Evidence of System Justification in Young Children

The near ubiquity of ingroup preference is consistent with the view that it is an automatic consequence of social categorization, possibly a basic foundation of intergroup relations. However, research with adults has demonstrated that automatic ingroup preference is notably absent among less dominant, less advantaged groups, an outcome predicted by System Justification Theory (Jost & Banaji, 1994). How basic is this tendency to justify existing social arrangements? Data from young children are crucial in addressing whether such an opposing orientation is itself a fundamental feature of intergroup social cognition. The developmental data summarized here suggest that knowledge about the relative status of one’s ingroup is absorbed and internalized sufficiently early in life, revealing system-justifying tendencies by age 5, the earliest age such questions have been examined to date. Across several studies summarized here young children from non-dominant groups failed to show an implicit ingroup preference, similar to their adult counterparts. We conclude that from an early age intergroup preferences are constrained by knowledge, implicit or explicit, about the relative status differences among groups and may suggest an orientation toward supporting existing social and political structures. The possibility that system-justifying tendencies may exist in even younger children remains open for future tests. Implicit ingroup preference is a ubiquitous phenomenon and it sits at the center of prominent theories that connect our understanding of self, identity, and intergroup relations (Jost & Hunyady, 2005; Tajfel & Turner, 1986; Turner, Brown, & Tajfel, 1979). Ingroup preference is seen not only in the context of real-world groups in which one holds longterm membership (e.g., race ⁄ethnicity, religion, nationality) but also in the context of arbitrarily created minimal groups, which by definition, lack intra-group interaction, knowledge about individual group members, or long-term identity. Research using the classic ‘minimal groups’ paradigm in which participants are randomly assigned to a group about which they have no prior knowledge has demonstrated that group membership is sufficient to engender explicit ingroup preference as well. Such a preference for the ingroup is manifested in the willingness to allocate greater resources to ingroup members, the tendency to positively stereotype the ingroup and to bestow more complex human emotions such as pride and shame to the ingroup (Cadinu & Rothbart, 1996; Gramzow, Gaertner, & Sedikides, 2001; Nesdale & Flesser, 2001; Tajfel, 1973, 1981; Turner, Brown, & Tajfel, 1979). Together, these results give us one of social psychology’s best known discoveries: that preference for one’s own group is a fact of human social relations. Because much of this evidence has been obtained on standard self-report measures, the role of demand characteristics including self-presentation could not be ruled out as a boundary condition only within which such results held true. Ample research has now demonstrated robust limits on the ability of individuals to know their thoughts and feelings and to accurately report on them (Bargh & Chartrand, 1999; Nisbett & Wilson, Social and Personality Psychology Compass 3 (2009): 1–9, 10.1111/j.1751-9004.2009.00214.x a 2009 The Authors Journal Compilation a 2009 Blackwell Publishing Ltd 1977). In contrast with explicit measures of preference such as self-report, implicit measures tend to capture preferences outside of conscious awareness and control. This ability is especially useful when studying preferences for socially charged constructs such as race because implicit measures are much more difficult to manipulate compared with their explicit counterparts. Additionally, these implicit measures of preference tend to predict different behaviors than their explicit counterparts, suggesting that both forms of preference have utility (Greenwald, Poehlman, Uhlmann, & Banaji, 2009). One such measure of implicit preference is the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998). The Implicit Association Test (henceforth, IAT) is a measure of semantic associations that has been used widely to assess unconscious thoughts including preferences, stereotypes, self-esteem and identity. The IAT is a reaction-time measure that assesses the strength of association between paired concepts using a very simple logic: the more two concepts are associated with one another, the easier it should be to group them together. The IAT compares the ease with which a person pairs particular concepts together (e.g., the concept good with the concept White and the concept bad with the concept Black) with the ease with which the opposite pairings are made (bad with White and good with Black). In this example, a preference for White relative to Black would correspond with faster and more accurate responses pairing the concepts good with White and bad with Black (to sample an IAT, please visit http://implicit.harvard.edu). Even when implicit measures of intergroup preference are used, the same result that has been obtained with explicit measures is seen. Using the IAT, we see that 75% of white American adults show an implicit preference for White over Black (Nosek, Banaji, & Greenwald, 2002). The size of this effect may have been impressive were it not for many other such demonstrations of implicit ingroup favoring preferences that have been reported over the past decade. For example, Korean and Japanese Americans who report little antipathy toward the other group nevertheless show robust and equivalent implicit ingroup preference; Christians likewise show preference for their religion over others; the young prefer the young; heterosexuals favor heterosexuals, and the list goes on (Nosek et al., 2007). Converging on the idea that an ingroup preference may be a psychological default, research suggests that the acquisition of an ingroup preference follows rapidly on the heels of social categorization (Bigler, 1995); as soon as a child can represent a social category, preference for the ingroup follows. For example, by age two, soon after the age when children appear to have a mature concept of gender, children report a stronger preference for their gender ingroup (Maccoby, 1988). And, by age four, the same age when children develop a representation of race commensurate with their adult counterparts, children report a stronger preference for their racial ingroup (Aboud, 1988; Hirschfeld, 1996). Furthermore, an ingroup preference is observed in children as young as age 3 on a variety of measures within a minimal groups paradigm (Bigler, Jones, & Lobliner, 1997). A 3-year-old randomly assigned to one of two novel social groups reveals greater preference for members of her new group. Again, the finding seems to be that once a category is represented, preference follows. The developmental evidence coupled with the adult findings suggest that a bias to favor the ingroup is a core aspect of human psychology, possibly an indirect indication of love for oneself (Greenwald & Banaji, 1995) that follows automatically from categorizing an individual as a member of one’s own group, an extension of oneself. Indeed, in all cultures studied, positively evaluating the self and one’s social group memberships seems to be the intuitive norm and is supported by extensive research (Banaji & Prentice, 1994; Greenwald, Pickrell, & Farnham, 2002; Taylor & Fiske, 1978; Yamaguchi et al., 2007). 2 System Justification in Children a 2009 The Authors Social and Personality Psychology Compass 3 (2009): 1–9, 10.1111/j.1751-9004.2009.00214.x Journal Compilation a 2009 Blackwell Publishing Ltd The absence of ingroup preference Among the reasons for using implicit measures of social cognition is the possibility that one may not know one’s own mind. This is certainly possible in the domain of intergroup relations where the preference for one’s own group is often at odds with the social demand to be considerate of others irrespective of group membership. For this reason, a result that is routinely observed in the data from white and black Americans is interesting. On self-report measures of race preference, whites tend not to report a strong explicit preference for their ingroup, typically not as strong a preference as black Americans do for their own group. On the other hand, white Americans show a robust implicit preference for their own group and far more so than do black Americans. In fact, the data from adult black Americans consistently show no implicit preference on average for the ingroup (Nosek, Banaji, & Greenwald, 2002), a finding we will return to later. Here we focus on the nature of implicit intergroup preferences among adults and then show how the question of origins can inform the theoretical explanation of the adult state. The first such result we obtained was at Yale University, where black American undergraduates showed no implicit ingroup preference. Taking the result to reflect the mind of this sample rather than black Americans more generally, we were surprised to discover in other more diverse samples of black Americans that a lack of implicit ingroup preference emerged in every case. In all samples tested, the number of black Americans showing an own group preference was matched by an equal number showing an outgroup (white) preference, bringing the group’s overall implicit preference to a null – no preference one way or the other. The most robust such result is available in the data from thousands of black Americans who have completed the race IAT at implicit.harvard.edu (Nosek et al., 2007). What began as a few demonstrations of a lack of implicit ingroup preference (e.g., absence of ingroup bias among African-American adults) has now been replicated dozens of times for quite divergent social categories. Indeed, a substantial body of literature now reveals the absence of an implicit ingroup preference for categories of race, body weight, gender, and sexuality to name just a few (Nosek et al., 2007). The noticeable absence of implicit ingroup preference among these varied groups has challenged psychologists for an explanation. Should we reevaluate the core assumption that a preference for the ingroup stems from an automatic consequence of self-identification? Or, should each of these findings be accounted for independently through minor adjustments and ⁄ or qualifications to this theory? System justification System Justification Theory (Jost, 1996; Jost, Banaji, & Nosek, 2004; Jost Fitzsimons, & Kay, 2004) in particular offers a useful framework for interpreting both the presence and absence of an implicit ingroup preference. According to the theory, social groups exist within a sociopolitical system and intergroup relations are, in part, influenced by a desire to maintain and or justify the status quo of that system. More importantly, System Justification Theory sets forth specific predictions concerning the conditions under which the presence or absence of an ingroup preference should be observed; namely whether one’s social group is considered dominant or not in the local cultural milieu. Specifically, both dominant and non-dominant groups should exhibit a preference for the culturally dominant groups because such a preference legitimizes the existing social and political order. The motivation for both groups is the same – to maintain a belief that System Justification in Children 3 a 2009 The Authors Social and Personality Psychology Compass 3 (2009): 1–9, 10.1111/j.1751-9004.2009.00214.x Journal Compilation a 2009 Blackwell Publishing Ltd the existing political structure is just simply because it exists. However, the consequence of this motivation produces quite divergent outcomes for dominant and non-dominant groups. In particular, dominant groups generally exhibit an ingroup preference because it is such attitudes and beliefs that support the system; non-dominant groups, on the other hand, exhibit a muted ingroup preference or outright outgroup preference because such attitudes and beliefs support or justify the existing system. Importantly, this pattern among non-dominant groups is only expected to be observed when comparing the ingroup to a higher status outgroup such that non-dominant groups would be expected to exhibit a positive ingroup bias when comparing their ingroup to an even lower-status outgroup (Lane, Mitchell, & Banaji, 2005; Rudman, Feinberg, & Fairchild, 2002). While the pattern of implicit preference data from adults described above is consistent with a system justification framework, these data do not undermine alternative proposals that privilege the self and an automatic preference for the ingroup. Indeed, it is possible that an automatic preference for the ingroup may characterize the acquisition of implicit social group attitudes, whereas System Justification Theory better characterizes how such attitudes change once formed. In particular, a preference for the ingroup may serve as a foundation for establishing intergroup evaluations. However, as one’s knowledge about the existing social and political system develops and knowledge about the cultural valuation of different groups becomes reinforced, then intergroup evaluations will change, serving to legitimize that system. Recent studies with children provide a critical opportunity to examine the role of an automatic ingroup preference and system-justifying beliefs in the acquisition and development of implicit intergroup preference. In particular, if children from non-dominant groups report a strong preference for their ingroup when adults from those groups show the opposite pattern, then this would point to an automatic ingroup preference as a psychological default. However, if children from non-dominant groups reveal similar patterns of preference as adults in their social groups (i.e., no ingroup preference among members of non-dominant group), then this would suggest that both the acquisition and development of intergroup attitudes are shaped by system-justifying beliefs. The age that children’s implicit preferences reflect status differences between groups will also have particular implications for our understanding of System Justification Theory and the origins of social group preferences. In particular, if young children prior to explicitly representing political systems show evidence of internalizing status information in the expression of their intergroup preferences, this would suggest that system-justifying beliefs might largely take the form of implicit representations. Alternatively, if children’s sensitivity to such status information emerges later in development, it would suggest that system-justifying beliefs may derive from more explicitly learned representations of power structures. Here is a case where developmental research can adjudicate between these contending theories while providing a more accurate framework for understanding the origins and nature of implicit intergroup preferences. Emergence of system justification in young children The past 5 years have witnessed increased interest in the development of implicit social cognition (Baron & Banaji, 2006; Dunham, Baron, & Banaji, 2006; Rutland, Cameron, Milne, & McGeorge, 2005; Sinclair, Dunn, & Lowery, 2005; Davis, Leman, & Barrett, 2007). A review of this literature yields a cogent picture – 5-year-olds, like their adult counterparts, reveal comparable magnitudes of implicit intergroup bias. In the case of race 4 System Justification in Children a 2009 The Authors Social and Personality Psychology Compass 3 (2009): 1–9, 10.1111/j.1751-9004.2009.00214.x Journal Compilation a 2009 Blackwell Publishing Ltd attitudes, where adult members of dominant groups report an implicit preference for the ingroup, children from these groups do so as well at magnitudes statistically indistinguishable from their adult counterparts. More importantly, where adult members of non-dominant groups report no implicit own group preference, children from these groups also fail to reveal such a preference. The mean level of implicit ingroup preference among 5-year-old children from non-dominant groups is also statistically indistinguishable from their adult counterparts, suggesting that by this young age children have already begun to internalize the cultural valuation of different social groups. In a first study, children ages 5–12 and adults were recruited from a predominantly Latino urban community in the United States (Dunham, Baron, & Banaji, 2007). Attitudes towards the ingroup category Latino was measured by contrasting it with two outgroups that differed in their relative cultural status, white (high-status) and black (lowstatus). Like their adult counterparts and consistent with previous findings among white children (Baron & Banaji, 2006; Rutland, Cameron, Milne, & McGeorge, 2005), Latino children showed early and robust implicit bias favoring the ingroup over a lower-status outgroup, black. However, the Latino-over-White bias was entirely absent (on average) from the youngest age tested (5-year-olds). That is, for the first time a case in which early implicit bias is not present at this age was discovered. Instead, an ingroup bias was only observed when the outgroup was non-dominant (black American). Both patterns of data (Latino > Black & Latino < White) were relatively stable across development, further evidence that such preferences are acquired early and undergo little change across development. These data suggest that young children from a non-dominant group are in fact sufficiently sensitive to status differences among groups involving their own, to show no ingroup preference. In a further study, implicit intergroup attitudes among black American 5–12-year-olds were measured (Baron, Shusterman, Bordeaux, & Banaji, 2004). This study explored whether black American children showed greater implicit ingroup preference when comparing their ingroup to the dominant outgroup, white. These children, like their adolescent and adult counterparts tested before (Baron, Shusterman, Bordeaux, & Banaji, 2004; Nosek, Banaji, & Greenwald, 2002), revealed no ingroup preference on average. Such a pattern reveals the early entry of implicit intergroup knowledge into a child’s mind. Together with the data from Latino children, these data suggest that similar to dominant groups, children from non-dominant groups exhibit greater own group preference when compared with another low-status group. However, unlike children from dominant groups, these children fail to show a preference for their own group when the comparison group is of high-status, suggesting they have internalized societal views of their group by the fifth year of life. A study involving a culture with relatively little exposure to status differences among certain outgroups would be particularly useful to tease out the developmental aspects of intergroup attitudes. Dunham, Baron, & Banaji (2006) examined whether Japanese 6-year-olds, 10-year-olds and adults living in a rural Japanese community with little exposure to the outgroups of whites and blacks, would similarly reveal an early sensitivity to status differences among these groups. For Japanese children in this community, white and black Americans ought to be equivalent outgroups. As such, an ingroup preference should be equally observed when one’s own group is compared to a dominant versus less-dominant outgroup in young children who cannot have learned the status difference between white and black. On the other hand, for adult Japanese, ingroup preference should be markedly greater when the ingroup (Japanese) is compared with black Americans (a known non-dominant group) compared with white Americans (a known System Justification in Children 5 a 2009 The Authors Social and Personality Psychology Compass 3 (2009): 1–9, 10.1111/j.1751-9004.2009.00214.x Journal Compilation a 2009 Blackwell Publishing Ltd dominant outgroup), since adults have had more opportunity to learn about the status differences between these two groups. Similar to the previous study with Latinos, implicit attitudes towards the ingroup (Japanese) were measured by contrasting it with two outgroups, white and black, in separate tests. When Japanese was compared with Black, a strong implicit preference for the ingroup was observed and the magnitude of this preference remained stable across development. Notably, when Japanese was contrasted with White, an implicit preference for the ingroup emerged in young children, however, the magnitude of this preference was attenuated in Japanese adults, pointing to a developmental change in the sensitivity to the cultural valuation of black and white. In particular, Japanese adults showed less ingroup bias than did Japanese children when the ingroup was compared with a higher status outgroup (white). These data suggest that ingroup preference is indeed the default state when the relative status of the outgroup is not sufficiently salient. Unlike Latino and black American children living in urban US communities, Japanese children had little to no exposure to either group in their rural community and hence no basis on which to differentiate between white and black outgroups. However, adult Japanese who are necessarily aware of dominance differences between white and black showed stronger preference for their own group when the contrasting group was black rather than white. Rather than children’s early implicit intergroup evaluations being driven entirely by a purported automatic preference for the ingroup, the developmental record suggests that from an early age children are sensitive to social dominance hierarchies embedded in their experienced social world. Collectively, these data suggest that by age 5 (the earliest age at which measures of implicit evaluation are currently adapted for test) children’s patterns of intergroup preference show sensitivity to knowledge, perhaps implicit given their early emergence, of the relative dominance of social groups within their cultural milieu. That this information shapes patterns of ingroup preference, specifically a lack of ingroup preference, is consistent with a main tenet of system justification theory. One alternative to consider is that such young children from non-dominant groups may hold dual preferences that are in conflict; a preference for the self and one’s ingroup on the one hand, and a preference for the dominant group on the other. Together, these two preferences may cancel each other producing what appears to be no mean level preference for either group.