read two readings and apply ideas to your own life

Ggnggnit

roccas_and_brewer--social_identity_complexity.pdf

Home >Social Science homework help >read two readings and apply ideas to your own life

Social Identity Complexity

Sonia Roccas The Open University of Israel

Marilynn B. Brewer Department of Psychology

Ohio State University

In this article, we introduce the concept of social identity complexity—a new theoreti- cal construct that refers to an individual’s subjective representation of the interrela- tionships among his or her multiple group identities. Social identity complexity re- flects the degree of overlap perceived to exist between groups of which a person is simultaneously a member. When the overlap of multiple ingroups is perceived to be high, the individual maintains a relatively simplified identity structure whereby mem- berships in different groups converge to form a single ingroup identification. When a person acknowledges, and accepts, that memberships in multiple ingroups are not fully convergent or overlapping, the associated identity structure is both more inclu- sive and more complex. In this article, we define the concept of social identity com- plexity and discuss its possible antecedents and consequences. Results from initial studies support the prediction that social identity complexity is affected by stress and is related to personal value priorities and to tolerance of outgroup members.

Recently researchers of group processes have ex- pressed increasing interest in the fact that most individ- uals are simultaneously members of multiple social groups. Although there has been some research on the effects of crosscutting social categories on ingroup bias (see Migdal, Hewstone, & Mullen, 1998; Urban & Miller, 1998), the majority of research on social iden- tity and intergroup relations has been conducted in the context of a single ingroup–outgroup categorization. Most researchers who study social identification agree in principle that people have multiple group identities (e.g., Stryker & Statham, 1985; Tajfel, 1978; see Deaux, 1996, for a review), but there has been rela- tively little research on the nature of the relationships among a particular person’s numerous ingroup identi- ties or on the effects of holding multiple social identi- ties on intergroup attitudes in general.

The effects of multiple categorization on percep- tion of other persons has been investigated to some extent in both the social cognition literature (e.g.,

Stangor, Lynch, Duan, & Glass, 1992) and in the literature on ingroup bias in evaluations of group members (for reviews, see Crisp & Hewstone, 1999; Urban & Miller, 1998). In both of these contexts it has been found that perceivers sometimes evaluate others on the basis of one dominant categorization and ignore or even inhibit alternative categorizations (e.g., Macrae, Bodenhausen, & Milne, 1995; Roth- bart & John, 1985), sometimes evaluate others on the basis of an additive combination of the different cat- egory memberships (e.g., Brown & Turner, 1979; Hewstone, Islam, & Judd, 1993), and sometimes cre- ate a compound category with emergent properties that are not predicted from the contributing categories considered separately (e.g., Brewer, 1988; Kunda, Miller, & Claire, 1990). One thing that has not previ- ously been taken into account in trying to explain these variations in perceptions of others is the way that the perceiver represents his or her own multiple category identities. For instance, how a person who is both White and Christian responds to another individ- ual who is Black and Christian may well depend on how the perceiver defines his or her racial and reli- gious identities as ingroups. Understanding the struc- ture of multiple social identities is important because representations of one’s ingroups have effects not only on the self-concept but also on the nature of re- lationships between self and others.

This article provides an initial investigation of the concept of social identity complexity—a new theoreti- cal construct that refers to the nature of the subjective

Personality and Social Psychology Review 2002, Vol. 6, No. 2, 88–106

Preparation of this article was begun while Sonia Roccas was a postdoctoral fellow at the Ohio State University with support from the Open University of Israel and the Mershon Center of Ohio State University. The research reported in this article was funded in part by National Science Foundation Grant SBR9514398 to Marilynn Brewer and by a grant from the Open University of Israel to Sonia Roccas.

Requests for reprints should be sent to Sonia Roccas, The Open University of Israel, 16 Klausner Street, Tel Aviv 61392, Israel, or Marilynn B. Brewer, Department of Psychology, Ohio State Univer- sity, 1885 Neil Avenue, Columbus, OH 43210.

representation of multiple ingroup identities. We begin by describing alternative models for how individuals may incorporate multiple group memberships in their overall social identity and in their conceptualization of ingroups and outgroups. We then introduce the concept of social identity complexity and array these different modes of identity representation on a continuum from simple to complex. Finally, we outline a theory of ante- cedents and consequences of social identity complex- ity and present some preliminary findings on the valid- ity of this theoretical framework.

Multiple Crosscutting Group Memberships and Social

Identity Complexity

Individuals belong to multiple social groups and to groups of different types. A recent article by Lickel et al. (2000) provided a useful taxonomy of social groups that is helpful in distinguishing among different forms of so- cial identity. Their analysis identified four general types of groups—intimacy groups, task groups, social catego- ries, and loose associations—that vary along a number of dimensions. Among these, intimate groups and small task groups are characterized by interpersonal connec- tions and face-to-face interactions among group mem- bers. As such, these are likely to be groups that are based on common bonds or personal ties among all of the group members (Prentice, Miller, & Lightdale, 1994). By contrast, large task groups (organizations) and social categories are more likely to be based on symbolic at- tachment to the group as a whole rather than on the per- sonal ties that exist among specific group members— what Prentice et al. referred to as “common-identity” groups. Shared membership in such large, symbolic groups engages collective identities that are depersonal- ized (Brewer & Gardner, 1996). In the words of self-cat- egorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), social identification is a process of depersonalization “whereby people come to perceive themselves more as the interchangeable exemplars of a social category than as unique personalities” (Turner et al., 1987, p. 50). Through such collective identities, in- dividuals become connected to others by virtue of their common attachment to the group rather than their per- sonal relationships. Social identity theory applies pri- marily to these large, collective ingroup identities, and it is this type of social identity that our theory of social identity complexity is intended to represent.

Multiple Ingroup Memberships: Objective Versus Subjective Representations

The actual degree of overlap between social catego- ries of which a person is simultaneously a member may

vary considerably. Some groups may be completely embedded in others (e.g., all Catholics are Christians), some may be completely orthogonal (e.g., Muslims and women) and some may overlap only slightly (e.g., corporate executives and women). When there is ex- tensive overlap between ingroups defined by different dimensions of categorization, identification is rela- tively simple—the individuals who constitute the in- group versus outgroups are the same for any categori- zation. If almost all Mormons live in Utah and almost all residents of Utah are Mormons, for instance, then the ingroup category based on religion comprises the same individuals as the ingroup based on state resi- dence. In this hypothetical situation, for a resident of Utah all other residents are seen as ingroup members whether religion or state of residence is the basis of ingroup–outgroup categorization.

When ingroups defined by different dimensions of categorization overlap only partially, however, the im- plications for social identification become more com- plex. In this case some of those who are fellow ingroup members on one dimension are simultaneously out- group members on the other. Consider the case of a woman who is a top manager: When the social context emphasizes the professional identity (e.g., a manage- ment conference), she is likely to perceive a male col- league as an ingroup member. Nonetheless, she may be aware that in different circumstances (circumstances that emphasize her identity as a woman) that same col- league is an outgroup member. It is these situations of crosscutting group memberships—where the constitu- tion and meaning of different ingroups do not com- pletely converge—that are of interest to us here. How do individuals construct their social identities in rela- tion to multiple, nonconvergent ingroup memberships?

The actual complexity of multiple, partially over- lapping group memberships may or may not be re- flected in the individual’s subjective representation of his or her multiple identities. For instance, a woman who is both White and Christian may think of her religious ingroup as composed primarily of White people, even though objectively there are many non- White Christians. Conversely, she may think of her racial ingroup as largely Christian, despite the fact that there are many Whites who embrace other reli- gions. By reducing the subjective inclusiveness of both ingroups to their overlapping memberships, the individual maintains a relatively simplified identity structure. When an individual acknowledges, and ac- cepts, the nonoverlapping memberships of her multi- ple ingroups, her subjective identity structure is both more inclusive and more complex.

When group identities do not converge, there are different ways in which the individual may structure his or her perception of the ingroups to reconcile the potentially competing implications for defining the so- cial self. Here we specify four alternative forms of

IDENTITY COMPLEXITY

identity structure that reflect different ways in which the relationships among multiple ingroups can be sub- jectively represented. The four structures are repre- sented schematically in Figure 1 for the case of two so- cial category identities (A and B).1 Each model of ingroup representation has implications for the inclu- sion or exclusion of others as members of the subjec- tive ingroup. This provides a link between representa- tions of multiple ingroups and different patterns of crossed categorization that have been identified in the intergroup relations literature (Hewstone et al., 1993; Urban & Miller, 1998).

Intersection. One way that an individual can achieve simultaneous recognition of more than one social identity and yet maintain a single ingroup rep- resentation is to define the ingroup as the intersection of multiple group memberships (Figure 1a). For in- stance, a female lawyer can define her primary social identity in terms of the compound combination of both sex and profession, an identity shared only with other women lawyers. In this representation, the com- pound category is a single, unique social identity with properties that make it distinct from either of the larger categories from which it is derived.

In this mode, multiple bases of group identifica- tion converge on a single social identity with one consolidated ingroup. Those who do not share the joint identities (e.g., lawyers who are male or women who are not lawyers) are outgroup members. Thus, the intersection model of ingroup representation cor- responds to the “conjunction/dissimilarity” pattern of responding to multiply categorized others (Urban & Miller, 1998) or what Hewstone et al. (1993) referred to as the “social exclusion” pattern. Only the con- junction of two group identities constitutes the per- ceiver’s ingroup; any other combination of category memberships are treated as outgroups.

Dominance. Another way that individuals can cope with competing social identities is to adopt one primary group identification to which all other poten- tial group identities are subordinated (Figure 1b). In this model, the ingroup is defined as those who share membership in this primary ingroup category; all other category memberships are essentially not social identities at all but simply aspects of the self as a member of the primary group. In other words, alter- native social identities are embedded within the pri- mary group identification (as sources of intragroup variation) but not extended to those outside that ingroup. For instance, a female lawyer who assigns primacy to her professional identity regards all law- yers as fellow ingroup members. Being a woman (or sailor, or Yale Law School graduate, etc.) is a charac- teristic that describes what kind of a lawyer she is, what makes her more or less similar to others in her ingroup category (and to the category prototype), but her social identity is not extended to women or Yale graduates in general.2

When one social identity takes precedence over all others, the individual should classify other people in terms of their membership in that one category. Those who share the dominant category membership are treated as ingroup members; those who are not in the category are outgroupers. This corresponds to the “category dominance” pattern of relating to multiply categorized others (Urban & Miller, 1998). An alter- native to complete dominance would be the “hierar- chical” pattern of multiple categorization (Brewer, Ho, Lee & Miller, 1987). In this form, anyone outside of the dominant category is treated as an outgroup member regardless of any other identities they may have. However, within the ingroup category further differentiations may be made on the basis of other shared identities. Thus, the woman lawyer may feel closer to other lawyers who are female than to those who are male, but she is still more identified with male lawyers than with females who are not a part of her profession.

Compartmentalization. If more than one group identity is important to an individual as a source of social identity, multiple identities can be activated and ex- pressed through a process of differentiation and isolation (Figure 1c). With compartmentalization, social identities are context specific or situation specific. In certain con- texts, one group membership becomes the primary basis of social identity, whereas other group identities become

ROCCAS & BREWER

1The model is not intended to be limited to consideration of only two identities at a time and can be extended to multiple cate- gory memberships.

Figure 1. Alternative structures of multiple ingroup represen- tations.

2Note that we are dealing here with the way in which the individ- ual herself conceptualizes her primary social identities and ingroups. An interesting extension of these ideas would consider the implica- tions of discrepancies between the way an individual defines his or her own social identity and how he or she is categorized by others in a given situation.

primary in different contexts. At the office, for instance, one’s professional identity may be the only relevant basis for ingroup–outgroup distinctions; shared identities based on sex, ethnicity, religion, or recreational group memberships are irrelevant and not activated in this set- ting. In the home, however, religious affiliation or cultural group membership may become the most important basis for shared identity and the social self. With this mode of identity structure, multiple nonconvergent identities are maintained, but the individual does not activate these so- cial identities simultaneously.

Context specificity is emphasized in Turner et al.’s (1987; Turner, Oakes, Haslam, & McGarty, 1994) self-categorization model of social identity in which social identities are presumed to be context depend- ent and essentially mutually exclusive. If a single so- cial identity is salient in a specific setting, then one would expect that social categorization to dominate both self-categorization and categorization of others as ingroup members or outgroup members within that situation. However, there may be situations in which more than one categorization is relevant and salient (e.g., both gender and occupation may be equally sa- lient in a work context). In that case, social identity theory predicts an additive pattern of evaluation of others as a function of their multiple group member- ships (Brown & Turner, 1979). That is, in- group–outgroup categorizations are combined addi- tively when responding to other individuals. Those who share both ingroup identities with the self are evaluated more positively than those who share only one common ingroup membership, who in turn are more positively evaluated than those who are outgroup members on both dimensions.

Merger. The final model for representation of multiple social group identities is one in which non- convergent group memberships are simultaneously rec- ognized and embraced in their most inclusive form (Figure 1d). In this mode, ingroup identification is ex- tended to others who share any of one’s important so- cial category memberships—social identity is the sum of one’s combined group identifications. For the fe- male lawyer, her identification with women as a social group crosses the boundary of lawyer and nonlawyer, and her identification with lawyers crosses the sex di- vide; therefore, both identity groups are important and salient across situations.

As the combination of social identities involving different constituents and different defining charac- teristics, such a merged ingroup identity is necessar- ily highly inclusive and diverse. In this mode, social identity transcends single categorical divisions be- tween people. Thus, the merger model goes beyond additivity of multiple ingroup memberships to what Urban and Miller (1998) referred to as the “equiv- alence pattern” of evaluating others with multiple

group memberships. The more social identities the individual has, the more inclusive the definition of ingroup becomes, to the point where no sharp in- group–outgroup distinctions are made on any dimen- sion and all others are evaluated equivalently.

Cognitive Consistency Resolution and Social Identity Complexity

There are parallels between the cognitive and mo- tivational processes involved in reconciling multiple group identities and the processes studied by cog- nitive consistency theorists (Abelson et al., 1968). More specifically, the four models of identity repre- sentation we identified previously correspond roughly to different modes of resolving inconsistency be- tween two incompatible beliefs or attitudes, as out- lined by Abelson (1959) and Kelman and Baron (1968). Intersection is similar to a form of cognitive differentiation in which the compatible elements of two cognitions are separated out and dissociated from the inconsistent elements (e.g., the smoker who deals with the potential inconsistency between smoking and health by focusing on the health benefits of smoking in preventing weight gain while denying the potential health risks associated with smoking). Dom- inance is analogous to the mechanism of “bolster- ing”—augmenting the strength and commitment to one cognition over the other. Compartmentalization is parallel to the mechanism of cognitive isolation and compartmentalizing in cognitive structures. Finally, merger is analogous to “transcendence,” the introduc- tion of some superordinate principle that makes the inconsistent cognitions compatible.

These different modes of inconsistency resolution and identity representation can be arrayed on a dimen- sion of cognitive complexity as defined by Tetlock (1983). Cognitive complexity is characterized by both differentiation and integration of potentially conflict- ing beliefs and values. The level of differentiation re- flects the degree to which inconsistencies are recog- nized (rather than denied or suppressed); integration reflects the level of resolution or reconciliation be- tween recognized inconsistencies.

Drawing on Tetlock’s (1983) conceptualization of levels of cognitive complexity, our theory orders the four modes of representation of multiple social identi- ties along a continuum from least to most complex. In- tersection is the least complex form because it reduces multiple, potentially diverse, group identities to a sin- gle, highly exclusive social identity. Dominance is also on the low-complexity end of the continuum because it suppresses inconsistencies within a single ingroup– outgroup dichotomization. Compartmentalization rep- resents the next level of complexity in that separate identities are acknowledged and differentiated, but without any attempt at reconciliation. Merger repre-

IDENTITY COMPLEXITY

sents the highest level of complexity because it pre- serves both differentiation and integration in an inclu- sive social identity.

Arraying the modes of identity representation along a continuum acknowledges that the specified types may differ more in degree than in kind. Domi- nance represents a somewhat more complex and in- clusive mode than intersection if the dominant group identity is a relatively large, heterogeneous social cat- egory (e.g., women) compared to a homogeneous compound categorization (e.g., WASP). However, if a dominant social identity is a relatively small exclu- sive group (e.g., a specific profession, a religious sect, or the like), then dominance is very similar to intersection in terms of complexity and inclusiveness. In either case, both dominance and intersection pro- vide a resolution that divides the social world along a single ingroup–outgroup category distinction.

Compartmentalization may also resemble domi- nance or intersection. Compartmentalized identities may themselves be compound social groups, for exam- ple, the person who identifies primarily as a woman law- yer at work but as a Catholic Republican in her family and social life. When specific group identities are very highly compartmentalized, the representation of ingroups is essentially a context-specific form of domi- nance. On the other hand, if compartmentalization is weak (i.e., different social identities “spill over” into dif- ferent situations or settings), then differentiated identi- ties may approach representation as merged identities. Finally, when merger is achieved by integrating multi- ple social identities into a single highly inclusive group identity (e.g., citizen of the world), the resolution resem- bles dominance, albeit with a highly complex and differ- entiated primary identity.

Further, individuals may adopt different modes of identity representation at different times, either during different periods of life or under different conditions or mental or emotional states. For example, a person who generally maintains a unified merged representation of his multiple social identities may resort to intersection, dominance, or compartmentalization in times of stress or insecurity. Thus, the subjective representation of multiple identities may reflect both individual differ- ences and situational factors.

Biculturalism: An Illustration

Illustrations of these alternative representations of multiple group identities can be drawn from the liter- ature on acculturation and bicultural identity (Berry, 1990, 1997; Berry & Sam, 1997; Birman, 1994; Phinney, 1990; Phinney & Devich-Navarro, 1997). The prototypic case here is the member of an immi- grant group or an ethnic enclave whose societal group membership (country of residence or citizenship) and ethnic–national group membership represent distinct

cultures and overlapping but nonconvergent social groups. As different cultures, the two group identities may represent different and sometimes conflicting sets of norms and values; as different constituencies they may place competing demands on individual loyalty and resources.

As identified in clinical case studies and empirical research on acculturating or adapting to a bicultural environment, individuals cope with the demands of competing cultural identities by adopting different forms of identity management (S. E. Cross, 1995; W. E. Cross, 1991; Sellers, Smith, Shelton, Rowley, & Chavous, 1998).

Hyphenated identities. One form of bicultural adaptation is to locate one’s cultural identity at the in- tersection of the ethnic and societal levels and thus form a blended bicultural identity (Birman, 1994; Phinney & Devich-Navarro, 1997). In this model, the ingroup is defined exclusively as those who share both ethnic heritage and residence in the host society. With this representation, terms such as African American, Latin American, and Korean American connote more than the combination of two separate group memberships; they represent unique cultural configurations derived from the specific experiences of enacting a particular ethnic-cultural identity within the American context. Such hyphenated identities are often associated with heightened ethnic conscious- ness and a type of nationalist ideology within the po- litical arena (Sellers et al., 1998). Ethnic politics rests on the assumption that the interests and needs of eth- nic subcultures can be understood and represented only by those who share the specific ingroup identity.

Cultural dominance. Another mode of coping with alternative cultural identities involves subordinat- ing one identity to the other. Assimilation to the host culture at the expense of ethnic cultural identity is one form of cultural dominance (Berry, 1990); at the other extreme is exclusive investment in one’s ethnic cultural identity with alienation from the culture of the host so- ciety—the separation strategy (Berry, 1990; Phinney & Devich-Navarro, 1997).

Compartmentalization. A third mode of adapt- ing to perceived conflict between alternative cultural group identifications is a kind of cultural “ambidex- trousness” in which the individual consciously activates different cultural identities in different contexts or so- cial settings—a pattern referred to as “alternating bi- culturalism” (LaFromboise, Coleman, & Gerton, 1993; Phinney & Devich-Navarro, 1997). This type of identity structure is illustrated by children of immigrant parents who alternate between the language used at home and the one used in the community (e.g., Harris, 1995) but is extended to other cultural practices, norms, and values

ROCCAS & BREWER

as well. Individuals who adopt this strategy have a sense of competence in both cultures (LaFromboise et al., 1993) but also an awareness of conflict between cultures that renders biculturalism sometimes problematic (Phinney & Devich-Navarro, 1997).

Integrated biculturalism. A fourth form of bi- culturalism is the concept of intercultural identity (Sussman, 2000). Unlike compartmentalization, in which the different cultures are experienced as basi- cally incompatible and situation specific, this form of biculturalism acknowledges multiple cultural identi- ties simultaneously—where membership, values, and norms of both groups are combined and integrated (Oyserman, Sakamoto, & Lauffer, 1998). It reflects a “humanist ideology” (Sellers et al., 1998) that may correspond to the internationalization of identity de- velopment in Cross’s (1991) model of racial identity.

Intercultural identity represents, according to Suss- man (2000), a “global identity shift … neither the inte- gration of home and host culture values (hybridization) nor the bicultural strategy which results from accultur- ation experience but rather an identity in which the (in- dividuals) define themselves as world citizens” (p. 368). This conceptualization clearly equates multicul- turalism with the acquisition of a more inclusive com- plex group identity than that represented by any com- ponent cultural identity alone.

Cognitive Structure and Processes: Underlying Assumptions of the Social

Identity Complexity Model

Like Tetlock (1983, 1986), we assume that greater levels of integrative complexity require effort-demand- ing cognitive strategies and resources. Social identity complexity is the product of a process of recognizing and interpreting information about one’s own ingroups. Having a complex social identity is dependent on two conditions: first, awareness of more than one ingroup categorization and second, recognition that the multiple ingroup categories do not converge. Reconciling the incongruences that are implied by this nonconvergence requires cognitive resources. Thus, like other forms of integrative complexity (Tetlock, Skitka, & Boettger, 1989; Woike & Aronoff, 1992), social identity complex- ity is subject to situational and motivational determi- nants as well as individual differences in cognitive style (Schroder, Driver, & Streufert, 1967).

Some individuals may be chronically high in social identity complexity. For such persons, integrative com- plexity in thinking about multiple ingroup identities may become automatized, requiring relatively little conscious effort or cognitive resources. In most cases, however, social identity complexity can be expected to vary as a function of the individual’s current motiva-

tion to think about his or her multiple ingroup identities and available cognitive resources to merge these iden- tities in an inclusive manner.

Integrative complexity is domain specific. Thus, individuals who have complex representations in one domain of the self-concept do not necessarily have representations of similar level of complexity in other domains. For example, intellectual self-complexity is distinct from religious self-complexity (Nielsen & Fultz, 1997), the complexity of the representation of the actual self is distinct from the complexity of the representation of the future self (Niedenthal, Setter- lund, & Wherry, 1992), and positive self-complexity is distinct from negative self-complexity. Moreover, the complexities of representations in different do- mains have distinctive consequences (Morgan & Janoff-Bulman, 1994).

Our conceptualization of social identity complexity derives from models of category representations. On one hand, social categories can be represented by the prototypical attributes of their members. Prototypes are abstracted representations of the central tendency, average or typical values of the members of a category (Smith, 1998). Based on this aspect of category repre- sentation, one way in which relationships between the ingroups of which a person is simultaneously a mem- ber can be conceptualized is the extent of perceived similarity between the prototypical attributes of those groups. On the other hand, social categories can also be represented in terms of category boundaries that deter- mine who are considered group members. In line with this type of model, another possible representation of the interrelationships between ingroups of which a per- son is simultaneously a member is the degree of per- ceived overlap between the memberships of the vari- ous ingroups.

Social Identity Complexity: Toward an Operational Definition

Although complexity of social identities may vary both within and between the four types of cognitive repre- sentation we have distinguished here, intersection and dominance can, in general, be classified as relatively low-complexity representations and compartmentaliza- tion and merger as relatively high complexity. Low com- plexity means that multiple identities are subjectively em- bedded in a single ingroup representation, whereas high complexity involves acknowledgment of differentiation and difference between ingroup categories.

More specifically, complexity involves understand- ing what people mean when they say that I am both “A” and “B.” Semantics make it difficult to distinguish whether the “and” in this self-description refers to the intersection of A and B (converging or embedded iden- tities), or the union of A and B as two ingroups. Be-

IDENTITY COMPLEXITY

cause of this semantic ambiguity, attempts to assess this distinction through direct self-report are likely to be met by confusion or social desirability biases. More indirectly, we can attempt to get at the subjective repre- sentation by assessing how the individual perceives the degree of overlap between different ingroup identities in terms of meaning and membership. Through such indirect assessment, we can attempt to identify where an individual falls between the extremes of identity convergence or identity complexity when two or more ingroup memberships are made salient.

For the purposes of operationalization, we focus on two aspects of the perceived relationships among in- groups that reflect the two aspects of category represen- tation described earlier. One aspect, based on prototype representation, is the perceived similarity of prototyp- ical values, norms, and other characteristics that define various ingroups. The second aspect, based on member- ship representation, is the perceived extent of shared membership across the different groups. These two as- pects give rise to two different manifestations of subjec- tive overlap between ingroups: overlap of group charac- teristics (the content of the prototypic representations of the respective groups) and overlap of group members (group composition). For either measure, a high degree of overlap signals a convergent representation of the ingroups as a single social identity; a low degree of per- ceived overlap reflects a more complex inclusive social identity. However, because these measures reflect dif- ferent aspects of ingroup representation (prototype and boundary definition), they represent two distinct types of complexity.

Shared Characteristics Across Multiple Ingroups

Individuals differ in the extent to which they perceive the prototypes of the groups of which they are simulta- neously members as similar to each other and featuring the same characteristics. The groups male and Ameri- can could serve as an example. Although these two groups are objectively orthogonal (only half of the Americans are males), they could be perceived as over- lapping in their values, norms, and other characteristics: Thus, a male American could conceptualize Americans as being characterized by prototypically male attributes such as physical courage and autonomy. To the extent that an individual sees his or her ingroups as highly simi- lar to each other and as sharing the same basic attributes and values, the different group identities are highly com- patible, and it is easy to conceive of the multiple group memberships as a single, relatively homogeneous in- group. In sum, the more a person perceives the groups to which he or she belongs as being similar to each other, the less complex is his or her social identity.

This operationalization of social identity complex- ity resembles a number of other identity constructs.

One construct with obvious relevance is that of self- complexity (Linville, 1985, 1987). Linville (1985, 1987) proposed that individuals differ in the complex- ity of the structure of their self-knowledge: A simple self-concept is represented in terms of a small number of self-aspects with overlapping features and attrib- utes. In contrast, a complex self-concept consists of a large number of aspects with independent features. So- cial identity complexity is similar to self-complexity in that both concepts describe the interrelations among aspects of self-knowledge. However, they differ in the type of knowledge at their focus: Self-complexity re- fers to the structure of the perception of personal attrib- utes, whereas social identity complexity refers to the structure of the perception of ingroups. Thus, when self-complexity is measured, individuals report their perception of their own personal attributes. In contrast, when social identity complexity is measured, individu- als report their perception of the prototypes of the groups to which they belong. Nonetheless, the two constructs represent analogous concepts at different levels of representation of the self—the personal and the collective selves (Brewer & Gardner, 1996).

Social identity complexity also partially overlaps with role identity concepts. According to role identity theory (Stryker, 1980; Stryker & Serpe, 1994; Stryker & Statham, 1985), each individual occupies multiple roles arranged hierarchically in order of salience. Role identity is the self-perception of individuals in a particular social position in relation to others who oc- cupy different roles within the same system. Role identity theory overlaps with social identity complex- ity in its interest in the consequences of the structural relations between the multiple roles that individuals occupy simultaneously within different social groups or contexts (e.g., Milkie & Peltola, 1999; Voydanoff & Donnelly, 1999). Because roles are relational iden- tities, role complexity derives from the multiple rep- resentations of self in relationships with others— what Brewer & Gardner (1996) referred to as the “in- terpersonal self”—whereas social identity complexity derives from the cognitive representation of one’s ingroups as a whole: the collective self.

Shared Membership Across Multiple Ingroups

Whereas similarity measures of social identity com- plexity (like self-complexity and role identity) assess identity in terms of the content of group representa- tions, a second manifestation of social identity com- plexity is the perceived overlap in the composition of group memberships. Some persons may perceive the different groups to which they belong as containing the same members. The groups Catholic and Italian could serve as an example. Although these two groups do not objectively share all of their members (many Italians

ROCCAS & BREWER

are not Catholic, and many Catholics are not Italian), some people may perceive them as highly overlapping: When they think about Italians they think about Catho- lics, and persons of different religious faith are not con- sidered “real” Italians.

As with overlap in group characteristics, high per- ceived overlap in group memberships implies that the different ingroups are actually conceived as a single convergent social identity. In this case, the subject boundaries of both ingroups are defined in such a way that they contain only those who share the other iden- tity as well. On the other hand, when overlapping membership between various ingroups is perceived to be relatively small, the boundaries of each ingroup are defined in such a way that they include members who do not share the other identities. In this case, the com- bined group identities are larger and more inclusive than any of the ingroups alone. In sum, the more a per- son perceives the groups to which he or she belongs as sharing the same members, the less complex is his or her social identity.

Together, these two dimensions—overlap of group prototypes and overlap of group membership—provide two indirect indexes of whether multiple ingroups are perceived in terms of a single convergent social iden- tity or as the union of different group identities. Be- cause content and composition of ingroup representa- tions are distinct properties of identities, the two indexes are not necessarily expected to be highly cor- related, and each would be expected to be only imper- fectly related to the complexity and inclusiveness of the individual’s subjective representation of ingroup identities. For instance, a respondent may recognize that two of his group memberships have very different prototypes (e.g., the typical male is different from the typical nurse), indicative of a complex, nonoverlapping identity representation. Yet the same respondent may

subjectively identify with an atypical subtype of both categories (male nurses), indicative of a simple, exclu- sive ingroup identity. Similarly, a respondent may rec- ognize that two of his or her groups do not share a high overlapping membership (e.g., he or she knows objectively that all Americans are not Christians), but still subjectively represent his or her identities as highly overlapping (e.g, “my” type of Americans are Christians). Nonetheless, across a relatively large num- ber and range of different group identities, high over- lap on either index connotes a relatively simple amal- gamation of multiple group identities, whereas low overlap implies a more complex social identity.

Antecedents of Social Identity Complexity

In this section we present the underlying cognitive model that led to specific predictions regarding fac- tors affecting identity complexity. We focus on those factors that are particularly relevant to the complexity of representations of one’s ingroups. Social identity complexity is determined by experiential and motiva- tional factors that influence the availability and acces- sibility of group representations that distinguish one ingroup from another and the cognitive resources available to integrate those multiple representations (cf. Higgins, 1996).

Figure 2 provides a summary representation of fac- tors that might affect the accessibility (both chronic and temporary) of more or less complex representa- tions of multiple group identities. As depicted in the figure, accessibility of a complex representation of so- cial identities is expected to be influenced by three types of factors. First we propose that the way individ- uals represent their ingroups is affected by the com-

IDENTITY COMPLEXITY

Figure 2. Hypothesized antecedents of social identity complexity.

plexity of their social experiences. In addition, we expect that social identity complexity is affected by stable individual differences in the motivation to attend to complex information. Finally, we expect that social identity complexity is affected by situational factors that temporarily affect attentional resources. Thus, we expect that the first two types of factors will cause sta- ble individual difference in the complexity of represen- tations of ingroups, whereas the third type of factors will cause temporary changes.

Complexity of Social Experiences

Most of the time, individuals are surrounded by oth- ers who are similar to themselves (Kelley & Evans, 1995). We are first exposed to our family members who naturally belong to the same race, religion, and socioeconomic status as ourselves. Youngsters go to school with children who live in the same neighbor- hood, and consequently homogeneity of the immediate social environment is maintained, albeit to a lesser de- gree. The immediate social environment within which most people are socialized is objectively less complex than the society as a whole. Thus, the local social struc- ture encourages the perception of relatively high simi- larity and overlap between ingroups, leading to a rela- tively simple social identity. To develop a complex social identity, special conditions are necessary—con- ditions that enhance the simultaneous awareness of more than one ingroup and the awareness that these ingroups overlap only partially.

The most obvious factor that may affect social iden- tity complexity is the actual complexity of the experi- enced social environment. Social environments in which different bases for ingroup–outgroup distinctions are crosscutting rather than convergent confront the indi- vidual with knowledge about the differences in meaning and composition of different social categorizations.

Living in a multicultural society, for instance, may enhance awareness that social categorization based on ethnic heritage and social categorization based on na- tional citizenship do not completely overlap and hence raises social identity complexity. Note however that, as our previous discussion of biculturalism suggests, liv- ing in a diverse, multicultural society may not always be sufficient to provide the conditions for complex identity formation. The impact of a multicultural environment may be experienced differently for different partici- pants depending on their actual exposure to diversity. In addition, the effects of a multicultural society on social identity complexity are likely to depend on the extent to which the society is stratified along ethnic lines and on prevailing norms concerning multiculturalism.

Living in a stratified society, especially when there is a high degree of congruence between status and ethnic group membership, may reduce experienced complex- ity even when the society is multicultural. In a society in

which members of different religious or ethnic groups engage in different occupations, for example, there is high objective overlap between the occupational and the ethnic group, and thus low social identity complexity can be expected. Moreover, in stratified societies pri- mary groups are usually composed of individuals who are members of the same ethnic group or social class, and thus contact with individuals who could be simul- taneously ingroup and outgroup members is minimal (Kelley & Evans, 1995; Massey & Denton, 1989).

Furthermore, the effect of living in a multicultural society on social identity complexity is also likely to be moderated by societal norms concerning multicultural- ism. When people of many cultural backgrounds live together, the cultural groups they form are often not equal in power. Accordingly, some groups are domi- nant, and their ideology may have extensive influence both on the actual diversity and on the perception of di- versity of the whole society. Some dominant groups are explicitly assimilationist and hold an ideology that promotes a single culture in the nation, whereas others are integrationist and explicitly encourage the mainte- nance of the cultural heritage of nondominant groups (Berry, 1997). It is likely that integrationist ideology enhances the social identity complexity of members of the dominant group because it encourages the various ethnocultural groups to express their diversity and raise its salience. Thus, when integrationist norms pre- vail, members of the dominant group are more likely to be aware of nonoverlap between their ethnic or racial group and the other groups to which they belong.

Of particular importance in determining social iden- tity complexity is the actual overlap and similarity be- tween one’s own ingroups. A person who is a member of groups that are highly similar in their attributes or have highly overlapping members is likely to have a simple representation of the interrelations between those groups. Thus, we suggest that simultaneous member- ship in groups that are similar or overlapping will result in low social identity complexity. However, being a member of groups that differ markedly in their objective attributes does not necessarily lead to a complex repre- sentation. Large differences between the characteristics of one’s own ingroups could result in low complexity if one ingroup dominates the representation of the interre- lationships between the various ingroups. This can be exemplified both with regard to the relative distinctive- ness of one’s various ingroups and their status: Distinc- tive ingroups are generally more salient and are more of- ten used for self-categorization (McGuire & McGuire, 1988; McGuire, McGuire, Child, & Fujioka, 1978). Thus, simultaneous membership in groups that vary in their distinctiveness may result in a representation of the social identity that is dominated by the most distinctive group, and other self-categorizations may be subservi- ent to it. Research on patterns of identities of immigrants supports this reasoning. For example, immigrants living

ROCCAS & BREWER

in neighborhoods where there are fewer members of eth- nic minorities are more likely to use ethnicity rather than other categories as a basis of social comparison (Mog- haddam, 1992).

A similar line of reasoning applies to differences in the relative status of ingroups. When there are status dif- ferences between ingroups, self-representations may be dominated by the highest status group. Thus, individu- als identify more with a specific high-status ingroup when they are simultaneously members of a low-status group than when they are simultaneously members of another high-status group (Roccas, 2001). In sum, when one’s ingroups are similar in their attributes and share their members, a simple representation of the social identity is likely. When the various ingroups differ from each other, however, two opposing outcomes may re- sult: either a complex representation that reflects the dif- ferences between the ingroups or a simple representa- tion, which is dominated by one of the ingroups.

Personal Attributes: Tolerance for Ambiguity

The chronic accessibility of the complexity of the structure of the groups of which one is member will also be affected by personal attributes. The most rele- vant factor is tolerance for ambiguity because a com- plex representation of the ingroups provides a less clear cut representation of the social world than a sim- ple representation. Individual differences in tolerance for ambiguity have been studied in terms of cognitive motives and values.

Cognitive Motives and Social Identity Complexity

There are several constructs that assess individual differences in the preference for confronting complex ambiguous information. These include need for closure (Kruglanski, 1990), uncertainty orientation (Sorrentino & Roney, 1999), personal need for structure (Neuberg & Newsome, 1993), and tolerance for ambiguity (Norton, 1974). These measures are likely to assess, at least in part, overlapping individual differences (e.g., Leone, Wallace, & Modglin, 1999; Neuberg, Judice, & West, 1997). They all have in common the need to create and maintain a simple structure and thus are expected to be related to the complexity of the representations of the in- terrelationships between ingroups. Here we exemplify the possible effects of cognitive motives on social iden- tity complexity by discussing two of these needs: the need for closure and uncertainty orientation.

Need for closure. Need for closure is a desire for a quick and definitive answer to any question or deci- sion rather than sustained uncertainty, confusion, or ambiguity (Kruglanski, 1990). The need for closure in-

duces a tendency to seek immediate and permanent an- swers. That is, individuals with a high need for closure seek answers urgently and strive for stable answers that avoid the necessity of future revisions and the uncer- tainty and ambiguity that revision process may entail.

A complex social identity lessens the possibility of obtaining firm and unequivocal answers that pertain to group membership of self and of others. When there is low overlap between ingroups, another individual may be simultaneously an ingroup member and an outgroup member. Consequently, there is no definite answer to questions such as “Is this person one of ‘us’ or one of ‘them’?” Similarly, when ingroups are perceived as highly dissimilar from each other, there is no definite answer to questions such as “Is this a ‘good’ group member?” When ingroups are characterized by differ- ent values and norms, the same individual may be con- sidered a good member of one group and an unsatisfac- tory member of another ingroup. Therefore, it is likely that individuals with high need for closure prefer to perceive their ingroups as similar to each other and as sharing their members and are thus likely to have rela- tively low social identity complexity.

The relationship between need for closure and social identity complexity has not been empirically examined directly. However, findings from a number of studies in- dicate that need for closure enhances the desire for a homogeneous social environment (Kruglanski & Web- ster, 1991; Kruglanski, Webster, & Klem, 1993). For instance, Kruglanski and Webster (1991) found that individuals in small groups placed under conditions ac- tivating high need for closure were more likely to reject a confederate who professed to hold a deviate opinion. Similarly, Kruglanski et al. (1993) found that individu- als with a high need for closure exhibited a stronger pref- erence for agreement with their dyadic partners relative to individuals with a low need for closure. Further, the need for closure has been demonstrated to be related to degree of ingroup favoritism and outgroup derogation (Shah, Kruglanski, & Thompson, 1998), supporting the idea that low tolerance for ambiguity is associated with clear ingroup–outgroup categorization.

Uncertainty orientation. Like need for closure, the concept of uncertainty orientation refers to indi- vidual differences in regulation of uncertainty and ambiguity (Sorrentino & Roney, 1999). At one ex- treme of the continuum are individuals who are cer- tainty oriented, who develop a style that actively avoids confronting uncertainty; at the other extreme, uncertainty-oriented individuals handle uncertainty by seeking out information aimed at confronting and resolving uncertainty or confusion. Based on this def- inition, we would expect certainty-oriented persons to avoid recognition of nonconvergent group identities and to represent multiple social identities as a single (joint) ingroup. On the other hand, uncertainty-ori-

IDENTITY COMPLEXITY

ented individuals should be more willing to seek out information about the self and the social environment that would lead to high social identity complexity.

Consistent with our hypothesis, uncertainty orienta- tion has been found to play a role in social categoriza- tion processes (Sorrentino, Hodson, & Huber, 2001). Certainty-oriented individuals, for instance, perceive their ingroups as more homogeneous than do uncer- tainty-oriented individuals, especially in intragroup contexts. Further, certainty-oriented individuals ex- hibit stronger ingroup bias under conditions of uncer- tainty. Both of these findings suggest that persons who are certainty-oriented use ingroup–outgroup distinc- tions to maintain clarity and avoid uncertainty, whereas uncertainty-oriented people are more tolerant of un- clear ingroup boundaries.

Values and Social Identity Complexity

The same line of reasoning that links social identity complexity with cognitive motives applies to its relation with personal value priorities. Values are cognitive so- cial representations of basic motivational goals, varying in importance, which serve as guiding principles in people’s lives (Kluckhohn, 1951; Rokeach, 1973; Schwartz, 1992). Based on universal requirements of human existence, Schwartz (1992, 1994, 2000; Schwartz & Sagiv, 1995) identified 10 motivationally distinct types of values and verified these distinct types in cross-cultural research in more than 60 countries.

The 10 value types form a circular structure that can be summarized into two basic conflicts.

Self-enhancement versus self-transcendence. Power and achievement value types are in conflict with universalism and benevolence value types. Both of the former emphasize pursuit of self-interests, whereas both of the latter involve concern for the welfare and interests of others.

Openness to change versus conservatism. Self- direction and stimulation value types are in conflict with security, conformity, and tradition value types. Both of the former emphasize independent action, thought and feeling, and readiness for new experience, whereas all of the latter emphasize self-restriction, order, and resis- tance to change.

Values affect the way in which individuals per- ceive and interpret information (Schwartz, Sagiv, & Boehnke, 2000). Hence values may affect the percep- tion of ingroups and the awareness of nonoverlap be- tween the various groups of which an individual is simultaneously a member. The value types most like- ly to affect accessibility of a complex social structure are conservatism and openness to change. Conserva- tism values express the motivation to avoid uncer- tainty, ambiguity, and instability. Individuals who

emphasize these values are motivated to perceive the environment in simple terms. Awareness that in- groups do not overlap and are dissimilar from each other contrasts with the motivation to avoid ambigu- ity and can evoke a sense of instability and insecurity. Hence, individuals who value conservatism are likely to develop a simple social identity. Conversely, those who emphasize openness to change are motivated to learn and explore and to make independent judg- ments based on their own experience. They are less likely to accept prevailing social norms and more likely to develop a complex social identity. We also expected that social identity complexity would be re- lated to the importance placed on universalism versus power values. Universalism values emphasize toler- ance and understanding of all people and thus express easiness with unclear ingroup boundaries. In contrast, power values express the motivation to legitimize hi- erarchical relations and to strive to control people and resources. Clear ingroup boundaries help to promote the pursuit of these values because they clarify hier- archical relations between individuals and groups. Thus, individuals who emphasize power values are likely to develop a simple social identity with clear distinction between ingroup and outgroup members.

Situational Factors

The socialization experiences and the stable individ- ual differences we have discussed produce long-term ef- fects that are likely to determine chronic accessibility of complex representations of multiple social identities. Above and beyond these chronic variables, situational factors that affect salience of specific ingroup identities and cognitive capacity are also likely to influence social identity complexity at least temporarily.

Distinctiveness. Individuals tend to pay particu- lar attention to attributes that make them distinctive. Thus, individuals shift their basis of categorization from context to context and tend to use the ingroup that is distinctive in each specific context (McGuire & McGuire, 1988). For example, the probability that a re- spondent describes himself or herself in terms of his or her gender is greater the higher proportion of people of the opposite gender present at the moment (Cota & Dion, 1986). Thus, we expect that in situations in which a specific ingroup is particularly distinctive, that group will play a dominant role in the representation of the social identity, resulting in a simple representation of the interrelationships among the various ingroups.

Cognitive load. Situational demands that place a heavy load on attention capacities, such as performing multiple tasks concurrently, usually have detrimental effects on information processing, retrieval, and analy- sis (e.g. Conway, Carroll, Pushkar, & Arbuckle, 1996;

ROCCAS & BREWER

Osterhouse & Brock, 1970; Petty, Wells, & Brock, 1976). Individuals are not usually constantly aware of all of their group memberships: Usually they are most aware of the categories that render the social context subjectively most meaningful—the social categories in which there is most similarity within groups and maxi- mum distinctiveness between groups (Oakes & Turner, 1990; Turner et al., 1987). Moreover, it is likely that the overlap between ingroups is more chronically accessi- ble than nonoverlap because individuals are usually surrounded by others who share their same ingroups. Thus, the awareness of complex social identities may require greater cognitive effort and attention than more simple ingroup representations. As a consequence, cognitive overload may affect the accessibility of in- formation that contributes to a complex social identity (simultaneous awareness of multiple ingroups and awareness of nonoverlap between them) more than the accessibility of information pertaining to simple social identity (awareness of only one ingroup or perception that all ingroups overlap extensively), resulting in a temporary reduction of social identity complexity.

Stress. A similar line of reasoning applies to the effects of acute stress on social identity complexity. Considerable research indicates that anxious people ap- pear to have diminished cognitive resources (for a re- view, see Wegner & Wenzlaff, 1996). Anxiety is espe- cially detrimental to performance on effortful tasks but has little effects on easy tasks or automatic ones (Ingram & Kendall, 1987; Kahneman, 1973; for a review, see Eysenck, 1992). In addition, extensive research indi- cates that stress causes the narrowing of attentional fo- cus (e.g., Neufeld & McCarty, 1994; Salmela & Ndoye, 1986; Steblay, 1992; but see Heuer & Reisberg, 1990, for a different view). Under stress, individuals often fo- cus on the central features of stimuli and neglect periph- eral characteristics. This could affect both aspects of so- cial identity complexity in that individuals under stress will tend to perceive their groups as largely overlapping and largely similar. These predictions are specific to negative moods that deplete cognitive processing. Thus, they could be extended to anger but not necessarily to sadness or happiness (Fiske, 1998).

Ingroup threat. A threat to an ingroup is likely to affect the representations of social identity through mul- tiple processes. First, it is likely to raise the salience of the threatened ingroup relative to that of other ingroups. Thus, the threatened ingroup may temporarily dominate social identity, and membership in the other ingroups becomes both less important and less differentiated. In addition, threat is likely to induce stress and deplete attentional resources and consequently lead to a simpli- fied representation of the ingroups. Consistent with our hypotheses, threat has been found to affect social cate- gorization processes (Rothgerber, 1997). For instance,

when there is a perception of threat, individuals perceive their ingroup as more homogeneous and perceive the self as more similar to the ingroup and more different from the outgroup. This finding suggests that under threat individuals prefer clear ingroup boundaries.

In sum, we propose that social identity complexity is affected both by chronic and by temporal accessibility of simultaneous membership in nonconvergent in- groups. Social identity complexity may thus vary both across individuals and across situations. Factors that af- fect chronic accessibility of multiple ingroups (e.g., the objective complexity of the social environment, per- sonal motivations, and values) will contribute to indi- vidual differences in social identity complexity. Factors that affect cognitive capacity to integrate ingroup repre- sentations (e.g., cognitive overload, stress) will con- tribute to variations in social identity complexity across situations, particularly for individuals who are not chronically complex in social identity.

Social Identity Complexity and Its Correlates: Preliminary Findings

In the initial stages of a program of research to as- sess social identity complexity and its correlates, we conducted questionnaire surveys in two cultural con- texts—the United States and Israel—to test some of the specific correlational predictions derived from our theoretical analysis of antecedent conditions.

Values and social identity complexity. Our first effort to develop and test measures for social identity complexity was a study conducted among American un- dergraduate college students. In the initial phase of the study, we asked respondents (n = 198) to check various social categories to which they belong (from a lengthy list of ethnic, religious, political, organizational, demo- graphic, and geographical social groups) and to indicate which of these group memberships were particularly important to them. The vast majority of students listed at least four or five different social identities as important to them, and most of these represented crosscutting so- cial categorizations (e.g., Catholic religion and Ohio cit- izen). Based on responses to this initial survey, we selected a subsample of respondents who identified themselves as White, American, college students, and affiliated with a large religious denomination, and who included these group memberships among their most important social identities. Social identity complexity measures were then computed with respect to these four social categories—nationality, ethnicity, religious de- nomination, and university.3

IDENTITY COMPLEXITY

3Note that by virtue of our selection rules, the target social cate- gories were equivalent in type and size across all respondents. Thus, our complexity measures reflected individual differences in subjec- tive representations of these social identities.

In the second phase of the study, we reminded re- spondents of their individual social group identities and asked them a series of questions about the relationships they perceived between all pairings of their ingroups (e.g., Catholics and Americans). One series of questions assessed their subjective impression of the extent of overlap in membership between each of their ingroups in each direction of comparison (e.g., “Of persons who are Catholic, how many are also university students?” “Of persons who are university students, how many are also Catholic?”). Judgments were made on a 10-point scale ranging from 1 (very few) to 5 (about half) to 10 (all). An index of overlap complexity was created by calculating the mean rating of proportion of overlap be- tween ingroups in which high values indicated greater overlap and less complexity in the representation of multiple identities.

A second series of questions assessed their subjec- tive impression of the extent of similarity between each of their ingroups. For every pairing of the four ingroup identities, participants were asked to indicate how much they agree that a typical member of one of the two ingroups is highly similar to a typical mem- ber of the other ingroup (e.g., in general, the typical American is very similar to the typical college stu- dent) using a 7-point rating scale ranging from 1 (strongly disagree) to 7 (strongly agree). An index of similarity complexity was created by computing the mean similarity ratings across all ingroup pairs, with higher scores indicating greater shared characteristics and lower complexity. The two measures of complex- ity were only slightly positively correlated (r = .17).

Respondents also completed the Schwartz Value In- ventory (Schwartz, 1992; Schwartz & Sagiv, 1995) rat- ing the importance they placed on 56 specific values representing the 10 types of values. From responses to this measure, we computed four value indexes that re- flected the two dimensions (openness vs. conservatism and power vs. universalism) we predicted to be related to social identity complexity. The standard indexes rec- ommended in Schwartz (1992, 1994) were used to measure the priority given to each type of values. Openness was computed as the average importance placed on creativity, freedom, independent, curious, choosing own goals, daring, a varied life, and an excit- ing life. Conservatism was defined as the average of humble, accepting my portion in life, devout, respect

for tradition, moderate, politeness, obedient, self-disci- pline, honoring parents and elders, family security, national security, social order, clean, and reciprocation of favors. Power was an index based on the importance of social power, authority, wealth, and preserving my public image, and universalism was derived from rat- ings of broad-minded, wisdom, social justice, equality, a world at peace, a world of beauty, unity with nature, and protecting the environment. We expected power and conservatism to be negatively related to social identity complexity and conversely that openness and universalism would be positively related to the com- plexity indexes.

Table 1 presents the results of the intercorrelational analysis between our measures of complexity and val- ues. Both similarity complexity and overlap complexity were related to the value dimensions in the predicted di- rection. Scores on the social identity complexity mea- sures were higher (lower complexity) for persons who ascribe relatively high importance to conservatism val- ues or to power, or both. Conversely, importance of openness to change and universalism values were asso- ciated with higher social identity complexity.

The pattern of intercorrelations in Table 1 was con- sistent with our analysis of the value antecedents of so- cial identity complexity. Values associated with low tolerance for ambiguity (conservatism as opposed to openness to change) are related to low complexity of ingroup representations. This relation was equivalent for both indexes of social identity complexity. In addi- tion, social identity complexity—particularly the over- lap complexity measure—proved to be associated with tolerance-related values (universalism as opposed to power). The size of the correlations were moderately positive, indicating that complexity is at least in part a reflection of value preferences but not equivalent to conservatism or universalism per se.

Mood and social identity complexity. We ex- amined the association of social identity complexity and stress in a survey administered to Israeli univer- sity students. The measures in Hebrew were devel- oped by translating and back translating the measures developed for the American samples. We selected a subsample of respondents who identified themselves as secular, Israeli, college students, and Jewish (n = 99). These groups were used for the complexity mea-

100

ROCCAS & BREWER

Table 1. Correlations of Values With Two Indexes of Social Identity Complexitya

Type of Value

Complexity Index Conservatism Openness Power Universalism

Similarity .20* –.25** .10 –.13 Overlap .17 –.24** .27** –.27**

aCorrelations are partialed on each respondent’s mean rating of all values to correct for scale use, as recommended in Schwartz (1992). *p < .05; **p < .01.

sures. The two measures of complexity were again only slightly positively correlated (r = .14).

The survey administered to the Israeli sample in- cluded a mood checklist, which was administered be- fore the complexity measures. The mood checklist con- sisted of 15 mood states that tapped four types of mood: Stress (e.g., worried, agitated), cheerful (e.g., happy, joyful), calm (e.g., calm, relaxed), and sad (e.g., sad, dis- appointed). Participants were instructed to check all the items that reflected their current mood. For each type of mood we computed a score by calculating the number of checked items divided by the total number of items re- flecting that mood.

We calculated the correlation between social iden- tity complexity scores and the presence of stress-re- lated mood. Both complexity indexes were signifi- cantly correlated with stress-related mood in the expected direction. That is, individuals who were ex- periencing stress had higher social identity complexity scores (lower complexity) than individuals who were not experiencing such emotions (r = .31, p < .01 for similarity complexity; r = .23, p < .05 for overlap com- plexity). Consistent with the hypothesis that mood ef- fects on social identity complexity would be specific to stress-related affect, the correlations between the other three mood indexes and both indexes of complexity were all nonsignificant (range = .00–.14).

In sum, findings indicate that experiencing a mood that depletes cognitive resources is associated with lower complexity of the representation of the social identity. This is consistent with past research (e.g. Easterbrook, 1959) indicating that strong emotional arousal narrows attention to focus on central cues to the exclusion of more peripheral stimuli. To our knowl- edge, the effects of stress on the complexity of self-rep- resentations in other self domains has not yet been ex- amined, but we would expect similar results for other domains as well.

Ingroup threat. To examine the causal link be- tween threat and complexity we conducted an experi- ment in which social identity complexity was mea- sured after threat was experimentally manipulated. The experiment was conducted in Israel using student respondents selected according to the same criteria of ingroup memberships as used in the survey study. We manipulated threat by raising either (a) the accessibil- ity of the threat of use of unconventional weapons in a possible future war with Iraq (high-threat condition) or (b) the accessibility of issues related to nature (low-threat condition). The accessibility of threat was manipulated with a four-item attitude questionnaire (e.g., “As long as Iraq has chemical weapons there is a chance that Israel will be attacked with such weap- ons” vs. “A nature walk can be relaxing”). In two separate replication experiments, we measured either similarity complexity (Experiment 1, n = 38) or over-

lap complexity (Experiment 2, n = 39) for the four targeted ingroups. Following the complexity measure, the mood checklist was administered (as in the survey questionnaire study).

Using the mood measures as a check of the conse- quences of the threat manipulation, high threat showed effects on stress-related mood in both experiments. In Experiment 1, the mean stress index for participants in the high-threat condition (M = .30) was significantly higher than that for low-threat participants (M = .00), t(36) = 2.73, p = .01. Similarly, in Experiment 2, the high-threat condition resulted in higher stress-related mood (M = .37) than the low-threat condition (M = .14), t(37) = 2.20, p < .05. Further, in both replications there was no significant effect of the threat manipulation on either positive moods (happy, calm) or non-stress-re- lated negative moods (sad).

The threat manipulation also had the predicted ef- fect on the two indexes of social identity complexity. High threat decreased complexity both for the similar- ity measure and for the overlap measure. As expected, participants in the high-threat condition in Experiment 1 perceived their ingroups as being more similar to each other (M = 4.20, n =18) than did participants in the low-threat condition (M = 3.57, n = 20), t(36) = 2.42, p < .05. Participants in the high-threat condition in Experiment 2 also perceived higher overlap between their ingroups (M = 6.68, n = 20) than participants in the low-threat condition (M = 5.91, n =19), t(37) = 2.56, p < .05.

Although the threat manipulation in both experi- mental replications affected both stress and social identity complexity as predicted, the condition effect on the two complexity measures remained signifi- cant even when stress effects were taken into ac- count in multiple regression analyses. In Experiment 1, the regression coefficient for the condition effect was .41, p < .05 after stress rating had been entered into the regression model. In Experiment 2, the re- sidual effect of condition was .38, p < .05 after stress was included. Thus, the threat manipulation affected the complexity of the representations of the ingroups above and beyond its effect on the stress experienced by the participants. We interpreted this as indicating that threat affected social identity com- plexity at least in part by raising the saliency of the threatened ingroup. To test this interpretation we computed both measures of identity complexity without including items related to the Israeli iden- tity. If the threat manipulation affected complexity through its impact on the salience of the threatened ingroup, the threat manipulation should have smaller effects on the complexity indexes based only on nonthreatened ingroups. Consistent with this inter- pretation, the threat manipulation had small and not statistically significant effects on the two new in- dexes of social identity complexity.

101

IDENTITY COMPLEXITY

Consequences of Social Identity Complexity

The importance of studying subjective representa- tions of social identity is closely related to its conse- quences. We expect that some of the implications of so- cial identity complexity are similar to the implications of the complexity of representations in other domains of the self-concept. Thus, we expect that complexity of the social identity may help individuals successfully con- front the affective implications of negative events re- lated to their social identity, just as personal self-com- plexity helps confront negative events related to the personal identity (Linville, 1987). In addition, we sug- gest that social identity complexity has unique implica- tions for intergroup relations. We focus our discussion on the latter.

Tolerance of Outgroup Members

Social identity complexity is based on chronic awareness of cross-categorization in one’s own social group memberships and those of others. A simple so- cial identity is likely to be accompanied by the percep- tion that any individual who is an outgroup member on one dimension is also an outgroup member on all oth- ers. In contrast, if an individual is aware that one of his or her ingroups only partly overlaps with any other of his or her ingroups, then we assume that individual is also aware that some of his or her ingroup members have crossed group memberships: They are ingroup members on one dimension but are simultaneously outgroup members on others. Making salient that an outgroup member on one category dimension is an ingroup member on another decreases bias by compar- ison with instances in which the latter information is not available (Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993). We suggest that this effect of social identity complexity can be extended to a tolerance for outgroups in general.

There are a number of theoretical reasons why a com- plex representation of ingroup categorization should in- fluence intergroup attitudes and behavior in ways that reduce bias and discrimination. First, crosscutting dis- tinctions make social categorization more complex and reduce the magnitude of ingroup–outgroup distinctions. According to social categorization theory (Deschamps & Doise, 1978; Doise, 1978; Vanbeselaere, 1991), pro- cesses of intracategory assimilation and intercategory contrast counteract each other when categories are crosscutting. Thus, the effects of intercategory accentu- ation are reduced or eliminated, and differences be- tween groups are minimized (or no greater than per- ceived differences within groups). This undermines the cognitive basis of ingroup bias. Second, partially over- lapping group memberships reduce the evaluative sig- nificance for the self of intergroup comparisons, thereby

undermining the motivational base for intergroup dis- crimination (Vanbeselaere, 1991). Third, multiple group memberships reduce the importance of any one social identity for satisfying an individual’s need for be- longing and self-definition (Brewer, 1991), again reduc- ing the motivational base for ingroup bias.

Finally, principles of cognitive balance (Heider, 1958; Newcomb, 1963) are also brought into play when ingroups and outgroups have overlapping membership. When another person is an ingroup member on one category dimension but belongs to an outgroup in another categorization, cognitive in- consistency is introduced if that individual is evalu- ated positively as an ingroup member but is also as- sociated with others who are evaluated negatively as outgroup members. In an effort to resolve such in- consistencies, interpersonal balance processes should lead to greater positivity toward the outgroup based on overlapping memberships.

In sum, both cognitive and motivational factors lead us to predict that complex social identities will be asso- ciated with reduced ingroup favoritism and increased tolerance and positivity toward outgroups in general.

Preliminary Findings: Social Identity Complexity and Tolerance

As an initial test of our hypotheses about social identity complexity and tolerance, the questionnaire surveys administered in the United States and Israel were expanded to include measures of tolerance to- ward outgroup members.

U.S. survey. Of the 122 respondents in the sur- vey conducted among American college students, ap- proximately half (n = 62) had an additional section at the end of their questionnaires in which they rated—on a 7-point scale of feelings of closeness ranging from 1 (not at all) to 7 (very much)—a series of target persons described by category memberships. Tolerance for outgroup members was computed by averaging the re- sponses to three targets. One of the targets was member of an outgroup defined by race, one was member of an outgroup defined by religion, and one was member of an outgroup defined both by race and religion. The de- scriptions were tailored for each participant according to his or her own race and religion.

As expected, tolerance toward outgroup members was higher for persons who had lower scores (higher complexity) on the overlap measure (r = –.32, p = .01). Similarity complexity correlated in the expected direc- tion, but the correlation was small and not significant (r = –.17, p > .10).

Results of multiple regression analyses confirmed that the relation between overlap complexity and toler- ance was significant even when controlling for per- sonal value priorities. Two values indexes were entered

102

ROCCAS & BREWER

in the analysis, derived from the two value dimensions in our previous study (see Table 1): (a) The difference between the importance of universalism values and the importance of power values reflected the relative value placed on tolerance, and (b) the difference between the importance of openness values and the importance of conservatism values reflected the relative value placed on independence from group norms. When overlap complexity was added to the model to predict tolerance ratings, the multiple correlation increased significantly (R = .42) above the correlation for values alone, F(1, 58) = 4.94, p < .05. Similarity complexity had no dis- tinctive contribution to the prediction of tolerance.

Israeli survey. In the Israeli sample, respondents reported their readiness to engage in social contact with recent immigrants from the former Soviet Union. We asked the students about the acceptability of contact with outgroup members—on a 5-point scale of readi- ness ranging from 1 (not at all) to 5 (very much)—in six domains: intermittent social relation, next-door neigh- bor, guest at one’s home, intimate friend, having a child play together, and having a child marry a recent immi- grant (α = .91).

Table 2 presents the obtained intercorrelations among the measures of stress-related mood, social identity complexity, and tolerance for contact. As ex- pected, both similarity complexity and overlap com- plexity were positively related to readiness to engage in social contact with outgroup members. In addition, as indicated earlier, both measures were correlated with reported current stress. However, results of mul- tiple regression analyses confirmed that the relation between the indexes of complexity and tolerance were significant even when controlling for reported stress. When similarity complexity was added to the model to predict tolerance ratings, the multiple corre- lation increased significantly (R = .32) above the cor- relation for stress alone, F(1, 94) = 8.94, p < .005. Similarly, when overlap complexity was added to stress, the multiple correlation (R = .29) was increased significantly, F(1, 94) = 7.05, p < .01. Further, each complexity index made a unique contri- bution to the prediction of tolerance. Thus, when overlap complexity was added to stress and similarity complexity, the multiple correlation (R = .40) in- creased significantly, F(1, 93) = 6.06, p < .05.

In sum, both the American and the Israeli survey confirm that social identity complexity is positively

related to tolerance toward outgroups. Results from the Israeli survey indicate that both similarity and overlap complexity have a distinctive contribution to the prediction of tolerance above and beyond the ef- fects of stress. Furthermore, results from the Ameri- can survey indicate that overlap complexity (but not similarity complexity) has a distinctive contribution to the prediction of tolerance above and beyond the contribution of personal value priorities. The two sur- veys differed in the cultural group examined and in the measures of tolerance. Taken together these re- sults provide support for our hypothesis that the com- plexity of representation of multiple ingroups is re- lated to intergroup prejudice.

Social identity complexity, tolerance, and threat. Experimental and field studies of ingroup identifica- tion indicate that when some external event threatens one of a person’s important social identifications, in- group bias intensifies. People who strongly identify with their ingroup show greater extremity in the fa- vorability of their evaluations of ingroup members when they perceive a threat to their identity (Brans- combe & Wann, 1996; Branscombe, Wann, Noel, & Coleman, 1993). This pattern has been largely attrib- uted to motivational factors. Social identity complex- ity, however, may serve as an alternative interpretation of what mediates this relation between external threat and ingroup bias. Threat may lower social identity complexity, and lower complexity may in turn result in lower tolerance toward outgroups.

Buffering effects of social identity complexity. On the converse side, high social identity complexity may help individuals confront threats to the status of any single ingroup. Individuals who are simulta- neously members of multiple groups that differ in their status can shift their locus of identity between their lower status and higher status groups (Roccas, 2001). Thus, high social identity complexity may serve as a buffer against aversive effects of threats to the status of a particular ingroup. Such a buffering role may be one factor in explaining the apparently small effects that membership in stigmatized groups has on well-being. Extensive research indicates that members of low-sta- tus groups do not have lower self-esteem than mem- bers of high-status groups, nor are they particularly dissatisfied with their lives (e.g. Crocker & Major, 1989; Diener & Diener, 1996). Possibly the debilitat-

103

IDENTITY COMPLEXITY

Table 2. Intercorrelations Among Stress-Related Mood, Social Identity Complexity, and Tolerance for Contact With Outgroup Members

Similarity Complexity Stress-Related Mood Tolerance

Overlap Complexity .14 .23 –.30 Similarity Complexity .31 –.31 Stress-Related Mood .13

ing effects of membership in one low-status group are mitigated by simultaneous membership in groups with relatively high status. Thus, for example, members of a discriminated group can enhance their self-esteem by identifying with a successful sports team.

Research (e.g. Dixon & Baumister, 1991; Linville, 1985, 1987; Morgan & Janoff-Bulman, 1994) stem- ming from self-complexity theory indicates that indi- viduals high on self complexity experience smaller negative reactions in response to painful events. These studies have examined threats that were specifically di- rected to individuals. We propose that social identity complexity may have an analogous buffering effect for threats directed to an ingroup. Support for this hypoth- esis was obtained in a recent study (Ruvolo, 1999) in which social identity complexity was experimentally manipulated by assigning participants to multiple (ver- sus single) social categories. Participants in the multi- ple identity condition were less affected by a social identity threat (receiving negative information about the ingroup) than participants with a single ingroup category who rated their ingroup less favorably after a threatening message.

Conclusion

We suggest that individuals differ in the complexity of their subjective representations of their multiple ingroups and indicate factors that could account for these differences. Individuals who live in a multicul- tural society that embraces an integrationist ideology are likely to have more complex representations of their multiple identities than individuals who live in a monocultural or a stratified society. Individuals who have a high need for closure or value motivational goals that emphasize maintenance of the status quo will have a simpler representation of the interrelations between their ingroups than individuals with the oppo- site motivations. We also propose that the representa- tion of one’s ingroups is likely to be affected by factors that limit attentional resources, such as stress or cogni- tive load. Our initial findings provide support for our hypotheses concerning the relations between social identity complexity, values, threat, and tolerance.

At this point, the development of our measures of social identity complexity is still rather crude and the data very preliminary. Although these results are en- couraging, validity of the measures should be investi- gated further, particularly with regard to how the re- lations of social identity complexity with external variables are affected by the specific groups included in the measure. However, if these findings are repli- cated with more refined measures and larger samples, they provide support for our thesis that awareness of ingroup diversity provides an effective formula for re- ducing intergroup prejudice.

References

Abelson, R. P. (1959). Modes of resolution of belief dilemmas. Jour- nal of Conflict Resolution, 3, 343–352.

Abelson, R. P., Aronson, E., McGuire, W. J., Newcomb, T. M., Rosenberg, M. J., & Tannenbaum, P. H. (1968). Theories of cognitive consistency: A sourcebook. Chicago: Rand McNally.

Berry, J. W. (1990). Psychology of acculturation: Understanding in- dividuals moving between cultures. In R. Brislin (Ed.), Applied cross-cultural psychology (pp. 232–253). Newbury Park, CA: Sage.

Berry, J. W. (1997). Immigration, acculturation, and adaptation. Ap- plied Psychology: An International Review, 46, 5–68.

Berry, J. W., & Sam, D. L. (1997). Acculturation and adaptation. In J. W. Berry, Y. H. Poortinga, & J. Pandey (Eds.), Handbook of cross-cultural psychology (2nd ed., Vol. 3, pp. 291–326). Bos- ton: Allyn & Bacon.

Birman, D. (1994). Acculturation and human diversity in a multicul- tural society. In E. Trickett, R. Watts, & D. Birman (Eds.), Hu- man diversity: Perspective on people in context (pp. 261–284). San Francisco: Jossey-Bass.

Branscombe, N. R., & Wann, D. L. (1996). Collective self-esteem consequences of outgroup derogation when a valued social identity is on trial. European Journal of Social Psychology, 24, 641–657.

Branscombe, N. R., Wann, D. L., Noel, J. G., & Coleman, J. (1993). In-group or out-group extremity: Importance of the threatened social identity. Personality and Social Psychology Bulletin, 19, 381–388.

Brewer, M. B. (1988). A dual process model of impression forma- tion. In R. Wyer & T. Srull (Eds.), Advances in social cogni- tion (Vol. 1, pp. 1–36). Hillsdale, NJ: Lawrence Erlbaum As- sociates, Inc.

Brewer, M. B. (1991). The social self: On being the same and differ- ent at the same time. Personality and Social Psychology Bulle- tin, 17, 475–482.

Brewer, M. B., & Gardner, W. (1996). Who is this “we”? Levels of collective identity and self representation. Journal of Personal- ity and Social Psychology, 71, 83–93.

Brewer, M. B., Ho, H.-K., Lee, J.-Y., & Miller, N. (1987). Social identity and social distance among Hong Kong school children. Personality and Social Psychology Bulletin, 13, 156–165.

Brown, R., & Turner, J. C. (1979). The criss-cross categorization ef- fect in intergroup discrimination. British Journal of Social Psy- chology, 18, 371–383.

Conway, M., Carroll, J. M., Pushkar, D., & Arbuckle, T. (1996). An- ticipated interaction, individual differences in attentional re- sources, and elaboration of behavior. Social Cognition, 14, 338–366.

Cota, A. A., & Dion, K. L. (1986). Salience of gender and sex com- parison of ad hoc groups: An experimental test of distinctive- ness theory. Journal of Personality and Social Psychology, 50, 770–776.

Crisp, R. J., & Hewstone, M. (1999). Differential evaluation of crossed category groups: Patterns, processes, and reducing in- tergroup bias. Group Processes and Intergroup Relations, 2, 1–27.

Crocker, J., & Major, B. (1989). Social stigma and self esteem: The self-protective properties of stigma. Psychological Review, 96, 608–630.

Cross, S. E. (1995). Self-construals, coping, and stress in cross-cul- tural adaptation. Journal of Cross-Cultural Psychology, 26, 673–697.

Cross, W. E. (1991). Shades of black: Diversity in African-Amerian identity. Philadelphia: Temple University Press.

Deaux, K. (1996). Social identification. In T. E. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic prin- ciples (pp. 777–798). New York: Guilford.

104

ROCCAS & BREWER

Deschamps, J.-C., & Doise, W. (1978). Crossed category member- ship in intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the social psychology of in- tergroup relations (pp. 141–158). London: Academic.

Diener, E., & Diener, C. (1996). Most people are happy. Psychologi- cal Science, 7, 181–185.

Dixon, T. M., & Baumister, R. F. (1991). Escaping the self: The mod- erating effects of self-complexity. Personality and Social Psy- chology Bulletin, 17, 363–368.

Doise, W. (1978). Groups and individuals: Explanations in social psychology. Cambridge, England: Cambridge University Press.

Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183–201.

Eysenck, M. W. (1992). Anxiety: The cognitive perspective. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 1, pp. 357–413). New York: McGraw-Hill.

Gaertner, S. L., Dovidio, J., Anastasio, P., Bachman, B., & Rust, M. (1993). The common ingroup identity model: Recategorization and the reduction of ingroup bias. In W. Stroebe & M. Hew- stone (Eds.), European review of social psychology (Vol. 4, pp. 1–26). London: Wiley.

Harris, J. C. (1995). Where is the child’s environment? A group so- cialization theory of development. Psychological Review, 102, 458–489.

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.

Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events: The accuracy of remembered minutiae. Memory and Cognition, 18, 496–506.

Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Per- sonality and Social Psychology, 64, 779–793.

Higgins, T. E. (1996). Knowledge activation: Accessibility, appli- cability, and salience. In T. E. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133–168). New York: Guilford.

Ingram, R. E., & Kendall, P. C. (1987). The cognitive side of anxiety. Cognitive therapy and research, 11, 523–536.

Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice Hall.

Kelley, J., & Evans, M. D. (1995). Class and class conflict in six western nations. American Sociological Review, 60, 157–178.

Kelman, H. C., & Baron, R. M. (1968). Determinants of modes of resolving inconsistency dilemmas: A functional analysis. In R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds.), Theories of cognitive consistency: A sourcebook (pp. 670–683). Chicago: Rand McNally.

Kluckhohn, C. (1951). Values and value-orientations. In T. Parsons & E. Shils (Eds.), Toward a general theory of action (pp. 388–433). New York: Harper & Row.

Kruglanski, A. W. (1990). Motivation for judging and knowing: Im- plications for causal attribution. In E. T. Higgins & R. M. Sor- rentino (Eds.), The handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 333–368). New York: Guilford.

Kruglanski, A. W., & Webster, D. M. (1991). Group members’ reactions to opinion deviates and conformists at varying degrees of proximity to decision deadline and of environmental noise. Journal of Personality and Social Psychology, 61, 212–225.

Kruglanski, A. W., Webster, D. M., & Klem, A. (1993). Motivated resistance and openness to persuasion in the presence or ab- sence of prior information. Journal of Personality and Social Psychology, 65, 861–876.

Kunda, Z., Miller, D. T., & Claire, T. (1990). Combining social con- cepts: The role of causal reasoning. Cognitive Science, 14, 551–577.

LaFromboise, T., Coleman, H. L. K., & Gerton, J. (1993). Psycho- logical impact of biculturalism: Evidence and theory. Psycho- logical Bulletin, 114, 395–412.

Leone, C., Wallace, H. M., & Modglin, K. (1999). The need for clo- sure and the need for structure: Interrelationships, correlates, & outcomes. Journal of Psychology, 133, 553–562.

Lickel, B., Hamilton, D. L., Wieczorkowska, G., Lewis, A., Sher- man, S. J., & Uhles, A. N. (2000). Varieties of groups and the perception of group entitativity. Journal of Personality and So- cial Psychology, 78, 223–246.

Linville, P. W. (1985). Self-complexity and affective extremity: Don’t put all of your eggs in one cognitive basket. Social Cogni- tion, 3, 94–120.

Linville, P. W. (1987). Self-complexity as a cognitive buffer against stress-related illness and depression. Journal of Personality and Social Psychology, 52, 663–676.

Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1995). The dis- section of selection in person perception: Inhibitory processes in social stereotyping. Journal of Personality and Social Psy- chology, 69, 397–407.

Massey, D. S., & Denton, N. A. (1989). Hypersegregation in U.S. Metropolitan areas: Black and Hispanic segregation along five dimensions. Demography, 26, 373–391.

McGuire, W. J., & McGuire, C. V. (1988). Content and process in the experience of self. In L. Berkowitz (Ed.), Advances in experi- mental social psychology (Vol. 21, pp. 97–144). San Diego, CA: Academic.

McGuire, W. J., McGuire, C. V., Child, P., & Fujioka, T. (1978). Sa- lience of ethnicity in the spontaneous self-concept as a function of one’s ethnic distinctiveness in the social environment. Jour- nal of Personality and Social Psychology, 36, 511–520.

Migdal, M. J., Hewstone, M., & Mullen, B. (1998). The effects of crossed categorization on intergroup evaluations: A meta-anal- ysis. British Journal of Social Psychology, 37, 303–324.

Milkie, M. A., & Peltola, P. (1999). Playing all the roles: Gender and the work–family balancing act. Journal of Marriage and the Family, 61, 476–490.

Moghaddam, F. M. (1992). Assimilation et multiculturalisme: Le cas de minorities au Quebec [Assimilation and multiculturalism: The case of minorities in Quebec]. La Revue Quebecoise de Psychologie, 13, 140–157.

Morgan, H. J., & Janoff-Bulman, R. (1994). Positive and negative self-complexity: Patterns of adjustment following traumatic versus non-traumatic life experiences. Journal of Social and Clinical Psychology, 13, 63–85.

Neuberg, S., Judice, T. N., & West, S. G. (1997). What the Need for Closure Scale measures and what it does not: Toward differenti- ating among related epistemic motives. Journal of Personality and Social Psychology, 72, 1396–1412.

Neuberg, S., & Newsome, J. T. (1993). Personal need for structure: Individual differences in the desire for simpler structure. Jour- nal of Personality and Social Psychology, 65, 113–131.

Neufeld, R. W. J., & McCarty, T. S. (1994). A formal analysis of stressor and stress-proneness effects on simple information pro- cessing. British Journal of Mathematical and Statistical Psy- chology, 47, 193–226.

Newcomb, T. (1963). Stabilities underlying changes in interpersonal attraction. Journal of Abnormal and Social Psychology, 66, 376–386.

Niedenthal, P. M., Setterlund, M. B., & Wherry, M. B. (1992). Possible self-complexity and affective reactions to goal-relevant evalua- tion. Journal of Personality and Social Psychology, 63, 5–16.

Nielsen, M. E., & Fultz, J. (1997). An alternative view of religious complexity. International Journal for the Psychology of Reli- gion, 7, 23–35.

105

IDENTITY COMPLEXITY

Norton, R. W. (1975). Measurement of ambiguity tolerance. Journal of Personality Assessment, 39, 607–619

Oakes, P. J., & Turner, J. C. (1990). Is limited information processing the cause of social stereotyping. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 1, pp. 111–135). London: Wiley.

Osterhouse, R. A., & Brock, T. C. (1970). Distraction increases yielding to propaganda by inhibiting counterarguing. Journal of Personality and Social Psychology, 15, 344–358.

Oyserman, D., Sakamoto, I., & Lauffer, A. (1998). Cultural accom- modation: Hybridity and the framing of social obligation. Jour- nal of Personality and Social Psychology, 6, 1606–1618.

Petty, R. E., Wells, G. L., & Brock, T. C. (1976). Distraction can en- hance or reduce yielding to propaganda: Thought disruption versus effort justification. Journal of Personality and Social Psychology, 34, 874–884.

Phinney, J. S. (1990). Ethnic identity in adolescents and adults: Re- view of research. Psychological Bulletin, 108, 499–514.

Phinney, J. S., & Devich-Navarro, M. (1997). Variations in bicultural identification among African American and Mexican American adolescents. Journal of Research on Adolescence, 7, 3–32.

Prentice, D. A., Miller, D. T., & Lightdale, J. R. (1994). Asymme- tries in attachments to groups and to their members: Distin- guishing between common-identity and common-bond groups. Personality and Social Psychology Bulletin, 20, 484–493.

Roccas, S. (2001). The effects of status on identification with multi- ple groups. Manuscript submitted for publication.

Rokeach, M. (1973). The nature of human values. New York: Free Press. Rothbart, M., & John, O. (1985). Social categorization and behav-

ioral episodes: A cognitive analysis of the effects of intergroup contact. Journal of Social Issues, 41(3), 81–104.

Rothgerber, H. (1997). External intergroup threat as an antecedent to perceptions in in-group and out-group homogeneity. Journal of Personality and Social Psychology, 73, 1206–1212.

Ruvolo, C. M. (1999). Benefits of identity complexity. Unpublished manuscript, United States Military Academy, West Point, NY.

Salmela, J. H., & Ndoye, O. D. (1986). Cognitive distortions during progressive exercise. Perceptual and Motor Skills, 63, 1067–1072.

Schroder, H. M., Driver, M. J., & Streufert, S. (1967). Human infor- mation processing. New York: Holt, Rinehart & Winston.

Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical advances and empirical tests in 20 coun- tries. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 1–65). New York: Academic.

Schwartz, S. H. (1994). Are there universal aspects in the content and structure of values? Journal of Social Issues, 50(4), 19–45.

Schwartz, S. H. (2000). [Multidimensional analyses of value struc- tures in 65 countries.] Unpublished raw data.

Schwartz, S. H., & Sagiv, L. (1995). Identifying culture-specifics in the content and structure of values. Journal of Cross-Cultural Psychology, 26, 92–116.

Schwartz, S. H., Sagiv, L., & Boehnke, K. (2000) Worries and val- ues. Journal of Personality, 68, 309–346.

Sellers, R. M., Smith, M., Shelton, J. N., Rowley, S., & Chavous, T. (1998). Multidimensional model of racial identity: A recon- ceptualization of African American racial identity. Personality and Social Psychology Review, 2, 18–39.

Shah, J. Y., Kruglanski, A. W., & Thompson, E. P. (1998). Member- ship has its (epistemic) rewards: Need for closure effects on in-group bias. Journal of Personality and Social Psychology, 75, 383–393.

Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of so- cial psychology (4th ed., Vol. 1, pp. 391–445). New York: McGraw-Hill.

Sorrentino, R. M., Hodson, G., & Huber, G. L. (2001). Uncer- tainty orientation and the social mind: Individual differ- ences in the interpersonal context. In J. Forgas, K. Wil- liams, & L. Wheeler (Eds.), The social mind: Cognitive and motivational aspects of interpersonal behavior (pp. 199–227). Cambridge, England: Cambridge University Press.

Sorrentino, R. M., & Roney, C. (1999). The uncertain mind: Individ- ual differences in facing the unknown. London: Psychology Press.

Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992). Categorization of individuals on the basis of multiple social features. Journal of Personality and Social Psychology, 62, 207–218.

Steblay, N. M. (1992). A meta-analytic review of the weapon focus effect. Law and Human Behavior, 16, 413–424.

Stryker, S. (1980). Symbolic interactionism: A social structural ver- sion. Menlo Park, CA: Benjamin/Cummings.

Stryker, S., & Serpe, R. T. (1994). Identity salience and psychologi- cal centrality: Equivalent, overlapping, or complementary con- cepts? Social Psychological Quarterly, 57, 16–35.

Stryker, S., & Statham, A. (1985). Symbolic interaction and role theory. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (pp. 311–378). New York: Random House.

Sussman, N. M. (2000). The dynamic nature of cultural identity throughout cultural transitions: Why home is not so sweet. Per- sonality and Social Psychology Review, 4, 355–373.

Tajfel, H. (1978). Interindividual behavior and intergroup behavior. In H. Tajfel (Ed.), Differentiation between social groups: Stud- ies in the social psychology of intergroup relations (pp. 27–60). London: Academic.

Tetlock, P. E. (1983). Accountability and complexity of thought. Journal of Personality and Social Psychology, 45, 74–83.

Tetlock, P. E. (1986). A value pluralism model of ideological rea- soning. Journal of Personality and Social Psychology, 50, 819–827.

Tetlock, P. E., Skitka, L., & Boettger, R. (1989). Social and cognitive strategies of coping with accountability: Conformity, complex- ity, and bolstering. Journal of Personality and Social Psychol- ogy, 57, 632–641.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categori- zation theory. Oxford, England: Basil Blackwell.

Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454–463.

Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization effects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908.

Vanbeselaere, N. (1991). The different effects of simple and crossed categorizations: A result of the category differentiation process or of differential category salience? In W. Stroebe & M. Hew- stone (Eds.), European review of social psychology (Vol. 2, pp. 247–278). Chichester, England: Wiley.

Voydanoff, P., & Donnelly, B. W. (1999). Multiple roles and psycho- logical distress: The intersection of the paid worker, spouse, and parent roles with the role of the adult child. Journal of Mar- riage and the Family, 61, 725–738.

Wegner, D. M., & Wenzlaff, R. M. (1996). Mental control. In T. E. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 466–492). New York: Guilford.

Woike, B. A., & Aronoff, J. (1992). Antecedents of complex social cognitions. Journal of Personality and Social Psychology, 63, 97–104.

106

ROCCAS & BREWER