by arbitrator(s), what is the total amount (including attorney’s fees) you would receive or pay?

AUTHORITY IN GROUPS

173

lntentions to acceptlreject award Do you think you will accept or reject the arbitration award? You had a choice to accept or reject the arbitrator(s) decision.

Outcome Perceived decision control: You had some control over the decision of the arbitrator(s). Perceived process control: You had some control over the way the arbitration hearing was conducted. Your lawyer had some control over the way the arbitration hearing was conducted.

MUSANTE, GILBERT, AND THIBAUT, 1983 Procedural justice lfactor scores with substantial loadings on the following questions] How satisfied are you with the procedure? Do you trust the procedure? Is the procedure fair? Was the trial fair? Are you satisfied with the trial? Was the trial verdict an objective representation of the facts in the case? Was the trial verdict fair? Do you accept the verdict?

PRUITT, PEIRCE, McGILLICUDDY, WELTON, AND CASTRIANO, 1990 Compliance with mediation Did the other party live up to all the terms of agreement?

Procedural fairness Was the mediation hearing fair?

Joint problem-solving [Coded from behavior in mediation hearing.]

Perceptions of mediator empathy [Measure not reported in paper.]

174

TOM R. TYLER AND E. ALLAN LIND

RASINSKI, 1988 Political activity related to specific government benefits and services Have you ever spoken to, written, or called a political official about government benefits or services? Have you ever written to a newspaper, magazine, television, or radio station about government benefits or services? Have you ever attended a political meeting or demonstration concerning government benefits or services?

General political activity Have you ever worked for a political candidate’s campaign? Have you ever displayed political signs, bumper stickers, or worn political buttons during an election period? Have you ever donated money to a political campaign? Have you ever tried to convince others who to vote for?

Procedural justice How fair are the procedures government agencies use to decide which groups of people will receive benefits. Are they very fair, somewhat fair, somewhat unfair, or very unfair? How often do you think the people in government consider the needs of all people when they decide who to give government benefits to? Would you say always, usually, sometimes, or never? How often do you think the people in government consider the needs of you or your family when they decide how much benefits you will receive? Would you say always, usually, sometimes, or never? Are the people in government usually impartial when they decide who will get benefits or do you think they favor some groups over others? Do you think government agencies usually have enough information about your needs when they make policies about benefits you will receive? Do you think government agencies usually have enough information about the needs of all people when they make policies about benefits you will receive? Do you think government agencies usually have enough information about the needs of all people when they make policies about which groups should get benefits?

Outcome favorability As you know, we all receive some type of benefit from the government. For example, the government maintains streets and parks and provides police and fire protec-

AUTHORITY IN GROUPS

175

tion. Public schooling is another benefit some people receive from the government. Some people even get special benefits like food stamps or social security. I would like you to think about whatever benefits you receive from the government. Overall, how much do you benefit from the government? Do you get a great many benefits, some benefits, a few benefits, or not many benefits at all? Now I am going to read a list of government benefits. For each one could you tell me whether you receive it now or received it any time in the past or never received it. a. Social Security b. Workmen’s compensation c. Veteran’s benefits d. Disability insurance e. Unemployment compensation f. AFDC. That is, aid to families with dependent children g. SSI. That is, supplemental security income h. General assistance. That is, locally provided public assistance to people not eligible for public aid i. Low-interest loans such as for education or home improvement j. Government grants such as for education, job training, or home improvement k. Medicare or Medicaid 1. Food stamps m. Anything I haven’t mentioned? Compared to 3 years ago (1980/1981) do you receive the same amount of benefits from the government, more benefits, or less benefits? Now I am going to name some groups of people. Please tell me if you think you receive more government benefits than the group, the same amount of benefits, or fewer benefits than the group? a. Elderly people over 65 b. Poor people c. People who are handicapped d. Middle-income people e. Unemployed people f. Rich people g. Minorities h. Sick people

Distributive fairness Do you feel that you receive as much benefit as you deserve or do you receive more or fewer benefits than you deserve? In your opinion do elderly people get more than their fair share of benefits from the government, less than their fair share of benefits, or do they get the benefits that they deserve? (Repeat questions substituting each of the seven remaining groups from above.)

I76

TOM R. TYLER AND E. ALLAN LIND

TYLER, 1984 Outcome favorability What was the final disposition of your case? (Charges dismissed/found innocent; found guilty, but sentence suspended; sentenced to serve supervised sentence; finedlgiven probation/given suspended sentence).

Distributive justice How fair was the outcome you received?

Procedural justice How just and impartial were the procedures used by the judge in trying your case?

Legitimacy How would you rate the performance of the judge in terms of: Overall performance of his duties Courtesy Honesty Fairness Overall, how fairly was your case handled? How good a job do you think the courts are doing in handling cases such as yours?

TYLER, 1990

Outcome favorability Calls to the police: Did the police solve the problem about which you called? Stopped by the police: Did the police arrest you or take you to a police station? Did the police cite you for a violation of the law? How serious was the violation? courts: How pleased were you with the outcome of your case?

Distributive justice How fair was the outcome [of your call]/[of your contact with the police]/[you received in court]?

AUTHORITY IN GROUPS

177

Did you feel that you received the outcome you deserved, a better outcome than you deserved, or a worse outcome than you deserved?

Procedural justice How fair were the procedures used [to handle your call]/[when you were stopped]/[when you went to court]? How fairly were you treated?

Legitimacy I have a great deal of respect for [the police/the courts]. On the whole, [police officers/judges] are honest. I feel proud of the [police/courts]. I feel that I should support the [police/courts]. Court decisions are almost always fair. The basic rights of citizens are well protected in the courts. People should obey the law even if it goes against what they think is right. I always try to follow the law even if I think it is wrong. Disobeying the law is seldom justified. It is difficult to break the law and keep one’s self-respect. A person who refused to obey the law is a menace to society.

Process control during the experience How much opportunity were you given to describe your problem before any decisions were made about how to handle it?

Decision control during the experience How much influence did you have over the decisions made by the [police/judge]?

Neutrality during the experience Did the methods used by the [police/judge] favor one person over another or were they equally fair to everyone involved? Were the [police/courts] honest in what they said to you? Did the [police/judge] do anything that you thought was dishonest or improper? Did they get the information needed to make good decisions about how to handle the issues involved? Did they try to bring the issues into the open so that they could be solved?

178

TOM R. TYLER AND E. ALLAN LlND

Trust How hard did your supervisor try to be fair to you?

Standing How politely were you treated? How much concern was shown for your rights?

Outcome consistent with others When you compare the outcome of your experience to the outcome that people generally receive in similar situations, did you receive about the same outcome as others, or did you receive a better or worse outcome than others generally receive?

Outcome consistent with expectations Was the outcome of your experience what you thought it would be or was it better or worse than you expected?

Correctability Do you know of any agency or organization that you could have complained to if you had felt that the [police/courts] were unfair?

Evaluation of authority How good a job are the [police/courts] doing? How satisfied are you with how well the [police/courts] solve problems and help those who come to them? How satisfied are you with the fairness of the outcomes that people receive when they deal with the [police/courts]? How satisfied are you with the fairness of the way the [police/courts] treat people and handle problems?

Compliance with laws Have you often, sometimes, seldom, or never done each of the following things during the past year: Made enough noise to disturb your neighbors? Littered in violation of the law?

AUTHORITY IN GROUPS

179

Driven an automobile while intoxicated? Driven over 55 miles per hour on the highway? Taken inexpensive items from stores without paying for them? Parked your car in violation of the law?

TYLER AND CAINE, 1981 Outcome favorability: students [Class grade]

Outcome favorability: Political authorities How much influence do you feel that you have over the decisions of the national government? To what extent do you think that the current policies of the national government benefit you?

Distributive justice: students Did the student receive the grade they deserved?

Distributive justice: political authorities Do you receive the outcomes you deserve?

Procedural justice: students How fair were the grading procedures used in the class?

Procedural justice: political authorities How fair are the procedures by which government decisions are made? How fair are the procedures by which government benefits are distributed?

TYLER, CASPER, AND FISHER, 1989 Outcome favorability Severity of sentence (no incarceration; some time in a jail; some time in a prison) Months sentenced to serve

TOM R. TYLER AND E. ALLAN LIND

180

Distributive justice How does your sentence compare to that received by others who have committed the same crime?

Procedural justice Did your lawyer: believe your story; fight hard fm you; tell you the truth; listen to you; give good advice; and want justice? Did the judge: tell the truth; follow the rules; try to determine if you were guilty; listen to all sides; act in an unbiased way; and want justice? Did the prosecutor: pay attention to your case; listen to all sides; want justice; and the tell the truth?

Attitudes toward authorities Own lawyer Fought hard for client Wants client to plead guilty Tells truth to client Listens to client Does not want client convicted Wants light sentence for client Wants justice for client Not primarily concerned about making money Does not want client punished Is on client’s side Prosecutor Tried to establish guilt or innocence Listened to all sides of the case Did not want defendants punished Wanted justice Did not simply want convictions Honest Not out to get defendant Judge Unbiased Not out to get defendant Followed legal rules Listened to all sides Honest Did not want all defendants punished Wanted justice

AUTHORITY IN GROUPS

181

Plea bargaining A good way to settle case? Benefits defendants? Lightens sentences Hurts the innocent

Interpersonal treatment Treatment by the police during arrest. Qpe of lawyer [public defendantlprivate]. Time the lawyer spent with defendant. Mode of deposition [plea bargainingltrial].

Potential sentence Prior record. Crime charged. Days in detention prior to trial.

TYLER, RASINSKI, AND McGRAW, 1985 [Preamble] The government provides citizens with many types of services and benefits, such as social security, Medicare and Medicaid, housing mortgage subsidies, veterans benefits, student loans, and unemployment and workman’s compensation.

Outcome favorability Do you receive a great many, some, a few, or no benefits from the government?

Distributive justice Do you feel that you receive a fair number of benefits or that you receive more or fewer benefits than you deserve?

Procedural justice How fair are the procedures by which the government decides who will receive government benefits? [Preamble] Consider the policies of the Reagan administration. First, let me ask you about the economic policies of the Reagan administration, that is, the way it has

182

TOM R . TYLER AND E. ALLAN LIND handled the problems of inflation, unemployment, and the role of government in the economy.

Outcome favorability In general, how much do you agree with the economic policies of the Reagan administration? How have those policies influenced your own economic situation? Would you say that they have helped you, hurt you, or that they haven't had much effect on you?

Distributive justice Do you feel that these policies are fair to you or that they benefit you more or less than is fair?

Procedural justice In deciding what national policies to implement, do you think that President Reagan . . . Considers the views of all sides? Takes enough time to consider policy decisions carefully? Has enough information to make good policy decisions? Is unbiased and impartial?

Trust in national government The government in Washington can usually be trusted to do what is right. Most federal government agencies do a good job. I respect the political institutions of the American government. My friends and I are well-represented in our political system.

TYLER AND SCHULLER, 1990 Outcome favorability How favorable was the outcome [of the experience] to you? How favorable was the outcome relative to what you expected prior to the experience?

Procedural justice How fair were the procedures used to handle the problem? How fairly were you treated by your supervisor?

AUTHORITY IN GROUPS

183

Distributive justice How fair was the outcome you received?

Process control during the experience How much opportunity were you given to describe your problem before any decisions were made about how to handle it?

Decision control during the experience How much influence did you have over the decisions made by your supervisor?

Neutrality during the experience Did the methods used by your supervisor favor one person over another or were they equally fair to everyone involved? Was your supervisor honest in what he said to you? Did your supervisor do anything that you thought was dishonest or improper? Did your supervisor get the information needed to make good decisions about how to handle the issues involved? Did your supervisor try to bring the issues into the open so that they could be solved?

Trusc How much consideration did your supervisor give to your views when decisions were made about how to handle the problem? How hard did your supervisor try to be fair to you? How hard did your supervisor try to take account of your needs in the situation?

Standing How politely were you treated? How much concern was shown by your supervisor for your rights?

Competence How hard did the authority try to Solve problem? Find satisfactory solution? Prevent future problems? Reduce anger?

.

TOM R. TYLER AND E. ALLAN LIND

Ethicalily How hard did the authority try to Consider your views Explain decision? Be fair to you? Take account of your needs?

..

Outcome consistent with others When you compare your outcome to the outcome people generally receive when they go to your supervisor with similar problems, did you receive about the same outcome as others, or did you receive a better or worse outcome than others usually receive?

Correctability Could you have complained to higher authorities in your organization? If you had complained, how likely is it that the complaint would have been effective? Could you have complained to your union? If you had complained, how likely is it that the complaint would have been effective? Could you have gone to court or complained to some government agency? If you had complained, how likely is it that the complaint would have been effective?

Legitimacy (see Organizational commitment)

Accept decision How willing are you to accept the decision?

Follow rules I follow work mles and instructions with extreme care. I sometimes miss deadlines set by the organization. I follow the policies established by my supervisor.

Intention to quit I often think about quitting my job. I will probably look for a job outside my current organization in the next year.

AUTHORITY IN GROUPS

185

I sometimes do things at work that I don’t agree with because I can’t afford to lose my job.

Intention to sue organization How much did you think about whether or not to complain to a government agency or go to court? Did you complain?

Supervisor competence He/she tries very hard to solve the problems he/she deals with. Helshe gets the information needed to make good decisions.

Supervisor trust He/she tries very hard to be fair. Helshe shows a real interest in trying to be fair to me. Helshe considers people’s views when making decisions.

Management evaluation Management can be trusted to make decisions that are also good for me. I trust management to treat me fairly.

Organizational commitment I am willing to put in a great deal of effort beyond that normally expected in order to help the organization I work for be successful. I talk up the organization I work for to my friends as a great place to work. I would accept almost any type of job assignment to keep working where I work now. I find that my values and the values of the organization I work for are very similar. 1 am proud to tell others that I am part of the organization I work for. The organization I work for really inspires me to do the very best job I can. I would recommend to close friends that they join the organization I work for.

Relative procedural desirability Was your problem handled in the manner you expected, or was it handled in a better or worse way than you expected?

Procedural desirability How satisfied were you with the procedures that your supervisor used to handle your problems?

186

TOM R. TYLER AND E. ALLAN LlND

References Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.),Advances in experimental social psychology (Vol. 2). New York: Academic Press. Adler, 1. W., Hensler, D. R., & Nelson, C. E. (1983). Simple justice: How litigants fare in the Pittsburgh Court Arbitration Progrum. Santa Monica, CA: RAND Corporation. Alexander, S., & Ruderman, A. (1987). The role of procedural and distributive justice in organizational behavior. Social Justice Research. 1, 177- 198. Barrett-Howard, E., & Qler, T. R. (1986). Procedural justice as a criterion in allocation decisions. Journal of Personality and Social Psychology, 50, 296-304. Belmont Report. (1979). Ethical principles and guidelines for the protection of human subjects of research. Federal Register, 44, 23192-23197. Bies, R. J. (1985). The predicament of injustice: The management of moral outrage. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 9, pp. 289-319). Greenwich, CT: JAI Press. Bies, R. J., & Moag, J. S. (1986). Interactional justice: Communication criteria of fairness. In R. J. Lewicki, B. H. Sheppard, & M. H. Bazerman (Eds.), Research on negotiation in organizations (pp. 43-55). Greenwich, CT: JAI Press. Bies, R. J., & Shapiro, D. L. (1987). Interaction fairness judgments: The influence of causal accounts. Social Justice Research, 1, 199-218. Bies, R. J., Shapiro, D. L., & Cummings, L. L. (1988). Causal accounts and managing organizational conflict. Communication Research, 15, 381-399. Blau, P. (1964). Exchange and power in social life. New York: Wiley. Block, R. L. (1971). Fear of crime and fear of police. Social Problems, 19, 91-101. Blumberg, P. (1980). Inequality of an age of decline. New York: Oxford University Press. Brett, J. M. (1986). Commentary on procedural justice papers. In R. Lewicki, M. Bazerman, & B. Sheppard (Eds.), Research on negotiation in organizations (Vol. 1, pp. 81-90). Greenwich, CT: JAI Press. Brewer, M. B., & Kramer, R. (1986). Choice behavior in social dilemmas: Effects of social identity, group size, and decision framing. Journal of Personality and Social Psychology, 50, 543-549. Caddell, P. (1979). Crisis of confidence: Trapped in a downward spiral. Public Opinion. 5 , 2-8. Cartwright, D., and Zander, A. (1953). Group dynamics. New York: Harper & Row. Casper, J. D., Qler, T.R., & Fisher, B. (1988). Procedural justice in felony cases. Luw and Society Review, 22, 483-507. Citrin, J. (1974). Comment: The political relevance of trust in government. American Political Science Review, 68, 973-988. Cohen, R. L. (1985). Procedural justice and participation. Human Relations. 38, 643-663. Collins, B. E., & Raven, B. H. (1969). Group structure: Attraction, coalitions, communication, and power. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (Vol. 4, pp. 102-204). Reading, MA: Addison-Wesley. Conlon, D. E., Lind, E. A,, & Lissak, R. 1. (1989). Nonlinear and nonmonotonic effects of outcome on procedural and distributive fairness judgments. Journal of Applied Social Psychology, 19, 1085- 1099. Converse, P. E. (1964). The nature of belief systems in mass publics. In D. Apter (Ed.),Ideologyund discontent (pp. 206-261). London: Collier-Macmillan. Crosby. F. (1976). A model of egoistical relative deprivation. Psychological Review, 83, 85-1 13. Crosby, F. (1982). Relative deprivation and working women. New York: Oxford University Press. Crosby, F. (1984a). Relative deprivation in organizational settings. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 6, pp. 51-93). Greenwich, CT: JAI Press.

AUTHORITY IN GROUPS

187

Crosby, F. (1984b). The denial of personal discrimination. American Behavioral Scienrist, 27, 371386.

Crosby, F., & Nagata, D. (1990, July). Denying person01 disadvantage. Paper presented at the annual meeting of the lnternational Society of Political Psychology, Washington, DC. Dawes, R. M., Orbell, J. M., Simmons, R. T., & van de Kragt, A. J. C. (1986). Organizing groups for collective action. American Political Science Review, 80, 1171-1 185. decarufel, A. (1981). The allocation and acquisition of resources in times of scarcity. In M. I. Lerner & S. C. Lerner (Eds.), The justice motive in social behavior. New York: Plenum. Deutsch, M. (1981). Justice in “the crunch.” In M. J. Lerner & S. C. Lerner (Eds.), The justice motive in social behavior. New York: Plenum. Earley, P. C., & Lind, E. A. (1987). Procedural justice and participation in task selection: The role of control in mediating justice judgments. Journal of Personality and Social Psychology, 52, 1148-1160.

Easton, D. (1965). A systems analysis of political life. Chicago: University of Chicago Press. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7 , 117-140. Folger, R. (1977). Distributive and procedural justice: Combined impact of “voice” and improvement on experienced inequity. Journal of Personality and Social Psychology, 35, 108-1 19. Folger, R. (1986). Rethinking equity theory: A referent cognitions model. In H. W. Bierhoff, R. L. Cohen, & J. Greenberg (Eds.), Justice in social relations (pp. 145-164). New York: Plenum. Folger, R., & Greenberg, J. (1985). Procedural justice: An interpretive analysis of personnel systems. In K. Rowland & G. Fenis (Eds.), Research in personnel and human resources management (Vol. 3, pp. 141-183). Greenwich, CT: JAI Press. Folger, R., & Konovsky, M. (1989). Effects of procedural and distributive justice on reactions to pay raise decisions. Academy of Management Journal, 32, 115-130. Folger, R., Rosenfield, D., Grove, J., & Corkran, L. (1979). Effects of “voice” and peer opinions on responses to inequity. Journal of Personality and Social Psychology, 37, 2253-226 I . French, 1. R. P., Jr., & Raven, B. (1959).The bases of social power. In D. Cartwright (Ed.), Studies in social power. Ann Arbor, MI: Institute for Social Research. Golding, W. (1962). Lord oftheflies. New York: Putnam. Greenberg, J. (1981). The justice of distributing scarce and abundant resources. In M. J. Lerner & S. C. Lerner (Eds.), The justice morive in social behavior. New York: Plenum. Greenberg, 1. (1989). Reactions to procedural injustice in payment distributions: Do the ends justify the means. Journal of Applied Psychology, 72, 55-61. Greenberg, J. (1 990a). Organizational justice: Yesterday, today, and tomorrow. JOUrMl of Management, 16, 401-434. Greenberg, I. (199Ob). Looking fair vs. being fair: Managing impressions of organizational justice. In B. Staw & L. Cummings (Eds.), Research in organizational behavior (Vol. 12. pp. 11 1157). Greenwich, CT: JAI Press. Greenberg, J., & Folger, R. (1983). Procedural justice, participation, and the fair process effect in groups and organizations. In P. Paulus (Ed.), Basic group processes (pp. 235-266). New York: Springer-Verlag. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Hollander, E. P. (1980). Leadership and social exchange processes. In K. J. Gergen, M. S. Greenberg, & R. H. Willis (Eds.), Social exchange: Advances in theory and research. New York: Plenum. Hollander, E. P., & Julian, J. W. (1978). Studies in leader legitimacy, influence and innovation. In L. Berkowitz (Ed.), Group processes. New York: Academic Press. Homans, G. (1961). Social behavior: Its elementary forms. New York: Harcourt, Brace, and World. Houlden, P. (1980). The impact of procedural modifications on evaluations of plea bargaining. Law and Society Review, 15, 267-292.

188

TOM R. TYLER AND E. ALLAN LIND

Kanfer, R., Sawyer, J., Earley, P. C., & Lind, E. A. (1987). Fairness and participation in evaluation procedures: Effects on task attitudes and performance. Social Justice Research. 1, 235249.

Katz, D., Gutek, B. A., Kahn, R. L., & Barton, E. (1975). Bureaucratic encounters. Ann Arbor, MI: Survey Research Center, Institute for Social Research. Kelley, H. H.,& Thibaut, J. (1978). Interpersonal relations: A theory of interdependence. New York: Wiley. Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2, 51-60. Kelman, H. C., & Hamilton, L. (1989). Crimes of obedience. New Haven, C R Yale. Kramer, R. D., & Brewer, M. B. (1984). Effects of group identity on resource use in a simulated commons dilemma. Journal of Personality and Sociat Psychology, 46, 1044-1057. LaTour, S. (1978). Determinants of participant and observer satisfaction with adversary and inquisitorial modes of adjudication. Journal of Personality and Social Psychology. 36, 15311545.

Laver, M. (1981). The polirics of private desires. New York: Penguin. Lerner, M. J. (1980). The belief in a jusr world. New York: Plenum, Lerner, M.J. (1981). The justice motive in human relations: Some thoughts on what we know and need to know about justice. in M. J. Lerner & S . C. Lerner (Eds.), Thejustice motive in social behavior. New York: Plenum. Lerner, M. J. (1982). The justice motive in human relations and the economic model of man: A radical analysis of facts and fictions. In V. J. Derlega & J. Grzelak (Eds.), Cooperation and helping behavior: Theories and research. New York: Academic Press. Leventhal, G. S . (1976). Fairness in social relationships. In J. Thibaut, J. Spence, & R. Carson (Eds.), Contemporary topics in social psychology. Morristown, NJ: General Learning Press. Leventhal, G. S . (1980). What should be done with equity theory? New approaches to the study of fairness in social relationships. In K. Gergen, M. Greenberg, & R. Willis (Eds.), Social exchange (pp. 27-55), New York: Plenum. Lewin, K., Lippitt, R., & White, R. K. (1939). Patterns of aggressive behavior in experimentally created “social climates.” Journal of Social Psychology, 10, 27 1-299. Lind, E. A., & Earley, P. C. (1992). Procedural justice and culture. International Journal of Psychology, 27, 227-242. Lind, E. A., Kanfer, R., & Earley, P. C. (1990). Voice, control, and procedural justice: Instrumental and noninstrumental concerns in fairness judgments. Journal of Personalityand Social Psychology, 59, 952-959. Lind, E. A., Kulik, C. T., Ambrose, M., & Park, M. (1991). Outcome and process concerns in organizational dispute resolution. American Bar Foundation Working Paper. Lind, E. A., Kurtz, S . , Musante, L., Walker, L., & Thibaut, J. (1980). Procedure and outcome effects on reactions to adjudicated resolutions of conflicts of interest. Journal of Personality and Social Psychology, 39, 643-653. Lind, E. A., C Lissak, R. 1. (1985). Apparent impropriety and procedural fairness judgments. Journal of Experimental Social Psychology, 21, 19-29. Lind, E. A., Lissak, R. E., & Conlon, D. E. (1983). Decision control and process control effects on procedural fairness judgments. Journal of Applied Social Psychology, 4, 338-350. Lind, E. A., MacCoun, R. J., Ebener, P. A., Felstiner, W. L. F., Hensler, D. R.,Resnik, J., & Qler, T.R. (1989). The perception of justice: Tort litigants’ views of trial, court-annexedarbitration, and judicial settlement conferences. Santa Monica, CA: Rand Corporation. Lind, E. A,, MacCoun, R. J., Ebener, P. A., Felstiner, W. L. F., Hensler, D. R.,Resnik, J., &Vler, T. R. (1989). The perception of justice: Tort litigants’ views of rrial. court-annexed arbitration, and judicial settlement conferences. Santa Monica. CA: RAND Corporation. Lind, E. A,, & Qler, T.R. (1988). The socialpsychology ofproceduraljustice. New York: Plenum.

AUTHORITY IN GROUPS

189

Lipset, S. M., & Schneider, W. (1983). The confidence gap: Business, labor, and management in the public mind. New York: Free Press. Lissak, R. 1. (1983). Procedural fairness: How employees evaluate procedures. Unpublished doctoral dissertation, University of Illinois, Champaign. Lissak, R. I., & Sheppard, B. H. (1983). Beyond fairness: The criterion problem in research on dispute intervention. Journal of Applied Social Psychology, 13, 45-65. MacCoun, R. J., Lind, E. A,, Hensler, D. R., Bryant, D. L., & Ebener, P. A. (1988). Alternative adjudication: An evaluation of the New Jersey automobile arbitration program. Santa Monica, CA: Institute for Civil Justice, RAND Corporation. Marshaw, 1. L. (1985). Due process in the administrative state. New Haven, CT:Yale. Masters, J. C., & Smith, W. P. (1987). Social comparison, social justice. and relative deprivation. Hillsdale, NJ: Erlbaum. McEwen, C. A,, & Maiman, R. J. (1984). Mediation in small claims court. Law andsociety Review. 18, 11-49. Messick, D. M., & Brewer, M. B. (1983). Solving social dilemmas. In L. Wheeler & P. Shaver (Eds.), Review of personalify and social psychology (Vol. 4, pp. 11-44). Beverly Hills, CA: Sage. Messick, D. M., & Sentis, K. (1983). Fairness, preference, and fairness biases. In D. Messick & K. Cook (Eds.), Equity theory: Psychological and sociological perspectives (pp. 61-94). New York: Praeger. Messick, D. M., Wilke, H.,Brewer, M. B., Kramer, R. M., Zemke, P.,& Lui, L. (1983). Individual adaptations and structural change as solutions to social dilemmas. Journal of Personaliry and Social Psychology, 44, 294-309. Miller, A. (1974). Political issues and trust in government: 1964-1970. American Political Science Review, 68, 951-972. Miller, A. (1979). The institutionalfocus of political distrust. Paper delivered at the annual meeting of the American Political Science Association, Washington, DC. Miller, A. (1983). Is confidence rebounding? Public Opinion, 6, 16-20. Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The measurement of organizational commitment. Journal of Vocational Behavior, 14, 224-247. Murdoch, P. (1967). The development of contractual norms in a dyad. Journal of Personality and Social Psychology, 6, 206-2 1 1. Murdoch, P., & Rosen, D. (1970). Norm formation in an interdependent dyad. Sociomerry, 3 3 , 2 6 4 275.

Musante, L., Gilbert, M. A., & Thibaut, J. (1983). The effects of control on perceived fairness of procedures and outcomes. Journal of Experimental Social Psychology, 19, 223-238. Olson, J. M., Herman, C. P., & Zanna, M. P. (1986). Relative deprivation and social comparison. Hillsdale, NJ: Erlbaum. Ophuls, W. (1977). Ecology and rhe politics of scarcity. San Francisco: Freeman. O’Reilly, C., & Chatman, J. (1986). Organizational commitment and psychological attachment: The effects of compliance, identification, and internalization on prosocial behavior. Journal of Applied Psychology, 71, 492-499. Parks, R. B. (1976). Police responses to victimization: Effects on citizen attitudes and perceptions. In W. Skogan (Ed.), Sample surveys of the victims of crime. Cambridge, MA: Ballinger. Pepitone, A. (1950). Motivational effects in social perception. Human Relations, 3, 57-76. Pruitt, D. G., Peirce, R. S., McGillicuddy, N. B., Welton, G. L., & Castrianno, L. M. (1990). Longterm success in mediation. Unpublished manuscript, State University of New York at Buffalo. Rasinski, K. A. (1988). Economic justice, political behavior, and American political values. Social Justice Research, 2, 61-79. Rasinski, K. A,, & bier, T. R. (1987). Fairness and vote choice in the 1984 Presidential election. American Politics Quarrerly. 16, 5-24.

190

TOM R. TYLER AND E. ALLAN LlND

Rasinski, K. A,, 'Qler, T. R., & Fridkin, K. (1985). Exploring dimensions of legitimacy: The mediating effects of personal and nonpersonal legitimacy on leadership endorsement and system support. Journal of Personaliry and Social Psychology, 49, 386-394. Raven, B. H., &French, J. R. P., Jr. (1958). Group support, legitimate power, and social influence. Journal of Personaliry, 21, 400-409. Robinson, J. P., & Shaver, P. R. (1973). Measures of social psychological attitudes (rev. ed.). Ann Arbor, MI: Survey Reseatch Center, Institute for Social Research. Samuelson, C. D., Messick, D. M., Rune, C. G., & Wilke, H. (1984). Individual and structural solutions to resource dilemmas in two cultures. Journal of Personality and Social Psychology, 47, 94-104. Schachter, S . (195 1). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology. 46, 190-207. Sears, D. 0. (1986). College sophomores in the laboratory: Influences of a narrow data base on psychology's view of human nature. Journal of Personality and Social Psychology. 51, 5 15530. Sears, D. 0.. 'Qler, T. R., Citrin, J., & Kinder, D. R. (1978). Political system support and public response to the energy crisis. American Journal of Political Science, 22, 56-82. Sheppard, B. H. (1984). Third-party conflict intervention: A procedural framework. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 6, pp. 141-190). Greenwich, CT: JAI Press. Sheppard, B. H., & Lewicki, R. J. (1987). Toward general principles of managerial fairness. Social Justice Research, 1, 161-176. Sherif, M., Harvey, 0. J., White, B. J., Hood, W. R., & Sherif, C. W. (1961). Intergroup conflict and cooperarion. Norman: University of Oklahoma Book Exchange. Sherif, M., & Sherif, C. (1953). Groups in harmony and tension. New York: Harper. Stouffer, S. A,, Suchman, E. A., DeVinney, L. C., Star, S. A., & Williams, R. M., Jr. (1949). The American soldier: Adjustment during army life (Vol. 1). Princeton, NJ: Princeton University Press. Strickland, L. H. (1958). Surveillance and trust. Journal of Personality, 26, 200-215. Tajfel, H. (1981). Social stereotypes and social groups. In I. T h e r & H. Giles (Eds.), Intergroup behavior. Chicago: University of Chicago Press. Tajfel, H., & Turner, J. (1979). An integrative theory of intergroup conflict. In W. Austin & S. Worchel (Eds.), The social psychology of intergroup relations. Monterey, CA: Brooks/Cole. Taylor, D. M., Wright, S. C., Moghaddam, D. M.,& Lalonde, R. N. (1990). The personal/group discrimination discrepancy. Personality and Social Psychology Bulletin, 16, 254-262. Thibaut, J. (1968). The development of contractual norms in bargaining: Replication and variation. Journal of ConJict Resolution, 12, 102- 112. Thibaut, J., & Faucheux, C. (1965). The development of contractual norms in a bargaining situation under two types of stress. Journal of Experimental Social Psychology, 1, 89-102. Thibaut, I., & Gruder, C. L. (1969). Formation of contractual agreements between parties of unequal power, Journal of Personality and Sociul Psychology, 11, 59-65. Thibaut, J., & Kelley, H. H. (1959). The social psychology of groups. New York Wiley. Thibaut, J., & Walker, L. (1975). Procedural justice: A psychological analysis. Hillsdale, NJ: Eribaum. Thibaut, I., & Walker, L. (1978). A theory of procedure. California Law Review, 66, 541566. Thomas, C. W., & Hyman, J. M. (1977). Perceptions of crime, fear of victimization, and public perceptions of police performance. Journal of Police Science and Public Administration, 5, 305-3 16. 'hrner, J. C. (1987). Rediscovering the social group: A self-categorization theory. New Yo&: Basil Blackwell.

AUTHORITY IN GROUPS

191

bier, T. R. (1984). The role of perceived injustice in defendant’s evaluations of their courtroom experience. Law and Society Review, 18, 51-74. Qler, T. R. (1987). Conditions leading to value expression effects in judgments of procedural justice: A test of four models. Journal of Personality and Social Psychology, 52, 333-344. Tyler, T. R. (1988). What is procedural justice?: Criteria used by citizens to assess the fairness of legal procedures. Law and Society Review, 22, 301-355. bler, T. R. (1989). The psychology of procedural justice: A test of the group value model. Journal of Personality and Social Psychology, 57, 333-344. Qler, T. R. (1990). Why people obey the law: Procedural justice, legitimacy. and compliance. New Haven, CT: Yale University Press. bler, T. R.. & Bies, R. (1990). Interpersonal aspects of procedural justice. In J. S. Carroll (Ed,), Applied social psychology in business settings. Hillsdale, NJ: Erlbaum. Tyler, T. R., & Caine, A. (1981). The role of distributional and procedural fairness in the endorsement of formal leaders. Journal of Personality and Social Psychology, 41, 642-655. Tyler, T. R., Casper, J. D., & Fisher, B. (1989). Maintaining allegiance toward political authorities: The role of prior attitudes and the use of fair procedures. American Journal of Political Science, 33, 629-652. Tyler, T. R., & Folger, R. (1980). Distributional and procedural aspects of satisfaction with citizenpolice encounters. Basic and Applied Social Psychology, 1, 281-292. Qler, T. R., & Lind, E. A. (1990). Intrinsic versus community-basedjustice models: When does group membership matter? Journal of Social Issues. 46, 83-94. Tyler, T. R., Rasinski, K., & McGraw, K. (1985).The influence of perceived injustice on support for political authorities. Journal of Applied Social Psychology. 15, 700-725. Tyler, T. R., Rasinski, K.. & Spodick, N. (1985). The influence of voice on satisfaction with leaders: Exploring the meaning of process control. Journal of Personality and Social Psychology, 48, 72-8 I . Tyler, T. R., & Schuller, R. (1990). A relational model of authority in work organizations: The psychology of procedural justice. Unpublished manuscript, American Bar Foundation. Qler, T.R., & Weber, R. (1982). Support for the death penalty: Instrumental response to crime or symbolic attitude. L a w and Society Review, 17, 21-45. Vidmar, N. (1990). The origins and consequences of procedural fairness. Law and Social Inquiry. 15, 877-892. Walker, L., LaTour, S., Lind, E. A., & Thibaut, J. (1974). Reactions of participants and observers to modes of adjudication. Journal of Applied Social Psychology, 4, 295-3 10. Walster, E., Berscheid, E., & Walster, G. W. (1973). New directions in equity research. Journal of Personality and Social Psychology, 25, 15 1- 176. Walster, E., & Walster, G. W. (1975). Equity and social justice. Journalof SocialIssues, 31, 21-43. Walster. E., Walster, G. W.. & Berscheid, E. (1978). Equity: Theory nnd research. Boston: Allyn & Bacon. Wright, J. (1981). Political disaffection. In S. Long (Ed.), Handbook of political behavior. New York: Plenum.

This Page Intentionally Left Blank

YOU CAN’T ALWAYS THINK WHAT YOU WANT: PROBLEMS IN THE SUPPRESSION OF UNWANTED THOUGHTS Daniel M. Wegner

1. Introduction Try to pose for yourself this task: not to think of a polar bear, and you will see that the cursed thing will come to mind every minute. (Fyodor Dostoevsky, Winrer Notes on Summer Impressions) Trouble is, a guy tries to shove it out of his head. That don’t work. What you got to do is kind of welcome it. (John Steinbeck, The Winrer of Our Disconrenr)

It is not easy to suppress a thought. One can demonstrate this to oneself in a moment merely by resolving to stop a thought about something, anything. The thought at once seems more important for this resolve, and even when other thoughts are brought to mind as distracters, the unwanted thought intrudes again and again. Still, it can be tempting to suppress thoughts under certain conditions. We imagine the placid mental landscapes that await us should we escape the unwanted thought. Ahead lies freedom from traumatic memories, from unpleasant emotions, from hateful addictions or untoward impulses, from fears, obsessions, and worries about the future-and we decide to suppress thoughts despite whatever adversity this choice might portend. This article is about this psychological deadlock-between the urge to suppress and the daunting task of suppression. A general account of thought suppression is offered that considers the pressures that instigate suppression, the process of suppression, and postsuppression processes and consequences. This account treats thought suppression as one of a class of phenomena of mental control-conscious attempts to control psychological contents and processes (Wegner, 1989; Wegner & Pennebaker, in press; Wegner & Schneider, 1989),

ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOWY. VOL. 25

I93

Copyright 0 1992 by Academic Press. Inc. All nghls of repmduclion in any form rcurved.

194

DANIEL M. WEGNER

and suggests how thought suppression may be involved in the larger enterprise of the self-control of emotion, action, and communication in social life.

11. Suppression in the Shadow of Repression

Sigmund Freud is responsible both for what we know about thought suppression, and for what we do not know. He became responsible for what we know by initiating the study of repression. In his most sweeping definition of this process, he held that “the essence of repression lies simply in the function of rejecting and keeping something out of consciousness” (Freud, 191511957, p. 105). Repression thus captured a variety of forms of motivated unawareness, both intentional and unintentional, with forgetting and without it. Although Freud’s study of repression has installed it in the vocabularies of psychologists everywhere, his focus on examples that entail unconsciously motivated, unintentional memory loss has yielded a modem psychology that is curiously stilted in its treatment of the conscious avoidance of unwanted thoughts (Erdelyi, 1990). In current use, repression is the term for unintentional, unconscious forgetting, and suppression has come to be understood as the intentional, conscious removal of a thought from subsequent conscious attention. Suppression is almost never studied, whereas a wide literature has formed around the concept of repression. Repression does seem to carry a certain excitement and mystique, perhaps because it is believed to involve processes that not only push thoughts out of conscious awareness but even erase memory from a position deep within the realm of unconscious motives. Although repression and suppression processes are by no means mutually exclusive or contradictory, the interest value of repression has tended to exclude suppression from scientific study. It is in this sense that Freud was responsible for what we do not know about suppression. Practically, the focus on a narrowly defined version of repression has been a nightmare. Examinations of processes of disattention have become limited to the study of an odd psychological event that, by definition, must occur without awareness, must spring from motives the person cannot acknowledge, and must result in the eradication of both some particular memory trace and the trace representing the event of repression itself. One of the few natural methods for studying repression so defined has been to ignore process and assume that it somehow happens to some people more than others. Such individual-difference research examines the characteristics of individuals who do not report thinking about unpleasant thoughts. Although this approach has yielded evidence of repression as a trait, in that reliable measurements of it can be made that converge with other measures of theoretical relevance (see, e.g., Davis, 1987; Weinberger,

THOUGHT SUPPRESSION

195

Schwartz, & Davidson, 1979), it remains the study of the absence of reported negative affect rather than the study of any active repressive process. The presumption of a powerful repressive process has been so sweeping, though, that research investigating it has carried on for many years in the face of little supportive evidence. It is now generally realized, however, that outside of case reports of some clinical amnesic phenomena (e.g., Breznitz, 1983; Rapaport, 1959), evidence for repression is thin indeed. Laboratory research indicates that amnesia induced through ego threat (Holmes, 1974, 1990),threat of physical harm (Sears, 1943), hypnosis (Kihlstrom, 1983), and even direct instruction (Bjork, 1989; Geiselman, Bjork, & Fishman, 1983; Weiner, 1966) remains at least an order of magnitude less pronounced than the memory losses envisioned by Freud. Even researchers who are sympathetic to the concept of repression still are not convinced by their own attempts at experimental demonstration (Erdelyi, 1990; Erdelyi & Goldberg, 1979). It is time to reconsider suppression. Thorough, unexpected memory loss, after all, is not the sole criterion of a person’s struggle to overcome unwanted thoughts (Greenwald, 1988). It may be that people do not commonly experience amnesia for unwanted thoughts. Rather, conscious struggles may be undertaken repeatedly against unwanted thoughts in everyday life in hopes that these thoughts can be kept from further conscious awareness. Conscious efforts to suppress unwanted thoughts are reported by normal individuals (Rachman & de Silva, 1978), as well as by those suffering from problems such as depression (Turner, Beidel, & Nathan, 1985), anxiety and worry (Borkovec, Robinson, Pruzinsky, & DePree, 1983), obsessions and compulsions (Rachman & Hodgson, 1980; Reed, 1985), addiction (Marlatt & Parks, 1982), obesity (Polivy, 1990), and posttraumatic stress (Ellis, 1983; Horowitz, 1975; Pennebaker, 1988; Tait & Silver, 1989). It makes sense that the consciously instigated mental disturbances accompanying suppression could have greater psychological impact than the unaccountable disappearances of memory accompanying repression, even though repression has long overshadowed suppression in the literature of psychology. Suppression is a street battle raging in the forefront of the mind, not a few random muggings hidden deep in its alleyways. The added benefit of this approach is thaE suppression is far easier to study. Because the onset and offset of conscious suppression can be precisely controlled through instruction, and because the course of its operation can be examined through standard self-report techniques and other probes, its full range of influence can be mapped. The study of suppression can begin where the study of repression seems to have come to a standstill. All that is needed are assumptions that (1) people can control their attempts to suppress unwanted thoughts, at least for the purposes of investigation, (2) they are able to report recurrences of the unwanted thoughts, and (3) the interesting result of attempted suppression is not

196

DANIEL M. WEGNER

always memory loss, but rather simple inattention to the unwanted thought. Far from the folly of looking for a lost object where the light is better, however, this emphasis on suppression may allow us for the first time to look beyond the shadow of repression to where there is any light at all.

111. Instigation of Suppression

Although it may seem that we elect to suppress thoughts arbitrarily, controlling ourselves by whim or simple choice, we are normally brought to this pass by very predictable conditions that place us in a unique state of mind. It is important to examine the character of these circumstances, and to explore the state of suppression readiness as well, if we are to understand the natural progression leading to each instance of suppression.

A. CIRCUMSTANCES OF SUPPRESSION Thought suppression is provoked when the person’s situation prompts the inhibition of some external expression of a thought. Normally, the external expressions people attempt to forestall in this way take the form of actions, communications, or emotional expressions. In each case, thought suppression is attempted as a preemptive strategy aimed at inhibition of the overt psychological consequences of the thought. 1. Action Inhibition

Thought suppression is used for behavioral self-control. When people are dieting (Polivy & Herman, 1985) or trying to abstain from addictive substances (Marlatt & Parks, 1982), for example, they may attempt thought suppression in these efforts. People suffering from unwanted compulsive actions report using thought suppression as a prophylaxis as well (Rachman & de Silva, 1978; Rachman & Hodgson, 1980). Although thought suppression might be of little use for avoiding involuntary acts such as sneezing or being startled, it could often be useful as a strategy for the inhibition of voluntary actions. The most well-developed research program on the inhibition of unwanted action is the work on delay of gratification by Mischel (see Mischel, Shoda, & Rodriguez, 1989). These studies indicate in general that the self-control of behavior-in particular, foregoing a small immediate reward for a larger delayed reward-is strongly influenced by the person’s thinking. The relationship between thinking and self-control is not always simple. Directing people to think

THOUGHT SUPPRESSION

197

about rewards by making them available for viewing (Mischel & Ebbesen, 1970), or by offering reminders of them (Mischel, Ebbesen, & Zeiss, 1972), for example, substantially undermines behavioral self-control among children. Presenting children with the rewards in an “abstract” format that deemphasizes their rewarding properties, however, (e.g., thinking of pretzel sticks as long, thin brown logs) can reverse this effect and in fact promote delay of gratification (Mischel & Baker, 1975). It seems, then, that the need to suppress thoughts should be most demanding when abstraction and other forms of cognitive distancing are difficult and the immediate pressure to act is strong. When this happens, people may intuit that thinking can soon lead to acting, and so attempt to forestall the process by stopping the thought of action. A child in one of these studies who sees the reward at arm’s length just crying out to be taken, for example, may find thought suppression is the last line of defense, the only apparent avenue to behavior control. Other examples suggest the same principle. Someone who wants to avoid striking out in anger, for instance, may find it particularly effective to suppress thoughts of this act only when the target of the anger is in close proximity (and sticking out its tongue). When the behavior is not in imminent danger of being elicited, thought is really not a problem. Allowing oneself the luxury of thoughts of chocolate, prime rib, or sex with forbidden partners, after all, may be highly enjoyable as long as one is locked in a vault at the time. Thoughts of these things just when such proscribed behaviors are immediately available, on the other hand, could be quite unwanted. It is also worth noting that thought suppression can be prompted when behaviors are desired rather than inhibited. In a class, for example, a student may find that irrelevant daydreams are getting in the way of learning the day’s lesson. Under these conditions, the social pressure to deploy attention appropriately and focus on the lesson can be almost palpable, and the pressure to suppress distracting thoughts must be proportionately strong. Because thought suppression is frequently instigated in service of concentration, those circumstances that promote concentration will bring suppression along (Wegner & Schneider, 1989). So, thought suppression may emerge both to aid behavioral inhibition and to foster behavior production. 2 . Communication Inhibition The desire to suppress a thought can arise in social circumstances as the result of attempts to prevent the verbal and nonverbal disclosure of thoughts or emotions to others. Keeping a secret, for instance, can require thought suppression, especially when one is in the presence of the audience from whom the secret is to be kept (Pennebaker, 1990). Wegner, Lane, and Dimitri (1992) have shown in this regard, for example, that people suppress thoughts of relationships that are

198

DANIEL M. WEGNER

secret. In a survey calling for descriptions of past relationships, respondents indicated that they were more likely to suppress thoughts about former relationships that were once secret-and that such secret loves were a continued target of obsessional thinking later on as well. This was also true for unrequited crushes. Those that were never revealed to anyone were more often the topic of suppression and subsequent obsession than crushes that had been revealed. Laboratory research by Wegner, Lane, and Dimitri was arranged to examine this effect through the manipulation of secrecy. Impromptu mixed-sex couples in this experiment were induced to keep a physical relationship secret. They were asked to engage in nonverbal communication with their feet and ankles under the table-in essence, playing “footsie”-while they competed in a card game with another couple. The couples who were in secret contact in this way reported greater attraction between partners following the session than did couples who did not have any contact at all. The secret contact couples also expressed greater attraction between partners than couples who experienced precisely the same physical relationship, but for whom the relationship was nonsecret. The imposition of secrecy heightened the degree to which partners in a physical relationship became preoccupied with one another, perhaps by virtue of thought suppression that the secrecy induced. Secrecy of sorts is required in many other social entanglements, and thought suppression is likely to result from these as well. Managing others’ impressions of oneself (Gilbert, Krull, & Pelham, 1988), deceiving others (DePaulo, 1992), watching others without their awareness (Olson, Barefoot, & Strickland, 1976), or avoiding prejudicial social judgments (Devine, 1989; Fiske, 1989; Showers & Cantor, 1985) are all ways in which one may be induced to suppress thoughts in the service of appropriate communication to others. These activities require the management of mind in service of social goals, and cannot by their nature be carried on with others’ full knowledge. Keeping thoughts suppressed at times when circumstances make their communication dangerous would seem to be one of the major intrapsychic activities accompanying successful social interaction.

3 . Emotion Inhibition

Thought suppression often occurs in the service of dampening or eliminating emotional expression-particularly if the emotion is negative. Reports of suppression often accompany emotional stimuli such as pain (Cioffi, 1991), loss (Coyne, 1989), or threat (Rachman, 1978), and are found as a result of stress in general (Selye, 1956; Wegner, 1988). Contemporary theories of mood control appear to agree with this assessment, as they converge on the idea that cognitive strategies aimed at the modification of moods are deployed primarily to avoid bad

THOUGHT SUPPRESSION

199

moods and approach good ones (e.g., Clark & Isen, 1982; Klinger, 1982; Zillman, 1988). This much was anticipated by Freud, of course, as his theory holds that processes of repression are instantiated by real or potential negative emotion (Freud, 1915/1957, 191411958). It is possible to move beyond his general idea that suppression arises only to quell the experience of unpleasant emotion, however, if it is recognized that thought suppression might be used to control emotional experiences of any kind-positive or negative-if social pressures militate against emotional expression. There do exist commonplace circumstances in which one may stifle a smile or generate a tear, after all, so a broadened conception of the strategic control of emotion seems necessary. In this view, it is not only negative emotion that instigates thought suppression, but the occurrence of emotional experience and/or expression of any kind that the individual deems somehow too extreme in the current circumstance-and thus worthy of elimination through thought suppression or other means (see, e.g., Hochschild, 1979, 1983; Salovey & Mayer, 1990). There are many social settings, for example, in which both negative and positive emotional expressions might be unsuitable. Lyman and Scott (1968) suggest that “coolness” is generally the most appropriate response to unknown social circumstances, and that the self-control of emotion must often be undertaken to this end. According to Goffman (1963), people who indulge in private emotions in public places are quickly stigmatized; he pointed out that this sort of behavior occurs only in settings allowing deviance (e.g., a mental hospital) or promoting estrangement (e.g., a waiting room filled with strangers). Such observations suggest that mood control through thought suppression may commonly be prompted by the anticipation of social interaction. When people expect to meet and interact with others, they may be inclined to avoid moods of any kind in the search for a self-presentation of emotional collectedness. Evidence that people do avoid moods in social situations has been provided by Erber and Wegner (1992). In this research, musical mood inductions were used to place subjects in good or bad moods, and subjects were led to expect that they would soon interact with another person or that they would spend some time alone. Subjects’ mood control strategies were then measured by allowing them to choose from among a variety of newspaper articles the ones they would prefer to read. Headlines were presented that had identified the articles in pretesting as depressing (e.g., “Man facing death penalty for killing tot”), neutral (e.g., “Shuttle workers load Galileo on Atlantis”), or cheerful (e.g., “Adventurer aborts attempt to cross Bering Strait in a tub”), and subjects’ selections were recorded. Those subjects who anticipated being alone selected articles to read that were consistent with their induced mood. However, those who anticipated social interaction selected articles that contrasted with their current mood-

200

DANIEL M. WEGNER

seemingly in an attempt to neutralize their mood in service of the upcoming interaction. This was even the case with subjects in a positive mood; they selected depressing articles in preparation for their interaction. Results of this kind suggest that there may be a variety of sources of mental control or thought suppression that transcend simple hedonic origins. Thought suppression may not always move the person directly toward positive affect, at least immediately, and may sometimes even seem to undermine pleasure. In essence, then, whereas extreme affective expression may well be allowed and even indulged at length when people are alone in their rooms, it must be cut short in many of the more constraining situations of everyday life. The suppression of thoughts as a preventative measure against emotional expression seems most likely to occur when the emotion expression is situationally maladroit. In sum, the circumstances of thought suppression are familiar to social psychologists as the social pressures that compel individuals to control themselves for social purposes. A history of psychoanalytic themes in thought suppression theorizing has often blurred this realization, emphasizing emotion and intrapsychic conflict at the expense of a full assessment of the multiple social pressures that regularly motivate individuals to seek detachment from their own thoughts. People seek to avoid thoughts because thoughts must be stopped to avoid the expression of unacceptable inner workings to others (Wegner & Erber, 1993). Thought suppression becomes an internal activity that prevents external expressions that people cannot or will not allow. This means that suppression is likely to be quite responsive to variations in social circumstance, and yet will over time become a habitual response to social settings that afford no options other than inhibition.

B. SUPPRESSION READINESS William James (1890) observed that consciousness appears fluid to us, a stream that is steadily flowing and always changing. This means that at any one point, we usually have more than one thought in mind. We may be focused on one idea, but meanwhile have in mind the “fringes” of other thoughts-the trails of thoughts that are now largely over and the hints and beginnings of thoughts that are soon to come. As we make our way along the stream of thoughts, we give each attention in turn, but never give all of our attention to any one thought. The contemporary literatures on conscious flow (Klinger, 1978; Singer, 1988) and on priming effects in memory (e.g., Marcel, 1983) provide ample evidence in support of James’s insight. People seem to think many things at once, and at many levels. The state of mind that prompts suppression, however, seems to represent an

THOUGHT SUPPRESSION

20 1

exception to this rule. The circumstances that precede suppression lead to entrainment of attention on some single thought. There is a tendency to focus on one thought in the experience of strong emotion, in the impulse to perform a compelling behavior, or in concentration on some burning secret. The pressure to express the thought meets head-on the desire to inhibit the expression, and the collision leads to a deadlock, a prolonged focus on the thought of what may imminently be felt, done, or said. In essence, then, we suppress a thought when we find ourselves with only one thought in mind-the thought we fear will produce an undesired expression of some kind. Historically, this mental state, the “fixed idea,” has been accorded broad importance as a source of psychopathology. James claimed that a wide array of exceptional mental states-from dreaming and trance to genius and demonic possession-could be traced to the case of narrowing of the normal range of consciousness. He observed that “In a healthy life, there are no single ideas” (Taylor, 1983, p. 18). Janet (1907) was similarly impressed by the ubiquity of the idPejxe in mental dysfunction, and held that it produced dissociative states of mind. More recently, Minsky (1987) has elaborated similar logic in suggesting that the mental equivalents of computer program “bugs” can bring thinking to a stop on some single idea and leave the person without a pathway to another thought. All of these views emphasize the role of attention entrainment in thought suppression. Quite simply, it seems that we attempt to stop a thought when the thought stops us. The state of suppression readiness, in other words, is a state in which the individual has become focused on some single idea-one that normally would lead to some unwanted act, emotion, or communication. This state is described in rich detail by a population of individuals who have significant experience with unwanted thoughts-people suffering from obsessivecompulsive disorder. They recount the dramatic paralysis of attention that occurs when the unwanted thought intrudes (e.g., Rapoport, 1989; Toates, 1990). Thought suppression is undertaken, then, both for these special cases and for normal individuals, as a controlled attempt at attention management designed to draw attention away from its one-track focus toward something else. The only escape from a fixed idea seems to be through thought suppression. As a first step toward suppression, one normally moves from direct awareness of the idea to a reflective awareness of one’s relationship to the idea (“I am thinking this thought too much”). Such reflective awareness is still fixed on the idea, of course, as it retains the idea even though it expresses a desire to avoid it; knowing that one is thinking about chocolate too much, for example, is also a case of thinking about chocolate. Ultimately, however, it is this reflective awareness that appears to prompt the attempt to dispel the thought and release the stream of consciousness to flow elsewhere. The fixed idea is pushed away following one’s realization that one’s thinking has become stalled on it.

202

DANIEL M. WEGNER

IV. Thought Suppression Processes How then does thought suppression proceed? It is likely that different kinds of suppression processes are involved depending on the timing of the suppression attempt. In what might be termed momenrary suppression, there is an attempt to stop a thought just as the thought is prompted by a stimulus. Logan (1983, 1988) and Logan and Cowan (1984) have made a series of studies of this process by the use of stop signals presented during various cognitive tasks. The conclusion of this research is that while actions can often be inhibited in midstream, thoughts once underway carry on to completion in a ballistic fashion. The processes of continuous suppression, on the other hand, appear to allow a somewhat greater measure of success. Given a period of minutes or hours to suppress a thought, people can attain lengthy intervals of effective distraction by adopting strategies that allow them to preempt future occurrences of the thought. When individuals are asked to report their thoughts as they try not to think of a white bear, for instance, they indicate multiple occurrences of the thought but not its constant presence (Wegner, Schneider, Carter, & White, 1987). This state of continuous suppression resembles the chronic dilemma people confront when they experience unwanted thoughts in everyday life, and it is this state that invites further analysis.

A. THE SUPPRESSION CYCLE The exercise of continuous thought suppression is reminiscent of the task of Sisyphus, the figure of Greek myth who was condemned forever to push a stone up a mountain only to have it roll back down. As a rule, people asked to suppress a thought (e.g., of a white bear) while they deliver verbal reports of their thinking show a cyclic sequence of activities with precisely this up-and-down character (Wegner et a]., 1987). Suppression begins with a plan to self-distract (e.g., “Okay, I’ll think of something else”), continues with the choice of a distracter (e.g., “I’ll think about the telephone”) and a period of concentration on distracter-relevant thoughts (e.g., “This phone is a portable, etc.”), and is completed with an intrusive return of the suppressed thought (e.g., “There’s the white bear again”). The cycle then repeats with a return to the plan to selfdistract (i.e., “Okay, I’ll think of something else”). It is possible to understand this cyclic reiteration as a product of the interplay of two cognitive processes instantiated by the intention to suppress (Wegner & Erber, in press). One is a controlled process, the other is automatic (see Bargh, 1984; Hasher & Zacks, 1979; Logan, 1988; Posner & Snyder, 1975; Shiffrin & Schneider, 1977). The first process is a controlled distracter search-a con-

THOUGHT SUPPRESSION

203

scious, attention-demanding search for thoughts that are not the unwanted thought. This process involves a significant allocation of cognitive resources to the task of finding distracters and maintaining their presence in mind by reviewing thoughts about them. The second process is an automatic target search-a relatively less attention-demanding process that searches for any sign of the unwanted thought. The controlled distracter search operates to replace the unwanted thought with a different thought. It is the more obvious of the two processes, as it is clear that this is what we are doing when we try to “think of something else.” A great deal of our attention can be invested in the pursuit of appropriate distracters during suppression (Gilbert et al., 1988), and this expenditure of cognitive capacity is what qualifies the process as “controlled” (see, e.g., Bargh, 1984). The automatic target search is a somewhat less obvious feature of intentional suppression, but its role in the suppression process is entailed by the need to prompt the operation of the controlled distracter search. The automatic target search is set to detect whether the controlled distracter search is necessary. To carry out the intention to suppress the thought, after all, requires that failures of the intention must be detected and then counteracted by the reinstatement of the controlled distracter search. This can happen if there exists an automatic process that tests constantly for indications of the unwanted thought. The two processes that operate in suppression can be understood as components of a feedback mechanism aimed at the control of thought (cf. Uleman, 1989). Normally, control systems that operate through feedback are composed of an operating process and a test process that detects the need for the operating process (Carver & Scheier, 1981; Miller, Galanter, & Pribram, 1960). The controlled search for distracters is the mental operating process that carries out the suppression, and this is the part of suppressing that seems effortful and requires conscious attentional guidance. Potential distracters must be identified, one must be chosen, and it must then be explored and elaborated-and all these subprocesses take cognitive capacity. The automatic target search can be understood as a relatively less effortful mechanism enlisted to test whether the operation of the controlled distracter search is needed at any particular moment. Like the “test” and “operate” components of a “test-operate-test-exit (TOTE)” unit (Miller et al., 1960), these two processes in combination create a purposive system that functions to suppress a thought. The interplay of the two processes is responsible for the cycle of suppression. Beginning with the intention to suppress, both processes are initialized. The automatic target search immediately indicates that the target is in consciousness, and thus initiates the controlled distracter search. The controlled distracter search brings a series of distracters to mind until one is selected that absorbs attention. At this point, attention is drawn from the controlled distracter search to the absorbing distracter itself, and in this sense the controlled distracter search is no

204

DANIEL M. WEGNER

longer functioning. There occurs here a plateau of indeterminate length that may be called “successful” suppression. In the background, however, the automatic target search is still looking continually for signs of the target in consciousness. When a reminder is encountered, however vague or distant, the automatic target search is tuned to register this discovery, and so to return the unwanted thought to consciousness. This reintroduces the cycle from the start and the controlled distracter search begins again. The automatic target search that is instituted to test for failures of the controlled distracter search, therefore, will often produce these very failures. The automatic target search should act in the same way as an externally encountered prime to make the unwanted thought highly accessible. It is usually observed in priming paradigms that the nonconscious priming of a thought promotes assimilation of incoming information to that thought (e.g., Lombardi, Higgins, & Bargh, 1987; Martin, 1986). During successful suppression, the automatic target search is indeed nonconscious, and it therefore could render the suppressed thought highly accessible to consciousness. The thought and its close associates should be activated with minimal prompting, and this should yield frequent remindings that in fact return the thought to conscious attention. Like other automatic processes, the automatic target search cannot be consciously inhibited (e.g., Logan & Cowan, 1984). So, it churns on for the entire suppression cycle, even when suitable distracters have been found and the individual has attained a plateau of successful suppression. This process thus explains the repeated intrusion of the thought both during the controlled distracter search and later when suppression has been achieved. There is evidence for the operation of both the controlled distracter search and the automatic target search in suppression, and there is evidence about several features of these searches that may impinge on the nature and success of suppression. Evidence relevant to each search process is considered in turn.

B. CONTROLLED DISTRACTER SEARCH The conscious search for distracters draws upon the individual’s perception and memory to replace the unwanted thought with another thought. This search is thus dependent on the richness of the perceived environment and on the accessibility of memories for its effectiveness. Certainly, the availability of potential distracters in the environment influences suppression directly. Knutson (1990) has found, for example, that people asked to shut their eyes and report their thinking during suppression indicate more frequent occurrences of the unwanted thought than do those allowed to keep their eyes open. If the person is in a particular environmental context, distracters reported during suppression will often be drawn from stimuli in this context (Wegner, Schneider, Knutson, & McMahon, 1991).

THOUGHT SUPPRESSION

205

The individual also turns to memory during suppression, and items are retrieved that might serve as reasonable distracters. The items chosen are, of course, highly reflective of the person's current mental contents. Items that are chronically or acutely accessible are likely to be retrieved and enlisted as distracters. If the person is suffering from depression, for example, it is likely that depressing distracters will be chosen (Wenzlaff, Wegner, & Roper, 1988). If the person is in a temporary good or bad mood, mood-relevant thoughts will often be selected as distracters (Wenzlaff, Wegner, & Klein, 1991). As a rule, people do not focus on one distracter alone. Although most past research on distraction per se has examined what happens to perception when people are given a single distracting stimulus or are asked to focus on a single thought (McCaul & Malott, 1984), this is not what people do naturally when they suppress a thought. Rather, a person suppressing a thought seems to be prompted by the cyclic recurrence of the thought to select a new distracter at each new failure to suppress (Wegner et al., 1987). In this sense, natural self-distraction appears to be unfocused rather than focused, meandering rather than centered on a target. Although unfocused search seems to yield only a slight disadvantage in suppression effectiveness in the short intervals of suppression studied in the laboratory (e.g., Wegner et al., 1987, Expt. 2), it could be that the failure to find a focus might produce significant deficits in suppression effectiveness in the longer term. If no one absorbing distracter is found, prolonged suppression will make it difficult to assemble a thematic approach for what to think or do on any given day. During unfocused self-distraction, targets of thought seem remarkably uninteresting. Nothing can be found that naturally absorbs attention, and one experiences the world as boring, shallow, and without value. In OfHuman Bondage, Maugham describes this state in a young man who is trying not to think about his lover's infidelity: He took a book and began to read . . . but he could make no sense of it, for his mind was elsewhere. He set his teeth and read on; he tried desperately to concentrate his attention; the sentences etched themselves in his brain by the force of his effort, but they were distorted by the agony he was enduring. . . . He discovered that the three pages he had read made no impression on him at all; and he went back and started from the beginning; he found himself reading one sentence over and over again; and now it weaved itself in his thoughts, horribly, like some formula in a nightmare. . . . He could not read any more now. He simply could not see the words. (1915/1991, pp. 386-387)

Unfocused self-distraction during suppression does have the interesting feature of making the person highly attentive to a variety of stimuli that might otherwise escape notice. The odd environmental object or event, the stray reminiscence, the whimsical hope, may each take center stage for the moment when suppression is the cardinal goal. This does not mean, however, that the controlled distracter search is entirely uninformed of the suppression target. Rather, distrac-

206

DANIEL M. WEGNER

ters may be sought according to some criterion of greatest anticipated semantic or affective distance from the suppression target. Wenzlaff et al. (1988) found that normal subjects attempting to suppress positive thoughts selected negative distracters when given a prepared list of positive, neutral, and negative topics, and generated negative distracters when such a list was not available. These subjects also selected and generated positive distracters when they were assigned the task of suppressing negative thoughts. Depressed subjects in this research, however, showed a specific deficit in their controlled distracter search. They were less able than other subjects to generate positive distracters when they were asked to suppress negative thoughts. (This deficit was not present when they were selecting distracters from a prepared list, so it is clear they had trouble in the attempt to generate the positive distracters rather than in the attempt to choose them.) It is not surprising, given this deficit, that depressed individuals in this research were found to suffer an unusual resurgence of an unwanted negative thought several minutes after they had started to suppress it (see also Howell & Conway, 1991). Unlike nondepressed people, the depressed subjects attempted to distract themselves from a negative thought with yet other negative thoughts-and so quickly found that the unwanted negative thought returned. Their distracters may have been too near the unwanted thought in affective tone, and so acted later as reminders and increased the likelihood of the unwanted thought’s recurrence. Depressed individuals did not suffer an inability to generate negative distracters to aid in the suppression of positive thoughts, and so were not subject to any general inability to suppress thoughts. The research to date on controlled distracter search suggests, then, that the search (1) is assisted by the external availability of potentially distracting stimuli, (2) draws on current mental contents or accessible items, (3) is typically unfocused rather than focused on one or a few stimuli, (4)may be more successful when it inclines toward distracters that are semantically or affectively distant from the unwanted thought, and ( 5 ) may be dependent on the individual’s capacity to generate new distracters internally that are sufficiently distant in this way.

C. AUTOMATIC TARGET SEARCH The operation of an automatic search for the unwanted thought during suppression can be inferred most obviously from the great frequency with which a suppressed thought recurs. For example, an undergraduate student I asked to think aloud while she tried not to think of the Statue of Liberty began talking about the books on the wall of the room and in a matter of moments mentioned that one was just the greenish tint of the Statue. She then spoke of her upcoming

THOUGHT SUPPRESSION

207

soccer practice and, in short order, added that the soccer team once visited New York but had not been to see where the Statue stands. In all of this, she was aware of the implications of these perceptions for her failure to suppress, yet remained profoundly sensitive to Statue relevance in almost everything. Observations of this kind have parallels in the various studies in which people think aloud during suppression. When people are asked to ring a bell to indicate the occurrence of the thought (Wegner et al., 1987), for example, the bell tolls again and again. When people sit in privacy and try not to think of an exciting thought such as sex, their electrodermal responses (skin conductance levels) indicate continued excitement-and by implication, the continued recurrence of the thought (Wegner, Shortt, Blake, & Page, 1990). Just because a mental process happens over and over, of course, does not necessarily mean it is automatic, and for this reason it is worth considering the evidence for the automatic target search in two specific domains-the cognitive hyperaccessibility of suppressed thoughts and psychophysiological dishabituation to suppressed thoughts.

1. Hyperaccessibility If the automatic target search is normally balanced by the controlled distracter search during suppression, the imposition of a cognitive load should undermine the controlled process and release the automatic target search to operate unimpeded. This means that suppression under cognitive load should make the suppressed thought unusually accessible to consciousness, likely to participate in perception and judgment far more than other thoughts. This idea was examined in two experiments by Wegner and Erber (in press). In Experiment 1, subjects were asked either to suppress or to concentrate on a particular target word (e.g.. house) as they performed a word association task. Some of their associative responses were called for under time pressure (a 3-sec response time limit) while others were prompted without such pressure (a 10-sec limit). Time pressure disturbs the operation of controlled, resource-dependent processes, while interfering only little with the operation of automatic processes (see, e.g., Bargh & Thein, 1985; Strack, Erber, & Wicklund, 1982). Over several trials, then, subjects’ tendency to respond with the target word to related prompts (e.g., home) was observed. Suppressing subjects who were under time pressure to report associates responded frequently with the target word to target-relevant prompts-blurting out the very word they had been trying not to think about (see Fig. I). They did this more often than did suppressing subjects who were not under time pressure to give their associations. Indeed, suppression with time pressure boosted responses of the target word over even the response level of subjects under time pressure who were actively trying to think about the target. This result is consistent with

208

DANIEL M. WEGNER

Fig. I . Prompted target responses per trial. Subjects concentrating on or suppressing a target word were prompted to give the target word in a word association task with either high or low time pressure. Based on data from Wegner and Erber (in press).

an interpretation of suppression in which an automatic target search process is released to sensitize the person to the unwanted thought when the controlled distracter search is undermined by time pressure. A second experiment by Wegner and Erber (in press) tested the hyperaccessibility prediction in the Stroop (1935) color-word interference paradigm. Subjects for this study were asked either to suppress or to concentrate on a target word while responding with key presses to indicate whether words on a computer screen appeared in red or blue. For the manipulation of cognitive load, some subjects did the task as they rehearsed a nine-digit number, whereas others did it while rehearsing only a one-digit number. Those subjects who were suppressing the target under high cognitive load showed interference with color naming when the target word appeared on the screen; their reaction times averaged 600 msec for targets and 561 msec for nontargets. Subjects who were suppressing the target with low load, or who were concentrating on the target in either load condition, showed no evidence of such interference; their reaction times for targets and nontargets differed by no more than 6 msec in any condition. Apparently, suppressing a word with cognitive load promoted hyperaccessibility of the target word. In a related experiment, Wegner and Erber (1991) tested the accessibility of suppressed thought by examining memory for distracters that had been suppressed during study. The intention to suppress a thought sometimes occurs, after

THOUGHT SUPPRESSION

209

all, in service of the desire to concentrate on something else (Wegner & Schneider, 1989). In trying to concentrate on a book, for instance, one might suppress thoughts of the music in the distance. Such implicit suppression during concentration should have the same properties as intentional suppression carried out in response to a direct instruction. Thus, when a cognitive load is imposed, items that are implicitly suppressed should be proffered greater attention. The music should be remembered better than if there had been no load at all. In this investigation, subjects were asked to study a fictitious map consisting of names of little-known African cities. Half the cities were highlighted in blue and half were highlighted in yellow, and subjects were asked to concentrate either on the blue items or on the yellow items. Half performed this task while rehearsing a nine-digit number, and the other half performed it without such rehearsal. Recognition memory for all items was assessed following this, and while the focal items were, of course, recognized more often than the implicitly suppressed ones, the influence of cognitive load on memory for suppressed items was as predicted. Memory was reliably greater for suppressed items studied under load than for those studied with no load. The results of the hyperaccessibility studies indicate that a suppressed thought becomes more rather than less accessible to consciousness when cognitive loads are imposed during suppression. The effect of this is that people say the very word they are trying not to think, attend to the very aspect of a stimulus they are trying not to consider, and encode the very distracting items they had hoped not to commit to memory. When the controlled distracter search is undermined while suppression is still underway, measures of cognitive accessibility indicate that the suppressed thought is very readily brought to mind.

2 . Dishabituation A somewhat different line of evidence on the automatic target search comes from findings on the enhancement of psychophysiological reactivity to an unwanted thought during suppression. As noted earlier, Wegner et a]. (1990, Expt. 1 and Expt. 2) observed that skin conductance level (SCL) becomes elevated for up to 3 min when a person first suppresses an exciting thought-just as it does when the person concentrates on that thought. In research examining the process over longer time intervals (Expt. 3), the typical subject trying not to think about sex was able to do so for several minutes at a time following the initial flush of excitement. And indeed, the mean SCL reduced to normal later in these sessions. The thought returned at a regular rate nonetheless, such that the cumulative record of thought recurrence over 30 min appeared generally linear when averaged across subjects. This was true for exciting thoughts of sex, and likewise for unexciting thoughts of the weather (see Fig. 2).

210

DANIEL M. WEGNER

TARQET THOUGHT

.o.WEATHER

*

0

SEX

I I I I I I I I I I I I I ( I I I I I I I I / I ; ; / ! I 5 10 15 20 25 30

MINUTE Fig. 2. Cumulative mean mentions in suppression. Subjects thinking aloud were asked either not to think about sex or not to think about the weather, but to mention it in case they did. Based on data from Wegner, Shortt, Blake, and Page (1990).

This experiment examined the consequences of further intrusions of the thought by measuring the correlation between SCL and mentions of the thought minute-by-minute as people reported their thoughts aloud over the course of the 30-min session. Subjects suppressing the exciting thought of sex showed a significant positive correlation between SCL and thought mentions over this time period. In essence, the subjects suppressing an exciting thought became reexcited by it each time it came back to mind. This is particularly interesting in light of the finding that no parallel correlation occurred for subjects who were actively trying to concentrate on the thought of sex. Occurrence of the thought during a particular minute yielded no accompanying increment in SCL. Such a correlation was also absent among subjects suppressing or concentrating on the unexciting thought of the weather (probably because the thought of weather was not arousing in either case). The unusual electrodermal reactivity of subjects suppressing the exciting thought suggests the operation of an automatic target search during suppression. Normally, repeated exposure to an exciting thought yields habituation. The person gets used to it. And certainly one would think that over the course of recurrences such as those shown in Fig. 2, such habituation would have plenty of opportunity to take place. The continued psychophysiological reactions to the exciting target item under conditions of suppression in this study reveal a lack of habituation. Although concentration on the exciting thought did drain the thought

THOUGHT SUPPRESSION

21 1

over time of its exciting character, suppression introduced a special vigilance or alertness that prolonged the capacity of the thought to produce excitement. The SCL reactions to intrusions during suppression could be the work of an automatic target search that counters habituation tendencies and keeps the person highly sensitive to the thought’s recurrence. The psychophysiological reactivity, then, marks not just the recurrence of an exciting thought, but the recurrence of an exciting thought that some part of the mind has been seeking. In summary, evidence from very different lines of research converges to suggest that the continuous suppression of a thought is accomplished in part by the operation of an automatic search process that makes the person ironically sensitive to that very thought. Disturbance of the controlled distracter search uncovers this automatic process in one way, in that the accessibilityof the unwanted thought is increased under these conditions. And psychophysiological recording yields evidence for this automatic search process in another way, in that suppression prompts heightened sensitivity to the return of suppressed exciting thoughts.

V. Postsuppression Processes Thought suppression busies the mind with a unique form of tumult. One’s attention careens in turn from a frantic search for distraction, then to disattention, and then to an alarmed recognition that the thought is back-and repeats this ritual again and again as long as suppression is continued. It is natural to suspect that even when this state has been abandoned, there may be subsequent psychological consequences ranging from the trivial to the profound. One family of consequences observed in research arises in the postsuppression rebound of the unwanted thought. When people are encouraged to think about some item that they previously suppressed, they commonly show a remarkable level of preoccupation with it-moreso than if they had never tried to suppress the thought. We turn now to the nature of this effect and its role in other psychological processes.

A. REBOUND EFFECTS

The rebound of a suppressed thought was initially observed by Wegner et al. (1987, Expt. 1) among subjects who had been asked to suppress the thought of a white bear. These subjects individually thought aloud for 5 min and rang a bell if the thought came to mind during suppression. As noted earlier. these subjects typically rang the bell and mentioned whire bear occasionally during this time. When these subjects were next asked to think about a white bear for a similar

212

DANIEL M. WEGNER

interval, they produced more mentions and more bell rings than did subjects who had simply been thinking of a white bear from the start. The rebound effect has since been observed repeatedly for bell rings and mentions during a think-aloud period (Wegner et al., 1987, Expt. 2; Wegner et al., 1991), and has also been observed when individuals in group testing write their ongoing thoughts and make check marks on paper for thought occurrences (Wenzlaff et al., 1991). These experiments have typically contrasted thought frequencies that occur when subjects are asked to think about something with those that occur when subjects are asked to think about something following a period of suppressing that thought. The rebound has also been observed under two other circumstances that suggest it is not limited only to self-reported thinking. In a study by Chandler and Wegner (1987), subjects were asked to write down their thoughts on any or all of a series of topics-some of which were related to white bears (e.g., iceberg) and some of which were not (e.g., gym shorts). Subjects performed this task while trying to think about a white bear, but subjects in one group did this after spending an initial 5-min period in silence under instructions not to think about a white bear. The postsuppression group showed greater preoccupation than did the group that had not suppressed, in that their written associations to the white bear-related prompts were reliably longer than those to unrelated prompts. Subjects who had not previously suppressed the thought of a white bear showed no such tendency. Another observation of the rebound was made in psychophysiologicalresearch by Gold and Wegner (1991). Subjects initially spent 9 min thinking about a former romantic relationship-an “old flame.” As might be expected, those who reported that they were still in love with this person showed an elevated SCL during this period compared with those who reported no lingering attachment. Thinking about “hot” flames induced a greater SCL than did thinking about “cold” flames. Subjects then spent a subsequent 9-min period either suppressing the thought of their old flame or suppressing the thought of the Statue of Liberty. All subjects were then given the opportunity to think again of their old flame in a third 9-min measurement period. As might be expected, the subjects who once again thought about their cold flame remained unexcited by these thoughts regardless of whether they had just suppressed them or had just suppressed the thought of the Statue. The individuals who had been aroused by their initial opportunity to think about the hot flame, but who were then directed to suppress the irrelevant thought of the Statue, were also relatively quiescent. Their SCL did not show any tendency to rise even while they thought about the hot flame once more. The tendency for SCL to become elevated with thoughts about the old flame was only clear in this final period for those subjects who had just previously suppressed the thought of the hot flame-the partner whose absence they lamented. In sum, people showed prolonged reactivity to their still-desired old

THOUGHT SUPPRESSION

213

flame only if they had previously been attempting to suppress thoughts of that person. This result suggests a postsuppression rebound in preoccupation with the suppressed thought. These observations are paralleled by a variety of observations of rebound-like effects in other research. One apparent analog, for example, can be found in the literature on diet-induced binge eating. Polivy and Herman (1985) report several lines of evidence indicating that initial inhibition of eating leads to subsequent eating disorders and obesity. To the degree that thought suppression is involved in inhibition of eating, and obsessive preoccupation is involved in obesity and eating disorders, these investigations suggest a behavioral reflection of the rebound effect. The finding that dieters are particularly inclined to overeat immediately after having broken the diet with a splurge is particularly reminiscent of the rebound effect. High levels of relapse following the violation of abstinence seem to be the rule with addictions such as smoking and alcoholism as well (Marlatt & Parks, 1982), so these phenomena also appear to follow a course like the suppression-induced rebound. Rebounds of emotion, in turn, are commonly observed in the case of inhibited emotional response to traumatic events (e.g., Janis, 1958; Lindemann, 1944; Stiles, 1987). An emotional response to an event such as a health emergency, death of a relative, rape, or other victimization may be blunted for any of a variety of reasons during the event itself, and lack of communication about the emotional quality of the event can follow as well. When this happens, subsequent emotional upheaval signaled by cognitive and physiological distress is often observed. Pennebaker ( 1990) has accumulated significant evidence indicating that the consequences of long-term inhibition can include both psychological disorder and impaired physical health. The rebound of anxiety and arousal has also been found in cases when people have used thought suppression as a strategy for the self-control of phobic or anxiety responses. Borkovec (1974) observed, for example, that snake-phobic individuals during systematic desensitization who mentally rehearsed turning around and running from a snake exhibited exaggerated heartrate responses to snake thoughts later on. Foa and Kozak (1986), Rachman (1980), and Rosenbaum, Beiderman, and Gersten (1989) recount a variety of related demonstrations of acute anxiety and/or fear exacerbated by initial attempts at behavioral or cognitive avoidance. One other set of rebound-like effects is found in the literature on the control of anger and violence. Megargee (1971) reviewed evidence suggesting that violent criminals show two contrasting patterns of response to violence-promoting situations-undercontrolled and overcontrolled. Whereas the undercontrolled individual responds too freely with violence across situations that prompt it only minimally, the overcontrolled person appears to hold back all aggressive tendencies until some one situation triggers an extremely violent episode. A similar pattern has been identified among violent sex offenders, in that suppression of

214

DANIEL M. WEGNER

exposure to sexual stimulation in adolescence appears to predict sexual offenses later on (Goldstein & Kant, 1973). These kinds of rebound effects, of course, are played out over much longer spans of time than those observed in the white bear situation. If the suppression-induced rebound effect observed in the white bear studies indeed operates in the same way as even a few of these many similar effects, however, it is worth understanding in some depth. One possibility is that the rebound occurs because the automatic target search initiated during suppression is not stopped when suppression is intentionally halted. Research examining the accessibility of suppressed thoughts after suppression is discontinued, however, suggests that the automatic target search is indeed brought to an end when the intent to suppress is abandoned (Wegner & Erber, in press). Thus, it is useful to examine the explanation for this effect that has been more clearly suggested by research to date: The rebound effect may be due to the bond that is formed between the unwanted thought and the distracters that are enlisted during the prior suppression. B . DISTRACTERS BECOME REMINDERS Suppression brings to mind many items other than the suppressed thought. As noted earlier, the controlled distracter search is normally unfocused, in that the person turns from one distracter to another and another as each fails to keep the unwanted thought away. The critical feature of such unfocused self-distraction is that it creates associations between the unwanted thought and all the various distracters. If one has focused in turn on the wall, the stock market, and a project of wiggling one’s ears as potential distracters from the thought of a white bear, for instance, the wall, market, and ears are now likely to be reminders of white bears, as is any other distracter that was used. This means that many of the person’s current mental contents become linked to the unwanted thought. These items can then serve as cues to remind the person of the thought in a later expression period when the thought is invited, so to yield the observed rebound effect. The creation of associations between a particular thought target and one’s other mental contents does not happen so readily when one is simply thinking about that target. The person who thinks about a white bear, for instance, may mention Eskimos or the zoo or snow, but will seldom leap to such seeming irrelevancies as the stock market or ear flexing. In a meandering trail of thought, the person might eventually reach such distracters, of course, and they could then become linked to the target. But the sheer frequency of such juxtapositions of target with other mental contents will be far less during concentration on the target than during the unfocused self-distraction that occurs in suppression. The

THOUGHT SUPPRESSION

215

suppression cycle guarantees a relatively constant reverberation of attention between target and distracters, and the connections forged in this fashion later can work both ways. Once formed as paths away from the target, they later can serve as easy paths back to the target. One test of this explanation of the rebound was offered by Wegner et al. ( 1 987, Expt. 2). This study called for some subjects to use afocused self-distraction strategy for suppression. They were told to try not to think of a white bear, but to think of a red Volkswagen in case they did. This instruction was intended to help subjects avoid using their current thoughts and context as distracters, and was expected to produce an attenuation of the rebound effect. In fact, this was observed both for bell rings and for think-aloud mentions. The results showed a rebound effect only among those subjects for whom no special strategy was suggested. Presumably, subjects given the red Volkswagen as a distracting focus were later unlikely to think about red Volkswagens very much during their opportunity to express the unwanted thought, and so escaped the unusual level of contextual reminding that underlies the rebound. If unfocused self-distractionoperates by forging connections between environmental features and the unwanted thought, then the continuity of context between suppression and later expression would seem to be a key condition for the rebound effect. This possibility was tested when Wegner et al. (1991) asked subjects to suppress or express thoughts of white bears in the context of a slide show featuring either classroom scenes or shots of household appliances. Subjects who next expressed white bear thoughts in a different slide-show context showed no evidence of rebound. However, when these same subjects were invited again to express white bear thoughts with the initial slide-show context reinstated, the rebound appeared. Those who had initially suppressed the thought later experienced a rebound of preoccupation with it-but only when they were once again exposed to the slide show during which the suppression had taken place. These findings deeply implicate context in the suppression-induced rebound effect. They indicate that the items on a person’s mind become bonded to the unwanted thought during suppression, such that later reinstatements of context that bring back whose items may have the effect of reintroducing the unwanted thought. At the same time, these findings implicate suppression in the Occurrence of context effects. They suggest that linkages formed between thoughts and a cognitive context may be strengthened when that thought is suppressed. This latter observation led Wenzlaff et al. (1991) to investigate the role of thought suppression in the bonding of thought and mood. They noted that a variety of research programs had examined, with mixed success, the possibility that thoughts experienced while a person is in a particular mood state might be more easily retrieved when that mood was experienced anew (e.g., Bower & Mayer, 1985). Research of this kind had not investigated suppressed thoughts,

216

DANIEL M. WEGNER

however, focusing instead only on thoughts that were given attention during a mood state. If suppression normally links the suppressed thought to context, then suppression of a thought during a mood state should link the thought to the mood such that the later reactivation of one would lead to the reinstatement of the other. In one experiment, Wenzlaff et al. (1991) induced subjects by music to experience either a positive or negative mood, and asked them to report their thoughts in writing while trying to think or not to think about a white bear. Later, all subjects were asked to think about a white bear and write their thoughts during a second mood induction (using different but equally moody music). These thought reports indicated that subjects who experienced similar moods during the periods of thought suppression and expression displayed a particularly strong rebound of the suppressed thought during the expression opportunity. Those who experienced different moods during initial suppression and later expression showed no evidence of a rebound effect. A second experiment tested the complementary connection-whether thought bonded to mood during suppression could later reinstate that mood. Initially, subjects who were in positive, negative, or neutral (music-induced) moods were asked to think or not to think about a white bear. Later, all subjects were asked to think about a white bear for a period, after which they reported their moods. The mood reports showed that subjects who had initially tried to suppress white bear thoughts experienced a reinstatement of the mood state that existed during the initial period of suppression. Those who first expressed white bear thoughts showed no evidence of such reinstatement. These findings, taken together, suggest that suppression creates a remarkable bond between a thought and the cognitive and emotional context in which the thought is suppressed. It would not be surprising if, for example, when a person suppresses some sad thought in an attempt to overcome a depressed mood, the thought becomes more deeply linked to the very mood state that its avoidance was designed to dismiss. In this way, further occurrences of the thought could induce depression, and depressive episodes could rekindle the thought, leading to an unfortunate and perhaps unwanted pathway between negative mental contents. Because suppression would not be as likely to accompany positive mental states or positive thoughts, however, such connections would be less inclined to form between such positive mental contents. This suggests the interesting hypothesis that negative moods are generally more “cognitive”-more populated with mood-congruent thoughts-than are positive moods. The contextualization of suppressed thoughts might be widely responsible for difficulties of self-control. The process could link suppressed thoughts of tobacco, alcohol, or drugs, for instance, with the mental states that accompany withdrawal from these substances, and so build up strong linkages between our cravings and thoughts of the items we crave. In suppressing a thought, the person may unwittingly play a role in making that thought more difficult to dispel in

THOUGHT SUPPRESSION

217

future instances. This suggests that one strategy for overcoming addictions is to quit away from home. Although behavior therapists sometimes suggest that selfcontrol should be conducted at home, in circumstances that will allow adequate generalization of training, there is little supporting evidence for the superiority of this (see, e.g., Polich, Armour, & Braiker, 1981). It may be instead that quitting bad habits in residential treatment facilities offers some advantage because the distracters found at the facility will no longer be present to act as reminders when the person returns home. The transformation of distracters into reminders, in sum, seems to offer a fruitful way of conceptualizing the occurrence of suppression-induced rebound effects. Evidence from several quarters converges to support the conclusion that in suppression, the alternative thoughts that are chosen as distracters themselves become connected to the unwanted thought. They then may provide a wide and strong bridge back to that thought, for whenever they recur the thought may soon cross again into consciousness.

VI. Conclusion When the Rolling Stones sang “You can’t always get what you want,” they provided a bit of relief by adding “but if you try some time, you just might find, you get what you need.” Regrettably, their refrain rings a false note in the world of thought suppression. The research to date suggests that you can’t always think what you want, and that even if you need ever so dearly to stop thinking of something, you still may not get what you need.

A. PROSPECTS FOR SUPPRESSION Suppression is demanding and complicated at a number of levels. First, it requires a careful search for distracting thoughts-a search that can easily go awry. The search for distracters can remain too near the unwanted thought in a semantic sense and so invite its return, or the search for distracters can fail to encounter an absorbing topic, skip from one item to another, and so create links between the unwanted thought and a variety of things that can later serve as reminders. And in the meantime, the search for distracters can be undermined by concurrent cognitive load and thus fail to perform its function assigned by the intention to suppress. If the controlled distracter search fails, the automatic target search surfaces to guide attention, and the unperturbed operation of this process virtually guarantees that the unwanted thought will be back. The difficulty of suppression is further complicated by the presence of emo-

218

DANIEL M. WEGNER

tion. Suppression often occurs in the hopes that emotion stripped of its basis in cognition will somehow dissipate. The emotional thought is suppressed, but despite this measure the physiological concomitants of emotion continue to emerge. These are likely to serve as reminders of the emotional thought, of course, and so they reintroduce the thought in their own way. In the bargain, moreover, the connection between the emotional thought and the emotional reaction is strengthened. Habituation to the thought is prevented by suppression, and intrusions of the thought during suppression serve only to stir up accelerated physiological response. Suppression of exciting thoughts seems to ensure that they will remain exciting. Finally, when the futility of suppression becomes tragically plain and the project is forsworn, the real trouble begins. Given license to think the unthinkable, the person will now be reminded of the formerly unwanted thought by the previous distracters used to dispel it. If the thought was suppressed at home, for instance, familiar things around the house may now serve as unintended reminders of the thought. If the thought was suppressed during a particular mood state, or perhaps because of it, the return of that mood will tend to elicit the thought and the thought will in turn provoke the mood. The embattled mental states from which thought suppression seems a reasonable solution will become married to the very thoughts that seem to be prolonging and inciting them. Maybe thought suppression is not worth the trouble. All of these difficulties seem to point to the conclusion that suppression is more of a problem than a solution. If we can’t always think what we want, or even what we need, it might be better if we simply learned to want what we think (Wegner, 1988). The paradoxical solution offered by some psychotherapists is that when thinking and emotion become deeply unwanted, it may be time to consider accepting these things (see, e.g., Ascher, 1989). A thought can only become unwanted because of the “unwanting,” after all, and the decision to set aside suppression and simply think about the discomforting or taboo items may be the first step to getting over the preoccupation with them. When people spend time thinking about their unwanted thoughts (Borkovec, Wilkinson, Folensbee, & Lerman, 1983) or disclosing them to others (Pennebaker, 1990), they often find that the thoughts lose their mysterious force and are easily set aside. Accepting unwanted thoughts and discontinuing suppression, however, may itself be demanding in light of the broad array of pressures to suppress that occur in the course of social life. It would be so pleasant if we could put out of mind our bad habits, our transgressions, our worries, our humiliations. Just think of our sheer effectiveness in social interaction if we could put out of mind our nervousness, our misgivings, our fear of having deceptions revealed, even our ruminations about little secrets and private opinions. We could interact freely and comfortably for once, without that ever-present cloud hanging overhead of the things we should not say or do or feel or think. On the other hand, it may be precisely these unwanted thoughts that make us moral and social human beings.

THOUGHT SUPPRESSION

219

B. IMPLICATIONS FOR SOCIAL LIFE The mental turmoil into which we are thrown by the suppression of thoughts is, at its center, a social phenomenon. We suppress thoughts because of the actual, imagined, or implied presence of others (to borrow a phrase from Allport, 1985), and this means that the problem of thought suppression is ultimately a problem of the individual’s adjustment to social life. When social circumstances force thought suppression, the person will be in a volatile and potentially antisocial state, ready to think exactly what should not be thought and so to disrupt the social setting in precisely the most inappropriate way. Of course, much of this turmoil will transpire only internally. The person who sees someone of another race, for example, may only briefly arrive at some prejudiced judgment, and then counter the judgment quickly with suppression. A remark based on a stereotype may be stifled, or a negative emotional response to the person may be overcome and neutralized in this way. None of this is particularly visible to an outsider, even though the internal conflict is highly disturbing to its host, and thought suppression is thus temporarily successful. A potential social conflict has been turned into a mental conflict, and so played out in images and ideas rather than in actual confrontation and social disharmony. This may be why thought suppression is such an inviting option in these circumstances-it is fast and, used as a temporary solution, can be very effective in preventing interpersonal conflict. The suppression of thoughts is likely to be a fleeting solution at best, though, as it gives way eventually to longer term consequences that can be insidious. On the outside, perhaps the first symptom of such inner conflict will be a general freezing or suppression of behavior (DePaulo, 1992). Then, more conspicuous manifestations of the suppressed thought can be expected. The person suppressing racist sentiments, for instance, may find that racially prejudiced opinions come to mind in just those situations where their expression would be least tolerable. Far from random, these slips may seem almost Freudian in character (cf. Baars, 1985), leaping forth in exactly the wrong circumstances. Not only despite the best intentions, but in fact because of them, the person finds that the most inappropriate thoughts course through the mind and so guide behavior. Suppression-induced slips of mind will arise primarily when the person is under cognitive load or stress. Just when the cognitive demands of social interaction are strongest (Gilbert, 1991), when there is the most to fear or desire (Jacobs & Nadel, 1985), or when alcohol clouds the vision (Steele & Josephs, 1990), we will find ourselves not expressing merely haphazard thoughts but instead spewing forth unintended bursts of the most unwanted thoughts we have. The mental control we exercise in response to social constraints will, under duress, backfire to knock down those constraints and leave us wondering at the power of pent-up thinking. Piety achieved through thought suppression can spawn profanity, equanimity achieved through thought suppression can yield

220

DANIEL M. WEGNER

violence, and tolerance achieved through thought suppression can turn into hate. Even when suppression is relaxed, its social consequences continue to be counted. The rebound guarantees that the thoughts we have suppressed in particular social circumstances will tend to come back to us, most probably when those circumstances arise again. So, if we have suppressed thoughts in order not to laugh at someone, we may well be reminded to laugh each time we are near that person again. If we have held back thoughts to aid in controlling our anger, lust, or despair, these things are also likely to rebound when we return to the people and situations that first prompted them. Ultimately, we will find that the social controls we have visited on our minds for convenience at one juncture become recurrent goads that make us think again and again of just what we should not. The experimental study of thought suppression, in sum, echoes some of the same implications that Freud drew from his psychoanalytic study of repression. Like repression, the suppression of thoughts acts to hide skeletons in the closet and so to provide for later capricious scarings. The paths of inference in these studies are strikingly different, of course. Freud reached the conclusion that repression is the source of neurosis by studying individual cases, and by developing a rich yet burdensome language for describing the circumstances of apparent forgettings. The present research program has used the direct expedient of telling people not to think, and has reached the initial conclusion that such thought suppression can be a surprisingly maladaptive response when it is used as a strategy for adaptation to social circumstances.

Acknowledgments This research was supported by National Science Foundation Grants BNS 88-18611 and BNS 90-96263. Ralph Erber, Stanley B. Klein, David J. Schneider, Toni Wegner, and Richard M. Wenzlaff made important contributions to the development of ideas in this article and their help is deeply appreciated. Thanks are due to Michael Ross and Timothy D. Wilson for their perceptive comments on an earlier draft.

References Allport, G . W. (1985). The historical background of modem social psychology. In G.Lindzey & E. Aronson (Eds.), Hundbook of social psychology (2nd ed., Vol. 1, pp. 1-80). Reading, MA: Addison-Wesley. Ascher, L. M. (Ed.). (1989). Therapeuric paradox. New York Guilford. Baars, B. 1. (1985). Can involuntary slips reveal one’s state of mind?-With an addendum on the problem of conscious control of action. In T. M. Shlechter & M. P. Toglia (Eds.), New directions in cognitive science (pp. 242-261). Norwood, NJ: Ablex.

THOUGHT SUPPRESSION

22 1

Bargh, J. A. (1984). Automatic and conscious processing of social information. In R. S. Wyer, Jr. & T. K. S ~ l (Eds.), l Handbook of social cognition (Vol. 3, pp. 1-43). Hillsdale, NJ: Erlbaum. Bargh, J. A., & Thein, R. D. (1985). Individual construct accessibility, person memory, and the recall-judgment link: The case of information overload. Journal of Personality and Social Psychology, 49, 1129-1 146. Bjork, R. A. (1989). Retrieval inhibition as an adaptive mechanism in human memory. In H. L. Roediger, 111 & F. I. M. Craik (Eds.), Varieties of memory and consciousness (pp. 309-330). Hillsdale, NJ: Erlbaum. Borkovec, T. D. (1974). Heart-rate process during systematic desensitization and implosive therapy for analog anxiety. Behavior Therapy, 5 , 636-641. Borkovec, T. D., Robinson, E., Pruzinsky, T., & DePree, J. A. (1983). Preliminary exploration of worry: Some characteristics and processes. Behaviour Research and Therapy, 21, 9- 16. Borkovec, T. D., Wilkinson, L., Folensbee, R., & Lerman, C. (1983). Stimulus control applications to the treatment of worry. Behaviour Research and Therapy, 21, 247-251. Bower, G. H., & Mayer, J. D. (1985). Failure to replicate mood-dependent retrieval. Bulletin of the Psychonomic Society, 23, 39-42. Breznitz, S. (Ed.). (1983). The denial of stress. New York: International Universities Press. Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation:A control-theory approach to human behavior. New York: Springer-Verlag. Chandler, G., & Wegner. D. M. (1987). The effect of rhought suppression on preoccupation with associated thoughts. Unpublished research data. University of Virginia, Charlottesville, VA. Cioffi, D. (1991). Beyond attentional strategies: A cognitive-perceptual model of somatic interpretation. Psychological Bulletin, 109, 25-41. Clark, M. S., & Isen, A. M. (1982). Toward understanding the relationship between feeling states and social behavior. In A. H. Hastorf & A. M. ken (Eds.), Cognitive social psycholoRy. (pp. 73- 108). New York: ElsevierINorth-Holland. Coyne, J. C. (1989). Employing therapeutic paradox in the treatment of depression. In L. M. Ascher (Ed.), Therapeutic paradox (pp. 163- 183). New York: Guilford. David, P. J. (1987). Repression and the inaccessibility of affective memories. Journal of Personality and Social Psychology, 53, 585-593. DePaulo, B . (1992). Nonverbal behavior and self-presentation. Psychological Bulletin, 111, 203243. Devine, P. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology. 56, 5- 18. Ellis, E. M. (1983). A review of empirical rape research: Victim reactions and response to treatment. Clinical Psychology Review. 3 , 473-490. Erber, R., & Wegner, D. M. (1991). On being cool and collected: Mood regulation in anticipation of social interaction. Manuscript submitted for publication. Erdelyi, M. H. (1990). Repression, reconstruction, and defense: History and integration of the psychoanalytic and experimental frameworks. In J. L. Singer (Ed.), Repression anddissociation (pp. 1-31). Chicago: University of Chicago Press. Erdelyi. M. H., & Goldberg, B. (1979). Let's not sweep repression under the rug: Toward acognitive psychology of repression. In J. F. Kihlstrom & F. J. Evans (Eds.), Functional disorders of memory (pp. 355-402). Hillsdale, NJ: Erlbaum. Fiske, S. T. (1989). Examining the role of intent, toward understanding its role in stereotyping and prejudice. In J. S. Uleman & I. A. Bargh (Eds.), Unintendedrhoughr (pp. 253-283). New York: Guilford. Foa, E. B., & Kozak, M. J. (1986). Emotional processing of fear: Exposure to corrective information. Psychological Bulletin, 99, 20-35. Freud, S. (1957). Repression. In J. Strachey (Ed.),The standard edition of the completepsychologi-

222

DANIEL M. WEGNER

cal works of Sigmund Freud (Vol. 14, pp. 146-158). London: Hogarth. (Original work published 1915) Freud, S. (1958). Remembering, repeating, and working-through. In J. Strachey (Ed.), The standard edition of the complete psychological works ofSigmund Freud (Vol. 12, pp. 145- 150). London: Hogarth. (Original work published 1914) Geiselman, R. E., Bjork, R. A., & Fishman, D. L. (1983). Disrupted retrieval in directed forgetting: A link with posthypnotic amnesia. Journal of Experimental Psychology: General, 112, 58-72. Gilbert, D. T. (1991). How mental systems believe. American Psychologist, 46, 107-119. Gilbert, D. T., Krull, D. S., & Pelham, B. W. (1988). Of thoughts unspoken: Social inference and the self-regulation of behavior. Journal of P ersonality and Social Psychology, 55, 685-694. Goffman, E. (1963). Behavior in public places. New York: Free Press. Gold, D. B., & Wegner, D. M. (1991). Fanning oldflames: Arousing romantic obsession through thoughr suppression. Paper presented at the meeting of the American Psychological Association, San Francisco. Goldstein. M. J., & Kant, H. S . (1973). Pornography and sexual deviance. Berkeley: University of California Press. Greenwald, A. G. (1988). Self-knowledge and self-deception. In J. S. Lockard & D. L. Paulhus (Eds.). Self-deception: An adaptive mechanism? (pp. 113-131). Englewood Cliffs, NJ: Prentice-Hall. Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental Psychology: General, 108, 356-388. Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of Sociology, 85, 551-575. Hochschild, A. R. (1983). The managed heart. Berkeley: University of California Press. Holmes, D. S. (1974). Investigation of repression: Differential recall of material experimentally or naturally associated with ego threat. Psychological Bulletin. 81, 632-653. Holmes. D. S. (1990). The evidence for repression: An examination of sixty years of research. In 1. L. Singer (Ed.), Repression and dissociation (pp. 85-102). Chicago: University of Chicago Press. Horowitz, M. (1975). Intrusive and repetitive thoughts after experimental stress. Archives of General Psychiatry. 32, 1457-1463. Howell. A., & Conway, M. (1991). Mood and the suppression of positive and negative selfreferent thoughts. Unpublished manuscript, Concordia University, Montreal, Canada. Jacobs, W. J., & Nadel, L. (1985). Stress-induced recovery of fears and phobias. Psychoiogical Review, 92, 512-531. James, W. (1890). Principles of psychology. New York: Holt. Janet, P. (1907). The major symptoms of hysteria. New York: Macmillan. Janis, I. (1958). Psychological stress. New York: Wiley. Kihlstrom, J. F. (1983). Instructed forgetting: Hypnotic and nonhypnotic. Journal of Experimental Psyrhology: General, 112, 73-79. Klinger, E. (1978). Modes of normal conscious flow. In K. S. Pope & J. L. Singer (Eds.), The srream of consciousness (pp. 226-258). New York: Plenum. Klinger, E. (1982). On the self-management of mood, affect, and attention. In P. Karoly & F. H. Kanfer (Eds.), Self-management and behavior change (pp. 129-164). New Yo&: Pergamon. Knutson, B. (1990). Thought suppression and access to distraction. Unpublished research data, Stanford University, Stanford, CA. Lindemann, E. (1944). Symptomatology and management of acute grief. American Journal of Psychiatry, 101, 141-148. Logan, G . D. (1983). On the ability to inhibit simple thoughts and actions: I. Stop-signal studies of

THOUGHT SUPPRESSION

223

decision and memory. Journal of Experimental Psychology: Learning. Memory, and Cognition. 9, 585-606. Logan, G . D. (1988). Toward an instance theory of automatization. Psychological Review, 95,492527. Logan, G. D., & Cowan, W. B. (1984). On the ability to inhibit thought and action: A theory of an act of control. Psychological Review, 91, 295-327. Lombardi, W. J., Higgins, E. T., & Bargh, J. A. (1987).The role of consciousness in priming effects on categorization: Assimilation versus contrast as a function of awareness of the priming task. Personality and Social Psychology Bulletin. 13, 41 1-429. Lyman, S. M., & Scott, M. B. (1968). Coolness in everyday life. In S. M. Lyman & M. B. Scott (Eds.), The sociology ofthe absurd (pp. 145- 157). Pacific Palisades, CA: Goodyear. Marcel, A. J. (1983). Conscious and unconscious perception: An approach to the relations between phenomenal experience and perceptual processes. Cognitive Psychology, 15, 238-300. Marlatt, G. P., & Parks, G. A. (1982). Self-management of addictive behaviors. In P. Karoly & F. H. Kanfer (Eds.), Self-management and behavior change (pp. 443-488). New York: Per gamon. Martin, L. L. (1986). Setheset: Use and disuse of concepts in impression formation. Journal of Personality and Social Psychology, 51, 493-504. Maugham, W. S. (1991). Of human bondage. New York: Bantam. (Original work published 1915). McCaul, K. D., & Malott, J. M. (1984). Distraction and coping with pain. Psychological Bulletin, 95, 516-533. Megargee, E. 1. (1971). The role of inhibition in the assessment and understanding of violence. In J. L. Singer (Ed.), The control of aggression and violence (pp. 125- 147). New York: Academic Press. Miller, G. A,, Galanter, E., & Pribram, K . H. (1960). Plans and the structure of behavior. New York: Holt. Minsky, M. (1987). Society of mind. New York: Simon & Schuster. Mischel, W., & Baker, N. (1975). Cognitive appraisals and transformations in delay behavior. Journal of Personality and Social Psychology. 31, 254-261. Mischel, W., & Ebbesen, E. B. (1970).Attention in delay of gratification. Journal ofPersonaliryand Social Psychology, 16, 329-337. Mischel, W., Ebbesen, E. B., & Zeiss, A. R. (1972). Cognitive and attentional mechanisms in delay of gratification. Journal of Personality and Social Psychology. 21, 204-218. Mischel, W., Shoda, Y.,& Rodriguez, M. L. (1989). Delay of gratification in children. Science, 244, 933-938. Olson, J. M., Barefoot, J. C., & Strickland, L. H. (1976). What the shadow knows: Person perception in a surveillance situation. Journal of Personality and Social Psychology, 34, 583589. Pennebaker, J. W. (1988). Confession, inhibition, and disease. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 22). San Diego: Academic Press. Pennebaker, J. W. (1990). Opening up. New York: Morrow. Polich, J. M., Armour, D. J., & Braiker, H. B. (1981). The course of alcoholism. New York: Wiley. Polivy, J. (1990). Inhibition of internally-cued behavior. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2, pp. 131-147). New York: Guilford. Polivy, J., & Herman, C. P. ( I 985). Dieting and binging: A causal analysis. American Psychologist, 40, 193-201. Posner, M. I., & Snyder, C. R. R. (1975). Attention and cognitive control. In R. I. Solso (Ed.), Information processing and cognilion. Hillsdale, NJ: Erlbaum. Rachman, S. (1978). Fear and courage. San Francisco: Freeman. Rachman, S. (1980). Emotional processing. Behaviour Research and Therapy, 18, 51-60.

224

DANIEL M. WEGNER

Rachman, S., & de Silva, P. (1978). Abnormal and normal obessions. Behaviour Research and Therapy, 16, 233-248. Rachman, S.,& Hodgson, R. J. (1980). Obsessions and compulsions. Englewood Cliffs, NJ: Prentice-Hall. Rapaport, D. (1959). Emotions and memory. New York: International Universities Press. Rapoport. I. L. (1989). The boy who couldn’t stop washing. New York: Dutton. Reed, G. F. (1985). Obsessional experience and compulsive behavior. Orlando, FL: Academic Press. Rosenbaum, J. F., Biederman, J., & Gersten, M. (1989). Anxiety disorders and behavioral inhibition. In J. S. Reznick (Ed.), Perspectives on behavioral inhibition (pp. 255-270). Chicago: University of Chicago Press. Salovey, P., & Mayer, J. D. (1990). Emotional intelligence. Imagination, Cognition, and Personality, 9, 185-211. Sears, R. R. (1943). Survey of objective studies of psychoanalytic concepts. New York: Social Science Research Council. Selye, H. (1956). The stress of life. London: Longmans Green. Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing: 11. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127- 190. Showers, C., & Cantor, N. (1985). Social cognition: A look at motivated strategies. AnnualReview of Psychology, 36, 275-305. Singer, I. L. (1 988). Sampling ongoing consciousness and emotional experience: lmplications for health. In M. J. Horowitz (Ed.), Psychodynamics and cognition (pp. 297-346). Chicago: University of Chicago Press. Steele, C. M., & Josephs, R. A. (1990). Alcohol myopia: Its prized and dangerous effects. American Psychologist. 45, 92 1-933. Stiles, W. 8 . (1987). “I have to talk to somebody”: A fever model of disclosure. In V. J. Derlega & J. H. Berg (Eds.), Self-disclosure: Theory, research, and therapy (pp. 257-282). New York: Plenum. Strack, F., Erber, R., & Wicklund, R. A. (1982). Effects of salience and time pressure on ratings of social causality. Journal of Experimental Social Psychology, 18, 58 1-594. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology. 18, 643-662. Tait, R., & Silver, R. C. (1989). Coming to terms with major negative life events. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 351-382). New York: Guilford. Taylor, E. (1983). William James on exceptional mental stares. Amherst: University of Massachussetts Press. Toates, F. (1990). Obsessional thoughts and behavior. Wellingborough, England: Thorsens. Turner, S . M., Beidel, D. C., & Nathan, R. S . (1985). Biological factors in obsessive-compulsive disorders. Psychological Bulletin, 97, 430-450. Uleman, I. S. (1989). A framework for thinking intentionally about unintended thoughts. In J. S. Uleman & J. A. Bargh (Eds.), Unintended fhought (pp. 425-449). New York: Guilford. Wegner, D. M. (1988). Stress and mental control. In S. Fisher & J. Reason (Eds.), Handbook of life stress, cognition, and health (pp. 685-699). Chichester: Wiley. Wegner, D. M. (1989). White bears and other unwanted thoughts. New York: VikinglPenguin. Wegner, D. M.,& Erber, R., (199 I). Hyperaccessibility of implicit/y suppressed thoughts. Unpublished manuscript, University of Virginia, Charlottesville. VA. Wegner, D. M., & Erber, R. (1993). Social foundations of mental control. In D. M. Wegner & 3. W. Pennebaker (Eds.). Handbook of mental control. Englewood Cliffs, NJ: Prentice-Hall.

THOUGHT SUPPRESSION

225

Wegner, D. M., & Erber, R. (in press). Hyperaccessibility of suppressed thoughts. Journal of Personality and Social Psychology.

Wegner, D. M., Lane, J., & Dimitri, S. (1992). Secret liaisons: The allure of covert relationships. Manuscript submitted for publication. Wegner, D. M., & Pennebaker, J. W. (Eds.), (1993). Handbook of mental control. Englewood Cliffs, NJ: Prentice-Hall. Wegner, D. M., & Schneider, D. J. (1989). Mental control: The war of the ghosts in the machine. In J. Uleman & J. Bargh (Eds.), Unintended thought (pp. 287-305). New York: Guilford. Wegner, D. M.,Schneider, D. J., Carter, S., 111, & White, L. (1987). Paradoxical effects of thought suppression. Journal of Personality and Social Psychology, 53, 5- 13. Wegner, D. M., Schneider, D. J., Knutson, B., & McMahon, S. (1991). Polluting the stream of consciousness: The influence of thought suppression on the mind’s environment. Cognitive Therapy and Research, 15, 141-152. Wegner, D. M., Shortt, J. W., Blake, A. W., & Page, M. S. (1990). The suppression of exciting thoughts. Journal of Personality and Social Psychology, 58, 409-4 18. Weinberger, D. A . , Schwartz, G. E., & Davidson, J. R. (1979). Low-anxious, high-anxious, and repressive. coping styles: Psychometric patterns and behavioral and physiological responses to stress. Journal of Abnormal Psychology, 88, 369-380. Weiner, B. (1966). Effects of motivation on the availability of memory traces. Psychological Bulletin, 65, 24-37.

Wenzlaff, R. M., Wegner, D. M.. & Klein, S. B. (1991). The role of thought suppression in the bonding of thought and mood. Journal of Personality and Social Psychology, 60,500-508. Wenzlaff, R., Wegner, D. M., & Roper, D. (1988). Depression and mental control: The resurgence of unwanted negative thoughts. Journal of Personality and Social Psychology. 55, 882-892. Zillman, D. (1988). Mood management: Using entertainment to full advantage. In L. Donohew, H. E. Sypher, & E. T. Higgins (Eds.),Communication, social cognition. and affect (pp. 147-171). Hillsdale, NJ: Erlbaum.

This Page Intentionally Left Blank

AFFECT IN SOCIAL JUDGMENTS AND DECISIONS: A MULTIPROCESS MODEL

Joseph Paul Forgas

I. Introduction What is the role of affect in our social judgments and decisions? This question has been of considerable interest to writers, artists, and philosophers as well as to laymen for centuries. Remarkably, scientific research on this topic is only a few decades old. Although the evidence for affective biases in social judgments is now overwhelming, theoretical explanations of this phenomenon still emphasize a number of conflicting processes, and there has been little in the way of conceptual integration in the literature. Several explanatory models have been proposed, each capable of accounting for some, but not all, of the available evidence (cf. Bower, 1991 ;Clore & Parrott, 199I ; Fiedler, 1991 ; Forgas, 199I a, in press-c; Isen, 1987; Mayer, 1986; Salovey, 1986; Schwarz & Bless, 1991a, I99 1 b). The major objective of this article is to propose a new, comprehensive model of social judgmental processes, based on a review of the historical and theoretical background of the field, as well as the results of our own empirical research program. Specifically, the model attempts to deal with the following questions: What processing strategies are available to people when performing a social judgment? Under what conditions are they most likely to be used? What is the role of affect in people’s processing preferences? How does affect influence the outcome of social judgments under each of the processing alternatives? The model distinguishes between four alternative processing strategies available to judges: ( I ) direct access of crystallized judgments, (2) motivated processing in the service of a preexisting goal, (3) heuristic or simplified processing, and (4) substantive or elaborate processing. The theory also specifies how eight specific features of the target, the judge, and the situation are likely to influence processing choices. 227 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOUXiY. VOL. 2S

Copyright 8 1992 by Academic Rcss. Inc. All rights of repmduclion in any form reserved.

228

JOSEPH PAUL FORGAS

The need for a such a comprehensive model of affect in social judgments has been increasingly evident in recent years (cf. Bower, 1991; Clore & Panott, 1991; Forgas, 1991a, in press-c; Kaplan, 1991; Mayer & Salovey, 1988). Both the rapidly accumulating empirical evidence and the growing convergence between the various explanations now offered underline the importance of an integrative conceptualization. A better understanding of how affect influences our social judgments is also of considerable practical importance. As Zajonc (1 980) argued, affect is likely to be the primary medium of interpersonal behavior. The way people judge and evaluate social stimuli lies at the heart of many areas of psychological inquiry, including work on attitudes (Eagly & Chaiken, 1991; Tesser & Shaffer, 1990; Zanna & Rempel, 1988), social cognition (Kruglanski, 1989; Wyer & Srull, 1989), persuasion (Chaiken, 1980; Petty & Cacioppo, 1986), intergroup behavior (Brewer & Kramer, 1983, social relationships (Levinger, 1990; Reis & Wheeler, 1991), stereotyping (Hamilton & Rose, 1980; Stangor & Ford, 1992), person perception (Brewer, 1988; Fiske & Neuberg, 1990), survey research (Schwarz & Strack, 1991), subjective wellbeing (Strack, Schwarz, & Argyle, 1991), self-perception (Sedikides, 1991), and the study of language and communication (Forgas, 1985b). Social judgments about ourselves and others are also the foundation of effective interpersonal behavior (Heider, 1958) and are of critical importance in such applied domains as organizational behavior (Isen & Baron, 1991), health psychology (Salovey, O’Leary, Stretton, Fishkin, & Drake, 1991), and clinical psychology (Mayer, 1986; Ottaviani & Beck, 1988). We shall begin with a brief overview of the historical background to research and theorizing on affect and social judgments. Next, a multiprocess theory of affect in social judgments will be outlined, seeking to account for both the informational and the processing consequences of affect. The main body of the article reviews and summarizes the results of our empirical research program on affective influences on social judgments and decisions, presenting evidence for both the informational and the processing consequences of moods.

A. AFFECT IN PSYCHOLOGY

To appreciate fully the problems of dealing with affect and judgmental processes within a single theoretical framework, it is perhaps necessary to consider briefly the historical background of research on affect in psychology. One of the most fundamental issues in psychology concerns the relationship between mental faculties such as cognition, conation, and affect (Hilgard, 1980). For most of its history, empirical psychology assumed that these faculties can be adequately studied in separation from each other. This fragmentation of psychology’s subject matter probably originated with German faculty psychology in the eighteenth

AFFECT

IN SOCIAL JUDGMENTS

229

century and has been greatly accelerated in the twentieth century (Hilgard, 1980). Neither of the two traditionally dominant paradigms in the discipline, behaviorism and the more recent cognitivist framework, assigned much importance to the study of affective phenomena. For radical behaviorism, the study of affect, by definition involving internal processes, was irrelevant. Insofar as affect was considered at all, this was mainly in terms of the motivational consequences of simple drive states such as hunger or thirst. Within the cognitive paradigm, affect was long thought to be tangential to its proper subject matter, traditionally defined as the study of essentially cold, rational ideation. Affect was at best seen as a potentially “disruptive” influence on normal (that is, affectless) cognition. Partly in reaction to the historical neglect of affective phenomena, during the past 10 years or so several attempts were made for the reintegration of affect into the mainstream of psychological inquiry. The issue of whether affect should be seen as part of the cognitive representational system or as a separate system in its own right remains central to contemporary conceptualizations (Fiedler & Forgas, 1988; Salovey & Mayer, 1990). Some theorists such as Zajonc (1980) argued for a “separate-systems’’ view, proposing that affective reactions often precede and are distinct from cognitive processes, a view that hinges on how broadly one defines the domain of cognition (Lazarus, 1984). If affect and cognition are at least partially independent response systems, affect may fulfill a direct informational role in judgments independent of its cognitive consequences. Such a separate-systems view is implied by some recent approaches to affect and social judgments (Clore & Parrott, 1991; Schwarz & Clore, 1988), derived from attributional principles (Zanna & Cooper, 1976). Despite the separation of affect, cognition, and conation in much of psychological theorizing, there is a parallel tradition recognizing the close interdependence of feeling, thinking, and behavior. These ideas were present in the writings of Descartes almost 350 years ago ( 164911961) and were also manifest in some of the work of William James (1890) (cf. Laird & Bresler, 1990). Several recent affect-cognition models also see affect as part of a single, integrated cognitive representational system (Bower, 1981; Bower & Cohen, 1982; Isen, 1984; Mayer, 1986; Salovey, 1986), in the sense that “the experience of an emotion is a cognition” (Laird & Bresler, 1991, p. 24). This cognitivist approach stimulated much of the recent work on affect and cognition, including much of the research on affect and social judgments considered here.

B. AFFECT, MOOD, AND EMOTION At present there seems to be little general agreement in the literature about the exact psychological definition of terms such as affect, feelings, emotions, or mood (Fiedler & Forgas, 1988; Forgas, 1991a; Frijda, 1986). For our purposes,

230

JOSEPH PAUL FORGAS

affect may be used as a broad and inclusive label to refer to both moods and emotions (Mayer, 1986; Petty, Gleicher, & Baker, 1991). Moods, in turn, are low

intensity, diffuse, and relatively enduring affective states without a salient antecedent cause and therefore have little cognitive content (e.g., feeling good or feeling bad) (cf. Clore & Parrott, 1991). Emotions, in contrast, are more intense, short-lived, and usually have a definite cause and clear cognitive content (e.g., annoyance, anger, or fear). Both moods and emotions may have informational value (Frijda, 1986; Schwarz, 1990). Although moods have been sometimes characterized as “mixed emotions,” much of the research and theorizing on affect and social judgments has in fact focused on low-intensity general mood states rather than specific emotions (Mayer, 1986; Sedikides, 1991). This may be because moods do seem to have a potentially more enduring, subtle, and insidious effect on memory, information processing, and judgments than do distinct emotions (Mayer, Gaschke, Braverman, & Evans, 1991). Diffuse positive and negative moods may also be directly linked to the ubiquitous and functionally significant good-bad dimension in evaluations (Osgood, 1969), and to the sadness-happiness dimension in clinical psychology (Ottaviani & Beck, 1988). For all of these reasons, judgmental research continues to focus on positivenegative moods rather than distinct emotions, an orientation that will also be followed here.

C. EARLY RESEARCH Although there has always been a commonsense awareness of the role of affect in social judgments, empirical evidence for this effect was rather scarce until the past 15 years or so. Yet in modem empirical psychology, the role of feelings in social judgments was recognized as early as 1940, when Razran showed that positive or negative affect (induced by receiving a free lunch, or being exposed to unpleasant odors) influenced judgments of soc.apolitica1 slogans. The first explanations of such mood effects on judgments emphasized either psychoanalytic or conditioning principles. For example, Feshbach and Singer ( 1957) suggested that a tendency by subjects made fearful through electric shocks to “perceive another person as fearful and anxious” (p. 286) was due to projection, as “suppression of fear facilitates the tendency to project fear onto another social object” (p. 286). Somewhat similar dynamic interpretations of the links between affect and cognition were proposed by Murray (1933) and others. In several other studies (Izard, 1964; Wehmer & Izard, 1962), affect was induced through the manipulated behavior of confederates. Generally,judgments and performance were more positive following positive mood induction. Selfperceptions and perceptions of social activities were also positively related to self-rated mood over time (Wessman & Ricks, 1966). In keeping with the chang-

AFFECT

IN SOCIAL JUDGMENTS

23 1

ing Zeitgeist in psychology, explanations of these mood effects eventually shifted from psychoanalytic to conditioning accounts. Theories proposed by Byme and Clore (1 970) and Clore and Byme ( 1974) imply that reinforcement and classical conditioning principles may account for affective influences on social judgments. Thus, evaluative social judgments may be conditioned by their temporal and spatial association with other stimuli eliciting unconditioned reactions. Griffith (1970) found that subjects exposed to excessive heat and humidity made more negative judgments of a target person, suggesting that “evaluative responses are . . . determined by the positive or negative properties of the total stimulus situation” (p. 240). Gouaux (1971) also reported mood effects on social judgments following exposure to happy or depressing films, concluding that “interpersonal attraction is a positive function of the subject’s affective state” (p. 40). Gouaux and Summers (1973) used spurious interpersonal feedback to manipulate mood, with similar results. Despite the early evidence for affective influences on social judgments, theoretical accounts based on conditioning and psychodynamic principles remained inconclusive. Mood effects in the absence of temporal contiguity between the affect-eliciting stimulus and the judgmental target seem inconsistent with conditioning accounts (Forgas & Bower, 1988; Mayer et a]., 1991; Wessman & Ricks, 1966). Although conditioning clearly plays a role in some affective reactions (Zanna, Kiesler, & Pilkonis, 1970), it fails as a complete theory of mood effects on social judgments because of the absence of a cognitive dimension detailing how such information integration occurs (Anderson, 1974; Kaplan, 1991). Paradoxically, psychoanalytic explanations seem closer to modem theorizing by suggesting that cognition in a sense becomes “infused” with affect (Feshbach & Singer, 1957). Explaining how such infusion comes about is precisely what recent cognitive and informational theories of affect and cognition have attempted to do (cf. Bower, 1983; Forgas, 1991a; Isen, 1984; Schwarz, 1990).

D. CONTEMPORARY ACCOUNTS Contemporary theories of affect and social judgments are predominantly based on cognitive principles. At present there are two major alternative approaches to explaining mood effects on judgments. The first approach is to focus on the qualitative, informational role of mood states in order to account for the significant mood-congruent judgmental biases found in many studies (Bower, 1991; Clore & Parrott, 1991; Forgas, Bower, & Krantz, 1984; Isen, 1987; Schwarz & Clore, 1988). The two most influential such informational theories are the affectpriming model (Bower, 1981, 1991; Branscombe, 1988; Forgas, 1991a; Forgas & Bower, 1988; Isen, 1984) and the affect-as-information model (Clore & Parrott, 1991; Schwarz & Clore, 1988). The affect-priming model is based on

232

JOSEPH PAUL FORGAS

memory principles, and assumes that mood has an indirect informational influence on social judgments, through the selective activation and greater availability of mood-primed associations in the constructive processing of social information. The affect-as-information model is based on attributional framework, and predicts that moods have a direct informational role, as judges rely on their affective state to infer a judgment. Both of these models can account for some of the mood-congruent bias found in many studies. In contrast with such informational models, there are also several processoriented explanations. These models typically focus on the influence of moods on processing strategies, instead of their informational effects and the qualitative outcome of the judgment. Most processing explanations propose a dichotomy of two distinct processing styles: simplified, effort-minimizing, heuristic processing strategies, and systematic, effortful, analytic processing styles (Fiedler, 1990; Isen, 1987; Petty et al., 1991; Schwarz, 1990). Few attempts have been made so far to evaluate critically the competing claims made by these theories, or to integrate informational and processing explanations within a single comprehensive framework. Yet with the rapid accumulation of empirical evidence, there is now a pressing need for such an integrative theoretical account. This is one of the main objectives of the multiprocess framework outlined below.

11. The Multiprocess Model of AtTeet

in Social Judgments The multiprocess framework proposed here makes several assumptions about the nature of social judgmental processes.

A. ASSUMPTIONS

1 . The Multiprocess Assumption The first assumption is that people can in fact choose between several alternative processing strategies when performing a social judgment. Although this is hardly a radical suggestion, it is in stark contrast with the “single-process” assumptions traditionally implied by most existing cognitive models (Forgas, 1981b, 1983b). Many social cognition models had their origins in informationprocessing cognitive psychology (cf. Wyer & Srull, 1981, 1989), assuming relatively robust, universal, and context-insensitive cognitive processes. The application of such single-process models to the complex and context-dependent domain of social judgments inevitably led to the formulation of more flexible

AFFECT IN SOCIAL JUDGMENTS

233

dual-process models in recent years. The multiprocess assumptions made here are clearly consistent with this trend (Brewer, 1988; Chaiken, 1980; Fiedler, 1991; Fiske & Neuberg, 1990; Kruglanski, 1989; Petty et al., 1991; Stangor & Ford, 1992).

2 . Processing Strategy Mediates Mood Effects A second assumption of the multiprocess model is that the nature of mood effects on social judgments should largely depend on what kind of processing strategy is chosen by a judge. Describing the alternative processing strategies that people use and the specific conditions leading to the adoption of those strategies is then one of the main objectives of the multiprocess framework. By allowing for different processing strategies, the model should be able to accommodate a range of complex and apparently inconsistent findings, including evidence for ( I ) mood-congruent as well as mood-contrast effects (Martin, 1986; Martin, Harlow, & Clore, 1990), (2) symmetric as well as asymmetric mood effects (Lewicka, 1992; Peeters and Czapinski, 1990; Taylor, 1991), (3) target specificity (self-other differences) in mood effects (Forgas et al., 1984; Forgas, Bower, & Moylan, 1990), (4) motivational influences on judgments (Clark & Isen, 1982; Forgas, in press-a), (5) encoding as well as retrieval effects (Mayer et al., 1991), and (6) informational as well as processing effects (Forgas, in press-b, in press-c).

3. Processing versus Informational Efsects

A third assumption implicit in the multiprocess model framework is the proposition that affect plays a dual role in judgments, first influencing the kind of processing strategy adopted, and then, given certain processing conditions, informing the outcome of the judgment due to indirect (priming) or direct (affectas-information)informational effects. For example, affect-priming effects should have a powerful informational impact on judgments when an elaborate, constructive processing strategy is adopted (Forgas, in press-a, in press-b), whereas affect-as-information models should give a better account of mood effects when simplified, heuristic processing is used (Schwarz, 1990). 4 . The Principle of Effort Minimization

The final assumption underlying the dual-process model is that people are principally lazy, effort-minimizing information processors. This means that the simplest and least effortful processing strategy should be adopted as long as it satisfies the minimum contextual requirements for a particular judgment. The model suggests that minimum acceptable processing standards should be deter-

234

JOSEPH PAUL FORGAS

mined by a combination of factors inherent in the target (familiarity, typicality, complexity), the judge (personal involvement, motivation, affective state, cognitive capacity), and the situation (demand effects, likelihood of public scrutiny, etc.). The assumption that people are “cognitive misers” is now commonly incorporated in most dual-process models of social information processing as a critical variable influencing processing preferences (Fiske & Neuberg, 1990; Petty et al., 1991; Weary, Marsh, & Gleicher, 1991). What are the principal processing strategies and the variables that influence processing choices? We shall look at this issue next.

B. THE FOUR PROCESSING STRATEGIES In aniving at a social decision or judgment, people appear to have at least four alternative kinds of processing strategies available to them. Affective influences on judgments will depend on which of these processing strategies is adopted. These processing options are (1) the direct-access strategy, involving the simple retrieval of preexisting, crystallized judgments; (2) the motivated processing strategy, designed to achieve a specific, potentially rewarding judgmental outcome; (3) the heuristic processing strategy, when subjects seek to simplify and short-circuit the substantive processing of relevant information, and (4) the substantive processing strategy, involving the detailed on-line processing and interpretation of the available information.

I . The Direct Access Strategy This is perhaps the most common and effective strategy used in many, if not most, everyday judgments. It simply involves the direct retrieval and use of preexisting, crystallized assessments or evaluations. Given its ubiquity, it is interesting that most models of social judgment pay little attention to this common judgmental approach and to the conditions facilitating its adoption. Directaccess strategies should be used when (1) a target is familiar and preexisting crystallized judgments are already available, and (2) the judge is not personally involved, and there are no strong cognitive, affective, motivational, or (3) situational forces mandating more elaborate processing. Direct-access judgments capitalize on the fact that people do have a rich repertoire of preformed global evaluative judgments to draw on, and probably prefer to use these instead of computing new opinions whenever the target is familiar and the conditions do not warrant more extensive processing. Directaccess judgments do not involve any computations and are thus different from category-based processing (cf. Fiske & Neuberg, 1990). Many everyday judgments, including most routine instances of stereotyping, are probably made in

AFFECT IN SOCIAL JUDGMENTS

235

this way. In the absence of constructive processing, the strongly cued retrieval of an existing crystallizedjudgment is likely to be fairly impervious to mood effects (Fiedler, 1991). In fact, such stored perceptions and evaluations may even be quite resistant to disconfirmation (Kelly, 1955; Swann, in press). However, even crystallized judgments were initially computed using constructive processes, at which time affect as well as other variables may well have influenced the judgmental outcome.

2 . The Motivated Processing Strategy Motivated processing occurs when a social judgment is driven by a specific preexisting goal, wherein the judge engages in selective, guided information search and integration strategies designed to support an antecedent motivational objective. In a real sense, motivated processing is perhaps the paradigmatic example when preferences do in fact come to guide inferences (Zajonc, 1980). Motivated processing is most likely when (1) the target is unfamiliar and no crystallized judgment is available and (2) the judge has a strong preexisting objective in computing the judgment, and employs a selective, guided search in support of a specific motivational goal. Motivated processing may be often used to achieve mood maintenance as well as mood repair, as people selectively look for and consider information diagnostic about desired outcomes (Clark & Isen, 1982). We found evidence for such motivated judgmental strategies in some recent studies of affiliative choices (Forgas, 1989, 1991~).Motivated processing as defined here involves not merely a generic motivation to be careful and accurate, but the superimposition of a specific, preexisting preference to guide information search and judgmental processes. There is now growing evidence for the motivated character of many cognitive processes and social judgments (Branscombe & Cohen, 1991; Forgas, 1991c; Martin, 1986; Tesser, 1986). Research on cognitive dissonance demonstrated motivational influences on self-perception and attitudes some time ago (Eagly & Chaiken, 1991; Wicklund & Frey, 1981; Zanna & Cooper, 1976). Motivated choice strategies have also been long recognized in the decision-making literature (Janis & Mann, 1977; Mann, 1992). In memory research, motivational states can enhance the accessibility of particular memories (Parrott & Sabini, 1990). Indeed, unbiased, nonmotivated information processing may well be far less common than usually assumed (Kunda, 1990). 3. The Heuristic Processing Strategy

Under some conditions, judgments will be computed with the minimum possible effort, without the full consideration of the available information, and using whatever shortcuts or simplificationsare readily available (Isen, 1984). Heuristic

236

JOSEPH PAUL FORGAS

processing is most likely to be used when (1) the target is simple or highly typical, (2) the judge has no specific motivation or personal involvement in the issue, has limited cognitive capacity, is more likely to be in a positive mood, and has little motivation to be accurate, and (3) the situation contains no demands for detailed processing. There is much evidence for heuristic processing in the social judgment literature. For example, judgments of persuasive messages may be based on their peripheral features (Petty et al., 1991; Schwarz & Bless, 1991b). Evaluations of a person may be determined by irrelevant associations with environmental variables (Griffith, 1970) or superficial similarity to an already familiar type of person (Forgas, in press-a, in press-b). Sometimes, judgments may be simply inferred from the perceiver’s prevailing affective state (cf. Clore & Parrott, in press; Schwarz & Bless, 1991a), probably the major source of moodcongruent judgmental effects when heuristic processing is used. 4. The Substantive Processing Strategy

Finally, substantive processing requires judges to select, learn, and interpret novel information about a target and to relate this information to preexisting knowledge structures in order to compute a judgment. It is this kind of open, online information processing and the underlying image of the “dutiful information processor” that is implicitly assumed by most information-processing theories and contemporary models of social cognition (Forgas, 1981b, 1983b; Wyer & Srull, 1981, 1989). Substantive processing is more likely to be adopted when (1) the target is atypical, unusual, or complex, (2) the judge has no specific motivation to pursue, has adequate cognitive capacity, is more likely to be in a negative mood, and is motivated to be accurate, possibly because of (3) explicit or implicit situational demands. Substantive processing is largely automatic and uncontrolled. Social and cognitive theories often analyze this process either in terms of memory models, focusing on the strategies involved in interpreting and assimilating novel information into a preexisting category system (Bower, 1991; Wyer & Srull, 1989), or in terms of information integration principles (Anderson, 1974; Kaplan, 1991). Affect can play a major informational role in substantive processing, through the indirect priming and activation of cognitive categories to be used in computing a judgment (Bower, 1991; Forgas, 1991a). In fact, the extent of mood-priming effects in a judgment may depend on the extent to which constructive, substantive processing is required to perform the task (Fiedler, 1991; Forgas, in press-c, in press-d).

C. IMPLICATIONS OF THE MODEL Distinguishing between these four alternative strategies has considerable bemefits for a comprehensive theory of social judgments. In terms of the model, the

AFFECT IN SOCIAL JUDGMENTS

237

influence of affect on judgments should depend on which strategy is adopted. This classification of processing strategies also highlights the fact that substan.tive processing, implicitly assumed to be the “standard” mental procedure by most cognitive as well as judgmental theories, is more likely to be merely a default option, used only when no other strategy is appropriate. This seems consistent with evidence from other domains suggesting that people are cognitive misers, and all things being equal, will prefer not to engage in extensive processing (Fiske & Neuberg, 1990; Weary et al., 1991). The model also suggests that many judgments previously simply considered as category based, particularly in the domain of stereotyping (cf. Fiske & Neuberg, 1990; Stangor & Ford, 1992), may in fact involve two or more quite different processing strategies, such as direct-access, motivated, or heuristic processing. For example, whereas direct-access stereotyping simply involves the recall of ready-formed opinions, heuristic processing involves some degree of construction wherein judgments are only partly determined by the categories used by the judge. Finally, motivated processing may also be implicated in stereotyping. Motivation for ego enhancement or in-group distinctiveness can play an important and relatively little studied role in stereotyping (cf. Brewer & Kramer, 1985; Tajfel & Forgas, 1982). These four processing alternatives also differ in the kind of information search strategies they imply (extensive search versus partial search), and the extent to which the judgment is or is not predetermined, as follows:

Prejudged Not prejudged

Full search

Partial search

Direct access Substantive

Motivated Heuristic

There are strong parallels between the processing combinations proposed here and Kruglanski’s ( 1 989) seminal lay epistemic theory (LET), a universal model of the way people seek information and generate and validate hypotheses. Whereas Kruglanski’s (1989) lay epistemic theory is a high-level theory that classifies knowledge acquisition in terms of a person’s long-term or short-term tendency to seek or avoid specific or nonspecific closure, our model is a middlelevel theory specifically designed to deal with affective influences on social judgments. The role of affect in processing has not been specifically addressed by LET; however, our four processing strategies do appear generally compatible with Kruglanski’s (1989) model. The pattern of simplified, heuristic processing in positive mood, and elaborate, substantive processing in negative mood, may also be accommodated by LET (A. W. Kruglanski, personal communication), although the specific conditions conducive to these strategies would need to be specified by middle-level theories, such as the present framework. The present model does in fact yield several specific testable predictions about the role of

238

JOSEPH PAUL FORGAS

factors associated with the target, the judge, and the situation in determining processing choices, as we shall see later (cf. Forgas, 1991b, 1991c, 1991d, in press-a). A critical aspect of the model is its focus on the variables predicting the use of particular processing strategies, a topic we shall turn to next.

D. VARIABLES DETERMINING PROCESSING CHOICES Figure 1 presents a comprehensive summary of the four processing strategies and the functions of the various predictor variables regulating processing choices (the numbers below refer to the numbers in Fig. 1). Which of the four alternative processing strategies will be adopted depends on a combination of three groups of variables associated with the target, the judge, and the judgmental situation (cf. Fiedler, 1991). Target features includefamiliarity (I), (cueing direct-access judgments to be made), and whether the stimulus is typical, ambiguous, or complex (4). Atypical, complex targets are more likely to receive substantive processing, increasing the likelihood of mood-primed associations influencing such judgments (Forgas, 1991b, 1991d, in press-a, in press-b). Judge features include the personal importance (2) of the judgment, the available cognitive capacity (3,and the affective (6) state of the judge at the time. Judges may also have a spec$c motivational goal (3) to guide a judgment (e.g., mood repair; Clark & Isen, 1982; Forgas, 1991c), or a general motivation to be accurate ( 7 ) and careful. Affect (6) in particular may reduce processing capacity (Ellis & Ashbrook, 1988; Isen, 1984), induce motivated processing (Forgas, 1991c), or influence motivation for accuracy (Schwarz, 1990). Finally, situational factors (8) may also influence processing choices through enhancing a judge’s motivation to be careful and accurate. Of course, not every variable need influence every processing choice, nor is the decision sequence necessarily invariant. However, the sequence of influences on processing choice depicted in Fig. 1 is likely to apply to a broad range of judgmental situations. 1. Familiarity. Identifying a judgmental target as familiar should have a major influence on processing choices. Complex social targets (such as other people) present us with numerous powerful recognition cues, strongly priming the retrieval of stored, crystallized evaluations once these are available in memory. Even superficial similarity to familiar others may be sufficient to trigger strong evaluative reactions. Greater liking for familiar others offers additional evidence for the status of familiarity as a strong retrieval cue. 2. Personal relevance. For targets identified as familiar, the simplest directaccess strategy should be adopted, but only if the judgment has no particular personal relevance or importance. The majority of our everyday social judgments are probably of low personal relevance, allowing such crystallized evaluations to

239

AFFECT IN SOCIAL J U D G M E N T S JUDGMENRUTARGEI

DIRECT ACCESS STRATEGY

MOTIVATED PROCESSING I . FAMILLAR? \

(CRYSTALLUPD)

fi

I

STRATEGY ------

(GOAL-DIRECIED)

-PRENDGeD -PARmAL SEARCH

m

I

ExM(pLE: MOODC O ~ O L ?

SUBSTANTIVE PROCESSING STRATEGY

(SIMPUPLED)

(-RATE)

+ -NUTPRENDGQ

-NOT PRENDGED -PARTULSeARCH

-EXIFNSTVESEARCH

EXAMPLE:

4

EXAMPLE:

h i 3 X I PRUQNG?

AFll3X-AS-IhTOlWATlOb 2

8. SrIlJAnONAL FACTORS: SOClAL DeSlRABlUTY NEED mu ACCURACY

AVAllABlLnY OF CRTERlA

Fig. 1. Outline of a multiprocess model for understanding mood effects on social judgments, showing the four principal processing strategies and the variables influencing processing choices.

be used. However, if the task is identified as of some personal relevance, further processing is required before a judgment is computed (Brewer, 1988). Despite the obvious intuitive appeal of personal relevance as a mediating variable in judgmental processing, few affect-cognition theories devote much attention to this variable. Yet there is evidence that simple manipulations of personal relevance can result in quite profound changes in processing strategies (Brewer, 1988; Forgas, 1991~). 3. Specific motivation. Not all judgments are open-ended: Often we seem to

240

JOSEPH PAUL FORGAS

know the kind of outcome we are looking for, even if we do not have all the necessary information and evidence to support it. In constructing a social judgment, one of the critical questions a judge must consider is whether there are special motivational forces mandating a particular judgmental outcome. The role of motivation in social cognition has been increasingly recognized in recent times (Argyle, 1991; Forgas, 1981b, 1983b, 1991c; Kunda, 1990). Common motivational forces guiding our judgmental strategies may include ego enhancement, ingroup favoritism, prejudice, need for affiliation, or need for consistency. Affect itself may be a source of specific motivation, with mood maintenance (in positive moods) or mood repair (in negative moods) as the most common motivated judgmental strategies (Clark & Isen, 1982). In a bad mood, motivated processing may involve the guided search for information helpful to the alleviation of an unpleasant state, as we have found in recent studies of affiliative choices (Forgas, 1989, 1991~).The critical feature of specific motivation as defined here is a qualitative bias toward certain kinds of information to support a preexisting, motivated preference. Such specific motivation is quite different from a general motivation to be accurate (to be considered later), which does not imply a qualitative preference for a particular kind of outcome. 4. Target typicality or complexity. Target qualities may have an important influence on whether simplified, heuristic, or elaborate, substantive processing strategies are used in computing a judgment. All things being equal, complex, atypical, and unusual targets should require substantive processing, whereas simple, typical targets can be more efficiently dealt with using heuristic processing. Past work on person prototypes provides some support for this proposition. We found that judges tend to adopt a simple, heuristic processing strategy when dealing with well-known, salient targets, whereas less familiar and less salient characters were processed using what appeared to be a substantive processing style (Forgas, 1983a, 1985a). More recently, we tested this hypothesis by measuring directly the effects of mood and target typicality on processing latencies. We found clear evidence for greater mood-priming effects to be associated with the more elaborate, substantive processing of atypical targets, confirming the important role this variable plays in triggering alternative processing strategies (Forgas, in press-a, in press-c, in press-d). 5. Cognitive capacity. The availability of cognitive processing capacity to execute a complex social judgment is one of the determinants of whether simple, heuristic, or elaborate, substantive processing is adopted. Several factors influence cognitive capacity. One possibility is that the additional information cued by both good and bad moods imposes greater cognitive demands on attention, memory, and processing capacity (Ellis & Ashbrook, 1988; Isen, 1984). This capacity-limit explanation cannot account for differential, asymmetric negative-as against positive-mood effects (Taylor, 1991). Mood may also induce attention deficits (Mackie & Worth, 1991; Stroessner, Hamilton, & Mackie, in

AFFECT

IN SOCIAL JUDGMENTS

24 1

press). In principle, both positive and negative moods may result in capacity limits and attention deficits, predisposing judges toward using heuristic rather than substantive processing. 6. Affective state. Feeling good or feeling bad is itself a significant influence on processing preferences. People in good moods seem to prefer simple, heuristic, and faster information-processing alternatives (Isen, 1984, 1987). In contrast, negative moods often trigger slower, more careful, and analytic processing styles both in normal and in clinical populations (Forgas & Bower, 1987; Ottaviani & Beck, 1988; Weary et a]., 1991). Several explanations for this processing asymmetry have been proposed (Taylor, 1991). Within the affect-asinformation framework, moods may function as evolutionary signals, informing us about states of the world that do or do not require careful monitoring or processing (Frijda, 1986). The result is loose, effort-minimizing, heuristic processing in good moods, and detailed, systematic, effortful, and analytic processing in dysphoria (Schwarz & Bless, 1991a). One potential conceptual problem with this model is its implication that the adoption of a processing style is essentially postjudgmental. That is, the decision of whether heuristic or substantive processing is required occurs only after a judgment based on the informational content of an unattributed mood state has already been computed. An important aspect of the model is the recognition that affect plays a multiple role in processing choices, some direct and some indirect. In addition to influencing directly processing preferences, good or bad mood may also contribute to a specific motivation for mood repair or mood maintenance, may influence cognitive capacity, and may serve to alert a person to be motivated to be accurate. In this way, affect may trigger both mood-congruentjudgmental effects (if heuristic or substantive processing is adopted) and mood-incongruentjudgmental effects. 7. Motivation for accuracy. A variety of forces may motivate a judge to attempt to be as accurate as possible in executing a judgment. The higher the motivation to be accurate, the more likely it is that substantive rather than heuristic processing strategies should be adopted. Evidence for the importance of such motivational forces comes from both memory (Martin et al., 1991) and judgmental tasks (Martin, 1986). One important influence on motivation for accuracy is temporary mood, as we have seen above: Negative mood in particular may serve to trigger a motivation to be careful, vigilant, and accurate (Schwarz, 1990;Weary et a]., 1991). Awareness of possible biasing factors in the availability of information may also motivate judges to “reset” the information array they consider relevant (Martin, 1986). Perhaps the most important source of a motivation to be accurate is the judgmental situation itself Certain kinds of judgments (such as those made publicly, those likely to be scrutinized, and those where clear criteria are available) by definition require a higher standard of accuracy. 8. Situational variables. The situational context of the judgment can, of

242

JOSEPH PAUL FORGAS

course, have a major impact on the way judges process information. Generally, situational influences include factors such as the likelihood of a judgment being made public, being scrutinized, and being compared to existing judgments, and other external demand effects. For most practical purposes, the role of situational factors is to increase or decrease the standards of accuracy required. In other words, we may think of situational influences as having a primary effect on the judges’ motivation to be accurate. By and large, the greater the situational demands for accuracy, the greater the motivation to be accurate, and the more likely it is that detailed, substantive rather than simple, heuristic processing strategies will be adopted. As the above discussion suggests, one of the most important consequences of the multiprocess framework is that it specifies the links between two different kinds of affective influences on judgments. (1) lnformationaf effects may be due to either the indirect (priming other cognitions; Bower, 1981) or direct informative functions of feelings (Schwarz & Clore, 1988). (2)’Processeffects in turn have to do with the role of feelings in influencing information-processing strategies, and ultimately judgments (Fiedler, 1991; Forgas, in press-c; Isen, 1984; Schwarz, 1990). We shall next consider the role of both kinds of processes in mood-congruent judgments before discussing the empirical evidence bearing on these issues.

111. Mechanisms of Mood Congruence in Judgments

This section will focus on the mechanisms underlying mood congruency in judgments, a robust finding now supported by extensive empirical evidence (see Bower, 1991; Forgas, 1991a; Forgas & Bower, 1988; Mayer et al., 1991; Schwarz & Clore, 1988; Sedikides, 1991, for recent reviews). In terms of the multiprocess framework, such mood congruency occurs because mood comes to indirectly (through primed associations) or directly (through misattribution) provide relevant information to be used in computing a judgment. The affectpriming account (Bower, 1981, 1991; Forgas & Bower, 1988; Isen, 1984) sees affect as indirectly influencing cognition (and ultimately, social judgments) through priming access to related cognitions. The alternative affect-as-information approach (Clore & Parrott, 1991; Schwarz & Clore, 1988) ascribes a direct informational role to affect. The multiprocess framework also suggests, however, that affect priming is most likely to occur in the course of elaborate, substantive processing, whereas affect-as-information processes are more likely when simplified, heuristic processing is adopted. The dual informational and processing role of affect in mood congruency as implied by the multiprocess framework is

AFFECT IN SOCIAL JUDGMENTS

243

summarized in Fig. 2. In this section the two main informational mechanisms underlying mood congruence, affect priming and affect-asinformation, and their processing implications will be considered. A. INDIRECT INFORMATIONAL EFFECTS OF MOOD 1. Affect-Priming Mechanisms

Affect priming assumes that social judgments involve constructive, or topdown, cognitive processing, wherein the ideas and interpretations of the judge are of prime importance. This constructivist view of social judgments is based on

Simplified heuristic processing

judgment

1

I I I

I

I I

I

store mood-primed associations

substantive processing

judgment

Fig. 2. Mechanisms of mood congruence in judgments: the processing (broken lines) and informational influence on affect on judgments (solid lines), suggesting mood-priming effects in substantive processing and affect-as-information effects in heuristic processing.

244

JOSEPH PAUL FORGAS

phenomenological psychology (Heider, 1958; Kelly, 1955) and the Gestalt tradition in social perception research (Asch, 1946), and is also supported by empirical work on implicit personality theories (Rosenberg & Sedlak, 1972). To the extent that social judgments are constructive, the availability of ideas and cognitions in the mind of the perceiver can play a major role in how complex and often indeterminate social stimuli are perceived, learned, and interpreted (Bower, 1983; Kelly, 1955). In terms of the affect-priming principle, moods can inform social judgments through selectively facilitating the access and retrieval of similar or related cognitive categories (Bower, 1981; Isen, 1984, 1987). Bower (1981), in his network model, proposed that affective states have a “specific node or unit in memory that . . . is also linked with propositions describing events from one’s life during which that emotion was aroused. . . . Activation of an emotion node also spreads activation throughout the memory structures to which it is connected” (Bower, 1981, p, 135). Thus, affect primes the recall and use of its associated constructs in the top-down processing of social judgments. Several consequences follow from this formulation. (1) The model predicts the superior recall of information that has been experienced in a matching affective state (mood-state-dependent memory effects).( 2 ) Information that is consistent with the current mood should also be more extensively encoded and learned, and better recalled due to the greater availability of mood-consistent associations (mood-congruentmemory and learning efsecrs). ( 3 ) People in a particular affective state tend to focus on affect-consistent rather than -inconsistent information (atrention effects). (4) Finally, the greater availability of mood-consistent associations should influence the top-down, constructive interpretation of inherently complex and ambiguous social information (associative effects). The multiprocess model implies that it is in the course of substantive processing that affect is most likely to play an indirect role in what is perceived, learned, and recalled, and how such information is interpreted, leading to an overall affect-consistent bias in judgments. Thus, good mood should lead to closer attention to matching information, better learning of such details, the interpretation of ambiguous information in a mood-consistent manner, and the better recall of such details later on, predictions that were supported in numerous studies (cf. Bower, 1991; Erber, 1991; Forgas, 1991a, 1991c, in press-b; Forgas & Bower, 1987, 1988; Forgas et al., 1984, 1990; Mayer et al., 1991).

2 . Evaluation of the Affect-Priming Hypothesis Affect priming is a broad and parsimonious principle with implications for a variety of cognitive processes. Despite strong cumulative evidence for affect priming in social judgments, memory effects seem less robust than initially expected (Blaney, 1986; Bower & Mayer, 1985). Weak mood-state-dependent

AFFECT IN SOClAL JUDGMENTS

245

memory effects were first attributed to the lack of sufficiently intense mood manipulations (Bower & Mayer, 1985). Evidence that surprisingly weak mood manipulations can produce significant (judgmental) effects (cf. Isen, 1984) detracts from the plausibility of the intensity explanation. Weak memory effects may also be due to the lack of causal belonging between the mood induction and the experimental task. This may be why mood effects are more reliably found with autobiographical material (where causal associations presumably exist) but not with incidental materials. However, evidence for mood effects in studies wherein causal association was unlikely, including Bower’s (1981) initial two-list learning experiment, detracts from the force of the causal belonging explanation. There have also been suggestions that mood priming may be difficult to obtain in conditions that are “antithetical to self-referencing” (Blaney, 1986, p. 232). We have found negative mood to have a stronger effect on judgments of the self than others (Forgas et al., 1984, 1990). a pattern also common in clinical depression (Ottaviani & Beck, 1988; Weary et a]., 1991). Within a multiprocess framework, such asymmetrical mood effects when judging self versus others may reflect the more substantive, elaborate processing of highly self-relevant information in dysphoria. Another issue is that affect priming is sometimes thought to imply that all “positively valenced material is more accessible . . . in positive . . . moods, and negatively valenced material is more accessible [in] negative moods” (Schwarz, 1990, p. 528). In this case, spreading activation to all similarly valenced contents would rapidly dissipate the priming effects of moods (a fan effect). In fact, Bower’s (1981) network model made no such claim. Mood is expected to prime not all similarly valenced cognitions, but cognitive contents experienced in a matching mood in the past, irrespective of their valence. Thus in a negative mood, both positive and negative, self-deprecatory and other-enhancing cognitions may simultaneously become more accessible. This may explain some of the self-other differences found in many studies of mood effects on social judgments in normal as well as in clinical populations (Forgas et al., 1984, 1990; Sedikides, 1991; Weary et al., 1991). The most recent version of the affect-priming model provides for a separate valence integration node for each judgmental target (Bower, 1991). Judgments are then a function of the accumulated balance of positive and negative valences at any one time. This view incorporates some information integration principles (Kaplan, 199I), and can explain such familiar judgmental effects as primacy and recency, or the persistence of evaluative impressions long after the data on which they are based are forgotten (Bower, 1991). Consistent with its roots in memory research, affect priming is essentially a single-process model, assuming automatic, substantive information processing on the part of the subjects. To the extent that people do have access to multiple processing strategies when computing a judgment, processing sryle appears to be

246

JOSEPH PAUL FORGAS

a critical factor in mediating mood-priming effects (Fiedler, 1991; Forgas, in press-c; Martin, 1986; Schwarz, 1990). It seems that affect “will influence cognitive processes to the extent that the cognitive task involves the active generation of new information as opposed to the passive conservation of information given” (Fiedler, 1990, pp. 2-3). The evidence shows more reliable mood effects on constructive tasks such as social judgments (cf. Forgas & Bower, 1988) rather than mere information retrieval (cf. Bower & Mayer, 1985). Within judgmental tasks, mood priming is stronger for targets that are unusual, atypical, and require more substantive processing (Forgas, 1991c, in press-b, in press-d; Mayer et al., 1991). In brief, affect priming is most reliably found in conditions that are conductive to a substantive processing style. In summary, affect priming is based on a rich tradition of research on memory and information processing, and offers a simple and parsimonious framework for accounting for many mood effects on social judgments that involve substantive processing strategies. The three most notable limitations of this model are that (1) it mainly applies to automatic, subconscious processes involved in substantive processing, (2) it regards affect as a component of a cognitive representational system with only indirect effects on cognition and judgments, and (3) the predictions of the model are not always readily falsifiable (Bower & Cohen, 1982). Alternative models assume different processing strategies and allow for a more direct informational role for feelings in social judgments, as we shall see below.

B . DIRECTION INFORMATIONAL EFFECTS OF MOOD I . The Misattribution Hypothesis Whereas affect priming deals with the indirect informational effects of mood during substantive processing, the affect-as-information model assumes that “rather than computing a judgment on the basis of recalled features of a target, individuals may . . . ask themselves: ‘How do I feel about it?’ [and] in doing so, they may mistake feelings due to a pre-existing state as a reaction to the target” (Schwarz, 1990, p. 529). In this process, affect is information: it directly functions as a judgment-simplifying heuristic as subjects consult their mood to infer a judgment (Clore & Parrott, 1991; Schwarz & Clore, 1988). This approach has close links to research on judgmental heuristics, as well as work on attributions (Zanna & Cooper, 1976), as judges “misattribute” their mood to the target, instead of its true and no longer salient cause. The affect-as-information framework predicts mood congruency when simplified, heuristic rather than substantive processing is used, and is thus essentially complementary to the affectpriming account.

AFFECT IN SOCIAL JUDGMENTS

247

2 . Evaluation of the Affect-as-Information Hypothesis Both affect priming and affect-as-information predict mood congruency in judgments, albeit under different processing assumptions. However, the affectas-information account seems to imply an all-or-nothing process. Here, mood either completely determines judgments (if mistakenly attributed to a target), or has no influence at all (if already attributed to another cause), with no provision for gradual, intermediate effects. The strong assertion that only unattributed moods may have judgmental consequences also makes this model readily falsifiable. The evidence on this point is still somewhat unclear. Several studies report that correctly attributed feelings failed to influence judgments (Clore & Parrott, 1991; Schwarz & Bless, 1991a; Schwarz & Clore, 1988). On the other hand, numerous studies that employ mood manipulation procedures, such as the Velten technique, hypnotic induction, behavior of a confederate, or manipulated feedback about performance, wherein subjects must have been highly aware of the correct source of their moods, nevertheless continue to find mood-based distortions in social judgments (Erber, 1991; Fiedler, 1990, 1991; Forgas & Bower, 1987; Forgas et a]., 1984; Sedikides, 1991), contrary to the implications of the model. Evidence for gradual, target-specific, or context-dependent mood effects (Forgas, in press-a, in press-c; Forgas et al., 1984, 1990; Mackie & Worth, 1991) also suggests that the all-or-nothing processes implied by affect-as-information models may need at least to be supplemented by other processing alternatives. Another issue is that the affect-as-information model predicts mood effects at the retrieval, or judgmental, stage only, whereas mood-priming models also make strong predictions about encoding, learning, and attention effects. To the extent that encoding effects have been reliably demonstrated in the literature (see Forgas, 1991a; Forgas & Bower, 1988 for a review), theories other than the affect-as-information model may be needed to account for such findings. Of course, within a multiprocess framework there is no fundamental reason why treating affect as a direct source of information would preclude its indirect role in information storage and retrieval as affect-priming models suggest (Kaplan, 199I), pointing to the complementary rather than incompatible features of affectpriming and affect-as-information models (Bower, 1991). As shown in Fig. 2, the multiprocess framework allows an integration of the indirect and direct informational effects of mood, in terms of a combination of affect-priming and affect-as-information principles. In addition to its informational effects, mood also has processing consequences (cf. Fig. 2), an issue we shall turn to next.

C. PROCESSING CONSEQUENCES The multiprocess model clearly recognizes that affect is not merely informative, but also influences people’s information-processing strategies (Fig. 2).

248

JOSEPH PAUL FORGAS

Several studies suggest that in a good mood, people reach a decision more quickly, use less information, tend to avoid demanding, systematic processing, and are more confident about their decisions (Forgas, 1989, in press-c; Isen, 1984, 1987; Schwarz, 1990). Relatively less attention has been paid to the processing consequences of negative moods, and what evidence exists suggests slower, more analytic, and vigilant processing strategies in dysphoria (Forgas, 1989; Forgas & Bower, 1987; Schwarz, 1990; Sedikides, 1991; Weary et al., 1991). These two processing styles are similar to heuristic and substantive processing as proposed in the multiprocess model, as well as several other recent processing dichotomies. For example, Fiedler (1990, 1991) suggested a global distinction between productive or constructive and reproductive or reconstructive processes, involving the “loosening” or “tightening” of informationprocessing strategies. Mood priming may be most marked when productive, elaborative, and loose cognitive processing occurs, as indeed found in some of our recent studies (see below, Forgas, in press-b, in press-c, in press-d). Somewhat related processing dichotomies in the recent social cognition literature include the distinction between category-based versus attribute-based processing (Fiske & Neuberg, 1990), impression formation set versus memory set processing (Wyer & Srull, 198 I), accuracy versus expectancy-confirming processing styles (Stangor & Ford, 1992), and central-route versus peripheral-route processing (Chaiken, 1980; Petty et a]., 1991). However, not all processing dichotomies make a clear distinction between the informational basis of judgments (e.g., attributes versus categories) and the processing strategy used (loose versus tight, top down versus bottom up, deductive versus inductive reasoning). In principle, any kind of informational input (e.g., mood) may be processed by any of the available processing options, as made clear by the multiprocess framework proposed here. But how do moods come to change the way we process information? There are several variables of interest here. 1. Affect and cognitive load. Cognitive psychology traditionally saw affect as a source of intrusion and distraction into normal (that is, cold) cognition. This may occur because the information cued by moods imposes cognitive strain, limiting attention, memory, and processing capacity. This may be why euphoric subjects engage in shorter, more superficial and less analytic processing strategies (Isen, 1984, 1987), or depressed subjects show a variety of cognitive distortions (Ellis & Ashbrook, 1988). As both good and bad moods may be a source of distraction and attention deficits (Mackie & Worth, 1991; Stroessner et al., in press), cognitive strain is not a plausible explanation of the often asymmetric processing consequences of good and bad moods reported in the literature (Forgas, 1991a; Taylor, 1991). 2. Affect as an evolutionary signal. In evolutionary terms, affect may inform us about states of the world that do or do not require careful monitoring or processing (Frijda, 1986). In an extension of the affect-as-information model,

AFFECT IN SOCIAL JUDGMENTS

249

Schwarz (1990; Schwarz & Bless, 1991a) suggests that positive moods signal that “all is well with the world,” cueing heuristic, effort-minimizing information processing, possibly allowing greater scope to engage in unusual, unorthodox, and creative thinking (Isen, 1984). Negative moods, in contrast, signal problematic, difficult situations, triggering careful, vigilant, analytic, and tight cognitive strategies. This account implies that processing choices are in a sense postjudgmental, a position that suggests a primary informational role for affect over cognition (Lazarus, 1984; Zajonc, 1980). 3. Affect and motivation. As suggested by the multiprocess model, affect may also guide information processing through its motivational consequences. A good mood may motivate the avoidance of effortful processing as a means of mood maintenance (Isen, 1984), whereas in bad moods, controlled processing may involve a motivated search for information relevant to the alleviation of an undesirable state (Clark & Isen, 1982), as found in our recent studies of mood effects on interpersonal preferences (see below, Forgas, 199lc). Personally relevant judgments are perhaps most likely to be influenced by motivational objectives (Forgas, 1991c;Tesser, 1986). We know that motivated thinking is well documented in areas such as cognitive dissonance, self-perception, decisionmaking, attitudes, and attributions (Eagly & Chaiken, 1991; Janis & Mann, 1977; Mann, 1992; Tesser, 1986; Wicklund & Frey, 1981; Zanna & Rempel, 1988). Motivation may also influence memory (Parrott & Sabini, 1990), and may induce either assimilation or contrast effects in information selectivity (Martin, 1986; Martin et al., 1991). In conclusion, it seems that the various processing and informational consequences of moods must be considered in conjunction with contextual influences. The multiprocess model was developed to provide an integrative framework within which these complex and often conflicting influences can be accommodated. As we have seen, mood-congruent judgment involves at least two distinct informational effects (affect priming and affect-as-information), each occurring in a different processing context (substantive versus heuristic processing). In the next section our empirical research program dealing with some of the informational and the processing consequences of moods will be summarized.

IV. The Research Program Over the past several years, our research on affective influences on social judgments had two complementary objectives: (1) to establish the conditions under which feelings indeed have a reliable and predictable influence on social judgments and (2) to analyze the cognitive processes that underlie such influences. These studies can be conveniently classified into three complementary

250

JOSEPH PAUL FORGAS

categories, in terms of their research focus: (1) studies exploring the role of informational, mood-priming effects in social judgments, (2) studies exploring the role of processing variables incorporated in the multiprocess model in social judgments, and (3) research on the extensions and practical applications of mood-congruent judgmental phenomena. The first group of studies explored the role of indirect informational mood effects in judgments and were primarily stimulated by affect-priming, associative network theories (Bower, 1981). We investigated the role of moods in the constructive processes of selective attention, learning, recall, and associations that are an essential part of the top-down nature of social judgments (Heider, 1958; Kelly, 1955). Studies in this group analyzed the conditions under which basic mood effect on judgments occur, ranging from studies of simple behavior interpretation tasks to more complex and demandingjudgments about interpersonal choices and attributions. However, our results eventually suggested a more complex picture than could be comfortably accommodated by an affect-priming framework, leading to the second series of experiments. The second group of studies addressed the role of different processing strategies in mediating mood effects on judgments, and several of these experiments test the implications of the multiprocess model outlined above. Studies in the second group explored the interactive role of mood and various features of the target, the judge, and the situation in cueing different processing strategies. Finally, the third group of investigations sought to extend these findings to applied judgments by children, by groups, and in various field settings. Most of these studies employ a two-stage procedure, wherein subjects sign up for two allegedly unrelated experiments. The first experiment-in fact, the mood manipulation-exposes subjects to mood-inducing stimuli, such as films, manipulated feedback about performance, hypnotic mood induction, or other positive or negative experiences. The second experiment asks subjects to perform a social judgment or decision under controlled conditions, in order to assess the impact of mood on judgmental outcomes. In all of these studies, mood induction procedures across experiments were varied in order to reduce the possibility of mood induction artifacts. In describing this research program we shall progress from simple, basic effects to more detailed analyses of specific judgmental processes and domains.

A. EVIDENCE FOR INFORMATIONAL (AFFECT-PRIMING) EFFECTS These studies investigated how affect priming may influence selective attention, learning, memory, and associative processes in social judgments. The domain of social judgments encompasses a variety of tasks, from complex

AFFECT IN SOCIAL JUDGMENTS

25 1

impression formation judgments and attributions, to the instant, automatic interpretation of ongoing social acts we observe. This research also explored the boundary conditions for mood-congruent judgmental effects.

I . Affect Priming in Impression Formation Impression formation is a typical example of a constructive social judgment wherein affect-priming principles can be expected to play a role. Perceiving another person involves selective attention to, and learning of information about, the target, the recall of relevant past information, and the interpretation of the available details in terms of primed cognitive constructs and associations. Affect should influence all of these processes, as we have seen above (cf. Bower, 1991; Forgas & Bower, 1988). Several experiments sought to establish the generality of the mood-congruent judgment phenomenon and to obtain direct evidence about mood-priming effects in the form of learning and judgmental latency data. In one such study of impression formation (Forgas & Bower, 1987), a happy, control, or sad mood was induced in subjects using false feedback about their performance on a prior, “unrelated” personality test. Subjects then read positive and negative details about a variety of other persons on a computer screen, before making impression formationjudgments about these targets. The computer recorded the time it took for happy, sad, or control subjects to read each item of positive or negative information, and to make each positive or a negative impression formation judgment. As predicted by affect-priming models, we did find a significant tendency for judgments to be biased in a mood-consistent direction. We also found specific support for the kind of processing differences predicted by affectpriming models. On the whole, subjects took longer to read and encode each item of mood-consistent rather than mood-inconsistent information about the targets (Fig. 3). This result is consistent with the notion that mood-consistent material should receive selective attention and more extensive processing and encoding, perhaps due to the larger number of associations required to link new information to already primed material in memory (Bower, 1981, 1991). Subsequent judgmental latency data were also consistent with affect-priming models. Mood-consistent judgments took significantly less time to make than did mood-inconsistent judgments, indicating the better prior learning and greater accessibility of mood-congruent constructs and associations used in forming an impression (Fig. 3). These reaction time results confirm that through “spreading activation, a dominant emotion will enhance the availability of emotioncongruent interpretations and the salience of congruent stimulus materials for learning” (Bower, 198 1, p. 45 1). However, these mood biases were not fully symmetrical (Taylor, 1991). We found weaker negative than positive mood effects, suggesting that additional normative or processing variables may have

252

JOSEPH PAUL FORGAS

Fig. 3. Mood effects on processing latencies: time taken in seconds to read a positive or a negative item and to make a positive or a negative judgment by people in happy or sad moods. After Forgas and Bower (1987).

influenced the results. In other studies we focused on simple, automatic behavior interpretation judgments, in order to explore the “boundary conditions” under which affective biases in judgments may still be demonstrated.

2 . Affect and the Interpretation of Social Behaviors The accurate monitoring and interpretation of simple social acts is an essential component of effective interpersonal behavior (Heider, 1958). Does affect influence such basic judgments? In one experiment, we sought to design a particularly challenging test of mood effects on such judgments, by (1) looking at the simplest kind of on-line judgments that are normally most likely to be data driven, and (2) by also providing subjects with objective, videotaped evidence about the social behaviors to be judged (Forgas et al., 1984). The study was carried out over two consecutive days. On the first day, pairs of subjects were videotaped whiIe engaging in four kinds of interviews with female confederates, of varying formality and intimacy (Forgas, 1979, 1982). The next day, the same subjects returned to participate in a “social perception experiment.” They were induced to feel happy or sad using a hypnotic mood induction procedure, and were then asked to look at the videotapes of the interaction episodes from the previous day. They were asked to monitor the tape and identify and score instances of positive/skilled and negative/unskiHed behaviors both for themselves and for their partner as they saw it on the videotape. We found a marked affective bias in these basic behavior interpretationjudgments across all four interaction episodes

AFFECT

IN SOCIAL JUDGMENTS

253

(Fig. 4). In a good mood, subjects saw more positive/skilled and fewer negative/ unskilled behaviors both in themselves and in their partners than did sad subjects. Sad subjects in contrast identified more negative than positive acts from the tapes, and were also markedly more critical of themselves than others, a pattern commonly found in depression (Roth & Rehm, 1980; Weary et a]. , 1991). Later judgments of the same tapes by observers who received no mood manipulation showed no such judgmental biases. There was also some evidence for mooddependent memory effects: Happy subjects recalled more details about easy, informal episodes whereas difficult, formal interactions were remembered relatively better by people in a sad mood. These results show that mood had a significant effect even on these most basic kinds of on-line judgments about social behaviors. It also seems from these results that self-related judgments may be particularly susceptible to negative affective biases. Within the affect-priming framework, this may be explained as due to the selective priming of self-deprecatory but other-enhancingcognitions in dysphoria, a pattern repeatedly found both in clinical and in normal populations (Forgas et al., 1990; Roth & Rehm, 1980). Further, the tendency for self-related information to receive more elaborate and substantive processing should also enhance the probability of mood-primed associations selectively influencing such judgments (Fiedler, 1990; Forgas, 1991b, in press-c). As this was a particularly challenging test of mood effects on very basic kinds of social judgments, it is likely that everyday judgments made of more complex stimuli and without the benefit of objective, videotaped evidence may be even more influenced by mood than was the case here.

Fig. 4. The effects of mood on perceptions of interactive behaviors in self and others: balance of mean number of positive minus negative acts seen for self and others. After Forgas, Bower, and Krantz (1984).

254

JOSEPH PAUL FORGAS

3. AfSect and Judgments about Achievement

Mood-primed associations should influence not only impression formation and behavior interpretation judgments, but also the way we make inferences about success and failure outcomes. In another series of experiments, we looked at the impact of mood on the way people explain doing well or doing badly in an achievement task, both for themselves and for others (Forgas et al., 1990). Attribution researchers have shown surprisingly little interest in studying perceiver characteristics such as mood, despite the roots of the paradigm in Heider’s (1958) phenomenological theories, and the clear role of motivation in achievement attributions (Weiner, 1980). Because causal attributions are not in themselves valenced, it is of particular interest to show mood influences on such judgments. In the first experiment on this topic, happy, sad, or neutral mood was induced by giving subjects bogus feedback about their performance on a test of verbal abilities, in an allegedly unrelated prior experiment. Subjects were then asked to make attributions for the success or failure outcomes of people who made risky or cautious choices when facing a typical life dilemma, such as succeeding or failing in a job, or winning or losing on an investment. As predicted, we found a clear mood-congruent effect, as subjects in a good mood were more likely to make benevolent attributions, crediting success to stable, internal causes, and blaming failure on external, unstable causes, than were sad subjects. Based on the intriguing evidence for self-other differences in earlier studies (Forgas et al., 1984), the next experiment explored whether mood has a differential influence on attributions for success and failure and to self versus others. The procedure was also changed to look at real life rather than hypothetical achievement outcomes here, and to assess cognitive as against motivational influences on attributions. Subjects were first shown happy, sad, or neutral videotapes, a procedure that was found effective in inducing different mood states in this and other experiments (Forgas, 1991a). Subjects then made attributions for a real-life achievement event, doing well or badly at a recent exam, and made these judgments for themselves as well as others in order to assess actor-observer differences (Forgas et al., 1990, Expt. 2). Results showed that in a positive mood subjects displayed a familiar self-serving bias, identifying stable causes for their own success, as well as the failures of others (Fig. 5). This pattern was entirely reversed in negative mood, when subjects became particularly critical of themselves, blaming stable and internal causes for their failure (Fig. 5). However, this bias did not generalize to others, who were in fact given more credit than the self for succeeding and less blame for failing by dysphoric subjects (see Fig. 5). This pattern seems clearly incompatible with motivational accounts, which should imply an increased self-serving bias in dysphoria. In later studies we again used videotapes to induce happy, neutral, and sad

255

AFFECT IN SOCIAL JUDGMENTS

1.5

POSITIVE MOOD

1.5 -Ef SELF

v)

NEGATIVE MOOD

A -OTHERS

1

1

5F: 3

0.5

m a I-

t

0

-0.5

SUCCESS

FA1LUR E

SUCCESS

FA1LURE

ACHIEVEMENT OUTCOME Fig. 5 . Mood effects on attributions for success and failure in an exam for self and others, to stable versus unstable causes. Positive values = stable causes. After Forgas. Bower, and Moylan (1990).

moods, but this time we asked subjects to make self-efficacy judgments and to also estimate the past and likely future exam performance of themselves and others. Results confirmed the earlier findings (Forgas et al., 1990; Expt. 3). Dysphoric subjects underestimated their own achievements and overestimated the achievements of others. This pattern was also repeated with self-efficacy judgments. Such results suggests that mood-induced attributional bias may come to determine eventual achievement, given the close links between self-efficacy judgments and subsequent performance (Kavanagh & Bower, 1985; Sedikides, 1991). These results also have implications for mood effects on social comparison judgments. It seems that negative mood may selectively induce selfdeprecating, upward comparison strategies, a pattern commonly reported in depression, as well as in several recent experimental studies of normal subjects (cf. Roth & Rehm, 1980; Swann, in press; Wheeler & Miyake, 1991). How can we best explain this self-specific bias for unfavorable attributions in sad moods? Within a multiprocess framework, we can account for these results in terms of the more extensive processing of self-related judgments and the preferential priming of self-deprecatory information in dysphoria, a pattern also commonly found in the clinical as well as the experimental literature (Forgas & Bower, 1987, 1988; Forgas et al., 1984; Ottaviani & Beck, 1988; Roth & Rehm, 1980). In contrast, motivational accounts of attribution would imply the opposite pattern, predicting increased self-serving bias when people need it most, in a

256

JOSEPH PAUL FORGAS

negative mood. Affect-as-information theories also cannot explain such targetspecific mood effects, as it is not clear why the “how do I feel about it?” heuristic would apply to both judgments of the self and others in a good mood, but only to judgments of the self in negative moods? One possibility is that these effects were mediated by different processing strategies. In a good mood, people may use careless, heuristic information processing, perhaps using their mood as information to infer more benevolent attributions both for themselves and others (Schwarz & Bless, 1991a). As suggested by the multiprocess model, negative mood in turn may lead to more substantive and elaborate processing, likely to further enhance the greater indirect availability of self-deprecating but otherenhancing cognitions primed by dysphoria. Several of our subsequent studies specifically addressed the role of mood and target characteristics in recruiting different processing strategies.

B. RESEARCH ON PROCESSING DIFFERENCES In terms of the multiprocess model, processing strategies should be dependent on an interaction of a person’s mood, and other factors, such as personal relevance, target typicality and familiarity and situational features. Such processing effects may in turn determine whether, and in what form, mood will be used as informative in computing a judgment. In one series of experiments, we looked at the interactive role of mood and personal relevance in the motivated processing of interpersonal decisions (Forgas, 199lc). 1 . Affect and Personal Relevance in Partner Choices

What role do feelings play in the interpersonal choices people make? Partner choice is an important and recumng aspect of daily life, and one of the most complex and demanding social judgments we make. Even long-term relationships can be thought of as based on a long sequence of positive partner choices (Levinger, 1990). In everyday life, choosing partners or colleagues is a critical task both in organizations and in our private lives. Pioneering work by Schachter (1959) showed that affect has a major influence on interpersonal choices. Anxious or frightened people seem not only to prefer the company of others, but tend to seek out those in a similar predicament, presumably as part of a motivated mood-control strategy (Clark & Isen, 1982). It seems that it is more rewarding to interact with partners who are in a matching rather than different mood from us (Locke & Horowitz, 1990). Remarkably, we know very little about the information-processing strategies people employ when making such motivated choices, despite growing evidence for the motivated character of many cognitive processes (Kunda, 1990; Martin et

AFFECT IN SOCIAL JUDGMENTS

257

al., 1991; Parrott & Sabini, 1990). In this series of studies, we asked people in happy, sad, or neutral moods to choose a partner either for themselves (personally relevant choice) or for another subject (personally irrelevant choice). We expected that when making a personally relevant choice, people in negative moods in particular may rely on motivated processing strategies in order to find a rewarding companion (Forgas, 1991~).In these studies we noted not only the kind of interpersonal choices subjects made, but also the kind of information they selected and the decision-making strategies they adopted (Forgas, 1989). In the first study in this series, subjects saw happy, sad, or affectively neutral videotapes as part of the mood induction, and were then asked to choose a partner for themselves or for another subject from four available candidates (Forgas, 1991c, Expt. I). Each candidate was described in terms of several social and task-related and desirable as well as undesirable characteristics. We found that the combination of sad mood and personal relevance triggered an apparently motivated decision strategy. Sad subjects when making a personal choice significantly preferred socially rewarding to task-competent partners, and also better remembered diagnostic information about such partners later on, suggesting their better learning of such details. In the follow-up study, the number of potential partners was increased and descriptions were provided on a series of cards making up “personal files” for each candidate (Forgas, 1991c, Expt. 2). Subjects were told to number and rate the relevance of each information unit they considered, in order to obtain a stepby-step record of the precise information search and decision strategies they used. Once again, we found strong evidence for motivated processing by dysphoric subjects when making a personal choice. These subjects significantly preferred socially skilled partners, while in all other conditions competent partners were chosen. It also took less time for sad subjects to make such a motivated, personal choice, whereas happy subjects were faster when making impersonal choices. Sad subjects also significantly preferred to look at social than task information during their deliberations, and were more likely to take consecutive steps in looking at information about one target (an impression formation strategy), instead of comparing several targets in terms of a particular feature (a comparison-by-features strategy). This result differs from previous findings with complex but personally uninvolving choices (lsen 8c Means, 1983), and suggests that motivated information search strategies do play a role in personally relevant judgments (Janis & Mann, 1977; Kunda, 1990). The procedure was further improved in later studies, wherein a computercontrolled procedure was used to present the target information and to record the choice latencies and step-by-step information search strategies used by subjects (Forgas, 1991c, Expt. 3). Results confirmed the use of motivated choice strategies. We found that dysphoric subjects making a personal choice were quicker to select social rather than task information, but studied such details at

25 8

JOSEPH PAUL FORGAS

greater length and more exhaustively (Fig. 6). These findings are generally consistent with our expectation that good moods and personally uninvolving choices are more likely to trigger shorter, simplified processing strategies, while the combination of sad mood and personal relevance is more likely to lead to motivated processing. In this case, dysphoria cued intentional, motivated search strategies and selective exposure to information about rewarding partners (Fig. 6). This strategy confirms other evidence for motivated thought in cognition research (Kunda, 1990; Martin et al., 1991; Petty & Cacioppo, 1986; Schwarz, 1990; Strack, Schwarz, & Gschneidinger, 1985; Tesser, 1986). It is interesting that motivated processing in dysphoria also led to faster and more efficient decisions, as subjects engage in a selective, guided search selecting only information relevant to their motivational objective. Motivated processing is probably a common and as yet little researched strategy in interpersonaljudgments both by normal and by depressed populations (Ottaviani & Beck, 1988; Weary et al., 1991) that must be incorporated in any comprehensive model of judgmental processes, as has been proposed here.

2 . Affect and Target Typicality in Processing Choices According to the multiprocess framework, target features are an important influence on processing choices (Fig. 1). Atypical targets in particular may selectively recruit more substantive processing strategies, and correspondingly greater mood-priming effects on such judgments. Personality traits commonly form recognizable, prototypical patterns, and categories such as “the loner,” the “housewife type,” or “the student activist” are often used to describe typical

Fig. 6. The effects of a mood and personal relevance on motivated processing: Dysphoric subjects making a personal affiliative choice selectively expose themselves to information about potentially rewarding partners.

AFFECT IN SOCIAL JUDGMENTS

259

persons. Most groups or subcultures develop their own repertoire of recognizable person types (Forgas, 1983a; Fussell, 1990), defined by a particular pattern of nonunique features (Rosch & Lloyd, 1978), and individual exemplars differ in the extent to which they approximate the ideal, prototypical pattern. It seems that being typical is sometimes helpful and sometimes not helpful in thinking about a person (Rojahn & Pettigrew, 1990). Some studies show that “the greater the prototypicality . . . the more easily would information about that person be encoded, retrieved and elaborated” (Cantor & Mischel, 1979, p. 191). Other evidence suggests that “acts that are unexpected or incongruent . . . are well remembered” (Hastie & Kumar, 1979, p. 31). In a series of experiments attempting to reconcile these conflicting models, we found that target features such as affective salience can play a critical role in triggering either heuristic, prototype-assimilation or substantive, prototype-contrast processing strategies (Forgas, 1983a, 1985a). We found that high salient prototypes were processed heuristically, as they “dominate impressions, leading to the poor processing of prototype-inconsistent characters. In the case of low salient prototypes the opposite effect was found, with novel and prototype-inconsistent information . . . leading to superior recall and predictions” (1983a, p. 156), consistent with the more substantive processing of these targets. Extrapolating from these results, a recent series of studies looked at the interactive role of mood and target prototypicality in triggering heuristic or substantive processing strategies (Forgas, in press-a). Generally, we expected that mood should have a greater influence on judgments about atypical rather than typical people, as atypical targets require more substantive, elaborate processing, with a greater scope for mood-primed associations to enter into such judgments. Accordingly, in these studies mood and target prototypicality were factorially manipulated in an attempt to assess precisely the interactive effects of these factors on judgments. In the first study, subjects feeling happy, sad, or neutral after watching videotapes read descriptions of nine target characters based on separately elicited and ecologically valid student types familiar in their social milieu (Forgas, in press-a, Expt. I). The prototypicality of the targets was manipulated so that three targets were described as highly prototypical (all features consistent with the type), three were highly atypical (half the features were contrary to the type), and three targets were intermediate (half the features were unrelated to the type). We found an overall significant mood-congruent effect on judgments. However, there was also a significant mood by prototypicality interaction. As expected, mood had a far greater impact on judgments of atypical characters, which require more systematic and detailed processing, than on judgments of typical characters (Fig. 7). The second experiment sought independent support for the hypothesized processing differences in terms of memory data (Forgas, in press-a, Expt. 2). The

260

JOSEPH PAUL FORGAS

PROTOTYPICAL 4.15

- 4.15

3.95

- 3.95

3.75

- 3.75

3.55

- 3.55

3.35

- 3.35

3.15

-S-

LIKABILITY

- 3.15

4- SELF-CONFIDENCE 2.95

-&+-

- 2.95

COMPETENCE I

2.75

Happy

Control

I

I

~

2.75

Sad

MOOD CONDITION Fig. 7. Mood effects on the perception of typical and atypical persons: Mood has a greater impact on judgments of atypical persons, requiring more substantive and elaborate processing.

procedure was essentially repeated, but thematically different mood induction films were used and the number of targets was changed from nine to six in order to reduce processing load. We expected and found that recall was overall better for atypical targets, probably due to the greater salience and informativeness of unusual, unexpected characters. We also found a significant interaction between mood and prototypicality. Mood had an overall greater effect on memory for atypical rather than typical targets. We also found evidence for an intriguing asymmetric mood effect here: Recall was comparatively better for atypical targets in a negative mood, and for typical people in positive mood (Fig. 8). This pattern suggests that encoding information about atypical people was facilitated by the kind of careful, substantive processing strategies usually associated with dysphoria (Schwarz, 1990), while learning about prototypical characters was apparently easier in a positive mood, which usually cues heuristic, schematic processing styles (Isen, 1984). Such asymmetric mood effects suggest a delicate interaction between moods and target features in the recruitment of different processing strategies. In terms of the multiprocess model, best recall performance seems to occur in conditions when both target typicality and mood trigger similar rather than conflicting processing styles (Fig. 8). The third experiment aimed to replicate the memory and judgmental results of the first two studies, and to provide direct evidence about processing differences due to mood and target typicality in the form of processing latency measures (Forgas, in press-a, Expt. 3). A computer-controlled stimulus presentation and

26 1

AFFECT IN SOCIAL JUDGMENTS

/

46 -

-TYPICAL

TARGET

0ATYPICAL TARGET

44-

2 el

r z t-

i

4240-

38-

!36p:

8

34Sad

MOOD CONDITION Fig. 8. Mood effects on memory for typical and atypical persons: There is evidence for better recall of typical people by subjects in a good mood, and of atypical people by subject in a bad mood. Higher values indicate better recall.

judgmental procedure was used to allow the collection of reaction time data. We also used a different mood induction method here, this time giving subjects manipulated feedback about their performance on a prior measure of verbal abilities, in order to reduce the likelihood of mood induction artifacts associated with any one method. We again found significant mood effects on judgments, which were markedly greater for atypical rather than typical targets. Encoding latencies were slower in negative than in positive moods, and when dealing with atypical rather than typical targets. Judgments of atypical targets also took longer to make than judgments of prototypical characters. These data help to establish that atypical targets did indeed receive more detailed and substantive processing at the encoding stage than did typical targets, giving greater scope for moodprimed associations to enter into such judgments. Memory data were also consistent with this interpretation, with relatively better recall of atypical targets in negative mood, and typical targets in positive mood (cf. Fig. 8). These results seem consistent with the multiprocess model, indicating that target typicality and mood have a joint effect on processing choices. The adoption of more substantive processing for atypical targets was the prerequisite for stronger mood-priming effects to occur. These findings also suggest that in the absence of any evidence for heuristic processing and reduced processing times, affect-as-information processes probably had no effect on these judgments. 3. Affect in the Perception of Relationships

Other experiments explored the role of mood and target typicality in judgments of pictorial rather than verbal stimulus materials. For example, judgments of

262

JOSEPH PAUL FORGAS

interpersonal relationships are frequently based on the observable physical characteristics of a couple, particularly at the early stages of unilateral awareness or surface contact (Levinger & Snoek, 1972). As interpersonal relationships are culturally defined and consensually shared within a social milieu, such judgments also reflect people’s ideas about the desirable features of typical relationships (Levinger, 1990). Generally, typical relationships are expected to be balanced, with a good overall match between the partners (Forgas & Dobosz, 1980). Reinforcement, social exchange, or balance theories of relationships all emphasize the importance of partner match, as do empirical studies of relationships (Heider, 1958; Forgas, 1988; Forgas & Dobosz, 1980). In these experiments we investigated mood effects on judgments of heterosexual relationships (Forgas, 199Id, in press-d). Target features were manipulated in terms of the match in physical attractiveness between the partners, creating typical, well-matched or atypical, ill-matched couples. We expected an overall mood-congruent bias in these judgments. We also expected mood effects to be greater in judgments of atypical, mismatched couples, as such targets presumably require more substantive, elaborative processing, with a correspondingly greater role for mood-primed associations to enter into such judgments. In one such study, videotapes were again used to induce happy, sad, or neutral moods in subjects. This was followed by impression formation judgments of relationships wherein the couples shown were either well-matched or ill-matched in terms of physical attractiveness. The evaluation of the relationship was significantly more positive by happy than by sad subjects. However, the effects of mood were again much greater when judging pictures of atypical (badly matched) rather than typical (well-matched) couples, consistent with our hypothesis that mood-primed associations play a greater role in judgments that require more substantive, elaborative processing. In terms of the multiprocess model, these results confirm that target characteristics can have a critical influence on determining processing strategies, and consequently the nature of mood effects on judgments. The findings also suggest that target typicality may be an important variable mediating mood effects on stereotypejudgments as well. We further explored this possibility in a subsequent experiment dealing with racial stereotypes. 4 . Affect and Stereotype Judgments

Social judgments in general, and stereotype judgments in particular, are constructive, wherein observable racial features often serve as the basis of categorybased inferences about people (Fiske & Neuberg, 1990; Tajfel & Forgas, 1982). There has been surprisingly little work on affective influences on stereotyping, although the cognitive (Hamilton & Rose, 1980) and motivational (Tajfel & Forgas, 1982) aspects of stereotypes have long been recognized. Recently, Mackie et al. (1989) found that incongruence between mood and the valence of

AFFECT IN SOCIAL JUDGMENTS

263

target information increases subjects’ reliance on illusory correlations in stereotype judgments, whereas Stroessner et a]. (in press) report that only subjects in a neutral mood selectively attended to infrequent behaviors by minority group members. Other work suggests that stereotyping is more likely when the task is complex, difficult, or there are other constraints (Kruglanski & Freund, 1983). In this study we expected that an unusual target context, such as encountering a person as part of a mixed-race dyad, should trigger a more detailed, substantive processing strategy, allowing a greater scope for mood-consistent associations to enter these judgments, than seeing the same person in a same-race dyad (Forgas, 1991d). This expectation is based on the rnultiprocess model and is supported by the results of the previous studies. In general, racially balanced relationships may be more typical, and good match-including racial match-seems a common feature of most relationships (Forgas & Dobosz, 1980). Deviations from prototypical, matched configurations are unusual and should lead to more detailed, substantive processing. In contrast, prototypical targets (a same-race dyad) should receive less substantive processing and should display weaker mood effects (Forgas, in press-d, Forgas & Moylan, 1991). In this study videotapes were again used to induce happy, sad, or neutral moods. Subjects were then asked to look at and form impressions of Caucasian or Asian males or females who were presented as part of a same-race or a mixedrace heterosexual dyad. Overall, judgments were more positive in a good than in a bad mood, and when the partner was of the same race rather than a different race. There was also a significant affect by racial match interaction. Mood effects on judgments were significantly greater when targets were encountered as part of an unusual target configuration, a mixed-race dyad, rather than in matched-race dyad (Fig. 9). This pattern is consistent with our expectation that mood should have a greater effect on judgments of targets that require more substantive, constructive processing strategies. This experiment provides convergent evidence for the important role target typicality can play in guiding processing choices. It seems that mood effects on stereotype judgments may be accentuated in conditions that are conducive to more constructive, substantive processing. The implications of these studies for everyday social judgments may be farreaching: Perceptions of unusual, atypical people or people encountered in an unusual context may be significantly more likely to be influenced by mood effects than judgments of typical targets. The multiprocess model offers a comprehensive framework for an understanding of such target-specific mood effects.

C. EXTENSIONS AND APPLICATIONS The research surveyed so far shows consistent and enduring mood effects on a variety of social judgments, due to both informational and processing influences. However, the evidence comes largely from single adult individuals making judg-

264

JOSEPH PAUL FORGAS

Fig. 9. The effects of mood on judgments of racially matched or racially unmatched couples in terms of likability and competence: Mood has a greater influence on judgments of mixed-race targets, which require more extensive substantive processing. Higher values indicate more positive evaluations.

ments in controlled environments. In several of our studies we sought to expand the generality of this phenomenon by looking at affective influences on social judgments in real-life field settings, in group settings, and also by children. 1 . Affect in Memory and Judgments by Children What role do moods play in the way young children, as against adults, learn, remember, and use information in their social judgments? There is some evidence that children are not only “cognitive aliens,” in Piaget’s terms, but may also deal with emotional stimuli differently from adults. In particular, the absence of strongly internalized adult norms about affect control may make findings difficult to generalize to children. In one of our studies we looked at mood effects on children’s memory and impression formation judgments. Children were presented with videotapes of two target children of their own age while experiencing happy or sad moods, after watching excerpts from cheerful or depressing films designed to influence their mood state (e.g., scenes depicting a birthday celebration, or the death of child’s pet dog). One day later, a matching or nonmatching mood state was induced using different films, and children were asked to recall and form impressions about the targets they saw on the previous day (Forgas, Burnham, & Trimboli, 1988). Results showed clear evidence for mood-state-dependent memory. Characters

AFFECT IN SOCIAL JUDGMENTS

265

seen in matching encoding and retrieval moods were remembered better. Different learning moods for the two targets also helped to improve memory for both, as mood presumably provided a helpful differentiating context for encoding (Bower, 1981). There was also a significant negativity bias in memory and judgments. This possibly reflects the weaker emotional socialization of children and their more restricted repertoire of emotional experiences. As children may not have fully internalized adult norms controlling expressions of negativity about others, their memory and judgments seemed to reflect fully the greater informativeness and salience of negative information about others. Results such as these suggest that mood effects on social judgments by adults may partly reflect adult patterns of emotional socialization, and may not be readily generalized to children.

2. Affect in Individual versus Group Judgments Not all social judgments are made by individuals. Many important decisions and judgments in our society are entrusted to groups or committees in the belief that their decision-making processes will be less biased and more representative than individual judgments. Yet almost all relevant research to date focused on individuals instead of groups (Argyle, 1991), despite evidence that group discussion per se may be the source of significant bias in judgments (Forgas, 1977, 1981a). In one of our recent studies (Forgas, 1990) subjects were presented with happy, sad, or neutral videotapes to induce the appropriate mood state, and were then asked to make individual or group judgments about nine person types (doctors, farmers, Catholics, Italians, etc.) on two different occasions separated by a 2-week interval. Results showed a clear mood-congruence effect, as judgments made by individual judges were more positive in a happy mood and more negative in a sad mood. However, mood had a markedly different effect on judgments made by groups. In a positive mood, groups made even more positive judgments than did individuals, consistent with research on group extremity shifts (Forgas, 1977) predicting the further polarization of opinions following group interaction due to normative and informational processes in the groups. Remarkably, such a polarization of judgments did not occur in a negative mood. In fact, it seems that on all three judgmental dimensions, group discussion did not increase the extremity of negative judgments, and for one dimension, it even resulted in a positive shift when compared to individual judgments. This effect may be explained in terms of the influence of dysphoria on the quality of the group interaction process. Previous research showed that a free, unconstrained, and involved group interaction is essential for an extremity shift to occur (Forgas, 1977, 1981a). Negative mood is probably incompatible with such an unconstrained interaction style, leading to the elimination of the group polarization effect. Further, group interaction should also lead to an increased

266

JOSEPH PAUL FORGAS

awareness of social norms controlling negative assessments of others, which should have a selective impact on constraining dysphoric subjects from forming more negative judgments. In other words, group interaction may also be influenced by the careful, analytic processing style normally associated with dysphoria, leading to an elimination of extremity shifts. Further research looking at the impact of positive and negative moods on group interaction strategies is needed to establish precisely the role of affect in such collective judgments (Argyle, 1991). Generally, our results suggest that there is an acute need to extend the currently predominant focus of social cognition research on isolated individuals to a more comprehensive treatment of complex, interactivejudgmental processes as they occur in everyday life (Forgas, 1981b). 3 . Mood and Judgments in Natural Situations

In addition to the extensive laboratory evidence summarized above, affective distortions in social judgments do of course occur in many natural, real-life situations. Affect can have a major impact on social judgments in everyday organizational settings (Isen and Baron, 1991), in clinical practice (Ottaviani & Beck, 1988), in medical contexts (Salovey et al., 1991), in judgments of subjective well-being (Strack et al., 1991), as well as in responses to survey questions (Schwarz & Strack, 1991). With notable exceptions (Mayer et a]., 1991; Schw a n & Clore, 1988), relatively few studies looked at the effects of naturally occurring mood on judgments. As part of our research program we also carried out a number of field studies, using unobtrusive techniques relying on nitturally occurring events as mood manipulations. Supporters of winning or losing sports teams, people who won or lost money while gambling, or people who received a small gift or a candy in a shopping center were our subjects, answering a variety of questions dealing with personal, social, or political judgments. In one representative study, almost 1000 visitors to happy, sad, or aggressive movies were interviewed immediately after leaving a movie theatre (Forgas & Moylan, 1987). They were asked to answer public opinion survey-type questions about their perception of topical issues, public figures, political parties, as well as their life satisfaction and expectations about the future. There were no significant differences between people tested before they saw each of the films. However, on all questions, subjects in a good mood after a happy film gave significantly more positive and lenient judgments and attributions than subjects who saw a sad or aggressive film, suggesting that mood-based distortions in social judgments are indeed a common-and probably much underestimated-feature of everyday life. Certainly there is much scope for further applied research to assess precisely the real-life implications of the kind of mood-based distortions in social judgments demonstrated by ourselves and others, a task that is now undertaken with increasing vigor (cf. Isen & Baron, 1991; Mayer et al., 1991; Salovey et al.,

AFFECT IN SOCIAL JUDGMENTS

267

1991). This work suggests that general moods may have a greater impact on natural judgments than do specific emotions, and vague, nonspecific questions are more likely to be subject to mood effects than specific, unambiguous items (Fiedler, 1991; Mayer et al., 1991). These results not only support the ecological validity of the phenomenon of mood-congruent judgments, but are also consistent with the implications of the multiprocess model (Forgas, 1991a).

V. Summary and Conclusions

Social judgments by definition involve high-level, abstract cognitive processes in the selection, learning, retrieval, and interpretation of information about target characteristics that are usually not directly observable, and the selective use of past memories and associations in this process (Heider, 1958; Kelly, 1955). In the first part of this article we surveyed early research on mood effects on social judgments, and reviewed current theoretical explanations of this effect. It was proposed that the judgmental consequences of affect are due to a combination of informational (affect priming or affect-as-information) and processing influences. An integrative multiprocess model was proposed here as a comprehensive account of mood effects on judgments, highlighting the role of contextual variables in processing choices. In the second half of this article a description of a series of our experiments suggested that temporary affective states do have a strong and consistent effect on a variety of social judgments, ranging from simple behavior monitoring tasks to complex attribution and personal preference tasks. Several studies produced evidence for the indirect informational effects of mood, due to the selective priming of mood-related associations. A second group of studies showed that feelings do have a crucial influence on the cognitive processes involved in social judgments (cf. Forgas & Bower, 1987, 1988). In particular, variables such as personal relevance or target typicality were found to interact with mood state in determining processing choices and the ensuing mood effects on memory and judgments. Most traditional as well as contemporary theories of social judgments are based on single-process assumptions-that is, the belief that there is one enduring way of processing information about social targets (Anderson, 1974; Asch, 1946; Wyer & Srull, 1989). Some recent models have proposed dual-process formulations to account for otherwise intractable judgmental phenomena (Brewer, 1988; Fiedler, 1991; Fiske & Neuberg, 1990; Kruglanski, 1989; Petty et al., 1991; Stangor & Ford, 1992). In contrast, the multiprocess model developed here specifies four information-processingstrategies available to compute a

268

JOSEPH PAUL FORGAS

social judgment, as well as the contextual variables that influence such processing choices. The processing strategies identified here were (1) the direct access strategy, when preexisting, crystallized judgments are simply retrieved from memory, (2) the motivated processing strategy, when the motivation to achieve a particular judgmental outcome overrides the usual processing alternatives, (3) the heuristic processing strategy, when judges choose to ignore or bypass relevant information in order to simplify the judgmental task and form an impression using various shortcuts, and (4) the substantive processing strategy, when judgments are computed on the basis of the selective processing of the available information, involving learning, assoc'iative, and memory processes. Processing choice is determined by a combination of judge, target, and situational characteristics, and as we have seen, a careful experimental analysis of processing choices is by no means impossible. The research evidence surveyed seems generally consistent with a multiprocess approach. In several studies, we found evidence for the operation of affective influences on substantive processing strategies, leading to the selective learning, interpretation, and recall of valenced information, and ultimately the distortion of judgments in an affect-consistent direction (cf. Forgas & Bower, 1987; Forgas et al., 1984). Actual differences in processing times and judgmental latencies in particular support the various mood-priming formulations when substantive processing is used (Forgas & Bower, 1987; Forgas, in press-a, in press-c). Other studies produced evidence indicating the role of mood and personal relevance in triggering motivated, controlled processing strategies in reaching a judgment. This was particularly the case when dysphoric subjects had the task of making partner choices (Forgas, 1989, 1991~).Under these circumstances, the motivation to obtain rewarding, pleasant partners was apparently paramount in the judgmental processes adopted. Others have reported strong evidence for mood effects on judgments consistent with a heuristic processing strategy (Clore & Parrott, 1991; Schwarz & Bless, 1991a; Schwarz & Clore, 1988). At the most general level, these results suggest that there is a broad and pervasive tendency for people to form social judgments in terms of their feelings at the time, and to adopt different information-processing strategies depending on their temporary affective state. Mood-induced differences in information selection, retrieval, and interpretation of the kind demonstrated here may have particularly important consequences for everyday social decisions and judgments. There is some recent evidence suggesting that mood-priming effects on cognition and judgments are most likely to be significant when the information base is complex and elaborate, selective and constructive processing is required, and the evidence is capable of supprting alternative interpretations (cf. Forgas et al., 1990). These conditions are met in many everyday social judgmental tasks both in people's private and in their working lives. This conclusion must be tempered by several qualifications. There

AFFECT IN SOCIAL JUDGMENTS

269

is growing evidence to suggest that positive and negative moods are not equally effective in influencing judgments. Negative mood effects are often less robust and more context dependent. Further, not all judgments are equally influenced by mood. Negative moods are more likely to influence self-relevant judgments, suggesting that in addition to affect priming (Bower, 1981), a variety of other social, motivational, or personal factors also influence how affect influences cognition and judgments. The role of affect in triggering different informationprocessing strategies remains one of the most important domains for future research. The studies reported in this article emphasize the complex, constructive character of social judgments and the interactive role of affect and other variables in determining processing choices. We may conclude that although much has been discovered about the informational and processing consequences of affect, not enough is known about the interaction of affect and stimulus characteristics. Several of our studies showed that complex, unusual, or atypical stimulus characters often elicit a more detailed, systematic, and substantive processing style, with significantly greater affective biases on such judgments. The proposed multiprocess framework should be particularly useful in explaining such contextspecific effects and in highlighting potentially fruitful areas for future research.

Acknowledgments Support by the Australian Research Council, the German Research Foundation, and ZUMA, Mannheim, Germany, and the contribution of Stephanie Moylan and Sandy Morrison to this project are gratefully acknowledged. I am indebted to Gordon Bower, Klaus Fiedler, Norbert Schwarz, Fritz Strack, Bob Wyer, and Mark Zanna for their stimulating ideas and comments on some of the work presented here.

References Anderson, N. H. (1974). Cognitive algebra: Integration theory applied to social attribution. In L. Berkowitz (Ed.), Advances in experimenral social psychology (Vol. 7, pp. 1-101). New York: Academic Press. Argyle. M. (1991). A critique of cognitive approaches to social judgments and social behaviour. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Asch, S. E. (1946). Forming impressions of personality. Journal ofAbnormal and Social Psychology, 41, 258-290. Blaney, P. H. (1986). Affect and memory: A review. Psychological Bulletin, 99, 229-246. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129-148.

270

JOSEPH PAUL FORGAS

Bower, G. H. (1983). Affect and cognition. Philosophical Transactions of the Royal Society of London, Series E . 302, 387-403. Bower, G. H. (1991). Mood congruity of social judgments. In J. P.Forgas (Ed.), Emotion andsocial judgments. Oxford: Pergamon. Bower, G. H., & Cohen, P. R. (1982). Emotional influences in memory and thinking: Data and theory. In M. S. Clark & S. T. Fiske (Eds.), Affect and cognition. Hillsdale, NJ: Erlbaum. Bower, G . H., & Mayer, J. D. (1985). Failure to replicate mood-dependent retrieval. Bulletin ofthe Psychonomic Society, 23, 39-42. Branscombe, N. (1988). Conscious and unconscious processing of affective and cognitive information. In K. Fiedler & J. P. Forgas (Eds.), Affect, cognition, and social behavior. Toronto: Hogrefe International. Branscombe, N., & Cohen, P. (1991). Motivation and complexity levels as determinants of heuristic use in social judgment. In J. P. Forgas (Ed.), Emotion and social judgments (pp. 145-160). Oxford: Pergamon. Brewer, M. (1988). A dual-process model of impression formation. In T. K. Srull & R. S. Wyer (Eds.), Advances in social cognition. Hillsdale, NJ: Erlbaum. Brewer, M., & Kramer, R. M. (1985). The psychology of intergroup attitudes and behavior. Annual Review of Psychology, 36, 219-243. Byrne, D., & Clore, G . L. (1970). A reinforcement model of evaluation responses. Personality: A n International Journal, 1, 103-128. Cantor, N., & Mischel, W. (1979). Prototypes in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 12). New York: Academic Press. Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766. Clark, M. S., & Isen, A. M. (1982). Towards understanding the relationship between feeling states and social behavior. In A. H. Hastorf & A. M. Isen (Eds.), Cognitive social psychology. New York: Elsevier/North-Holland. Clore, G. L., & Byrne, D. (1984). The reinforcement affect model of attraction. In T. L. Huston (Ed.). Foundations of interpersonal attraction (pp. 143-170). New York Academic Press. Clore, G . L., & Parrott, G. (1991). Moods and their vicissitudes: Thoughts and feelings as information. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Descartes, R. (1961). Passions of the soul: Essential works of Descartes. (L. Blair, Trans.). New York: Bantam Books. (Original work published 1649). Eagly, A,, & Chaiken, S. (1991). The psychology of attitudes. New York: Harcourt Brace Jovanovich . Ellis, H. C., & Ashbrook, T. W. (1988). Resource allocation model of the effects of depressed mood state on memory. In K. Fiedler & J. P. Forgas (Eds.), Affect, cognition and social behaviour. Toronto: Hogrefe International. Erber, R. (1991). Affective and semantic priming: Effects of mood on category accessibility and inference. Journal of Experimental Social Psychology, 27, 480-498. Feshbach, S., & Singer, R. D. (1957). The effects of fear arousal and suppression of fear upon social perception. Journal of Abnormal and Social Psychology, 55, 283-288. Fiedler, K. (1990). Mood-dependent selectivity in social cognition. In W. Stroebe & M. Hewstone (Fds.), European review of social psychology (Vol. 1, pp. 1-32). Chichester: Wiley. Fiedler, K. (1991). On the task, the measures and the mood: Research on affect and social cognition. In J. P. Forgas (Ed.), Emorion and social judgments. Oxford: Pergamon. Fiedler, K., & Forgas, J. P. (Eds.). (1988). Affect, cognition and social behavior. Toronto: Hogrefe International. Fiske, S. T., & Neuberg, L. (1990). A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpreta-

AFFECT IN SOCIAL JUDGMENTS

27 1

tion. In M. Zanna (Ed.), Advances in experimental social psychology. San Diego: Academic Press.

Forgas. J. P. (1977). Polarisation and moderation of person perception judgements as a function of group interaction styles. European Journal of Social Psychology. 7, 175-187. Forgas, J. P. (1979). Social episodes: The study of interaction routines. London & New York: Academic Press. Forgas, J. P. (1981a). Responsibility attribution by groups and individuals: The effects of the interaction episode. European Journal of Social Psychology, 11, 87-99. Forgas, J. P. (Ed.).(1981b). Social cognition: Perspectives on everyday understanding. London & New York: Academic Press. Forgas, J. P. (1982). Episode cognition: Internal representations of interaction routines. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 15, pp. 59- 104). New York: Academic Press. Forgas, J. P. (1983a). The effects of prototypicality and cultural salience on perceptions of people. Journal of Research in Personality. 17, 153-173. Forgas, J. P. (1983b). What is social about social cognition? British Journal of Social Psychology. 22, 129-144. Forgas, J . P. (1985a). Person prototypes and cultural salience: The role of cognitive and cultural factors in impression formation. British Journal of Social Psychology, 24, 3- 17. Forgas, J. P. (Ed.) (1985b). Language and social siruarions. New York: Springer. Forgas, 1. P. (1988). The role of physical attractiveness in the interpretation of facial expression cues. Personality and Social Psychology Bulletin, 13, 478-489. Forgas, J. P. (1989). Mood effects on decision-making strategies. Ausrralian Journal of Psychology, 41, 197-214. Forgas, J . P. (1990). Affective influences on individual and group judgments. European Journal of Social Psychology. 20, 441-453. Forgas, 1. P. (Ed.) (1991a). Emotion and social judgments. Oxford: Pergamon. Forgas, J. P. (1991b). Affective influences on the perception of balanced and unbalanced relationships. Manuscript, University of New South Wales. Forgas, J. P. (1991~).Affective influences on partner choice: Role of mood in social decisions. Journal of Personality and Social Psychology. Forgas, J. P. (1991d). Mood effects on stereotype judgments. Manuscript, University of New South Wales. Forgas, J. P. (in press-a). On mood and peculiar people: Affect and person typicality in impression formation. Journal of Personality and Social Psychology. Forgas, J. P. (in press-b). Mood effects on the perception of atypical persons. European Journal of Social Psychology. Forgas, J. P. (in press-c). Affect and person perception: Empirical evidence and an integrative model. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology. Chichester: Wiley. Forgas, J. P. (in press-d). On making sense of odd couples: Mood effects on the perception of atypical relationships. Personality and Social Psychology Bulletin. Forgas, J. P., & Bower, G. H. (1987). Mood effects on person perception judgments. Journal of Personality and Social Psychology. 53, 53-60. Forgas, J. P., & Bower, G. H. (1988). Affect in social judgments. Australian Journaloffsychology, 40, 125-145. Forgas, J . P., Bower, G. H., & Krantz, S. (1984). The influence of mood on perceptions of social interactions. Journal of Experimental Social Psychology, 20, 497-5 13. Forgas, 1. P., Bower, G . H.,& Moylan, S. J. (1990). Praise or blame? Affective influences on attributions for achievement. Journal of Personality and Social Psychology, 59, 809-818.

272

JOSEPH PAUL FORGAS

Forgas, J. P., Bumham, D., & Trimboli, C. (1988). Mood, memory and social judgements in children. Journal of Personality and Social Psychology, 54, 697-703. Forgas, J. P., & Dobosz, B. (1980). Dimensions of romantic involvement: Towards a taxonomy of heterosexual relationships. Social Psychology Quarterly. 43, 290-300. Forgas, J. P., & Moylan, S. J. (1987). After the movies: The effects of transient mood states on social judgments. Personality and Social Psychology Bulletin, 13, 478-489. Forgas, J. P., & Moylan, S. J. (1991). Affective influences on stereotype judgments. Cognirion & Emotion, 5, 379-395. Frijda, N. H. (1986). The emotions. Cambridge: Cambridge University Press. Fussell, S.R. (1990). Social categories on campus. Manuscript, Princeton University, Princeton, NJ. Gouaux, C. (1971). Induced affective states and interpersonal attraction. Journal of Personality and Social Psychology, 20, 37-43. Gouaux, C . , & Summers, K. (1973). Interpersonal attraction as a function of affective states and affective change. Journal of Research in Personality, 7 , 254-260. Griffith, W. ( 1970). Environmental effects on interpersonal behavior: Ambient effective temperature and attraction. Journal of Personality and Social Psychology. 15, 240-244. Hamilton, D. L., & Rose, T. L. (1980). Illusory correlation and the maintenance of stereotypic beliefs. Journal of Personality and Social Psychology, 39, 832-845. Hastie, R., & Kumar, P. A. (1979). Person memory: Personality traits as organizing principles in memory for behavior. Journal of Personality and Social Psychology, 37, 25-38. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Hilgard, E. R. (1980). The trilogy of mind: Cognition, affection, and conation. Journal of the History of the Behavioral Sciences, 16, 107-1 17. Isen, A. (1984). Towards understanding the role of affect in cognition. In R. S.Wyer & T. K. S ~ l l (Eds.), Handbook of social cognition (Vol. 3). Hillsdale, NJ: Erlbaum. Isen, A. (1987). Positive affect, cognitive processes and social behaviour. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 20, pp. 203-253). Orlando, FL: Academic Press. Isen, A. M., & Baron, R. A. (1991). Positive affect as a factor in organizational behaviour. In B. M. Stan & L. L. Cummings (Eds.), Research in organizational behaviour. New York: JAI Press. Isen, A. M., & Means, B. (1983). The influence of positive affect on decision making strategy. Social Cognition, 2, 18-3 I. Izard, C. E. (1964). The effect of role-played emotion on affective reactions, intellectual functioning and evaluative ratings of the actress. Journal of Clinical Psychology, 20, 444-446. James, W. (1890). Principles of psychology. New York: Holt. Janis, I. L., & Mann, L. (1977). Decision making: A psychological analysis of conflict, choice and commitment. New York: Free Press. Kaplan, M. F. (1991). The joint effects of cognition and affect on social judgment. In J. P. Forgas (Ed.),Emotion and social judgment. Oxford: Pergamon. Kavanagh, D. J., & Bower, G. H. (1985). Mood and self-efficacy: Impact of joy and sadness on perceived capabilities. Cognitive Therapy and Research, 9, 507-525. Kelly, G. A. (1955). The psychology of personal constructs. New York: Norton. Kruglanski, A. W. (1989). Lay epistemics and human knowledge: Cognitive and motivational bases. New York: Plenum. Kruglanski, A. W., & Freund, T. (1983). The freezing and unfreezing of lay inferences: Effects of impressional primacy, ethnic stereotyping and numerical anchoring. Journal of Experimenral Social Psychology, 19, 448-468. Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 331-350. Laird, J. D., & Bresler, C. (1990). William James and the mechanisms of emotional experience. Personality and Social Psychology Bulletin, 16, 636-65 1.

AFFECT IN SOCIAL JUDGMENTS

273

Laird, J. D., & Bresler, C. (1991). The process of emotional experience: A self-perception theory. In M. Clark (Ed.),Review of personality and social psychology. Beverly Hills, CA: Sage. Lazarus, R. S. (1984). On the primacy of cognition. American Psychologist, 39, 124-129. Levinger, G. (1990, July). Figure versus ground: Micro and macro perspectives on personal relationships. Invited address at the 5th International Conference on Personal Relationships, Oxford. Levinger, G., & Snoek, J. D. (1972). Attraction in relationships: A new look at interpersonal attraction. Momstown, NJ: General Learning Press. Lewicka, M. (Ed.). (1992). Positive-negative asymmetry. Special Issue of the European Journal of Social Psychology. Locke, K. D., & Horowitz, L. M. (1990). Satisfaction in interpersonal interactions as a function of similarity in level of dysphoria. Journal of Personality and Social Psychology, 58, 823831.

Mackie, D., Hamilton, D. L., Schroth, M. E., Carlisle, C. J . , Gersho, B. F., Meneses, L. M., Nedler, B. F., & Reichel, L. D. (1989). The effects of induced mood on expectancy-based illusory correlations. Journal of Experimental Social Psychology, 25, 524-544. Mackie, D., & Worth, L. (1991). Feeling good, but not thinking straight: The impact of positive mood on persuasion. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Mann, L. (1992). Stress, affect and risk taking. In J. F. Yates (Ed.), Risk-taking behavior. Chichester: Wiley. Martin, L. L. (1986). Setheset: Use and disuse of concepts in impression formation. Journal of Personality and Social Psychology, 51, 493-504. Martin, L. L., Harlow, S., & Clore, G. (1991). The use and misuse ofmood in recall. Manuscript, University of Georgia, Athens. Mayer, J. D. (1986). How mood influences cognition. In N. E. Sharkey (Ed.), Advances in cognitive science (Vol. 1). Chichester: Ellis Honvood. Mayer, J. D., Gaschke, Y. N., Braverman, D. L., & Evans, T. W. (1991). Mood congruent judgment and natural mood. Manuscript, University of New Hampshire, Durham. Mayer, J. D., & Salovey, P. (1988). Personality moderates the interaction of mood and cognition. In K. Fiedler & J. P. Forgas (Eds.), Afect, cognition and social behaviour. Toronto: Hogrefe International. Murray, H. A. (1933). The effects of fear upon estimates of the maliciousness of other personalities. Journal of Social Psychology. 4, 310-329. Osgood, C. (1969). On the whys and wherefores of E, P, and A. Journal of Personality and Social Psychology. 12, 194- 199. Ottaviani, R., & Beck, A. T. (1988). Cognitive theory of depression. In K. Fiedler & J. Forgas (Eds.), Afect. cognition and social behavior. Toronto: Hogrefe International. Parron, W. G., & Sabini, J. (1990). Mood and memory under natural conditions: Evidence for mood incongruent recall. Journal of Personality and Social Psychology, 59, 321-336. Peeters, G., & Czapinski, J. (1990). Positive-negative asymmetry in evaluations. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology. Chichester: Wiley. Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York: Springer-Verlag. Petty, R. E., Gleicher, F., &Baker, S. (1991). Multiple roles for affect in persuasion. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Razran, G. H. S. (1940). Conditioned response changes in rating and appraising sociopolitical slogans. Psychological Bulletin, 37, 481. Reis, H. T., & Wheeler, L. (1991). Studying social interaction with the Rochester Interaction record. In M. Zanna (Ed.), Advances in experimental socialpsychology (Vol. 24). San Diego: Academic Press.

274

JOSEPH PAUL FORGAS

Rojahn, K.,& Pettigrew, T. F. (1990, June). Memory for schema-consistent and schema-inconsistent information: A meta-analysis. Paper presented at the general meeting of the European Association of Experimental Social Psychology, Budapest. Rosch, E., & Lloyd, B. B. (Eds.). (1978). Cognition and categorization. Hillsdale, NJ: Erlbaum. Rosenberg, S., & Sedlak. A. (1972). StnICtural representations of implicit personality theory. In L. Berkowitz (Ed.), Advances in experimental social psychotogy (Vol. 6). New York: Academic Press. Roth, D.. & Rehm, L. P. (1980). Relationships among self-monitoring processes, memory, and depression. Cognitive Therapy and Research, 4, 149- 157. Salovey, P. (1986). Cognitions about the self Connecting feeling states and social behavior. In P. Shaver (Ed.), Self, situations and social behavior (Vol. 6). Beverly Hills, CA: Sage. Salovey, P., & Mayer. J. D. (1990). Emotional intelligence. Imagination. Cognition and Personality. 9, 185-211. Salovey, P., O'Leary,A., Stretton, M., Fishkin, S., & Drake, C. A. (1991). Influence of mood on judgments about health and illness. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Schachter, S. (1959). The psychology of afjiliarion. Stanford, CA: Stanford University Press. Schwan, N. (19%). Feelings as information: Informational and motivational functions of affective states. In E. T. Higgins & R. Sorrentino(Eds.), Handbook ofmorivation and cognition: Foundations ofsocial behaviour (Vol. 2). New York Guilford. Schwarz, N., & Bless, H. (1991a). Happy and mindless, but sad and smart? The impact of affective states on analytic reasoning. In J. P. Forgas (Ed.), Emotion and social judgments. Oxford: Pergamon. Schwarz, N., & Bless, H. (1991b). Mood and persuasion: Affective states influence the processing of persuasive communications. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 24). San Diego: Academic Press. Schwarz, N.. & Clore, G. L. (1988). How do I feel about it? The informative function of affective states. In K. Fiedler & J. P. Forgas (Eds.), Affect, cognition, and social behavior. Toronto: Hogrefe International. Schwarz, N., & Strack, F. (1991). Context effects in attitude surveys: Applying cognitive theory to social research. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2). Chicester: Wiley. Sedikides, C. (1991). Mood effects on self-perception. In M. Clark (Ed.), Review ofpersonalify and social psychology. Beverly Hills, CA: Sage. Stangor, C., & Ford, T. (1992). Accuracy and expectancy-confirmingprocessing and the development of stereotypes. In W. Stroebe & M. Hewstone (Eds.), European review ofsocialpsychology. Chicester: Wiley. Strack, F., Schwarz, N., & Argyle, M. (Eds.). (1991). Subjective well-being. Oxford: Pergamon. Strack, F., Schwarz, N., & Gschneidinger, E. (1985). Happiness and reminiscing: The role of time perspective, affect and mode of thinking. Journal of Personality and Social Psychology, 49, 1460- 1469. Stroessner,S. J., Hamilton, D. L., & Mackie, D. M. (in press). Affect and stereotyping:The effect of induced mood on distinctiveness-based illusory correlations. Journal of Personality and Social Psychology. Swann, W. B., Jr. (in press). Seeking 'truth' and finding despair: Some unhappy consequences of negative self-concept. Current Directions in Psychological Sciences. Tajfel, S . , & Forgas, J. P. (1982). Social categorisation: Cognitions, values and groups. In J. P. Forgas (Ed.), Social cognition. London & New York Academic Press. Taylor, S. (1991). Asymmetrical effects of positive and negative events: The mobilization-minirnization hypothesis. Psychological Bulletin, 110, 67-85.

AFFECT IN SOCIAL JUDGMENTS

275

Tesser, A. (1986). Some effects of self-evaluation maintenance on cognition and action. In R. M. Sorrentino & E. T. Higgins (Eds.), The hundbook of motivation and cognition: Foundations of social behaviour (pp. 435-464). New York: Guilford. Tesser, A,, & Shaffer, D. R. (1990). Attitudes and attitude change. Annual Review of Psychology, 41, 479-523.

Weary, G., Marsh, K. L., & Gleicher, F. (1991). Social-cognitive consequences of depression. In G. Weary, F. Gleicher, & K. L. Marsh (Eds.), Control motivation and social cognition. New York: Springer. Wehmer, G., & Izard, C. E. (1962). The effect of self-esteem and induced affect on interpersonal perception and intellective functioning. Nashville, TN: Vanderbilt University. Weiner, B. (1980). Human motivation. New York: Holt, Rinehart, & Winston. Wessman, A. E., & Ricks, D. F. (1966). Mood and personality. Experimental Aging Research, 10, 197-200.

Wheeler, L., & Miyake, L. (1991). Mood efects on social comparison processes. New York: University of Rochester. Wicklund, R., & Frey, D. (1981). Cognitive consistency: motivational vs. non-motivational perspectives. In J. P. Forgas (Ed.), Social cognition: Perspectives on everyday understanding (pp. 141160). London: Academic Press. Wyer, R. S., & Srull, T. K. (1981). Category accessibility: Some theoretical and empirical issues concerning the processing of social stimulus information. In E. T. Higgins, C . P. Herman, & M. P. Zanna (Eds.), Social cognition: The Ontario Symposium (Vol. 1). Hillsdale, NJ: Erlbaum. Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in irs social context. Hillsdale, NJ: Erlbaum. Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35, 151-175. Zanna, M. P., & Cooper, J. (1976). Dissonance and the attribution process. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New directions in attribution research (Vol. 1). Hillsdale, NJ: Erlbaum. Zanna, M. P., Kiesler, C. A,, & Pilkonis, P. A. (1970). Positive and negative attitudinal affect established by classical conditioning. Journal of Personality and Social Psychology, 13, 224238.

Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar-Tal & A. Kruglanski (Eds.), The social psychology of knowledge. New York: Cambridge University Press.

This Page Intentionally Left Blank

THE SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

Thomas Blass

I. Introduction The untimely death of Stanley Milgram on December 20, 1984, at the age of 51, marked the end of one of the most productive, eclectic, and innovative careers in social psychology. Although he was best known both among psychologists and the public at large for his landmark studies on obedience to authority (Milgram, 1963, 1974, 1984d), Milgram's span of interest took in a variety of other, less controversial, phenomena. Thus, he left his mark in the form of innovative research on topics as wide-ranging as urban psychology (Milgram, 1970a, 1970b),' the effects of televised antisocial behavior (Milgram & Shotland, 19731, subway norms (Milgram & Sabini, 1978), the lost-letter technique (Milgram, 1969c; Milgram, Mann, & Harter, 1965), the small-world method (Korte & Milgram, 1970; Milgram, 1967b; Travers & Milgram, 1969), crosscultural differences in conformity (Milgram, 1960, 1961), crowds and collective behavior (Milgram, Bickman, & Berkowitz, 1969; Milgram, Liberty, Toledo, & Wackenhut, 1986), mental maps (Milgram, Greenwald, Kessler, McKenna, & Waters, 1972; Milgram & Jodelet, 1976, 1977), the psychology of photography (Milgram, 1976b), the familiar stranger (Milgram, 1977b), and person perception (Milgram, 1984b; see also Colligan, 1980; Herbert, 1985). It is possible to distill from Milgram's work a set of underlying unifying features, themes, and metatheoretical assumptions that, in their totality, add up to a distinctive approach to social psychology. My goal in this article is to present 'This work (Milgram, 1970a), as well as most of the other articles published by Milgram up to the mid-1970s, was reprinted in an anthology with commentary (Milgram, 1977e). In most instances, when I cite such an article, I give the source as the reference, rather than the anthology, except when the original publication is relatively inaccessible. Milgram (1970b) is a somewhat longer version of Milgram (1970a). ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLDGY, VOL. 25

277

Copyright Q 1992 by Academic Ress. Inc. All rights of reproduction in any form reserved.

27 8

THOMAS BLASS

the underlying characteristics of Milgram’s approach to the study of social behavior, and then to explore, in detail, the impact of his major research programs, thereby facilitating a fuller understanding of his contributions to social psychology as well as to some other disciplines.

11. Underlying Characteristics of His Approach A. A PHENOMENOLOGICAL APPROACH An understanding of Milgram’s work needs to begin with the recognition that Milgram took a phenomenological approach to the study of social behavior. That is, he treated subjective experience as valid data on which to build more systematic analyses of the phenomenon in question. It was this willingness to treat personal experience of, and intuitions about, everyday events seriously that led him into behavioral territories unexplored by others. Thus, for example, in speculating about what might happen if New York subway riders were asked to give up their seats, Milgram and Sabini (1978, p. 33) state “Common sense suggests that it is impossible to obtain a seat on the subway simply by asking for it.” Even the obedience studies were ultimately gmunded in everyday observation. Milgram (1963) introduced his first experimental report of his obedience research by linking it to the Holocaust and to earlier writings on authority, obedience, social power, and suggestion. But he concludes his introduction by stating that his work “derives, in the first instance, from direct observation of a social fact; the individual who is commanded by a legitimate authority ordinarily obeys. Obedience comes easily and often. It is a ubiquitous and indispensable feature of social life” (p. 372). We can also see the central role that observation played in Milgram’s approach in his article, “The Experience of Living in Cities” (Milgram, 1970a). In it he asks what contribution psychology could make toward understanding city life and how one should proceed empirically. And he answers that “observation is the indispensable starting point. Any observer in the streets of midtown Manhattan will see (i) large numbers of people, (ii) a high population density, and (iii) heterogeneity of population. These three factors need to be at the root of any sociopsychological theory of city life, for they condition all aspects of our experience in the metropolis” (Milgram, 1970a, p. 1461). By allotting a primary role to everyday observation in generating research, Milgram pursued a largely atheoretical approach in most of his studies. This approach was already evident in his doctoral dissertation, in which he criticized social psychology for considering hypothesis testing to be the primary function of experimentation (Milgram, 1960, p. 194). As Tavris (1974b) later put it,

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

279

“Most psychologists test hypotheses; Milgram asks questions” ( p. 75). When he did invoke theory, he did so usually after the research was conducted. Thus, rather than the more typical hypothesis-generating function that theory plays in social psychology, for Milgram theory came into play in an attempt either to integrate or to explain findings he already obtained. An example of the former is his introduction of the theoretical concept of “overload” in relation to his studies of city versus small-town behavior (Milgram, 1970a) in order to bring “a grab bag of experimental studies . . . under a more rigorous theoretical discipline” (Milgram, 1981b, p. 14). A prime example of the latter, of the introduction of a theoretical concept to serve in an explanatory capacity, is the “agentic state” construct that Milgram introduced after the completion of the obedience experiments in an attempt to explain the behavior of his obedient subjects. According to Milgram, obedience is made possible through the subject’s entry into a different experiential state-a state of agency-in which a person no longer views himself as acting autonomously but rather as an agent of an authority. A main feature of the agentic state is a shift in responsibility-the subject sheds responsibility for his actions and hands it over to the person in charge. Actions, no matter how destructive they may be, are no longer guided by the usual moral considerations but rather by how well they carry out the authority’s commands (see Milgram, 1974, chaps. 10-12). Milgram first used the agentic-state explanation 7 years after the completion of his obedience studies. It made a brief first appearance in a magazine article by Meyer (1970), in which Milgram is quoted as saying “Just as water can turn to ice under certain conditions of temperature, a person can move to the state of mind that I call agency . . . the critical thing is that you see yourself as the instrument of the execution of another person’s wishes” ( p. 132). Even the idea of relinquishing responsibility to an authority as an important concomitant of obedience did not take on a central role in Milgram’s thinking until 1967 (Milgram, 1967a), although both it and the notion of becoming an agent were already anticipated, though not developed, earlier (see Milgram, 1964b, p. 851; 1965c, pp. 72-73). There was, however, no mention of responsibility or of agency in Milgram’s (1963) first report on the obedience research. There is also another sense in which Milgram’s approach can be characterized as phenomenological. In studying some facet of social behavior, Milgram would usually present a phenomenological analysis of it that would yield a description of it in terms of its essential, defining properties. Sometimes, in fact, his contribution to our understanding of a phenomenon was purely analytic without involving any empirical research whatsoever. A case in point is an article Milgram wrote exploring the significance of the news (Milgram, 1977f). In it he defines news as “information about events that are going on outside immediate experience” (p. 167). It is a means for extending our eyes and ears to encompass, and prepare us to deal with, a broader range of Occurrences. News can be seen as the consciousness of a society, and totalitarian governments repress the news for the

280

THOMAS BLASS

same reason troubled personalities keep certain thoughts out of awarenessbecause of deep-seated insecurities. Milgram notes two further characteristics of the news. First, today news serves more as a source of entertainment than as a source of usable information, and it enables us to indulge our taste for thrills and excitement under the cloak of respectability and enlightenment. Second, news has become increasingly ritualized: the time-slot allotted for the evening news broadcast must be filled even if there is nothing substantive to be communicated. In a similar vein, Milgram (19778) advances our understanding of another phenomenon-fanaticism-through a conceptual analysis of its defining properties, rather than by empirical means. “Fanaticism,” he points out, “is applied to the state of mind of those who are wholeheartedly committed to a set of beliefs and are condemned for it” (p. 58). The fanatic is condemned because his enthusiasm is seen as excessive, a mark of intellectual rigidity and lack of critical judgment, and as a barrier to reasonable persuasion. The maintenance of the social order is predicated on the possibility of resolving conflicts through mutual accommodation and influence. The fanatic is seen as a threat by virtue of his intensity of beliefs, which makes him impervious to influence by the larger society. Milgram notes that fanaticism differs from other forms of deviancy such as criminality. Whereas both society and the criminal recognize that the latter is breaking the law, the fanatic sees the problem in the surrounding society, not in his own behavior and beliefs. From his perspective it is they who “have not grasped the truth and thus are behaving strangely, and even destructively” (p. 59). There is a natural history to every episode of fanaticism, beginning frequently with a drastic change in belief and behavior when the person first “sees the light.” This often evokes criticism and attempts to convince the fanatic to return to his previous lifestyle. As a result the fanatic may harden in his belief, his fanaticism may peter out for lack of social support, or a group of like-minded believers may form a community for mutual reinforcement and to attain legitimacy. The larger society might respond to a fanatical group in a number of different ways: It could eventually capitulate to its beliefs, or the group may be isolated in a ghetto, made to emigrate, or be destroyed. Conceptual analysis also figured importantly as an accompaniment to Milgram’s research. Thus, for example, although undoubtedly the quantitative outcome-an obedience rate of 65%-was the most compelling feature of Milgram’s (1963) first report on his obedience research, an important adjunct was his analysis of the features of the experiment likely to be promotive of obedience, because it helped explain what at first seemed incomprehensible. Among others, he mentions the reputation of the sponsoring “background authority,” Yale University; the fact that the experiment has a laudable purpose, the study of learning and memory; and the sense of obligation on the part of the subject to the experimenter, strengthened by the payment involved, to help the latter complete the experiment.

SOCIAL PSYCHOUXiY OF STANLEY MILGRAM

28 1

B. AN EXPERIMENTAL STRATEGY: OUTCOME MEASURES THAT ARE OBSERVABLE, DISCRETE, AND DICHOTOMOUS Despite the usefulness of his conceptual analyses, Milgram was, first and foremost, an experimentalist who was able to explore a wide variety of social psychological phenomena ranging from conformity (Milgram, 1960, 1961, I964a) and obedience (Milgram, 1 9 6 5 ~to ) the lost-letter technique (Milgram, 1969c) and the drawing power of crowds (Milgram et al., 1969) within the structural confines of the randomized experiment. For Milgram the use of the experimental method was clearly an essential feature of social psychology, as the following quote reveals: “The creative claim of social psychology lies in its capacity to reconstruct varied types of social experience in an experimental format, to clarify and make visible the operation of obscure social forces so that they may be explored in terms of the language of cause and effect” (Milgram, 1977e, p. 1). At the same time, for Milgram the choice of method was dictated by the requirements of the behavior being studied, not vice versa, and when the problem called for it he readily used less powerful research methods. When they were informative enough, he even used simple demonstrations; for example, when wanting to determine how strangers would respond to asking to have their picture taken, his students simply went up to people and asked them “May I take your picture?” (Milgram, 1976b, p. 10). If there is a unifying characteristic to Milgram’s data-gathering techniques, it is not in the types of research designs or techniques he used, but rather in the nature of the target behaviors he studied. In most of his studies, the outcome measure was discrete and dichotomous: It was essentially a yes or no answer to the question of whether a given behavior occurred. Did New York City subway riders give up their seats when requested to (Milgram & Sabini, 1978)? Were lost letters differing in the admirability of their targets mailed or not (Milgram et al., 1965)? Would strangers, approached either in midtown Manhattan or in a small town, reciprocate an extended hand with a handshake (Milgram, 1977e, pp. 17lS)? Would viewers of a prime-time TV program, with or without an antisocial act embedded in it, break into and steal money from a charity display (Milgram & Shotland, 1973)? Would a subject commanded to inflict increasingly harsher punishment on a learner obey fully or disobey and stop at some point short of the highest possible voltage (Milgram, 1974)? Even Milgram’s other measure of obedience-the specific break-off point on the shock continuum-can be seen as the accumulation of the outcomes of a series of discrete, dichotomous response choices at each voltage level, i.e., whether to press the shock lever or not. This undoubtedly was a major factor accounting for the compellingness of much of Milgram’s research. In contrast to the relativism and ambiguity inherent in many continuous measures (e.g., a point on a numerical scale), the discrete, observable

282

THOMAS BLASS

acts comprising most of Milgram’s findings lent them a quality of absoluteness, clarity, and finality that made their implications directly discernible to both lay and professional readers.

C. THE SALIENCE OF MORAL ISSUES Another factor that has contributed to the compellingness of Milgram’s work is that he often emphasized the moral aspects of the behavior under discussion. Thus, for example, he might see an observed action as representing the resolution of a conflict between two behavioral alternatives, one of which the person ought to have chosen according to moral, or at least normative, standards. Clearly, of all of Milgram’s work, the most heavily imbued with moral significance was his obedience research. In the obedience studies Milgram confronted two moral issues, the first by design and the second largely (though not solely) in response to criticisms of the ethics of the experiments. The first moral issue was embedded in the conflict situation Milgram’s subjects found themselves in-a conflict between obeying legitimate authority and following the dictates of one’s conscience (Milgram, 1963, 1974, 1984d). The moral dimension has been a continually salient feature of discussions of the subject’s behavior in the obedience experiments, starting with Milgram himself, who described his obedient subjects as acting “in what appeared to be a shockingly immoral way” (Milgram, 1964b, p. 849). In his earliest experimental report, Milgram (1963) already pointed to the moral issue addressed by his work: “Subjects have learned from childhood that it is a fundamental breach of moral conduct to hurt another person against his will. Yet, 26 subjects abandon this tenet in following the instructions of an authority who has no special powers to enforce his commands” (p. 376). Milgram (1965a, p. 130) asked 20 college students to render a moral judgment about the experiment. After describing the concrete details of the experiment, Milgram asked “How should you perform in the experimental situation? Should you go through to the end of the experiment or should you break off at some point?” Not one subject suggested complete obedience (450 V) as appropriate, the highest stopping point mentioned being 255 V, with the median being 150V. Moral judgment also plays a salient role in the analyses of the obedience experiment by Milgram’s students, Sabini and Silver (1982, 1985), who see the subjects’ obedience as stemming from the fact that to disobey is to reproach the authority for morally reprehensible behavior. As they put it, “Any justification they might have offered for refusing to continue would have involved an explicit or implicit condemnation of the authority” (Sabini & Silver, 1982, p. 38). The second moral issue raised by the obedience experiments had to do with

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

283

Milgram’s treatment of his subjects. Beginning with Baumrind (1964), a number of writers have debated the ethics of placing subjects in the extremely stressful situation Milgram created and of deceiving them about the true purpose of the experiment (e.g., Abse, 1973; Crawford, 1972; Elms, 1982; Kelman, 1967; Milgram, 1964b, 1973b, 1974, 1977h; Miller, 1986). Milgram told an interviewer (R. I. Evans, 1976, p. 352) that he was “totally astonished” by the criticisms evoked by his obedience experiments. In an effort to dispel the claim of some critics that he was insensitive to ethical matters, in his collection of his writings Milgram (1977e) included an excerpt from his doctoral dissertation (Milgram, 1960) comparing the conformity levels of Norwegians and Frenchmen. The excerpt describes a questionnaire given to Norwegian subjects to assess their reactions to the experiment. Included in it was a question about whether the subject felt that the experiment was ethical or not. (Most thought it was neither ethical nor unethical.) Another question, similar to one he used later in the obedience experiments (Milgram, 1964b), asked subjects to indicate on a scale ranging from very glad to very sorry how they felt about being in the experiment. (The majority was either glad or very glad to have been in the conformity experiment, a finding similar to that obtained in the obedience experiments [Milgram, 1964bl.) A moral perspective was not restricted to the obedience research-it also pervaded many of Milgram’s other writings. Thus, for example, he (Milgram, 1970a, 1970b) saw withdrawal from moral and social involvement with others as one of the consequences of the stimulus overload, characteristic of urban life. A moral perspective is also evident in a “think” piece that Milgram wrote about the Kitty Genovese incident (Milgram & Hollander, 1964). In it, he points out that, in condemning the 38 nonintervening witnesses to the attack, we should not lose sight of the fact that they did not commit the murder but only failed to stop it. “It is no more than clear thinking to bear in mind the moral difference” (Milgram & Hollander, 1964, p. 602). We also find here a theme that is to recur in many places (e.g., Milgram, 1967a, 1974, 1977e, pp. 20, 92), i.e., that people are not always able to translate their moral principles into action. For example, here is how Milgram (1982) describes the significance,of LatanC and Darley’s research on bystander intervention: “The investigators showed that it is not just a general ‘morality’ that determines whether a person will help someone in trouble, but that the exact details of the situation significantly shape their responses” (p. 50). When Milgram proposed a field experiment on television violence to the CBS network, he saw the main character in the planned stimulus program as being “confronted with a moral dilemma” (Milgram & Shotland, 1973, p. 77). Milgram even brought a moral viewpoint to phenomena one would not normally conceptualize in those terms. Thus, he saw the Asch group-pressure experiments as representing a “dilemma of truth versus conformity,’’ one that the person had

284

THOMAS BLASS

to resolve in a manner that was either “consistent with or in opposition to moral values” (Milgram, 1977e, p. 92). Even the innocuous act of mailing a lost letter (Milgram, 1969c) seemed to have moral overtones for Milgram.

D. THE PRIMACY OF SITUATIONAL DETERMINANTS Earlier I noted that Milgram was an experimentalist who was able to apply the experimental method to a wide variety of phenomena. Milgram was also an experimentalist in the sense that situational factors had primacy in his thinking as potential determinants of behavior, and the understanding of a phenomenon was to be achieved through the experimental variation of relevant conditions in order to identify features of the immediate situation.that either promote or inhibit the behavior under focus. Thus, for example, in his last published statement about the obedience research (Milgram, 1984d, p. 446; 1987, p. 774), he stated that “the crux of Milgram’s inquiry is a set of experimental variations which examine the variables which increase or diminish obedience.” More generally, he noted in another context that “social psychologists make use of experimental variation, in which one systematically alters the situation to find the causes of behavior” (Milgram & Sabini, 1979, p. 74). Milgram was a situationist, par excellence, and his obedience research has historically provided strong ammunition for the situational emphasis of mainstream social psychology (see Blass, 1991d). The importance he placed on situational manipulations was already discernible in his doctoral dissertation (Milgram, 1960, 1961). In it, he set up an analog of the Asch group-pressure paradigm in Norway and France to study cross-national differences in conformity. But he used judgments of auditory stimuli, rather than the visual linejudgment task of the original Asch studies. Specifically, on each trial subjects had to indicate which of a pair of tones was longer. A unanimous majority of five was simulated through tape-recorded answers that were incorrect on designated trials. Milgram found the Norwegians more conforming than the French: The former yielded to the incorrect judgments of the bogus majority on 62% of the trials, whereas the latter conformed only 50% of the time. In addition to establishing this basic finding, Milgram conducted a series of experimental variations. He found that the Norwegians always conformed more than the French, but also that changes in the experimental conditions affected both nationalities in similar ways. Thus, for example, both groups became less conforming when accurate responding was made more important by telling subjects that the experimental outcome would be applied to the design of safety signals on airplanes. They both became more conforming, however, in an experi-

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

285

mental condition in which vocal criticism from the other “subjects” would be heard whenever a subject gave a nonconforming response. Despite the importance Milgram placed on situational factors in understanding social behavior, he did not negate personality and individual differences, as was the case with some situationist viewpoints early in the person-situation debate (see Blass, 1977, 1984). One can readily find examples of individual dispositions receiving recognition in Milgram’s writings. An early example can be found in an experiment in which he studied the effects of group pressure (rather than an authority’s commands) on subjects’ willingness to inflict increasingly severe shocks on a “learner” (Milgram, 1964a). The main finding was that the maximum shock elicited by two confederates playing the roles of fellow “teachers” was significantly higher than the one given by subjects in a control condition in which they were free to set the shock level on each trial. Thus, whereas only 1 subject out of 40 (2.5%) ever gave the highest (450 V) shock in the control condition, 7 out of 40 (17.5%) yielded to peer pressure to give the 450-Vshock. But an almost equally salient feature of the findings were the large individual differences Milgram found in response to group pressure, with “some subjects following the group closely, others resisting effectively” (p. 140).2A more recent example of attention to personal dispositions by Milgram comes from his writings on the psychology of photography (Milgram, 1976b, 1977c), in which he states that one of the greatest challenges for photography research is the delineation of the psychological makeup of the professional photographer. More generally, Milgram saw a dispositional approach complementing his own orientation to the study of social behavior: The implicit model for [my] experimental work is that of the person influenced by social forces while often believing in his or her own independence of them. It is thus a social psychology of the reactive individual, the recipient of forces and pressures emanating from outside oneself. This represents, of course, only one side of the coin of social life, for we as individuals also initiate action out of internal needs and actively construct the social world we inhabit. But I have left to other investigators the task of examining the complementary side of our social natures. (Milgram, 1977e, p. 1)

E. THE INFLUENCE OF SOLOMON ASCH Milgram did his graduate work in social psychology at Harvard, and the person there who had the greatest intellectual influence on him was Solomon Asch, who was there as a visiting faculty member. Milgram served as Asch’s 2Judith Waters, a former student and research assistant of Milgram, notes that Milgram very briefly considered giving personality measures to subjects in the obedience experiments, but decided against it because doing so might have contaminated the expenmental results (personal communication, August 13, 1990).

286

THOMAS BLASS

teaching assistant and later also worked for him at the Institute for Advanced Study in Princeton in 1959-1960 (see R. I. Evans, 1976, p. 347; Tavris, 1974a, p. 77). Milgram (in Tavris, 1974a, p. 77) described Asch as “a brilliant, creative man, who possessed great philosophic depth. He is certainly the most impressive social psychologist I have known.” Milgram’s admiration of Asch also extended to his conformity experiment. He considered it “the decisive paradigm” for studying group effects and “as a kind of permanent intellectual jewel” (Milgram, 1977e, p. 152; see also Milgram, 1978). It is thus no surprise that in his doctoral dissertation-the cross-cultural study of conformity-Milgram ( 1960, 1961) used an adaptation of the Asch paradigm. More importantly, Milgram credits the Asch conformity experiment as one of the two factors that gave rise to his obedience paradigm. One factor was his general concern about authority, one that was “forced upon members of my generation, in particular upon Jews such as myself, by the atrocities of World War 11. . . . The impact of the holocaust on my own psyche energized my interest in obedience and shaped the particular form in which it was examined” (Milgram, 1977e, pp. 92-93). The other factor was his attempt to adapt the Asch conformity experiment to study something more consequential than judging lengths of lines. Here is how he put it to interviewer Tavris (1974a, p. 80) (see also Milgram, in R. I. Evans, 1976, pp. 347-348): I wondered whether groups could pressure a person into performing an act whose human import was more readily apparent, perhaps behaving aggressively toward another person, say by administering increasingly severe shocks to him. But to study the group effect you would also need an experimental control; you’d have to know how the subject performed without any group pressure. At that instant, my thought shifted, zeroing in on this experimental control. Just how far would a person go under the experimenter’s orders? It was an incandescent moment, the fusion of a general idea on obedience with a specific technical procedure.

In addition to these clear-cut or explicitly stated connections to Asch, there are certain similarities between the approaches of Asch and Milgram that suggest additional instances of influence. Perhaps the most noteworthy is a shared philosophical predilection for seeing the bottle of human possibilities as half full rather than half empty. Much like Asch (e.g., Asch, 1952, p. 484), Milgram showed a preference for viewing social behavior as primarily rational, a product of the sensible weighing of available alternatives rather than of human limitation and irrationality (see, for example, Milgram & Toch, 1969, pp. 568-569).

F. INFLUENCE OF THE THEATER The dramaturgical perspective of Goffman, suggesting that social behavior can be understood as performances staged to create a desired effect on others, has

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

287

influenced a number of self-presentational perspectives in contemporary social psychology. Milgram also shows this influence when he draws on Goffman to help identify one of the “binding factors” or psychological inhibiting mechanisms that he believed prevented the subject in the obedience experiment from defying the experimenter. According to Goffman (1959), “when an individual projects a definition of the situation and thereby makes an implicit or explicit claim to be a person of a particular kind, he automatically . . . oblige[s others] to value and treat him in the manner that persons of his kind have a right to expect” (p. 13). Milgram (1974) points out that it is impossible for the subject to disobey the experimenter without at the same time violating the latter’s selfdefinition as a competent authority. “Thus, the subject fears that if he breaks off, he will appear arrogant, untoward, and rude. Such emotions, although they appear small in scope alongside the violence being done to the learner, nonetheless help bind the subject into obedience” (Milgram, 1974, p. 150). But it was not in its role as a metaphor for social life that the theater found its most pervasive influence in Milgram’s work. Rather, it was in his experimentsboth in his exquisite attention to technical details and staging as well as in their intended effects on their audience (i.e., readers of his reports). In the obedience experiments, he used professional industrial engravers on his “shock generator” to ensure its perceived authenticity, and his experimenter was deliberately clothed in a gray lab coat for its ambiguous connotations about the nature of the wearer’s authority, rather than white, which would signify that he was a medical technician (Meyer, 1970). Much like a director of a play, he chose his actors carefully and went through many rehearsals with them (Tavris, 1974b). A statement he once made (Milgram, 1976c, p. 24) is an apt summary of his (and many other social psychologists’) approach to experimentation: “Although experiments in chemistry and physics often involve shiny equipment, flasks, and electronic gear, an experiment in social psychology smacks much more of dramaturgy or theater.” For Milgram, the theater also provided a perspective from which to view a goal of experimentation and the means one sometimes has to use to achieve it. This can be seen most clearly in an exchange that Milgram had with Abse, a British playwright. Abse (1973) had written a play, The Dogs of Pavlov, which was inspired by the obedience experiments. In an introductory essay, Abse vigorously condemned Milgram’s use of deception and the induction of stress, which, he believed, victimized and degraded the subject. He argued that many people “may feel that in order to demonstrate that subjects may behave like so many Eichmanns the experimenter had to act the part, to some extent, of a Himmler” (Abse, 1973, p. 29). In his rebuttal (Milgram, 1973b), which appears together with the script and Abse’s introduction in one volume, Milgram expresses surprise at Abse’s harsh criticisms of his use of what Milgram prefers to call “technical illusions” (rather than deception) because “as a dramatist you surely understand that illusion may serve a revelatory function, and indeed, the

288

THOMAS BLASS

very possibility of theater is founded on the benign use of contrivance” (p. 39). He goes on to point out that both the playwright and the experimenter use artifice as a means toward beneficial ends-entertainment and intellectual enrichment, in the case of the theater, and the revelation of truths that are hard to get at, in the case of the experiment, i.e., that subjects were far more obedient than one would have expected. Milgram notes a further parallel between the stage and the laboratory: In both, the participants find the use of pretense acceptable. Theater goers do not feel tricked, for example, that an old man turns out to be quite young when the greasepaint is removed. In a similar vein, most subjects in the obedience experiments felt that the experience had been worthwhile, once it was explained to them.

G. STYLE, AS WELL AS SUBSTANCE So far, my discussion of Milgram’s work has focused on substance-the content of his research and writings-but style was also an integral part of his approach. His writing was invariably lucid and readable, no matter who his intended audience was-whether it was fellow psychologists reading a journal report of one of his experimentsor readers of a general-circulationmagazine. His judicious use of metaphor, simile, and the memorable turn of the phrase gave even his scientific journal articles a literary quality that deepened their impact on the reader. For example, in referring to the Holocaust as an antecedent of his obedience research, he said: “It has been reliably established that from 19331945 millions of innocent persons were systematically slaughtered on command. Gas chambers were built, death camps were guarded, daily quotas of corpses were produced with the same efficiency as the manufacture of appliances” (Milgram, 1963, p. 371). Another example, in a much lighter vein, is his description of a person’s contrasting responses to two different kinds of photographs: “Examine a photograph of yourself taken away from home. . . . Count up the associations elicited by each detail of the photograph. In contrast to the garden of meanings that bloom from the snapshot taken at home, you will find this one a desert” (Milgram, 1977a, pp. 63, 65). The literary merits of Milgram’s writings have received recognition in a number of ways: His book, Obedience to Authority: An Experimental View (Milgram, 1974), was nominated for a National Book Award; parts of it were adapted for a literary magazine, Harper’s (Milgram, 1973a); and his writings have been reprinted in anthologies used in writing courses (Chittenden & Kiniry, 1986; Comley, Hamilton, Klaus, Scholes, & Sommers, 1984; Eastman et al., 1988; Stubbs & Barnet, 1983). Although much of his writing was marked by an economy of expression, Milgram’s prose became rich and expansive when occasioned by the subject

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

289

matter. And in fact it has been suggested that his too vivid descriptions of his subjects’ emotional turmoil in the obedience studies helped convince many readers that his subjects had been mistreated (Pattullo, 1984). One typically finished reading a piece of writing by Milgram feeling not only that one had gained some new insights about social behavior but also that the process was a thoroughly enjoyable one. Contributing to the pleasurable experience was the wry sense of humor that peppered his work throughout his career. Thus, in his very first publication, his cross-cultural study of conformity (Milgram, 1961), he noted that it was easier to simulate a majority using taperecorded voices than to use confederates because “tapes do not have to be paid by the hour and they are always available” (p. 47). And, later in his study of the cognitive maps of New Yorkers (Milgram et al., 1972), here is how he comments on the finding that Queens residents were four times as likely to recognize a street in Manhattan than in their own borough: “Taxi drivers are reputed to fear entering Queens lest they never find their way out. And with good reason, when even the people who live in Queens are lost in their home borough compared with the sense of place they experience in Manhattan!” (p. 199). Milgram was at his best when his humor turned to sarcasm and he used it as a deadly weapon in scholarly combat with Orne and Holland (1 968). Criticizing the obedience experiments from a demand characteristics perspective, Orne and Holland argued that Milgram’s subjects saw through the deception but pretended not to because they did not want to ruin the experiment. Milgram (1972) replied “Orne’s suggestion that the subjects only feigned sweating, trembling, and stuttering to please the experimenter is pathetically detached from reality, equivalent to the statement that hemophiliacs bleed to keep their physicians busy” (p. 140). A biting wit enlivened his writings until the very end. In one of his last publications, Milgram (Milgram & Martin, 1985) commented on a theoretical article by wondering whether it was very interesting or simply “the commonplace raised to a high order of technical expression.” And, “it is not always easy to distinguish a grand theory from one that is only grandiose” (pp. 139, 141). Milgram was an effective communicator, ably “giving psychology away” through his writings and interviews. As he did not shy away from appearing in mass-circulation magazines, he was able to reach a wide a ~ d i e n c eFor . ~ example, an excellent concise article about his obedience research appeared in TV Guide (Milgram, 1976c) to coincide with the airing of The Tenrh Level, a television drama based on the obedience experiments. Milgram had an apprecia3He was also an effective communicator of social psychology via the medium of film. He produced or coproduced a total of six educational films. In addition to Obedience, they are Inviration to Social Psychology, Human Aggression, Nonverbal Communication, Conformity and Independence, and the award-winning The City and the Sey. The films merit a detailed analysis, which is beyond the scope of this article.

290

THOMAS BLASS

tion for the journalists’ craft, and he edited a book consisting of newspaper and magazine articles on psychological topics meant for the introductory psychology course (Milgram, 1975).

111. Shortcomings and Flaws

So far, my overview of Milgram’s work in terms of its underlying themes and characteristics has been generally favorable. But much like an aerial photograph that reveals both topographical continuities and disturbances not visible through a close-up view, an overview of the whole corpus of Milgram’s work discloses not only its positive unifying qualities but also its shortcomings and flaws. One way to characterize a general deficiency in Milgram’s work that a broad view reveals is that it is marked by a number of contradictions and inconsistencies, among which are the following salient discrepancies: 1. For someone so energetic in generating new ideas, Milgram’s reuse of some of his material in different publications comes as a surprise. For example, we find much of the same content, with either the same or paraphrased wording, in four articles on the psychology of photography appearing in four different places (Milgram, 1976a, 1976b, 1977c, 1977d). As another example, most of the first chapter of the obedience book (Milgram, 1974) appeared earlier in Milgram (1967a) and the opening paragraphs were from Milgram (1963). 2. Given the generally lucid quality of Milgram’s writing and the meticulous attention to detail in the design of his experiments, it is surprisingly difficult to gain a full, coherent understanding of the details of his obedience experiments due, to a large extent, to the unusual, piecemeal way they were published.

Milgram began the obedience research in 1960 and completed it in 1963. Initially, he published a series of reports on the work within a 3-year period (Milgram, 1963, 1964a, 1965a, 196%) containing descriptions of varying detail of 12 experimental conditions (as well as brief references to several more). During this time, two additional articles appeared, one describing pilot procedures and findings (Milgram, 1964c) and the other (Milgram, 1964b) presenting a rebuttal of Baumrind’s (1964) criticisms, which also contained new information on the well-being of his subjects, in the form of postexperimental questionnaire data and an excerpt from a brief psychiatric report (Errera, 1972). His documentary film, Obedience, also made its appearance at this time (Milgram, 1965b). In 1966, Elms and Milgram reported the results of a comparison of a sample of obedient and defiant subjects on a number of personality and other self-report measures, which Elms had administered several months after the subjects’ participation in the experiments. Milgram (1967a) published

SOCIAL P S Y C H O W Y OF STANLEY MILGRAM

29 1

an article describing his obedience research in a small-circulation British magazine. Although no new data were presented, it was here that Milgram first conveyed the importance of relinquished responsibility in the obedience process. In 1968, the most comprehensive report up to that time (Milgram, 1965c) was reprinted in the International Journal of Psychiatry together with four evaluative articles, two of which were favorable (M. Erickson, 1968; Etzioni, 1968) and two which were critical of the obedience work (Masserman, 1968; Orne & Holland, 1968). The same issue also contained a brief reply to his critics by Milgram (1968), most of which was directed at Orne and Holland (1968), the longer of the two critical articles. The latter had argued that Milgram’s subjects probably saw through the deception. Later, Milgram (1972) wrote a much longer rebuttal of Orne and Holland, which included newly reported postexperimental data in support of Milgram’s claim that most subjects believed the shocks were genuine and very painful. A full description of the research program did not appear until 1974 (Milgram, 1974). This book presented 19 variations in similar detail (i.e., in addition to the verbal descriptions, frequency distributions of break-off points for each condition appeared in tabular form). Nine of these conditions were described for the first time. Ten had appeared in earlier reports, but no cross-referencing-which would have enabled the reader to integrate all the findings and publications-was provided. Reading the various reports of the obedience experiments also leaves one with a lack of sense of closure. First, this is due to the brief mention of a couple of conditions without a precise quantitative report of their outcomes, i.e., a condition in which the experimenter is completely absent and tape-recorded instructions are used and another in which the experimenter is absent during the first part of the experiment, giving his instructions by phone, but then reappears when the subject refuses to continue (Milgram, 196%). Second, at the end of one of his reports (Milgram, 1965c, p. 71) Milgram lists a number of variations that were to be reported elsewhere. All do appear in Milgram (1974) except for one, referred to as an experiment that “concerned the personal relationship between victim and subject” (Milgram, 1965c, p. 71). This description does not seem to match any of the variations reported later in Milgram (1974) nor in any of Milgram’s other publications. (See Blass, 1991d, for a review of the obedience research.) Although “a foolish consistency” may be “the hobgoblin of little minds,” as Emerson once said, one cannot help but wonder about the following additional contradictions in Milgram’s work: 3. Why are self-reports considered credible some of the time but not other times? In his reply to Orne and Holland’s (1968) criticisms of the internal validity of the obedience studies, Milgram (1972, Table 2-3) presented the results of a follow-up questionnaire given to subjects a year after their participation, which revealed that only 2.4% of the subjects were sure that the learner was not really

292

THOMAS BLASS

getting shocked. But he was not willing to discard the data from these kinds of subjects because their stated disbelief may well have been an after-the-fact rationalization to justify their obedience. In another context (Milgram, 1964b)-in attempting to justify the ethics of the obedience research-he also draws on questionnaire responses of subjects for supportive data. Milgram found 83.7% of his subjects stating that they were glad or very glad to have participated, 15.1% were neither sorry nor glad, and only 1.3% had misgivings about it. Initially (Milgram, 1964b) he noted that “such findings are to be interpreted cautiously, but they cannot be disregarded” (p. 849). Later, however, caution seemed to have been cast aside, with Milgram (1977h) dismissing the criticism that subjects’ positive reactions may simply have been a product of cognitive dissonance. In fact, in his later writings, subjects’ positive evaluations of their experience were made to bear the main burden for the ethical justification of the obedience studies: “The central moral justification for allowing a procedure of the sort used in my experiment is that it is judged acceptable by those who have taken part in it. Moreover, it was the salience of this fact throughout that constituted the chief moral warrant for the continuation of the experiments” (Milgram, 1973b, pp. 40-41; see also Milgram, 19771.1, pp. 21-22).4 4. Why did a researcher, who for most of his career had, by word and deed, stressed the importance of studying behavior directly (as opposed to what people say they will do), study cognitive maps without also demonstrating their links to actual behavior? Milgram conducted three systematic studies of the mental maps of the residents of two major cities, New York City and Paris. The primary focus of the first of two New York City studies (Milgram et al., 1972) was to determine the recognizability of different parts of the city by its inhabitants. To achieve this goal, Milgram and his students systematically sampled and photographed over 150 scenes throughout the five boroughs of the city. A group of New Yorkers, proportionately drawn from all the five boroughs, were subsequently asked to identify the locations depicted in the pictures. The Paris study (Milgram, 1984a; Milgram & Jodelet, 1976, 1977) used a number of other techniques besides scene recognition-e.g., having subjects draw subjective maps of the city-in order to tap Parisians’ cognitive representations of their city. Milgram also conducted a second study of New Yorkers’ images of their city using hand-drawn maps and responses to a questionnaire published in New York magazine (Duncan, 1977; “Mental Maps,” 1975; Milgram, 1984a). Although the studies were imaginative 4Judith Waters (personal communication, August 13, 1990) adds: “I read all of the responses when I was preparing to illustrate the book [Milgram, 19741. Many of the comments sound very defensive or as if they (the respondents) were trying to impress him (and Yale). I remember one, in particular, who said that participating in the experiment was the most important event of his life. One wonders about the rest of that existence!”

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

293

and many of the results fascinating, the connections to actual behavior, though assumed, were never dem~nstrated.~

IV. Enduring Contributions Although Milgram’s work did have some shortcomings, they were clearly overshadowed by his enduring contributions to social psychology, as well as to some other disciplines, to which we now turn. Specifically, I will assess the impact of each of his major research foci in turn.

A. OBEDIENCE TO AUTHORITY Within social psychology, and perhaps even psychology as a whole, it would be hard to match the durability and breadth of interest evoked by the obedience research. From the very beginning, it caught the attention of both psychologists and the press (e.g., Baumrind, 1964; “Could We Be,” 1964; Kaufmann, 1967; Sullivan, 1963) and, to the present day, continues to stimulate research and analysis (Blass, 1990a, 1990b, 1990c, 1991a, 1991b, 1991d; Blass & Krackow, 1991; Gibson, 1991; Meeus & Raaijmakers, 1987; Miller, 1986; Nissani, 1990; Schurz, 1985) and influence-or, at least, figure prominently in-theorizing about obedience-related phenomena (Billig, 1990; Haritos-Fatouros, 1988; Kelman & Hamilton, 1989; Staub, 1989). A significant portion of Mixon’s (1989) book is devoted to refuting Milgram’s view that his subjects’ actions represented destructive obedience and to presenting an alternate conceptualization, building on earlier writings (e.g., Mixon, 1971, 1976). Not only does the work continue to be cited in most introductory texts (see Gorenflo & McConnell, 1991) and virtually all social psychology texts, but it also appears in some junior and senior high school curricula on the Holocaust (Strom & Parsons, 1978; Teaching About the Holocaust, 1985) and in one aimed 5John Sabini, a former student and research assistant of Milgram, provides a perspective that can help diminish the inconsistency between the cognitive map studies and Milgram’s other work: “It is true that Milgram never demonstrated connections to behavior [in the cognitive map studies]. But as you . . . point out early in the paper, Milgram wasn’t a behaviorist. He was interested in capturing aspects of experience. It is also true, of course, that his favored strategy was to find some dramatic behavior to externalize that experience. But in the case of mental maps I think he thought that in themselves they captured something of the subjects’ experience of the cities. The spirit of his thinking about mental maps seems clearest to me not in his articles about them, but in his writing about photography” (personal communication, August 20, 1990).

294

THOMAS BLASS

more generally at sensitizing students about the dangers of blind obedience (Bushman, 1985). Unmatched, too, perhaps, are the intense opinions, pro and con, that the obedience research has generated. For example, Roger Brown (1974) stated that the “series of experiments . . . probably constitutes the most important social psychological research done in this generation” ( p. 42), and Herrnstein (1974) regarded them as “among the most original in all of social psychology” (p. 82). In a similar vein, Eysenck (1974) noted that “Many people believe that an experimental social psychology is either impossible, sterile, or useless. That this belief is mistaken has been obvious for quite some time. . . . The important contributions of Stanley Milgram during the early sixties present us with another outstanding example of the possibilities inherent in the use of the experimental method in this field” (p. 21). More recently, Ross (1988) articulated the importance of the obedience research as follows: “Perhaps more than any other empirical contributions in the history of social science, they have become part of our society’s shared intellectual legacy-that small body of historical incidents, biblical parables, and classic literature that serious thinkers feel free to draw on when they debate about human nature or contemplate human history” (p. 101). On the other hand, Wrightsman (1974) gave this assessment of the obedience work: “When I consider that the initial study was a demonstration and not even an experiment, that the research program lacked any initial theory or tests of significance, and that many of its conclusions are subject to alternative explanations, I am saddened that it is the obedience study that will go down in history as reflecting the 1960s in social psychological research” ( p. 804). Kohlberg (1974) also took a critical stand toward the obedience experiments, though on different grounds: “Milgram himself was unwittingly the moral victim of the ‘authority of science,’ just as his subjects were. Serving the authority of science under the banner of ‘objectivity,’ he, himself, inflicted pain on others for the greater social welfare. . . . Milgram failed to treat his subjects as persons, the basic meaning of the moral attitude, just as most of his subjects failed to treat the victim as a person” (p. 607). The most severe-and surprisingly strident-criticism of the obedience work that I have encountered came from Bruno Bettelheim (in Askenasy, 1978, p. 13 1): I detest it, as 1 detest Zimbardo’s experiments. These experiments are so vile, the intention with which they were engaged is so vile, that nothing these experiments show has any value. , . . They are in line with the human experiments of the Nazis. . . . Having been one of those experimented with, I can see no redeeming merit in these experiments. Milgram, Zimbardo did them to promote their own professional advancement.

It is a testament to the broader human significance of the obedience work that disciplines outside of psychology, e.g., economics (Akerlof, 1991), education

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

295

(Atlas, 1985), sociology (Damico, 1982), political science (Helm & Morelli, 1979, 1985), and philosophy (Patten, 1977), as well as political and social commentators (e.g., Giffin, 1983; Karnow, 1971; Krauthammer, 1985; Shnayerson, 1973; S. J. Singer, 1991), have found relevance in it. By 1981, the first journal report on the obedience research (Milgram, 1963) had become a “citation classic,” having been cited over 255 times since 1966 (Milgram, 1981a). The continuing interest in the obedience studies has transcended national boundaries, as evidenced by the fact that Obedience to Authority: An Experimental View (Milgram, 1974) has been translated into 1 1 languages (with multiple editions appearing in some cases), i.e., German (three editions), French (two editions), Japanese (two editions), Dutch, Danish, Italian, Spanish (two editions), Swedish, Portuguese (published in Brazil), Indonesian, and Serbo-Croatian (published in Yugoslavia). There is also a British edition of the book. Milgram continued to write and speak about the topic long after he had completed the research (CBS News, 1979; Milgram, 1979, 1983, 1984d, 1987; Muson, 1978). Not surprisingly, in a recent article on “psychological literacy” (Boneau, I990), Milgram’s obedience experiment is listed among psychology’s “top 100” terms and concepts. Nor should it be surprising that one of the exhibits comprising an innovative psychology exhibition, which opened at the Ontario Science Center in Toronto in March, 1991, is about the obedience research (“Exhibition Explores,” 1991). (In it, visitors entering a tiled passageway are instructed by signs to walk on the black squares only. Almost everyone obeys. Afterward, they get an introduction to the obedience experiments, which includes seeing the original “shock” generator used by Milgram.) But it is quite remarkable that professions as diverse as marketing (Armstrong, I975), accounting (Dirsmith, 1983), nursing (Redfearn, 1982), and management (Laurent, 1987) can derive practical lessons from the obedience studies for their respective practitioners. Even more remarkable is a recent account of a social psychologist’s trial testimony in which the lessons derived from the obedience research (as well as some other classic social-psychological studies, such as those on group polarization, bystander intervention, and deindividuation) actually saved lives. In two trials in South Africa involving a total of 13 defendants accused of murder during mob actions, the courts accepted obedience to authority and several other social psychological phenomena as extenuating factors, with the end result that nine of the accused were saved from the death sentence (Colman, 1991). The obedience work dramatically demonstrated that people are much more likely to obey the orders-perhaps even the destructive orders-of a legitimate authority than was thought.6 Milgram’s (1965c, p. 74) suggestion that this involves allowing the authority to define reality for us, “the control of [our] 6Though this remains an enduring lesson, there is some evidence (Kaufmann & Kooman, 1967; Mixon, 1971) suggesting that it is in need of some qualification.

296

THOMAS BLASS

perceptions,” is both thought-provoking and disturbing. Although, as I will show, the exact parallels between the behavior of his subjects and those of the Nazis and their collaborators remain open to debate, there is no question that the obedience experiments have promoted a greater awareness of the Holocaust and efforts to understand its determinants. I hope that such consciousness raising can serve a preventive function. As a Holocaust survivor, I am especially appreciative of the potential value of the obedience experiments in this regard. The ethical controversy surrounding the obedience experiments has left an indelible mark on experimentation with human subjects. It, together with a handful of other controversial studies, led to the formulation of the American Psychological Association’s ethical principles for research with human subjects and to the federally mandated creation of Institutional Review Boards for the protection of human subjects. Although psychologists have disagreed about the necessity of such regulations, there is currently a greater sensitivity to the wellbeing of the research participant than was the case in the past, as evidenced, for example, by the quadrupling of the number of studies reporting debriefings from 1964 to 1980 in the Journal of Abnormal and Social Psychology and Journal of Personality and Social Psychology (Ullman & Jackson, 1982). As Miller (1986) aptly concluded after reviewing the ethical debate surrounding the obedience studies, “To say that we are better off for having consjdered the ethical dimension of the obedience paradigm may sound trite, but it must certainly be the case. Among the beneficiaries, doubtless, are legions of research subjects who have been treated in a more humane and informed manner by researchers who, in the past two decades, have been enlightened by studying the issues reviewed here” (p. 138). The obedience studies are often used to illustrate ethically problematic research in introductory, social, and methods texts, but what is often overlooked is that Milgram (1963) undertook procedures “to assure that the subject would leave the laboratory in a state of well-being. A friendly reconciliation was arranged between the subject and the victim, and an effort was made to reduce any tensions that arose as a result of the experiment” (p. 374). In this regard, it is interesting to note that Harris (1988) credits Milgram (1964b) with the first published use of the term “debriefing” in an experimental context.

I . Impact on Social Psychological Research and Conceptualizations: Variety and Numbers We now turn to a more focused look at the discernible influences of the obedience paradigm on contemporary social psychological research and thinking. Over the years, the obedience studies have been used by social psychologists to crystallize concepts, to clarify phenomena, and to marshal1 support for viewpoints on a variety of issues. Thus, when Freedman (1969) argued against role playing as an alternative to the standard laboratory experiment, he drew on the

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

297

oft-reported gap between predictions and findings in the obedience experiment as major support. (Ironically, it later turned out that the strongest evidence supporting the research viability of role playing came from three role-playing variations of the obedience experiment that obtained levels of obedience comparable to Milgram’s findings [Geller, 1975, 1978; Mixon, 1971; O’Leary,Willis, & Tomich, 1970; see also Albrecht, 19731.) When a number of critics declared a crisis in social psychology during the late 1960s and early 1970s and accused it of triviality and irrelevance, defenders of mainstream social psychology would invariably counter by citing the obedience work as a prime example of substance and social relevance. And when Aronson and Carlsmith (1968) introduced the distinction between mundane and experimental realism, didn’t their choice of the obedience experiment as an illustration of the latter help make what they meant vividly clear? When Gergen (1973) argued for the historical and cultural boundedness of many social psychological phenomena, he used the obedience research as one of his supporting pieces of evidence, suggesting that the findings were dependent on prevailing attitudes about authority. The obedience paradigm also came into play in relation to another of Gergen’s (1973) contentions, i.e., the operation of enlightenment effects. Gergen argued that “sophistication as to psychological principles liberates one from their behavioral implications” ( p. 3 13). To provide a concrete test of whether or not obtaining psychological knowledge can affect one’s behavior, as Gergen claimed, Shelton (1982) conducted an ingenious study in which she first taught her subjects about the basic obedience experiment and its outcome. Then she had them serve as experimenters in a similar “learning” experiment in which they had to supervise another subject who was supposed to teach a verbal-learning task to a learner by means of increasing voltages of shock on each subsequent error. Although both teacher and learner were confederates, the experimenter-subject was told that only the latter was. As the voltages of “shock” and the learner’s expression of pain increased, the teacher, on the verge of crying and complaining of stomach pains, begged the experimenter-subject to stop the experiment. In spite of this, out of 24 subjects who began the experiment, 22 went to the end, commanding the teacher to continue to the maximum shock-450 V. Apparently, the subject could not see that his or her obedience to the investigator was similar to the teacher’s obedience to him or her. The obedience research played an evidential role in Ross’s (1977) introduction of the fundamental attribution error, i.e., the tendency for perceivers to overemphasize dispositions and underemphasize situational determinants in explaining another person’s behavior. Ross argued that the obedience research and Asch’s conformity studies were the successes they were among psychologists “because they demonstrated control by a situational factor that the reader had previously assumed to be too weak to exert such control” (p. 187).

298

THOMAS BLASS

The obedience experiment also played a role, though a lesser one, in another type of social perceptual phenomenon-stereotyping. One of the earliest studies providing evidence for the beauty stereotype was one by Miller, Gillen, Schenker, and Radlove (1974). In that study, subjects were given photographs of individuals differing in attractiveness and were asked to indicate the maximum shock these individuals would have given the learner had they been in the obedience experiment. Miller and colleagues found that the less attractive persons were seen as giving higher shocks than the more attractive ones. More recently, I have been engaged in a research program exploring perceivers’ attributions about the behavior of the interactants in the obedience scenario, using an edited version of Milgram’s (1965b) documentary film, Obedience (Blass, 1990a, 1990b, 1991b). For example, one study (Blass, 1990b) attempted to replicate a finding by Tyler and Devinitz (1981). In their experiment, subjects read about a theft from a dormitory room that was described as either occurring quite often (to 1 out of 20 students) or very infrequently (to only 1 out of 200 students). Tyler and Devinitz also manipulated outcome severity by varying the value of the stolen item (either $5 or a stereo worth $300). They found that the theft victim was held more responsible for the rarely occurring theft than for the more frequent one, but that severity of outcome had no effect on responsibility attributions. My study was a conceptual replication of the QlerDevinitz study aimed at determining if frequency rather than severity would still influence attributions even in a highly involving task. I showed subjects a short version of Obedience, which ends with a participant continuing to increase the voltage on the shock generator. The film was stopped right after the participant gives the 180-V shock. Outcome frequency and severity were varied in a factorial design via differential information given to subjects. Some subjects were informed that the subject in the film went no further, whereas others were told that he continued to the final shock (mild versus severe outcome). Outcome frequency was varied by informing subjects that he was either 1 of 15 or of 65 out of a total of 100 subjects who did this. Confirming Qler and Devinitz’s results, I found that outcome frequency, but not severity, had a significant effect on responsibility attributions. Another recent instance of the influence of the obedience paradigm can be seen in Cialdini’s (1985, 1987) organizing framework for the various sources of interpersonal influence. Drawing on the findings of experimental social psychology and the tactics used by compliance professionals, such as car salesmen, Cialdini formulated the following six principles of compliance: reciprocation, commitment and consistency, social proof, liking, scarcity, and authority. His prime support for the power of the authority principle is, of course, the obedience studies (see Cialdini, 1985, pp. 175-180). We have seen the variety of ways the obedience paradigm has been implicated in social psychological research and thought. But ultimately the prime indicators

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

299

of the impact of a paradigm are, first, the extent to which it has stimulated further research involving the paradigm, and, second, the increment in knowledge of the phenomenon provided by that research. Over the years, a number of investigators have conducted replications of the Milgram obedience experiment. By my count, there are about 40 reports of replications in the literature other than Milgram’s own studies. This figure includes both methodological and conceptual replications as well as role-playing versions of the obedience experiment. Although many of the reports involve more than one experimental condition, this is still a rather modest yield considering the amount of attention the obedience studies have received over the years, and in comparison to the number of studies generated by some other social psychological paradigms, such as the risky shift (Cartwright, 1973). Undoubtedly, the advent of federal regulations and the American Psychological Association’s code of ethics on research with human subjects in the early and mid-1970s effectively curtailed the further use of the obedience paradigm in the United States so that, ironically, within a year or two of the publication of Obedience to Authority: An Experimental View (Milgram, 1974), such research by Americans had ground to a halt. It is thus no coincidence that the most recent replications have been conducted in countries other than the United States, i.e., Scotland (Burley & McGuinness, 1977), Jordan (Shanab & Yahya, 1977, 1978), Spain (Miranda, Caballero, Gomez, & Zamorano, 1981), Canada (Shelton, 1982), Austria (Schurz, 1983, and Holland (Meeus & Raaijmakers, 1987). 2 . Impact on Advancing the Understanding of the Dynamics of Obedience More important than their sheer numbers is the advance in our knowledge of the dynamics of obedience to authority, beyond Milgram’s work, that the replications have provided, to which we now turn. Specifically, I will review the accumulated evidence pertaining to three salient questions relating to the Milgram obedience research: Is the subjects’ behavior in the Milgram-type experiment a manifestation of obedience or of aggression? What is the current status of the “agentic-state” ~ o n c e p t ?Did ~ Milgram demonstrate the operation of destructive obedience? a. Is I t Obedience or Aggression? One of the most clear-cut and consistent findings in the obedience literature is that when the subject is allowed to choose the shock level on each trial, rather than being required by the authority to increase it continuously, the subject typically remains at the lower end of the voltage scale and only rarely does a subject use the highest voltage switch. In 7Most of the material that follows pertaining to these two questions was presented at the symposium on Milgram’s contributions to social psychology held at the 1990 APA meetings in Boston, Massachusetts (Blass, 1990~).

300

THOMAS BLASS

such a variant (Milgram, 1974, Expt. l l ) , Milgram found that the average highest shock level chosen by the subjects was 5.5 (on the 30-step shock scale), and only 1 subject out of 40 used the highest shock lever. This is to be contrasted with the appropriate comparison baseline condition (Expt. 5), in which the average maximum shock administered was 24.55, with 65% of the subjects being fully obedient (see Milgram, 1974, Table 3, p. 60). Similar results have been found by others. In an obedience study conducted by Mantell (1971) in Munich, Germany, he found that only 7% of his subjects in a self-decision condition gave the highest shock, compared to an 85% obedience rate in his baseline condition, a voice-feedback variant. In Jordan, Shanab and Yahya (1977) found that only 16% of the subjects (public school children) in a self-decision condition delivered the highest shock, a figure significantly lower than the obedience rate in a standard condition (73%). In an experiment using the same procedures with an adult sample (Jordanian college students), the analogous figures were 12.5 and 62.5%, respectively (Shanab & Yahya, 1978). It should be noted that the shock levels reported by Shanab and Yahya in their selfdecision conditions are higher than those found by Milgram (1974, Expt. 11) and Mantell (1971). This may have been due to a feature of the self-decision condition unique to Shanab and Yahya (1977, 1978), i.e., although subjects were free not to shock at all on any given trial, once they did give a shock, it had to be one step higher on each subsequent error. Another study (Kilham & Mann, 1974), conducted in Australia, also found very low shock levels attained in a self-decision condition. The mean shock level was 3.12, a figure that was significantly lower than the ones in their more standard conditions. A doctoral dissertation by Shalala ( 1974) comprised a rather complicated obedience experiment. Subjects were military personnel at Fort Knox, Kentucky, and the experiment involved seven different conditions. One of them was a selfdecision condition (SD,), which appears to be somewhat comparable to the ones in the experiments described so far. In this condition, the subjects, low-ranking Army tank crewmen, were told by the experimenter, a lieutenant colonel, to “use their own judgments as to when, if, and how much punishment” the learner was to get for making mistakes (p. 50). No subject in this condition gave the highest shock on the scale, compared to a 30% obedience rate in a voice-feedback condition. The distinction between a condition in which a subject is free to choose the shock intensity with which to “punish” a “learner” and one in which a subject is commanded to increase the intensity by one step on each subsequent error is fundamentally important, for two reasons. One is that this constitutes the methodological difference between the Buss aggression paradigm and the Milgram obedience paradigm. Although both Buss and Milgram independently developed a shock machine that allowed for shocks of increasing voltage levels to be

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

30 1

administered, the form of their typical use has varied. Buss and other aggression researchers have allowed the subject to choose from among several shock intensities on each punishment trial and studied the average shock levels (and sometimes durations) chosen as a function of different antecedent conditions (e.g., frustration, presence of cues suggesting violence). Intrinsic to the Milgram obedience format, however, is the incremental nature of the procedure, so that the focus of interest is on the average break-off point or the number of subjects proceeding to the highest possible shock level. The large difference between a self-decision procedure and the standard incremental format in punishment levels typically administered also has theoretical importance. A possible contending explanation, noted by Milgram, for the fact that his subjects were willing to use very high shock intensities to punish a learner is that subjects’ aggressive tendencies rather than the experimenter’s orders were causal (Milgram, 1974, p. 70). That is, according to this view, the experiment merely provides the opportunity for the expression of hostile impulses that people normally have to hold in check. Thus, from this perspective, the apparently severe punishments subjects were willing to inflict on another person would not be a manifestation of the power of an authority to exact obedience, but rather a demonstration of what happens when a person’s aggressive instincts are given free rein. The low shock intensities delivered by subjects in Milgram’s self-decision experiment (Expt. 1 l), and similar findings by the other researchers, which I have described, argue strongly against this alternative explanation. In a self-decision condition, inflicting severe punishments was as acceptable as when subjects were ordered to increase the shock level for each subsequent punishment. Yet in the former, subjects typically restricted their punishments to the low end of the voltage scale. As Milgram (1974) put it, “Insofar as the experiments tell us something about human nature, the revelation on how men act toward others when they are on their own is here. Whatever leads to shocking the victim at the highest level cannot be explained by autonomously generated aggression but needs to be explained by the transformation of behavior that comes about through obedience to orders” (p. 72). It should be noted, parenthetically, that these findings do not completely rule out the operation of individual differences in hostility or aggressiveness. Although subjects in self-decision conditions typically stay at the lower ends of the scale, there is still some variability there that could be due to dispositional differences in aggressiveness. The results of three other variations conducted by Milgram (1974) also speak against the primary determinative role of aggressive tendencies in his experiments. One was Experiment 12, in which the experimenter calls a halt to the experiment but the “learner” wants it to continue, claiming that “it would be an affront to his manliness” not to finish it. A second was Experiment 13, in which the experimenter gets called away, after giving instructions to the subject and to a

302

THOMAS BLASS

confederate playing the role of another subject. He tells them to go ahead with the experiment until the learner learns all the word pairs correctly but says nothing about which shock levels they should use. After the experimenter leaves, the confederate apparently thinks up the idea of increasing the shock on each subsequent error and orders the subject to follow this procedure. In the third experiment (Expt. 14), the authority ends up as the victim, with an ordinary person giving orders to the subject to shock him. This reversal comes about as follows: After the experimenter describes the punishments to be used, the learner (a confederate) says he is afraid of shocks and is willing to participate only after he sees someone else-for example, the experimenter-go through the same procedures. Because of the difficulty he has had in obtaining subjects, the experimenter agrees, provided that the learner serve in the experiment right afterward. The experiment proceeds with the subject administering the shocks. At 150 V, the experimenter demands to be let out but the confederate insists that they continue. Both in this experiment and in the one in which the learner wants to continue but the experimenter says to stop (Expt. 12), no subject continued giving shocks after the experimenter called a halt to the proceedings, and in Experiment 13, in which a peer was giving the orders, only 4 out of 20 subjects ended up giving the highest shock level. Milgram summarizes the significanceof these three experiments as follows: “Those who argue that aggressive motives or sadistic instincts are unleashed when the command to hurt another person is given must take account of the subjects’ adamant refusal to go on in these experiments. It is not what subjects do but for whom they are doing it that counts” (p. 104). So the evidence is solidly supportive of Milgram’s view that the power of the authority, rather than aggressive tendencies, was the primary determinant of the punishing behavior of his subjects. b. What Is the Current Status of the Agentic-State Concept? When we turn to the literature pertaining to the agentic-state concept, however, the evidence is clearly not as supportive. Milgram (1974) had argued that obedience involves divesting oneself of the responsibility for one’s actions and relinquishing it to the authority in charge, and he referred to this process as a shift from a state of autonomy to an agentic state. It was seen by him as the process that enables a person to act destructively, without the usual moral constraints on his or her actions. Although logically compelling, the evidence supporting the shift-in-responsibility process has been weak or contradictory. Milgram’s (1974, pp. 203-204) own evidence comes from postexperimental information gathered from subjects in his proximity series (Expts. 1 through 4 in Milgram, 1974). He asked each participant to divide responsibility among the experimenter, the victim, and himself for giving the victim “shocks against his will.” In line with his agenticstate notion, defiant subjects ascribed a higher percentage of the responsibility to themselves (48.4%)than did obedient subjects (36.3%).However, this finding is

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

303

weakened by the fact that obedients (38.4%) did not relinquish any more responsibility to the experimenter than did defiants (38.8%). Obedients did, however, place twice as much responsibility (25.3%) on the victim than did defiants (12.8%). Although this is an interesting finding given what we know historically about the dehumanization of victims of mass violence, it is not intrinsic to the agentic-shift process. Mantel1 and Panzarella (1976) had three experimental conditions in the previously mentioned obedience experiment conducted in Munich, Germany: a standard, baseline condition, a “delegitimizing model” condition, and a selfdecision condition. Although obedience varied as a function of condition, how much responsibility the subject assigned to himself was not significantly related to amount of obedience. In an experiment by Tilker (1970), subjects were made observers of a Milgram-type experiment and the amount of responsibility for the welfare of the learner was varied. Amount of assigned responsibility did not have a significant overall effect on how much the observer protested the teacher’s actions. However, all five subjects in a condition in which they were given total responsibility and could see and hear the victim physically interfered with the experiment. Also contradictory is the previously mentioned experiment by Kilham and Mann (1974) conducted in Australia. On the one hand, consistent with a shift-in-responsibility process, was their finding that obedience was significantly reduced when the subject had only the indirect role of transmitting the experimenter’s orders rather than actually pressing the shock levers. Milgram (1974, Expt. 18) had obtained a similar result. On the other hand, Kilham and Mann’s experiment had an unusual feature that should have led to increased obedience according to Milgram’s agentic-state conceptualization. Their fourth and final prod to be used on reluctant subjects was “I’m taking full responsibility, so you have no option but to continue with the experiment” (p. 699). (It should be noted that although Milgram’s experimenter apparently occasionally accepted responsibility explicitly [see Milgram, 1965b, 1965c, 1974, pp. 73-77], this was not one of Milgram’s prods nor otherwise a planned feature of the experimental script.) Despite this explicit reference to responsibility-taking by the experimenter, the percentage of subjects who were fully obedient (28%) is the lowest reported in the literature for a standard, baseline condition. A final study, a recent obedience experiment by Schurz (1985) in Austria, is supportive of Milgram. Instead of using electric shock, her subjects gave what they thought was increasingly painful ultrasound to a learner. At the highest points on the scale, this form of stimulation was supposedly capable of causing injury to the skin. Supporting Milgram’s view, she found that disobedient subjects were more willing to take responsibility for their actions than those who were fully obedient. It remains a challenge for future thought and analysis to reconcile this set of findings with Milgram’s agentic-state model. One direction this reconciliation might take is the possibility that only some subjects need to relinquish responsi-

304

THOMAS BLASS

bility in order to be obedient, a possibility suggested by the findings of an obedience experiment conducted in Italy by Ancona and Pareyson (1968).*In their experiment, subjects were instructed to give increasingly painful shocks via a 22-step 330-V shock generator. The victim, played by a professional film actor, was both visible and audible and responded with increasingly intense cries of pain and body movements as the subject increased the voltage, and by 240V the victim was begging “Stop now!” Altogether, 34 out of 40 subjects (85%) were fully obedient. On the basis of the subjects’ answers to a postexperimental questionnaire and the clinical examination of the postexperimental interviews by three judges, Ancona and Pareyson were able to classify the obedient subjects into two groups of 17 each. They report that subjects in the first group participated because they trusted the authority behind the experiment, were preoccupied with the victim’s welfare, and accepted responsibility for any negative consequences of their own actions. On the other hand, the second group of obedients participated in the experiment out of obedience to an order, downplayed the suffering of the victim, focused on their own reactions rather than the victim’s, and denied any responsibility for their own actions. Although some caution is called for because clinical judgment was involved in distinguishing the two groups, these findings are important. They suggest that shedding responsibility is a concomitant of obedience only for some people (i.e., not everybody has to enter into an agentic state in order to be obedient), which might explain the conflicting results pertaining to responsibility-taking and to relinquishing that I just reviewed. c. Did Milgram Demonstrate Destructive Obedience? Unquestionably, one of the most important factors behind the continuing and widespread interest in the obedience work is the disturbing lesson Milgram derives from it about human nature, i.e., just how easily normal individuals can be made to carry out inhumane commands. In particular is the implication that anybody could have acted like the Nazis and willingly participated in the destruction of European Jewry during the Second World War. The potential relevance of his work to the Holocaust was a continuing theme in Milgram’s writings, beginning with the opening paragraph of the first report of his findings (Milgram, 1963). Although Milgram was certainly sensitive to the differences between the conditions in his laboratory and those during the Nazi era (“Is a match flame comparable to the Chicago fire of 1898?,”Milgram, 1974,p. 175), he felt that both events shared a common psychological process: “In the laboratory, through a set of simple manipulations, ordinary people no longer perceived themselves as a responsible part of the causal chain leading to action against a person. . . . What we find in common among soldier, party function*Mythanks to Alfonso Campolattaro for putting much time and effort into translating this article from the Italian.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

305

ary, and obedient subject is the same limitless capacity to yield to authority and the use of identical mental mechanisms to reduce the strain of acting against a helpless victim” (Milgram, 1974, pp. 175-176). And, in 1979, appearing on the TV news program, “Sixty Minutes” (CBS News, 1979), Milgram stated, on the basis of his findings, that “if a system of death camps were set up in the United States of the sort we had seen in Nazi Germany, one would be able to find sufficient personnel for those camps in any medium-sized American town” ( pp. 7-8). Milgram’s analysis has been adopted or at least gets sympathetic treatment not only by many introductory-level texts, but also by more advanced and scholarly works, e.g., Sabini and Silver (1980), Berger (1983), Miller (1986), and Hirsch (1988). A book called Are We All Nazis? by Askenasy (1978) answers that question largely on the basis of Milgram’s work. More recently, in Kelman and Hamilton’s (1989) book, Crimes of Obedience, the obedience work is used as the model for the difficulty people have in defying authority. The validity of Milgram’s assertion that there is a parallel between the behavior of his laboratory subjects and that of the Nazis under Hitler rests on the assumption that obedient subjects in his obedience experiments interpreted the situation the way he had intended it. But is this assumption correct? More specifically, there are two related but separate questions that bear on this issue. First, did subjects believe that the shocks were real? Second, even if they believed that the shocks were genuine, did they also believe that they were harming the victim? In other words, did their actions represent “destructive obedience” as Milgram (1963, p. 371) contended, or merely that they were inflicting pain but not permanent injury? The question about the believability of Milgram’s procedures was first raised by Orne and Holland (Holland, 1967; Orne & Holland, 1968), who argued that the demand characteristics of the setting-i.e., the incongruity of a calm experimenter prodding the subject to go on despite the protests of the learner-revealed to the subject the make-believe nature of the proceedings. If this interpretation of subjects’ behavior in the Milgram experiments is correct, it calls for a differing view regarding the possible parallels between them and Nazi Germany. As Holland put it, “Milgram’s subjects were obedient, but they were obedient only in that they pushed levers. Death-camp executioners were obedient, but they were obedient in that they pushed levers knowing that in doing so they were committing murder. No real relationship may be drawn between these two cases unless it can be established that Milgram’s subjects ‘knew’ that they were shocking the learner as they had been told. This has not been established” (Holland, 1967, p. 71). Milgram (1968, 1972) presented an effective rebuttal of Orne and Holland’s critique, drawing on postexperimental data (Milgram, 1972) to show that most of his subjects believed that the shocks were real and extremely painful. It should

306

THOMAS BLASS

also be noted that if we examine the replications of the obedience studies that have been conducted outside of North America-in Jordan (Shanab & Yahya, 1977, 1978), West Germany (Mantell, 1971), Spain (Miranda et al., 1981), Italy (Ancona & Pareyson, 19681, Austria (Schurz, 1985), and Australia (Kilham & Mann, 1974)-we can note that, in all cases except the Australian replication, the obedience rates were as high or higher than the ones reported by Milgram (1974) for comparable conditions. Subjects in these countries were probably less familiar than North Americans with psychological experimentation and therefore less likely to have the kind of problem-solving set postulated for subjects in psychology experiments by Orne and Holland. Their high rates of obedience speak strongly in favor of the credibility of the obedience experiment for most subjects. Even if the shocks in the obedience experiments were perceived as real, and therefore the learner’s protests genuine, can we also assume that subjects obeyed believing that their actions were destructive, i.e., injuring or killing the victim? The fit between the behavior of subjects in an obedience experiment and the murderous actions of a Nazi stormtrooper is contingent on the assumption that obedient laboratory subjects saw their actions as potentially deadly. This assumption has been questioned by several writers. Mixon (1971, 1976, 1989) has argued that, although subjects might be willing to cause pain, they would not harm another individual. If subjects were certain that the learner was being harmed, they would not obey the experimenter. According to Mixon, subjects obey the experimenter, knowing that in a scientific experiment the safety of participants is assured and trusting the experimenter when he says that although the shocks may be painful, they are not dangerous. Disobedience takes place, according to Mixon, when the learner’s protests make the subject question the experimenter’s reassurances. Mixon’s view of the obedient subjects in a Milgram experiment is therefore a more benign one, explaining their behavior without recourse to any assumptions about their destructiveness. The empirical evidence in support of Mixon’s position comes from a nonactive role-playing procedure. In it, Mixon (1971) read subjects a description of the procedure in Milgram’s (1963) experiment, but stopped reading at the point where the learner pounds on the wall and discontinues providing answers, and then asked his (Mixon’s) subjects to indicate what they believed took place from this point on. By varying relevant details in the description read to different groups of subjects, Mixon was able to show that predicted defiance increased as the harmfulness of the procedure became more certain. Thus, when the script described the experimenter as a competent person and as giving the reassurance that the shocks might cause pain but not permanent tissue damage, subjects predicted that 50% of the teachers would be fully obedient. But when, in another script, the experimenter was described as “decidedly agitated” and, rather than allaying the teacher’s qualms by reassuring him that the shocks do not cause

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

307

permanent damage, indicates instead that he considers the learner’s health to be irrelevant, not one subject predicted full obedience (Table 3, p. 47). Greenwood (1983) also makes a similar distinction between causing pain and causing harm, arguing that Milgram went “too quickly from the assumption that his subjects were convinced they were inflicting painful electric shocks, to the assumption that they were convinced they were inflicting harmful electric shocks” (p. 242). In a conceptual analysis of the Milgram experiments, Patten (1977) also argues for a discontinuity between obedience of the laboratory subject and obedience to commit mass murder. Patten distinguishes between the kind of authority represented by Milgram’s experimenter and the kind wielded by a Hitler. The former possesses what Patten calls “expert-command’’ authority. That is, he commands obedience by virtue of his presumed expertise regarding learning and shock devices. The latter, more worrisome, kind of authority wields a “simplecommand” authority, i.e., whose power to command is based on legal or quasilegal considerations, not because of any special knowledge of the task at hand. According to Patten, knowing how one might respond to expert-command authority cannot inform us about the person’s behavior vis-a-vis a simple command authority. Thus, the viewpoints just presented all share the assumption, explicitly or implicitly, that, in contrast to the murderous obedience that took place during the Third Reich, obedient subjects in the Milgram experiments typically do not believe they are harming their victim. The force of this argument is diminished, however, by the findings of two experiments that used the Milgram paradigm. In one study (Ring, Wallston, & Corey, 1970), using “painful” sound rather than shocks, 91% of the subjects were obedient-the highest rate reported in the literature for a standard obedience experiment. This high rate of obedience occurred despite the fact that the experimenter’s behavior conveyed the impression of possible harm to the learner. He acted as if he had not expected the learner’s complaints: At various points in the procedure, he shook his head in perplexity, and wiped his forehead. In Munich, Germany, Mantell (1971) conducted a study, referred to earlier, consisting of a set of obedience experiments, one of which, comparable to Milgram’s voice-feedback condition, yielded an 85% obedience rate. Altogether, 101 subjects served as participants. A postexperimental questionnaire revealed that 26 subjects believed that the learner was “either dead or might have died,” 23 believed that he lost consciousness, and 5 believed that he suffered some sort of damage. On the other hand, 36 believed that he had undergone severe pain but “was all right,” whereas 9 felt that the learner was “just fine.” Thus, approximately half of the participants believed that they had harmed the victim, some with lethal consequences. The findings are clouded somewhat by the fact that Mantell does not break them down by condition nor by whether or not the subject

308

THOMAS BLASS

was obedient, though he does state that “the subjects’ estimates of the consequences for the [learner] correlate relatively well with the number of shocks given and therefore the number of screams heard: the more shocks given, the more severe the estimates” (p. 109). In sum, whether one considers obedient laboratory subjects as having acted destructively, with its attendant implications about a pervasive human propensity to obey malevolent authority, depends on which arguments and evidence one gives more credence to-those consistent with Milgram’s position or the ones in line with Mixon’s view.

B . CONFORMITY Milgram also made a number of important contributions to another form of social influence besides obedience, i.e., conformity. Although his conformity research broke new ground, it has rarely been recognized as such, having been overshadowed by the obedience research. Milgram’s conformity experiments were groundbreaking in the following three ways. First, his cross-cultural study of conformity (Milgram, 1960, 1961), described earlier, was the first of its kind. Second, using an adaptation of his laboratory paradigm, he showed that peer pressure can induce an individual to give increasingly more severe shocks to a learner (Milgram, 1964a), and thereby extended Asch’s line-judgment findings to a more consequential type of conformity. And, third, in a study aptly titled “Liberating Effects of Group Pressure,” Milgram (1965a) showed that group pressure could yield laudable outcomes, and thereby departed radically from previous laboratory studies of conformity, which typically demonstrated the negative, constricting effects of group influence. In this variant of the obedience experiment, there were three teachers-the real subject and two confederates. Partway into the shock series, the latter defy the experimenter and refuse to continue. Fully 90% of the subjects followed suit, dropping out at some point before the end. According to Milgram (1974, p. 118), no other variant was as effective in weakening the authority’s power as this one, and he draws an important lesson from his findings: “The mutual support provided by men for each other is the strongest bulwark we have against the excesses of authority” (Milgram, 1974, p. 121). In a sense, then, Milgram provided a powerful antidote to enable individuals to withstand an authority’s unwanted influence-an optimistic finding that sometimes gets lost amidst the typically gloomy textbook presentations of the obedience studies. At the same time, this study’s implications were not lost on several researchers, who produced a creative diversity of studies inspired by it. Thus, Kudirka (1965), a student of Milgram, found that the experimenter’s power to induce subjects to eat bitter, quinine-soaked, crackers was drastically

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

309

weakened when they observed a peer (a confederate) dissenting. When subjects had to confront the experimental authority by themselves, only 26% were defiant, but when they saw a confederate drop out of the experiment, more than twice as many (58%) were able to defy the experimenter’s orders. Feldman and Scheibe (1972) also showed that defiant peers could help free the subject from an experimenter’s demands to carry out another kind of distasteful request-to complete a highly intrusive personality questionnaire. In one condition, naive subjects start filling out the questionnaire together with a confederate posing as a subject. Soon the confederate, visibly perturbed by the questions, announces, “This is no one’s business but my own,” tears up the questionnaire, and leaves the room. In a second condition, he is joined by another confederate 2 minutes later, while in a third condition, a total of three confederate-subjects, one after another, depart, without completing the questionnaire. In a fourth condition, no confederates were present in the room, and all the subjects ended up completing the experiment. But in the conditions containing confederates, as the number of dissenting confederates increased, so did the number of subjects who defied the experimenter. A more recent study (Goldstein, 198l), a doctoral dissertation supervised by Milgram, showed that peers could effectively weaken or strengthen the power of the “authority” of television. Subjects watched a series of television messages dealing with a variety of topics, e.g., film reviews and health messages, which had been taped off the air, together with a confederate. The confederate, who was either a friend or a stranger, was instructed to either agree with, disagree with, or not to comment on designated messages right before the subject was to give her judgment. Goldstein found that the peer’s comments significantly modified the subject’s reactions to the television messages. In the “disagreement” condition, the subject showed significantly less agreement with the TV messages than in the “no-comment’’ condition, whereas the confederate’s agreeing comments elicited significantly more agreement than the latter. Furthermore, friends were more effective than strangers in influencing subjects’ judgments. A final instance of a study that built on Milgram’s (1965a) conformity-to-defiant-peers findings was a study by Hunt (1979) involving hypnosis. Hunt found that when a subject underwent a hypnotic induction together with a confederate who refuses to continue and leaves the room about halfway into the induction, this significantly lowered the subject’s degree of obedience to the hypnotist’s authority, i.e., reduced the “depth” of hypnosis he or she attained. Thus, in various ways, these four studies reflect the impact of one aspect of Milgram’s work on conformity. Each study not only extended the domain of applicability of Milgram’s (1965a) findings but also contains the seeds for further extension and inquiry. Thus, for example, Goldstein’s (1981) television paradigm could readily be adapted to and tested for its effectiveness as a technique for counteracting antisocial and other negative forms of television fare. A further contribution to our understanding of social influence comes from

310

THOMAS BLASS

Milgram’s (1974, pp. 113-1 15) conceptual analysis of conformity and obedience, yielding some clarifying distinctions between the two of them. First, obedience always involves social inequality, with the person of higher status prescribing the behavior of a person of lower status. Conformity, on the other hand, represents a person yielding to a group of peers, who have no special prerogatives to regulate his behavior. Second, conformity to a group involves imitation, whereas yielding to an authority does not. Third, although in conformity the pressure exerted is only implicit, obedience represents a reaction to explicit commands. Fourth, when people yield to group pressure, they typically deny a causal role to members of the group. In contrast, when a subject obeys the orders of an authority, the former readily admits the latter’s influence.

C. CITY LIFE Milgram’s move to the City University of New York in 1967 soon produced his article, “The Experience of Living in Cities” (Milgram, 1970a), which was “seminal to the growing field of urban psychology” (Sabini, 1986, p. 1378). In it, Milgram applied the concept of overload to help understand adaptation to city life, and through illustrative field experiments on urban-nonurban differences in helping behavior conducted with his students, he demonstrated that it was possible to apply a disciplined, social psychological approach to the study of citytown differences. Based on an invited address Milgram gave at the annual convention of the American Psychological Association in Washington on September 2, 1969, the article achieved the status of a citation classic within a decade (Milgram, 1981b). It has been one of the most oft-cited articles in both the urban social behavior and helping literatures (Korte, 1990), and has also been discussed in some textbooks on urban sociology (Fischer, 1984; Karp, Stone, & Yoels, 1977). Building on earlier writings by the urban sociologists, Simmel (190311950) and Wirth (1938), Milgram presented the concept of overload to describe the state of the individual typically confronted by the massive, densely packed, and heterogeneous population of a city. According to Milgram (197Oa, p. 1462), the concept of overload “drawn from systems analysis, refers to a system’s inability to process inputs from the environment because there are too many inputs for the system to cope with, or because successive inputs come so fast that input A cannot be processed when input B is presented. When overload is present, adaptations occur. The system must set priorities and make choices. . . . City life, as we experience it, constitutes a continuous set of encounters with overload, and of resultant adaptations.” The phenomenon of the “familiar stranger” (Milgram, 1977b)-people we see regularly but never interact with, such as fellow commuters at a train station-represents one such adaptation.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

311

Why has the concept caught on? A possible reason, which I believe has a lot of merit, has been articulated by Little (1987, p. 222): There is a general thesis of considerable popularity within environmental psychology to the effect that humans have limited capacity for processing information, that contemporary environments, particularly urban ones, provide a surfeit of such information, and that this situation in part creates much of the malaise of current living. Certainly the most influential statement of this thesis was Milgram’s (1970a) elegant formulation of the insidious effects of information input overload in generating urban pathology. . . . Milgram’s ability to relate urban indifference and hostility to the vagaries of information processing struck a responsive chord in social psychologists who. no doubt, valued the blend of cognitivism, contextualism, and social relevance in Milgram’s perspective.

“The Experience of Living in Cities” has been influential both in theory development and in stimulating research. The overload concept is one of the underpinnings of Altman’s (1975) theory of crowding. According to Altman, crowding takes place when various privacy-regulation mechanisms fail to attain our desired level of privacy. Geller (1980), a student of Milgram, added the concept of optimal level of stimulation to the overload concept to produce a more “balanced” formulation, which sees urban stimulation as potentially resulting in positive as well as negative outcomes. And Cohen’s (1978) approach to the study of urban and environmental stressors extended Milgram’s model to encompass physical as well as social sources of overload. In a recent assessment of the overload concept, Korte (1990) states that the 1970s marked a significant increase in research conducted on urban-nonurban differences in social behavior and expresses the belief that many of the researchers involved had been inspired by Milgram’s work, often as a result of studying with him as his graduate students. What does the future hold for Milgram’s overload concept? According to Korte (1990, p. 8), it “may depend on the occurrence of additional insights and understandings of this fascinating notion. Since computer systems were in part an inspiration to the thinking of Milgram and others on this point, perhaps future developments in computer systems will revitalize and reshape the overload concept. The field of urban social psychology has been enriched by Stanley Milgram and the input overload concept, as well as by the research that has resulted in this area, and I am hopeful that this contribution will prove to be of lasting value.” Milgram’s “cognitive maps” studies reflected another aspect of his interest in the psychology of urban life. In fact, Milgram’s (1970a) article ends with a section titled “Cognitive Maps of Cities,” which essentially previews the technique used in Milgram et al. (1972) in which New Yorkers were asked to identify photographs of geographically sampled locations throughout the five boroughs of New York City. Although not nearly as influential as his urban overload concept, “A Psychological Map of New York City” (Milgram et al., 1972) brought

312

THOMAS BLASS

increased precision and rigor to the study of the “imagability” of cities pioneered by the urban planner, Lynch (1960); its findings, together with those of the methodologically richer cognitive-mapping study of Paris (Milgram & Jodelet, 1976, 1977), represent substantive contributions to the accumulated body of knowledge on environmental cognition (for reviews, see G. W. Evans, 1980; Golledge, 1987).

D. TELEVISION AND ANTISOCIAL BEHAVIOR When Milgram turned to the study of television violence (Milgram & Shotland, 1973), he did so in a manner that remains unique to the present day. Virtually all experimental research on the effects of televised antisocial behavior uses already-existing fare as stimulus materials. In a typical study, subjects are randomly assigned to watch either a program that is clearly violent in content or one that is free of violence, and are subsequently compared on some dependent measure of aggression. A built-in problem with this approach is that the chosen programs may also differ in other qualities besides the presence or absence of antisocial themes. Milgram and Shotland (1973) departed radically from this traditional approach by manipulating the independent variable via the systematic variation of the content of one episode of a popular prime-time television series. Specifically, Milgram was able to get the cooperation of CBS to produce a particular episode of the dramatic series, Medical Center, with three different endings. In that episode, the plot revolves around the misfortunes of a hospital orderly, Tom Desmond: He loses his job, with a sick wife and an infant to support, and his boat, which he had hoped would provide some income, is repossessed. %o versions end with a destructive act, the orderly repeatedly smashing open fund-raising collection boxes distributed by the Medical Center and stealing the money they contained. In one variant he is caught by the police; in the other, he eludes capture. The third version had a prosocial ending, with Desmond putting money into, rather than stealing from, a collection box. The two antisocial versions also showed the orderly placing two abusive phone calls to the medical director of the hospital. Another episode, with a completely different storyline, without any suggestion of violence whatsoever, served as a baseline, control condition. Milgram and Shotland embedded these programs in a series of field experiments in which both viewing and the opportunity to imitate the antisocial acts occurred in real-life setting^.^ The study is not without shortcomings. One of the eight experiments, for T h e first seven experiments used the Medical Cenrer episodes. In order to determine whether depicting the antisocial act as an actual, rather than a fictional, event made a difference, the eighth study embedded it in a videotape of a local newscast.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

313

example, recruited subjects from New York’s Times Square area, a group that apparently contained vagrants, alcoholics, and drug addicts-not a very generalizable sample. I have a copy of the prosocial version of the Medical Center episode used by Milgram and Shotland. I regularly show it to my social psychology classes and also in a winter session course on the social psychology of Stanley Milgram which I have created (Blass, 1991~).Toward the end of the film, whenever Desmond fixes his gaze on a charity box, his potential prey, the soundtrack accompanies it with a “boing-boing” sound to bring it into dramatic focus. But the sound, not unlike the kind one hears when a cartoon character bounces back after running into a brick wall, creates an unintended melodramatic effect, as evidenced by the laughter I sometimes hear from my students as they watch that scene. Also, as a number of writers have correctly pointed out, no theoretical context is provided by Milgram and Shotland for their study and only minimal attention is given to prior relevant research on observational learning and media influence (see Comstock, 1974; Dominick, 1975; Feshbach, 1975). Nevertheless, its unique methodology and scope (a total of eight integrated field experiments and, in one of them, millions of at-home viewers as potential subjects) make its null-effects findings deserving of special attention in any discussions of media influences. Yet a look at reviews of the literature on TV violence and viewer aggression shows that Milgram and Shotland’s research program has had very little impact on the debate about the harmful effects of television. Most reviews, if they mentioned it at all, have dismissed it because of methodological problems (Comstock, Chaffee, Katzman, McCombs, & Roberts, 1978; Geen & Thomas, 1986; Huesmann, 1982; Liebert, Sprafkin, & Davidson, 1982; but see also Freedman, 1988; Kaplan & Singer, 1976, for more favorable reactions to the study). In focusing largely on the flaws in the Milgram-Shotland television experiments, these reviews overlooked what their unique design was able to achieve-a degree of control over the independent variable rarely attained in experiments on the effects of TV violence, a design feature that is crucial if one wants to be able to draw causal conclusions.

E. THE LOST-LETTER TECHNIQUE In a brief note, Milgram (Milgram et a]., 1965) introduced the lost-letter technique as a method “for assessing community orientations toward political groups and other institutions” ( p. 437). The technique involves “losing” stamped, addressed envelopes in various locations-on sidewalks, in stores, telephone booths, and under windshield wipers of cars. By varying the name of the addressee on the letters, and comparing the proportions of the letters mailed to each target, prevailing attitudes toward various groups or organizations can be measured unobtrusively. The technique had originated with Merritt and Fowler

3 14

THOMAS BLASS

(1948); however, they used it to study the personal honesty of the finders of the letters, whereas Milgram’s innovation was to use rate of returns “as a clue to how people felt and-more important-how they would act toward different political organizations” (Milgram, 1969c, p. 3 1). .In the first experiment, conducted in New Haven, Connecticut, 400 letters were distributed, with 100 letters each bearing the following designations: Friends of the Communist Party, Friends of the Nazi Party, Medical Research Associates, and a Mr. Walter Carnap. The number of letters returned was 25, 25, 72, and 71, respectively. Milgram and his students went on to use the technique to predict the winner of the 1964 Presidential election, to study attitudes toward racial integration in Connecticut and North Carolina, to determine the political allegiances of the Chinese residents of Honk Kong, Bangkok, and Singapore, and to identify pockets of neo-Nazi sentiment in Munich, Germany (Milgram, 1969~). The lost-letter technique has been the most widely used nonreactive measure of attitude and opinion. By 1981, over 30 studies had used it (Webb, Campbell, Schwartz, Sechrest, & Grove, 1981; see it, as well as Sechrest & Belew, 1983, for reviews), and it continues to be used for a variety of purposes, from studying the Catholic-Protestant conflict in Northern Ireland (Kremer, Barry, & McNally, 1986) to measuring attitudes toward abortion (Kunz & Fernquist, 1989). Perhaps more important than the number of studies that have employed it is the fact that, more generally, the lost-letter technique has helped sensitize social psychologists about the value of nonreactive measures in attitude measurement. The technique has a number of limitations, some of which were identified by Milgram (1969a, 1969c; Milgram et al., 1965): There is no control over the mediating processes leading to the return of the letters; it should work for highly polarized and emotionally charged issues but not for subtle ones; because a lot of uncontrolled yariance is involved, response rates cannot provide exact information about population parameters, and for the same reason, large numbers of letters need to be distributed in order to be able to obtain between-condition differences; its primary usefulness is when the respondents’ awareness that they are being studied would seriously distort their answers, an opinion echoed in Sechrest and Belew’s (1983) cautionary evaluation of the lost-letter technique: It seems like a relatively simple nonreactive measure on the surface, but factors such as location of the drop sites, preservation of a stamp, potential curiosity arousal of the finders in response to some addresses, and the possibility of money inside the envelopes all can affect returns and interfere with concluding anything but finders’ attitudes. However, the method should not be disregarded completely. Ir very well may prove to be rhe optimal method, or at least a useful supplement to other methods. when rhe issues for which attirudes are being assessed are of a highly sensitive and rhus reacrive narure. What is

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

315

absolutely essential, however, is that potential users of the technique do systematic methodological testing before employing the measure for actual attitude assessment. (p. 42; emphasis added)

F. THE SMALLWORLD METHOD In 1979, communication researcher Everett M. Rogers was a visiting professor at a university in Mexico City. One day, while having a conversation with a student named Pedro Felix Hernandez, Rogers complimented him on the excellence of his English. The student replied that this was a result of having lived with a family in Iowa when he was there as an exchange student. The conversation continued (Rogers & Kincaid, 1981, p. 107): ROGERS:Oh, where in Iowa did you live? PEDRO:With a farm family in Collins. ROGERS:Collins? That’s the community I studied for my Ph.D. dissertation at Iowa State University! What is the family’s name? PEDRO:Robert Badstubner. ROGERS (in amazement): Why, that farmer was one of the opinion leaders in my investigation of the diffusion of 2.4-D weed spray! ROGERS A N D PEDRO(in unison): What a small world!

Two people meeting for the first time and discovering that they have someone illustrated in the just-quoted conversation between the professor and the student-happens surprisingly often, and has been referred to as the small-world phenomenon (Kochen, 1989). The more general question these kinds of encounters pose is the small-world problem, which Milgram (1967b) formulated as follows: “Starting with any two people in the world, what is the probability that they will know each other? . . . Another question one may ask is: Given any two people in the world, person X and person Z, how many intermediate acquaintance links are needed before X and Z are connected?” ( p. 62). l o To provide an answer, Milgram devised the small-world method, which involves the following features: The name of a randomly selected person (the target), usually in a distant city, is given to a sample of men and women (starters). The task of the starters is to send a folder to the target person using only a series of friends and acquaintances who would be more likely to know the target person than they (the starters) would. The folder could only be sent to a person

in common-as

‘There is a slightly longer version of this article (Milgram 1969b). with a different set of illustrations, including a reproduction of the contents of the folder that was used in the mailings.

316

THOMAS BLASS

whom the sender knows on a first-name basis. To keep track of the course of the folder, it contained a roster to which each subject had to add his or her name and tracer postcards to be mailed to Milgram. Among the main findings obtained by Milgram and colleagues (Korte & Milgram, 1970; Milgram, 1967b; Travers & Milgram, 1969) using the smallworld method, were the following: 1. The average number of intermediariesin completed chains ranged from 4.4 to 5.9. 2. Subjects were much more likely to send the mailing on to friends or acquaintances than to relatives. 3. Only a fraction of the chains were completed. For example, in a study with starters in Nebraska and a stockbroker in Boston as the target person, only 24% of the mailings eventually reached him. 4. In an experiment composed of white starters in Los Angeles who had to reach either a white or black target in New York City, a chain was two and a half times as likely to reach the former than the latter. Milgram’s reports on the small-world method stimulated a good deal of research, and, by 1979, there were already enough studies using the technique to merit a review article (Bernard & Killworth, 1979). Although a few social psychologists other than Milgram and his colleagues have used the small-world method (e.g., Bochner, Duncan, Kennedy, & Orr, 1976; Shotland, 1976), its primary impact has been among social-network researchers, a group composed largely of sociologists, political scientists, and anthropologists. Killworth and Bernard (197811979), for example, were inspired by the small-world method to create their own technique for examining social networks, called the reverse small-world technique. In this method, respondents are given a large number of targets and are asked to indicate their choice of the first link in a potential chain to reach each target. In another modification of the small-world method, Guiot (1975) substituted telephone calls for mailings, and achieved a much higher completion rate (85%) than did Milgram and colleagues. Studies of social networks have typically relied on self-report measures, e.g., questions asking subjects to list individuals with whom they have a particular kind of relationship (Marsden, 1990). Through the small-world method, Milgram has enabled network researchers to use subjects’ actions as a source of information about their social ties. Perhaps the most salient indicator of the degree of influence of the small-world method on the field of social-network research is the recent publication of an edited volume, The Small World (Kochen, 1989), which has as its subtitle, “A volume of recent research advances commemorating Ithiel de Sola Pool, Stanley Milgram, Theodore Newcomb.” (Unpublished working papers by Pool and Kochen presenting a theoretical model of the small world had stimulated

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

317

Milgram’s interest in the problem.) A number of the chapters in the book deal with the small-world method. Kochen (1989) describes Milgram’s (1967b) first article on the subject as a “seminal paper” (p. ix), and states that the technique “established an experimental paradigm for this field” (p. 149, and Kadushin (1989) notes that it “remains one of the critical tools of network analysis” (p. xxiv). A chapter by Klovdahl(1989) explores the limitations of the small-world method (see also B. H. Erickson, 1978).

V. Conclusions American psychology has had a love-hate relationship with Stanley Milgram. Although his obedience research is a classic of modem psychology and a “must” topic for the introductory-level course, the American Psychological Association never gave him its Award for Distinguished Scientific Contributions, an award it has given to several important social psychologists over the years. I I Social psychologists will invariably invoke the results of the obedience experiments whenever we need to affirm that our field can reveal something about social behavior that is not predictable from common sense. Yet most writings on the history or theories of social psychology make only passing mention of Milgram-if they mention him at all. These lacunae would seem to be at odds with the fact that references to his writings by scholars and researchers continue unabated: An on-line bibliographic search using the Institute for Scientific Information’s “Social Science Citation Index” data base reveals that, through October, 1991, he was cited 46 times in 1991 alone and a total of 1720 times since 1972. Yet, Milgram must take at least part of the blame if the conventional, “establishment” forms of recognition eluded him. I believe it had a lot to do with the impression he created among some psychologists of a dilettante who flitted from one newsworthy phenomenon to the next, not staying with any one long enough to probe it with adequate depth. Even those with a more favorable view of Milgram’s accomplishments are not likely to think of him foremost as someone who did programmatic, time-consuming research. Yet he did exactly that, three different times in his career. His obedience research comprised over 20 interrelated studies spanning the years 1960 to 1963. The television study, consisting of eight sequential field experiments, was first proposed to CBS on April 23, 1969. Most of 1970 was devoted to correspondence with the writer of the Medical “He did win, in 1964, the annual Socio-Psychological Prize of the American Association for the Advancement of Science for his article “Some conditions of obedience and disobedience to authority” (Milgram, 196.5~).

318

THOMAS BLASS

Center episode that was to serve as the stimulus program and to the actual production of the film, with Milgram sitting “on the sidelines at MGM studios acting as a censor whenever an element was introduced that did not meet the needs of the experiment” (Milgram & Shotland, 1973, p. 74). The experiments themselves were conducted from September, 1970, through November, 1971. To conduct his dissertation research on conformity, Milgram spent 1957- 1958 in Norway and 1958-1959 in France, carrying out a total of 14 experimental variations. Including subjects who participated in pretesting the procedures at Harvard during the summer of 1957, a total of 390 participants served in the conformity experiments (Milgram, 1960). Milgram was equally at home publishing in magazines as in journals, which made him an effective disseminator of psychological information to the public. At the same time one sometimes almost got the feeling that for Milgram magazines such as Psychology Today (e.g., Milgram, 1977d) and Omni (Milgram, 1984c) were functionally equivalent to the Journal of Personality and Social Psychology. Thus, the first article about the small-world method was published not in a scientific journal but in Psychology Today (Milgram, 1967b). I believe that similarly odd publication decisions were at least partially responsible for two of the three aforementioned research endeavors not getting their proper recognition as sequentially developed programmatic efforts. I have already mentioned, in a previous section, the piecemeal nature of the publication of the obedience research: It took Milgram a full 9 years after the appearance of the last journal report on the obedience work (Milgram, 1965~)to reveal (in Milgram, 1974) the complete research program, which included experimental conditions not described in earlier publications. With his cross-cultural research on conformity (his doctoral dissertation, Milgram, 1960), Milgram completely flouted the norms of scientific publication because the only place he published it was in ScientificAmerican (Milgram, 1961). Because readers do not normally expect a magazine to be the primary publication vehicle for the presentation of original research findings, for a long time many psychologists remained unaware of the conformity studies and the extensive program of research they represented. Despite the lack of attention to Milgram in most writings on the history of social psychology, he can readily be placed in a historical context. His phenomenon-centered approach represents a continuation, through Asch, of the Gestalt tradition. At the same time, his boundless confidence in the possibility of studying a wide range of social phenomena scientifically makes him supremely Lxwinian. The innovative work of some of Milgram’s students and colleagues bears his imprint. Examples that come to mind are Gaertner and Bickman’s (1971) wrongnumber technique; Korte’s operationalization of the urban overload concept (Korte, Ypma, & Toppen, 1975); Takooshian, Haber, and Lucido’s (1977) lostchild study; and Mann’s (1981) attempt to account for the formation of baiting

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

3 19

crowds at episodes of threatened suicide. The work of Sabini and Silver (e.g., Sabini, 1976; Sabini & Silver, 1982, 1985; Silver, 1977) has continued and amplified Milgram’s focus on the moral aspects of social behavior. Milgram also had a lasting influence on his students in less easily identifiable ways. For example, Judith Waters, a former student and research assistant, told me that every time she conducts a study she asks herself: “Is this how Stanley would have done it?” Milgram spent most of his professional career ( 1967- 1984) at the Graduate Center of the City University of New York. During that time, he supervised more doctoral dissertations in social psychology than any other CUNY faculty member. Many more students undoubtedly came into contact with him through coursework, seminars, or as a member of their dissertation committees. Yet his students have not had the kind of impact on social psychology as have those of Festinger and Schachter. 1 am not sure why this is, but one possibility is contained in a tribute by one of Milgram’s former students, which she wrote in the acknowledgments section of her doctoral dissertation: Although the late Stanley Milgram did not live to see this dissertation completed. he brought an element of realism and vision to this work. Stanley understood the strengths and limitations of social psychology. Likewise, he had a true appreciation of the strengths and limitations of human beings, and had the wonderful ability to elicit the most out of both his students and science. Although he had dijjiculfy in transmitting his innate creativity to students, as a teacher-exemplar he taught me basic tools of social psychology. But most of all I will cherish Stanley’s altruistic nature and the positive reinforcement I received from him. (Fogelman, 1987, p. ix, emphasis added)

Milgram sensitized us to the hidden workings of the social world. He showed the difficulty people often have of bridging the gap between intentions and actions. Even moral principles are not invariably translated into behavior but can have their potential power overridden by momentary situational pressures. Although people can be responsive to the precise and subtle details of the immediate situation, people are generally not aware of their power and typically think of themselves as free of social pressures. Social norms can often have a compelling effect on our behavior, wielding their power by means of the unexpected amount of inhibitory anxiety generated by their violation. While some critics were chipping away at the bedrock findings of social psychology through claims of bias and artifact, Milgram was expanding its proper domain by turning his attention to such topics as photography, Candid Camera, mental maps of cities, and subway norms. To dispel the gloom created by periodic pronouncements about a crisis in social psychology, one could always share vicariously in the “pleasurable activity of experimental invention” (Milgram, 1977e, p. 2) through exposure to such innovations as the lost-letter technique and the small-world method.

320

THOMAS BLASS

What is remarkable is that despite the pessimism occasioned by the obedience findings, Milgram was able to maintain a positive, hopeful view of human potential: “Though enormous pressures may bear down on a person to abandon critical intelligence, dispense with conscience, and surrender humanity, that person will often prove hardy and resilient, transcend the pressure of the moment, and reaffirm the power and integrity of one’s own spirit. Well, as our experiments show, it does not always happen this way. But it remains a worthy ideal” (Milgram, 1977e, p. 155). This, perhaps, is his greatest legacy.

Acknowledgments I thank Judith Waters, John Sabini, Stanley Feldstein, and an anonymous reviewer for their helpful comments on an earlier version of this article.

References Abse, D. (1973). The dogs of Pavlov. London: Valentine. Mitchell & Co. Akerlof, G. A. (1991). Procrastination and obedience. American Economic Review, 81, 1-19. Albrecht, M. 1. (1973). Conformity and dissent in the absence of consensus. Unpublished master’s thesis, St. Louis University, St. Louis, MO. Altman, I. (1975). The environment and social behavior. Monterey, CA: Brooks/Cole. Ancona, L., & Pareyson, R. (1968). Contributo allo studio della aggressione: La dinamica della obbedienza distruttiva [Contribution to the study of aggression: The dynamics of destructive obedience]. Archivio di Psicologia. Neurologia, e Psichiatria. 29, 340-372. Armstrong, J. S. (1975). [Review of Obedience to authority].Journal of Marketing, 39, 125. Aronson, E., & Carlsmith, J. M. (1968). Experimentation in social psychology. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology: Vol. 2. Research methods (2nd ed., pp. 1-79). Reading, MA: Addison-Wesley. Asch, S. E. (1952). Social psychology. E n g l e w d Cliffs, NJ: Prentice-Hall. Askenasy, H. (1978). Are we all Nazis? Secaucus, NJ: Lyle Stuart. Atlas, A. J. (1985). Enlightenment eclipsed: German schools and society. Doctoral dissertation, Boston University, Boston. (University Microfilms International No. 8602726) Baumrind, D. (1964). Some thoughts on ethics of research: After reading Milgram’s “Behavioral study of obedience.” American Psychologist, 19, 421 -423. Berger, L. (1983). A psychological perspective on the Holocaust: Is mass murder part of human behavior? In R. L. Braham (Ed.),Perspectives on the Holocaust (pp. 19-32). Boston: KluwerNijhoff. Bernard, H.R., & Killworth, P. D. (1979). A review of small world literature. Sociologicul Symposium, 28, 87-100. Billig, M. (1990). Psychological aspects of Fascism. Patterns of Prejudice, 24, 19-31. Blass, T. (1977). On personality variables, situations, and social behavior. In T.Blass (Ed.), Personality variables in social behavior (pp. 1-24), Hillsdale, NJ: Erlbaum.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

32 1

Blass, T. (1984). Social psychology and personality: Toward a convergence. Journal of Personality and Social Psychology, 47, 1013-1027. Blass, T. (1990a. March 30). Attribution of responsibility and trust in the Milgram obedience experiment. Paper presented at the annual meeting of the Eastern Psychological Association, Philadelphia. Blass, T. (l990b, June 8). Judgments about the Milgram obedience experiment support a cognitive view of defensive attribution. Paper presented at the annual convention of the American Psychological Society, Dallas, TX. Blass, T. (199Oc, August 13). Obedience to authority: Some issues and significance. In T. Blass (Chair), Perspectives on Stanley Milgram’s contributions to social psychology. Symposium conducted at the annual meeting of the American Psychological Association, Boston. Blass, T. (1991a). Psychological perspectives on the perpetrators of the Holocaust: The role of situational pressures, personal dispositions. and their interactions. Holocaust and Genocide Studies, in press. Blass, T. (1991b). Right-wing Authoritarianism and role as predictors of attributions about obedience to authority. Under review. Blass, T. (1991c, April 14). Teaching a course on the social psychology of Stanley Milgram. Paper presented at the annual meeting of the Eastern Psychological Association, New York. Blass, T. (1991d). Understanding behavior in the Milgram obedience experiment: The role of personality, situations, and their interactions. Journal of Personality and Social Psychology. 60,398413. Blass. T., & Krackow. A. (1991, June 14). The Milgram obedience experiments: Students’ views vs. scholarly perspectives and actual findings. Paper presented at the annual convention of the American Psychological Society, Washington, DC. Bochner, S . , Duncan, R., Kennedy, E., & Orr, F. (1976). Acquaintance links between residents of a high rise building: An application of the “small world” method. Journal of Social Psychology, 100, 277-284. Boneau, C. A. (1990). Psychological literacy: A first approximation. American Psychologist. 45, 891-900. Brown, R. (1974, February 24). Authority [Letter to the editor]. New York Times, Book Review section, pp. 42-43. Burley, P. M., & McGuinness, J. (1977). Effects of social intelligence on the Milgram paradigm. Psychological Reports, 40, 767-770. Bushman, B. J. (1985). A curriculum on obedience to authority. Unpublished master’s project, Utah State UniversityIWeber State College, Logan. Cartwright, D. (1973). Determinants of scientific progress: The case of research on the risky shift. American Psychologist, 28, 222-23 I . CBS News (1979, March 31). Transcript of Sixty Minutes segment, “I was only following orders,” pp. 2-8. Chittenden, P., & Kiniry, M. (Eds.). (1986). Making connections across the curriculum: Readings for analysis. New York: St. Martins Press. Cialdini, R. B. (1985). Infuence: Science and practice. Glenview, IL: Scott, Foreman. Cialdini, R. B. (1 987). Compliance principles of compliance professionals: Psychologists of necessity. In M. P. Zanna, J. M. Olson, & C. P. Herman (Eds.), Social influence: The Ontario symposium (Vol. 5 , pp. 165-184). Hillsdale, NJ: Erlbaum. Cohen, S. (1978). Environmental load and the allocation of attention. In A. Baum, J. E. Singer, & S. Valins (Eds.), Advances in environmental psychology (Vol. 1, pp. 1-29). Hillsdale, NJ: Erlbaum. Colligan, D. (1980, March). The high priest of guerilla psych [sic]. Omni, pp. 108-110, 112-115. Colman, A. M. (1991). Crowd psychology in South African murder trials. American Psychologisr, 46, 1071-1079.

322

THOMAS BLASS

Comley, N. R., Hamilton, D., Klaus, C. H., Scholes, R., & Sommers, N. (Eds.). (1984). Fields of writing: Readings across the disciplines. New York: St. Martin’s Press. Comstock, G . A. (1974). Review of Television and antisocial behavior: Field experiments. Journal of Communications. 24, 155-158. Comstock, G . A,, Chaffee, S., Katzman, N., McCombs, M., & Roberts, D. (1978). Television and human behavior. New York: Columbia University Press. Could we be Nazi followers? (1964, January). Science Digest, pp. 81-82. Crawford, T. (1972). In defense of obedience research: An extension of the Kelman ethic. In A. G . Miller (Ed.), The social psychology of psychological research (pp. 179- 186). New York: Macmillan. Damico, A. J. (1982). The sociology of justice: Kohlberg and Milgram. Political Theory, 10, 409434. Dirsmith, M. W.(1983). “Obedience” in the classroom. Journal of Accounting Education, 1, 4150. Dominick, J. R. (1975). Review of Television and antisocial behavior: Field experiments. Journal of Broadcasting, 19, 114- 116. Duncan, S. (1977, December 19). Mental maps of New York. New York. pp. 51-62. Eastman, A. M. et al. (Eds.). (1988). The Norton reader: An anthology of exposiroryprose (7th ed.). New York: Norton. Elms, A. C. (1982). Keeping deception honest: Justifying conditions for social scientific research stratagems. In T. L. Beauchamp, R. R. Faden, R. J. Wallace, Jr., & L. Walters (Eds.), Ethical issues in social science research (pp. 232-245). Baltimore, MD: Johns Hopkins University Press. Elms, A. C., & Milgram, S. (1966). Personality characteristics associated with obedience and defiance toward authoritative command. Journal of Experimental Research in Personality, 1, 282-289. Erickson, B. H. (1978). Some problems of inference from chain data. In K. F. Schuessler (Ed.), Sociological methodology I979 (pp. 276-302). San Francisco: Jossey-Bass. Erickson, M. (1968). The inhumanity of ordinary people. International Journal of Psychiatry, 6, 277-279. Errera, P. (1972). Statement based on interviews with forty “worst cases” in the Milgram obedience experiments. In J. Katz (Ed.), Experimentation wirh human beings ( p. 400). New York: Russell Sage. Etzioni, A. (1968). A model of significant research. International Journal of Psychiatry, 6,279-280. Evans, G . W.(1980). Environmental cognition. Psychological Bulletin, 88, 259-287. Evans, R. 1. (1976). [Interview with] Stanley Milgram. In R. I. Evans (Ed.), The making of psychology: Discussions with creative contributors ( pp. 346-356). New York: Knopf. Exhibition explores social psychology research. (199 I , September/October). Psychological Science Agenda, p. 16. Eysenck, H. J. (1974, June 8). Doing as they were told [Review of Obedience to authority: An experimental view]. The Guardian, pp. 21-22. Feldman, R. S., & Scheibe, K. E. (1972). Determinants of dissent in a psychological experiment. Journal of Personality, 40,331-348. Feshbach, S. (1975). Television and the null hypothesis [Review of Television and antisocial behavior: Field experiments]. Contemporary Psychology, 20, 2 12-2 13. Fischer, C. S. (1984). The urban experience (2nd ed.). Orlando. FL: Harcourt Brace Jovanovich. Fogelman, E. (1987). The rescuers: A sociopsychological study of altruistic behavior during the Nazi era. Doctoral dissertation, City University of New York. (University Microfilms International No. 8801711) Freedman, J. (1969). Role playing: Psychology by consensus. Journal of Personality and Social Psychology. 13, 595-604.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

323

Freedman, J. L. (1988). Television violence and aggression: What the evidence shows. In S. Oskamp (Ed.), Applied social psychology annual (Vol. 8, pp. 144-162). Newbury Park, CA: Sage. Gaertner, S., & Bickman, L. (1971). Effects of race on the elicitation of helping behavior: The wrong number technique. Journal of Personality and Social Psychology, 20, 2 18-222. Geen, R . G., & Thomas, S. L. (1986).The immediate effects of media violence on behavior. Journal of Social Issues, 42(3), 7-27. Geller, D. M. (1975). A role-playing simulation of obedience: Focus on involvement. Doctoral dissertation, City University of New York. (University Microfilms No. 76-276) Geller, D. M. (1978). Involvement in role-playing simulations: A demonstration with studies on obedience. Journal of Personality and Social Psychology, 36, 2 19-235. Geller, D. M. (1980). Responses to urban stimuli: A balanced approach. Journal of Social Issues, 36(3), 86-100. Gergen, K. J. (1973). Social psychology as history. Journal of Personality and Social Psychology, 26, 309-320. Gibson, J. T. (1991). Training people to inflict pain: State terror and social learning. Journal of Humanistic Psychology, 31, 72-87. Giffin, B. (1983, November 26). Are the American people living in Jonestown? Los Angeles Times, Section 11, p. 2. Goffman, E. (1959). The presentation of serf in everydav life. Garden City, NY: Doubleday. Goldstein, E. V. (1981). The mediation of television messages by personal inj7uence: An experimental paradigm. Doctoral dissertation, City University of New York (University Microfilms International No. 81 12357) Golledge, R. G. (1987). Environmental cognition. In D. Stokols & I. Altman (Eds.), Handbook of environmental psychology ( pp. 131- 174). New York: Wiley. Gorenflo, D. W., & McConnell, J. V. (1991). The most frequently cited journal articles and authors in introductory psychology textbooks. Teaching of Psychology, 18, 8- 12. Greenwood, J. D. (1983). Role-playing as an experimental strategy in social psychology. European Journal of Social Psychology, 13, 235-254. Guiot, J. M. (1975). A modification of Milgram’s small world method. European Journal of Social Psychology, 6, 503-507. Haritos-Fatouros, M. (1988). The official torturer: A learning model for obedience to the authority of violence. Journal of Applied Social Psvchology. 18, I 107- 1120. Hams, B. (1988). Key words: A history of debriefing in social psychology. In J. G. Morawski (Ed.), The rise of experimentation in American psvchology (pp. 188-212). New Haven, CT: Yale University Press. Helm, C., & Morelli, M. (1979). Stanley Milgram and the obedience experiment: Authority, legitimacy, and human action. Political Theory, 7 , 321-345. Helm, C., & Morelli, M. (1985). Obedience to authority in a laboratory setting: Generalizability and context dependency. Political Studies. 33, 610-627. Herbert, W. (1985, March). Cyranoids: Artificial selves. Psychology Today, pp. 6-7. Herrnstein, R. J. (1974, June). Measuring evil [Review of Obedience to authority]. Commentary, 57, 82, 84, 86-88. Hirsch, H. (1988). Why people kill: Conditions for participation in mass murder. In H. Hirsch & J. D. Spiro (Eds.), Persistent prejudice: Perspectives on anti-Semitism (pp. 43-62). Fairfax, VA: George Mason University Press. Holland, C. D. (1967). Sources of variance in the experimental investigation of behuvioral obedience. Doctoral dissertation, University of Connecticut, S t o m . (University Microfilms No. 69-2146). Huesmann, L. R. (1982). Television violence and aggressive behavior. In D. Pearl, L. Bouthilet. & J. Lazar (Eds.), Television and behavior: Ten years of scientific progress and implications for the eighties: Vol2. Technical reviews (pp. 126-137). Rockville, MD: NIMH.

324

THOMAS BLASS

Hunt, S. M. (1979). Hypnosis as obedience behavior. British Journal of Social and Clinical Psychology, 18, 21-27. Kadushin, C. (1989). The small world method and other innovations in experimental social psychology. In M. Kochen (Ed.), The small world (pp. xxiii-xxvi). Norwood, NJ: Ablex. Kaplan, R. M., & Singer, R. D. (1976). Television violence and viewer aggression: A reexamination of the evidence. Journal of Social Issues, 32(4), 35-70. Karnow, S. (1971, March 30). Calley, Eichmann, both obedient. St. Petersburg Times, p. 12A. Karp, D. A,. Stone, G. P., & Yoels, W. C. (1977). Being urban: A socialpsychological view of city life. Lexington, MA: Heath. Kaufmann, H. (1967). The price of obedience and the price of knowledge [Comment]. American Psychologist, 22, 321-322. Kaufmann, H., & Kooman, A. (1967). Predicted compliance in obedience situations as a function of implied instructional variables. Psychonomic Science, 7 , 205-206. Kelman, H. C. (1967). Human use of human subjects: The problem of deception in social psychological experiments. Psychological Bulletin, 67, 1-1 1. Kelman, H. C., & Hamilton, V. L. (1989). Crimes of obedience: Toward a social psychology of authority and responsibility. New Haven, CT: Yale University Press. Kilham, W., & Mann, L. (1974). Level of destructive obedience as a function of transmitter and executant roles in the Milgram obedience paradigm. Journal of Personality and Social Psychology, 29, 696-702. Killworth, P. D., & Bernard, H. R. (1978/1979). The reverse small-world experiment. Social Networks. 1, 159-192. Klovdahl, A. S. (1989). Urban social networks: Some methodological problems and possibilities. In M. Kochen (Ed.), The small world (pp. 177-210). Norwood, NJ: Ablex. Kochen, M. (Ed.). (1989). The small world. Norwood, NJ: Ablex. Kohlberg, L. (1974). A moral experiment is not ethical unless educative. Phi Delta Kappan, 55,607, 644. Korte, C. (1990, August 13). The urban overload concept. In T. Blass (Chair), Perspectives on Stanley Milgram’s contributions to social psychology. Symposium conducted at the annual meeting of the American Psychological Association, Boston. Korte, C., & Milgram, S. (1970). Acquaintance networks between racial groups: Application of the small world method. Journal of Personalify and Social Psychology, 15, 101-108. Korte, C., Ypma, I., & Toppen, A. (1975). Helpfulness in Dutch society as a function of urbanization and environmental input level. Journal of Personality and Social Psychology, 32, 9961003. Krauthammer, C. (1985, March 29). Experiment in Jamestown. The Washingron Post, p. A23. Kremer, J., Bany, R., & McNally, A. (1986). The misdirected letter and the quasi-questionnaire: Unobtrusive measures of prejudice in Northern Ireland. Journal of Applied Social Psychology, 16, 303-309. Kudirka, N. Z. (1965). Defiance of authority under peer influence. Doctoral dissertation, Yale University, New Haven, CT. (University Microfilms No. 65-12, 969) Kunz, P. R., & Fernquist, R. M. (1989). Opinions on abortion as measured by the lost-letter technique. Psychological Reporrs, 65, 1343- 1346. Laurent, J. (1987). Milgram’s shocking experiments: A case in the social construction of “science.” Indian Journal of History of Science, 22, 247-272. Liebert, R. M., Spratkin, J. N., & Davidson, E. S. (1982). The early window; Eflects of television on children and youth (2nd ed.). New York: Pergamon. Little, B. R. (1987). Personality and the environment. In D. Stokols & 1. Altman (Eds.), Handbook of environmental psychology (pp. 205-244). New York: Wiley. Lynch, K. (1960). The image of the city. Cambridge, MA: M.I.T. Press.

SOCIAL PSYCHOLOGY O F STANLEY MILGRAM

325

Mann, L. (1981). The baiting crowd in episodes of threatened suicide. Journal of Personality and Social Psychology, 41, 703-709. Mantell, D. M. (1971). The potential for violence in Germany. Journal of Social Issues. 27(4), 101112. Mantell, D. M., & Panzarella, R. (1976). Obedience and responsibility. British Journal of Social and Clinical Psychology, 15, 239-245. Marsden, P. V. (1990). Network data and measurement. Annual Review of Sociology, 16, 435-463. Masseman, J. (1968). Debatable conclusions. International Journal of Psychiatry. 6, 28 1-282. Meeus, W. H. J., & Raaijmakers, Q. A. W. (1987). Administrative obedience as a social phenomenon. In w. Doise & S. Moscovici (Eds.), Current issues in European social psychology (Vol. 2, pp. 183-230). Cambridge: Cambridge University Press. Mental maps of a city (1975, November 10). New York. pp. 49-5 I . Menitt, C. B., & Fowler, R. G. (1948). The pecuniary honesty of the public at large. Journal of Abnormal and Social Psychology, 43, 90-93. Meyer, P. (1970, February). If Hitler asked you to electrocute a stranger, would you? Probably. Esquire, pp. 73, 128, 130, 132. Milgram, S. (1960). conformity in Norway and France: An experimental study of national charmreristics. Doctoral dissertation, Harvard University, Cambridge, MA. Milgram, S. (1961, December). Nationality and conformity. Scientific American, pp. 45-51. Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology. 67, 371-378. Milgram, S . (1964a). Group pressure and action against a person. Journal of Abnormal and Social Psychology. 69, 137-143. Milgram, S. (1964b). Issues in the study of obedience: A reply to Baumrind. American Psychologist, 19, 848-852. Milgram, S. (1964~).Technique and first findings of a laboratory study of obedience to authority. Yale Science Magazine, 39, 9-1 1, 14. Milgram, S. (1965a). Liberating effects of group pressure. Journal of Personaliry and Social Psychology, 1, 127-134. Milgram, S. (1965b). Obedience (a filmed experiment). Distributed by the New York University Film Library Milgram, S. (1965~).Some conditions of obedience and disobedience to authority. Human Relations, 18, 57-76. Milgram, S. (1967a). Obedience to criminal orders: The compulsion to do evil. Patterns of Prejudice, 1, 3-7. Milgram, S. (1967b, May). The small-world problem. Psychology Today. 1, 60-67. Milgram, S. (1968). Reply to the critics. International Journal of Psychiarry. 6, 294-295. Milgram, S. (1969a). Comment on "A failure to validate the lost letter technique." Public Opinion Quarterly. 33, 263-264. Milgram, S. (1969b). Interdisciplinary thinking and the small world problem. In M. Sherif & C. W. Sherif (Eds.), Interdisciplinary relationships in the social sciences (pp. 103- 120). Chicago: Aldine. Milgram, S. ( 1 9 6 9 ~June). The lost-letter technique. Psychology Today, pp. 30-33, 66, 68. Milgram, S. (1970a). The experience of living in cities. Science. 167, 1461-1468. Milgram, S. (1970b). The experience of living in cities: A psychological analysis. In F. F. Korten, S. W. Cook, & J. 1. Lacey (Eds.), Psychology and the problem of sociery. Washington, DC: American Psychological Association. Milgram, S. (1972). Interpreting obedience: Error and evidence (A reply to Ome and Holland). In A. G. Miller (Ed.), The social psychology of psychological research (pp. 138-154). New York: Free Press.

326

THOMAS BLASS

Milgram, S. (1973a. December). The perils of obedience. Harper’s, pp. 62-66, 75-77. Milgram, S. (1973b). Responses I and 11. In D. Abse, The dogs of Pavlov (pp. 37-44, 125-127). London: Valentine, Mitchell & Co. Milgram, S. (1974). Obedience to authority: An experimental view. New York: Harper & Row. Milgram, S. (Ed.). (1975). Psychology in today’s world. Boston: Educational Associates. Milgram, S. (1976a. June 3). The image freezing machine. New Society, pp. 519-522. Milgram, S. (1976b, NovemberDecember). The image-freezing machine. Society, pp. 7-12. Milgram, S. ( 1 9 7 6 ~August 21). Obedience to authority: A CBS drama deals with the shocking results of a social psychologist’s experiments. TV Guide, pp. 24-25. Milgram, S. (1977a, January). City families: Britons and Americans talk about their images, frozen forever on film. Psychology Today, pp. 59-63, 65. Milgram, S. (1977b). The familiar stranger: An aspect of urban anonymity. In S. Milgram, The individual in a social world: Essays and experiments (pp. 5 1-53). Reading, MA: AddisonWesQ Milgram, S. (1977~).The image freezing machine. In S. Milgram, The individual in a social world: Essays and experiments (pp. 339-350). Reading, MA: Addison-Wesley. Milgram, S. (1977d, January). The image-freezing machine. Psychology Today, pp. 50,52,54, 108. Milgram, S. (1977e). The individual in a social world: Essays and experiments. Reading, MA: Addison-Wesley. Milgram, S. (1977f). Reflections on the news. Anrioch Review, 35, 167-170. Milgram, S . (1977g). The social meaning of fanaticism. Ercetera, 34, 58-61. Milgram, S. (1977h). Subject reaction: The neglected factor in the ethics of experimentation. Hastings Center Report. 7 , 19-23. Milgram, S. (1978). A work of great potential influence [Review of Social influence and social change]. Contemporary Psychology, 23, 129- 131. Milgram, S. (1979). PrBface a la deuxieme Bdition franGaise [Preface to the second French edition]. Soumission d I’aurorire‘: Un point de vue expkrimental. Paris: Calmann-LBvy. Milgram, S. (1981a, March 2). This week’s citation classic: Milgram, S. Behavioral study of obedience, Current Contents, Social and Behavioral Sciences, p. 18. Milgram, S. (1981b, December 14). This week’s citation classic: Milgram, S. The experience of living in cities. Current Contents, Social and Behavioral Sciences, p. 14. Milgram, S. (1982, May). [Contribution to] Understanding psychological man: A state-of-thescience report. Psychology Today, pp. 49-51. Milgram, S. (1983). Reflections on Morelli’s “Dilemma of obedience.” Metaphilosophy. 14, 190194. Milgram, S. (1984a). Cities as social representations. In R. M. Farr & S. Moscovici (Eds.), Social representations (pp. 289-309). Cambridge & Paris: Cambridge University Press & Editions de la Maison des Sciences de 1’Homme. Milgram, S. (1984b, August 26). Cyranoids. Talk delivered at the annual convention of the American Psychological Association, Toronto, Canada. Milgram, S. (1984~.October). Network love. Omni, 7 , pp. 34, 148. Milgram, S. (1984d). Obedience. In R. J. Corsini (Ed.), Encyclopedia of psychology (Vol. 2, pp. 446-447). New York: Wiley. Milgram, S. (1987). Obedience. In R. J. Corsini (Ed.), Concise encyclopedia of psychology (pp. 773-774). New York: Wiley. Milgram, S., Bickman, L., & Berkowitz, L. (1969). Note on the drawing power of crowds of different size. Journal of Personality and Social Psychology, 13, 79-82. Milgram, S., Greenwaid, J., Kessler, S., McKenna, W., & Waters, J. (1972, March-April). A psychological map of New York City. American Scientist, pp. 194-200. Milgram, S., & Hollander, P. (1964, June 15). The murder they heard. The Nation. pp. 602-604. Milgram, S . , & Jodelet, D. (1976). Psychological maps of Paris. In H. M. Proshansky, W. H.

SOCIAL PSYCH0UK;Y OF STANLEY MILGRAM

327

Ittelson, & L. G . Rivlin (Eds.), Environmental psychology: People and their physical settings (2nd ed., pp. 104-124). New York Holt, Rinehart, & Winston. Milgram, S., & Jodelet, D. (1977, November 3). The way Parisians see Paris. New Society, pp. 234237. Milgram, S., Liberty, H. J., Toledo, R., & Wackenhut, J. (1986). Response to intrusion into waiting lines. Journal of Personality and Social Psychology, 51, 683-689. Milgram, S . , Mann, L., & Harter, S. (1965). The lost-letter technique: A tool of social research. Public Opinion Quarterly, 29, 437-438. Milgram, S., & Martin, A. I. (1985). The role of the self in biological and cultural systems: A response to Csikszentmihalyi and Massimini. New ldeas in Psychology, 3, 139-142. Milgram, S . , & Sabini, J. (1978). On maintaining urban norms: A field experiment in the subway. In A. Baum. J. E. Singer, & S. Valins (Eds.), Advances in environmental psychology (Vol. 1, pp. 31-40). Hillsdale, NJ: Erlbaum. Milgram, S., & Sabini, J. (1979). Candid Camera. Society, 16, 72-75. Milgram, S., & Shotland, R. L. (1973). Television and antisocial behavior: Field experiments. New York: Academic Press. Milgram, S., & Toch, H. (1969). Collective behavior: Crowds and social movements. In G . Lindzey & E. Aronson (Eds.), The handbook of social psychology (2nd ed., Vol. 4, pp. 507-610). Reading, MA: Addison-Wesley. Miller, A. G. (1986). The obedience experiments: A case study of controversy in social science. New York Praeger. Miller, A. G., Gillen, B., Schenker, C., & Radlove, S. (1974). The prediction and perception of obedience to authority. Journal of Personality, 42, 23-42. Miranda, F. S. B., Cabellero, R. B., Gomez, M. N. G., & Zamorano, M. A. M. (1981). Obediencia a la autoridad [Obedience to authority]. Psiquis. 2, 212-221. Mixon, D. (1971). Further conditions of obedience and disobedience to authority. Doctoral dissertation, University of Nevada, Reno. (University Microfilms No. 72-6477) Mixon, D. ( 1976). Studying feignable behavior. Representative Research in Social Psychology, 7 , 89- 104. Mixon, D . (1989). Obedience and civilization: Authorized crime and the normality of evil. London, England: Pluto Press. Muson, H. (1978, January). Blind obedience. Psychology Today, pp. 12, 112. Nissani, M. (1990). A cognitive reinterpretation of Stanley Milgram’s observations on obedience to authority [Comment]. American Psychologist, 45, 1384- 1385. O’Leary,C., Willis, F. N., & Tomich, E. (1970). Conformity under deceptive and non-deceptive techniques. Sociological Quarterly. 11, 87-93. Ome, M. T., & Holland, C. H. (1968). On the ecological validity of laboratory deceptions. International Journal of Psychiatry, 6, 282-293. Patten, S. C . (1977). Milgram’s shocking experiments. Philosophy, 52, 425-440. Pattullo. E. L. (1984). Institutional review boards and social research: A disruptive, subjective perspective, retrospective and prospective. In J. E. Sieber (Ed.), NfH readings on the protection of human subjects in behavioral and social science research: Conference proceedings and background papers (pp. 10- 17). Frederick, MD: University Publications of America. Redfearn, D. J. ( 1982). Predictors of nurses’ compliance with physicians‘ inappropriate orders. Ann Arbor, MI: UMI Research Press. Ring, K., Wallston, K.,& Corey, M. (1970). Mode of debriefing as a factor affecting subjective reactions to a Milgram-type obedience experiment: An ethical inquiry. Representative Research in Social Psychology, 1, 67-85. Rogers, E. M., & Kincaid, D. L. (1981). Communication networks: Toward a new paradigm for research. New York: Free Press. Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution

328

THOMAS BLASS

process. In L. Berkowitz (Ed.), Advances in experimenral social psychology (Vol. 10, pp. 173219). New York: Academic Press. Ross, L. D. (1988). Situationist perspectives on the obedience experiments [Review of The obedience experiments: A case srudy of conrroversy in social science]. Coniemporary Psychology, 33, 101-104.

Sabini, J. P. (1976). Moral reproach: A conceptual and experimenral analysis. Doctoral dissertation, City University of New York. (University Microfilms No. 76-30,419). Sabini, J. P. (1986). Obituary: Stanley Milgram (1933-1984). American Psychologisr, 41, 13781379. Sabini, J. P., & Silver, M. (1980). Destroying the innocent with a clear conscience: A sociopsychology of the Holocaust. In J. E. Dimsdale (Ed.), Survivors, vicrims, and perpetrators: Essays on the Nazi Holocaust (pp. 329-358). Washington, DC: Hemisphere. Sabini, 1. P., & Silver, M. (1982). Moralities of everyday life. Oxford: Oxford University Press. Sabini, J. P.,& Silver, M. (1985). Critical thinking and obedience to authority. National Forum, 65, 13-17. Schurz, G. ( 1985). Experimentelle Uberpriifung des Zusammenhangs zwischen Personlichkeitsmerkmalen und der Bereitschaft zum destruktiven Gehorsam gegeniiber Autoritaten [Experimental examination of the relationships between personality characteristics and the readiness for destructive obedience toward authority]. Zeirschrifr fir Experimenrelle unpr Angewandre Psychologie. 32, 160- 177. Sechrest, L., & Belew, J. (1983). Nonreactive measures of social attitudes. In L. Bickman (Ed.), Applied social psychology annual (Vol. 4, pp. 23-64). Beverly Hills, CA: Sage. Shalala, S. R. (1974). A siudy of various communication settings which produce obedience by subordinates to unlawful superior orders. Doctoral dissertation, University of Kansas, Lawrence. (University Microfilms No. 75-17,675) Shanab, M. E., & Yahya, K. A. (1977). A behavioral study of obedience in children. Journal of Personality and Social Psychology, 35, 530-536. Shanab, M. E., & Yahya, K . A. (1978). A cross-cultural study of obedience. Bulletin of the Psychonomic Society, 11, 267-269. Shelton, G. A. (1982). The generalizarion of undersranding to behavior: The role of perspective in enlighrenment. Unpublished doctoral thesis, University of British Columbia. Shnayerson, R. (1973, December). Conscience and Congress: The dilemmas of obedience. Harper’s Magazine, p. 61. Shotland, R. L. (1976). University communication networks: The small world method. New York: Wiley. Silver, M. (1977). The social consrruction of envy. Doctoral dissertation, City University of New York. (University Microfilms No. 77-20,498) Simmel, G. (1950). The metropolis and mental life. In K. H. Wolff (Ed. and Trans.), The sociology of Georg Simmel. New York: Free Press. (Original work published 1903) Singer, S. J. (1991, July 17). Obey-but at what cost? The Baltimore Sun, p. 15A. Staub. E. (1989). The roots of evil: The origins of genocide and other group violence. New York: Cambridge University Press. Strom, M. S., & Parsons, W. S. (1978). Facing hisrory and ourselves: Holocaust and human behavior (rev. ed., Vol. 3). Brookline, MA: Authors. Stubbs, M., & Barnet, S. (Eds.). (1983). The Little, Brown reader (3rd ed.). Boston: Little, Brown. Sullivan, W. (1963, October 26). 65 percent in test blindly obey order to inflict pain. The New York Times, p. 10. Takooshian, H., Haber, S., & Lucido, D. J. (1977, February). Who wouldn’t help a lost child? You, maybe. Psychology Today, 67-68, 88. Tavris, C. (1974a, June). The frozen world of the familiar stranger [Interview with Stanley Milgram]. Psychology Today, pp. 71-73, 76-78, 80.

SOCIAL PSYCHOLOGY OF STANLEY MILGRAM

329

Tavris, C. (1974b, June). A sketch of Stanley Milgram: A man of I .OOO ideas. Psychology Today, pp. 74-75.

Teaching about the Holocaust and genocide: Introduction. The Human Rights Series: Vol. I . (1985). Albany: University of the State of New York, State Education Department. Tilker, H. A. (1970). Socially responsible behavior as a function of observer responsibility and victim feedback. Journal of Personality and Social Psychology, 14, 95- 100. Travers, J., & Milgram, S. (1969). An experimental study of the small world problem. Sociometry, 32, 425-443. S l e r , T. R., & Devinitz, V. (1981). Self-serving bias in attribution of responsibility: Cognitive versus motivational explanations. Journal of Experimental Social Psychology, 17, 408-41 6. Ullrnan, D., & Jackson, T. T. (1982). Researchers’ ethical conscience: Debriefing from 1960 to 1980 [Commenl]. American Psychologisf, 37, 972-973. Webb, E. J., Campbell, D. T., Schwartz, R. D., Sechrest, L., & Grove, J. B. (1981). Nonreactive measures in the social sciences (2nd ed.). Boston: Houghton-Mifflin. Wirth, L. (1938). Urbanism as a way of life. American Journal of Sociology. 44, 1-24. Wrightsman, L. S. (1974). The most important social psychological research in this generation? [Review of Obedience to authority: An experimental view]. Confemporaty Psychology, 19, 803-805.

This Page Intentionally Left Blank

THE IMPACT OF ACCOUNTABILITY ON JUDGMENT AND CHOICE: TOWARD A SOCIAL CONTINGENCY MODEL Philip E. Tetlock

I. Introduction The study of judgment and choice has been an academic growth industry for a quarter of a century now. The magnitude of the research effort is reflected in the reference lists of recent literature reviews. Since 1970, there have been well over 3000 experimental studies of social cognition and causal attribution (Fiske & Taylor, 1991; Hastie, 1983; Higgins & Bargh, 1987; Markus & Zajonc, 1985; Nisbett & Ross, 1980; Showers & Cantor, 1985). There has been an even greater surge of interest in behavioral decision theory, with offshoots in such diverse fields as medical diagnosis, marketing, accounting, finance, meteorology, and the law (Abelson & Levi, 1985; Einhom & Hogarth, 1981; Hogarth, 1990; Kahneman, Tversky, & Slovic, 1982; Payne, Bettman, & Johnson, in press). Although most research takes the form of experiments on cognitive processes, it is possible to explore judgment and choice from a variety of methodological and theoretical vantage points. Many social psychologists focus on interpersonal and small-group processes (Bales, 1970; Janis, 1982, 1989; Moscovici, 1985). Most important decisions are not the product of isolated information processors; they are the product of intensive interactions among members of groups. Working at a still higher level of analysis, many investigators in organizational behavior, political science, and economics have preferred a macro or social-system perspective on decision-making. Both individuals and small groups are constrained by institutional norms and procedures (Allison, 1971; Katz & Kahn, 1978; Lindblom, 1959; March & Olsen, 1989; March & Simon, 1958; Pfeffer, 198I). Investigators in this tradition often view experimental studies with skepticism; they recommend studying decision-making in the actual institutional settings in which it occurs. One’s choice of level of analysis has profound consequences for how one ADVANCES IN EXPtRlMENTAL SOCIAL PSYCHOLIXIY. VOL 25

33 I

Copyright 0 1992 by Academic h s s . Inc. All rights of repduclion in MY form reserved.

332

PHILIP E. TETLOCK

thinks about thought. Drawing on general systems theory, I have elsewhere compared the different levels of analysis to looking through a microscope at different magnifications (Tetlock, 1989a). At the most intense magnificationsuch as provided by laboratory experiments on cognitive processes-one can observe the phenomenon in detail. The price is, however, the inability to view the phenomenon within a broader systems context. At less intense magnification-such as provided by field studies of organizational decision-making-one can see context but only at the cost of slighting subsystem detail. Not surprisingly, communication is difficult. What investigators find exciting at one level is invisible to investigators working at other levels. One can study decision-making at a purely cognitive level without ever referring to research on group dynamics, role theory, or bureaucratic politics. Conversely, one can study bargaining, coalition formation, and incrementalism in organizations without ever referring to cognitive research on knowledge structures and judgmental heuristics. The most general goal of this article is to encourage communication across levels of analysis. Despite impressive advances in social cognition, group dynamics, and organizational behavior, the linkages among these traditions are still primitive. A major challenge confronting judgment and choice theorists is the creation of conceptual frameworks that bridge these traditionally separate levels of analysis (cf. George, 1980a, 1980b; Staw, Sandelands, & Dutton, 1981). The analysis that follows is divided into four sections. The first section begins with a quick overview of experimental cognitive research on judgment and choice. I organize my discussion around the notion that empirical work is ultimately grounded in “hard-core” assumptions about the subject matter. Two hardcore assumptions have proved particularly fruitful in the study of judgment and choice-the view of the person as intuitive psychologist and as intuitive economist. Working from these metaphorical foundations, investigators have articulated and refined a wide range of testable middle-range theories that, in turn, have stimulated a wide range of data collection efforts. The obvious successes of these research programs should not, however, blind us to their equally obvious shortcomings. The psychologist and economist metaphors are like beacons: They highlight some aspects of judgment and choice, but leave equally important aspects in the dark. The second section proposes an alternative and largely complementary metaphor for studying judgment and choice: the politician whose primary goal is to maintain the positive regard of important constituencies to whom he or she feels accountable. Working from this premise, I propose a social contingency model that identifies (1) three major strategies that people use to cope with demands for accountability from significant others in their lives (the acceptability heuristic, preemptive self-criticism, and defensive bolstering) and (2) personality and situational moderators of these coping strategies. I review the major predictions of

ACCOUNTABILITY

333

the model and the growing body of experimental and field research relevant to those predictions. The third section explores the normative and policy implications of the social contingency model. Although accountability has been presented as a panacea for numerous social ills (schools that do not teach, overstaffed middle management, unresponsive government bureaucracies, etc.), efforts to improve performance by enhancing accountability often produce perverse effects. The social contingency model suggests some reasons why such efforts backfire. I attempt to identify the conditions under which each of the three major strategies of coping with accountability is likely to prove adaptive or maladaptive from either an individual or institutional point of view. The final section explores the concept of error and bias in judgment from a broad functionalist perspective that takes into account the roles that people play in social and political systems. Many response tendencies that look flawed within a narrowly cognitive or microeconomic perspective can be viewed as prudent responses to the social world within which people live and work. I question the normative status of several hypothesized errors or biases within the cognitive research program and suggest possible interactions between cognitive and social contextual variables.

11. The Metaphorical Roots of the Cognitive

Research Program In this postpositivist era, it is no longer heretical-indeed, it is almost obligatory-to observe that initial assumptions about human nature influence how we design, execute, and interpret research. Lakatos’s (1970) concept of the research program offers a useful framework for thinking about the influence of these starting assumptions (see also Suppe, 1973). He argues that ‘underlying all research programs are hard-core premises that alternately inspire and inhibit theoretical creativity. Hard-core premises inspire creativity by highlighting phenomena that otherwise would have been neglected and by pointing to ways of articulating middle-range theories that are sufficiently specific and falsifiable to guide actual empirical work. Lakatos calls this function the “positive heuristic” of a research program. Hard-core assumptions inhibit creativity by limiting the range of research questions that investigators pose and by encouraging a rigid, defensive attitude toward research findings that challenge hard-core assumptions. Lakatos calls this function the “negative heuristic” of a research program. The negative heuristic directs researchers to develop a protective belt of auxiliary hypotheses that shelters the hard core from refutation.

334

PHILIP E. TETLOCK

Although the research literatures on judgment and choice are voluminous and in some cases forbiddingly technical, these literatures have not outgrown their roots in hard-core assumptions. Implicitly or explicitly, researchers still make assumptions about the underlying goals or functions of judgment and choice. One influential hard-core metaphor has been that of the social perceiver as intuitive psychologist who strives for “cognitive mastery of the causal structure of the environment” (Kelley, 1967, p. 193). This metaphor contains the seeds of an ambitious research agenda as well as a set of normative criteria for evaluating judgment and choice. The key question becomes: What kind of intuitive psychologist is the average perceiver? How good a job does the intuitive psychologist do in forming hypotheses about the causes of behavior, testing hypotheses dispassionately against the evidence, and abandoning favorite hypotheses in the face of contradictory evidence? The psychologist metaphor should not, of course, be confused with the theories derived from it. It is possible to spin out many testable middle-range theories of people’s competence as intuitive psychologists. The most optimistic middle-range theorists depict people who rely on lay versions of formal logical and statistical procedures to arrive at conclusions (Azjen & Fishbein, 1975; Jones & Davis, 1965; Jones & McGillis, 1976; Kelley, 1967, 1971). The most pessimistic theories depict people as “cognitive misers” who are prone to a variety of judgmental failings (Fiske & Taylor, 1991; Nisbett & Ross, 1980). There are also middle-range theories near the midpoint of the optimismpessimism continuum that emphasize people’s capacity to shift from simpler to more complex cognitive strategies in response to situational demands (Chaiken, 1980; Fiske & Neuberg, 1990; Kruglanski, 1990; Kunda, 1990; Langer, 1990; Maheswaran & Chaiken, 1991; Petty & Cacioppo, 1986). Another metaphor has profoundly shaped the development of the decisionmaking literature: that of the intuitive economist whose primary goal in life is the maximization of subjective expected utility. Once again, the hard-core metaphor highlights an ambitious research agenda as well as a set of normative criteria for evaluating the quality of judgment and choice. The key question becomes: How effective or ineffective are people as intuitive economists? Working from this metaphorical premise, it is again possible to articulate and refine many testable middle-range theories of the intuitive economist-some more optimistic and some more pessimistic (Kahneman, Knetch, & Thaler, 1986; Kahneman & Tversky, 1979; Thaler, 1985). The optimists downplay the frequency and severity of violations of the normative axioms of rational choice theory (sometimes even by reformulating the axioms); the pessimists stress the robustness of the violations, even when people are making repeated, high-stakes decisions in competitive market environments. The preponderance of the evidence currently favors a moderately pessimistic assessment of our skills as both intuitive psychologists and economist. As intuitive psychologists, we appear to be too quick to draw strong conclusions about

ACCOUNTABILITY

335

others’ personalities from fragmentary and unrepresentative evidence (Jones, 1979; Ross, 1977), too slow to revise our hypotheses in response to new evidence (Nisbett & Ross, 1980), too confident in the correctness of our judgments in many settings (Fischhoff, 1982), and too quick to loose faith in genuinely diagnostic evidence when that evidence is embedded in irrelevant or distractor variables (Nisbett, Zukier, & Lemley, 1981; Zukier, 1982). Our record as intuitive economists is hardly more impressive. We try to avoid the painful value tradeoffs that expected utility maximization demands we confront. We give far too much weight to considerations that normative theory says we should ignore. Far from ignoring sunk costs, for example, we often escalate our commitments to failing policies (Arkes & Blumer, 1985; Brockner & Rubin, 1985; Staw, 1980). And we often ignore variables to which normative theories say we should attend. In many situations, people are as oblivious to opportunity costs as they are attentive to sunk costs. There is also substantial convergence in the theoretical explanations advanced for our shortcomings as intuitive psychologists and economists. People, it is widely agreed, are limited-capacity information processors who rely on a variety of inferential shortcuts to understand an otherwise impossibly complex environment and to make otherwise hopelessly difficult decisions. We rely on the availability and representativeness heuristics to make inferences about causality and probability, we rely on noncompensatory choice rules to escape dissonant tradeoffs, and we rely on our preconceptions (“existing knowledge structures”) in interpreting new evidence. The price of cognitive economy, in this view, is increased susceptibility to error and bias.

111. The Political Research Program

The psychologist and economist metaphors have stimulated important discoveries. Both research programs are, however, incomplete. In the pursuit of highly abstract laws of information processing, the psychologist and economist research programs have largely ignored the social settings in which people make most decisions. Subjects in laboratory studies of cognitive processes rarely feel accountable to others for the positions they take. They function in a social vacuum (or as close an approximation to a social vacuum as can be achieved) in which they do not need to worry about the interpersonal consequences of their conduct (“How will others react if I do this? How effectively can I justify my views if challenged?”). Such issues are simply seen as irrelevant to the explanatory goals of the research program. Such issues are absolutely central, however, to a research program grounded in the politician metaphor. The politician research program begins where the cog-

336

PHILIP E. TETLOCK

nitive research programs leave off. The starting point for analysis is the information processor-with whatever cognitive limitations he or she may possess-in an environment structured by social and organizational systems to which the individual belongs and with which the individual may identify. Whereas the central objective of the cognitive research programs is to identify fundamental laws of human thought, the central objective of the political research program is to identify the behavioral strategies that people have developed for coping with fundamental features of everyday decision environments-features likely to be present in some form in all social settings. Like the psychologist and economist research programs, the politician program rests on hard-core assumptions concerning its subject matter. And like the other metaphors, the politician metaphor contains the intellectual seeds of both a research agenda and a set of normative criteria for assessing the quality of judgment and choice. The key questions become: What kind of intuitive politician is the average person? How well do people do in placating the constituencies to whom they feel accountable? And what cognitive and social strategies do people use in coping with accountability demands?’

A. THE HARD CORE OF THE POLITICIAN RESEARCH PROGRAM

Given the crucial role that underlying assumptions play in guiding empirical work, especially in the early stages of a research program, these assumptions deserve to be spelled out in detail. The first assumption deals with the nature of

‘There are, of course, costs as well as benefits in shifting from one metaphorical framework to another in the study of judgment and choice. The major disadvantage is the opportunity cost of diverting talent from the pursuit of “normal science“ within an established metaphor to initiating a normal science tradition within a novel metaphorical framework. In this case, I would argue that the advantages substantially outweigh these costs. There are two major advantages: ( I ) The politician metaphor highlights psychological processes that other metaphors obscure. Many well-replicated judgmental biases interact in complex ways with accountability manipulations. Moreover, it is not just a simple matter of accountability making things better. Accountability can exacerbate bias, attenuate bias, or have no effect-depending on the type of accountability manipulation and the type of bias under investigation. (2) In addition to identifying boundary conditions on established effects, the politician research program offers us an alternative set of normative criteria for judging effects to be emors or biases. Effects that look like judgmental flaws from one metaphorical perspective often look quite prudent, even reasonable, within another perspective (these normatively controversial effects include the fundamental attribution error, the dilution bias, and ambiguity aversion). To be sure, one can take this line of argument too far and become an apologist for virtually every judgmental shortcoming documented in the literature. One can extend Quine’s “principle of charity” to the point of implausibility. My goal here is to strike a reasonable balance among alternative (metaphorically grounded) conceptions of rationality.

ACCOUNTABILITY

337

the social world in which people make up their minds; the second assumption deals with the goals and motives that drive the decision-making process. 1 . Assumption 1 :Accountability of Conduct as a Universal Feature of the Natural Decision Environment Attribution theorists have long noted that in everyday life people are presumed to be agents of their actions-they are responsible for what they do (cf. Heider, 1944; Jones & Davis, 1965). It makes sense to ask people for the reasons underlying their actions because it is assumed that people possess the power to monitor and control their conduct in accord with self-generated plans of action (another once heretical idea that is now widely accepted; Bandura, 1977; Schlenker, 1982). The accountability of conduct is a sociocultural adaptation to the problem of how to coordinate relationships among individuals who are capable of observing, commenting on, and controlling their own actions. As such, accountability is a key component of the solution to the primal sociological riddle of how society is possible (cf. Scott & Lyman, 1968). Organized social life cannot exist without some regularity. This regularity is provided by shared rules, norms, and social practices (Weick, 1979). Accountability is a critical rule and a norm enforcement mechanism-the social psychological link between individual decision-makers on the one hand and social systems on the other. Expectations of accountability are an implicit or explicit constraint on virtually everything people do (“If I do this, how will others react?”). Failure to act in ways for which one can construct acceptable accounts leads to varying degrees of censure, depending on the gravity of the offense and the norms of the society (cf. Pfeffer, 1981; Tetlock, 1985). Although one can make a powerful case for the universality of accountability (Semin & Manstead, 1983), the specific norms and values to which people are held accountable vary dramatically from one culture or time to another. As C. Wright Mills (1940) noted, when people leave groups and join new ones, they must learn new vocabularies of motives-new rules for generating acceptable explanations of behavior. These vocabularies of motives vary as a function of both the micro and the macro context. The micro context includes ideologies and values that characterize distinctive organizations within societies (e.g., IBM) and rules within organizations (e.g.. occupants of boundary roles, such as union negotiators, who must be responsive to conflicting constituencies). The macro context refers to cultural ideologies and values (e.g., science, Protestantism, capitalism, Marxism) and societal ones (e.g., nationalism). An important empirical task within the politician research program is systematic ethnographic work to characterize the normative beliefs and values that define the standards of accountability in particular decision-making settings (cf. Pepitone, 1976).

338

PHILIP E. TETLOCK

2 . Assumption 2: People as Approval and Status Seekers People seek approval and respect for many reasons, including both symbolic and tangible rewards and punishments. Theories of impression management and self-esteem maintenance have emphasized symbolic motives (e.g., Baumeister, 1982; Greenwald, 1980; Schlenker, 1982; Stryker & Gottlieb, 1981). The following motives are especially important here: 1. The motivation to protect and enhance one’s social image or identity. One of the most influential motivational assumptions in social science is that people seek approval and respect as ends in themselves. In the words of Linton (1945, p. 9): “The need for eliciting favorable responses from others is an almost constant component of personality. Indeed it is not too much to say that there is very little organized human behavior which is not directed toward its satisfaction in at least some degree.” Zetterberg (1957, p. 189) has even proposed that “the maximization of favorable attitudes from others is the counterpart in sociological theory to the maximization of profit in economic theory.” A warehouse of findings in experimental social psychology attests to the influence of this view of human nature, including work on ingratiation (Jones & Wortman, 1973), conformity (Hare, 1976), and strategic attitude shifts (Cialdini, Petty, & Cacioppo, 1981). 2. The motivation to protect and enhance one’s self-image. This assumption has an equally venerable status in social and personality psychology (Allport, 1937; Sherif & Cantril, 1947). People do not seek approval as an end in itself, but rather as a means of bolstering their own self-worth on important dimensions of evaluation (e.g., intelligent, conscientious, likable). There is a substantial literature on the ego-defensive tactics that people use to dissociate themselves from negative outcomes and to associate themselves with positive outcomes (Greenwald, 1980; Schlenker, 1982; Tetlock & Levi, 1982). 3. The motivation to acquire power and wealth. Social exchange theorists (e.g., Blau, 1964) have emphasized a third motive for seeking approval and respect: the desire to control desirable material resources. Researchers in organizational behavior have been most sensitive to this motivational dimension of human nature: Decision-makers in organizations can be fruitfully viewed as actors in competition for scarce resources within a rule-governed political contest for power (Pfeffer & Salancik, 1978). Much of what people in organizations do can be understood as tactical maneuvers to legitimize their claims on scarce resources. Any complete analysis of the decision-maker as politician should take account of all three of these sources of motivation. Unfortunately, we do not have a good understanding of the relative importance of these motives or of the conditions under which one versus another becomes dominant. Attempts to disentangle motives have, moreover, proved frustrating (Tetlock & Manstead, 1985). In most

ACCOUNTABILITY

339

situations, the three motives are closely intertwined; for example, impressing others will boost not only one’s social image but also one’s self-image (as advocates of the looking-glass self hypothesis would argue) and one’s material standing in the world (one’s social image largely determines one’s interpersonal market value; Blau, 1964). Conversely, improvements in one’s material standing will tend to enhance both self-image and social image (as observers of conspicuous consumption have long been aware; Veblen, 1889).

B. STRATEGIES FOR COPING WITH ACCOUNTABILITY The politician research program rests on two hard-core assumptions: The first posits that accountability of conduct is a universal problem of social life; the second posits that people seek the approval and respect of those to whom they are accountable. These hard-core postulates direct empirical work in radically different directions from the cognitive research programs. The central questions become: What type of politician is the average person? What strategies do people use to cope with accountability demands from important constituencies? What are the consequences of these strategies? Just as there are middle-range theories that depict people as more or less effective intuitive psychologists and economists, so one can easily imagine theories that depict people as more or less effective politicians. At one end of the effectiveness continuum would be theories that posit Machiavellian levels of political cunning. People, in this view, are incessant schemers who actively seek out information about the expectations and preferences of others, carefully calculate the impact of possible decisions on others, anticipate potential objections, and craft accounts to preempt these objections. At the other end of the effectiveness continuum would be theories that portray people as hopelessly inept politicians who may try to maintain good working relations with important constituencies, but who instead frequently wind up antagonizing them. Indeed, it may be reasonable to extrapolate from work within the economist and psychologist traditions that people often are poor politicians. Insofar as people fail on cognitive tasks critical for “good judgment” defined within a psychological or economic framework (revising beliefs and commitments, discounting for situational pressures, etc.), there is a good chance they will also fail on cognitive tasks critical for skillful political maneuvering. The social contingency model developed here falls somewhere near the midpoint of the effectiveness continuum. The model is an eclectic creation that borrows assumptions from a number of theoretical traditions. The model posits, for instance, that the cognitive miser image of the social thinker provides a useful first approximation for predicting how people cope with accountability predica-

340

PHILIP E. TETLOCK

ments. All other things being equal, people prefer least-effort solutions. They simply adopt positions likely to gain the favor of those to whom they feel accountable (a coping strategy labeled here as the acceptability heuristic). The social contingency model also posits, however, that the solutions to accountability predicaments are not always so straightforward. In some situations, it is not obvious what the most acceptable option is. The model predicts that, when people do not know the views of the prospective audience and are under no pressure to justify past actions, accountability will motivate people to abandon their cognitive-miserly ways and to become relatively flexible, self-critical, and multidimensional thinkers. Finally, the model identifies a third major coping strategy. When people have irrevocably committed themselves to a course of action, accountability will again motivate cognitive effort. The result will not, though, be self-critical, flexible, and complex thought. Rather, the result will be rigid, defensive, and evaluatively consistent thought. Accountability will prompt people to generate as many reasons as they can why they are right and potential critics wrong. We should be careful, however, in making normative judgments of these three coping strategies. Many scholars emphasize the adaptiveness of preemptive selfcriticism and the maladaptiveness of the acceptability heuristic and of defensive bolstering (e.g., Janis, 1989). From the standpoint of the social contingency model, however, it is possible to make strong normative cases for and against each coping strategy. The theoretical challenge is identifying the conditions under which each strategy is likely to prove more or less adaptive-a challenge to which we shall return after reviewing the activating conditions for each coping strategy.

C. THE ACCEPTIBILITY HEURISTIC: THE COGNITIVE MISER IN SOCIAL CONTEXT The simplest coping strategy is to make decisions that one is confident others will accept. This strategy is compatible with the cognitive miser hypothesis. Often the socially acceptable option is obvious, likely to come to mind quickly, and likely to be bolstered by supportive arguments readily available in the environment (especially true in group polarization and groupthink situations). The acceptability heuristic allows one to avoid much “unnecessary” cognitive work (analyzing the pros and cons of alternative course of actions; interpreting complex, often contradictory, patterns of information; making difficult trade-offs). All one needs to do is to adopt the salient acceptable option. Laboratory and field studies suggest that people frequently do exactly that: they choose the most clearly defensible action open to them. For instance,

ACCOUNTABILITY

34 1

several experiments on negotiation behavior have found that negotiators who expect to justify bargaining outcomes to the groups they represent have much more difficulty arriving at mutually beneficial compromise agreements than do negotiators who are not under such pressure (Klimoski, 1971; D. Pruitt, 1981). The most plausible explanation is that accountability to constituents (who presumably favor tough negotiation stands) induces concern for appearing strong by refusing to make concessions. Negotiators respond by employing competitive bargaining tactics that, while obstacles to resolving conflicts of interest, are effective in protecting their images in the eyes of constituents (Camevale, 1985). In a similar vein, experimental work on ingratiation reveals the willingness of people to tailor their opinions to those of others-especially high-status others (Jones & Wortman, 1973). In our own experimental work, we have documented that, when subjects know the views of the audience and do not feel locked into any prior attitudinal commitment, they shift their views toward the prospective audience (Tetlock, 1983a; Tetlock, Skitka, & Boettger, 1989). Subjects accountable to a liberal audience report substantially more liberal attitudes than do subjects accountable to a conservative audience. Moreover, most subjects do not internalize these attitude shifts. When we content analyze confidential thought protocols, accountability to a known audience affects neither the number nor the self-rated importance of the liberal and conservative thoughts reported. When we assessed attitudes in a later follow-up interview, there was virtually no evidence that these strategic attitude shifts had persisted over time. Consistent with the “elasticband” findings of Cialdini, Levy, Herman, Kozlowski, and Petty (1976), most people quickly snap back to their original attitudinal positions. The acceptability heuristic is not limited to laboratory experiments; it also operates in high-level policy-making settings. The political necessity of defending one’s conduct is a key constraint on how policy-makers choose among the options confronting them. In the words of C. Wright Mills (1940), “Often anticipations of appropriate conduct will control conduct (‘If I did this, what could 1 say? What would they say?’). Decisions may be, wholly or in part, delimited by answers to such queries.’’ The political scientists Snyder, Bruck, and Sapin (1962) took a similar stand in their pioneering analysis of the foreign policy-making process: “The decision to perform or not to perform a given act may be taken on the basis of available answers to the question, ‘What will be said?’.” Consistent with these propositions, historical case studies of government decisions references to policy-makers assessing possible lines of defense against critics and opponents (Anderson, 1981; Graber, 1976). For instance, after Egypt nationalized the Suez Canal in 1956, British leaders tried to avoid open collusion with Israel because they thought it could not be justified to their home constituencies. The British therefore delayed the invasion until after the initial Israeli strike

342

PHILIP E. TETLOCK

into Egypt when they erroneously thought that they could assume the role of peacemaker. In Goldmann’s (1971) words: “The British searched for an acceptable justification. Having found one, actions were modified accordingly.” Similarly, Kennedy rejected direct American participation in the 1961 Bay of Pigs invasion because he felt it would be awkward to justify after he had pledged abstinence from such a conflict. The decision meant that air cover for the invasion would be inadequate and would contribute to the failure of the mission. These examples illustrate what Snyder et al. (1962, p. 183) called “the continual interaction between considerations of what to do and what to say . . . Statecraft from this point of view is the art of combining the desirable and the justifiable.” Pfeffer (1981) stresses the importance of justifiability of policy options in managerial decision-making. He argues that the primary task of managers is to make what is going on in the organization meaningful to the participants and to develop a social consensus around the activities undertaken. A critical determinant of managerial success is the ability to convince both internal (within the organization) and external observers that the operations of the organization are consonant with prevailing social values and system. Management involves more than making sense of the world-it requires mobilizing social support for the policies that management endorses. Over time, the search for courses of action that can be readily justified may become scripted or ritualized, thus requiring even less cognitive effort. For instance, Meyer and Rowan (1977) have argued that the formal structures of many organizations in postindustrial society reflect the “justificatory myths” of the political environment, not the rational demands of the work activities. They cite the example of personnel services. Such services not only accomplish the instrumental functions of matching people to jobs, they also communicate to a wide range of constituencies that the organization follows legitimate procedures-such as equal opportunity programs or personality testing-and makes a good-faith effort to achieve collectively defined (often mandated) goals. In overview, the acceptability heuristic complements and fleshes out the cognitive miser characterization of the decision-maker. The most salient consideration in many decisions is the justifiability of policy options to others. The cognitive research program tells us that people often use a small number of items of information in making up their minds; the social contingency model tells us that decision-makers’ estimates of the probable reactions of those to whom they are accountable will be prominent among the few items considered. The cognitive research program focuses on process (how people think), the social contingency model focuses on content (what people think). Although this may seem a congenial division of labor, the distinction is far from airtight. Demands for accountability not only affect what people think; demands for accountability also affect how people think.

ACCOUNTABILITY

343

D. PREEMPTIVE SELF-CRITICISM Although the acceptability heuristic can be cognitively economical and socially adaptive, its usefulness is limited to settings in which one can easily figure out what others want or expect. What happens, however, when people find themselves in normatively ambiguous situations in which they do not know what the audience wants and they are not locked into any prior attitudinal commitments? Numerous studies have shown that under these conditions accountability can induce people to abandon their cognitive-miserly ways. Predecisional accountability to unknown audiences frequently motivates vigilant, complex, and self-critical thinking (Chaiken, 1980; Cvetkovich, 1978; Hagafors & Brehmer, 1983; McAllister, Mitchell, & Beach, 1979; Rozelle & Baxter, 1981; Tetlock, 1983a; Tetlock et al., 1989). Tetlock (1983a) and Tetlock et al. (1989) hypothesized that when people know the views of the audience, they rely on the acceptability heuristic and simply shift their views toward the audience. By contrast, when people know nothing about the prospective audience, they need to think through the issue much more carefully to arrive at a defensible position. Accountability to unknown others motivates people to consider arguments on both sides in order to prepare themselves for a variety of critical reactions to their views. To test these hypotheses, we conducted experiments in which subjects reported their positions on controversial policy issues (capital punishment, university tuition, affirmative action) under one of four conditions: expecting the positions they took to be confidential or expecting to justify their positions to a liberal, conservative, or unknown audience. In addition, subjects reported their thoughts (confidentiality always guaranteed) on each issue prior to committing themselves to positions. These thought protocols were then subjected to detailed content and structural analysis designed to assess the “integrative complexity” of subjects’ thinking on the issues: How many aspects or dimensions of each issue did they distinguish? Did they interpret issues in dichotomous, good-bad terms, or did they recognize positive and negative features of stands on both sides of the issues?) Subjects coped with pressures to justify their opinions in two qualitatively distinct ways: strategically shifting their public positions on the issues (thus making the task of justification easier) and thinking about issues in more multidimensional ways (thus preparing themselves for possible counterarguments). They relied on strategic attitude shifts (the acceptability heuristic) when they felt accountable to an audience with known liberal or conservative views. Not surprisingly, subjects accountable to a liberal audience expressed more liberal views whereas subjects accountable to a conservative audience expressed more conservative ones. Accountability to a known audience had no impact, however, on the complexity of private thoughts on the policy issues. The reverse pattern emerged

344

PHILIP E. TETLOCK

among subjects who felt accountable to an unknown audience. Here accountability had no impact on the complexity of private thoughts. Subjects were more tolerant of evaluative inconsistency (recognizing good features of rejected policies and bad features of accepted ones) and aware of difficult value trade-offs (e.g., the need to deter crime and protect the lives of the innocent, the need to remedy past racial injustices without creating new ones). Subjects accountable to the unknown audience engaged in preemptive self-criticism. They attempted to anticipate arguments that critics could raise against their positions. This cognitive reaction can be viewed as an adaptive strategy to protect both one’s selfimage and social image. Expecting to justify one’s views to an unknown audience raised the prospect of failure: The other person might find serious flaws in one’s position. In order to reduce the likelihood of such an esteem-threatening and embarrassing event, subjects demonstrated their awareness of alternative perspectives. The implicit message was, “You can see I’m no fool. I may believe this, but I understand the arguments on the other side.” Several other studies-in which subjects have felt accountable to an individual with unknown or, at least, difficult-to-guess policy preferences-have reported similar effects (Chaiken, 1980; McAllister et al., 1979). From a social cognition perspective, a key question becomes: Does motivating people to think in more integratively complex ways affect well-replicated judgmental shortcomings? Are complex thinkers any less susceptible to belief perseverance, the fundamental attribution error, and overconfidence? The evidence reported later indicates that predecisional accountability to unknown audiences does indeed attenuate, sometimes even eliminate, certain biases. It would be wrong, however, to conclude that integrative complexity is always beneficial. Integrative complexity also sometimes exacerbates effects that many observers view as biased or flawed.

E. THE RATIONALIZATION HEURISTIC: THE COGNITIVE MISER ON THE DEFENSIVE The previous discussion focused on situations in which the desire for approval and respect motivated complex patterns of thinking. In all these situations, people had no basis for inferring the policy preferences of the audience, thus greatly reducing the usefulness of the low-effort acceptability heuristic. Another critical ingredient was also present. People realized that they would need to explain their conduct before they had committed themselves to a course of action. Predecisional accountability, combined with normative ambiguity, promoted vigilant analysis of options. Other combinations of circumstances can trigger very different coping responses. For instance, imagine that people are accountable not for decisions they have yet to make, but for decisions they have already made. Imagine, moreover,

ACCOUNTABILITY

345

that these decisions have led to undesirable consequences (lower profits, bad publicity, employee dissatisfaction, etc.). Here, the same basic motive-the desire for approval and respect-will lead not to a forward-looking rationality, but rather to a backward-looking rationality, a defensive search for ways of rationalizing past conduct. The primary concern of decision-makers will be to portray earlier actions in the best possible light. Tetlock et al. (1989) illustrated how a seemingly minor variation in the timing of an accountability manipulation can dramatically influence the coping strategies activated. In this study, subjects reported their thoughts on four controversial issues either before or afer they had made a commitmentto attitudinal stands on those issues. Some subjects believed their attitudinal stands were private; others believed that they would later be asked to justify their stands to an audience with unknown, liberal, or conservative views. There was a strong interaction between when subjects reported their thoughts and the accountability manipulation. Subjects who felt accountable and reported their thoughts after making attitudinal commitments became much less integratively complex than both unaccountable subjects who reported their thoughts after making commitments and than unaccountable or accountable subjects who reported their thoughts prior to taking a stand. Once accountable subjects had publicly committed themselves to positions, the major function of thought became generating as many justifications for those positions as they could. As a result, the integrative complexity of thoughts plunged (subjects were far less likely to concede legitimacy to other points of view) and the number of proattitudinal thoughts increased (subjects generated more reasons why they were right). It might be tempting to downplay the importance of these postdecisional cognitive defenses. Postdecisional accounting, however, goes far beyond mere verbal posturing (Brockner & Rubin, 1985; Staw, 1980; Teger, 1980). In a variety of experimental paradigms (The Dollar Auction, The Waiting. Game, Business Simulations), investigators have demonstrated that the need to justify policies that have worked out badly places great pressure on decision-makers to increase their behavioral commitments to these failing policies (a prediction that follows from both cognitive dissonance and impression management theories; Schlenker, 1982; Wicklund & Brehm, 1976). For instance, in a series of business simulation experiments, Staw found that pressures to account for poor past decisions can reinforce decision-makers’ commitments to earlier actions, increasing the rigidity and inflexibility of the policymaking process. Staw (1976) studied the tendency of decision-makers to escalate commitments to a policy following the receipt of disappointing feedback on the effectiveness of the policies. One group of subjects (the personally responsible group) was instructed to allocate research and development funds to one of two operating divisions of a company. They were then given the results of their initial decision (successful or unsuccessful) and asked to make a second allocation

346

PHILIP E. TETLOCK

decision. Another group of subjects received the same information, but did not make the initial allocation decision themselves (this decision had supposedly been made earlier by a financial officer of the firm). The results revealed a significant interaction between the success-failure and personal responsibility manipulations. Subjects allocated the most money to the failing division of the company when they felt personally responsible for having directed funds to that division in the initial allocation decision. These findings suggest that subjects sought to justify a poor decision by escalating their commitment to it. Staw’s earlier work-inspired by the cognitive dissonance theory-treated justification as an intrapsychic process (people are concerned with protecting their self-images as competent and rational beings). A later simulation experiment by Fox and Staw (1979) explicitly focused on the impact of political pressures for accountability on the decision-making process. In this simulation, subjects were placed in an administrative situation in which they possessed high or low job security and in which they believed that the board of directors (to whom they were accountable) was receptive or unreceptive to the policies they had been pursuing. Fox and Staw hypothesized that decision-makers would feel the greatest need to escalate their commitment to a failing policy when they were most vulnerable (low job security and a skeptical board of directors). The results supported these hypotheses. Decision-makers who were worried about keeping their jobs and fending off high-level critics within the organization were most likely to escalate their commitment to their initial decision and were most inflexible in defending their original positions. There are many reasons to suppose, then, that postdecisional accountabilityfar from encouraging complex, self-critical thought-actually exacerbates many judgmental biases and defects of the cognitive miser. Demands for accountability sometimes motivate people to bolster previous decisions, to be overconfident in the correctness of those decisions, to overassimilate new evidence, and to deny difficult value trade-offs, particularly when the trade-offs require acknowledging flaws in one’s past decisions (cf. Festinger, 1964; Kiesler, 1971).

F. HYBRID RESPONSES TO ACCOUNTABILITY We have identified an assortment of coping strategies, including strategic attitude shifts, mindless endorsement of organizational rituals and myths, thoughtful analysis of available evidence and options, and the rigid defense of earlier choices. The antecedent conditions for activating each coping strategy rarely, however, appear in pure form, but rather in varying degrees and combinations in any given situation. The coping strategies are thus not as mutually exclusive as presented here. In many situations, the views of those to whom one is accountable are neither perfectly obvious nor completely unknown. We have

ACCOUNTABILITY

341

suspicions (held with varying degrees of confident) about what others want to hear. In other situations, we are accountable not to one but to many individuals whose views may be in harmony or in conflict. Preliminary evidence from our experimental work suggests that these forms of accountability will trigger multiple coping responses, including both the search for a least common denominator acceptable policy and vigilant information processing to anticipate objections to that policy (see also Carnevale, 1985). In still other situations, we may be called upon to account both for actions we have already performed and for decisions we have yet to make. The views of those to whom we are accountable may be only partly known. Again, multiple coping responses will probably be activated-for instance, a search for a policy acceptable to key constituencies in conjunction with complex thought and the development of rationalizations that render new decisions consistent with previous commitments. We should also not focus solely on situational determinants of coping strategies. Cognitive stylistic and motivational variables influence responses to accountability. Tetlock et al. (1989) found that social anxiety and need for approval predicted not only how motivated people were to gain the approval of prospective audiences, but also the strategies they used to achieve that goal. High social anxiety subjects were more likely to become integratively complex in response to predecisional accountability demands from an unknown audience and integratively simple in response to postdecisional accountability demands from known and unknown audiences. Cognitive style variables such as dogmatism and need for cognition are likely predictors of coping response thresholds (e.g., when decision-makersengage in flexible, self-critical information processing as opposed to rigid bolstering of preferred options). A comprehensive theoretical analysis will ultimately have to incorporate such individual difference variables.

IV. Normative Assessments of Coping Strategies A widely held misconception is that accountability is a fail-safe cure to a wide range of social ills. If the graduates of our schools cannot read, then we should mandate compulsory testing and link funding to performance on those tests. If our military cannot defeat a guerrilla insurgency, then we should link promotions to tangible indicators of success such as body counts and percentage of population and territory controlled. If professors are aloof and indifferent toward undergraduates, then we should link promotions to undergraduate course evaluations. From government bureaucracies to academia to the military to corporate capitalism, the clarion call can be heard: “Hold the rascals accountable.” Although one may sympathize with the frustration behind such calls, this

348

PHILIP E. TETLOCK

simplistic view fails for a number of reasons. One problem is that most social indicators of performance (to which we hold people accountable) are easily corrupted. Teachers may teach the content of tests; military officers may inflate statistical reports of progress; professors might entertain rather than educate undergraduates. A second problem is that the criteria we “annoint” (J. Darley, personal communication, 1991) as measures of progress are often only probabilistically related to the objectives we are trying to achieve. The productivity of doctors is imperfectly captured by the number of patients they see or the number of procedures and drugs they prescribe; the long-term effectiveness of chief executive officers of corporations is weakly correlated with the latest quarterly profit figures; the long-term effectiveness of government leaders is only loosely connected to the most recent trends in crime, economic indicators, or international relations. By annointing the wrong standard of performance in our accountability system, we may wind up encouraging the wrong kinds of behaviors (rewarding doctors who examine many patients superficially or who overprescribe expensive procedures, in the process increasing both patient mortality and health care inflation). Or we may wind up holding people responsible for outcomes outside of their control (e.g., cyclical economic trends that increase or decrease profitability, demographically induced trends in crime or unemployment, etc.) and failing to hold people responsible for outcomes under their control (not rewarding doctors for practicing preventive medicine that saves lives and money in the long run, not rewarding corporate leaders for investing in infrastructure and modernization programs that are critical for long-term competitiveness, etc.). In designing accountability systems, we should of course be sensitive to both the corruptibility of our standards and the danger of annointing the wrong standards. The social contingency model underscores this point by highlighting the importance of characterizing precisely the incentives that accountability systems create. Each of the three major coping strategies can be viewed as a reasonable response to a particular configuration of circumstances that affects the costs and benefits of available social and cognitive responses. In addition, each coping strategy can be viewed as either adaptive or maladaptive for either the individual decision-maker or the institutions the decision-maker inhabits. In the following sections, we summarize the functional pros and cons of each coping strategy.

A. THE ACCEPTABILITY HEURISTIC The acceptability heuristic is, perhaps, the least inspiring strategy for coping with accountability. This strategy does, however, have obvious adaptive value for the individual decision-maker. People, quite correctly, view opinion conformity as a reliable means of gaining the approval and respect of others. Although

ACCOUNTABILITY

349

limiting conditions exist (one should not be too sycophantic), we generally evaluate others more positively to the degree their attitudes are similar to our own. We see similar others as more likable and intelligent than dissimilar others (Byme, Nelson, & Reeves, 1966) as well as more deserving of promotion (Baskett, 1973). The acceptability heuristic may also have adaptive value for the groups or organizations to which the individual belongs. To the degree group members practice this heuristic, it will be relatively easy to reach consensus and unanimity. And, to the degree group members feel accountable to constituencies who have a good understanding of the issues, decision-makers who take the views of these constituencies into account are likely to do a better job than decision-makers who ignore these constituencies. Indeed, under such conditions, the acceptability heuristic may check a wide range of judgmental fallacies that decision-makers never would have corrected on their own. To put the point bluntly, one is less likely to ignore base rates or to commit the conjunction fallacy if one’s boss is well-versed in probability theory and has made clear the relevant standards of performance evaluation. The acceptability heuristic can, of course, also have highly dysfunctional effects, from both an individuar and an organizational perspective. From the standpoint of maximizing individual career prospects, a decision-maker who relies solely on the acceptability heuristic runs the risk of appearing uncreative, unnecessary, or spineless. From an organizational standpoint, decision-making groups that consist solely of practitioners of the acceptability heuristic are likely candidates for groupthink (Janis, 1982, 1989). If no one is willing to voice unpopular doubts or misgivings, the likelihood of serious miscalculations may increase dramatically. Finally, just as it is possible for decision-makers to recommend the right solution because they are accountable to a wise audience, it is equally possible for them to recommend the wrong solution because they are accountable to a shortsighted audience. The acceptability heuristic may cause doctors to overprescribe drugs and tests for their patients (because they expect to be accountable to judges and juries who are susceptible to the certainty-ofhindsight effect) or it may cause the chief financial officers of corporations to argue against costly long-term restructuring that will adversely affect cash flow and immediate profitability (because they expect to be accountable to an investment community that focuses on maximizing quarterly returns). The acceptability heuristic implies that decision-makers can be no better as well as no worse than the constituencies to whom they are accountable. Several experiments have illustrated the dysfunctional effects of the acceptability heuristic. For instance, Adelberg and Batson (1978) found that accountability impaired effective helping whenever agency resources were inadequate to assist everyone in need and difficult choices were necessary. Accountability did so by distracting help-givers from the problems they were supposed to be solving

350

PHILIP E. TETLOCK

and focusing their attention on the need to justify what they were doing. Accountable help givers were more likely to avoid hard-to-justify, but necessary, decisions concerning which applicants for aid (in this case, student loans) would receive assistance. They made a much higher proportion of wasteful decisionsgiving ineffectively small amounts of money to large numbers of students. In an experimental simulation of Food and Drug Administration (FDA) decisions to permit drugs into the United States pharmaceuticals market, Tetlock and Boettger (1991) asked both accountable and unaccountable subjects to judge the acceptability of an anticlotting drug that (1) was or was not already on the market (the status quo manipulation) and (2) according to initial research, would save 300, 600, or 900 lives but at the cost of side effects (hemorrhaging) that would claim 0, 100, or 300 lives. We found a significant three-way interaction between accountability, the status quo manipulation, and the magnitude of the expected risk. Relative to unaccountable subjects, accountable ones were especially reluctant to accept a drug that would impose suffering on an identifiable patient population, even if the number of lives saved greatly exceeded the number of lives lost. Moreover, this differential response of accountable and unaccountable decision-makers was much more pronounced when judging a drug not yet on the market than when judging a drug already on the market. Accountable decisionmakers were also more likely than their unaccountable counterparts to try to escape the decisional dilemma that had been created for them (cf. Janis & Mann, 1977, on defensive avoidance). Accountable subjects confronted by close-call cost-benefit decisions on a drug not yet on the market were more likely to procrastinate (by deferring the decision until further research had been done, although many people would die in the interim and there was little prospect of more conclusive evidence from the further work) and to “buckpass” (to refer the decision to another government agency). One can, of course, attach very different normative labels to these effects. From a harsh perspective, accountability encourages moral and political cowardice; from a more charitable perspective, accountable subjects were appropriately circumspect about their qualifications to make decisions that would profoundly affect the well-being of others. Defenders of the judgment policies of accountable subjects may, however, have the more difficult assignment. When confronted by a drug already on the market, accountable subjects were almost as willing as unaccountable subjects to trade off lives saved from prescribing the drug for blood clots against lives lost from side effects. Accountable subjects found the trade-off especially aversive only when they believed the drug was not on the market. The key difference between accountable and unaccountable subjects does not lie in moral objections to utilitarian cost-benefit reasoning; rather it resides in the asymmetrical treatment of both omissions versus commissions and losses versus gains (Spranca, Minsk, & Barron, 1991; Tversky & Kahneman, 1982). Most accountable subjects were willing to save 600 lives at the cost

ACCOUNTABILITY

35 I

of 100 when the drug was on the market but balked at the same cost-benefit ratio when the drug was not yet on the market. The losses one imposes by changing the status quo loom larger than the gains one confers. As the thought protocol evidence indicated, accountable subjects were primarily concerned about the patients who would die from banning an available drug or about the patients who would die from side effects from introducing a currently unavailable drug. This pattern of attentional focus is what one would expect from decision-makers who were preoccupied with justifying their choices and anticipating the cohesiveness, size, and anger of the relevant audiences. It is not hard to imagine the furor from patients whose lives have been saved (and might again be) by a drug that the FDA now wants to withdraw from the market (although that protest will be offset by support from the victims of side effects). It is also not hard to imagine the protests of those who will suffer the side effects from a newly introduced drug. When those who would have benefited from the drug do not know who they are (as is generally the case when the drug is not yet on the market), the safest political strategy is to procrastinate and defer responsibility. Only when this normally quiescent interest group is mobilized (e.g., the recent surge of action among AIDS patients who seek out risky, unapproved drugs to forestall what is currently the certainty of painful death) is it possible to highlight the opportunity costs of procrastination and buckpassing (Graham, I99 1). In short, simple normative generalizations about the acceptability heuristic fail us. Whether one applauds the results of the heuristic depends on whether one approves or disapproves of the standards and priorities of the relevant audience. Focusing on the heuristic as decision-making procedure, one’s normative assessment is colored by one’s assumptions about whether the decision-making benefits (savings in cognitive effort, emotional strain, social controversy, and time) are outweighed by the potential liabilities: Would decision-makers have done a better job if they had disregarded the views of the relevant constituencies and thought through the issues on their own?

B. PREEMPTIVE SELF-CRITICISM Predecisional accountability to unknown audiences often motivates people to anticipate objections that others might raise and to incorporate those objections into their own position. The result is more evaluatively inconsistent, flexible, and multidimensional thinking. The normative question is whether motivating integrative complexity increases, decreases, or has no effect on well-replicated biases in human judgment. Once again, a highly conditional assessment is in order. Preemptive selfcriticism attenuates, even eliminates, certain biases (belief perseverance, the fundamental attribution error, and overconfidence) but magnifies others (the

352

PHILIP E. TETLOCK

dilution effect, the tendency to pay undue attention to worst-case scenarios, and the tendency to take weak initial negotiating stands). From one normative perspective, preemptive self-criticism represents flexibility, open-mindedness, and sophistication; from another, it looks like confusion, vacillation, and weakness.

I . The Benefits of Complexiry Inducing preemptive self-criticism has proven to be a powerful method of eliminating judgmental biases that are resistant to other types of experimental manipulations and incentives.2 I consider three examples here: belief perseverance, the fundamental attribution error, and overconfidence. a. Reducing Belief Perseverance. The prevailing view of the person within the cognitive research program has been that of a theory-driven thinker who relies heavily on preconceptions in interpreting new information. Considerable evidence buttresses this view that people are sometimes slow to revise their initial impressions of events, even in the face of directly contradictory evidence (e.g., Nisbett & Ross, 1980). *It is important to distinguish carefully among independent variables that, on intuitive grounds, one might suppose have similar effects. As we have seen, accountability is not simply a generic motivator of cognitive work. Different types of accountability can have radically different effects. Contingency theory predictions need to be based on detailed analyses of the situational incentives for particular kinds of cognitive processing. Three examples underscore this point. Superficially, cognitive tuning and accountability manipulations are quite similar. Placing subjects in a transmission set (leading them to expect to communicate their opinions to others) might not seem very different from making subjects accountable (leading them to expect to justify their opinions to others). The two manipulations, however, sometimes have very different effects (cf. Zajonc, 1960; Tetlock, 1985). Subjects in transmission sets, for example, form more extreme dispositional attributions in an essay attribution paradigm than do subjects in a no-set control condition (cf. Harvey, Harkins, & Kagehiro, 1976). By contrast, subjects given accountability instructions make less extreme and more discriminating patterns of causal attributions (Tetlock, 1985). Expecting to communicate one’s impressions of an event places a premium on one’s ability to generate succinct and readily comprehensible descriptions of that event; expecting to justify one’s impressions of an event places a premium not only on communicating one’s opinions, but also on defending those opinions against possible counterarguments. The former manipulation encourages people to suppress ambiguity and to present issues in sharp, polarized terms. The latter manipulation encourages people to express complex, many-sided opinions that are difficult to refute and easy to justify. One might also expect an accountability manipulation to have effects very similar to those of a manipulation of decision importance. Although these two independent variables do sometimes have similar effects (e.g., McAllister et al., 1979). they also sometimes have quite dissimilar effects. For instance, the effects of accountability on overconfidence are very different from the effects of decision importance on overconfidence reported by Siebert (1974). She found that increasing the importance of the decision-making task exacerbated overconfidence-a result that she interpreted in terms of Hull-Spence theory (the tendency for arousal to increase the likelihood of dominant responses). Obviously, given the many differences between the two manipulations, there are many possible explanations for the diverging results. Perhaps Siebert’s manipulation (course grades were thought to

ACCOUNTABILITY

353

Belief perseverance is not, however, an immutable law of thought. Tetlock (1983b) found that accountability can, under certain conditions, prevent first impressions from dominating final judgments (although it cannot reverse the effect once it has occurred). He presented subjects with a long list of arguments from a murder trial, half of which cast doubt on the defendant’s guilt and half of which suggested that the defendant was indeed guilty. He also varied the order in which subjects received the evidence: an exonerating/incriminating, an incriminating/exonerating, and a randomly alternating order of presentation. Subjects who did not expect to justify their judgment of the defendant’s guilt showed a substantial primacy effect: Early evidence had greater impact on subjective probability ratings of guilt than did later evidence. Subjects who expected to justify their judgment of the defendant’s guilt before viewing the evidence were, however, immune to the primacy effect. Order of presentation made no difference. Moreover, accountability did not eliminate the primacy effect by merely affecting the types of judgments subjects were willing to express (e.g., accountability did not turn people into fence-sitters who were unwilling to commit themselves to any position). Two lines of evidence argued strongly against be at stake) was more powerful than the Tetlock and Kim (1987) manipulation and produced levels of arousal that interfered with, rather than facilitated, self-critical, integratively complex thought. Arousal is related in a curvilinear fashion to integrative complexity, with moderate levels most conducive to complex functioning (Schroder, Driver, & Streufert, 1967). The Tetlock and Kim manipulation may have created an optimum level of arousal, and Siebert’s, a superoptimal level. It is also possible, however, that the two manipulations differ in qualitative ways. More important than the general arousing properties of accountability may be the specific coping responses activated by the need to justify one’s view. Accountability may serve as a signal to subjects to take the role of the other toward their own mental processes and to give serious weight to the possibility that their preferred answers might be wrong. In this view, accountability does not simply motivate thought; it functions as a social brake on judgmental biases that occur in our less reflective moments. A third example is the superficial similarity between accountability and the personal involvement manipulations employed by Petty and Cacioppo (1986), in their experimental work on attitude change. The elaboration-likelihood model predicts that personal involvement will amplify “dominant cognitive responses” to a persuasive message. If the message contains strong arguments, personally involved subjects will be all the more persuaded; if the message contains weak arguments, personally involved subjects will be less persuaded. Petty and Cacioppo (1986) have obtained the predicted interactions between personal involvement and other independent variables in numerous studies. The social contingency model, by contrast, predicts that predecisional accountability to unknown audiences will act as a check on dominant cognitive responses by motivating preemptive self-criticism. This type of accountability may actually cancel out the effects of personal involvement. The social contingency model is not always, however, in tension with the elaboration-likelihood model. Postdecisional accountability-which appears to motivate defensive bolstering and the generation of proattitudinal thoughts-may have effects quite similar to personal involvement. These examples suggest that contingency theories need (I)to specify the limits on the flexibility of judgment and choice processes and (2) to clarify distinctions among independent variables that are often loosely classified as motivators of cognitive work or effort, including personal involvement, decision importance, transmission-receiver set, monetary incentives, and reversibility of outcomes.

354

PHILIP E. TETWCK

such a response bias artifact. First, accountability per se was not sufficient to eliminate the primacy effect. Subjects who realized they were accountable only after exposure to the evidence displayed primacy effects comparable in magnitude to those of unaccountable subjects. Only accountability prior to the evidence destroyed the primacy effect. Second, subjects who realized they had to justify their views prior to the evidence recalled significantly more case information than did subjects who felt unaccountable or accountable only after exposure to the evidence. A response bias interpretation cannot explain these effects on memory. Taken as a whole, the data strongly suggest that preexposure accountability induced people to become more vigilant information processors who were willing to revise initial impressions in response to changing evidence. Rozelle and Baxter (1981) also reported evidence that accountability can encourage data-driven and discourage theory-driven information processing. They noted that previous work on person perception had shown that characteristics of the perceiver are frequently more important determinants of person perception than are characteristics of the person being judged. Perceivers tend to offer undifferentiated descriptions of stimulus persons (a given perceiver tends to see different stimulus persons as similar to each other) as well as idiosyncratic descriptions of stimulus persons (little overlap exists in the descriptions that different perceivers offer of the same stimulus person). This pattern is exactly what one would expect if people were theory-driven thinkers who rely on their own implicit theories of personality and give little weight to actual properties of the persons being judged (cf. Bourne, 1977). The earlier work was conducted in settings in which people did not believe their judgments would have important consequences for either themselves or others. Rozelle and Baxter explored the impact on persons perception of whether people felt their decisions would be important (influence applicants’ admission to graduate school) and of whether perceivers believed they would later have to justify their decision to a faculty review committee. Under high importance and high accountability, they found a reversal of the typical finding of undifferentiated and idiosyncratic descriptions of stimulus persons. The following patterns emerged: ( 1) differentiated perceiver descriptions of stimulus persons (low within-judge-overlap of descriptive characteristics assigned to others) and (2) substantial agreement among judges in the descriptions offered of the same stimulus persons. In brief, accountability appeared to sensitize perceivers to “what was actually out there.” 6. A Social Check on the Fundamental Attribution Error. Many social psychologists believe there is a systematic bias in the person perception process: a pervasive tendency among observers to overestimate personality or dispositional causes of behavior and to underestimate the influence of situational constraints on behavior (Jones, 1979; Nisbett & Ross, 1980; Ross, 1977). Indeed, Ross (1977) was sufficiently confident in the robustness of the phenomenon to label it

ACCOUNTABILITY

355

the fundamental attribution error. Jones (1 979) used a somewhat less judgmental term, the overattribution effect. One widely accepted explanation for the effect emphasizes people’s reliance on simple, highly overlearned, judgmental heuristics in social perception tasks. Heider (1958) proposed the earliest variant of this explanation. Behavior, he argued, engulfs the perceptual field. The person and his or her behavior form a natural perceptual-cognitive gestalt. What could be more obvious than the basic palpable act that without the actor there can be no act? And what could be simpler than to ascribe behavioral dispositions to the actor that render the actoract linkage intelligible (aggressive people behave aggressively, intelligent ones, intelligently, etc.)? Nisbett and Ross (1980) noted the compatibility of this explanation with recent work on heuristics. Personality traits are generally the most cognitively available and representative explanations for behavior. People prefer dispositional explanations because such explanations are typically the first ones to come to mind, and people rarely bother to consider less obvious situational ones. People are content with top-of-the-head causal interpretations of behavior (Fiske & Taylor, 1991). The most widely cited evidence for the fundamental attribution error comes from the attitude-attribution paradigm (Jones, 1979). In a typical experiment, subjects are presented with written or spoken statements of opinion that have allegedly been made by a target person under conditions of high or low choice. The task of observers is to infer the “true” attitude of the target person who expressed these opinions. Tetlock (1985) conducted an essay attribution experiment that explicitly manipulated whether subjects felt accountable for their attributional judgments and when they learned of being accountable. The essay topic in this case concerned minority quota systems in college admissions. Subjects were sometimes confronted by an essay that advocated a quota system and sometimes by an essay that opposed such a system. In addition, subjects learned either that the essay writer had little choice concerning the position advocated (a requirement for participating in an experiment) or that the writer had freely chosen to take the position advocated. The final independent variable was accountability. Subjects either did not feel accountable for their causal attributions, learned they were accountable prior to exposure to the evidence on which they would be basing their judgments, or learned of being accountable only after exposure to the evidence. The classic overattribution effect was replicated when subjects did not feel accountable for their attributional judgments or when subjects learned of being accountable only after exposure to all the evidence. Although subjects in these conditions made less confident inferences about the essay writer’s attitudes in the low-choice than in the high-choice conditions, they still made quite strong inferences about underlying attitudes in the low-choice conditions. Second, sub-

356

PHILIP E. TETLOCK

jects who learned of being accountable prior to exposure to the evidence successfully resisted the overattribution effect. When these subjects were informed that the essay writer did not choose the position advocated, they refrained from drawing correspondent inferences about the writer’s underlying attitudes from the essay. It is also worth noting that preexposure accountability did not make subjects indiscriminately cautious. Preexposure accountability subjects drew every bit as extreme inferences about the essay writer’s underlying attitudes in the high-choice condition as did unaccountable or postexposure accountability subjects. These results are difficult to reconcile with a response bias interpretation. If accountability merely transformed people into fence-sitters, it should not have mattered whether subjects learned of being accountable before or after exposure to the evidence. It should also not have mattered whether subjects made inferences about low- versus high-choice essay writers. The specificity of the effects suggests that the accountability manipulation did not just shift some general response threshold; rather, it encouraged people to become more differentiated and circumspect thinkers about the causes of behavior. c. Inducing Appropriate Conjidence. A substantial literature indicates that people are often excessively confident in the correctness of their factual judgments and predictions (Fischhoff, 1982). Tetlock and Kim (1987) investigated the impact of accountability demands on cognitive processing in a personality prediction task. We presented subjects with the responses of actual test-takers to 16 items drawn from Jackson’s Personality Research Form and then asked subjects to perform three tasks: (1) to form written impressions of each test-taker, (2) to predict how each of these individuals responded to an additional 16 items drawn from the same tests, and (3) to assign confidence ratings to their predictions. Three experimental conditions were of special interest: no accountability (subjects learned all of their responses would be anonymous), preexposure accountability (subjects learned of the need to justify their predictions prior to forming impressions of any of the test-takers), and postexposure accountability (subjects learned of being accountable only after forming impressions and reporting them, but not before making predictions and assigning confidence ratings to those predictions). The rationale for the pre- and postexposure accountability manipulations was the same as in the belief perseverance and overattribution studies. If accountability affects how people think about the evidence, then it should make a substantial difference when subjects learn of being accountable. If accountability merely affects responses thresholds, then it should not matter. Several effects emerged. First, preexposure accountability subjects formed more integratively complex impressions of the test-takers. They were more likely to recognize contradictory evidence on the test-taker’s standing on personality

ACCOUNTABILITY

357

dimensions and to offer multidimensional characterizations of the test-takers. Preexposure accountability subjects were more likely to say things such as “This person is friendly and outgoing in some situations, but not in others” or “This person likes the company of others, but will give priority to his work when push comes to shove.” Second, preexposure accountability subjects made more accurate predictions of the actual responses of the test-takers to the remainder of the tests. Relatively few people have such extreme scores on the Affiliation or Achievement scales that one can do a good job predicting their responses to individual items from global, undifferentiated trait labels. It helps, in most cases, to be more integratively complex. Third, preexposure accountability subjects reported more realistic levels of confidence than no-accountability or postexposure accountability subjects. Moreover, this was not just a matter of accountability making people more cautious. Three aspects of evidence allowed us to eliminate this interpretation: 1. Postexposure accountability subjects did not display a reduction in the overconfidence effect. Postexposure accountability subjects reported confidence ratings that were virtually identical to unaccountable subjects. 2. The confidence ratings of preexposure accountability subjects were not just lower, they were also better calibrated. It is possible-using the Brier-Murphy procedure-to partition the covariance between accuracy scores and confidence ratings into three additive components (a knowledge component, a calibration component, and a resolution component). The calibration component is the weighted average of the mean square differences between the proportion of correct predictions in each category and the probability value of that category. The resolution component is a measure of the judge’s ability to assign confidence ratings to predictions in such a way that the proportions of correct answers in differentcategories are maximally different-it is the variance of correct predictions across the confidence categories. 3. Not only were confidence ratings better calibrated, the superior calibration was not achieved at a cost in resolution (another sign that subjects were not just indiscriminately bunching up all of their confidence ratings at the low end of the probability scale in response to accountability demands).

A final result is worth noting. Analysis of covariance indicated that when we statistically controlled for the effects of preexposure accountability on the integrative complexity of the initial impressions subjects formed of the test-takers, the improvements in both accuracy and calibration were substantially reduced (although still significant). This result suggests that preexposure accountability is an effective debiasing manipulation, at least in part, because of its capacity to motivate complex, self-critical thought.

358

PHILIP E. TETLOCK

2 . Potential Dysfunctional Effects of Motivating Integrative Complexity One might conclude at this point that motivating integrative complexity and improving judgment and choice are synonymous. That conclusion would be premature. There are some settings in which encouraging integrative complexity, far from debiasingjudgment, appears to make matters worse. Tetlock and Boettger (1989), for instance, have shown that preexposure accountability to an unknown audience motivates integrative complexity (replicating previous studies) but also increases bias-in this case, the tendency to make inappropriately regressive predictions when diagnostic evidence is accompanied by diverse bits of nondiagnostic information (what Nisbett et al., 1981, and Zukier, 1982, have called the dilution effect). In one scenario, Tetlock and Boettger asked subjects to predict the grade-point averages of target students. Subjects received either only diagnostic evidence (e.g., a student studies 3 or 31 hours per week) or diagnostic evidence plus a host of irrelevant information (e.g., the student is a cheerful person, plays tennis three or four times a month, and has never dated anyone for longer than 2 months). This information had been carefully selected from pretesting; pretesting subjects were virtually unanimous that the information was useless. Nonetheless, this “useless” information caused subjects to be substantially less confident in their predictions. Moreover, accountability exacerbated the dilution effect. Accountable subjects tried to be “good,” complex information processors and to integrate both diagnostic and nondiagnostic evidence in making predictions about grade-point average. By motivating integrative complexity, the accountability manipulation sent our subjects off on inferential wild goose chases. One accountable subject concluded from the information that the student is cheerful, that the student is therefore well adjusted and likely to do well in school; another subject concluded from the fact the student occasionally plays tennis that he is in reasonably good shape and likely to be successful. Another subject concluded from the fact the student had never dated anyone longer than 2 months that the student is therefore emotionally unstable and likely to be doing poorly in school. In short, preexposure accountability to an unknown audience motivated subjects to be more integratively complex, but it did not make them wiser-it did not make them more discriminating consumers of the information at their disposal. Another example of the arguably maladaptive effects of integrative complexity comes from the Tetlock and Boettger (1991) study that examined judgments of the acceptability of a drug on the United States pharmaceutical market. As noted earlier, accountable subjects were much more responsive to the level of risk posed by the drug, and especially so when they believed the drug had not yet been admitted into the market. Analysis of covariance indicated that subjects who thought about the issues in more integratively complex ways were largely

ACCOUNTABILITY

359

responsible for the three-way interaction between the status quo manipulation, the accountability manipulation, and the level of risk manipulation. Examination of the thoughts reported by these subjects revealed a preoccupation with worstcase scenario thinking and explicit concern for what they would say to those who would be injured by their decisions. There is, of course, nothing immoral or irrational about such concerns. It is noteworthy, however, that these subjects were much more tolerant of risk created by a drug that was already on the market (the status quo condition). Removing a drug with a high benefit-cost ratio from the market would antagonize those constituencies who currently benefit from it. Introducing a drug with an equally positive benefit-cost ratio focuses attention on those constituencies who would be hurt. In short, accountability pressures that motivate integrative complexity make people reluctant to put themselves out on a limb and take stands that require painful trade-offs. The Tetlock and Boettger (1991) study, for instance, found that accountable subjects (especially in the nonstatus quo conditions) were not only more hesitant to accept risks from the new drug, they were also more likely to search for ways of passing responsibility for the decision to others (buckpassing) and of deferring the decision (procrastination). Integrative complexity can be a sign of confusion and vacillation.

C. THE RATIONALIZATION HEURISTIC Postdecisional accountability often triggers defensive bolstering-efforts to generate as many justifications as possible. The normative costs and benefits of this coping strategy look like mirror images of those for preemptive selfcriticism, The danger of defensive bolstering is that one will rigidly persevere with a failing policy in a vain effort to recoup sunk costs. By generating supportive cognitions, decision-makers who practice defensive bolstering blind themselves to changing policy in ways that would better promote their own values and those of the institutions they represent. Of course, it is not always a good idea to abandon a policy at the first sign of trouble. Decision-makers who practice defensive bolstering as opposed to preemptive self-criticism are more likely to stick with a fundamentally good policy that has recently run into serious shortterm difficulties. There is a fine, normative line between principled determination “to stay the course” and stubborn refusal “to acknowledge the facts.”

V. Another Look at Normative Issues From a social contingency perspective, each of the three major coping responses to accountability is appropriate under some circumstances and inap-

360

PHILIP E. TETLOCK

propriate under others. An Aristotelian golden mean is lurking here. One should be sensitive to the views of important constituencies, but avoid appearing so chameleonic that one loses their trust and confidence. One should be self-critical, but not to the point of paralysis. And one should stick by one’s principles but not to the point of dogmatism and self-righteousness. In discussing the pros and cons of different coping strategies, I have implicitly accepted the characterization of certain effects as errors or biases within the research literature. In some cases (e.g., overconfidence), such characterizations may be quite apt. The social contingency model of judgment and choice reminds us, however, that response tendencies that look like errors and biases given one set of assumptions about a subject’s beliefs and goals often look thoughtful and prudent given another set of functionalist assumptions. Before labeling an effect a cognitive flaw, one should consider (1) the interpersonal, institutional, and political goals that people are trying to achieve by making judgments of a particular type (e.g., Do people seek to achieve causal understanding or to express their moral approval/disapproval? Do people attempt to maximize expected utility or to minimize risk of serious criticism?) and (2) whether cognitive strategies that serve people well in everyday life may lead them seriously astray in laboratory experiments on judgment and choice. I present several examples of such normative reassessment here: the fundamental attribution error, biases in legal award and compensation decisions, the dilution effect, ambiguity aversion, the attraction effect, the conjunction fallacy, the overuse of base rates and underuse of individuating information, and the status quo effect.3

A. THE FUNDAMENTAL ATTRIBUTION ERROR

From a purely cognitive perspective, it is useful to think of people as intuitive scientists who seek cognitive mastery of causal relationships. Overestimating dispositional causes and underestimating situational causes look very much like 3Equally difficult are the normative issues that arise in assessing coping responses to accountability in negotiation settings. One simple hypothesis is that those forms of accountability that motivate integratively complex thought increase the likelihood of reaching mutually beneficial compromise solutions to mixed-motive conflicts (cf. D. G. Ruin & Lewis, 1975). Although one can make a reasonable empirical and logical case for this hypothesis, the hypothesis oversimplifies the nonnative issues at stake. Motivating the negotiators for one’s own side to be integratively complex (to appreciate the concerns of the other side and develop compromises that take those concerns into account) may be a poor idea when one confronts an implacably hostile opponent who cannot be appeased. The search for a viable integrative solution in such situations may be as quixotic as attempts to integrate all the stimulus evidence at one’s disposal in the “dilution” conditions of Tetlock and Boettger (1989). There are some negotiation contexts in which one is better off relying on simple decision rules that protect one from ruthless exploitation.

ACCOUNTABILITY

36 1

errors in this framework. If we assume, however, that people are intuitive politicians who have incentives to hold others strictly accountable for their conduct, the same judgmental pattern takes on a different normative hue. One way of pressuring other people to behave is by indicating to them that one has a low tolerance for justifications or excuses and that one will treat their behavior as automatically diagnostic of underlying intentions and personality attributes. In one study discussed earlier, Tetlock ( 1985) found that accountable subjects resisted the overattribution effect. The interpretation was that accountability had improved the quality of judgment by motivating subjects to be more discriminating and thoughtful. If accountability had merely motivated people to be more punitive and to make more dispositional attributions for behavior, it should not have mattered whether subjects learned of being accountable prior to or only after exposure to the stimulus evidence. The social contingency model warns us not to assume that accountability will always have such an effect. In some situations, accountability may make people into more thoughtful, intuitive scientists; in other situations, accountability may transform people into more punitive, intuitive lawyers or politicians. Much hinges on exactly how subjects define the experimental situation (“Is the experimenter primarily interested in my ability to make subtle causal distinctions, or in the moral and political soundness of my judgment?”). A testable hypothesis is that when subjects feel their moral judgment is of primary interest, accountability demands will motivate them to hold people more rigorously responsible for their actions. As Axelrod (1986) notes, people are not only expected to act in accord with prevailing norms, they are also expected to censure those who violate norms (a norm to enforce norms, or a metanorm). Insofar as accountable subjects feel that their moral mettle is being tested (their willingness to apply metanorms), they may be more motivated to hold others responsible and to reject situational explanations or excuses. This theoretical argument does not require any radical revision in the social contingency model. The argument simply asserts that accountable subjects will tend to rely on the acceptability heuristic when they can infer the views of the prospective audience. In this case, subjects may assume that the prospective audience expects them to hold people responsible for deviant or untoward conduct.

B. BIASES IN COMPENSATION DECISIONS In a series of ingenious experiments, Ritov, Hodes, and Baron (1990) have shown that observers direct more money to victims of misfortune when the misfortune is caused by a person rather than natural causes and when the misfortune is the result of a malfunction rather than a routine procedure. Consider the following example: A train approaches a tree that has fallen across the tracks and

362

PHILIP E. TETU)CK

could cause a derailment. Either an engineer or an automatic stopping device is responsible for detecting obstacles and stopping the train. The engineer or electromechanical device does exactly what he or it is supposed to (stop the train, but one person is injured) or the engineer or electromechanical device fails to do what he or it is supposed to (the train hits the tree and one person is injured). Observers want greater compensation when the engineer was responsible and when the injury occurred as a result of failure to detect the tree. Ritov et al. (1990) view these findings as evidence of the influence of normatively “irrelevant variables” on judgment. From a social contingency perspective, however, subjects have not fallen prey to a “bias.” They are trying to restore the original social contract between the victim and perpetrator. That contract specified that, in return for paying the fare, the owners of the train would transport the passengers to point,X and would do so as safely as current technology and nature permit. When either the engineer or electromechanical device fails to do the assigned job, that contract has been violated. In a modified replication of this study, we found similar results, with one noteworthy exception. Accountable subjects made more generous compensation awards whenever the accident was the result of a malfunction (either technical or human) that could conceivably be prevented in the future by creating incentives for better performance. When the accident was the result of an uncontrollable act of nature (the tree suddenly falling in front of the tree so that it was physically impossible to stop) or the routine operation of safety procedures, accountable subjects were actually less forthcoming in their awards than unaccountable subjects. These results suggest that accountable subjects are not simply engaging in generous grandstanding or anticorporate posturing; they seemed especially attuned to whether a norm had been violated and, if so, they were especially willing to make large compensation awards to restore the social status quo ante. In short, accountable subjects were vigilant, metanorm enforcers, society’s watchdogs.

C. THE DILUTION EFFECT From a multiple-regression perspective, there is little justification for reducing one’s confidence in the predictive power of a diagnostic cue when that cue is accompanied by a host of nondiagnostic ones. From a social functionalist perspective, however, this erosion of confidence may be justified. Far from representing a bias, the dilution effect may constitute a rational response to the interpersonal and institutional context. The presentation of information in dilution experiments can be likened to a conversation between researcher and subject in which subjects assume, following Grice’s (1975) “axioms of conversation,” that the information presented is indeed relevant to the task at hand. This assumption is hardly unreasonable. In most conversations, people refrain from

ACCOUNTABILITY

363

making statements that are utterly irrelevant to the interaction (indeed, the tendency to ramble on in disjointed fashion is taken as a sign of madness). Given that the experimenters deemed it appropriate to include an assortment of evidence in their communicationsto subjects, one would expect a good Bayesian to attach a high prior probability to the evidence being relevant to the task. The dilution effect-and its magnification in the accountability conditions-may be as much an expression of demand characteristics (“The experimenter expects me to use all the evidence in preparation for the conversation”) as of judgmental heuristics. In a recent experiment, we disentangled these diverging normative interpretations of the dilution effect by activating or deactivating conversational norms (Tetlock & Boettger, 1991). The experiment replicated the work of Tetlock and Boettger (1989), with the addition of one major independent variable: whether subjects were told that the information presented had been screened for relevance (explicit priming of conversational norms), whether subjects were told that the information presented had been randomly sampled from a computer data base (explicit deactivation of conversational norms), and whether subjects were told nothing one way or the other about the relevance of the information. As before, accountability magnified the dilution effect, in the no-priming condition and, to a lesser extent, in the explicit-priming-of-conversational-axiomscondition. Accountability did not, however, magnify the dilution effect when subjects were warned that the information had been randomly pulled together. These results suggest that accountability motivated subjects to integrate the available information only when there was a reasonably presumption of relevance. When that presumption was removed, compared to unaccountable subjects, accountable subjects actually became more resistant to the dilution effect.

D. AMBIGUITY AVERSION Ellsberg (1961) noted that one or more axioms of utility theory are violated whenever people make choices solely to avoid ambiguity. Ambiguity and vagueness about probabilities, even when expressed as second-order probabilities, are formally irrelevant in decision analysis but play a large role in everyday decision-making. There is considerable evidence, for example, that avoidance of ambiguity affects public policy. Insurers are unwilling to cover chemical and waste-processing firms because it is so difficult to specify the distribution of anticipated claims (Kunreuther, 1987). Einhom and Hogarth (198 1) have also suggested that ambiguity aversion explains why some technologies are feared more than their first-order probabilities of failure or accident warrant. One interpretation is that dislike of ambiguity is simply a pure preference, like the preference for a “fair” coin over a coin biased in some unknown way. Another interpretation is suggested by the theory proposed by Loomes and

364

PHILIP E. TETLDCK

Sugden (1982), in which transitivity is sacrificed to accommodate feelings of regret. Decision-makers, in this scheme, avoid ambiguity because they can anticipate the deep regret they would feel if a worse-case scenario materialized (if they decided to place bets on a coin that was severely biased against them). It should not be surprising from a social contingency model perspective that accountability often exacerbates ambiguity avoidance (as Curley, Yates, & Abrams, 1986, found). Accountable decision-makers are more likely to anticipate how difficult it would be to justify choosing an option that led to a worstcase outcome. One would stand accused of recklessness.

E. THE ATTRACTION EFFECT Imagine a choice between two major alternatives, A and B. A is superior to B on one major dimension of evaluation (say, less expensive) whereas B is superior to A on the other major dimension of evaluation (say, quality). One confronts an especially difficult trade-off, because the options have been carefully calibrated to have offsetting strengths and weaknesses. People find the choice much easier, however, if one introduces a (logically irrelevant) third option, C, that is inferior in both cost and quality to B, but inferior in only one respect to A. B becomes more attractive vis-8-vis A by virtue of being the dominant option with respect to c. From a rigorous rational actor perspective, people who fall prey to the attraction effect have committed an error (the principle of regularity). Why should a difficult trade-off become easier by adding an inferior third option? From a social functional perspective, the same behavior makes a good deal of sense. As Simonson (1989) notes, it has become easier to justify selecting B than A. B “dominates” C whereas A does not. Simonson shows experimentally that holding subjects accountable for their choices leads to a magnification of the attraction effect. There is a kind of social rationality at work. People seem to reason, “I’m less likely to be blamed if I choose the option that dominates at least one other option than if I choose the option that requires trade-offs no matter what comparison I made. In justifying B, I can point to its unequivocal superiority over C.” From social experience in both giving and receiving accounts (Schlenker, 1982; Tetlock, 1981), people have learned that one is less likely to blamed when one can offer a clear-cut, sensible-sounding set of reasons for one’s conduct.

F. THE CONJUNCTION FALLACY In an influential series of studies, ’ILersky and Kahneman (1982, 1983) found that lay judgments of probability systematically violated the conjunction rule,

ACCOUNTABILITY

365

which they considered to be the “simplest and most fundamental quantitative rule of probability” (1983, p. 294). They demonstrated, for example, that many subjects rated the conjunction of two events as more probable than one of the individual events. According to the conjunction rule, the probability that A&B is true cannot be greater than the probability that A is true because the conjunction is logically included by the single element. A subset cannot be larger than its superset, therefore any element cannot have a greater probability of being a member of a subset than its superset. The most cited example of the conjunction fallacy involves a description of a woman who appears to be a social activist. Tversky and Kahneman (1983) asked subjects to rank or rate the probability that the woman is a bank teller as opposed to a bank teller who is active in the feminist movement. One version of the task runs as follows: Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations. Subjects were then asked to check which of the following two alternatives was more probable: Linda is a bank teller (t) Linda is bank teller and is active in the feminist movement (t&f) In this version, 85% of subjects rated the conjunction of constituents (t&f) as more probable than the single constituent (t). Tversky and Kahneman (1983) argued that subjects made this error in probabilistic reasoning because they relied on the representativenessheuristic. Because, “Linda is a bank teller and is active in the feminist movement” (t&f) resembles the personality description of Linda better than “Linda is a bank teller” (t), subjects judge that it is more likely that she is a bank teller and a feminist (t&f) than a bank teller (t). In a series of studies, Dulany and Hilton (1989) challenged the characterization of the conjunction effect as a fallacy. They argued that the experimental manipulations designed by ’hersky and Kahneman ( 1983) to activate the representativeness heuristic also activate conversational norms that render subjects’ judgments quite reasonable. Dulany and Hilton find that those subjects who are most likely to commit the conjunction fallacy are also the most likely to draw “implied negation” and “empirical possibility” implications from the phrasing of the conjunct. Subjects implicitly assume that the same norms that regulate conversations in everyday life regulate the encounter with the experimenter.

366

PHILIP E. TETLOCK

When the experimenter presents information, they assume that the experimenter is trying to make the most informative statement possible (thus observing the Grician maxim of quantity), would only make statements relevant to the topic under discussion (thus observing the Grician maxim of relevance), would not say anything that he or she knows to be false (thus observing the maxim of truthfulness), and would refrain from making statements for which he or she lacks adequate evidence (thus observing the maxim of quality). When the experimenter asks subjects to judge the relative likelihood of Linda being a bank teller and of Linda being a bank teller and a feminist, many subjects attached an “implied negation” interpretation to the former statement, understanding it to mean Linda is a bank teller but not a feminist. Dulany and Hilton also show that they can drastically reduce the size of the conjunction effect when they ask subjects to say how strongly they believe the conjunction as opposed to both the “logical possibility” and the “epistemic uncertainty” phrasings of the constituent. With these revised phrasings, only about a third as many subjects display the conjunction fallacy as compared to the original Tversky and Kahneman (1983) research. These resuIts underscore the importance of conversational pragmatics in judgment and choice research. People bring to the experiment an elaborate repertoire of assumptions concerning how people communicate with each other. In everyday life, these assumptions serve people reasonably well. In experiments, these assumptions can lead people wildly astray. There is no reason, from this perspective, to expect accountability manipulations to attenuate the conjunction effect. If anything, accountability manipulations may magnify the effect by further underscoring the conversational nature of the interaction.

G. OVERUSE OF INDIVIDUATING INFORMATION Just as conversational norms cast doubt on the irrationality of the dilution and conjunction effects, conversational norms also help to explain the apparent overreliance on nondiagnostic individuating information at the expense of base rate information. For instance Schwartz, Strack, Hilton, and Naderer (1991) manipulated the applicability of Grice’s cooperative principle to explore whether conversational norms contribute to the relative neglect of base rate information. They presented the nondiagnostic individuating information either as a statement that had been written by a psychologist for this particular task or as a random sample of information drawn by a computer. As one would expect from a Grician analysis, subjects relied more on the personality information and less on base rate information in the former than in the latter case, presumably because the human communicator but not a computer is supposed to conform to conversational norms and to provide information that is informative, truthful, and relevant.

ACCOUNTABILITY

367

Krosnick, Li, and Lehman (1990) reported seven experiments that provided further support for a conversational norm interpretation of base rate neglect. They noted that most earlier studies confounded the base rate/individuating information manipulation with order of presentation by always giving the base rate information first. They demonstrated a receny effect and made a strong case that this effect is the result not of differential availability of information in memory, but rather of subjects’ inferences (based on conversational conventions) that the experimenter believes that subjects should rely most on the information presented last. From a social contingency perspective, experimental researchers frequently violate fundamental norms of conversation (e.g., by presenting information to subjects that is not relevant in a context that suggests otherwise) or activate norms that mislead subjects (e.g., diverting attention from relevant information). The danger here is that researchers will draw spurious inferences about cognitive limitations and shortcomings. The alternative interpretation is that subjects are simply using the cues available to them in a reasonable manner.

H. THE STATUS QUO EFFECT A fundamental property of the rational choice model, under certainty or uncertainty, is that only preference-relevant features of the alternatives should influence the individual’s decision. Thus, neither the order in which the alternatives are presented nor any arbitrary labels they may carry should affect the individual’s choice. Of course, in real-world decision problems, the alternatives often come with influential labels. Indeed, one alternative inevitably canies the label status quo-that is, doing nothing or maintaining one’s current or previous decision is almost always a possibility. In a series of experimental studies of financial decision, Samuelson and Zeckhauser ( 1987) have shown that faced with new options, decision-makers often stick disproportionately with the status quo alternative. This preference for the status quo is not, of course, automatic evidence of irrationality. Decisionmakers might prefer the status quo in order to avoid transaction costs or uncertainty. Samuelson and Zeckhauser, however, eliminated these alternative explanations experimentally. From a political perspective, the preference for the status quo has a number of additional rational sources. Decision-makers may prefer to persist with current policy in order to justify the sacrifices that they have already made on behalf of that policy. Such retrospective rationality is highly consistent with the defensive bolstering strategy of coping with anticipated accountability demands. Another factor contributing to commitment to the status quo is regret avoidance. Occasionally people find themselves in the unpleasant position of regretting the conse-

368

PHILIP E. TETLOCK

quences of previous decisions. These “lessons of experience” teach people to avoid, if possible, regrettable consequences. Indeed, there is substantial evidence (Bell, 1982; ’hersky, & Kahneman, 1982) that regret avoidance influences decision-making. For instance, people tend to avoid decisions in which they could appear after the fact to have made the wrong choice, even if in advance the decision appeared correct given the information available (see the discussion of ambiguity aversion, Section V,D). And, as ’hersky and Kahneman (1982) argue in their discussion of the simulation heuristic, people feel greater regret for bad outcomes that are the result of new actions than for similar outcomes resulting from inaction. Avoidance of decision regret is therefore an additional cause of the status quo effect. It favors adherence to status quo norms or routine behavior at the expense of innovation, and it reinforces the inclination to conform to social norms. This reasoning suggests that accountability demands should magnify rather than attenuate the status quo effect. This prediction is consistent with what Tetlock and Boettger (1991) found in their experimental simulation of decisions to admit or reject drugs for the United States pharmaceutical market. Subjects in general, and accountable subjects in particular, were much more tolerant of risk from drugs already on the market than from drugs not yet admitted onto the market. This prediction is also consistent with several psychological studies of policy controversies. Ritov et al. (1990) report a set of experiments on the reluctance to vaccinate that reveal people to be much more tolerant of the risks of doing nothing than they are of the risks of action. Deaths caused by intentional human action (vaccination) are weighed more heavily than deaths caused by allowing an epidemic to run its natural course-a tendency that medical malpractice laws apparently encourage. And Howard, Matheson, and North (1972) report an insightful case study of why the United States government was so reluctant to seed hurricanes to divert storms from more to less populated areas. Although seeding hurricanes makes good sense in a cost-benefit framework, it makes much less sense in a legal and political system that is more tolerant of sins of omission (let nature take its course) than sins of commission (imagine the protests of the victims of the diverted hurricanes and the legal liability problems). In short, political institutions can both reflect and amplify the cognitive processes that fuel the status quo effect.

VI. Concluding Comments This article makes the case for a new metaphor in the study of judgment and choice: that of the politician. The arguments advanced have been empirical,

ACCOUNTABILITY

369

theoretical, and normative. At an empirical level, the politician metaphor sensitizes us to social and institutional boundary conditions on cognitive tendencies often hypothesized (at least by default) to be universal. At a theoretical level, the politician metaphor alerts us to new ways of thinking about cognitive processes. A major function of thought is the generation of good reasons or justifications for what one plans to do or has already done. At a normative level, the politician metaphor warns us not to view accountability as a panacea, but rather to give careful thought to the incentives that accountability systems create. The metaphor also raises serious doubts about the classification of many response tendencies as errors or biases within the cognitive research program. In addition, this article advances a testable middle-range theory predicated on the politician metaphor: the social contingency model of judgment and choice. This model does not map neatly into any of the traditional levels of analysis: the individual, the small group, the organization, and political system. The unit of study is the individual in relation to these social milieux. As such, the model is an eclectic creation. The model borrows, qualifies, and elaborates on the cognitive miser image of the thinker that has been so influential in experimental work on social cognition. The model adopts the approval and status-seeker image of human nature that has been so influential in role theory, symbolic interactionism, and impression management theory. The model draws on sociological and anthropological theory concerning the necessary conditions for social order in positing accountability to be a universal feature of natural decision environments. Taken together, these disparate themes define a distinctive approach to judgment and choice that identifies three major strategies of coping with accountability predicaments, the conditions under which each is likely to be activated, and the functional pros and cons of each strategy. The social contingency model is not tightly linked to any particular methodology (in contrast to the cognitive research program, which is closely tied to laboratory experimentation). The theoretical eclecticism of the model demands a corresponding commitment to methodological eclecticism. Controlled experimentation can play a key role by clarifying situational determinants of the strategies people adopt for coping with accountability. Investigators can isolate the effects of independent variables (e.g., pre- versus postdecisional accountability, knowledge of the views of those to whom one is accountable), systematically assess interactions between independent variables, and test detailed models of the processes mediating relationships between independent and dependent variables. But laboratory work by itself is inadequate. Other methods are equally crucial. We need to assess individual differences in propensities to adopt coping strategies and the possible interactive effects of personality and situational variables. We need field experiments to assess the generality and robustness of laboratory findings as well as the effects of normative interventions such as

370

PHILIP E. TETWCK

multiple advocacy on organizational decision processes. We need comparative (cross-sectional and historical) ethnographic studies of organizational culture to document the diverse forms that accountability relationships take and the styles of decision-making associated with these relationships. To state an often violated dictum, one’s choice of method should be tailored to the theoretical problem at hand. The social contingency model poses problems that cross disciplinary boundaries and that require a plurality of methodologies. In closing it is worth considering the potential problem of proliferating metaphors in social psychological theory. There is, in principle, no limit to the range of metaphors one can draw on in formulating the hard-core premises of research programs. At what point do we reach the point of diminishing returns? My answer is that we reach this point when we have vigorous research programs that address the full range of human functions of judgment and choice: to help people advance their material self-interest (the economist), to help people make sense of the surrounding world (the psychologist), to help people get along with others (the politician), and to help people live with themselves (satisfaction of underlying personality needs and values). This list bears a marked and by no means coincidental resemblance to the classic functionalist theories of attitudes. As a final goal of theory building, of course, mere open-mindedness is not enough. Empirical turf disputes among research programs grounded in different metaphors are bound to arise-as indeed they already have in controversies over attribution biases (cold cognitive versus hot motivational explanations; Tetlock & Levi, 1982), over “dissonance,” “reactance,” and “equity” effects (intrapsychic versus impression management explanations; Tetlock & Manstead, 1985), and over evidence of human concern for the well-being of others (egoistic versus altruistic explanations; Caporael, Dawes, Orbell, & van de Kragt, 1989; Etzioni, 1988; Tetlock, 1989b). Such disputes are notoriously difficult to resolve and raise fundamental questions about the nature of progress within our science. When has a metaphor been overextended? There are no precise or infallible indicators, but there are suggestive clues. A metaphor has been exhausted when its advocates cease to make new empirical discoveries and devote increasing proportions of their time to post hoc (semantic relabeling) interpretations of anomalous findings. To paraphrase a proverb of William Blake, one never knows one has had enough of a research program until one has had more than enough.

Acknowledgments I appreciate the helpful comments of Randall Peterson and Mark Zanna on an earlier version of this article. I also owe a debt to John Darley for our conversations.

ACCOUNTABILITY

37 1

References Abelson, R. P., & Levi, A. (1985). Decision-making and decision theory. In E. Aronson & G. Lindzey (Eds.), Handbook of socialpsychology (3rd ed., pp. 23 1-3 10). New York: Random House. Adelberg, S., & Batson, C. D. (1978). Accountability and helping: When needs exceed resources. Journal of Personality and Social Psychology, 36, 343-350. Allison, G. T. (1971). Essence of decision. Boston: Little, Brown. Ailport, G. W. (1937). Personality: A psychological inrerprefarion. New York: Holt. Anderson, P. A. (1981). Justifications and precedents as constraints in foreign policy decisionmaking. American Journal of Political Science, 25, 738-761. Arkes, H. R., & Blumer, C. (1985). The psychology of sunk costs. Organizarional Behavior and Human Performance, 35, 124-140. Atkinson, R. C., & S h i f i n , R . M. (1968). Human memory: A proposed system and its control processes. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and morivations: Advances in rheory and research (Vol. 2). New York: Academic Press. Axelrod, R. (1986). The evolufion of cooperation. New York Basic Books. Azjen, I., & Fishbein, M. (1975). A Bayesian analysis of attribution processes. Psychological Bullerin, 82(2), 261-277. Bales, R . F. (1970). Personality and interpersonal behavior. New York: Holt, Reinhart, & Winston. Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215. Baskett, G. D. (1973). Interview decisions as determined by competency and attitude similarity. Journal of Applied Psychology, 57, 343-345. Baumeister, R. F. (1 982). A self-presentational view of social phenomena. Psychological Bullerin, 91, 3-26. Bell, D. E. (1982). Regret in decision making under uncertainty. Operarions Research. 30(5), 961981. Blau, P. (1964). Exchange and power in social life. New York: Wiley. Bourne, E. (1977). Can we describe an individual’s personality? Agreement on stereotype versus individual attributes. Journal of Personality and Social Psychology, 35, 863-872. Brockner, J., & Rubin, J. Z. (1985). Enfrapmenr in escalaring conflicrs: A social psychological analysis. New York: Springer-Verlag. Byrne, D., Nelson, D., & Reeves, K. (1966). Effects of consensual validation and invalidation on attraction as a function of verifiability. Journal of Experimental Social Psychology. 2 , 98- 107. Caporael, L. R., Dawes, R., Orbell, J. M., & van de Kragt, J. C. (1989). Selfishness examined: Cooperation in the absence of egoistic incentives. Brain and Behavioral Sciences, 12, 683698. Camevale, P. J. D. (1985). Accountability and the dynamics of group presentation. In E. J. Lawler (Ed.), Advances in Group Process (Vol. 2, pp. 227-248). Camevale, P. J. D., Pruitt, D. G., & Seilheimer, S. D. (1981). Looking and competing: Accountability and visual access in integrative bargaining. Journal of Personality and Social Psychology, 40,111-120. Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766. Cialdini, R . B., Levy, A,, Herman, C. P., Kozlowski, L. T., & Petty, R . E. (1976). Elastic shifts of opinion: Determinants of direction and durability. Journal of Personality andSocial Psychology, 34,663-672.

372

PHILIP E. TETLOCK

Cialdini, R. B., Petty, R. E., & Cacioppo, J. T. (1981). Attitude and attitude change. AnnualReview of Psychology, 32, 357-404. Curley, S. P., Yates, F., & Abrams, R. A. (1986). Psychological sources of ambiguity avoidance. Organizational Behavior and Human Decision Processes, 38, 230-256. Cvetkovich, G. (1978). Cognitive accommodation, language, and social responsibility. Social Psychology, 41, 149-155. Dulaney, D. E., & Hilton, D. J. (1991). Conversational implicature, conscious representation, and the conjunction fallacy. Social Cognition, 9(1), 85-1 10. Einhorn, H., & Hogarth, R. M. (1981). Behavioral decision theory. Annual Review of Psychology, 31, 53-88. Ellsberg, D. (1961). Risk, ambiguity and the Savage axioms. Quarterly Journal of Economics, 75, 643-669.

Etzioni, A. (1988). The moral dimension. New York: Free Press. Festinger, L. (Ed.). (1964). Conflicts, decision, and dissonance. Stanford, CA: Stanford University Press. Fischhoff, B. (1982). Debiasing. In D. Kahneman, P. Slovic, & A. 'hersky (Eds.), Judgment under uncertainty: Heuristic and biases (pp. 422-444). New York: Cambridge University Press. Fischhoff, B., Lichtenstein, S., Slovic, P., Derby, S., & Keeney, R. (1981). Acceptable risk. New York Cambridge University Press. Fischhoff, B., Slovic, P., & Lichtenstein, S. (1980). Knowing what you want: Measuring labile values. In T. Wallsten (Ed.), Cognitive processes in choice and decision behavior (pp. 117141). Hillsdale: NJ: Erlbaum. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Influence of information and motivation on attention and interpretations. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1-74). San Diego: Academic Press. Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: McGraw-Hill. Fox, F., & Staw, B. M. (1979). The trapped administrator: The effects of job insecurity and policy resistance upon commitment to a course of action. Administrative Science Quarterly. 24, 449471.

Funder, D. C. (1987). Errors and mistakes: Evaluating the accuracy of social judgment. Psychological Bulletin. 101, 75-91. George, A. L. (1980a). Presidential decision-making in foreign policy: The effective use of information and advice. Boulder, CO: Westview Press. George, A. L. (1980b). Domestic constraints on regime change in U.S. foreign policy: The need for policy legitimacy. In 0. R. Holsti, R. M. Siverson, & A. L. George ( a s . ) , Change in the international system. Boulder, CO: Westview Press. Graber, D. (1976). Verbal behavior and politics. Urbana: University of Illinois Press. Graham, M. (1991, January). The quiet drug revolution. Atlantic Monthly, 256, 34-40. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603-618. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics: 3 Speech acts (pp. 41-58). New York: Academic Press. Hagafors, R., & Brehmer, B. (1983). Does having to justify one's decisions change the nature of the judgment process? Organizational Behavior and Human Performance. 31, 223-232. Hare, A. P. (1976). Handbook of small group research (2nd ed.). New York: Free Press. Harvey, J. H., Harkins, S. G., & Kagehiro, D. K. (1976). Cognitive tuning and the attribution of causality. Journal of Personality and Social Psychology, 34, 708-715. Hastie, R. (1983). Social inference. Annual Review of Psychology, 34, 161-188. Heider, F. (1944). Social perception and phenomenal causality. Psychological Review, 51,358-374.

ACCOUNTABILITY

373

Heider, F. (1958). The social psychology of interpersonal relations. New York: Wiley. Higgins, E. T.,& Bargh, J. A. (1987). Social cognition and social perception. Annual Review of Psychology, 38, 369-425. Hogarth, R. (1990). Insights in decision making: A tribute to Hillel J . Einhorn. Chicago: University of Chicago Press. Howard, R., Matheson, J. E., &North, D. W. (1972). The decision to seed hurricanes. Science, 176, 1191-1202.

Janis, 1. L. (1982). Groupthink (2nd ed.). Boston: Houghton Mifflin. Janis, 1. L. (1989). Crucial decisions: Leadership in policy making and crisis management. New York: Free Press. Janis, I. L., & Mann, L. (1977). Decision making: A psychological analysis of conflict, choice, and commitment. New York: Free Press. Jones, E. E. (1979). The rocky road from acts to dispositions. American psychologist. 34(2), 107117.

Jones, E. E., & Davis, K. E. (1965). From acts to dispositions: The attribution process in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp, 220-266). New York: Academic Press. Jones, E. E., & Wortman, C. (1973). Ingratiation: An artriburional approach. Morristown, NJ: General Learning Press. Kahneman, D., Knetch, J., & Thaler, R. (1986). Fairness and the assumptions of economics. In R. M. Hogarth & M. W. Reder (Eds.), Rational choice: The conflict between economics and psychology (pp. 101-1 126). Chicago: University of Chicago Press. Kahneman, D., & 'hersky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica. 47, 263-291. Kahneman, D., 'hersky, A,, & Slovic, P. (1982). Judgment under uncertainfy. Cambridge, England: Cambridge University Press. Katz. D., & Kahn. R. L. (1978). The socialpsychology oforganizations (2nd ed.). New York: Wiley. Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska Symposium on Motivation (Vol. 15, pp. 192-240). Lincoln: University of Nebraska Press. Kelley, H. H. (1971). Attribution in social interaction. In E. E. Jones, D. E. Kanause, H.H.Kelley, R. E. Nisbett. S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior. Morristown, NJ: General Learning Press. Kiesler, A. C. (1971). The psychology of commitment. New York: Academic Press. Klimoski, R. J. (1971). The effects of intragroup forces on intergroup conflict resolution. Organizational Behavior and Human Peflormance. 8, 363-383. Krosnick, J., Li, F., & Lehman, D. (1990). Conversational conventions, order of information on, and the effect of base rates and individuating information on social judgment. Journal of Personality and Social Psychology, 59, 1140- 1152. Kruglanski, A. W. (1990). Lay epistemic theory in social-cognitive psychology. Psychological Inquiry. Kunda, Z. (1990). The case for motivated reasoning. Psychological Bullerin, 108(3), 480-498. Kunreuther, H.(1987). The failure of EIL coverage: Gridlock in environmental insurance. Environment 29(1), 18-20, 31-34. Lakatos, 1. (1970). Falsification and the methodology of scientific research programs. In I. Lakatos & A. Musgrave (Eds.), Criticism and the growth of knowledge (pp. 91-196). Cambridge, MA Cambridge University Press. Langer, E. (1990). Mindfulness. Reading, MA: Addison-Wesley. Lindblom, C. E. (1959). The science of "muddling through." Public AdministrationReview, 19,7988.

Linton, R. (1945). The cultural background of personality. New Yo& Appleton-Century.

374

PHILIP E. TETLOCK

Loomes, G., & Sugden, R. (1982). Regret t h e o j : An alternative theory of rational choice under uncertainty. Economic Journal, 92, 805-824. Maheswaran, D., & Chaiken, S. (1991). Promoting systematic processing in low-motivation settings: Effects of incongment information on processing and judgment. Journal of Personality and Social Psychology, 61, 13-25. March, J . , & Olsen (1989). Rediscovering institutions. Stanford, CA: Stanford University Press. March, I., & Simon, H. (1958). Organizations. New York Wiley. Markus, H., & Zajonc, R. B. (1985). The cognitive perspective in social psychology. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (Vol. 1, pp. 137-230). New York: Random House. McAllister, P. W., Mitchell, T. R., & Beach, L. R. (1979). The contingency model for the selection of decision strategies: An empirical test of the effects of significance, accountability, and reversibility. Organizational Behavior and Human Performance. 24, 228-244. McGuire, W. J. (1968). Theory of the structure of human thought. In R. P. Abelson et al. (Eds.), Theories of cognitive consistency: A sourcebook (pp. 140- 162). Chicago: Rand McNally. Meyer, 1. W.. & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340-363. Mills, C. W. (1940). Situated identities and vocabularies of motives. American Sociological Review, 5 , 904-913. Moscovici, S . (1985). Social influence and conformity. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed., pp. 347-412). New York: Random House. Neuberg, S. L., & Fiske, S. T. (1987). Motivational influences on impression formation: Dependency, accuracy-driven attention, and individuating information. Journal of Personality and Social Psychology. 53, 431-444. Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgment. New York: Appleton-Century-Crofts. Nisbett, R. E., Zukier, H., & Lemley, R. (1981). The dilution effect: Nondiagnostic information. Cognitive Psychology, 13, 248-277. Payne, I. W., Bettman, 1. R., & Johnson, E. I. (in press). Behavioral decision research: A constructive processing perspective. Annual Review of Psychology. Pepitone, A. (1976). Toward a normative and biocultural social psychology. Journal of Personality and Social Psychology, 34, 611-653. Peny, R. E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary approaches. Dubuque, IA: Brown. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123-205). Orlando, FL: Academic Press. Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement as a determinant of argument-based persuasion. Journal of Personality and Social Psychology, 41, 847-855. Pfeffer, J. (1981). Management as symbolic action: The creation and maintenance of organizational paradigms. In L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 3). Greenwich, CT: JAI Press. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations:A resource dependence perspective. New York: Harper & Row. Pruitt, D. (1981). Negotiation behavior. New York: Academic Press. Pruitt, D. G., & Lewis, S. A. (1975). Development of integrative solutions in bilateral negotiation. Journal of Social Psychology, 31(4), 621-633. Rapport, A. (1964). Fights. games, and debates. Ann Arbor: University of Michigan Press. Ritov, I., Hodes, & Baron, J. (1990). Reluctance to vaccinate: Omission bias and ambiguity. Journal of Behavioral Decision Making, 3, 263-278.

ACCOUNTABILITY

375

Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution process. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 10, pp. 174221). New York: Academic Press. Rozelle, R. M., & Baxter, J. C. (1981). Influence of role pressures on the perceiver: Judgments of videotaped interviews varying judge accountability and responsibility. Journal of Applied Psychology, 66, 437-441. Samuelson, W., & Zeckhauser, R. (1987). Status quo bias in decision making. Journal of Risk and Uncertainly. 1, 7-59. Schlenker, B. R. (1982). Translating actions into attitudes: An identity-analytic approach to the explanation of social conduct. In L. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 15, pp. 194-248). New York: Academic Press. Schroeder, H., Driver, M.,& Streufert, S. (1967). Human informarionprocessing. New York: Holt, Rinehart, & Winston. Schwartz, N., Strack, F., Hilton, D., & Naderer, G. (1991). Baserates, representativeness, and the logic of conversation: The contextual relevance of irrelevant information. Social Cognition. 9, 67-84.

Scott, M., & Lyman, S. (1968). Accounts. American Sociological Review, 33, 46-62. Semin, G. R., & Manstead, A. S. R. (1983). The accountabiliry of conduct: A social psychological analysis. London: Academic Press. Shaw, M. E., & Costanzo, P. (1982). Theories of social psychology (2nd ed.). New York: McGrawHill. Sherif, M.. & Cantril, H. (1947). The psychology of ego-involvements. New York: Wiley. Showers, C., & Cantor, N. (1985). Social cognition: A look at motivated strategies. Annual Review of Psychology. 36, 215-305. Siebert, J. E. (1974). Effects of decision importance on ability to generate warranted subjective uncertainty. Journal of Personality and Social Ps.vchology, 30, 688-694. Simonson, J. (1989). Choice based on reasons: The case of attraction and compromise effects. Journal of Consumer Research, 16, 158- 174. Snyder, R. C., Bruck, H. W., & Sapin, B. (Eds.). (1962). Foreign policy decision-making. New York: Free Press. Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Cambridge, MA: Harvard University Press. Spranca, M., Minsk, E., & Baron. J. (1991). Omission and commission in judgment and choice. Journal of Experimental Social Psychology. 27, 76- 105. Staw, B. M. (1976). Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance. 16, 27-44. Staw, B. M. (1980). Rationality and justification in organizational life. In B. M . Staw and L. Cummings (Eds.), Research in organizational behavior (Vol. 2). Greenwich, C T JAI Press. Staw, B. M., Sandelands, L. E., & Dutton, J. E. (1981). Threat-rigidity effects in organizational behavior: A multilevel analysis. Administrative Science Quarterly. 26, 501-524. Stryker, S., & Gottlieb, A. (1981). Attribution theory and symbolic interactionism: A comparison. In J. H. Harvey, W. Ickes, & R. F. Kidd (Eds.), New directions in attribution theory (Vol. 3). Hillsdale, NJ: Erlbaum. Suppe, F. (1973). The structure of scientific theories. Urbana: University of Illinois. Svenson, 0. ( 1979). Process descriptions of decision making. Organizational Behavior and Human Performance, 23, 86-1 12. Teger, A. (1980). Too much invested to quit. New York: Pergamon Press. Tetlock, P. E. (1981). Influence of self-presentation goals on attributional reports. Social Psychology Quarterly. 44,300-3 1 I .

376

PHILIP E. TETLOCK

Tetlock, P. E. (1983a). Accountability and the complexity of thought. Journal of Personality and Social Psychology. 45, 74-83. Tetlock, P. E. (1983b). Accountability and the perseverance of first impressions. Social Psychology Quarterly. 46,285-292. Tetlock, P. E. (1985). Accountability: A social check on the fundamental attribution error. Social Psychology Quarterly, 48, 227-236. Tetlock, P. E. (1989a). Methodological themes and variations. In P. E. Tetlock, R. Jervis, C. Tilly, P. Stem, & J. Husbands (Eds.), Behavior, society, and nuclear war (Vol. I ) . New York Oxford University Press. Tetlock, P. E. (1989b). The selfishness-altruismdebate: In defense of agnosticism. Brain and Eehavioral Sciences, 12, 723-725. Tetlock, P. E., & Boettger, R. (1989). Accountability: A social magnifier of the dilution effect. Journal of Personality and Social Psychology: Attitudes and Social Cognition, 57, 388-398. Tetlock, P. E., & Boettger, R. (1991). Accountability ampliJFes the status quo effect when changes create victim. Unpublished manuscript, University of California, Berkeley. Tetlock, P. E., & Kim, J. (1987). Accountability and overconfidence in a personality prediction task. Journal of Personality and Social Psychology: Attitudes and Social Cognition. 52, 700-709. Tetlock, P. E., & Levi, A. (1982). Attribution bias: On the inconclusiveness of the cognitionmotivation debate. Journal of Experimental Social Psychology, 18, 68-88. Tetlock, P. E., & Manstead, A. S. R. (1985). Impression management versus intrapsychic explanations in social psychology: A useful dichotomy? Psychological Review. 92, 58-79. Tetlock, P. E., Skitka, L., & Boettger, R. (1989). Social and cognitive strategies of coping with accountability:Conformity, complexity, and bolstering. Journal of Personality and Social Psychology: Interpersonal Relations and Group Dynamics, 57, 632-64 1. Thaler, R. (1985). Mental accounting and consumer choice. Marketing Science, 4(2), 199-214. 'hersky, A., & Kahneman, D. (1982). Judgments of and by representativeness. In D. Kahneman, P. Slovic, & A. 'hersky (Eds.), Judgments under uncertainry: Heuritics and biases (pp. 84100). New York: Cambridge University Press. 'hersky, A., & Kahneman, D. (1983). Extension versus intuitive reasoning: The conjunction fallacy in probability judgement. Psychological Review, 90, 293-315. 'hersky, A,, and Kahneman, D. (1986). Rational choices and the framing of decisions. Journal of Business, 59(4), 25 1-278. Veblen, T. (1889). The theory of the leisure class. New York: Macmillan. Weick, K. E. (1979). Cognitive processes in organizations. In B. M. Staw (Ed.), Research in organizational behavior. Greenwich, CT: JAI Press. Wicklund, R. A., & Brehm, J. W. (1976). Perspectives on cognitive dissonance. Hillsdale, NJ: Erlbaum. Zajonc, R. B. (1960). The process of cognitive tuning in communications. Journal of Abnormal and Social Psychology, 61, 159- 167. Zetterberg, H. L. (1957). Compliant actions. Acta Sociologica, 2, 188-192. Zukier, H. (1982). The dilution effect: The role of the correlation and the dispersion of predictor variables in the use of nondiagnostic information. Journal of Social Psychology, 43, 11631174.

INDEX

A

theory development, 4-5, 8-9, 14-15 value priorities, 55, 57 Acquaintance process, behavioral confirmation and, 87-88, 93, 101 functions, 73-75 perceivers, 77, 81 Action inhibition, thought suppression and, 196-197, 201-202 Adaptation accountability and, 337 Milgram and, 310 Adaptiveness, accountability and, 342-344, 348-349 Addictions, thought suppression and, 213 Affect-as-information model, social judgments and, 231 mood congruence, 242-243, 247-248 research progress, 256, 261 Affect in social judgments, 227-229, 267269 contemporary accounts, 23 1-232 early research, 230-231 mood and emotion, 229-230 mood congruence, 242-243 direct informational effects, 246-247 indirect informational effects, 243-246 processing consequences, 247-249 multiprocess model, 232 assumptions, 232-234 implications, 236-238 strategies, 234-236 variables, 238-242 research program, 249-250 extensions, 263-267 informational effects, 250-256 processing differences, 256-263 Affect-priming model, social judgments and, 231-232

Abstractness, values and, 4 Accessibility, thought suppression and, 207209, 214 Accountability, 33 1-333, 368-370 cognitive research metaphors, 333-335 coping strategies, 347-348 acceptability heuristic, 348-35 1 preemptive self-criticism, 351-359 rationalization heuristic, 359 normative issues, 359-360 ambiguity aversion, 363-364 attraction effect, 364 attribution error, 360-361 biases in compensation decisions, 361362 conjunction fallacy, 364-366 dilution effect, 362-363 individuating information, 366-367 status quo effect, 367-368 political research, 335-336 acceptability heuristic, 340-342 coping strategies, 339-340 hard core, 336-339 hybrid responses, 346-347 preemptive self-criticism, 343-344 rationalization heuristic, 344-346 Accuracy affect in social judgments and, 238, 241242, 248 behavioral confirmation and, 99- 100 relational model of authority and, 139 Achievement affect in social judgments and, 254-256 values and empirical studies, 23, 27-28 theory assessment, 40-44,48 377

378

INDEX

mood congruence, 242-246 research progress, 250-256 Affective state, affect in social judgments and, 238 Age behavioral confirmation and, 68 values and, 18-20, 48, 54 Agency, Milgram and, 279 Agentic state, Milgram and, 279, 299, 302304 Aggression Milgram and, 299-302 thought suppression and, 213 Altman, Milgram and, 31 I Ambiguity aversion, accountability and, 360, 363-364 Ambiguous targets, affect in social judgments and, 238 Amnesia, thought suppression and, 195 Antisocial behavior, television and, 3 12-3 13, 3 17-3 I8 Anxiety accountability and, 347 affect in social judgments and, 256 thought suppression and, 213 Arousal, values and, 7, 14, 44 Asch, Solomon, influence on Milgram, 284286, 297, 318 Associations, affect in social judgments and, 250, 259, 261, 268 Associative effects, affect in social judgments and, 244 Attention affect in social judgments and mood congruence, 244, 247-248 multiprocess model, 240-241 research progress, 250-251 thought suppression and, 215, 217 instigation, 197, 20 I processes, 203-205 Attitudes accountability and, 370 coping strategies, 349, 355-356 political research, 342-346 affect in social judgments and, 235, 240, 249 behavioral confirmation and context, 95-97, 104 disconfirmation, 88-89 functional analysis, 71-72

Milgram and, 313-314 relational model of authority and, 118-1 19, I24 empirical studies, 125-128, 130, 137 procedural justice, 141, 158 Attraction effect, accountability and, 360, 364 Attributes affect in social judgments and, 248 behavioral confirmation and context, 96-97, 99, 101 disconfirmation, 88-89 implications, 93 Attribution error, fundamental, accountability and, 352, 354-356, 360-361 Attributions accountability and, 331, 355, 370 overattribution effect, 355-356, 361 affect in social judgments and, 267 mood congruence, 246, 249 research progress, 251, 254-255, 266 behavioral confirmation and, 102 Milgram and, 298 Authority Milgram and characteristics, 278-279, 282, 287 contributions, 293-308, 3 10 relational model of. see Relational model of authority in groups Autocratic groups, relational model of authority and, 118-119 Automatic target search, thought suppression and, 217 postsuppression processes, 2 14 processes, 203-204, 206-21 I Autonomy, values and, 7

B Behavior accountability and, 331, 345, 354-355, 361 affect in social judgments and, 252-253, 263 Milgram and, 319 characteristics, 278-279, 282, 284-286, 289 contributions, 297-298, 310 relational model of authority and, 119-120, 129-133, 158, 162 thought suppression and, 196-197, 213 values and, 1-2, 10, 17, 48, 51

INDEX Behavioral confirmation, motivational foundations of, 67, 105-106 disconfirmation, 83-84 personal identities, 85-89 threatened identities, 89-91 trading places, 84-85 functional analysis, 70-73 context, 95- 105 implications, 91-95 functions, 73-74 social interaction, 76-77 social knowledge, 74-75 mechanisms, 69-70 perceivers, 77-78 analysis, 78-79 correlational evidence, 79-80 experimental evidence, 80-83 social interaction, 67-69 Behavioral disconfirmation, 83-85 functional analysis, 72-73 context, 95-96, 101-102 implications, 92-93 identities personal, 85-89 threatened, 89-91 Belief pemverence, accountabilityand, 352-354 Beliefs behavioral confirmation and, 88-91, 99-100 relational model of authority and, 166 procedural justice, 155, 157-158 social psychology of authority, 124, 137 Beneficence, relational model of authority and, I60 Benevolence affect in social judgments and, 254, 256 relationalmodelofauthorityand,155-156.163 values and empirical studies, 23, 27, 29-30, 32, 36 theory assessment, 38-43, 46 theory development, 1 1- 12, 15 value priorities, 53, 55-57 Bias suppression, relational model of authority and, 139 Boundary conditions, affect in social judgments and, 252

C Carefulness, affect in social judgments and, 238

379

Categories, affect in social judgments and, 235, 244, 248, 262 Causal attribution, accountability and, 331, 355 Causal belonging, affect in social judgments and, 245 Causality, accountability and, 334-335, 360 Chinese culture, values and, 48 Choice, accountability and, see Accountability Circumscribed accuracy, behavioral confirmation and, 100 City life, Milgram and, 310-312, 318 Coercion, relational model of authority and, 1 I8 Cognition, see also Social cognition accountability and, 347, 349 affect in social judgments and, 228-229, 239, 268-269 mood congruence, 242, 244, 248-249 research progress, 253, 258, 266 Milgram and, 3 I2 relational model of authority and, 119, 134, 164-166 thought suppression and, 218 Cognitive capacity, affect in social judgments and, 240-241, 244 Cognitive demands. affect in social judgments and, 240 Cognitive dissonance accountability and, 345-346 affect in social judgments and, 235, 249 Milgram and, 292 Cognitive distancing. thought suppression and, 197 Cognitive load affect in social judgments and, 248 thought suppression and, 207-209, 217, 219 Cognitive maps, Milgram and, 289, 292, 31 I 312 Cognitive misers accountability and, 342, 344-346 affect in social judgments and, 234, 237 Cognitive processes accountability and, 331-333, 369-370 cognitive research metaphors, 333-335 coping strategies, 348, 352, 356 normative issues, 360, 367-368 political research, 335-336, 339-345, 347

380

INDEX

affect in social judgments and, 229, 267 mood congruence, 243-246, 249 multiprocess model, 232, 234-238, 241 research progress, 249, 251, 254, 256, 262 behavioral confirmation and, 70, 75, 94, 97, 106

relational model of authority and, 122, 157, 166 thought suppression and, 202-203, 207, 213. 215-216 Cognitive strain, affect in social judgments and, 248 Coherence, values and, 10 Cohort effect, values and, 55 Commitment, Milgram and, 298 Communal societies, values and, 56-57 Communication inhibition, thought suppression and. 196-198, 201 Compatibility behavioral confirmation and, 102 values and, 58, 60 empirical studies, 3 I , 35-36 theory assessment, 42-43, 47 theory development, 4-5, 14- I5 Compensation decisions, bias in, accountability and, 360-362 Competence relational model of authority and, 156, 158, 166

values and, 7, 9 Complex targets, affect in social judgments and, 238, 240 Complexity accountability and, 352-357 integrative, accountability and, 343-345, 347, 358-359 Compliance Milgram and, 298 relational model of authority and, 163, 165 empirical studies, 124, 130-131, 133, 137 procedural justice, 144, 152, 161-162 social psychology of authority, 117-119, I24 Comprehensiveness, values and, 2 Conation, affect in social judgments and, 228229 Concentration, thought suppression and, 197, 209

Conditioning, affect in social judgments and, 230-231 Confidence, accountability and, 356-357 Configurational verification, values and, 2 I Confirmation, behavioral, see Behavioral confirmation, motivational foundations of Conflict relational model of authority and, 115-1 16, 135- I37 thought suppression and, 219 values and, 60 empirical studies, 35-36 theory assessment, 41-44, 47 theory development, 4, 15 Conformity Milgram and, 318 characteristics, 281, 283-284, 286, 289 contributions, 297, 308-310 relational model of authority and, 152 restrictive, 4-5 values and empirical studies, 23, 27, 29, 31-32, 35 theory assessment, 38-44, 46 theory development, 9-10, 15 value priorities, 54, 57-58 Conjunction fallacy, accountability and, 360, 364-366 Conservation, values and, 43 Consistency Milgram and, 298 relational model of authority and, 139 Constraints, behavioral confirmation and, 9091, 94, 96, 98 Constructive processing, affect in social judgments and, 235-236, 246, 248, 250, 263 Context accountability and, 333, 337, 340-342 affect in social judgments and, 232, 247, 249, 265, 267-269 behavioral confirmation and, 95- 105 thought suppression and, 215-216 Context of measurement, values and, 46-47 Contingency model, social, see Accountability Continuous suppression, thought suppression and, 202, 211 Contractual societies, values and, 56-57 Control relational model of authority and procedural justice, 143, 150, 152, 154155

INDEX research, 144-149 thought suppression and, 220 Controlled distracter search, thought suppression and, 202-207, 211, 217 Conversational norms, accountability and, 362-367 Coping strategies, accountability and, 332, 347-348, 369 acceptability heuristic, 348-35 1 normative issues, 359-360, 367 political research, 336, 339-340, 344-347 preemptive self-criticism, 351-359 rationalization heuristic, 359 Correctability, relational model of authority and, 139 Correspondence bias, behavioral confirmation and, 75 Cost-benefit decisions, accountability and, 350-351, 359 Crowding, Milgram and, 311, 319 Cues accountability and, 363 affect in social judgments and, 235, 238, 250, 260 behavioral confirmation and, 76-77, 86, 88 thought suppression and, 214 Culture accountability and, 337 values and, 2-3, 48, 59-60 empirical studies, 17-18, 23, 26, 30-31, 35-36 theory assessment, 37, 40, 42, 46-48, 50-5 1

theory development, 9- 12

D Debriefing, Milgram and, 296 Decision control, relational model of authority and, 158 control issues, 145- 147 noncontrol issues, 150, 154 theories, 138-139 Decision-making accountability and, 331-332, 334 coping strategies, 348-351, 359 normative issues, 363-364. 367-368 political research, 337-338.342.344-346 affect in social judgments and, 249, 257, 265

38 1

relational model of authority and procedural justice, 138, 141, 147-149, 160-161 social psychology of authority, 121, 126127, 134-135 Defense mechanisms, behavioral confirmation and, 97, 101, 103 Defensive responses, behavioral confirmation and, 90-91, 96, 104 Delegitimizing model, Milgram and, 303 Democratic groups, relational model of authority and, 118 Depression affect in social judgments and, 245, 253 thought suppression and, 199-200, 205206, 216 Destructive obedience, Milgram and, 299, 304-308 Detachment, values and, 11, 26-27, 38 Dilution effect, accountability and, 360, 362363, 366 Direct-access strategy, affect in social judgments and, 234-235, 237 Direct functions, affect in social judgments and, 242 Discrimination, relational model of authority and, 156 Dishabituation, thought suppression and, 207, 209-21 1 Disposition accountability and, 354 behavioral confirmation and, 88-90, 95, 99 Dispute resolution, relational model of authority and, 138-139, 145 Distracters, thought suppression and, 193, 197, 217 postsuppression processes, 2 14-2 I7 processes, 202-208, 21 I Distraction, affect in social judgments and, 248 Distributive fairness, relational model of authority and, 123, 125, 127-128, 133, 135 Distributive justice relational model of authority and, 119, 121122, 131-132 measures, 167-186 Dominance, behavioral confirmation and, 90 Dual-process model, affect in social judgments and, 233-234 Dysphoria, affect in social judgments and, 254-258, 260, 265, 268

382

INDEX

E Eating disorders, thought suppression and, 213 Education, values and, 47 Effectiveness, relational model of authority and, 119 Effort minimization, affect in social judgments and, 233-234 Egocentric bias, relational model of authority and, 152-153 Emotion affect in social judgments and, 229-230, 265, 267 Milgram and, 287, 289 thought suppression and, 218-219 Emotion inhibition, thought suppression and, 196, 198-201, 213, 216 Encoding, affect in social judgments and, 233, 25 1, 265 Enjoyment, values and, 4-5, 8 Equity theories, relational model of authority and, 121-122 Equivalence of meaning, values and, 3 Ethical principles for research, Milgram and, 296, 299 Ethicality, relational model of authority and, 160-161, 163 noncontrol issues, 155-157 theories. 139, 142 Ethics, Milgram and, 283 Evolutionary signals, affect in social judgments and, 248-249 Expectancy, affect in social judgments and, 248 Expectancy confirmation, 68 Expectations behavioral confirmation and, 68, 77, 94 context, 99-100, 102 disconfirmation, 84-8 1 perceivers, 78, 82 relational model of authority and, 120, 122, 157, 162 Experience, values and, 1-2, 7 Extraverts, behavioral confirmation and, 8788, 91, 99

F Facilitative distortion, relational model of authority and, 156

Fairness, relational model of authority and, 163 empirical studies, 125-127, 130-135 implications, 160- 162 measures, 167-186 noncontrol issues, 152 procedural justice, 138-139, 144, 146, 148, 158 theories, 121-123 False consciousness, relational model of authority and, 162 Familiar stranger, Milgram and, 310 Familiarity, affect in social judgments and, 238 Fanaticism, Milgram and, 280 Feedback affect in social judgments and, 231 thought suppression and, 203 First-order inferences, relational model of authority and, 166 Focused self-distraction, thought suppression and, 205, 215 Freud, Sigmund, thought suppression and, 194, 199, 220 Friendship, behavioral confirmation and, 103I04 Functional analysis, behavioral confirmation and, 70-73 context, 95-105 implications, 9 1-95 Fundamental attribution error, accountability and, 352, 354-356, 360-361

G Gender, values and, 18-20, 48 Generality, values and, 2, 47-49 Getting along, behavioral confirmation and, 74, 76-77, 88 Global accuracy, behavioral confirmation and, 100

Goals affect in social judgments and, 235, 238 behavioral confirmation and, 97, 101 relational model of authority and, 134, 140, 142, 147-148 values and empirical studies, 17, 27 theory assessment, 40 theory development, 4, 7-13, 15 value priorities, 52, 59

INDEX Goffman, Milgram and, 286-287 Group dynamics, accountability and, 332 Group pressure, Milgram and, 285, 308 Group process, relational model of authority and, 162, 165 Group-value models, relational model of authority and, 140-141, 149, 158 Groups, relational model of authority in, see Relational model of authority in groups

H Habituation, thought suppression and, 2 10211, 218 Harmony, values and, 26-27, 38, 43, 48 Hedonism relational model of authority and, 144 values and empirical studies, 23, 27-28, 32 theory assessment, 38, 41-44, 46, 48 theory development, 8, 14-15 value priorities, 54, 57 Heuristic processing strategy, affect in social judgments and, 268 mood congruence, 248-249 multiprocess model, 234-237, 240-242 research progress, 256, 259-260 Hollow justice, relational model of authority and, 162 Hyperaccessibility, thought suppression and, 207-209

I Identification, relational model of authority and, 118 Identity, relational model of authority and, 142-143 Illusion, Milgram and, 287 Imagability, Milgram and, 3 12 Importance, affect in social judgments and, 238 Impression formation accountability and, 356-357 affect in social judgments and, 251-252, 254, 257, 262, 264, 268 relational model of authority and, 165 Impression management accountability and, 338, 345

383

behavioral confirmation and, 76, 79, 86 relational model of authority and, 166 Independence, values and, 4-5, 7 Indirect functions, affect in social judgments and, 242 Individuating information, accountability and, 360, 366-367 Information, individuating, accountability and, 360, 366-367 Information processing accountability and, 331, 335-337, 354, 358 affect in social judgments and, 256, 267269 mood congruence, 245-249 multiprocess model, 232, 234-236, 241242 values and, 17 Information search, affect in social judgments and, 237, 257 Informational consequences, affect in social judgments and, 228 Informational effects, affect in social judgments and, 267 mood congruence, 243-247, 249 multiprocess model, 233-234 research progress, 250-256 Informational theories, affect in social judgments and, 231 Inhibition, thought suppression and instigation, 196-201 postsuppression processes, 21 3 processes, 202, 204 Institutional norms, accountability and, 33 I Instrumental goals, values and, 15 Instrumental models, relational model of authority and, 138-142, 144-145 Instrumental values, 16 empirical studies, 17, 23, 36-37 theory assessment, 49-50 Integrative complexity, accountability and, 343-345, 347, 358-359 Intense mood manipulation, affect in social judgments and, 245 Interaction behavioral confirmation and, 67-69, 73, 106 context, 96-98, 100-102, 104-105 disconfirmation, 85, 87-88 functions, 73-77

384

INDEX

implications, 92-93, 95 perceivers, 77-82 relational model of authority and, 120 thought suppression and, 199-200 Internalization, relational model of authority and, I18 Interpersonal relationships, affect in social judgments and, 262 Intervention, behavioral confirmation and, 85, 87 Introverts, behavioral confirmation and, 8788, 90, 99

J Judge, affect in social judgments and, 234236, 238, 241-242 Judgment accountability and, see Accountability social, affect in, see Affect in social judgments Justice models, relational model of authority and, 121-123 Justice, procedural, see Procedural justice

K Knowledge, behavioral confirmation and functional analysis, 94-98, 101-102, 104 perceivers, 79-80, 82

L Labels, accountability and, 350 Language, values and, 17-21 Lay epistemic theory, affect in social judgments and, 237 Learning, affect in social judgments and, 244, 247, 250-251. 268 Legal authority, relational model of authority and, 125-126, 130 procedural justice, 150-151, 155, 158 Legal judgments, relational model of authority and, 129-130 Legitimacy Milgram and, 280 relational model of authority and, 117, 163, 165-166 empirical studies, 124, 126, 129-134, 137

measures, 167-186 procedural justice, 137, 151, 158, 162 social psychology of authority, 118- 124 Legitimate authority, Milgram and, 278, 282 Legitimate power, relational model of authority and, 118, 135-136, 162 Liking, Milgram and, 298 Lost-letter technique, Milgram and, 281, 313315, 319

M Macro context, accountability and, 337 Managerial authority, relational model of authority and, 126-127 procedural justice, 150-152, 156, 158 Matching mood, affect in social judgments and, 245 Maturity, values and, 4-5, 11, 13, 39 Memory accountability and, 354, 367 affect in social judgments and, 232, 267268 mood congruence, 244-246, 248-249 multiprocess model, 235, 238, 240-241 research progress, 250, 253, 259-260, 264-265 thought suppression and, 193- 195, 204205, 208-209 Mental control, thought suppression and, 193 Micro context, accountability and, 337 Milgram, Stanley, 277-278, 317-320 characteristics Asch, influences of, 285-286 experimental strategy, 28 1-282 moral issues, 282-284 phenomenology, 278-280 situational determinants, 284-285 style, 288-290 theater, influences of, 286-288 contributions, 293 city life, 310-312 conformity, 308-3 I0 lost-letter technique, 313-315 obedience to authority, 293-308 small-world method, 315-317 television, 312-313 shortcomings, 290-293 Misattribution hypothesis, affect in social judgments and, 246

385

INDEX Momentary suppression, 202 Mood affect in social judgments and, 229-232, 267-269 extensions, 264-267 multiprocess model, 235, 240-241 research progress, 251-263 thought suppression and, 218 instigation, 198-199, 201 postsuppression processes, 215-216 processes, 205 Mood congruence, affect in social judgments and, 232, 242-243 informational effects, 243-247 multiprocess model, 233, 242 processing consequences, 247-249 research progress, 250-251, 254, 259, 262, 265. 267 Mood-consistentjudgments, affect in, 25 1 Mood-priming effects, affect in social judgments and, 268 multiprocess model, 236, 238 research progress, 250-25 I , 258-259, 262 Moral issues, Milgram and, 282-284, 303, 319 Motivated processing strategy, affect in social judgments and, 234-235, 257, 268 Motivation accountability and, 361, 370 coping strategies, 35 1, 358 political research, 338, 343-344, 346347 affect in social judgments and, 249, 268269 multiprocess model, 233, 236-242 research progress, 254, 256, 258, 262 values and, 60 empirical studies, 17, 22-23, 27-28, 31, 36 theory assessment, 37-40, 42-46 theory development, 9- 13 value priorities, 52-54, 59 Motivational foundations of behavioral confirmation, see Behavioral confirmation, motivational foundations of Motivational types, values and, 4-7, 50, 54, 57,59 Multidimensional scaling, values and, 2 1-22 Multiprocess model, affect in social judgments and, see Affect in social judgments

N Nationalism, values and, 58-59 Negative expectations, behavioral confirmation and, 100 Negative thoughts, thought suppression and, 206 Negotiation, accountability and, 341 Neutrality, relational model of authority and, 165 noncontrol issues, 149-150, 152, 154, 157158 procedural justice, 158-159 theories, 141-143 Norms accountability and, 331, 333-335, 359-360, 369 ambiguity aversion, 363-364 attraction effect, 364 attribution error, 360-361 compensation decisions, 36 1-362 conjunction fallacy, 364-366 coping strategies, 347-359 dilution effect, 362-363 individuating information, 366-367 political research, 336-337, 343-344 status quo effect, 367-368 affect in social judgments and, 263 Milgram and, 282, 319 relational model of authority and, 1 16- 117, 162 values and, 9-10, 40, 50-51

0 Obedience Milgram and, 317, 320 characteristics, 278-284, 287-289 contributions, 293-308 shortcomings, 290-292 relational model of authority and, 117 procedural justice, 158, 161 social psychology of authority, 117-1 18, 130, 133, 137 values and, 4 Obesity, thought suppression and, 213 Obsession, thought suppression and, 198, 20 1 Opinions, relational model of authority and, 118

386

INDEX

Organizational behavior accountability and, 332, 336-337, 342, 349, 370 relational model of authority and, 131-132, 134, 137, 155 Organizational commitment, relational model of authority and, 126-127, 150 Outcome favorability, relational model of authority and, 123 empirical studies, 125, 127-132 measures, 167- 186 procedural justice, 143, 146, 149-152, 157 Outcome frequency, Milgram and, 298 Outcomes, relational model of authority and, 134-135, 138, 153 Overattribution effect, accountability and, 355-356, 361 Overconfidence, accountability and, 352 Overload, Milgram and, 3 1 1, 3 I8 Overuse of base rates, accountability and, 360

P Partner choices, affect in social judgments and, 256-258, 268 Peer counselors, behavioral confirmation and, 82 Peer pressure, Milgram and, 285, 308-310 Perceivers affect in social judgments and, 244 behavioral confirmation and, 67-69, 73 context, 98-101, 103 disconfirmation, 84-91 functions, 73, 75-83 implications, 92-95 relational model of authority and, 120 Perception, accountability and, 355 Personal identity, behavioral confirmation and, 85-89 Personal relevance, affect in social judgments and, 238-239, 256-258, 269 Personality accountability and, 332, 335, 369 coping strategies, 354-356 normative issues, 361, 365 political research, 338 affect in social judgments and, 244, 251, 258 behavioral confirmation and, 68-69, 71, 75-76

context, 95, 97, 100, 104 disconfirmation, 85-86, 88-89 perceivers, 79-81 Milgram and, 280, 285, 309 Phenomenology, Milgram and, 278-280 Philosophy, values and, 10 Phobic response, thought suppression and, 2 13 Photography, Milgram and, 285 Political authority, relational model of authority and, 127-128, 132 Political behavior, relational model of authority and, 132-133 Political research, accountability and, 335-339 acceptability heuristic, 340-342 coping strategies, 339-340 hybrid responses, 346-347 preemptive self-criticism, 343-344 rationalization heuristic, 344-346 Politics, values and, 58-59 Population, Milgram and, 278 Postsuppression processes, thought suppression and, 193, 211 distracters, 214-217 rebound effects, 21 1-214 Postsuppression rebound, thought suppression and, 21 1 Power behavioral confirmation and, 78-79, 82, 90, 104 relational model of authority and, 115-1 18, 143, 156, 164 values and empirical studies, 23, 27-28 theory assessment, 37, 40-41, 43-44, 48 theory development, 4, 8-9, 14-15 value priorities, 55, 57 Predictability, behavioral confirmation and, 99 Preemptive self-criticism, accountability and, 343-344, 351-359 Primacy accountability and, 353-354 affect in social judgments and, 245 Primary groups, values and, 57 Priming affect in social judgments and, 267, 269 mood congruence, 242-247 multipmcess model, 233, 236, 238 research progress, 250-256, 261 thought suppression and, 204 Probability, accountability and, 364-365

INDEX Procedural desirability, relational model of authority and, 123, 132 Procedural fairness, relational model of authority and, 123, 158, 163 control issues, 144-146, 149 empirical studies, 125- 128, 131 - I35 implications, 160- I62 noncontrol issues, 150-151, 157 Procedural justice, relational model of authority and, 137, 164 concerns, 158-159 control issues, 144- 149 empirical studies, 125, 127, 129-135 implications, 160- 162 measures, 167-186 noncontrol issues, 149- 158 theories. 122-123, 138-144 Procedures, relational model of authority and, 134-135 Process control. relational model of authority and control issues, 144-149 procedural justice, 154, 157, 159 theories, 138-139 Processing capacity, affect in social judgments and, 238, 248 Processing effects, affect in social judgments and, 233, 242, 261 Processing strategies, affect in social judgments and, 256-260, 262-263, 268-269 Processing style, affect in social judgments and, 245 Productive processes, affect in social judgments and, 248 Prosocial type, values and, 4-5, 11-12 Prototypes, affect in social judgments and, 259-260 Psychoanalysis, affect in social judgments and, 230-23 I Psychology, behavioral confirmation and, 7071, 79, 92, 106 context, 96-98, 100-101, 105 Public choice theory, relational model of authority and, 119

R Race affect in social judgments and, 263 behavioral confirmation and. 68

387

Rapport, behavioral confirmation and, 82 Rational choice theory, accountability and, 334, 367 Rationalization heuristic, accountability and, 344-347, 359 Reaction time affect in social judgments and, 266 thought suppression and, 208 Rebound effects, thought suppression and, 21 1-217, 220 Recall, affect in social judgments and, 250, 259-260, 264 Recency, affect in social judgments and, 245 Regression analysis, relational model of authority and, 150-151 Regret avoidance, accountability and, 367368 Reinforcement, affect in social judgments and, 23 1 Relational model of authority in groups, 162165, 167-186 implications, 165- 166 procedural justice, 137 control issues, 144- 149 implications, 160- 162 noncontrol issues, 149- 158 relational concerns, 158- 159 theories, 138-144 social psychology of authority, 115- I17 effective functioning, 117-1 19 empirical studies, 124- 137 legitimacy, 119 theories, 119-124 Relationships, affect in social judgments and, 261-263 Relative deprivation. relational model of authority and, 121-122 Religion, values and, 10, 59 empirical studies, 17-20 theory assessment, 38, 40, 47-48 Reminders, thought suppression and, 214-217 Repeated exposure, relational model of authority and, 163 Representativeness, relational model of authority and, 139 Reproductive processes, affect in social judgments and, 248 Restrictive conformity, values and, 4-5 Retrieval, affect in social judgments and, 265, 268

388

INDEX

mood congruence, 244, 246-247 multiprocess model, 233, 235, 238 Reward power, relational model of authority and, 118 Rokeach Value Survey, values and, 4, 8, 17, 36 Roles, behavioral confirmation and, 98 Rule creation, relational model of authority and, 115-116

S Scarcity, Milgram and, 298 Second-order inferences, relational model of authority and, 166 Security relational model of authority and, 157 values and, 57 empirical studies, 23, 27, 29, 32 theory assessment, 38, 41-44, 47 theory development, 4-5, 9, 12, 15 Selective attention, affect in social judgments and, 250-251 Self-actualization, behavioral confirmation and, 97 Self-conceptions, behavioral confirmation and, 90-91 Self-consciousness, behavioral confirmation and, 89 Self-control, thought suppression and, 196197, 216-217 Self-criticism, accountability and, 343-344, 346, 351-360 Self-decision, Milgram and, 300-301 Self-direction, values and empirical studies, 23, 27-29, 32 theory assessment, 38, 41-42, 46, 48 theory development, 4-8, 14-15 value priorities, 54, 57-58 Self-enhancement, values and, 5 , 43-44 Self-esteem accountability and, 338 behavioral confirmation and, 103-104 Self-fulfilling prophecy, behavioral confirmation and, 67-70, 74, 95, 97, 106 Self-image, accountability and, 338-339, 344 Self-interest accountability and, 370 relational model of authority and, 120

Self-monitoring, behavioral confirmation and, 80, 88 Self-perception affect in social judgments and, 230, 235 behavioral confirmation and, 90 Self-respect, relational model of authority and, 161 Self-restraint, values and, 5 Self-transcendence, values and, 43-44 Shock generation, Milgram and, 287 Simulation heuristic, accountability and, 368 Sinusoid curve of value associations, 54-57, 59 Situational variables accountability and, 332, 347, 354, 360361, 369 affect in social judgments and, 236, 238, 241-242, 256 Milgram and, 284-285, 297 Skin conductance level, thought suppression and, 209-212 Small-world method, Milgram and, 313-315, 319 Smallest Space Analysis (SSA), values and, 52 empirical studies, 21-23, 28, 30, 32, 36-37 theory assessment, 37, 40,42, 45-46, 4849 Social adjustment, behavioral confirmation and, 95-96, 104 Social cognition accountability and, 33 1-332, 344, 369 affect in social judgments and, 232, 236, 248, 266 relational model of authority and, 162, 164166 Social comparison, affect in social judgments and, 255 Social contingency model, see Accountability Social dilemmas, relational model of authority and, 136 Social exchange theories, relational model of authority and, 119-121 Social identity, accountability and, 338 Social image, accountability and, 338-339, 344 Social judgments, affect in, see Affect in social judgments Social knowledge, behavioral confirmation and context, 95, 98-99

INDEX functions, 74-75, 77 theory assessment, 37-38, 41-43, 46, implications, 92-93 48 perceivers, 77 theory development, 7-8, 14-15 Social life, thought suppression and, 2 19-220 value priorities, 52, 54, 57, 59 Social network, Milgram and, 316 Structural equation analysis, relational model Social order, values and, 4 of authority and, 129 Social proof, Milgram and, 298 Substantive processing strategy, affect in social Socialization judgments and, 268-269 affect in social judgments and, 265 mood congruence, 242, 244-246, 248 values and, 51 multiprocess model, 234, 236-237, 240Societal harmony, values and, 48 242 Society research progress, 253, 256, 258-263 Milgram and, 280 Subway, Milgram and, 278, 281, 319 relational model of authority and, 133, 135- Suppression 137, 163-164 defined, 194 procedural justice, 155, 161 thought, see Thought suppression values and, 56-57 Symbols Specific motivation, affect in social judgments relational model of authority and, 140, 143 and, 238-240 values and, 10 Spirituality, values and empirical studies, 26-27, 30, 36 T theory assessment, 38-39 theory development, 10- I 1, 13, 15 Target specificity, affect in social judgments Spreading activation, affect in social judgments and, 233, 256 and, 251 Targets Stability, behavioral confirmation and, 99 affect in social judgments and, 267 Standing, relational model of authority and, 165 mood congruence, 246-247 noncontrol issues, 149-150, 152-155 multiprocess model, 234-236, 238 procedural justice, 158-159, 161-162 research progress, 256, 259-260, 262theories, 141-144 263 Status behavioral confirmation and, 67-69, 73 Milgram and, 310 context, 98-101 relational model of authority and, 140-141, disconfirmation, 84-86, 88-91 143, 153 functions, 73, 75-83 values and, 9, 57 implications, 92-94 Status quo effect, accountability and, 360, Teleonomic theory, behavioral confirmation 367-368 and, 71 Status recognition, relational model of authori- Television, antisocial behavior and, 312-313, ty and, 141 3 17-318 Status seekers, accountability and, 338-339, Terminal goals, values and, 15 369 Terminal values, 16 Stereotypes, behavioral confirmation and, 68, empirical studies, 17, 23, 36-37 94-95, 102-103 theory assessment, 49-50 Stereotyping Theater, Milgram and, 286-288 affect in social judgments and, 234, 237, Thought suppression, 193- 194, 217 262-263 implications, 219-220 Milgram and, 298 instigation, 196 Stimulation, values and circumstances, 196-200 empirical studies, 23, 27-28 readiness, 200-201

390

INDEX

postsuppression processes, 21 1 distracters, 214-217 rebound effects, 21 1-214 processes, 202 automatic target search, 206-21 1 controlled distracter search, 204-206 cycle, 202-204 prospects, 217-2 18 repression, 194-196 Time pressure, thought suppression and, 207 TOTE, thought suppression and, 203 Tradition, values and empirical studies, 23, 27, 29-30, 32, 36 theory assessment, 37, 39-44 theory development, 10, 15 value priorities, 57-59 Transaction costs, relational model of authority and, 120-121 Trust, relational model of authority and, 162, 165 noncontrol issues, 149-150, 152, 154-157 procedural justice, 158-159 theories, 141, 143 "pica1 targets, affect in social judgments and, 238, 240, 256, 258-263 U

Unfairness, relational model of authority and, 148 Unfocused self-distraction, thought suppression and, 205, 214-215 Universalism, values and empirical studies, 23, 27, 29-30, 36 theory assessment, 39, 41-43, 46-48 theory development, 11-15 value priorities, 53, 57-58 Universals, values and, see Values Urban psychology, Milgram and, 310-312, 318

V Valence integration, affect in social judgments and, 245 Value associations, sinusoid curve of, 54-57, 59 Value content, 2, 5-13, 22, 50 Value hierarchies, 3

Value priorities, 1-3, 51-52, 60 indexes of value types, 52-53 theory assessment, 47, 50-51 value structure, 54-59 Value structure, 3 empirical studies, 22 hypotheses, 54-59 theory assessment, 47-50 theory development, 5, 13-15 Values, 1-3, 59-60 accountability and, 337, 342 behavioral confirmation and, 88-89, 95, 103-105 empirical studies methods, 17-23 results, 23-37 theory-based survey, 16-17 relational model of authority and, 118, 134, 137, 144 survey, 60-62 theory assessment dynamic structure of value types, 41-47 nature of value types, 37-41 normative ideals, 50-5 1 terminal versus instrumental values, 4950 universal value structure, 47-49 theory development early version, 3-5 terminal versus instrumental values, 1516 value content, 5- 13 value structure, 13-15 value priorities, 51-52 indexes of value types, 52-53 value structure hypotheses, 54-59 Violence television and, 312-313, 317-318 thought suppression and, 213, 220 ViItuous behavior, values and, 48 Voice, relational model of authority and, 139, 144, 147 Voluntary compliance, relational model of authority and, 118-119, 137, 163, 165 Volunteers, behavioral confirmation and, 104-

I05

W Word association, thought suppression and, 207

CONTENTS OF OTHER VOLUMES

Volume 1 Cultural Influences upon Cognitive Processes Harry C. Triandis The Interaction of Cognitive and Physiological Determinants of Emotional State Stanley Schachter Experimental Studies of Coalition Formation William A. Camson Communication Networks Marvin E. Shaw A Contingency Model of Leadership Effectiveness Fred E. Fiedler Inducing Resistance to Persuasion: Some Contemporary Approaches William J . McGuire Social Motivation, Dependency, and Susceptibility to Social Influence Richard H . Walters and Ross D. Purke Sociability and Social Organization in Monkeys I and Apes William A. Mason Author Index-Subject Index

From Acts to Dispositions: The Attribution Process in Person Perception Edward E. Jones and Keith E . Davis Inequality in Social Exchange J . Stacy Adams The Concept of Aggressive Drive: Some Additional Considerations Leonard Berkowitz Author Index-Subject Index

Volume 3

Mathematical Models in Social Psychology Robert P. Abelson The Experimental Analysis of Social Performance Michael Argyle and Adam Kendon A Structural Balance Approach to the Analysis of Communication Effects N. T. Feather Effects of Fear Arousal on Attitude Change: Recent Developments in Theory and Experimental Research Irving L. Janis Communication Processes and the Properties of Language Volume 2 Serge Moscovici Vicarious Processes: A Case of No-Trial Learn- The Congruity Principle Revisited: Studies in the Reduction, Induction and Generalization of ing Persuasion Albert Bandura Percy H.Tannenbaum Selective Exposure Author Index-Subject Index Jonathan L. Freedman and David 0. Sears Group Problem Solving Volume 4 L. Richard Hoffman Situational Factors in Conformity The Theory of Cognitive Dissonance: A Current Vernon L.Allen Perspective Social Power Elliot Aronson John Schopler

39 1

392

CONTENTS OF OTHER VOLUMES

Attitudes and Attraction Donn Byrne Sociolinguistics Susan M. Ervin-Tripp Recognition of Emotion Nico H. Frijda Studies of Status Congruence Edward E. Sampson Exploratory Investigations of Empathy Ezra Stotland The Personal Reference Scale: An Approach to Social Judgment Harry S. Vpshaw Author Index-Subject Index

Volume 5 Media Violence and Aggressive Behavior: A Review of Experimental Research Richard E. Goranson Studies in Leader Legitimacy, Influence, and Innovation Edwin P. Hollander and James W. Julian Experimental Studies of Negro-White Relationships Irwin Katz Findings and Theory in the Study of Fear Communications Howard Leventhal Perceived Freedom Ivan D. Steiner Experimental Studies of Families Nancy E. Wmler and Elliot G. Mishler Why Do Groups Make Riskier Decisions than Individuals? Kenneth L. Dion, Robert S. Baron. and Norman Miller Author Index-Subject Index

Volume 6 Self-Perception Theory Daryl J. Bem Social Norms, Feelings, and Other Factors Affecting Helping and Altruism Leonard Berkowitz The Power of Liking: Consequences of Interpersonal Attitudes Derived from a Liberalized View of Secondary Reinforcement Albert J. Lott and Bernice E. Lott

Social Influence, Conformity Bias, and the Study of Active Minorities Serge Moscovici and Claude Faucheux A Critical Analysis of Research Utilizing the Prisoner’s Dilemma Paradigm for the Study of Bargaining Charlan Nemeth Suuctural Representationsof Implicit Personality Theory Seymour Rosenberg and Andrea Sedlak Author Index-Subject Index

Volume 7 Cognitive Algebra: Integration Theory Applied to Social Attribution Norman A. Anderson On Conflicts and Bargaining Erika ApfeIbaum Physical Attractiveness Ellen Bersheid and Elaine Walster Compliance,Justification,and CognitiveChange Harold B. Gerard, Edward S. Connolley, and Roland A. Wilhelmy Processes in Delay of Gratification Walter Mischel Helping a Distressed Person: Social, Personality, and Stimulus Determinants Ervin Staub Author Index-Subject Index

Volume 8 Social Support for Nonconformity Vernon L. Allen Group Tasks, Group Interaction Process, and Group Performance Effectiveness: A Review and Proposed Integration J. Richard Hackman and Charles G. Morris The Human Subject in the Rychology Experiment: Fact and Artifact A r k W.Kruglanski Emotional Arousal in the Facilitation of Aggression through Communication Percy H. Tannenbaum and DolfZillman The Reluctance to Transmit Bad News Abraham Tesser and Sidney Rosen Objective Self-Awareness Robert A. Wicklund

CONTENTS OF OTHER VOLUMES Responses to Uncontrollable Outcomes: An Integration of Reactance Theory and the Learned Helplessness Model Camille B. Wortman and Jack W. Brehm Subject Index

Volume 9 New Directions in Equity Research Elaine Walster. Ellen Berscheid, and G. William Walster Equity Theory Revisited: Comments and Annotated Bibliography J. Stacy Adams and Sara Freedman The Distribution of Rewards and Resources in Groups and Organizations Gerald S. Leventhal Deserving and the Emergenceof Forms of Justice Melvin J. Lerner, Dale I: MiNer, and John G . Holmes Equity and the Law: The Effect of a Harmdoer’s “Suffering in the Act” on Liking and Assigned Punishment William Austin, Elaine Walster, and Mary Kristine Utne Incremental Exchange Theory: A Formal Model for Progression in Dyadic Social Interaction L. Lowell Huesmann and George Levinger Commentary George C. Homans Subject Index

Volume 10 The Catharsis of Aggression: An Evaluation of a Hypothesis Russell G. Geen and Michael B . Quanty Mere Exposure Albert A. Harrison Moral Internalization: Current Theory and Research Martin L. H o m n Some Effects of Violent and Nonviolent Movies on the Behavior of Juvenile Delinquents Ross D. Parke, Leonard Berkowirz, Jacques P. Leyens, Stephen G. West, and Richard Sebasrian The IntuitivePsychologistand His Shortcomings: Distortions in the Attribution Process Lee Ross

393

Normative Influences on Altruism Shalom H. Schwartz A Discussion of the Domain and Methods of Social Psychology:’WoPapers by Ron Harre and Bany R. Schlenker Leonard Berkowirz The Ethogenic Approach: Theory and Practice R. Harre On the Ethogenic Approach: Etiquette and Revolution Barry R. Schlenker Automatisms and Autonomies: In Reply to Professor Schlenker R. Harm Subject Index

Volume 11 The Persistence of Experimentally Induced Attitude Change Thomas D. Cook and Brian F. Flay The Contingency Model and the Dynamics of the Leadership Process Fred E . Fiedler An Attributional Theory of Choice Andy Kukla Group-Induced Polarization of Attitudes and Behavior Helmut Lamm and David G. Myers Crowding: Determinants and Effects Janet E. Stockdale Salience: Attention, and Attribution: Top of the Head Phenomena Shelley E. Taylor and Susan T. Fiske Self-Generated Attitude Change Abraham Tesser Subject Index

Volume 12 Part I. Studies in Social Cognition Prototypes in Person Perception Nancy Cantor and Walter Mischel A Cognitive-AttributionalAnalysis of Stereotyping David L. Hamilton Self-Monitoring Processes Mark Snyder

394

CONTENTS OF OTHER VOLUMES

Part 11. Social Influences and Social Interaction Architectural Mediation of Residential Density and Control: Crowding and the Regulation of Social Contact Andrew Baum and Stuart Valins A Cultural Ecology of Social Behavior J. W. Berry Experimentson Deviance with Special Reference to Dishonesty David P. Farrington From the Early Window to the Late Night Show: International Trends in the Study of Television's Impact on Children and Adults John P. Murray and Susan Kippar Effects of Prosocial Television and Film Material on the Behavior of Viewers J. Phillipe Rushton Subject Index

Dialectic Conceptions in Social Rychology: An Application to Social Penetration and Privacy Regulation Irwin Altman, Anne Vinsel, and Barbara B. Brown Direct Experience and Attitude-Behavior Consistency Russell H.Fazio and Mark P. Zanna Predictability and Human Stress: Toward a Clarification of Evidence and Theory Suzanne M. Miller Perceptual and Judgmental Processes in Social Contexts Arnold Upmeyer Jury Trials: Psychology and Law Charlan Jeanne Nemeth Index

Volume 15

Volume 13 People's Analyses of the Causes of AbilityLinked Performances John M. Darley and George R. Goethals The Empirical Exploration of Intrinsic Motivational Processes Edward I. Deci and Richard M. Ryan Attribution of Responsibility: From Man the Scientist to Man as Lawyer Frank D. Fincham and Joseph M . Jaspars Toward a Comprehensive Theory of Emotion Howard Leventhal Toward a Theory of Conversion Behavior Serge Moscovici The Role of Information Retrieval and Conditional Inference Processes in Belief Formation and Change Robert S. Wyer, Jr. and Jon Hartwick Index

Balance, Agreement, and Positivity in the Cognition of Small Social Structures Walter H. Crockett Episode Cognition: Internal Representations of Interaction Routines Joseph P. Forgas The Effects of Aggressive-Pornographic Mass Media Stimuli Neil M. Malamuth and Ed Donnerstein Socialization in Small Groups: Temporal Changes in Individual-Group Relations Richard L. Moreland and John M . Levine Translating Actions into Attitudes: An IdentityAnalytic Approach to the Explanation of Social Conduct Barry R. Schlenker Aversive Conditions as Stimuli to Aggression Leonard Berkowitz Index

Volume 14

Volume 16

Verbal and Nonverbal Communicationof Deception Miron Zuckerman. Bella M. DePaulo. and Robert Rosenthal Cognitive, Social, and Personality Processes in the Physiological Detection of Deception William M. Waid and Martin T. Orne

A Contextualist Theory of Knowledge: Its Implications for Innovation and Reform in Rychological Research William J. McGuire Social Cognition: Some Historical and Theoretical Perspectives Janet Landman and Melvin Manis

CONTENTS OF OTHER VOLUMES Pwddigmatic Behaviorism: Unified Theory for Social-Personality Psychology Arthur W. Staats Social Psychology from the Standpoint of a Structural Symbolic Interactionism: Toward an Interdisciplinary Social Psychology Sheldon Stryker Toward an Interdisciplinary Social Psychology Carl W. Backman Index

Volume 17 Mental Representations of Self John F. Kihlsrrom and Nancy Cantor Theory of the Self Impasse and Evolution Kenneth J . Gergen A Perceptual-Motor Theory of Emotion Howard Leventhal Equity and Social Change in Human Relationships Charles G. McClintock, Roderick M. Kramer. and Linda J. Keil A New Look at Dissonance Theory Joel Cooper and Russell H. Fazio Cognitive Theories of Persuasion Alice H. Eagly and Shelly Chaiken Helping Behavior and Altruism: An Empirical and Conceptual Overview John F. Dovidio Index

Volume 18 A Typological Approach to Marital Interaction: Recent Theory and Research Mary Anne Fitzpatrick Groups in Exotic Environments Albert A. Harrison and Mary M. Connors Balance Theory, the Jordan Paradigm, and the Wiest Tetrahedon Chester A . Insko The Social Relations Model David A. Kenny and Lawrence La Voie Coalition Bargaining S. S. Komorita When Belief Creates Reality Mark Snyder Index

395

Volume 19 Distraction-Conflict Theory: Progress and Problems Robert S. Baron Recent Research on Selective Exposure to Information Dieter Frey The Role of Threat to Self-Esteem and Perceived Control in Recipient Reaction to Help: Theory Development and Empirical Validation Arie Nadler and Jeffrey D. Fisher The Elaboration Likelihood Model of Persuasion Richard E. Petty and John T. Cacioppo Natural Experiments on the Effects of Mass Media Violence on Fatal Aggression: Strengths and Weaknesses of a New Approach David P. Phillips Paradigms and Groups Ivan D. Steiner Social Categorization: Implications for Creation and Reduction of Intergroup Bias David A. Wilder Index

Volume 20 Attitudes, Traits, and Actions: Dispositional Prediction of Behavior in Personality and Social Psychology Icek Ajzen Prosocial Motivation: Is It Ever Truly Altruistic? C. Daniel Batson Dimensions of Group Process: Amount and Structure of Vocal Interaction James M. Dabbs, Jr. and R. Barry Ruback The Dynamics of Opinion Formation Harold B. Gerard and Ruben Orive Positive Affect, Cognitive Processes, and Social Behavior Alice M. Isen Between Hope and Fear: The Psychology of Risk Lola L. Lopes Toward an Integration of Cognitive and Motivational Perspectives on Social Inference: A Biased Hypothesis-Testing Model Tom Pyszczynski and Jeff Greenberg Index

396

CONTENTS OF (ITHER VOLUMES

Volume 21 Introduction Leonard Berkowitz Part 1. The Self as Known Narrative and the Self as Relationship Kenneth J. Gergen and Mary M. Gergen Self and Others: Studies in Social Personality and Autobiography Seymour Rosenberg Content and Process in the Experience of Self William J. McGuire and Claire K McGuire Information Processing and the Study of the Self John F. Kihlstrom. Nancy Cantor, Jeanne Sumi Albright. Beverly R. Chew, Sianley B. Klein, and Paula M. Niedenihal Part 11. Self-Motives Toward a Self-Evaluation Maintenance Model of Social Behavior Abraham Tesser The Self: A Dialectical Approach Carl W Backman The Psychology of Self-Affirmation: Sustaining the Integrity of the Self Claude M. Sieele A Model of Behavioral Self-Regulation:Translating Intention into Action Michael F. Scheier and Charles S. Carver Index

Volume 22 On the Construction of the Anger Experience: Aversive Events and Negative Priming in the Formation of Feelings Leonard Berkowitz and Karen Heimer Social Rychophysiology: A New Look John T.Cacioppo, Richard E. Petty, and Louis G. Tassinary Self-DiscrepancyTheory: What Patterns of SelfBeliefs Cause People to Suffer? E. Tory Higgins Minding Matters: The Consequences of Mindlessness-Mindfulness Ellen J. Langer The Tradeoffsof Social Control and Innovation in Groups and Organizations Charlan Jeanne Nemeth and Barry M . Staw

Confession, Inhibition, and Disease James W. Pennebaker A Sociocognitive Model of Attitude Structure and Function Anthony R. Pratkanis and Anthony G. Greenwald Introspection, Attitude Change, and AttitudeBehavior Consistency: The Disruptive Effects of Explaining Why We Feel the Way We Do Timothy D. Wilson, Dana S. Dunn. Dolores Krafr, and Douglas J. Lisle Index

Volume 23 A Continuum of Impression Formation, from Category-Based to Individuating Processes: Influences of Information and Motivation on Attention and Interpretation Susan ‘I: Fiske and Steven L. Neuberg Multiple Processes by Which Attitudes Guide Behavior: The MODE Model as as Integrative Framework Russell H. Fazio PEAT An Integrative Model of Attribution Processes John W.Medcof Reading People’s Minds: A TransformationRule Model for Predicting Others’ Thoughts and Feelings Rachel Karnwl Self-Attention and Behavior: A Review and Theoretical Update Frederick X. Gibbons Counterfactual Thinking and Social Perception: Thinking about What Might Have Been Dale T. Miller, William Turnbull. and Cathy McFarland Index

Volume 24 The Role of Self-Interest in Social and Political Attitudes David 0.Sears and Carolyn L. Funk A Terror Management Theory of Social Behavior: The Psychological Functions of SelfEsteem and Cultural Worldviews Sheldon Solomon, Jeff Greenberg, and Tom Pyszczynski

CONTENTS OF OTHER VOLUMES

Mood and Persuasion: Affective States Influence the Processing of Persuasive Communications Norbert Schwarz, Herberi Bless, and Gerd Bohner A Focus Theory of Normative Conduct: A Theoretical Refinement and Reevaluation of the Role of Norms in Human Behavior Robert B. Cialdini, Carl A. Kallgren. and Raymond R. Reno

397

The Effects of Interaction Goals on Person Perception James L. Hilton and John M . Darley Studying Social Interaction with the Rochester Interaction Record Harry T.Reis and Ludd Wheeler Subjective Construal, Social Inference, and Human Misunderstanding Dale W. Grifln and Lee Ross Index

This Page Intentionally Left Blank