Contributions to the social identity approach to conformity and intragroup differentiation

Contributions to the Social Identity Approach to Conformity and Intragroup Differentiation António José Miguel Cameira PhD dissertation supervised by Prof. Dr. José M. Marques and Prof. Dr. Michael A. Hogg Faculty of Psychology and Educational Sciences University of Porto 2005 Dedicated to my father 3 Acknowledgements This work would never be possible without the support of Profs. José Marques and Michael Hogg. Prof. José Marques introduced me to Social Psychology and was an endless source of knowledge in his classes, talks, writings and opinions; he also provided me experiences that gave meaning to this knowledge. The educational opportunities he provided me are countless. Prof. Michael Hogg’s writings and talks were fundamental for my education. He sponsored me beyond expectations, providing opportunities and experience, namely, during my staying in the University of Queensland under his supervision. For these reasons, I am deeply indebted to them both. I also wish to acknowledge the contributions of Profs. Nyla Branscombe and Dominic Abrams to my socio-psychological education with their illuminating talks, writings and personal conversations. For similar reasons, I also wish to thank Profs. Fabio Lorenzi-Cioldi, Jorge Vala, and Maria Benedita Monteiro. Thanks also to Prof. Gabrielle Poeschl for helping to create the conditions for me to work. The discussions with my colleagues Isabel Pinto and Rui Serôdio were important moments to clarify points of view, and improve down-to-earth but critical experimental procedures. I wish to thank them for their companionship along the production of this work. Finally, I wish to thank my family, especially Claudia, my mother and my kids, for their support in this process. I am sure that without their moral support, the present work would not be possible. This PhD was financially supported by the Faculty of Psychology and Educational Sciences, the Centre of Psychology of the University of Porto, and Project PRAXIS XXI/P/PSI/13192/98 of the Portuguese Foundation for Science and Technology 4 Resumo Contributos para uma abordagem de identidade social à conformidade e à diferenciação intragrupal No presente trabalho, pretendemos obter evidência para o carácter implícito da influência informacional de referência (e.g., Hogg & Turner, 1987; Turner, 1987; 1991) e da diferenciação normativa intragrupal visando o estabelecimento de uma identidade social positiva, a Dinâmica de Grupos Subjectiva (e.g., Marques, Páez & Abrams, 1998; Marques, Abrams & Serôdio, 2001). No primeiro estudo empírico, examinamos a conformidade às normas implícitas dos grupos em função da sua credibilidade. Na primeira experiência, os participantes (n = 26), depois de observarem a selecção de imagens de um júri de peritos, caracterizada por uma norma implícita, avaliaram outras imagens semelhantes ou às aprovadas ou às rejeitadas pelo júri. Os resultados sugerem que a norma implícita do júri afectou as avaliações subsequentes dos participantes. A segunda experiência (n = 35) mostrou que apenas os estudantes das áreas de Letras mas não os de Ciências, estereotipicamente considerados, respectivamente, credível e não credível na apreciação estética, se conformaram mais à norma implícita do endogrupo que à do exogrupo. Na terceira experiência, os participantes (n = 37), foram categorizados num de dois grupos artificiais, respectivamente, credível e não credível na tarefa. Os resultados mostraram que a conformidade explícita, a identificação ao grupo e a percepção de credibilidade dos grupos estavam correlacionadas, particularmente quando o grupo era credível. Na quarta experiência (n = 57) replicámos os resultados anteriores, e verificámos ainda que a conformidade explícita e a implícita diferiam em magnitude mas não nas normas endossadas. Globalmente, os resultados são consistentes com o modelo da Influência Informacional da Referência. O segundo estudo examina a possibilidade de diferenciação normativa implícita. Na primeira experiência, os participantes (n = 31) categorizados em dois grupos artificiais, sobreavaliaram os membros do endogrupo relativamente aos do exogrupo não os membros que adoptavam as normas implícitas do endogrupo sobre os que adoptavam as do exogrupo. Na segunda experiência (n = 25), focada nas normas, obtivemos uma sobreavaliação dos membros que adoptaram a norma endogrupal implícita relativamente aos que adoptaram a norma exogrupal. Os resultados das duas experiências são discutidos à luz do modelo da Dinâmica de Grupos Subjectiva. 5 Résumé Contributions à l’approche d’identité sociale à la conformité et à la différentiation intragroupe Dans le présent travail, on vise obtenir évidence pour le caractère implicite de l’influence informationnelle de référence (e.g., Hogg & Turner, 1987; Turner, 1987; 1991) et de la différentiation normative intragroupe destinée à réussir une identité sociale positive, la Dynamique des Groupes Subjective (e.g., Marques, Páez & Abrams, 1998; Marques, Abrams & Serôdio, 2001). Dans la première étude, on examine la conformité endogroupale implicite en fonction de sa crédibilité. Dans la première expérience, les participants (n = 26), après avoir observé la sélection d’images d’un jury de experts, caractérisée par une norme implicite, ont jugé des images similaires ou aux approuvées ou aux rejetées par le jury. Les résultats suggèrent que la norme implicite du jury a affecté les jugements subséquents des participants. La seconde expérience (n = 35) a montré que les étudiants d’Humanités, mais non ceux de Sciences, stéréotypiquement considérés, respectivement, crédibles et non crédibles à l’appréciation esthétique, se sont conformé plus à la norme implicite de l’endogroupe qu’à celle de l’exogroupe. Dans la troisième expérience, les participants (n = 37), ont été catégorisés en un de deux groupes artificiels, respectivement, crédible et non crédible à la tâche. Les résultats ont montrés que la conformité explicite, l’identification au groupe et la perception de crédibilité des groupes étaient associées, particulièrement quand le groupe était crédible. Dans la quatrième expérience (n = 57) on a répliqués les résultats antérieurs et on a encore vérifié que la conformité explicite et celle implicite différait en intensité mais non aux normes soutenues. Globalement, les résultats sont consistent avec le modèle de l’Influence Informationnelle de Référence. La seconde étude examine la possibilité de différentiation normative implicite. Dans la première expérience, les participants (n = 31) catégorisés en deux groupes artificiels, ont surévalué des membres endogroupales relativement aux membres exogroupales mais pas les membres qui adoptaient la norme implicite de l’endogroupe sur ceux qui adoptaient celle de l’exogroupe. Dans la seconde expérience (n = 25), on a obtenue une surévaluation des membres qui adoptaient la norme implicite endogroupale relativement à ceux qui adoptaient la norme exogroupale. Les résultats des deux expériences sont discutés en base du modèle de la Dynamique des Groupes Subjective. 6 Abstract Contributions to the Social Identity Approach to Conformity and Intragroup Differentiation In the present work, we aim to obtain evidence for the implicit character of referent informational influence (e.g., Hogg & Turner, 1987; Turner, 1987; 1991) and of normative intragroup differentiation aimed to achieve a positive social identity, Subjective Group Dynamics (e.g., Marques, Páez & Abrams, 1998; Marques, Abrams & Serôdio, 2001). In the first study, we examined conformity to implicit group norms as a function of their credibility. In the first experiment, participants (n = 26), after observing the selection of abstract pictures made by a jury of experts, characterized by an implicit norm, judged other pictures similar to either the ones approved or rejected by the jury. The results suggest that the implicit norm of the jury affected participants’ subsequent evaluations. The second experience (n = 35) showed that Arts students, but not Sciences students, stereotypically considered, respectively, credible or non-credible in esthetical appreciation, conformed more to the implicit norm of the in-group than to that of the out-group. In the third experience, participants (n = 37), were categorized in one of two group artificial groups, respectively, credible and non-credible in the task. The results showed that explicit conformity, group identification and perception of groups’ credibility were correlated, particularly when the group was credible. In the fourth experience (n = 57), we replicated the previous findings, and confirmed that explicit and implicit conformity differed in degree but not in the endorsed norms. In the whole, the results are consistent with the Referent Informational Influence model. The second study focused on the possibility of implicit normative differentiation. In the first experiment, participants (n = 31) categorized in two artificial groups upgraded in-group over out-group members, but not members that endorsed the implicit norm of the ingroup over those that endorsed that of the outgroup. In the second experience (n = 25), focused on the norms, we found an upgrading of members that adopted the implicit ingroup norm relatively to members that adopted the equivalent out-group norm. The results of the two experiments are discussed at the light of Subjective Group Dynamics. 7 INDEX INTRODUCTION................................................................................................................................... 11 CHAPTER 1 THE SOCIAL IDENTITY APPROACH .............................................................................................. 15 Metatheoretical Conceptions....................................................................................................................... 15 SOCIAL IDENTITY THEORY....................................................................................................................... 18 Social Categorization....................................................................................................................... 18 Social Identity and Social Comparison............................................................................................ 21 Intergroup and Interpersonal Comparisons................................................................................................. 22 Intergroup and Interpersonal Behavior ........................................................................................... 23 Social Categorization and Intergroup Differentiation..................................................................... 24 Strategies for Positive Group Differentiation .............................................................................................. 25 SELF-CATEGORIZATION THEORY ............................................................................................................. 28 The Hierarchical Model of the Self.................................................................................................. 28 Relationships among Self-Categorizations................................................................................................... 28 The Salience of Group Memberships ............................................................................................... 30 Depersonalization ............................................................................................................................ 32 Social Attraction............................................................................................................................... 33 Motives for Self-Categorization ....................................................................................................... 35 Conclusion........................................................................................................................................ 37 CHAPTER 2 GROUP INFLUENCE AND CONFORMITY...................................................................................... 38 Sherif’s (1936) Studies on the Formation of Norms......................................................................... 38 Asch’s (1951) Studies on Conformity............................................................................................... 40 Theory of Informal Social Communication...................................................................................... 41 Group Locomotion....................................................................................................................................... 42 The Achievement of Uniformity.................................................................................................................... 43 The Distinction between Normative and Informational Influences.................................................. 44 Deutsch and Gerard’s (1955) Experiment................................................................................................... 45 Minority Influence............................................................................................................................ 47 Tests to the Conversion Model..................................................................................................................... 49 Group Polarization .......................................................................................................................... 50 The Traditional Model of Social Influence....................................................................................... 52 Objections to the Traditional Model of Influence ............................................................................ 54 THE SOCIAL IDENTITY APPROACH TO SOCIAL INFLUENCE ........................................................................ 56 Referent Informational Influence ..................................................................................................... 56 Conformity as Assimilation to In-Group Representations............................................................................ 57 Informational and Normative Nature of Group Influence............................................................................ 59 Social Influence Phenomena from a Social Identity Perspective................................................................. 60 Conclusion................................................................................................................................................... 64 CHAPTER 3 SOCIAL IDENTITY, IMPLICIT PROCESSING AND GROUP INFLUENCE.............................. 67 THE ASSOCIATION BETWEEN SELF AND IN-GROUP REPRESENTATIONS ...................................................... 67 IMPLICIT IN-GROUP BIAS........................................................................................................................ 69 IMPLICIT OUT-GROUP DEROGATION....................................................................................................... 70 IMPLICIT PREJUDICE .............................................................................................................................. 71 STUDY 1 CONFORMITY TO IMPLICIT GROUP NORMS .............................................................................................. 74 OVERVIEW OF THE STUDY ....................................................................................................................... 76 EXPERIMENT 1 ACCEPTANCE OF A CREDIBLE SOURCE’S IMPLICIT NORM......................................................................... 76 Method ............................................................................................................................................. 77 Results.............................................................................................................................................. 82 Discussion........................................................................................................................................ 84 8 EXPERIMENT 2 CONFORMITY TO IMPLICIT NORMS AS A FUNCTION OF SOURCE’S CREDIBILITY AND GROUP MEMBERSHIP.85 Pilot Study 1..................................................................................................................................... 85 Pilot Study 2..................................................................................................................................... 86 Main Experiment.............................................................................................................................. 89 Method ............................................................................................................................................. 89 Results.............................................................................................................................................. 89 Discussion........................................................................................................................................ 92 EXPERIMENT 3 CONFORMITY TO EXPLICIT NORMS AS A FUNCTION OF SOURCE’S CREDIBILITY AND SOURCE’S GROUP MEMBERSHIP.......................................................................................................................................... 94 Method ............................................................................................................................................. 95 Results.............................................................................................................................................. 97 Discussion...................................................................................................................................... 100 EXPERIMENT 4 EFFECTS OF SOURCES’ CREDIBILITY AND GROUP MEMBERSHIP ON CONFORMITY TO EXPLICIT AND IMPLICIT NORMS ..................................................................................................................................................102 Method ........................................................................................................................................... 102 Results............................................................................................................................................ 102 Discussion...................................................................................................................................... 110 CONCLUSION.........................................................................................................................................111 CHAPTER 4 SOCIAL IDENTITY AND NORMATIVE DIFFERENTIATION....................................................113 TRADITIONAL RESEARCH ON NORMATIVE DIFFERENTIATION ...................................................................113 Group Pressures towards Uniformity........................................................................................................ 113 Differential Reactions to Deviance............................................................................................................ 115 Status and Deviance................................................................................................................................... 116 Positive Differentiation of Minorities......................................................................................................... 117 SOCIAL IDENTITY APPROACHES TO NORMATIVE DIFFERENTIATION..........................................................118 Intragroup Differentiation as a Function of Members’ Prototypicality......................................... 120 Prototypicality and Attractiveness............................................................................................................. 120 Prototypicality and Leadership Endorsement............................................................................................ 122 Differentiation as a Function of Members’ Contributes for Positive Social Identity..................... 124 Black-Sheep Effect..................................................................................................................................... 124 Subjective Group Dynamics....................................................................................................................... 126 The Development of Sensitivity to In-Group Deviance.............................................................................. 131 Emotional Consequences of Own Normative and Deviant Behavior......................................................... 132 Conclusion................................................................................................................................................. 134 STUDY 2 IMPLICIT NORMATIVE DIFFERENTIATION OF GROUP MEMBERS...............................................................136 EXPERIMENT 5 Method ........................................................................................................................................... 137 Results............................................................................................................................................ 139 Discussion...................................................................................................................................... 144 EXPERIMENT 6 Method ........................................................................................................................................... 146 Results............................................................................................................................................ 148 Discussion and Conclusions........................................................................................................... 151 GENERAL CONCLUSION...................................................................................................................153 Group Influence ............................................................................................................................. 153 Normative Intragroup Differentiation............................................................................................ 158 REFERENCES .......................................................................................................................................161 APPENDIX 1 PRE-TEST OF THE MATERIAL USED IN EXPERIMENTS 1-5...................................................182 APPENDIX 2 PRE-TEST TO EXPERIMENT 1.........................................................................................................186 9 INDEX OF TABLES TABLE 1. REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES GROUPED BY BALANCE CONDITION (EXPERIMENT 1). ..................................................................................................................82 TABLE 2. REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES ORDERED BY BALANCE CONDITION AND PARTICIPANTS’ GROUP (EXPERIMENT 2).. ..................................................................... 90 TABLE 3. REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES IN THE IMPLICIT NORM CONDITION ORDERED BY BALANCE CONDITION AND PARTICIPANTS’ GROUP (EXPERIMENT 4). ............. 105 TABLE 4. MEANS OF CONFORMITY TO IN-GROUP AND OUT-GROUP NORMS (STANDARD DEVIATION IN ITALICS) AS A FUNCTION OF GROUPS’ COMPETENCE AND TYPE OF NORM (EXPERIMENT 4).................... 106 TABLE 5. CORRELATIONS BETWEEN RELATIVE IN-GROUP IDENTIFICATION, RELATIVE PERCEIVED COMPETENCE AND RELATIVE IN-GROUP CONFORMITY WITHIN EACH BETWEEN-PARTICIPANTS CELL (EXPERIMENT 4).................................................................................................................................... 109 TABLE 6.REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES (EXPERIMENT 5). .............. 141 TABLE 7. REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES (EXPERIMENT 6). ............. 149 TABLE 8. REPORTS ON THE DISTINCTION BETWEEN THE TWO TYPES OF PICTURES (PRE-TEST). ...................... 184 TABLE 9. REPORTS ON THE DISTINCTION BETWEEN THE TWO SERIES OF PICTURES GROUPED BY BALANCE CONDITION (PRE-TEST TO EXPERIMENT 1). ........................................................................................... 188 16 Kunda, 1986). The emphasis on the individual is also visible on approaches to social phenomena focusing the interpersonal dimension of social relationships (for critic perspectives, cf., e.g., Doise, 1986; Hogg, 1992; Hogg & Abrams, 1988). The interactionist paradigm. The interactionist paradigm emphasizes the dynamic nature of the self and the idea that individuals’ self-conception, and, consequently, their opinions, judgments and behavior change in the course of social interaction. The content of the mind is the product of social interaction, a socialization process during which individuals internalize symbols, meanings, values and norms (cf., e.g., Mead, 1934). Thinking and meaningful action are possible only to the use of these tools, the social contents of the mind. As Vigotsky (1925) emphasized, “All higher mental functions are the essence of internalized relations of a social order, a basis for the social structure of the individual. Their composition, genetic structure, method of action – in a word, their entire nature – is social; even in being transformed into mental processes, they remain quasisocial. Man as an individual maintains the functions of socializing.” (p.106) It is, thus, impossible to dissociate individuals from the social groups to which they belong. “Individuals in their multiplicity cannot be opposed to or in reality distinguished from society: individuals are society and society is the natural form of being of human individuals. […] The fallacy that the individual may be opposed to society arises from a legitimate but different contrast between a particular individual and others, resulting in the idea of uniqueness or individuality […] There is no such thing as the pre-social, asocial, purely biological, ‘as if isolated’ individual except as analytic, fictional abstraction.” (Asch, 1952, p. 239) The object of Social Psychology. For the interactionist paradigm, the object of Social Psychology is not that of the individual Psychology or that of Sociology, but the mental representations and processes produced by social interaction. “The task of Social Psychology, as a part of psychology (the science of the individual mental processes), is not to provide social explanations of 17 behavior (this can be left to sociology, politics, etc.), nor to provide ‘psychological explanations’ of, i.e. ‘to psychologize’ social behavior, but to explain the psychological aspects of society. This equates with understanding the structures and processes whereby society is psychologically represented in and mediated by individuals’ minds.”(Turner & Oakes, 1986, pp. 239-240) More specifically, the object of Social Psychology is the “socially mediated cognition, phenomenologically experienced as the perception of a shared, public, objective world”, the “psychological representations of the interaction, interactors, setting etc. shared by the interactors”, and the “intersubjective world of shared social meanings.” (Turner & Oakes, 1986, p. 240). Concepts of group. The two metatheories result in different conceptions of group. According to Turner (1987), the two conceptions derive from a fundamentally different meaning attributed to interdependence. In the interaccionist approach, from Lewin (1947) to Sherif (1967), or Asch (1952), interdependence is considered “in the sense of ‘functional unity’, or ‘dynamic system’, or ‘mutual psychological field’” (Turner, 1987, p. 20). Interdependence refers thus to the proprieties of the group, as a functional whole, and their effect on the members. In the individualistic metatheory, group interdependence refers to individuals’ dependence on others to satisfy their individual needs. “In recent years the emphasis has been on forms of interdependence related to or assumed to be related to the satisfaction of individuals’ needs or, which is the same thing theoretically, the achievement of rewarding outcomes, i.e., motivational interdependence. By motivational interdependence is meant the idea that actions and characteristics of others relevant to the satisfaction of one’s needs are functionally related by the structure of the situation to actions and characteristics of one’s own relevant to their needs. Thus, at one extreme, people may co-operate to achieve some otherwise unattainable goal; at the other, they may associate simply because they find each other’s company mutually rewarding.” (Turner, 1987, p. 20) 18 In the individualistic conception of interdependence, the group is the result of the interpersonal relationships among members (e.g., Hogg, 1992; 1993; Hogg & Turner, 1985a; 1985b; 1987; Turner, 1984; Turner et al, 1987). “The group whole, the social norm, is assumed to be exactly the sum of its parts, the members’ individual opinions – there is no gain or loss in collective wisdom […] A group is merely the product of interpersonal relations and processes or, more precisely, the same thing as relatively stable relations of interpersonal co-operation, attraction and influence between people” (Turner, 1987, p. 25). In contrast, for the interactionist approach, the group is a meaningful entity relatively independent from its individual components, possessing different proprieties from individuals (Asch, 1952; Lewin, 1947; Sherif, 1967). Both Social Identity Theory and Self-Category Theory reject the individualistic metatheory and assume the principles of the interactionist approach to social phenomena. Social Identity Theory Social Identity Theory (e.g., Tajfel & Turner, 1979; 1986; Tajfel, 19781; 1981; 1982; Turner, 1975) proposes three basic processes to understand group behavior: social categorization, social identification and social comparison (cf. Tajfel, 1978; Tajfel, 1982; Tajfel & Turner, 1979). Social Categorization Categorization is a pervasive phenomenon that fulfills a basic individual need: the perception of clear and predictable environments (cf. Bruner, 1957). It simplifies and systematizes the environment, ordering it in a manner that makes sense for the individual, thus helping to structure its causal understanding (Tajfel, 1978). Inductive and deductive aspects of categorization. To achieve a clear perception of the environment, the cognitive system processes stimuli inferentially. Two aspects of inferential processing are crucial to achieve a simplification of the environment: the 1 In the present chapter, the reference Tajfel (1978) stands for Tajfel’s Introduction and three chapters in the edited book “Differentiation between social groups”. 19 inductive and the deductive aspects. The inductive aspect consists in recognizing shared attributes in a number of objects in order to form a category. The deductive aspect consists in associating the category attributes to the attributes recognized in an object. The inferential processing of stimuli implies the modification of the environment because it is based on the omission, as well as, overattribution of features to the objects and events so that they may be integrated in the pre-existing structures. In Bruner’s (1957) terms, “perceptual categorization of an object or events permits one to ‘go beyond’ the proprieties of the object or event” (p. 129). By making the environment a predictable world, categorization allows action to can take place expediently, without a prior time- and energy-consuming interpretation of the objects and circumstances involved (cf. Tajfel, 1972). As an illustration of the deductive transformation of stimuli, Tajfel and Wilkes (1963) asked their participants to judge the length of each line in various series of eight lines whose length progressed geometrically. The series were either classified, i.e., labeled A or B, so that the shorter lines belonged to category A and the longer ones belonged to category B; classified randomly, i.e., the categories A and B assigned to each line were not correlated to the lines’ length; or unclassified. The results showed that participants in the criteria categorization condition exaggerated the differences in length between the two categories. Moreover, these participants also tended to judge the lines within each category to be more similar to one another, than the participants in the other conditions did. Tajfel and Wilkes (1963) dubbed the accentuation of intracategorical similarities and of inter-categorical differences, the accentuation effect. The Accentuation Effect in Social Categories The categorization of physical objects and the categorization of people in continuous dimensions – defining attributes such as traits or personality characteristics – follow identical rules (e.g., Doise, 1976). However, the structuring of social environment differs, in several aspects, from the categorization of the physical environment (cf. Deschamps, 1984; Doise, Deschamps & Meyer, 1978; Doise & Sinclair, 1973; Eiser & Stroebe, 1972; Tajfel, Sheik & Gardner, 1964). 20 For instance, there is often higher involvement of the perceiver in social categorization (cf. Tajfel, 1969; 1970; 1972). Burris and Branscombe (2005) provide a recent illustration of the effect of self-relevance on the accentuation of intercategorical differences. The authors found that the distances between the city of the respondent and another city in the same country (U.S.A.) were estimated as smaller than that to a city situated in a neighboring country (Canada or Mexico), even when the former distances were objectively larger than the latter. Similar to any categorization, the notion of countries entails an accentuation effect of the distance between the cities of different countries, what the authors dubbed the border effect. Most significantly, the authors also found that when estimating the distances between two cities located in two foreign countries (Mexico and El Salvador), the border effect disappeared. As the authors concluded, “crossing the psychological boundary between self and not-self creates a visual illusion that distorts on-line distance estimates” (p. 305). The accentuation of intercategorical similarities is also affected by the involvement of self. For instance, Simon (1993) claims that when the in-group is relevant for the self, individuals are more to accentuate the similarities within the ingroup than within out-groups. The author and his colleagues (e.g., Simon & Brown, 1987; Simon & Mummendey, 1990; Simon & Pettigrew, 1990) found that the minority group members accentuate the similarities within the in-group whereas majority group members accentuate the similarities within the out-group (cf. Chapter 4). The results are explained by the higher relevance of the minority in-group membership for the self as regards that of the majority (cf. Brewer, 1991, 1993). In sum, the involvement of the self in social categorization confer it distinct characteristics as regards categorization of objects. Social values also play a major role in the structuring of social environment (e.g., Maass & Schaller, 1991; Oakes, Haslam & Turner, 1994; Tajfel & Forgas, 1981), although the categorization of objects is also affected by the social value ascribed to them (e.g., Bruner & Goodman, 1947). The Value Dimension of Social Categorization As Tajfel and Forgas (1981) put it, “no social domain of any importance is classified in a completely value-free manner, as currently prevalent cognitiveindividualistic research would lead us to believe” (p. 126). Social categories are always, 21 to some extent, charged with an evaluative connotation. In fact, the evaluative dimension seems to precede the descriptive dimension in the formation of categories. As Tajfel and Forgas (1981) claim, “categorizations of people, groups of people and social events in terms of value differentials are probably one of the earliest and most basic forms of social categorization” (p. 119). For instance, Tajfel and Jahoda (1966) asked British 6-7 years old children to indicate among plastic squares of different sizes, which matched the surface of America, France, Germany, and Soviet Union; then, the authors asked the same children to indicate their preferences for each of these four countries. Children were more consensual in preferring America and France to Germany and Soviet Union than to say that America and Soviet Union are larger than France and Germany. In sum, the four countries were distinguished based on the associated value more than through their objective attribute. As the authors claimed, a consensus on the relative value of the nations is likely to confer as much legitimacy to the differentiation between the four countries, as a descriptive attribute such as size. As Tajfel (1969, p. 87) remarks, “[…] ‘bad’ and ‘good’ even ‘liked’ or ‘disliked’ become incontrovertible statements of fact not different in their mode of assimilation from, for example, ‘large’ or ‘small’”. Above, we mentioned the effects of self-involvement in social perception. We shall now address how social categories affect self-perception by becoming meaningful aspects of individual identity. The categorized perception of the social environment and the value associated to social categories affects both how individuals perceive others as belonging to social categories and how they perceive themselves in relation to others. As Tajfel (1972) puts it, the categorization of social environment provides a system that creates and defines the particular place of the individual in the society, his/her social identities. Social Identity and Social Comparison Social identity consists of those aspects of “an individual’s self-image that derive from social categories to which he perceives himself as belonging” (Tajfel & Turner, 1979, p. 40). If it is true that, “the image or concept that an individual has of himself or herself is infinitely more complex, both in its contents and its derivations, than social identity as defined and circumscribed here” (Tajfel, 1978, p. 63), it is also 22 true that some aspects of that image are the result of membership of certain social groups and categories, and that some of these aspects will determine individual’s behavior in the course of his/her life. Cognitive, evaluative and emotional components of social identity. Social identity comprises three components, “[…] a cognitive component, in the sense of knowledge that one belongs to a group; an evaluative one, in the sense that the notion of group and/or of one’s membership of it may have a positive or negative value connotation; and an emotional component in the sense that the cognitive and evaluative aspects of the group and one’s membership of it may be accompanied by emotions (such as hatred, like or dislike) directed towards one’s own group and towards others which stand in certain relations to it.” (Tajfel, 1978, p. 28) Both the cognitive and evaluative dimensions of social identity are established in comparison to out-groups. In fact, social identifications “are to a very large extent relational and comparative: they define the individual as similar to or different from, as “better” or “worse” than, members of other groups” (Tajfel & Turner, 1979, p. 40). Intergroup comparisons are thus an inevitable outcome of social identity. Intergroup and Interpersonal Comparisons Intergroup comparison is similar to social comparison as described by Festinger (1954). However, in intergroup comparison, individuals compare with others not as individuals but as members of groups (cf. Tajfel, 1978). The hypothesis of a drive to evaluate one’s abilities is seen in terms of individuals need to evaluate themselves as members of a group. Furthermore, intergroup comparison resembles interpersonal comparison only in what refers to comparison of abilities, not opinions (cf. Turner, 1975). According to Festinger (1954), individuals are motivated to ascertain the accuracy of their opinions, but when comparing abilities, they are motivated to achieve positive positions. As Turner (1975) remarks, more than achieving a precise position of the in-group as regards the out-group, individuals are motivated to achieve a positive social comparison, “the important comparative dimensions for social identity parallel those of abilities rather than opinions, i.e., they are value-laden […] the need to evaluate 23 himself in society is more correctly expressed as a need to make a favorable or positive evaluation of the individual in society.” (p. 9). Intergroup and Interpersonal Behavior The behavioral outcome of social identity is intergroup behavior. Tajfel (1978) defines intergroup behavior as “all behavior of two or more individuals towards each other [that] is determined by their membership of different social groups or categories.” (p. 41). For Tajfel (1978), during a limited social interaction, intergroup behavior appears combined with interpersonal behavior in a varying degree depending on individuals’ perception of the situation. The author proposes a continuum of interindividual behavior – intergroup behavior, in which the interpersonal pole refers to ‘any social encounter between two or more people in which all the interaction that takes place is determined by the personal relationships between the individuals and their respective individual characteristics’; and the intergroup pole refers to ‘all behavior of two or more individuals towards each other determined by their group membership of different social group or categories’ (p. 41). The variability of social interactions along the continuum predicts that, with few exceptions, social situations fall between the two extremes and, depending on the pole they tend to, the situation may be classified as having more an intergroup or an interpersonal character. Variability–uniformity continuum. Intergroup behavior is characterized by less variability in the relations with other persons. “The nearer is a social situation (as interpreted by members of a group) to the intergroup extreme of the interpersonal-intergroup continuum, the more uniformity will the individual members of the groups concerned show in their behavior towards members of out-groups. Conversely, the nearer is the situation to the interpersonal end, the more variability will be shown in behavior towards members of out-groups” (Tajfel, 1978, p. 44-45) The uniformity of intergroup behavior is associated with the perceptual accentuation of the differences between in-group and out-group and of the similarities among their respective members (cf. above). “The nearer is a social situation to the intergroup extreme, the stronger tendency will there be for members of the in-group to 24 treat members of the out-group as undifferentiated items in an unified social category, i.e., independently of the individual differences between them.” (Tajfel, 1978, p. 45). Social Categorization and Intergroup Differentiation Social identification implies that members compare with other groups motivated to achieve a distinct and positive position of the in-group. A consequence of the need for positive group differentiation is the establishment of competitive intergroup relations. In their seminal study, Tajfel, Flament, Billig and Bundy (1971) showed that this process emerged even in minimal laboratory conditions whose goal was “to establish minimal conditions in which an individual will, in his behavior, distinguish between in-group and out-group” (Tajfel, 1978, p. 77). Tajfel and colleagues (1971; Experiment 1) categorized the participants allegedly as a function of their over- or under-estimation of number of projected dots. Subsequently, without being given the opportunity to interact among them, participants, individually and anonymously, awarded amounts of money to three pairs of recipients based solely on their group membership. The analysis of the intergroup choices in matrices presenting members of the two groups showed that participants awarded significantly more points to in-group than to out-group recipients. The mere categorization of the participants was enough to produce a discriminatory intergroup behavior. The result is more significant when compared with the results of the same group choices in which both the recipients were either in-group or out-group members. In these sequences, participants used a more fair strategy. In Experiment 2, participants were randomly assigned to one of the two groups, based on their alleged preference for the abstract paintings of Klee and Kandinsky. The results showed that, similar to the intergroup choices in the previous experiment, the majority of the participants favored in-group recipients. As the authors put it, “[…] in a situation devoid of the usual trappings of in-group membership and of all the vagaries of interacting with the out-group, the Ss still act in terms of their group membership and of in-group categorization. Their actions are unambiguously directed at favoring the members of their in-group as 25 against the members of the out-group. This happens despite the fact that an alternative strategy – acting in terms of the greatest common good – is clearly open to them at a relatively small cost of advantages that would accrue to members of the in-group.” (p. 173) Although fairness was also an important strategy, suggesting that participants’ responses were affected by an equalitarian social norm, the achievement of positive ingroup differentiation was a major directive for their responses. The tendency to favor the in-group over the out-group in evaluations and behavior was later dubbed in-group bias, “the laboratory analog of real-world ethnocentrism” (Tajfel & Turner, 1979, p. 38). As the authors remarked, “not only are incompatible group interests not always sufficient to generate conflict [but] these conditions are not always necessary for the development of competition and discrimination between groups” (p. 38). Tajfel and colleagues’ (1971) results are of major importance in the development of Social Identity Theory (cf. Tajfel, 1972; Tajfel & Turner, 1979). They are at the origin of a whole line of research aimed to understand the minimal conditions for group behavior (e.g., Diehl, 1990). Early relevant research addressed the assumption of interpersonal similarities and differences in minimal groups (cf., e.g., Allen & Wilder, 1975; Billig & Tajfel, 1973; Diehl, 1988) or the involvement of the self in intergroup differentiation (cf. Turner, 1975; 1978). More recently, Otten, Mummendey and Blanz (1996), found that when allocating negative outcomes (lists of meaningless syllables to memorize), did not discriminated out-group members in the same way as when allocating positive outcomes (money). Participants in the negative outcome conditions used fairer intergroup strategies, except when the value of group membership was threatened; in this case, participants reacted favoring the in-group over the out-group in the allocation of both positive and negative outcomes. Strategies for Positive Group Differentiation Tajfel and colleagues’ (1971) experiments depicted the strategies adopted in laboratory groups. In the real world, the strategies for positive group differentiation vary as a function of the status of the in-group and might not involve competitive intergroup relations. The social values ascribed to each group in the social system, affects the way the respective members experience their social identity (Tajfel, 1972; Tajfel & Turner, 32 perceiver readiness and normative fit – reflecting an individual’s motives, desires, memories, knowledge, habits and so forth – provide definite internal psychological constraints on self-category variation” (Turner et al, 1944, pp. 459-60). However, the alleged stability of the self derives more from the relative invariability of the social contexts in which individual normally interact than from the predominance of personal aspects. The relatively stable set of roles occupied by individuals in their social network, and the recurrent social contexts in which individuals move, lead to the frequent activation of some self-categories in detriment of others. However, this is due to the stability of the social context rather than to the stability of cognitive structures. Depersonalization Depersonalization is the phenomenological outcome of social selfcategorization. An inclusive level of self-categorization implies that there is a perceptual amplification of the similarities between the self and the categorical attributes (the prototype) and a perceptual reduction of the differences from these attributes (cf., e.g., Haslam & Turner, 1992; Simon, Pantaleo & Mummendey, 1995). “Just as categorizing others causes us to perceive others stereotypically, categorization of the self causes us to perceive ourselves stereotypically. It is in this way that the individual comes to describe the self in terms of the group’s defining characteristics, experience the feelings and emotions of the group, and to acquire the goals and needs of other group members.” (Hogg & Turner, 1987, p. 149) Depersonalization and deindividuation. Depersonalization is conceptually different from deindividuation, which corresponds to a psychological process of reduced self-awareness (e.g., Diener, 1980; Prentice-Dunn & Rogers, 1982). Selfawareness includes attention to aspects, such as one’s attitudes and norms and increases the capacity for self-regulation. Deindividuation is thus associated with anti-normative and disinhibited behavior (but, cf. Reicher, 1987). “[Depersonalization] is not a loss of individual identity, nor a loss or submergence of the self in the group, and nor any kind of regression to a more primitive or unconscious form of identity. It is the change from the personal to 33 the social level of identity, a change in the nature and content of the self-concept corresponding to the functioning of self-perception at a more inclusive level of abstraction.” (Turner et al, 1987, p. 51) The salience of a self-category does not imply self-attention in the same way that self-attention does not necessarily imply a shift to the personal self. As Abrams and colleagues (e.g., Abrams, 1994; 1999; Abrams & Masser, 1998) argue, the two processes correspond to different and independent dimensions of the self. The shift to a social self-categorization and depersonalization is an adaptive response to changes in the context (e.g., Turner et al, 1994), and individuals are generally not aware of the process. However, this does not mean that we cannot be self-conscious when acting as group members. Planned and strategic group behavior, for instance, requires a focus on the collective rather than on the personal self (Abrams, 1999). Social Attraction A consequence of depersonalized perception of self and others are the positive feelings towards other in-group members, a phenomenon that Hogg and colleagues (e.g., Hogg, 1987; 1992; 1993; Hogg & Turner, 1985a; 1985b) dubbed social attraction: “a complex of positive feelings, a positive attitude, towards fellow group members, which is accentuated under circumstances that enhance one’s sense of belonging to the group” (Hogg, 1992, p. 100). Social attraction and interpersonal attraction. Social attraction is different from interpersonal attraction, although both entail positive feelings towards other individuals. Personal attraction is specific of interpersonal relationships. It is a positive feeling generated by distinct personal traits and other personal characteristics that do not relate to any significant social category. In contrast, social attraction is a positive feeling that emerges among individuals that define themselves as members of the same group. In fact, as Hogg (1992) put it, social attraction “is actually attraction to the group as that group is embodied by specific group members, so that the object of positive attitude and feelings is not actually the unique individual person, but the prototype that s/he embodies.” (p. 100). As Hogg (1992) remarks, personal and social attraction may coexist. For instance, in small interactive groups (as groups of friends or work teams), 34 in which members interact repeatedly, people may feel attracted to each other based on common group membership and based on each other’s distinctive traits. Group cohesiveness. The above distinction is important to analyze group cohesiveness. Traditionally, group cohesiveness was measured by the number of friendships and other interpersonal relationships among members (e.g., Lott & Lott, 1955; Festinger, Schachter & Back, 1950). In contrast, Hogg (e.g., 1987; 1992) argues that group cohesiveness is generated by attraction to the group as a whole, not by interpersonal attraction among members. Group cohesiveness is a group propriety, which is related with social identity. This view explains why large social groups as nations, parties or religious sects may also be evaluated in terms of cohesiveness. Empirical evidence. Hogg and Turner (1985a) provided empirical support for the distinction between interpersonal and social attraction. The authors categorized or not participants allegedly according to their similarity with other members in personality traits that were mostly attractive or unattractive. The results showed that, independently of the attractiveness of the traits that defined the groups, participants were more attracted to in-group than to out-group members. Most significantly, categorized participants considered themselves similar to in-group attractive members but not to equivalent out-group attractive members, whereas non-categorized participants considered themselves similar to both in-group and out-group attractive members (cf. also, Hogg, Cooper-Shaw & Holzworth, 1993; Hogg & Hardie, 991; Hogg, Hardie & Reynolds, 1995). In sum, social attraction is motivated by a salient common group membership leading to the depersonalized perception of self and others. People feel attracted to each other because they have a general feeling of sharing the same social and moral standards, the same assumptions about social reality, about the values that should rule one’s life, etc. Social attraction is thus an outcome of self-categorization and of depersonalized perception of self and others. We shall elaborate more on social attraction in Chapter 4 while addressing the issue of intragroup differentiation. 35 Motives for Self-Categorization What motives lead individuals to self-categorize? As seen above, individuals attempt to achieve positive social differentiation of the in-group as regards out-groups (Tajfel, 1978). Individuals strive for a clear perception of the in-group boundaries and for a positive intergroup comparison. The motives invoked for self-categorization are thus associated with the need for self-enhancement (e.g., Abrams & Hogg, 1988; Hogg & Abrams, 1990) and with the need for uncertainty reduction (e.g., Hogg & Abrams, 1993; Hogg, 2000). Brewer (1991; 1993) proposes another relevant motive for selfcategorization the achievement of optimal distinctiveness. Self-esteem motive. The first hypothesis raised was the motivation for positive self-esteem. According to Abrams and Hogg (1988; Hogg & Abrams, 1990), the selfesteem hypothesis has two corollaries: Low self-esteem can motivate people to identify with groups that raise self-esteem, or it can improve the evaluation of groups to which they belong. However, the relationship between self-esteem and self-categorization or social identification has proven difficult to establish empirically (Abrams & Hogg, 1988; Hogg & Abrams, 1990, 1993). Social identification is only one of the ways in which positive self-esteem can be achieved and despite the attempts to operationalized self-esteem associated with the collective self separated from self-esteem associated with the personal self (e.g., Crocker & Luhtanen, 1990), it has been difficult to measure the distinct impact of collective self-esteem in self-categorization (e.g., Rubin & Hewstone, 1998). Uncertainty Reduction Motive The uncertainty reduction hypothesis refers to the epistemic need for perceptual clarity. This need refers to subjective certainty, about one’s perceptions and beliefs (cf. Festinger, 1950; 1954) and about one’s position in the social environment (Tajfel & Turner, 1979). Certainty about ourselves in each situation enables us to know how to think, feel and behave. According to the uncertainty-reduction model (Hogg, 2000; 2001; Grieve & Hogg, 1999; Hogg & Mullin, 1999), self-categorization and depersonalization entails that “individuality and concomitant unshared cognitions, feelings, and behaviors are replaced by an in-group prototype that prescribes shared 36 cognitions, feelings, and behaviors” (p. 92). Through self-categorization, we become certain about our cognitions, feelings, and behaviors. Empirical evidence. As evidence for the uncertainty reduction hypothesis, Grieve and Hogg (1999) found that participants reported less uncertainty, than in the pre-test, when they were categorized into minimal groups and allocating points to ingroup and out-group members in Tajfel and colleagues’ (1971) matrices. Participants in the non-categorized condition did not show such uncertainty reduction (cf. also, Mullin & Hogg, 1998). Hogg and Mullin (1999) further manipulated uncertainty orthogonally to topic relevance and categorization. Specifically, participants were asked to give their opinion about topics highly relevant to the self such as, career, family and success, or low relevant to the self such as, using always the same toothpaste brand, and were given the idea that those opinions could be either more or less correct or subjective. The authors found higher in-group identification in high importance than low importance condition and more need for in-group validation of their opinions in high uncertainty than in low uncertainty conditions, particularly, in high importance conditions. Optimal Distinctiveness Motive Brewer (1991; 1993) proposes that individuals identify with groups because they wish to avoid isolation and stigmatization, and to maintain social ties with other individuals (the need for assimilation and belonging). However, they also wish to maintain some level of distinctiveness to avoid anonymity and depersonalization (the need for differentiation; e.g., Codol, 1984; Snyder & Fromkin, 1980). The existence of these two opposed needs of self-assimilation to and differentiation from others determines that individuals will identify more with groups that satisfy both needs in simultaneous. Optimal social identity corresponds to a balance between inclusion and differentiation, where the need for assimilation is satisfied within the group, and the need for distinctiveness is satisfied through intergroup comparisons. A corollary of this hypothesis is that optimal distinctiveness is achieved in mid-size groups but not in large or in too small groups. As evidence for this idea, Lau (1989) obtained a U-shaped relationship between the demographic density of Black individuals in the United States, and their reported identification with the Black race. In parts of the country where the 37 number of Afro-Americans is very low, a raise of density is associated with an increase of identification. Conversely, where density of Afro-Americans is very high, the increase of density corresponds to a decrease of identification. Conclusion In this chapter, we briefly reviewed Social Identity Theory and Self- Categorization Theory. Both theories issue from a metatheoretical background that stresses the social dimension of cognitive and emotional processes. A basic distinction is established between group and interpersonal behavior ensuing from social and personal identities, respectively. The categorized perception of the social environment, the comparisons between social categories, and the resulting evaluations and associated emotions are the underpinnings of social identities. Social identities are as important as personal identities in determining individuals’ attitudes, judgments, and behavior. Social identities, or social self-categories, are activated in a dynamic interaction between the individual’s and contextual characteristics. The self-concept is thus a flexible structure that shifts from personal to social self-categories. Depersonalization is the perceptual outcome of the shift to a social self-concept. This shift implicates that thinking, feeling, and judging are regulated against the defining standards and norms of the salient in-group, the group prototype. Social self-categorization implies that individuals are more attracted to in-group than to out-group members independently of their attractiveness at an interpersonal level. The motives for social identification are likely to be related to different individual needs, such as the need for a clear perception of reality and the reduction of uncertainty about it, the need for associated positive self-esteem and the balance between the needs for assimilation and individuality. The social identity approach, including Social Identity Theory and Self- Categorization Theory, presents an alternative view on several phenomena, among which, group influence, the subject of the next chapter. 38 CHAPTER 2 GROUP INFLUENCE AND CONFORMITY In this chapter, we shall focus first on early contributes and on the traditional approach to group influence. After a perspective on the metatheoretical underpinnings of the traditional approach, we shall focus on the social identity perspective on the topic. The literature on group influence has two major references: the studies of Sherif (1936) on the formation of social norms and Asch’s (1951) study on conformity. Festinger’s (1950) theory of Informal Social Communication is among the former attempts to theorize on group influence. These early references were the basis for Deutsch and Gerard’s (1955) distinction between informational and normative influences, an important model in the traditional literature on group influence. Two other fields of research are central in group influence literature: minority influence and group polarization. We shall begin by briefly reviewing these theories and research. Sherif’s (1936) Studies on the Formation of Norms In his studies, Sherif (1936) investigates the processes through which norms become an objective reference for individuals’ perceptions and judgments. “Once frames of reference are established and incorporated in the individual, they enter as important factors to determine or modify his reactions to the situations that he will face later – social, and even non social at times, especially if the stimulus field is not well structured.” (Sherif, 1936, p. 86) Sherif (1936) used the autokinetic effect, a perceptual illusion in which a stationary light appears to move erratically in the darkness. In this situation, there is no objective reference to estimate the light movement. In several sessions occurring in successive days, participants were asked to make, alone, an estimate of the movement from the point in which the light appeared to the point in which it disappeared. The results showed that participants developed a standard estimate, a personal norm, and that their subsequent judgments oscillated around this norm. As Sherif remarks, when no references are available to base their judgments, individuals establish their own 39 standards, “the general psychological tendency to experience things in relation to some frame of reference” (p. 81). In subsequent group sessions, participants made their judgments in groups of two or three. The results showed that participants’ estimates gradually converged to a common estimate, the group norm, in a process of reciprocal influence that evolved with the participants’ interactions. For the other half of participants, the order of sessions was reversed. First, they were included in groups and then, in subsequent sessions, they made their judgments alone. These participants quickly developed a group norm, which was kept, with negligible variations, in the subsequent individual sessions. Variations of the procedure. The variations to Sherif’s paradigmatic study are innumerous attempting to check several aspects of group influence. We shall mention only a few that are representative for the present purposes. For instance, Sherif (1936) asked a confederate to maintained his position in the exchange of estimates. Because of his reluctance to change, the other participants gradually converged to the confederate’s estimate indicating that the group norm is not the simple average of individuals’ personal norms. Other variations tested the endurance of the group norms. Rohrer, Baron, Hoffman and Swander (1954; see also Bovard, 1948) replicated Sherif’s (1936) procedure in two moments, the latter one year after the former. In the second moment, judgments were always made individually. The results showed that group norms had preponderance over personal norms on these later judgments. In another variation of the procedure, McNeil and Sherif (1976) showed that the persistence of a norm depends on its arbitrariness. In group sessions, participants made their estimates with three confederates that made either moderate or extreme estimates. Participants in the extreme estimates condition, initially conformed to the group but gradually, abandoned the group norm, moderating their estimates. In sum, Sherif and colleagues’ studies revealed that social phenomena, such as norms, have characteristics of their own that cannot be explained by individual processes. Norms are the product of social interaction rather than the average of individual opinions. Some individual have more influence than others do; however to be influential, their opinions have to be considered acceptable by the group. Norms have 40 lasting effects on the members’ opinions because, as Sherif puts it, they “[…] have a reality of their own, independent of this or that individual member of society; and this immediate reality may be verified easily by the resistance one meets when he deviates considerably from the well-established norms of his time” (1936, p. 58). Asch’s (1951) Studies on Conformity In Asch’s (1951) experiments, participants were placed in a situation in which the group consensus differed from their own perceptions aiming to “study the social and personal conditions leading individuals to resist or submit to collective pressures when these are contrary to reality” (Asch, 1951, p. 231). Specifically, the group consisting of a naïve participant and seven confederates was instructed to compare a stimulus line with three other lines with different length in 18 trials. Only one line had the same length as the stimulus line. Participants were told to voice their judgments of which line of the set of three had the same length as the stimulus line, in order so that the naïve participant responded always in last. Confederates were instructed to emit unanimously erroneous judgments. Conformity was measured by the amount of participant’s incorrect judgments. In the end of the session, participants were interviewed, checking on their opinion about the reasons for yielding or not to the group. In a control condition, participants made their judgments alone. The results showed that whereas in the control condition, participants gave, in average, less than one error in all trials, in the group condition, 32% of participants made judgments consistent with the majority’s erroneous judgments in half or more than half of trials. One fourth of participants made correct judgments in all trials. The post-hoc interviews revealed that participants conformed for different reasons. A small number of participants reported, what was called perceptive deformation, i.e., they perceived the group’s consensus as correct. The majority of conformers reported lack of confidence on their own judgments as regards the group consensus (judgment deformation). The remaining conformers reported that they were conscious that they were responding erroneously but could not bear to appear different from the group (action deformation). 41 Experimental variations. Asch (1956) conducted several variations concerning the support to non-conformity. In one case, the author instructed one confederate placed in the fourth position of the group to respond always correctly. The frequency of errors decreased to 5.5%. In another variation, one confederate was instructed to respond first correctly and, in the middle of the session, to join the group consensus. The errors’ proportion rose to 28.5%. In a third variation, a confederate was instructed to leave the majority, near the end of the experiment, and start making correct judgments. Conformity decreased to 8.7%. In a fourth variation, one confederate was instructed to respond in an intermediate point between the majority’s judgments and correct judgments. Errors’ frequency decreased and most errors were moderate. In another condition of the experiment, in which confederates made extreme errors unanimously, most errors were extreme. Asch (1956) also varied the size of the majority to 2, 3, 4, 8, 16 confederates. With two confederates, conformity rises to 12.8%. Majorities of 4, 8 and 16 did not induce conformity more than a majority of 3. In another variation, conditions were reversed. With one confederate only, conformity is negligible. One confederate was instructed to give extreme errors in a majority of 16 naïve participants. The majority rejected and mocked at his responses. Asch’s (1951) seminal study is paradigmatic in conformity research and originated innumerous variations. The reinterpretation of the results (cf. Moscovici & Faucheux, 1972) originated another line of research, minority influence, which we shall address below in the text. Theory of Informal Social Communication Festinger’s (1950) theory of Informal Social Communication is a leading theory of group influence and an important reference of the traditional approach. The theory focuses on the pressures the group exerts on its members towards uniformity of opinions within the group. Pressures towards uniformity have two functions: to ascertain social reality and to facilitate group locomotion. 48 is, participants that were exposed to the minority opinion were more likely to name green the intermediary blue-green disks than participants that were not exposed to the minority opinion. Moreover, participants that did not yield to the minority’s direct influence were more likely to yield to its indirect influence than those that had yielded to the minority’s direct influence. Minority influence: Normative or informational? The theory is not clear on whether minority influence entails normative or informational influence. For instance, Mugny and Papastamou (1980) presented evidence for the idea that when the majority members’ attention is focused on the (minority) source’s characteristics, influence decreases as regards to when they are focused only in their arguments. In addition, evidence suggests that consistent minorities exert more influence than inconsistent minorities (e.g., Mugny, 1982). As Maass and Clark (1984) argue, consistent positions are associated with informational influence because they convey the impression of confidence and certainty leading to the attributions of credibility and competence to the source. Together, these results indicate that minorities exert mainly informational influence. However, other research shows that the perception of minority as holding idiosyncratic opinions decreases substantially their persuasiveness (cf. Papastamou, 1986; Moscovici & Lage, 1976). These results indicate that minority influence depends also on the perception of a consensual point of view, and consequently, of a normative opinion (cf. Mugny & Pérez, 1991). As Moscovici (1976) put it, minority voices “call attention to the existence of a coherent point of view, to something powerful and, of course, to a norm. In a word, it forcefully indicates the nomic quality of an individual or group” (p. 139). The processing of indirect influence. According to the theory (cf. Moscovici & Personnaz, 1991), minorities’ indirect influence results from a combination of systematic information processing and unintentional acceptance. Whereas majorities’ sources are, from the outset, validated by consensus and do not motivate a systematic analysis of their arguments and opinions, minorities, because they do not benefit from consensual validation, induce a thorough analysis of the message content. According to Moscovici and Personnaz (1991), the cognitive conflict generated by the influential 49 message, is solved differently depending on whether the source is a majority or a minority. In the case of a majority source, individuals solve the conflict immediately by complying with the source. When the source is a minority, no such solution is available because the source has no normative impact over receivers. Consequently, individuals feel uncertain and unable to decide whether to yield or not to the source. According to the authors, the conflict is solved unconsciously by adopting the source’s opinion. As Moscovici (1980) put it, “The more intense the pressure, the greater the effects obtained [by the majority] on the direct, overt level, in short, on the level of the most superficial acceptance, and [by the minority] on the indirect, latent level, leading on the whole to an acceptance that may be so deep that the subject is not even aware of it.” (p. 216) Tests to the Conversion Model The assumptions about the difference nature of minority influence as regards majority influence have not obtained consistent empirical support (e.g., Doms & van Avermaet, 1980; 1985; Kruglansky & Mackie, 1990; Mackie, 1987; Martin, 1995, 1998; Sorrentino, King & Leo, 1980; Wolf, 1985). Both majorities and minorities seem to be able, in certain circumstances, to exert their influence through normative and informational influences. In fact, several models emphasize the similarities rather than the differences between minority and majority influences and reject the idea of two modes of processing group influence (cf. Latané & Wolf, 1981; Tanford & Penrod, 1984; Kruglansky & Mackie, 1990). Recently, Wood, Lundgren, Ouellete, Busceme and Blackstone (1994) used different methods to meta-analyze minority vs. majority studies, including direct and indirect methods of assessing private acceptance. The authors concluded that whereas according to some methods, majority and minority sources generated approximately the same amount of private indirect change, with other methods of analysis, minority influence generated substantially more private indirect change than majority influence, suggesting that the effect “proved fragile and easily muted by a variety of not well-understood moderating factors” (Wood et al, 1994, p.336). 50 Despite the above-mentioned theoretical problems, research on minority influence showed that the influence of groups is more complex than was traditionally depicted (cf. Levine, 1980; Wood, 1999). It challenged the traditional assumption that individuals are motivated to reduce uncertainty about reality at all costs even if pressing deviant members towards the group consensus (Festinger, 1950). Another phenomenon associated with the influence of groups is group polarization. Consistent with the distinction between two processes of group influence, the traditional explanations of group polarization emphasize either the normative or the informational aspects of conformity. Group Polarization Group polarization can be defined as the tendency for group discussion to strengthen the prevailing response tendency within a group. More precisely, if we compare individuals’ opinions after group discussion to their opinions before group discussion, “the average postgroup response will tend to be more extreme in the same direction as the average of the pregroup responses” (Myers & Lamm, 1976, p. 603). Polarization increases when there is a need for the group to reach a consensus, as compared to when there is no need for consensus, and to when people are merely exposed to others’ opinions. The magnitude of polarization is negligible when there is no initial dominant position, i.e., when the average opinion is neither negative nor positive. As such, the direction and magnitude of the shift may be predicted by the prediscussion average. In any case, the prediscussion and postdiscussion average opinions are generally highly correlated (cf. Myers, 1982). Value theory. One of the traditional explanations of group polarization, called value theory (cf. Turner, 1991), emphasizes the social value ascribed to one pole over the other, so that people polarize in the more socially desirable direction. For instance, Baron and Roper (1976), in an experiment using Sherif’s (1936) paradigm, found larger estimations when the experimenter previously defined large estimations as a sign of intelligence. The model stresses the social comparison nature of group polarization. Individuals adopt moderate initial opinions because they underestimate the positions of other members of the group. After finding that others hold more extreme positions than their own, they polarize in order to recover their personal self-esteem. 51 Persuasive arguments theory. Another model of group polarization, the persuasive arguments theory, explains the phenomenon through informational processes (e.g., Burnstein & Vinokur, 1973; 1975; 1977). The basic assumption is that when asked to express their opinion in some issue, individuals draw on pro and con arguments to form their opinion. Pro and con arguments derive from a cultural pool shared by most individuals. The more arguments favor one direction, the more opinions shift in that direction. Group discussion results in extreme opinions because individuals exchange their respective arguments and reinforce the validity of their initial opinions. For instance, Burnstein and Vinokur (1975) found that individuals polarized only when they had not been experimentally distracted and could think on others’ arguments. The authors argue that distraction prevented the possibility of considering the new arguments as a reinforcement of their initial position. The model also focus on the persuasive characteristics of the arguments - the more the arguments are original, consistent and valid, the larger the polarization they produce. Problems in the traditional approaches. Both models of group polarization have not obtained clear empirical support. The two alternative explanations of the phenomenon have invalidated each other because evidence for one model generally invalidates the other model. For instance, Burnstein and Vinokur (1973) obtained polarization in a situation where social comparison was prevented: participants polarized after arguing, by request of the experimenter, against their own initial positions. Also, merely reading the arguments substituted group discussion in polarizing participants’ initial positions. In turn, Blascovich, Ginsburg and Veach (1975) showed that the exchange of arguments is not necessary for group polarization. The authors asked participants to make 20 successive bets of blackjack, alone or in group. In the group condition, they could discuss or not discuss the bets after each hand. There was a shift to riskier bets in the group relative to the individual conditions, but the extent of the risk was not higher from the discussion to the non-discussion condition. Group discussion added nothing to the comparison between own and others bets. Theoretical critiques. The two models have also been contested in theoretical grounds. In what regards value theory, one problem is that polarization is considered an individual strategy of self-presentation and the interactive context of the group is not accounted for. As Whetherell (1987) contends, the ascription of one of the two opposed 52 positions is seen as issuing from abstract social values rather than an emergent norm of the group. In what regards persuasive arguments theory, the assumptions about arguments’ persuasiveness are conceptually biased. Persuasiveness is defined as an inherent quality of the arguments rather than an attribute obtained in the context of the group. As Whetherell (1987) remarks, “Properties of arguments such as validity, triviality, or even originality do not exist in the abstract […] A message from a group one supports and identifies with will be perceived quite differently than similar messages from a rival out-group to which one is strongly opposed” (p. 148). The explanations of group polarization are centered either in individual strategies to convey a positive self-image or in the informational influence of others in reinforcing a personal opinion. However, the group context in which polarization occurs is never accounted for in these explanations. Moreover, the evidence presented suggests the operation of a single process of influence encompassing social comparison and informational aspects. We shall now focus on commonalties of the above reviewed perspectives on group influence, and address some of the limitations of the traditional approach as a whole. The Traditional Model of Social Influence As seen above, the distinction of different forms of group influence has been a prevailing tendency in traditional research (e.g., Cialdini & Trost, 1998; Wood, 1999). This tendency is visible in the explanation of conformity, minority influence or group polarization phenomena. Turner (1991) dubbed this theoretical tradition, ‘the dualprocess model’2. The dual-process model. Deutsch and Gerard (1955) distinction between processes of conformity is paradigmatic of the dual-process model. In their model, 2 The dual-process model is congruent with other theories: Kelley’s (1952) distinction between the comparative and normative functions of reference groups; Thibaut and Strickland’s (1956) distinction between task set where others’ are ‘mediators of fact’ and group set where the individual is motivated to achieve and maintain membership; Jones and Gerard’s (1967) distinction between informational dependence leading to comparative appraisal (i.e., self-evaluation through comparison with others), and outcome dependence leading to reflected appraisal (i.e., inferences about others evaluate the self based on their reactions). 53 informational influence is considered true influence leading to private acceptance and internalization, and to lasting opinion and attitude change. This influence is informational in essence: messages are influential to the extent that they provide evidence on reality; the others are informative to the extent that they are reliable and credible sources; the underlying motive is the desire to be correct and attain subjective validity to beliefs. The process has the following causal structure: the ambiguous nature of the stimulus makes it difficult to do direct tests to physical reality leading to subjective uncertainty; uncertainty generates dependence on others to obtain valid information; informational dependence causes receptiveness to influence, people conform to other’s opinions considered as evidence about reality (e.g., Festinger, 1950; Jones & Gerard, 1967). The classic examples are Sherif’s (1936) studies. These studies represent the typical situation of informational influence leading to internalization (cf. Allen, 1965). Normative influence corresponds to the type of influence, in which individuals conform in appearance, but not necessarily in private, to others’ expectations. It is, specifically, a group process, of conformity to social pressures, based on others’ power to reward and punish. It is motivated by a desire of acceptance and approval. The individual is more concerned with the consequences of their actions in terms of how the group will react, than with the content of the action per se; it is ‘conformity’ in the sense of submission to the group pressure but, in essence, it is tactic and instrumental more than irrational and emotional (e.g., Deutsch & Gerard, 1955; Thibault & Strickland, 1956). Some specific ‘conformers’ may have an irrational and emotional urge to belong, but in general, group attraction is based on interdependence to achieve shared goals. Conformity is assumed functional for the group to achieve its goals. The causal structure initiates in others’ power to reward and punish (for instance, accepting or rejecting the individual in the group), which generates the need of social approval and the fear to be different; therefore, under conditions of surveillance, in which, one may be identified and being held liable for non-conformity, the individual tends to meet the expectations, producing conformity to the group norm; dependence on others, gives them the power to control the individual’s public behavior (e.g., Jones & Gerard, 1967; Kelman, 1958). The base of influence is not the need to validate beliefs, 54 but a social relation of power and the need to be socially accepted. The classic example is Asch’s (1951) conformity paradigm, in which participants give erroneous responses in order to agree with the group. Multi-process models. The dual-process model expands to three or more components. Kelman (1958) distinguishes three influence processes: compliance, which is based on others’ power to mediate rewards and costs; identification, which is based in attraction to others and may lead to acceptance of others’ values, provided the relationship is maintained; internalization, which corresponds to informational influence of the dual-process model. French and Raven (1959) distinguish five influence processes, reflecting different forms of dependence and power: power to reward (capacity to mediate the distribution of positive and negative outcomes), coercive power (capacity to apply punishments to those that not obey to requirements), reference power (influence based in identification, attraction or respect for the power holder), expert power (power deriving from the assumption that the power holder possesses superior capacity and knowledge) and legitimate power (authority deriving from the power holder legitimate right to demand obedience). The differences between dual-process and multi-process models relate to the degree of specification of the types of dependence and to the characteristics of the influence process rather than to the conception of influence. All models see influence as a process involving power and dependence and all models make distinctions between normative and informational processes. The general tendency is to distinguish between strictly social (normative) and cognitive (informational) motives. Objections to the Traditional Model of Influence Normative and informational processes are rarely separable in empirical results. Several variables, such as, group interdependence, cohesiveness, and unanimity, exert their effects in conformity by both processes. The first type of objections to the traditional model is metatheoretical. The distinction between ‘normative’ and ‘informational’ influences, is individualistic by nature because it tends to equate the former as group pressure, compliance or non-informative conformity (a passive and irrational process) and implies a continuous ‘conformity conflict’ between the (correct) individual and the (incorrect) group (cf. Turner, 1991). Informational influence, similar 55 to Festinger’s (1950) test of social reality, is seen as a secondary process, a surrogate to individual tests to reality. Contrarily to Sherif’s (1936) approach, influence is not seen as a group process, but the average of individual movements to each other, in the exchange of their private knowledge. In sum, individual perception is seen as primary, valid, and normal, whereas social influence is indirect, coercive and useful only when the former fails. Second, for Turner (1991), the model has failed in (1) dealing with minority influence, (2) providing a simple explanation to group polarization, and (3) accounting for some results of Asch’s (1951) paradigm of conformity. On the first aspect, the traditional model bases on the principle that influence was always exerted by the majority. For Moscovici & Faucheux (1972; cf. also, Moscovici, 1976), the traditional model suffers from a conformist bias adopting the majority point of view, and assuming that all conformity implies conformity or deviance. The function of social influence is to shape the individual to the social system, to its roles and norms. This model is unable to explain social change and the influence of marginal groups, subversive elites or oppressed minorities. As for the second aspect, the traditional conception of group norm as the average of the individual opinions is irreconcilable with the phenomenon of group polarization. Indeed, evidence consistently shows that the group norm is often more extreme than the average of individual opinions in the same direction (Myers & Lamm, 1976). To explain the phenomenon, traditional theories have argued either for the existence of socially valued positions (e.g., Baron & Roper, 1976), or for the existence of a cultural pool of pro and con arguments (e.g., Burnstein & Vinokur, 1977). However, none of the two models is able to account for the phenomenon in all its dimensions. Finally, concerning the third objection, Turner (1991) considers that Asch’s (1951) paradigm refutes the hypothesis that subjective uncertainty reflects the ambiguity of stimuli. As the author argues, uncertainty arises from disagreement, or anticipated disagreement, with similar others (cf. also Moscovici, 1976). Stimulus ambiguity produces subjective uncertainty, which leads to informational dependence from others, to mutual influence and to the formation of shared norms. Perceptual 56 ambiguity originates uncertainty to the extent that individuals expect similar others to perceive a stimulus as ambiguous as they do. The Social Identity Approach to Social Influence The concept of self-categorization allows a more simple and holistic explanation of social influence phenomena, such as conformity, minority influence and polarization. The Self-Categorization Theory approach to social influence phenomena is based on the idea that social interaction induces the formation of psychological groups. Psychological group formation. For Self-Categorization Theory, the shift to a categorized perception of the social environment corresponds to the formation of psychological groups (cf. Turner, 1984). In the process of group formation, individuals change from personal to group behavior and relate to others not as unique individuals but as in-group or out-group members. Psychological group formation is in the origin of social relations rather than its consequence. Groups may emerge from social interactions but, this is not a necessary condition and in some circumstances, it may even not be sufficient (cf. Turner, 1984). In a social categorization perspective, the group is defined as “a collection of people that share the same social identification or define themselves in terms of the same social category membership” (Turner, 1984, p. 530). This definition applies to small, interactive groups of friends or work teams, as well as to large social categories such as nationalities, races, gender and age groups. Referent Informational Influence Self-categorization is the basic process of conformity and other social influence phenomena (cf. Turner, 1987; Hogg & Turner, 1987). Perceiving the self and others as members of their respective categories implies thinking and action in accordance with the attributes associated with the category. In Self-Categorization Theory, categories’ attributes correspond to group norms in the same way as to perceive oneself as embodying the categories’ attributes is equivalent to perceive oneself as a normative member of the in-group (cf. Hogg & Turner, 1987). This process is dubbed Referent Informational Influence and is described in three stages (cf. also Turner, 1982): “First, individuals categorize and define themselves as members of a distinct social category or assign themselves a social identity; second, they form 57 or learn the stereotypic norms of that category. They ascertain that certain ways of behaving, perceiving and believing are criterial attributes of category membership; that certain appropriate, expected or desirable behaviors are used to define the category as distinct from other categories, and, finally, they assign these norms to themselves and thus their behavior becomes more normative (conformist) as their category membership becomes salient.” (Hogg & Turner, 1987, p. 149) The theory’s central assumption is that agreement with identical others in relevant aspects in a given situation creates subjective validity, “Subjective validity (Festinger, 1950; Kelley, 1967), one’s confidence in the objective validity of one’s opinions, attitudes, beliefs, etc. (also termed subjective certainty, competence, correctness, etc.) is a direct function of the extent to which similar people (in relevant aspects) in the same stimulus situation are perceived, expected, or believed to agree with one’s own response.” (Turner, 1987, p. 73) Conversely, subjective uncertainty “is a direct function of the extent to which similar others are not perceived, expected, or believed to respond similarly to oneself in the same stimulus situation” (Turner, 1987, p. 73). Through consensual validation, uncertainty about reality is reduced (cf. McGarty, Turner, Oakes & Haslam, 1993). Moreover, consensus tends to lead to external attributions and to the reification of beliefs (Kelley, 1967; cf. also, Asch, 1952; Moscovici, 1976; Sherif, 1936). As Hogg and Turner (1987) put it, “social consensus or agreement leads to the external attribution of the shared response – that is, it is perceived to reflect some external public invariance in the object of perception and is thus perceived to be ‘objectively’ valid, correct, appropriate, required, demanded, etc.” (p. 149). Conformity as Assimilation to In-Group Representations Referent Informational Influence refers to conformity to a representation of the in-group rather than to actual behavior of in-group members. “While overt behavior provides cues concerning the normative tendency of a group or social category, the self-categorization process ensures that what 64 of the phenomenon. The theories emphasize either social comparison or informational motives. Self-Categorization Theory proposes an explanation that is based on a single process of influence. Specifically, the theory conceives group polarization as conformity to prototypical group positions (cf. e.g., Abrams, Wetherell, Cochrane, Hogg & Turner, 1990; McGarty, Turner, Hogg, David & Whetherell, 1992; Whetherell, 1987). Polarization represents assimilation to in-group positions and differentiation from out-group positions: the convergence to the member(s) that hold(s) positions simultaneously more similar to those of other in-group members and more distant from out-group positions. It is the need for distinctiveness, which leads members to shift to an extreme position in the direction opposed to that of the out-group. In this perspective, there is a cognitive basis for the idea that extreme positions are more attractive than moderate positions, as argued by the theory of value (e.g., Lamm & Myers, 1978; Baron & Roper, 1976). Extreme positions are more attractive because they contribute to the differentiation of the in-group from the out-group. The evidence obtained for persuasive arguments theory is also accounted for (e.g., Burnstein & Vinokur, 1973; 1975; 1977): prototypical members are more able than nonprototypical members are to persuade other members to change their opinions (cf. van Knippenberg, Lossie & Wilke, 1994; van Knippenberg & Wilke, 1992). Prototypical positions are thus both more attractive and influential leading other members to extremize their opinions. Conclusion Traditional approaches established theoretical distinctions in group influence based on the effects of public versus private contexts, or task versus group orientations. Often, group norms are depicted as arbitrary, subjective and uninformed opinions deviating from objective standards and conformity to the group is associated with public compliance rather than with internalization. When group norms and consensus are viewed as providing subjective validity to members’ opinions and attitudes, the group is viewed as an external device to which individuals resort to when personal testing is difficult. The nature of the constructs that are manipulated to produce informational influence, such as the recipient’s uncertainty, the source’s credibility or the objectivity of the task are rarely questioned. 65 Traditional explanations of group polarization are based on distinct influence processes: whereas one model favors social comparison processes, the other model favors informational processes. However, none model accounts for the phenomenon in the whole. Also, the minority influence model focused on the distinction between group influences. Although opposing to some traditional assumptions about conformity, the model stresses the existence of two forms of conformity: majority influence is associated with compliance and minority influence with internalization. The model also claims that minorities are more able than majorities to exert indirect influence, that is, the unconscious conversion to their positions. This claim has not obtained consistent empirical support. Nevertheless, the minority influence phenomenon has revealed some of the inconsistencies of the dominant model namely its assumption about the function of group influence in the preservation of social stability. Self-Categorization Theory provides an integrated account of social influence phenomena. Central to the theory is the concept of social categorization. Social categorization is a pervasive process that has effects on social perception and selfconception. By accentuating the similarities between the members of a category and the differences between categories, individuals form a stereotype of each category in which the characteristics common to the members of the category and differences to members of other categories are accentuated. The assignment of the self to one of those categories leads to a change in the self-concept so that individuals perceive themselves not as unique individuals but as undifferentiated members of the category. They conform to the norms of the in-group, the attributes of the group representation, resulting in stereotyped judgments, opinions, and behavior. In this perspective, group influence is both normative and informative. It does not depend on other members’ surveillance because the in-group opinion is the reference for private opinions. Individuals conform to the group norm because it conveys the more adequate and objective account of reality. As such, in-group influence is a single process, which is responsible for both majority and minority, direct or indirect, influence. The extent into which a group is able to exert influence on members, directly or indirectly, depends more on how much it represents for members’ identity than on its relative size. 66 The social identity approach to social influence has further support in research on group polarization. The shift to polarized positions subsequent to group discussion is explainable by the higher attractiveness and persuasion of prototypical members, because these members contribute to the distinctiveness of the group as regards the outgroup. 67 CHAPTER 3 SOCIAL IDENTITY, IMPLICIT PROCESSING AND GROUP INFLUENCE Recent research has focused on automatic processes3 associated with social identity. Automatic processes “develop out of frequent and consistent experience in an environmental domain” (Wegner & Bargh, 1998, p. 463). Unlike controlled processes, automatic processes have the advantage of not requiring attention and not being hindered by the overload of the working memory (cf. Kahneman, 1973; Posner & Snyder, 1975). An automatic response in a usual situation is the prevailing response when conscious processes do not intervene to modify the response (cf. Schneider & Shiffrin, 1977; Shiffrin & Schneider, 1977). Some automatic processes are not even controllable, that is, once there are enough cues to initiate them, they run until completion (e.g., Bargh, 1989; 1994; 1996). Thus, only vital and highly adaptive processes become automatic (e.g., Bargh, 1989; 1994; 1996; Bargh & Ferguson, 2000; Bruner, 1992; Vigotsky, 1925). The findings showing that social identity processes operate at an implicit level indicate their crucial role in the determination of individual behavior. Implicit Social Identity Processes Research on implicit social identity processes has focused in two major issues: the association between the in-group representation and the self-representation; and the automaticity of the in-group bias. The Association between Self and In-Group Representations The social identity approach postulates that the collective self is as much part of the self-concept as the personal self (e.g., Simon, 1997; Turner et al, 1987; 1994). As such, there should be strong associations between the attributes of an in-group stereotype and the self-concept. For instance, Smith and Henry (1996) first asked participants to rate 90 traits as a function of their description of themselves, and, 3 As pointed out by Greenwald and Banaji (1995), the pair of terms implicit-explicit is used in some literatures, whereas the pairs automatic-controlled, or even unconscious-conscious and indirect-direct, are used in other literatures. Like these authors, we assume that the first terms and the second terms of the pairs, respectively, are equivalent. Therefore, the terms implicit and automatic, as well as the terms explicit and controlled will be used interchangeably. 68 depending on participants’ significant group membership, of Liberal arts and engineering majors’ students, or Greeks fraternity/sorority members. Then they were asked to respond to the same test in a reaction time test. They should press the button ‘yes’ or ‘no’ when the trait word appeared in the computer screen, depending on whether the trait described themselves or not. The authors found that participants were faster in pressing the ‘yes’ button to indicate that trait-words were self-descriptive, and also made less errors, when traits matched the stereotype of the in-group (as indicated by their initial ratings in the conventional measure) than when they did not matched this stereotype. Moreover, the effect is not found when the similarity/dissimilarity of self to the out-group stereotype is accounted for. The fact that traits on which the self and in-group are similar are readily accessible for self-description confirms the Self-Categorization assumption that social identities are important components of the self-concept (cf. also, Smith, Coats & Walling, 1999). In the previous experiment, it was assumed that the stereotypic self-descriptors were made salient by the initial task. To confirm the idea that the shifts on selfperception correspond to shifts on the social context (e.g., Turner et al, 1987; 1994), Onorato and Turner (2004) used a modified version of Markus’ (1977). In Markus’ (1977) original procedure, participants were classified as a function of their responses to a questionnaire in which they had to rate themselves in three scales referring to trait description items related to the dependence vs. independence dimension (leader / follower, individualist / conformist, independent /dependent). The participants that rated themselves in the independent pole of the scale were classified as Schematics Independents; participants that rated themselves in the dependent pole were classified as Schematics Dependents; and participants that rated themselves in the middle of the scale were classified as Aschematic. In a second session, participants were presented with several trait-words (some of which related with the independence / dependence dimension) on a screen and had to push the yes or no button depending on whether they found that the trait presented was or not self-descriptor. The results showed that dependent participants responded faster to say that dependent trait-words were selfdescriptors than independent trait-words, and the opposite for the Independent participants. 69 To prove that, consistent with self-categorization principles, in some situations social identity has a more powerful effect on information processing than personal selfaspects, Onorato and Turner (2004) formed groups of participants of the same sex and same type – Independent, Dependent, or Aschematic. Participants were to reach a consensus on the degree to which some independent / dependent traits were more typical of men or women thus priming participants with a situation that made salient the stereotypical attributes of their own and the other gender group. They then completed a modified version of the Markus’ (1977) latency task. Specifically, participants were asked to respond whether the trait (independent, dependent, masculine or feminine) appearing in the screen was descriptor of him/her as compared to the other gender group (‘Us’ or ‘Them’). The results showed that irrespectively of their personal schema type (Independent, Dependent or Aschematic), both male and female participants were faster to respond stereotypically than non-stereotypically confirming the idea that social group memberships are as important to self-definition as personal traits. Implicit In-Group Bias In the origin of in-group bias is the need for a positive social identity (e.g., Tajfel, 1978). The recurrence of in-group biases is reflected in the stronger association of positive attributes to in-group and stronger association of negative attributes to outgroups. For instance, Perdue, Dovidio, Gurtman and Tyler (1990) found that nonsense syllables unobtrusively paired with in-group designating pronouns, such as we or us, were rated as more pleasant than syllables paired with out-group designators, such as, they or them. Moreover, the subliminal exposition to in-group designators decreased the reaction time in the classification of positive adjectives whereas the exposition to outgroup designators facilitated the classification of negative adjectives. Finally, the subliminal exposition to in-group designators facilitated the classification of positive person descriptors as compared to a neutral subliminal prime. The results suggest that the use of in-group or out-group pronoun designators in the speech prompts automatic positive or negative attitudes towards in-group and out-group targets, respectively. 70 Other research (Otten & Wentura, 1999) showed that group membership even when defined through minimal conditions in the laboratory automatically entails positive feelings towards the in-group. Otten and Wentura (1999) found that after being categorized through a typical minimal group procedure, participants’ classification of adjectives was affected by prior subliminal exposure to in-group and out-group labels. They made more errors in classifying negative adjectives when primed with the ingroup label and in classifying positive adjectives when primed with the out-group label than in the reversed conditions. The results suggest that in-group and its members are automatically associated to positive and pleasant feelings and/or the out-group and its members are associated with negative unpleasant feelings. Implicit Out-Group Derogation Consistent with the above-described results, Cameira, Serôdio, Pinto and Marques (2002) obtained out-group derogation as an implicit response to a salient intergroup situation. In their experiment, the authors manipulated the group membership of the experimenter as regard the participants and checked the implicit influence of this variable in subsequent choices. Specifically, small groups of Psychology or Social Work students interacted with an experimenter that ostensibly presented herself as Psychologist or Social Worker conducting a survey on Community Intervention. After responding to a questionnaire aimed to increase the salience of their group membership, participants were asked to choose between two teams to work with in a second phase of the alleged study. The teams were composed by three members whose faces, depicted in the questionnaire sheet, and were either similar or different from the first experimenters’ face (similarity of the team members’ face to the experimenter’s face had been tested earlier). After making their choice, participants were asked to mark among several hypotheses the reasons of their choices. The predictions were that participants would choose more the similar than the different team when the experimenter was an in-group member. The results did not meet the predictions but revealed that participants chose the different team more when the experimenter was an out-group member than when the experimenter was an ingroup member. This effect increased with in-group identification. The authors interpreted these results in terms of group contexts: whereas the out-group experimenter 71 condition portrayed an intergroup context, the in-group experimenter condition portrayed an intra-group context. Hence, judgments and evaluations based on group membership were more likely in the former than in the latter context. Participants attributed to chance or to the attractiveness of the chosen team the reasons of their preference. The results suggest that the implicitly learned association biased the choices of high-identified participants in the intergroup context. Implicit Prejudice Research on implicit prejudice provides innumerous examples of automatic ingroup bias or out-group derogation. Most important, this research has shown that implicit (non-controlled) attitudes towards discriminated groups are often not correlated with explicit (controlled) attitudes. This absence of correlation between the two variables has been interpreted as the result of a conscious need to show socially desirable non-prejudiced responses affecting solely the controllable measure (e.g., Dovidio, Kawakami, Johnson, Johnson & Howard, 1997; Fazio, Jackson, Dunton & Williams, 1995; Greenwald & Banaji, 1995). For instance, Devine (1989) compared the automatic and controlled attitudes of White individuals towards Blacks. First, the author showed that both prejudiced and non-prejudiced individuals (as determined by their responses to the Modern Racism Scale) were knowledgeable of the negative attributes commonly ascribed to Black individuals. Subsequently, all participants rated a story character in several aggressiveness-related traits after being subliminally primed with words stereotypical of the Black group and with neutral words. The results showed that both prejudiced and non-prejudiced participants rated the story character as more aggressive when primed with Black-related words than when primed with neutral words. Finally, participants were asked to list their thoughts about the Blacks racial group under anonymous conditions. Results show that prejudiced participants had a marked tendency to generate stereotype-congruent responses whereas non-prejudiced individuals showed the opposite response pattern, that is, they generate beliefs that contradicted the stereotype of Blacks. In addition, most prejudiced participants referred to the aggressiveness or violence topic whereas only a minority of non-prejudiced participants did. 72 The author concluded that both prejudiced and non-prejudiced individuals are aware of the stereotype of Blacks and are automatically affected by it. However, whereas the prejudiced individuals have an explicit attitude congruent with the stereotype, non-prejudiced individuals, consciously oppose to this implicit influence, responding counter-stereotypically. Other research suggests that implicit and explicit attitudes may also be correlated when there are no normative reasons to conceal prejudice. Greenwald, McGhee and Schwartz (1998) designed the Implicit Association Test drawing on the automatic activation of the attitude towards a social group upon the presentation of a stimulus item associated with the group. Typically, the procedure has several phases that include the classification, in their respective labels, of clearly positive or negative adjectives as “good” or “unpleasant” and of words related with two opposing social groups. Greenwald and colleagues (1998; Experiment 2) compared the IAT results of Korean American and Japanese American participants (ethnic groups with a history of antagonism). As expected, the Japanese participants were slower to react when their ethnic group label was associated with the “unpleasant” label (and the Korean with the “pleasant” label) than in the reversed situation. The opposed pattern of results was obtained with the Korean participants. The participants also answered to a series of questions measuring their involvement in the Japanese or Korean social networks, respectively. The authors found that the results of the explicit measure were positively correlated with the IAT results. A replica of the above study Greenwald and colleagues (1998; Experiment 3) using the Blacks vs. White labels and related stimulus words (typical Black and White names), revealed a stronger association, among White participants, between the Blacks group and its negative evaluation and lower correlation with the explicit measure than in the Korean and Japanese groups. As the authors conclude, these results indicate the existence of more concealed hostility between Whites and Blacks as compared with the concealed hostility between Koreans and Japanese, which is consistent with existence of social norms aimed to protect Blacks from Whites’ prejudice, and that these norms have some acceptance among Whites. 73 In sum, the above-described findings indicate that the activation of positive attitudes towards in-group members and/or negative attitudes towards out-group members tend to be automatic processes. In our empirical research, we attempted to extend the application of non-controlled methodologies to other social identity processes. Specifically, in the first four-experiment study, we examined conformity to the implicit group norms. 80 Figure 1. Examples of the two types of pictures used in the experiments. The top six pictures are examples of the vertical type and bottom six pictures are examples of the horizontal type. 81 Tests of awareness. After observing the sample of pictures, participants were told that the experimenters wished to known whether they were able to distinguish between approved and rejected pictures. This corresponded to the recognition test of awareness. Participants were asked to observe twenty unlabelled pictures that allegedly had been extracted from the lot of approved and rejected pictures judged by the jury. After observing each picture for five seconds, participants reported whether it had been approved or rejected by the jury. The experimenter controlled the presentation of each picture after ensuring that the participants have finished with the previous one, although mentioning the need for “a somewhat fast rhythm”. The number of pictures correctly classified represented the extent to which the respondent was aware of the series’ distinctive details. Twenty, or close to twenty, correctly classified pictures meant that the respondent had a correct criterion for the classification and was aware of the details. None or close to none correctly classified pictures meant that the respondent based on an incorrect criterion. Ten or close to ten correctly classified pictures meant that the respondent did not used any criterion and answered by chance. After this task, participants answered to the question “Have you found any distinction between the approved and the rejected pictures?” and, in the affirmative case, they were asked to describe the distinction(s). This task corresponded to the ‘written report’ test of awareness. Judgment task. Subsequently, participants were told that, in the present research, the experimenters wanted to know “the opinion of common people about this particular kind of pictures”. Participants were then asked to judge twenty pictures that allegedly had been produced recently and had never been judged. There were 10 vertical and 10 horizontal pictures and their order of presentation was randomized. Participants observed each picture for 5 seconds and then reported their judgment in a 7-point scale (1= Do not like it; 7= Like it very much). After this task, participants were debriefed and thanked for their participation. In the debriefing, two participants verbally reported the distinction between the approved and rejected pictures. The experimenter made sure that they had reported the distinction in the questionnaire in order to discard their data from the analysis. 82 Results Recognition and recall. The data of two participants that found the graphic details of the pictures were removed from the analysis. We computed a recognition index by summing the number of pictures correctly recognized (minimum=0; maximum = 20). The mean recognition rate is not different from chance, M = 9.41, SD = 2.89, t(23) < 1, suggesting that participants were unaware of the distinction between the two series of pictures. No. Approved Pictures Rejected Pictures Balance 1 1 More undefined and irregular More geometrical 2 More creative - 6 Darker Lighter Balance 2 15 More detailed Less denser, less esthetically organized 17 Lighter Darker 18 More interesting, with abstract background They all have an F; meaningless 19 Darker More blank spaces, some represent letters 20 More variety of color and curves More black or white color 21 Less larger with more variations of the pattern Larger with a more uniform pattern 22 Less organized pattern More organized patterns 23 Less “noise” density Opposed to the approved ones 24 Better esthetics and straight drawings No esthetics, no straight drawings Table 1. Reports on the distinction between the two series of pictures grouped by balance condition (Experiment 1). The responses in bold characters refer pictures’ details resembling the actual distinction. 83 The analysis of the 12 reports of the participants that reported having found a distinction between approved pictures and rejected pictures revealed that one participant mentioned a distinction close to the actual one (cf. Table 1 – participant # 18). However, the fact that the participant only classified correctly 12 pictures suggests that the distinction was not used systematically in the subsequent tasks. The three participants reporting differences in the general tonality of the two types of pictures had opposite opinions indicating that this difference was not objective. Indeed, the differences reported by the participants match previous findings showing that, when believing that there is an hidden structure in the stimuli, participants tend to formulate rules based on their most obvious features (e.g., Reber, 1976). Conformity. We averaged the ratings of approved and rejected pictures into two new variables of preference for approved pictures and preference for rejected pictures, respectively (Norm). In the context of the task, we assumed that conformity to the credible source’s opinion is represented by the preference of accepted pictures as regard rejected pictures. Therefore, we conducted a Norm x Balance ANOVA, in which Norm was a within-participants factor. The analysis revealed a significant main effect of Norm, F(1, 22) = 4.61, p = .043, η2 = .17; all others effects, highest, F(1, 22) = 2.30, ns. As predicted, participants showed higher preference for the approved pictures than for the rejected pictures, M = 4.03, SD = .67 and M = 3.78, SD = .595. Conformity and recognition. To check on the effect of awareness in the participants’ conformity we divided participants by median-split of the recognition index and created a new two-level variable. ‘Good recognizers’ recognized an average of 12.86 pictures and ‘bad recognizers’, an average of 8.00 pictures. We then conducted an ANOVA with the new variable as a between-participants factor and Norm as withinparticipants factor. The interaction of the two factors was non-significant indicating that the judgments of ‘good recognizers’ did not differed from those of ‘bad recognizers’, F(1, 22) < 1. 5 We conducted the same analysis removing the data of participant #18 that reported a distinction similar to the actual one. Results still show a preference for approved pictures, M = 4.00 and M = 3.74, F(1, 22) = 4.46, p = .046, η2 = .17. 84 Discussion Confirming the hypothesis, the results suggest that individuals tend to regulate their judgments by those of credible sources. After being exposed to pictures that had been approved or rejected by a jury of experts, participants preferred pictures that were similar to approved ones. Consistent with long lasting evidence showing that people follow the opinion of credible sources our results indicate that this acceptance does not depend on the capacity to articulate that opinion. It must be stressed that the context created in the experiment favored an automatic reliance on the jury’s opinion. Participants were asked to make judgments on several very similar pictures, thus, involving the implicit demand to differentiate among the pictures in terms of preference. Although, esthetical judgments are generally perceived as subjective, the previous mention of a jury, and of approved and rejected pictures, may have led to the inference that that was an objective task. Judgments perceived as objective prompts a search for external validation rather than reliance on internal standards (e.g., Insko, Drenan, Solomon, Smith & Wade, 1983) especially when external sources are considered competent (e.g., Aronson, Turner & Carlsmith, 1963; Bochner & Insko, 1966; Di Vesta, 1959; Samelson, 1957). Moreover, the similitude of the pictures is likely to induce the perception of lack of agreement in the judgments and consequently more uncertainty (cf., e.g., Hogg & Turner, 1987; Turner, 1991). Uncertainty also induces the search for external information to anchor one’s judgments (cf., e.g., Allen, 1965; Coleman, Blake & Mouton, 1958; Croner & Willis, 1961; Crutchfield, 1955; Fisher & Lubin, 1958; Kelley & Lamb, 1957; Sherif, 1936). Other results show that the knowledge of the jury’s criterion, even if implicit, did not contribute for the judgments. The fact that virtually all participants were unable to articulate the differences between the two kinds of pictures, and that their respective level of recognition did not affected their judgments suggests that these latter were not consciously processed. The traditional assumption that informational influence is accepted in a relatively automatic way is thus confirmed. 85 Experiment 2 Conformity to Implicit Norms as a Function of Source’s Credibility and Group Membership The results of the previous experiment suggest that individuals automatically accept informational influences. In the present experiment, we test the general hypothesis that automatic conformity applies also to normative influences. In this experiment, the credible source is either an in-group or an out-group. Our first prediction is that implicit conformity to the credible source occurs only when the source is an in-group. However, in the present experimental context, credibility is also a relevant attribute of the source given the task participants are to perform. Our second prediction is thus that non-credible group members will not conform to any of the groups. We assume that, in a situation in which credibility is relevant, non-conformity to a credible out-group source is the result of normative influences. We took advantage of the fact that students enrolled in the Psychology course issue from either Arts or Sciences high-school areas and that this fact corresponds to a significant social categorization among sophomores. Compatible with the nature of the experimental task, esthetic aptitude is a stereotypical attribute that differentiates the two groups: Arts students are normally considered more knowledgeable than Sciences students in esthetics matters. To confirm this idea, using an unobtrusive method, we conducted two pilot studies in the same population of the main experiment. The first pilot study was aimed to establish which personality traits were associated with credible appreciations of the abstract pictures used in the experiment; the second pilot study was aimed to examine the association of those personality traits to Arts and Sciences students and, indirectly, the ascription of credibility in the task to both groups. Pilot Study 1 Forty Psychology and Educational Sciences undergraduate students (28 female and 12 male; ages ranging from 18 to 24 years old), after observing a reproduction of two of the pictures we used in our experiments (cf. Figure 1), were asked to write, at least, five traits of “the people that could have a credible opinion about this kind of 86 pictures”. Respondents produced 73 different traits, 32 of which with more than one mention. Three categories were formed by aggregating traits with a similar semantic meaning. ‘Creative’ (frequency = 26) was aggregated with ‘Creative mind’ (frequency = 2); ‘Artistic’ (frequency = 10) was aggregated with ‘Art knowledgeable’ (frequency = 5) and with ‘Esthetic’ (frequency = 1); ‘Sensitive’ (frequency = 8) was aggregated with ‘Subjective’ (frequency = 8). The categories received the name of the most frequent trait. The analysis of the frequencies of the categories and remaining traits, revealed an accentuated decrease between the fifth more mentioned trait, Abstract (frequency = 15) and the sixth more mentioned trait (frequency = 8). We thus decided to select the following traits: creative (28), imaginative (19), artistic (16), sensitive (16), and abstract (15). The five selected traits (named ‘Relevant traits’) were embedded in a longer list of 14 traits including five traits with an opposed meaning (named ‘Opposed traits’), ‘realistic’, ‘practical’, ‘objective’, ‘rational’, and ‘concrete’, and four filler-traits, ‘fond of nature’, ‘athletic’, ‘lazy’ and ‘active’. The 14 trait-list was pre-tested to ensure that Relevant and Opposed traits had identical values for Psychology undergraduates. We then proceed with the next pilot study. Pilot Study 2 Fifty-two undergraduate students attending an introductory course to Psychology (44 female and 8 male; age range 18-28 years old, M = 19.02, SD = 1.73) and ensuing either from the high-school Arts area (23) or from the high-school Sciences area (29), were asked to answer a questionnaire about the characteristics of Psychology sophomores. Half of the respondents were asked to first think about students ensuing from the high-school Arts area and then to mark into what extent they possessed, in their opinion, each of the 14 traits, in 7-point scales (1=nothing at all; 7=very much). Then they were asked to think about students ensuing from the high-school Sciences area and to rate those students on the same 14 traits. For the other half of the respondents the target groups order was reversed: they were asked to rate first Science area students and then Arts area students. 87 Results We first averaged the ratings of the Relevant traits relative to Sciences students (Cronbach’s α = .75) and to Arts students (Cronbach’s α = .75) into two new variables, Sciences Relevant traits and Arts Relevant traits. Then, we averaged the ratings of the opposed traits relative to Sciences students (Cronbach’s α = .80) and to Arts students (Cronbach’s α = .74) into two new variables, Sciences Opposed traits and Arts Opposed traits. We then conducted an Area x Order x Target Group (Arts vs. Sciences) x Traits (Relevant vs. Opposed) ANOVA, in which Target Group and Traits were withinparticipants factors. The Traits x Target Group interaction revealed significant, F(1, 48) = 188.52, p < .001, η2 = .80. Indeed, Arts students were rated higher in relevant traits than Sciences students, M = 5.11, SD = .77 and M = 3.31, SD = .80, respectively, F(1, 51) = 112.34, p < .001, η2 = .69, and were rated lower in opposed traits, M = 3.56, SD = .76 and M = 5.65, SD = .77, F(1, 51) = 151.12, p < .001, η2 = .75. The analysis of differences across Traits also revealed significant: Arts students were rated higher in relevant than in opposed traits, F(1, 51) = 104.61, p < .001, η2 = .67, and Sciences students were rated higher in opposed than in Relevant traits, F(1, 51) = 182.44, p < .001, η2 = .78. There was no interaction with Area, F(1, 50) = 1.61, ns, η2 = .007, showing that the respondents rated the target groups similarly independently of their respective membership group (cf. Figure 2)6. In the whole, the results supported our initial idea; we may assume thus that, in the main experiment, Arts students are perceived as a more credible source than are Sciences students. 6 The analysis also revealed a significant main effect of Traits, F(1, 50) = 26.68, p < .001, η2 = .35, showing that the respondents, evaluated both groups as possessing the opposed traits more than the relevant traits, M = 4.21, SD = .48 and M = 4.60, SD = .46, respectively. There were no effects of Target Group or Area, F(1, 50) = 1.62, ns, F(1, 50) = 1.31, ns, respectively. The main effect of Traits was qualified by an interaction with Area, F(1, 50) = 4.22, p = .05, η2 = .08, showing that although Sciences students judged both target groups higher in the opposed items than in the relevant items, M = 4.23, SD = .41 and M = 4.48, respectively,SD = .45, t (28) = 2.61, p = .01, Arts Students judgments were more extremized, M = 4.19, SD = .59 and M = 4.76, SD =.42, respectively, , t (22) = 4.32, p < .001. Finally, an interaction of Area and Target Group, F(1, 50) = 25.71, p < .001, η2 = .34, revealed that Sciences students rated their own group higher than the out-group, M = 4.60, SD = .45 and M = 4.11, SD = .45, respectively, t (28) = 4.63, p < .001, and the same occurred among Arts students, M = 4.33, SD = .48 and M = 4.62, SD = .50, respectively, t (22) = 2.61, p = .02. 88 In-Group Out-Group Sciences Participants Figure 2. In-group and out-group evaluations in relevant and opposed traits as a function of respondents’ high-school area (Pilot study 2). 1 2 3 4 5 6 7 Evaluation In-Group Out-Group Arts Participants 1 2 3 4 5 6 7 Evaluation Relevant Traits Opposed Traits 89 Main Experiment Method Participants and design. Thirty-five students of a Psychology introductory course (29 female and 6 male; ages ranging from 17 to 20 years old, M = 18.30, SD = .80) volunteered to participate. Eighteen ensued from the high-school Arts area, and the remaining 17 participants ensued from the high-school Sciences area. We employed a 2(Group: Sciences vs. Arts) x 2(Norm: In-group vs. Out-group) mixed design in which Norm was a within-subject factor. Procedure. The procedure was similar to the procedure of the previous experiment with the following exceptions. As they arrived to the laboratory, participants were asked their high-school area and then indicated to sit at one table of the two existing in the room. Three or four Sciences students sat at one table, and three or four Arts students sat at the other table. This procedure was aimed to make their group membership salient. Each group sat around a large table facing a computer screen where the pictures were displayed. The two groups were sitting so that they could see only the screen on their table. The experiment was presented as part of a study on the esthetical preferences of students ensuing from the Sciences and the Arts school areas. In the learning phase, pictures were labeled Sciences or Arts depending on whether they allegedly were preferred by large samples of students of each respective area. The distractive piece of music, used in the previous experiment, was played in the background with the excuse that that was the normal procedure of the experiment. Results Recognition and recall. As in the previous experiments, we computed a recognition index by summing the pictures correctly recognized (minimum = 0; maximum = 20). The one-way ANOVA of Group on the recognition index was nonsignificant, F(1, 34) < 1, and the mean recognition was not different from chance, M = 10.07, SD = .65, t(34) < 1. In the written report, fifteen participants declared having found a distinction between the two types of pictures, but none reported the actual graphic details. Furthermore, one participant reported a distinction, vertical as opposed to horizontal 96 into two groups. Participants were told that: “Recent research has found the existence of an important dimension of personality that can be detected by a simple perceptive test. In the present session, you will complete this test. We will not say more about this personality dimension now except that it is the origin of two very distinct perceptive types. Your results will be communicated to you in the next session as well as an explanation of this personality dimension.” The bogus test that participants were asked to complete, consisted in the observation of briefly presented ambiguous pictures, after which participants had to describe the first figure they saw in each picture 7. One week later, participants were randomly distributed by three sessions of 11- 13 participants each. As they arrived, they were asked to sit facing one of two screens existent in the room. The two groups were disposed in the room back to back, so that they could not see each other. Participants were then given the feedback of their results. Half of the participants were randomly assigned to the Proficient type and the other half was assigned to the Progressive type. Participants also read the characteristics of both perceptive types: Proficient were described as “creative, imaginative, artistic, sensitive and abstract” and Progressive were described as “realistic, objective, practical, concrete, and rational”. We used the traits obtained in Pilot Study 2 to ensure that the description of one of the artificial groups matched the characteristics valued in the esthetic judgment of the pictures and that the description of the other group matched the characteristics with the opposed meaning. To reinforce the credibility of the Proficient type in the task, participants were also told that, “The two perceptive types differ in several aspects. Differences in various aptitudes are currently being investigated. The present investigation is aimed to study the esthetic taste of Proficient and Progressive. Previous results have shown that the esthetic taste of the Proficient persons is, in general, of high quality, similar to that of experts in fine art, whereas the taste of Progressive persons is, in general, of low esthetic quality.” Participants observed a graph of bars depicted in their booklets reinforcing the idea of Proficient’s superior esthetic taste as opposed to Progressive’s inferior esthetic taste. 7 The procedure to create artificial groups was based on Marques and colleagues (2001a; Experiments 2 - 3). 97 The subsequent procedure was identical to the one of the previous experiment with the following exception: To ensure that participants became aware of the group norms, when introducing the learning task, the experimenter noticed that: “Interestingly, the pictures preferred by Proficients present always this detail (the graphic detail was pointed out in a sample picture displayed on the screen), and the pictures preferred by Progressives present always this other detail (the other graphic detail was pointed out in another sample picture displayed on the screen).” Finally, participants were asked to register their group membership and to answer a brief questionnaire including the following questions: “To what extent do you think that Proficients are competent in the just finished task?” and “To what extent do you think that Progressives are competent in the just finished task?”. Participants also answered the following six questions aimed to measure their identification with ingroup and out-group. The questions were extracted from recent research on social identity (e.g., Hogg & Hains, 1996; Marques, Abrams & Serodio, 2001), and included, “To what extent do (would) you like to have your (the other) perceptive type?”, “To what extent do you feel similar to the characteristics of your (the other) perceptive type?”, “To what extent do you define yourself as Proficient / Progressive?” All items were answered in 7-point scales with labeled ends (1 = Absolutely nothing; 7 = Very much). In the end of the session, participants were thoroughly debriefed, ensuring that every participant clearly understood that the perceptive test and perceptive types were false. Results Group identification. All participants were aware of their group membership and of the norms of the two groups. The items referring to in-group identification were averaged into a single index of in-group identification (Cronbach’s alpha = .80), M = 6.46, SD = 1.31. The items related with identification to the out-group could not be averaged (Cronbach’s alpha = .57). Therefore, we retained the item “To what extent do you feel similar to the characteristics of the other perceptive type?” because it had the highest inter-item correlation, r = .50 (all the others, highest, r = .37). We subsequently conducted a Group (Competent vs. Non-competent) X Identification (In-group vs. Outgroup) ANOVA with the latter as within-participants factor. The results show that all 98 participants identified more with the in-group than with the out-group, M = 6.46, SD = 1.31 and M = 3.03, SD = 1.30, respectively, F(1, 35) = 167.48, p < .001, η2 = .83. The significant interaction with Group, F(1, 35) = 28.90, p < .001, η2 = .45, indicates that members of the competent group identified more with the in-group relatively to the outgroup, respectively, M = 7.25, SD = .97 and M = 2.55, SD = 1.00, respectively, F(1, 35) = 182.57, p < .001, than members of the non-competent group, respectively, M = 5.53, SD = 1.00 and M = 3.59, SD = 1.42, respectively, F(1, 35) = 26.47, p < .001. Perceived competence of the groups. To examine the differences in perceived competence of the two groups, we conducted a Group (Competent vs. Non-competent) X Perceived Competence (In-group vs. Out-group) ANOVA, with the latter as withinparticipants factor. The results showed a main effect of Perceived competence, M = 5.30, SD = 1.15 and M = 4.46, SD = 1.63, respectively, F(1, 35) = 6.08, p = .02, η2 = .15. The Group x Perceived competence interaction, F(1, 35) = 21.73, p < .001, η2 = 38, revealed that, whereas members of the competent group perceived the in-group as more competent than the out-group, respectively, M = 5.70, SD = .86 and M = 3.60, SD = .43, respectively, F(1, 35) = 27.64, p < .001, members of the non-competent group did not perceive one group more competent than the other, M = 4.82, SD = 1.29 and M = 5.47, SD = 1.23, respectively, F(1, 35) = 2.23, ns. Conformity. We averaged the ratings of the pictures into two new variables of conformity to in-group norm and conformity to out-group norm (Norm). As in the previous experiments, we began by conducting a Group x Norm ANOVA, with Norm as a within-participants factor. The analysis revealed a main effect of Norm, F(1, 35) = 18.89, p < .001, η2 = .35, qualified by an interaction with Group, F(1, 35) = 18.90, p < .001, η2 = .35. In general, participants preferred in-group pictures to out-group pictures, M = 4.71, SD = .80 and M = 3.87, SD = 1.04, respectively. As expected, the Group x Norm interaction indicates that in-group conformity was significant among the participants who had been categorized in the competent group, M = 4.99, SD = .72 and M = 3.43, SD = .86, respectively, F(1, 35) = 41.12, p < .001, but not among those who had been categorized in the non-competent group, M = 4.39, SD = 1.02 and M = 4.39, SD = .79, respectively, F(1, 35) < 1. 99 Competent Non Competent Participants’ group Figure 4. Means of conformity to in-group and out-group norms as a function of groups’ competence in the task (Experiment 3). Competent group members conformed more to in-group norms than noncompetent group members, F(1, 35) = 5.93, p = .02. Conversely, non-competent group members conformed more to out-group norms than competent group members, F(1, 35) = 9.63, p = .004 (cf. Figure 4). Comparing the rates of in-group and out-group pictures with the middle of the scale, competent participants preferred in-group pictures, t (19) = 6.18, p < .001, and rejected out-group pictures, t(19) = 2.97, p = .008, whereas noncompetent participants did not preferred or rejected in-group or out-group pictures, highest t (16) = 2.03, ns. Conformity, identification and perceived groups’ competence. To examine the relationships between the three variables we first computed three indexes. The index of Relative In-Group Competence was obtained by subtracting the score of perceived competence of the out-group from the score of perceived competence of the in-group so that the more positive the more the participant perceives the in-group as superior to the out-group. The index of Relative In-group Identification was obtained by subtracting In-group Norm Out-group Norm 1 2 3 4 5 6 7 Conformity 100 identification from the out-group from identification to the in-group so that the more positive the more the participant identified to the in-group relatively to the out-group. Finally, the index of Relative In-group Conformity was obtained by subtracting outgroup conformity from in-group conformity so that the more positive the more the participant conforms to the in-group relatively to the out-group. Relative In-group Conformity was significantly correlated to Relative In-group Identification, r = .57, p < .001, and to Relative In-Group Competence, r = .36, p = .03. Relative In-group Identification was marginally correlated to Relative In-Group Competence, r = .28, p = .09. We also correlated the three indexes within each Group condition. Relative In-group Conformity was correlated to Relative In-group Identification in the competent group condition, r = .52, p = .02, n = 20; other correlations, highest, r = .28, ns. Discussion The results support the idea that informational influence is inseparable from normative influence and that both aspects concur to conform to in-group norms: identification is associated with conformity directly, thus increasing normative influences, and indirectly because it relates to the perceptions about the validity of group norms, thus increasing informational influences. However, the results also indicate that the above relationship is stronger when in-group credibility is externally recognized. When group’s credibility is questioned by external sources the relationship among in-group identification, perceived competence and conformity decreases. Several aspects may have contributed for this result namely the conflict between opposed motives. This conflict is also reflected in the perceptions of in-group and out-group competence. Indeed, even if the ascriptions of credibility came from a neutral or impartial source, the experimenter, non-competent members distorted the information and perceptually leveled the two groups. The perception of leveled credibility is consistent with their non-preference for one of the two group norms. If compared with the results of Experiment 2, competent group members displayed more in-group conformity as well as rejection of out-group norms. The fact 101 that the same did not occurred in non-competent group members suggests that explicit norms do not boost exclusively normative influences. The larger influence of explicit in-group norms as regards implicit norms seems to depend mainly on its perceived validity, not of self-imposed pressure to comply. It could be argued that there were some noticeable differences between Experiments 2 and 3 preventing the comparison of their respective results. In the next experiment, we search further validation of our hypotheses by comparing directly the implicit and explicit influence of competent and non-competent groups. 102 Experiment 4 Effects of Sources’ Credibility and Group Membership on Conformity to Explicit and Implicit Norms In the present experiment, we categorized the participants in two artificial groups with different competence in the task, using the same procedure as in Experiment 3, and added a condition in which participants were unaware of the groups’ norms. Our predictions are the same of the previous experiments: competent group members will conform to in-group norms and non-competent group members will not conform to any of the group norms, that is, there will be no differences between the patterns of implicit and explicit conformity. As argued before, explicit norms are more prescriptive (Marques et al, 1998) and enhance the need for positive distinctiveness (Turner et al, 1988). Consequently, we expect an increase of in-group conformity from the implicit to the explicit condition in the case of competent group members for whom the in-group is a valid reference. In what concerns the relationships of in-group identification, perceived competence with conformity we expect positive correlations, at least, when norms are explicit and the group is competent. Method Participants were 57 high-school pupils (37 female and 20 male; M = 16.61, SD = 2.65) that volunteered to participate. We used a 2(Norm Type: Explicit vs. Implicit) x 2(Group: Competent vs. Non-competent) x 2(Norm: In-group vs. Out-group) design, in which the latter factor was within-participants. The procedure was the same as in the previous experiment with the exception that, whereas participants in the explicit norm condition were called attention to the distinctive graphic details of the two groups’ preferred pictures, participants in the implicit norm condition were not informed of the details. Results Identification. All participants recalled their correct perceptive type. To inspect on the participants’ in-group identification we averaged the three items on identification to the in-group into an index of in-group identification (Cronbach’s alpha = .85), and averaged the items on identification to the out-group into an index of out-group 103 Competent Non-Competent Participants’ Group identification (Cronbach’s alpha = .82). We then conducted a Group x Norm Type x Identification (to the in-group vs. to the out-group) ANOVA in which Identification was a within-participants factor. The analysis yielded a single significant effect of the within-participants factor, F(1, 53) = 45.91, p < .001, η2 = .46; all the other effects, highest, F(1, 53) = 2.90, ns, η2 = .05. Participants identified more with the in-group than with the out-group independently of the fact that they belonged to the credible or to the not credible group, M = 5.28, SD = 1.27 and M = 3.37, SD = 1.34, respectively. Figure 5. Means of attribution of competence in the task for in-group and out-group as a function of the categorization in the competent or in the non-competent group (Experiment 4). Perceived groups’ competence. We then inspected the responses to the two items that measured the perceived competence of the two groups in the task. A Norm Type x Group x Perceived Competence (of the in-group vs. of the out-group) ANOVA, in which Perceived Competence was a within-participants factor, revealed a significant effect of Group, F(1, 53) = 10.82, p = .002, η2 = .17. All the other effects were nonsignificant, highest, F(1, 53) = 3.70, ns. Whereas competent participants evaluated the in-group as more competent in the task than the out-group, F(1, 55) = 14.44, p < .001, η2 = .21, M = 5.42, SD = 1.31 and M = 3.68, SD = 1.33, respectively, non-competent In-group Out-group 1 2 3 4 5 6 7 Attribution of competence 104 participants did not evaluate the in-group as more competent than the out-group, F(1, 55) < 1, M = 4.77, SD = 1.61 and M = 4.31, SD = 1.59, respectively (cf. Figure 5). Recognition and recall. We computed a recognition index by summing the number of pictures correctly recognized (minimum = 0; maximum = 20). The Norm Type x Group ANOVA on the recognition index yielded a significant main effect of Norm Type, F(1, 56) = 67.31, p < .001, η2 = .56. Other effects were all non-significant, highest effect, F(1, 56) = 3.46, ns. In the implicit condition, recognition was not different from chance, t(28) < 1, M = 10.14, SD = 2.59, whereas in the explicit condition, recognition was superior, t(27) = 9.87, p < .001, M = 17.54, SD = 4.04. In the implicit condition, 23 participants reported having found a distinction between the two series of the pictures but none was able to articulate the actual difference, or a distinction close to the actual difference. Five participants reported that Proficient pictures were lighter than Progressive pictures, but three were in one balance condition and two were in the other balance condition (cf. Table 3). One participant reported that the Progressive pictures were darker than the Proficient pictures. In the explicit condition, all participants mentioned the graphic details except for eight participants that, as revealed in the debriefing, thought they should mentioned other features apart from the details. 105 No. Group Proficient Pictures Progressive Pictures Balance 1 16 Progr More creative but hardly comprehensible More comprehensible 17 Progr Very complicated More comprehensible 18 Progr Clearer more blank spaces Darker 19 Progr Composed by smaller objects Composed by larger objects 21 Progr With smaller characters With larger characters 25 Progr Lighter with more blank spaces Darker 27 Progr Funnier and more elaborated More simple 20 Profic More declarative and comprehensible Harder to understand 22 Profic More explicit, show creativity More practical (abstract) 23 Profic Explicative, interesting, enhance the beauty of the pictures and of reality More practical, simple and abstract 28 Profic With beautiful figures with imagination More practical and uglier Balance 2 5 Progr More round and complex More abstract and simpler 6 Progr Simpler More complex 7 Progr More complicated and artistic Simpler 9 Progr Black dots are more concentrated More white spaces 12 Progr More abstract With some figures like animals, etc. 13 Progr Lighter Darker 14 Progr Lighter with more dots Darker with more squares 1 Profic - More realistic with real forms 3 Profic - Smaller 4 Profic More varied texture Less varied texture 10 Profic More irrational, e.g., with waves, oscillations Less irrational 11 Profic Lighter Darker Table 3. Reports on the distinction between the two series of pictures in the implicit norm condition ordered by balance condition and participants’ group (Experiment 4). Note: Profic = Proficient; Progr = Progressive. 112 In the second study, we shall examine the operation of implicit processes in normative differentiation. The next chapter is dedicated to review relevant literature on intragroup differentiation processes that are related with social identity. 113 CHAPTER 4 SOCIAL IDENTITY AND NORMATIVE DIFFERENTIATION Not all members conform to the influence of the group and its norms. The extent into which members conform or deviate from group norms is a major cause for intragroup differentiation, dubbed normative differentiation in the remaining of the chapter8. Normative differentiation was early investigated in the context of small group dynamics (e.g., Cartwright & Zander, 1968) and in the context of minority influence (e.g., Moscovici, 1976; 1985). We shall briefly review these traditional approaches to normative differentiation. Traditional Research on Normative Differentiation Classical references in the study of reactions to group deviance in small groups are Schachter’s (1951) and Festinger, Schachter and Back (1950) experiments. Theoretically, Festinger’s (1950) theory of informal social communication is also a significant benchmark of this field. We have already tackled some of Festinger’s (1950) principles in the previous chapter; therefore, we shall briefly review the hypotheses more relevant for the present topic and proceed with subsequent developments. Group Pressures towards Uniformity Festinger (1950) presents several postulates regarding group deviance and the group pressures towards uniformity. The author proposes that three group variables are related to the magnitude of pressures to uniformity: the perceived opinion discrepancy among group members, the relevance of the issue to the group function, and group cohesiveness. Two other variables related with the deviant members are predicted to increase pressures to uniformity: The perception of deviants as group members or the desire that they will be group members, and the perception that communication pressure will change their opinion. The theory also assumes that opinion change in deviant members is a function of the magnitude of the group pressure, their desire to remain in the group and the perceived support for their opinion in other groups. 8 We use the term proposed by Marques, Abrams, Páez and Martinez-Taboada (1998b) 114 In his classical experiment, Schachter (1951) provided empirical support for some of Festinger’s (1950) postulates. The author asked participants to participate in a group discussion, allegedly, aimed to form a club. There were 32 groups in four experimental conditions: high or low group cohesiveness and relevance or irrelevance of the issue of discussion. The participants were asked to discuss the treatment that should be given to Johnny Rocco, a juvenile delinquent. The case was presented in a sympathetic way to induce participants’ favoring of a kindhearted treatment as opposed to punishment. Clubs were formed of five to seven participants and three confederates: the mode, which agreed with the group during all the discussion; the slider, which initially disagreed with the group, defending a harsh punishment of the delinquent, but allowing to be gradually persuaded by the group; and the deviant, which disagreed with the group during all the discussion, defending a harsh punishment of the delinquent. Relevance was manipulated by the relevance of the discussion topic to the purpose of the club. For half of the participants, the purpose of the club was either editorial, to advise national magazines on articles and policy, or case study, to advise a group of lawyers, judges and social workers on the treatment of delinquents. For the other half of the participants, the club was aimed to form a local theatre company or a local radio station. The discussion of the juvenile delinquent case was made relevant for the former two clubs but not for the two latter clubs. Cohesiveness was manipulated by the attractiveness of the club to the members. The case-study club was formed by participants that previously showed high interest in joining the club, and the editorial club was formed by participants that previously showed high interest in joining the case-study club. The theatre club was formed by participants that previously showed high interest in the theatre activity, and the radio club was formed by participants that previously showed high interest in joining the theatre club. The analysis of communications during the discussion revealed that communications to deviants increased linearly during the discussion and were stronger in the high relevance and high cohesiveness conditions than in the low relevance and low cohesion conditions. In the high cohesiveness / high relevance condition, there was a slight decline by the end of the discussion. Communications to sliders increased in the beginning of the discussion; when sliders shifted to the group position, communication decreased steadily until the end of the discussion. Finally, communications to modes 115 were insignificant during all the discussion. The results also show that, among the confederates, modes received more sociometric choices than sliders and these latter more than deviants did. A similar pattern emerged in the assignment to three committee positions in the club: modes received more nominations to the position with an intermediate prestige; sliders received nominations to either the most prestigious or the least prestigious of the three positions; finally, deviants received most nominations for the less prestigious position except in the low cohesiveness and low relevant condition. The results obtained by Schachter’s (1950; cf. also, Festinger & Thibaut, 1951) confirmed those of Festinger et al’s (1950) field study concerning the unpopularity of members that deviate from the group standards and do not contribute to the group, especially in cohesive groups. As referred in the previous chapter, Festinger (1950) postulates that groups allot time and energy communicating with members that deviate from the group consensus and, ultimately, reject them because deviant members jeopardize the validity of the group consensus about social reality and the possibility that the group reaches its goals. In contrast, normative members uphold the validity of group consensus and contribute for group locomotion. Normative members endorse the group opinions and goals, consequently, they are more attractive and are assigned to more prestigious positions than are deviant members. Normative members also change their opinions less than deviant members, and communicate more to deviants members than to other normative members (Festinger, Gerard, Hymovitch, Kelley & Raven, 1952). Differential Reactions to Deviance Israel (1956) and Collins and Raven (1969) presented relevant contributes to the study of group reactions to deviance. Israel (1956) pointed out that, reactions to deviance could not be reduced to a general negative reaction. The author distinguished between inclusive reactions, including attempts to change the deviant’s opinion, and exclusive reactions, including isolation or expulsion of the deviant from the group (cf. also Orcutt, 1973). Israel (1956) also argued that individuals do not always seek corroborating support for their beliefs. Often individuals are motivated to evaluate the objectivity of their opinions and choose to compare with opinions different from them. Collins and Raven (1969) also contemplated the instances in which individuals are 116 motivated to re-evaluate their own opinions upon disagreement with others. The authors distinguish between different strategies aimed to deal with disagreement: Individuals may change their own opinions; attempt to persuade the other; minimize the importance of the disagreement; rejecting or derogating the other; and distinguish the other as a person from his or her opinions. Status and Deviance Another perspective, which tackles the issue of group deviance, is that of Hollander (e.g., 1958; 1960). Hollander (1958) argues that members acquire status in the group by the accumulation of positive impressions in other members, what the author dubbed ‘idiosyncratic credit’. These credits result from positive contributes of the member to the group in the form of competent performance, guiding the group in successful directions, etc. Idiosyncratic credit allows the member to show nonconformity to the group consensus and to be influential to change the group norms. As Hollander (1960) argues, ‘A person gains credits, i.e., rises in status, by showing competence and conforming to the expectancies applicable to him at the time. Eventually his credits allow him to nonconform with greater impunity’ (p. 365). The model was applied to understand how leaders are able to change the group norms without the resistance from the followers (e.g., Hollander, 1985). An important specification of the model is that for a competent member to earn idiosyncratic credit for nonconformity is that s/he must previous show conformity to the norms of the group. In an experiment aimed to confirm the model, Hollander (1960) formed groups of six participants that were to work in 15 trials on a complex matrix in order to obtain the maximum payoff for the group. Each group had to agree previously on the rules for the procedure in the trials. One confederate deviate from the group in following conditions: he contested and opposed, or not, the group in this preliminary phase; and later in the experiment, he contested and violated, or not, the rules established by the group to the procedure. In all conditions, the confederate proved to be highly competent in the task thus contributing for maximizing the group payoff. The results showed that groups accepted the (competent) opinion of the confederate the most when he conformed to the group before and during the trials and accepted it the least when he contested the group in the two moments. More 117 importantly, the group was more influenced by the member’s opinion when he violated the rules but had previously agreed with the group, than when he opposed to the group in the preliminary phase and conformed to the rules during the trials. Hollander’s (1958; 1960) model was subjected to several tests that refined its predictions relatively to the deviant’s status or existence of formal punitions (e.g., Wahrman, 1970) or the degree of non conformity of the deviant (e.g., Wiggins, Dill & Schwartz, 1965). Positive Differentiation of Minorities Another field of research that is relevant for the issue of normative differentiation is that of minority influence. As described in the previous chapter, minorities can produce opinion change as effectively as majorities (e.g., Moscovici & Lage, 1972; Moscovici, Lage & Naffrechoux, 1969; Mugny, 1982; Mugny & Pérez, 1991; Nemeth & Wachtler, 1974). Although it is assumed that, in general, they provoke negative feelings in majority members, “a minority is bound to be rejected and disliked” (Moscovici, 1976, p. 199), minorities’ opinions and arguments, conveyed in particular modes of communication, are positively considered and evaluated by majorities’ members. As Moscovici (1976) proposes, whereas majority influence is accepted because majority sources are attractive and therefore liked, minority influence is accepted because minority sources are admired and recognized for their assurance. For Moscovici (1976), attractiveness and influence are two independent aspects that are correlated in the case of majorities and uncorrelated in the case of minorities. For instance, in the second part of Moscovici, Lage and Naffrechoux’s (1969) experience, in which two confederates consistently diverged from the majority of four naïve participants (cf. previous chapter), the authors asked participants to judge the other five members regarding their competence, assurance, preference to integrate a future experiment and capacity to lead the group in a future experiment. The results showed that the judgments of the deviant/minority members did not differed from those of the majority in all respects except in assurance, in which the first of the pair of deviants scored higher than did all the other members. Nemeth and Wachtler (1973) simulated a trial with a five person-jury, in which one confederate (to be deviant in the forthcoming interaction) either chose to sit in the head chair or was assigned to the head chair at the beginning of the session. The 118 confederate was able to change the group consensus in the first but not in the second condition. In a post-experimental questionnaire, the ‘willful’ deviant (but not the ‘forced’ one) was judged as more consistent, more independent, more strong-willed, to have induced them to think more and reassess their own positions than the other members of the group. The deviant was also considered less perceptive, less warm, less reasonable and fair, and was less liked than the other members were. The minority influence research (see also, Levine & Ranelli, 1978; Nemeth & Wachtler, 1974) showed that deviant members are evaluated negatively in dimensions traditionally used to measure differentiation within majorities, such as, sociometric choices, appointment to leading positions, but they may be positively evaluated in dimensions relevant for the change of consensus, such as, respect, independence, certainty or influence. We have succinctly reviewed traditional approaches to normative differentiation, which addressed processes occurring in small interactive groups. We shall now focus on social identity models of intragroup differentiation, whose principles apply to large social groups as well as to small interactive groups. Social Identity Approaches to Normative Differentiation As seen in Chapter 1, the social identity approach tends to stress the role of intergroup differentiation and intragroup assimilation, rather than intragroup differentiation (e.g., Doise, Deschamps & Meyer, 1978; Eiser & Stroebe, 1972; Tajfel, 1959; cf. Chapter 1). However, several aspects mediate the perception of differentiation within in-group or out-group, such as, the relative status of the groups, and the relevance and valence of the comparative dimension (cf. Simon, 1992). For instance, Simon and Pettigrew (1990) manipulated the size of the in-group as regards the outgroup in artificially created groups and found that participants perceived more in-group variability in the case of majorities and more out-group variability in the case of minorities. Participants also perceived the behavior of minority out-group or in-group members to be more predictable than that of majority members (cf. also, Simon & Brown, 1987; Simon & Mummendey, 1990). These results are consistent with the idea that minorities reveal stronger in-group identification than majorities, resulting in a higher level of self-stereotyping (e.g., Brewer, 1991; 1993; Mullen, 1991). Other 119 research related to group status (e.g., Lorenzi-Cioldi, 1998; Lorenzi-Cioldi, Eagly & Stewart, 1995) and power (e.g., Fiske & Dépret, 1996; Stevens & Fiske, 2000) revealed that low status or powerless group members tend to self-stereotype more than high status and powerful group members. The above evidence opposes the prevailing idea that the tendency to differentiate among in-group members and to homogenize out-group members leads the processing of information on groups (e.g., Ostrom, Prior & Simpson, 1981; Quattrone & Jones, 1980) and in the formation of group representations (e.g., Linville, Fischer & Salovey, 1989; Judd & Park, 1988; Park, Judd & Ryan, 1991). As Simon (1992) pointed out, the perception of relative group homogeneity is socially mediated and, as such, a ‘hypothetically variable process’ that is not reducible to a general cognitive process. More important for our present concerns, the above results indicate that social identity plays a decisive role in the differentiation within the in-group (cf. also, Simon, Pantaleo & Mummendey, 1995). Indeed, research on several intragroup phenomena has revealed that social identity processes are also relevant in the intragroup context (cf. Hogg & Hardie, 1991; 1992; Marques, Yzerbyt & Leyens, 1988; Marques & Yzerbyt, 1988). As Hogg (1996) pointed out, the social identity view on intragroup processes distinguishes from the traditional view because it is based on distinct assumptions, namely: “(1) A group is a collection of people who categorize themselves in terms of the same social categorization; (2) interpersonal processes are conceptually separate from group processes; (3) the intergroup social comparative context is an integral feature of the analysis; and (4) a complete explanation requires articulation of cognitive and social processes.” (p. 71) One of the former social identity approaches to intragroup differentiation was that of Hogg and colleagues (e.g., Hogg, Cooper-Shaw & Holzworth, 1993; Hogg & Hardie, 1991; 1992). These authors investigated the effect of members’ prototypicality in their social attractiveness (cf. also, Hogg, Hardie & Reynolds, 1995; for reviews, see Hogg, 1996a; 1996b). The effects of prototypicality on leadership endorsement were also the focus of extensive research (e.g., Fielding & Hogg, 1997; Hains, Hogg & Duck, 1997; Hogg, Hains & Mason, 1998; for a review, see Hogg, 2001a). Differentiation 120 based on prototypicality matches the scope of the present analysis because, for Self- Categorization Theory, the group prototype corresponds to the defining norm of the group (cf. Chapter 1). Intragroup Differentiation as a Function of Members’ Prototypicality As seen above, Self-Categorization Theory sustains that individuals are motivated to hold a clear perception of their social environment. Individuals feel confident when others’ behavior is consistent with their representations of the respective groups. Certainty about one’s representations of the social world corresponds to certainty about one’s own position in the world and to certainty about others’ behavior; the perception that others’ behavior is predictable from the recognition of their group membership (cf. Hogg, 2000; 2001; Turner et al, 1987). Prototypicality and Attractiveness Whereas typical in-group or out-group members affirm the validity of people’s beliefs on the social world, atypical in-group or out-group members invalidate them and generate uncertainty. A consequence of this need to reduce to uncertainty is that members’ social attractiveness varies with their prototypicality, i.e., members will be the more socially attractive the more they represent the group both by being similar to other in-group members and by differentiating from out-group members (cf. Hogg, 1992; 1996; Whetherell, 1987). Hogg, Cooper-Shaw and Holzworth (1993) measured the prototypicality of the self and other members, self-involvement in the group in work teams of an Australian organization. The authors also collected the respondents’ sociometric choices based on the group or on friendship. The authors found that respondents that ascribed more importance to the group were not considered particularly attractive, on an interpersonal basis; conversely, those that did not feel the group highly important, attracted more choices based on friendship. In general, group-based attraction for other members was more related with their prototypicality than with their personal attractiveness, whereas interpersonal attraction was not related with group variables. Categorical salience and identification. Higher attraction to prototypical members is associated with the distinction between interpersonal and social attraction, 121 an issue addressed in Chapter 1 (cf. Hogg, 1987; 1992; 1993; Hogg & Turner 1985a; 1985b). As social attraction is intimately related with the salience of common group membership, the positive differentiation of prototypical members should occur in salient group contexts rather than in salient interpersonal contexts and should be more evident in members highly identified to the group. To inspect these hypotheses, Hogg and Hardie (1991) asked the members of an Australian football team to rank their teammates in terms of prototypicality, social attraction and personal attraction following instructions that made salient their group membership or their interpersonal relationships. The results revealed that the members’ prototypicality was more associated with liking and popularity under salient team instructions, than in conditions that rendered individuality and interpersonal relationships salient. In addition, judgers who identified more strongly with the team differentiated more among other members based on prototypicality than did less identified judgers. Hogg and Hardie (1992) also confirmed the hypothesis that group salience determines the preference for prototypical members, using the autokinetic paradigm with four-person groups. In the groups that converged more in their judgments of estimates, that is, those in which the in-group category was more salient, participants reported more attraction for more prototypical members than for less prototypical members. Hogg, Hardie and Reynolds (1995) manipulated the dimensions in which the participant and the target member were similar. The experiment was presented as a study on consumer behavior and participants were first categorized in one of two groups of consumer style, ‘Visual’ or ‘Tactile’, as function of their responses to a bogus projective test. After testing an alleged new product, participants were led to believe that they would continue the testing with another partner (interpersonal condition) or a four-person group (group condition). They were then asked to form an impression on their partner or a member of the group. The participant and the target member could be similar in two dimensions (preferences in appearance and flavor of the product): one dimension was stereotypical of the in-group and the other of the out-group. Participants were similar to the target in either one or other dimensions, in both or neither. The results showed that the more strongly participants identified with their group the more they favorably evaluated the in-group target when the target was similar in the in-group stereotypical dimension, thus being in-group prototypical. This relationship 128 Therefore, group members often are not fully aware of the existence of norms until the norm is violated. The different characteristics of denotative and prescriptive norms entails that their violation is differently considered. In-group members that conform to prescriptive norms uphold the value of the group whereas members that deviate from these norms undermine this value. This idea is consistent with results obtained by Marques and colleagues (1998b) concerning reactions to deviance from high-salient and low-salient norms. The authors conducted an experiment allegedly about decision making in juries. In a first session, participants examined a murder case involving six suspects and asked to rank the suspects according to their responsibility in the crime. They were also informed that there were two patterns of decision making that could be diagnosed through their responses. In the second session, participants received their results informing them as being typical of one of the decision patterns. The salience of the norms was manipulated in the following way. In the high-salience condition, participants were told that to belong to a pattern people should present a specific ranking of responsibility. The in-group typical ranking (in-group norm) matched exactly the participant’s ranking whereas the out-group typical ranking (out-group norm) was the opposite of the participant’s sequence. The participants then read the answers given by five individuals belonging either to same or other pattern. The responses of four individuals matched their own pattern and those of the fifth individual were close to the out-group norm. In the low-salience condition, participants were simply shown the members’ responses and no information was given regarding the group norms. The authors found that participants upheld in-group normative members and downgrade in-group deviant members, significantly more than equivalent out-group members, only when norms were high salient. When norms were low salient, participants did not differentiate between normative and deviant members and showed an overall preference for in-group members in detriment of out-group members. In another experiment using the same general procedure, Marques and colleagues (1998b) further found that the extremized intragroup differentiation found within the in-group when norms were perceived to be prescriptive, was accentuated when participants believed their judgments of members were to be seen by a typical in-group member and were attenuated when they believed the examiner was a typical out-group member. 129 Together, these results suggest that group norms become prescriptive as members become more aware of their importance for in-group identity, leading to strategic evaluation of in-group members. Relevance of the prescriptive norm. Deviance from prescriptive norms leads to subjective group dynamics if the norm is perceived to be meaningful for a positive group identity. For instance, Marques, Abrams and Serôdio (2001a) first ensure that students were generally supportive of student initiation practices. Then, they asked Psychology and Law students to state their opinion on a continuum of seven statements from totally support to totally rejection of student initiation practices, and to indicate the position above which they would disagree. They were then showed a bogus distribution of responses of a sample of Psychology or Law students that either validated or undermined the generic norm. Then, they were asked to evaluate Psychology or Law students (depending on their own course, in-group or out-group members) that agreed with their own opinion (what the authors dubbed, subjective normative members) and another in-group or out-group member whose opinion was above their rejection threshold (subjective deviant member). The results showed that when the norm was validated, deviant members were derogated and normative members were upgraded but evaluations did not vary with the targets’ group membership. However, when the norm was undermined, in-group deviant members were more derogated than similar outgroup members were, and normative members were more upgraded than similar ingroup members were. As the authors affirm, when the norm was generally upheld, deviant behavior was likely to affect the norm validity and consequently the value of ingroup identity (cf. also, Marques 1990). Relevance of group membership. Extremized differentiation among in-group members depends not only of the relevance of the norm, but also of the relevance of group membership either circumstantially or as a chronic self-defining attribute. Branscombe, Wann, Noel and Coleman (1993) showed that when the group is not meaningful, individuals extremize their evaluations of normative and deviant members more in the out-group than in the in-group. Participants were asked to evaluate the author of an article about a basketball game between participants’ own university team and a rival university team. The author was either a fan of the participants’ in-group or out-group team and was either a loyal fan (supporting the team in all occasions) or 130 disloyal (supporting the team only when it succeeded). In the game reported in the article, the in-group team either lost or won the game representing a threat or not to the group identity. The results showed that participants extremized their evaluations of the outgroup loyal and disloyal authors as regards the in-group authors when their level of ingroup identification was low. High identifiers presented the reversed pattern of evaluation: they extremized their evaluations of the in-group loyal and disloyal author relatively to the comparable out-group authors. The extremized evaluations of loyal and disloyal in-group targets were more evident when the in-group team had lost the game in the report. As Branscombe and colleagues (1993) claim, when the target’s actions have few or no implications for the judger’s identity, out-group extremization occurs, whereas when the target represents a threat to the judger both contextually and because this identity is a meaningful reference, the judgments of in-group members are extremized as regards those of out-group members. The results above, taken as a whole, suggest that the importance of the dimension in which members evaluate deviant members and the importance of the group for them are determinant of the motivation to protect the group. More evidence for this idea comes from results indicating that prototypical deviant members are more rejected than their non-prototypical counterparts (Serôdio & Marques, 2005). Specifically, members that deviate from norms that are important to define a positive group identity (prescriptive norms) are more derogated when these members are denotatively more typical of the group than when they are atypical. Conversely, group members that endorse prescriptive norms are more upgraded when they are perceived as more typical of their group than when they are not. In a similar vein, Pinto and Marques (2005) showed that ‘full members’ (i.e., members that have long joined the group, and learned and assimilated in-group norms) are more derogated when they deviate from prescriptive norms than ‘novices’ (i.e., members that are just entering the group and have not assimilated in-group norms). Taken as a whole, these results suggest that the more members are descriptive of the group as a whole the more they will be rejected when they deviate from prescriptive norms. In other words, the more members are perceived as representative of the group, the more they have the responsibility to convey a positive image of the group, the more their deviant behavior endangers the 131 positivity of the group, and the more it is difficult for the group to differentiate from the deviant behavior, by recategorizing them as out-group members (Marques et al, 1998a; 1998b). Derogation of deviants and increased identification. The aim of the subjective group dynamics process occurring at the intragroup level is to reestablish a positive group identity and to obtain again subjective validity for the in-group norms undermined by deviant behavior. This would imply a reinforcement of in-group identification after the derogation of the deviant member. Marques and colleagues (1998a) presented supporting evidence for this hypothesis. Using the same general procedure described above, the authors, after categorizing participants into two decision-making patterns as regards a murder case, measured their identification to the in-group before and after they evaluate in-group or out-group normative and deviant members. Results showed that participants reported more in-group identification after than before the judgments and that normative differentiation predicted post in-group identification. In other words, the more they distinguished between normative and deviant in-group members the more they subsequently identified with the group. The Development of Sensitivity to In-Group Deviance Extreme intragroup differentiation as a strategy to uphold the validity of ingroup norms seems to appear later in individuals’ development. Support for this idea comes from evidence presented by Abrams, Rutland, Cameron and Marques (2003b), suggesting that older children are more sensitive than are younger children, to the implications of in-group deviance. Specifically, the authors compared evaluations of 6-7 years old to 10-11 years old children of same Summer School (in-group) and other Summer School (out-group) children, that either praised their own Summer School (normative member) or praised the other Summer School (deviant member). The results show that whereas younger children showed simple in-group bias, favoring in-group over out-group members, but did not differentiate between normative and deviant target members, older children also showed in-group bias while differentiating between ingroup normative and deviant members. However, the result showing that younger children were less sensitive to deviance than older children must be pondered against the differential importance that a 132 norm such as that of loyalty has for younger and older children. As the authors recognize, “It is conceivable that in the case of strongly socially prescribed norms (e.g., gender- or age-related behavioral norms), younger children might be more adept at recognizing deviance and its implications” (p. 172). Notwithstanding, Abrams, Rutland and Cameron (2003a) obtained similar results with norms whose relevance for participants was assumedly independent of their age strata – the norm of supporting national against foreign football team. Taken as a whole, evidence indicates that, contrarily to in-group bias, intragroup differentiation depends on an increased consciousness of social norms and the negative impact that in-group deviance may have on group identity. In sum, the research on the black-sheep effect and on the Subjective Group Dynamics model distinguishes from the traditional approach to reactions to deviance by its focus on social identity processes. Subjective Group Dynamics model also distinguishes from Self-Categorization Theory because it focuses on the role of norms to establish a positive social identity additionally to clear-cut group boundaries. A consequence of this focus is that the model postulates that intragroup norms may be salient in intergroup contexts, without the re-categorization of the social situation as function of these norms. Whereas for Self-Categorization Theory a salient norm defines the significant categorization in the contexts, for the Subjective Group Dynamics model group norms have, at least, two functions: One function is the differentiation between social categories (denotative or descriptive norms) and the other is the establishment of a positive group identity (prescriptive norms). Other research on group norms is consistent with this distinction. Emotional Consequences of Own Normative and Deviant Behavior The distinction between in-group descriptive and prescriptive norms (cf. also, Cialdini, Kallgren & Reno, 1991) was addressed by Christensen and colleagues (2004) in terms of the emotions induced by conformity to or deviance from these norms. In one experiment, the authors manipulated the descriptive or prescriptive nature of the violated or endorsed norm, and the salience or non-salience of the group to which the norm referred. The experiment was presented as a study on the causes for good and bad performance. The task consisted in obtaining the best performance in a Rubik’s Cube 133 puzzle. Each participant worked in two separate desks with one confederate that simulated to be unable to perform the task successfully. Participants then read the instructions before reporting their feelings on the task. They were recalled that they had worked separately from the partner (the data of participants that offered help to the confederate were not included) and, depending on conditions, they were informed that most people did or did not offered help (violation or conformity of descriptive norm) or that they had not supported the idea of helping others or that they had supported the idea of not cheating (violation or conformity of prescriptive norms). These were the instructions of the non-group condition. In the salient in-group condition, the instructions were the same but instead of ‘most people’, they referred to ‘most students of the participant’s university’. As expected, the results indicated a general tendency to feel more positive emotions after conforming to, than after deviating from, the norm. This difference was significant only when the norm referred to a salient in-group revealing its relation with social identity processes. Moreover, the difference was significant only when the norm was prescriptive suggesting its higher meaning for the value of group identity as regards that of descriptive norms. The results also reinforced the relation between the reported emotions and group identity by revealing an interaction with level of in-group identification. Whereas high identifiers reported more positive emotions when they had conformed to the in-group prescriptive norm than they had deviated from it, the reports of low identifiers do not show such effect. The results of the descriptive in-group norm condition revealed no such effects (cf. also, Costarelli, 2005). The above results suggest that a salient group membership entails the need to present in-group prescriptive but not the need to present in-group descriptive behavior. The results are consistent with the Subjective Group Dynamics’ postulate that deviance from in-group prescriptive norms threatens the value of the group identity. The effect of the context on norms prescriptiveness. In the experiments described above, Marques and colleagues’ (1998b) results suggest that descriptive norms may become incidentally prescriptive. In some contexts, irrelevant descriptive norms may become important to assert the group identity. Specifically, the authors found that participants reacted more extremely to a deviant from a norm when it was perceived as 134 important for the in-group definition, than when it was perceived as an undifferentiating attribute of the group. Christensen and colleagues (2004; Exp. 2) obtained results consistent with these, in the following conditions: in one situation, a norm described the in-group and the out-group behavior; in another situation, the same norm was descriptive only of the in-group, as opposed to another norm that described the outgroup. In the latter situation, the descriptive norm became prescriptive to differentiate the group from the out-group. Participants completed an initial bogus task, selecting the most needed items to survive after a plane crash. The responses allegedly revealed different survival strategies. In one condition, participants were told that students of their own University and those of another rival University had opposed task strategies and that their own strategy was typical of the in-group. Another condition was similar to the previous one but participants were told that their own strategy was typical of the other University students. In another condition, participants were told that the student of both Universities used the same strategy and their own strategy was identical to it. In the fourth condition, participants were told that their responses differed from the one used by the students of both Universities. There were thus four experimental conditions: conformity or deviance from differentiating or non-differentiating descriptive norms. The results showed that participants reported more positive emotions when they conformed than when they deviated from the norm. The effect was significant only when the norm was differentiating and increased from low to high-identified participants. These results again support the Subjective Group Dynamics model’s assumption that the prescriptive character of group norms derives from its role in upholding the positive distinctiveness of group identity. Conclusion In this Chapter, we reviewed research and theory on intragroup differentiation related to group norms. We began with a reference to traditional research relevant for this field focusing mainly in small group processes. In contrast with this research, the social identity approach focuses on the role of social categorization and social identification in differentiation within groups. 135 We reviewed two social identity approaches of intragroup differentiation. One approach emphasizes individuals’ need for cognitive clarity and the positive differentiation of prototypical members, which contribute for the distinctiveness of the in-group. Differentiation based on prototypicality applies to perceptions of members’ social attractiveness (e.g., Hogg et al, 1993; 1995) or endorsement to leadership positions (e.g., Hogg, 2001a). The other approach emphasizes individuals’ need for positive social identity. According to the Subjective Group Dynamics model (e.g., Marques et al; 1998a; 1998b; 2001), group norms accomplish two functions: the description of the expected group behavior and the maintenance of the group positive value. The former are denotative norms whereas the latter are prescriptive norms. Normative differentiation within the ingroup is more likely to emerge as function of prescriptive norms than as function of descriptive norms. Moreover, differentiation between in-group normative and deviant members is more extreme than that between the equivalent out-group members, suggesting the importance of prescriptive norms for group identity. The model postulates that the salience of group prescriptive norms does not imply the decreased salience of the intergroup context. Members may be aware of their group membership while differentiating between in-group members that conform and those that deviate from the group norms. The findings of research on the emotions induced by own normative and deviant behavior (Christensen et al, 2004), are consistent with the model. 136 Study 2 Implicit Normative Differentiation of Group Members In our second study, we explore the hypothesis that, in intergroup contexts, individuals differentiate between members that conform and those that deviate from the implicit group norms. We base our predictions on the Subjective Group Dynamics model (e.g., Marques et al, 1998a; 2001b). According to the model, normative differentiation within the in-group is a strategy aimed to protect a positive social identity. Individuals upgrade in-group normative members that contribute for a positive group identity, and downgrade deviant members that undermine this identity. Normative differentiation is thus more extremized in the in-group than in the out-group. Consistent with these assumptions, the model holds that derogation of in-group deviants does not correspond to a re-categorization of the deviant member as a function of the violated norm (Marques et al, 1998b). Normative differentiation occurs simultaneously with intergroup differentiation: The derogation of in-group deviants occurs when both group membership and group norms are salient in the context. Marques and colleagues’ Experiment 3 (1998b) provides evidence for the above idea. The experiment was presented as part of a larger research on jury decisionmaking. After ranking the characters involved in murder case according to their responsibility in the crime, participants were informed about the existence of two decision-making patterns. In the next session, participants were categorized into one of these two patterns. Participants were then asked to analyze the rankings of five other persons. However, before this task, half of the participants were informed that previous studies demonstrated that members of the participant’s decision pattern made the same ranking as the participant’s and that for the opposed decision-making pattern the typical ranking was opposed to that of the participant (salient norm condition). The other half of participants was not informed about the group norms (non-salient norm condition). Participants observed then the rankings of five other persons. In the in-group condition, four persons presented exactly the same ranking as the participant’s (in-group normative members) and one person presented the opposed ranking (in-group deviant member). In the out-group condition, four persons presented the ranking opposed to participant’s 137 (out-group normative members) and one person presented the same ranking (out-group deviant member). The results showed that participants evaluated the in-group, as a whole, more positively than they evaluated the out-group. In the non-salient norm condition, the evaluation of members reflected the same in-group favoritism and in-group members were better evaluated than were out-group members independently of being normative or deviant. However, in the salient norm condition, the in-group normative member was upgraded relatively to the out-group normative member and the in-group deviant member was downgraded relatively to the out-group correspondent member. These results suggest that extremized normative differentiation involves the salience of both categorization and of the in-group norm. When solely the categorization is salient, conventional in-group bias emerges with the typical accentuation of intergroup differences and intragroup similarities of salient intergroup contexts (e.g., Doise, Deschamps & Meyer, 1978; Simon, 1992). Previous evidence showed that in-group favoritism is an implicit response in salient intergroup contexts (cf. Cameira et al, 2002; Devine, 1989; Dovidio et al, 1997; Fazio et al, 1995; Greenwald & Banaji, 1995; Otten & Wentura, 1999; Perdue et al, 1990). In the present study, we aim to check on whether intragroup normative differentiation possesses the same implicit features, that is, if participants upgrade members that conform over those that deviate from implicit group norms. Consistent with the Subjective Group Dynamics model, we expect differentiation between normative and deviant members to be extremized within the in-group relatively to the out-group, revealing the need to protect in-group identity. Despite the discontinuity between the first and the present study in dependent variables, for the sake of clarity, we numbered the two experiments, Experiments 5 and 6, respectively. Experiment 5 Method Participants and design. Thirty-one high school pupils (23 females and 8 males; ages ranging from 17 to 21 years old, M = 17.65, SD = .88) volunteered to participate. The design was a 2(Group: In-group targets vs. Out-group targets) x 2(Member: 144 conducted a Group x Member x Similarity ANOVA on Favorability. The analysis revealed a significant full interaction, F(1, 27) = 13.63, p = .001, η2 = .34. The decomposition of the interaction indicated that the Group x Similarity interaction was significant only relatively to normative members, F(1, 27) = 10.93, p = .003, η2 = .29 whereas the evaluations of deviant members did not differ significantly, F(1, 27) = 2.32, ns. Further analysis showed that only in-group normative members were differently rated by high and low similar participants, respectively, F(1, 16) = 15.01, p = .001, η2 = .50; out-group normative members were not differently rated by the two types of participants, F(1, 13) < 1 ns (cf. Figure 9). In addition, among highsimilar participants, the in-group normative members were upgraded relatively to normative out-group members, F(1, 13) = 24.21, p < .001, η2 = .67, whereas low similar participants did not differentiated between the two targets, F(1, 16) = 1.87, ns. Evaluations of members and recognition. To inspect the role of recognition in the ratings of the targets, we divided participants into ‘good’ and ‘bad’ recognizers by median-split of the recognition index, M = 12.60, SD = 1.80 and M = 7.19, SD = 1.42, respectively, creating a new two-level variable, Recognition. A Group x Recognition x Member ANOVA on Favorability did not revealed effects involving Recognition, all effects, F(1, 27) < 1. Discussion The results did not support the idea of implicit normative differentiation. Contrary to expected, participants differentiate more within the out-group than within the in-group, favoring the deviant over the normative out-group member. The result may have been due to the high salient categorization suggested by several other results. For instance, participants’ evaluations were in-group biased, favoring in-group over outgroup members as a whole. Similarity to the in-group significantly moderated this tendency. Participants that reported high similarity to the in-group favored in-group normative members more than did participants that reported less similarity to the ingroup. Seemingly, participants evaluating in-group members were more concerned in differentiating the in-group from the out-group than in differentiating within the ingroup. As such, they presented in-group bias and assimilated in-group members. Participants evaluating out-group members also presented a form of in-group bias by 145 positively differentiating supporters of the in-group norm from supporters of the outgroup norm. The next experiment was conducted experiment individually, not in group sessions, with a more involving scenario in order to increase the focus on the norms as regards the categorization. 146 Experiment 6 The results of the previous experiment did not support the hypothesis of implicit intragroup differentiation suggesting more concern of participants in differentiating the in-group from the out-group than in differentiating normative from deviant members. An explanation ventured to these results was that the salience of norms was reduced relatively to that of the categorization. In the present experiment, we introduced several changes to the procedure and material to deal with this possibility. To increase participants’ (first-year Psychology students) interest in the group norms we told them that the second session was part of a study aimed to prepare educational material for kindergarten children. The stimuli used to convey the group norms were thus changed to fit the new cover story. In addition, participants completed the experiment individually and the instructions and stimuli were administered through an interactive computer program. Method Participants and design. Twenty-five students enrolled in an introductory course of Psychology (23 females and 2 males; ages ranging from 17 to 21 years old; M = 18.16, SD = .80) volunteered to participate. The design was again a 2(Group: In-group targets vs. Out-group targets) x 2(Member: Normative vs. Deviant) mixed design, in which Member is a within-subject factor and Group is a between-participants factor. Procedure. The experiment occurred in two sessions. The first session, aimed to prepare a convincing context for the subsequent categorization of participants into two groups, was identical to that of the previous experiment. Three days later, participants arrived alone at a scheduled time to the laboratory. They were then handed their personal report including the results of the perceptive test with their perceptive style (Deductive or Inferent), an index showing that they possessed most of their style characteristics (90%), and the information about the characteristics of each of the two styles described in the previous experiment. The second part of the experiment was presented as part of larger study whose results were to be used in the preparation of educational material for kindergarten children. Participants were asked to step into a soundproof box and sit at a table with a PC monitor and keyboard. They were informed that all they needed to proceed was 147 presented in the screen and that, except when instructed otherwise, responses were given on the keyboard. After starting the program, which would also save the participant’s answers, the experimenter left the box. Participants read “Previous studies have ascertained that the two perceptive styles have different esthetic preferences even among very similar images. In one of these studies, a group of Deductives and a group of Inferents separately agreed on the best among several versions of the same picture to include in a book for kindergarten children. The versions chosen by each group were always different from the ones chosen by the other group.” Participants then observed the versions purportedly preferred by Deductives and the versions preferred by Inferents. Figure 10. The three pairs of pictures presented to participants (Experiment 6). Each pair presents only one single difference. The pair at the left differs in the position of the ball; the pair at the center differs in the girl positioned at the right of the picture; the pair at the right differs in the position of the second ‘o’. The two versions of three pictures, depicted in Figure 10, were presented sequentially for two seconds with a two-second blank interval in between the two, and labeled with group that had preferred it. Participants were then asked to write down the differences they had found between the two versions. Subsequently, participants were informed that the present study was aimed to investigate the way the two styles made impressions of persons. They were then asked 148 to form an impression of two persons that had taken part of the alleged groups, based solely on the pictures they had personally chosen after the respective group arrived to consensus, Person A and Person B. The two targets were presented as belonging both either to the Deductive group (in-group targets condition) or to the Inferent group (outgroup targets condition). Person A’s choices, presented sequentially for two seconds each, were the three versions preferred by participant’s in-group, thus, depending on whether it was presented as an in-group or out-group member, Person A was either normative or deviant, respectively. They were then asked to register from 1 to 9 the impression produced by the person, in which 1= Very bad impression and 9= Very good impression. The same procedure was used for Person B. Person B’s choices were the three versions preferred by participant’s out-group, thus depending on whether it was presented as an in-group or out-group member, Person B was deviant or normative, respectively. To control for order effects in the evaluations of normative and deviant members, the normative choices of Person A and Person B were balanced evenly across conditions. Subsequently, participants were asked to register whether, in their opinion, each of the six versions, presented again for two seconds each, was preferred by Deductives or Inferents. This task corresponded to the test of recognition. Finally, participants were asked to complete a questionnaire measuring in-group and out-group identification in the same items as those of the previous experiments. In the end of session, participants were fully debriefed. Results Identification. All participants recalled correctly their perceptive type in the end of the experiment. To examine participants’ identification, we first averaged the items of in-group identification creating an in-group identification index, Cronbach’s alpha = .86, and averaged the items on out-group identification creating an out-group identification index, Cronbach’s alpha = .82. We then conducted a Group x Identification (with the in-group vs. with the out-group) ANOVA, in which Identification was a within-participants factor. The analysis yielded a single significant effect of Identification, F(1, 23) = 43.20, p < .001, η2 = .65, all the other effects, F(1, 149 23) < 1. Participants identified more with the in-group than with the out-group, respectively, M = 5.42, SD = 1.03 and M = 3.20, SD = 1.17. No. Group Deductive Pictures Inferent Pictures 1 Infer More movement More centered and balanced 2 Deduc Out of focus - 5 Infer More disorganized - 10 Deduc Different positions of the objects in the two versions - 11 Deduc More brilliant - 15 Deduc - Less defined, more confuse 17 Deduc - Dance: more movement 20 Infer - Sharper pictures 21 Infer Clearer - 22 Infer More organized - Table 7. Reports on the distinction between the two series of pictures (Experiment 6). Responses in bold mention aspects close to the actual distinctions. Note: Deduc= Deductive; Infer= Inferent Awareness tests. We first computed a recognition index by summing all the correctly recognized pictures, that is, the number of pictures of which the participant recognized the two versions. Thus, the index varied between 0 and 3. In average, participants recognized less than one picture, M = .80, SD = .82, and the recognition index did not vary across condition, F(1, 24) < 1. In what refers to recall of the distinctions, the analysis of the written reports indicate that none participant could describe any of the distinctions (cf. Table 7). One participant reported a distinction referring two of the three actual distinctions (# 10). However, the participant was not able to precise the distinctions; moreover, she could not classify correctly none of the three pairs of pictures in the recognition task indicating that the participant had only a vague notion of the distinction. 150 Impression In-Group Out-Group Targets’ Group Figure 11. Impression of normative and deviant members as a function of their group membership (Experiment 6) Impression of members. To examine the differences in the item measuring the impression of members, we conducted a Group x Member ANOVA in which represented the impression made on normative and deviant target-members. There was a single effect, the Group x Member interaction, F(1, 23) = 16.01, p < .001, η2 = .41, highest other effect, F(1, 23) = 1.98, ns. The analysis of the interaction revealed that ingroup normative targets were better evaluated than in-group deviant targets, M = 7.23, SD = 1.01 and M =6.15, SD = 1.14, respectively, F(1, 23) = 8.29, p < .008, and outgroup normative targets were worst evaluated than out-group deviant targets, M = 5.58, SD = 1.31 and M = 6.67, SD = 1.37, respectively, F(1, 23) = 7.74, p < .01 (cf. Figure 8). The in-group normative member was better evaluated than the out-group equivalent, F(1, 24) = 12.48, p = .002, η2 = .35, and deviant in-group and out-group members were similarly evaluated, F(1, 24) = 1.04, ns. All members were evaluated significantly above the middle of the scale except for the normative out-group member, t(11) = 1.54, ns, all other members, lowest, t(12) = 3.64, p = .003 (cf. Figure 11). 1 2 3 4 5 6 7 8 9 N ormative Membe r Deviant Member 151 Impression of members and in-group identification. To examine the impact of in-group identification on impression, we divided participants by median-split of the ingroup identification index, M = 4.81, SD = .73 and M = 6.50, SD = .35, respectively for ‘low’ and ‘high identifiers’, creating a new two-level variable, Identification. We then performed a Group x Identification x Member ANOVA. There were no effects involving Identification, all effects, F(1, 21) < 1, suggesting that in-group identification did not played a significant role in the judgments of the targets. Impression of members and recognition. To examine the role of recognition of the norms, we proceeded similarly, dividing participants by median-split of the recognition index into two groups, ‘good’ and ‘bad’ recognizers, with a average recognition of M = .42 , SD = .51 and M = 2.00, SD = .00, respectively, creating a new two-level variable, Recognition. We then performed a Group x Recognition x Member ANOVA, which did not yield significant effects involving Recognition, highest effect, F(1, 21) = 1.52, ns, suggesting that recognition of the norms did not played a significant role in the judgments. Discussion and Conclusions With the changes to the experimental procedure aimed to increase the importance of the group norms, the results partially supported the hypothesis of implicit normative differentiation. Specifically, participants favored members who endorse implicit in-group norms over those who support equivalent out-group norms. However, this result cannot be interpreted as the result of Subjective Group Dynamics. The Subjective Group Dynamics model postulates that the need to protect group identity derives from the simultaneous salience of social self-categorization and of the violated group norm (cf., e.g., Marques et al, 1998a; 2001b). The simultaneous salience of group and norm is revealed by the extremized evaluations of in-group normative and deviant members as regards their out-group equivalents. Our results present only one aspect of the expected effect: in-group normative members were favored over out-group normative members but in-group deviants were not downgraded relatively to out-group deviants. Other results suggest that, in the present experiment, the initial categorization was less salient than in the previous experiment. For instance, in-group targets were not 152 favored over out-group targets, as a whole, as in the previous experiment. Moreover, the level of in-group identification did not affected judgments, as in the previous experiment. Jointly, these two results suggest that the intergroup context before the judgments was less salient than in the previous experiment. The isolated conditions in which the tasks were completed as well as the more involving scenario seemed to have induced, as intended, an increased salience of the group norms but at costs of a decreased salience of the initial categorization. Participants seemed to be more concerned in differentiating as a function of the norms than as a function of the groups. Given the scarce indications of the initial self-categorization at the time of judgments, a re-categorization explanation could also apply to the results. The salience of the norms could have led participants to interpret implicitly the situation as the opposition between two new categories, derogating targets that endorse the in-group norm over those that endorse the out-group norm independently of their initial group membership. 153 GENERAL CONCLUSION Previous research has shown that attitudes, opinions and judgments are implicitly affected by group memberships. Our aim, in the present work, was to extend this investigation on social identity implicit processes to group influence and normative intragroup differentiation. We shall now review our major results, outline some of their implications and draw some prospects for future research. Group Influence Group influence is a major social-psychological field of research. Classical experiments in Social Psychology, such as those of Sherif (1936) and Asch (1951), addressed the influence that group norms have in individuals’ judgments. It is therefore a significant field to examine implicit processes. In addition, implicit processing is a relevant issue in the theory of group influence. Minority influence theories emphasize the distinction between direct and indirect influence (e.g., Moscovici, 1976; 1980; 1985; Mugny & Pérez, 1991; Pérez & Mugny, 1998), and compliance is often associated to deliberate behavior whereas internalization is associated to implicit assimilation (e.g., Cialdini & Trost, 1998; Kelman, 1958; Moscovici, 1976; Wood, 1999). Sherif (1936) examined the way individuals achieve collective frames of reference, social norms, and how these norms become objective stimuli. His study is paradigmatic of a Social Psychology aiming to understand the social processes generalized to small groups and large social groups such as societies or cultures. Norms are seen as social products, which cannot be reduced to the average of the individual contributes. Social influence is considered a genuine group phenomenon not an accumulation of interpersonal influences. Subsequent research did not pursue Sherif’s wide-range purpose and became more concerned with influence phenomena that emerge in small interactive groups. Traditional social influence literature is characterized by an individualistic bias because it places the individual at the center of the influence process. For instance, Moscovici (1976) remarked that, by distinguishing social reality from physical reality, Festinger’s (1950) assigns the predominance in testing reality to individuals’ personal means;