Perceived Consensus as a Foundation of Racial Stereotyping

In four experiments, it was demonstrated that the expression of racial stereotyping and prejudice is highly affected by perceptions about the extent to which stereotypic beliefs are consensually shared by members of relevant reference groups. Providing feedback that others held more favorable beliefs toward African Americans than individuals originally estimated increased the expression of positive and reduced the expression of negative stereotypes, whereas providing information that others held less favorable beliefs than originally estimated increased the expression of negative and reduced the expression of positive stereotypes. The observed stereotype change was large in magnitude, and it was found to hold up over a period of one week. Consensus effects were stronger among people who were low rather than high in initial confidence about their stereotypic beliefs and among people who were exposed to information about the opinions of ingroup rather than outgroup members. Finally, providing information that others agreed with the individual’s own stereotypes made the stereotypes more resistant to subsequent attempts to change them. Consensus Feedback and Stereotype Change 5 Perceived Consensus as a Foundation of Racial Stereotyping Stereotypes are developed and changed both as a result of information that comes from indirect sources, such as parents, peers, and the media, as well as through direct contact with members of stereotyped social groups. Although it is frequently acknowledged that both of these sources are determinants of stereotyping and prejudice, current research has tended to focus almost exclusively upon the influence of direct intergroup contact. This is the focus of virtually all current models of stereotype formation (cf. Eagly & Kite, 1987; Hamilton & Gifford, 1976; Hoffman & Hurst, 1989) and stereotype change (Hewstone & Brown, 1986; Rothbart & John, 1992; Stephan, 1985). Furthermore, current reviews of the stereotyping and prejudice literatures (Fiske, 1998; Brewer & Brown, 1998; Hamilton & Sherman, 1994) have paid virtually no attention to the possibility that stereotypes are learned or changed through information provided by socially relevant ingroups. There are theoretical and practical limitations to the assumption that stereotypes are determined primarily by direct intergroup contact. First, a large body of research has demonstrated that stereotypes and prejudice can be developed about groups with which the individual has had very little or even no direct contact. Research also indicates that stereotypes are by and large very difficult to change through exposure to individual exemplars who disconfirm existing beliefs. For example, in the intergroup contact literature, attitudes are found to change through exposure to group members only in very limited conditions (Stephan, 1985). And even when contact leads to a change in attitudes toward individual group members, attitudes toward the group as a whole generally do not follow (cf. Hewstone & Brown, 1986). Secondly, the underlying assumption that social stereotypes are primarily data-driven has tended to isolate the study of social stereotyping from other social psychological approaches. A guiding principle of social psychology has been that the perception of everyday reality is shaped less by direct experience with the targets of social perception than through processes of social influence and subjective comparisons with the opinions expressed by others (Allport, 1935; 1985; Asch, 1952; Bar-Tal, 1990; Berscheid, 1966; Festinger, 1957; Hardin & Higgins, 1996; Jones, 1985; Kelman, 1961; Lewin, 1952; Sherif, 1936; Turner, 1991). By focusing on “bottom-up” information processing mechanisms, researchers run the risk of underestimating the extent to which social stereotypes get their power from the fact that they are shared with others. Early theories of social stereotyping were predicated on the notion that consensuality was an essential feature of stereotypes (Allport, 1954; Katz & Braly, 1933), and that assumption has been retained by some contemporary authors (Allen & Wilder, 1980; Ehrlich, 1973; Abrams & Hogg, 1988; Jost & Banaji, 1994; Schaller & Conway, in press; Tajfel, 1981; Oakes et al., 1994; Stangor & Schaller, 1996). But, with only two recent exceptions (Haslam, Oakes, McGarty, Turner, Reynolds, & Eggins 1996; Wittenbrink & Henly, 1996), the possibility that stereotypes might be formed or altered purely on the basis of perceptions of other people's stereotypic beliefs has not been empirically examined. This is also the focus of the present research. Consensus Feedback and Stereotype Change 6 In one study, Wittenbrink and Henly (1996, Experiment 3) used a response format manipulation to give high and low prejudiced white participants -as determined on the basis of scores on McConahay, Hardee, & Batts’ (1981) Modern Racism Scale -the expectation that other individuals believed that African Americans had either a large proportion or small proportion of negative characteristics. Participants then completed the Modern Racism Scale again. Results indicated that high prejudiced participants expressed more favorable attitudes toward African Americans after they had been provided with positive (as opposed to negative) feedback about the beliefs of others, but initially low prejudiced individuals did not show any change as a result of the opinion feedback. In another relevant study, Haslam, Oakes, McGarty, Turner, Reynolds, and Eggins (1996) found that people changed their stereotypes of national groups so that they were more similar to the beliefs allegedly held by members of a desirable ingroup (other “unprejudiced” students at one’s college), and they changed their stereotypes away from the beliefs allegedly held by an undesirable outgroup (“prejudiced” people). Although these lines of research suggest that it might be possible to change stereotypical beliefs by providing consensus information about the stereotypes held by others, they are not conclusive in this regard. The research by Haslam et al. (1996) is problematic because the attitudes of the reference group (prejudiced versus non-prejudiced) were deliberately confounded with the source of the communication (ingroup versus outgroup), making it impossible to know the extent to which group membership versus attitudinal content was important in producing stereotype change. Also, in this study the lack of a control group makes it impossible to determine which source of feedback produced change from initial opinions. The Wittenbrink and Henly (1996) study is also somewhat difficult to interpret because only participants who were high in initial levels of prejudice were found to have changed their beliefs as a function of receiving consensus information, and there is no clear reason why this should have been so. Did low prejudiced people fail to accept the validity of others' beliefs, did they fail to identify with the reference group that served as the source of consensus information, were their attitudes more clearly established, or did this occur for still some other reason? The theory of "self-categorization" (Turner, 1987) and the theory of "shared reality" (Hardin & Higgins, 1996) both suggest that attitudes and beliefs are formed and transformed through processes of social validation by others. According to these approaches, individuals may be expected to hold stereotypic beliefs to the extent that they perceive relevant others to hold those same beliefs. If the prop of perceived social support were removed, then it is possible that those beliefs would come crashing down. Yet despite the theories’ agreement regarding the importance of social validation, the two perspectives stress somewhat different mechanisms by which perceived consensus might contribute to the formation and transformation of social stereotypes. According to Hardin and Higgins (1996), people create their beliefs through interaction with and verification on the part of others in order to create a "shared reality." Thus, when an individual is faced with uncertainty, he or she may look to the beliefs of others for purposes of information and validation. Because the accuracy of social stereotypes is difficult to assess and socially sensitive, people might be especially likely to be influenced by others when it comes to stereotyping. The notion that people look to others as a way of reducing uncertainty and acquiring knowledge about Consensus Feedback and Stereotype Change 7 the world has a long and distinguished tradition in social psychology (e.g., Asch, 1952; Festinger, 1954), and it is associated with the concept of “informational influence”. It is also possible that people might change their beliefs not as a way of achieving accuracy in perception per se, but rather as a way of stressing identification or affiliation with other ingroup members -a type of “normative influence”. In this case, changes in attitudes, beliefs, and behaviors occur to the extent that the relationship with the person or group is an integral aspect of the person’s self-concept (e.g., Kelman, 1961). Social identity and self-categorization theories (Haslam et al., 1996; Tajfel & Turner, 1979; Turner, 1987) emphasize the significance of social groups in defining personal identities, proposing that individuals are strongly motivated to share beliefs with other members of the ingroup (Abrams & Hogg, 1988; Turner, 1991). Thus, subjective information about the beliefs of others may be effective in changing stereotypes because of needs to identify and affiliate with other people. As Turner (1991) has pointed out, informational and normative processes are not mutually exclusive. It seems likely that people glean more information from trustworthy sources with whom they identify and that people affiliate with others who are valued sources of knowledge and information. The research reported here leads to the conclusion that epistemic and affiliative needs both play a role in consensus effects with regard to racial stereotyping. The primary goal of the present research was to demonstrate the powerful effect that perceptions of consensus have on the personal endorsement of stereotypes. Experiment 1 demonstrates that racial stereotypes may be undermined by presenting people with information that breaks down the presumed consensus surrounding those stereotypes. Experiment 2 replicates the effect and demonstrates that consensus information is more influential among people who report low rather than high confidence in their initial stereotypic beliefs. Experiment 3 demonstrates that people are more influenced by information about ingroup consensus as compared with outgroup consensus and that stereotype consensuality is related to the expression of racial prejudice one week later in a different experimental context. This study rules out alternative explanations for consensus effects, such as demand characteristics and cognitive anchoring. Finally, Experiment 4 confirms that consensually backed racial stereotypes are more resistant to subsequent attempts at persuasion.