As should be clear by now, negative stereotypes can have devastating effects. There have been ample demonstrations throughout history of open and intended, often murderous discrimination arising from negative group stereotypes. The social psychological research documented in this chapter has focused on more subtle influences on judgment whose consequences can nevertheless be quite pernicious. Stereotypes can influence the way members of stigmatized groups are perceived, understood, and treated by others. Even people who have no prejudicial intent can inadvertently view stereotyped individuals through the lenses of negative stereotypes, without even realizing that they are doing so. For their part, stereotyped individuals often assume that they are operating under a cloud of stereotype-based suspicion of inferiority, and this can cause them to experience enough anxiety to undermine their performance.
As you contemplate the vast social costs of mistaken negative stereotypes, you may be wondering if anything can be done to correct such misperceptions. Is it possible to change people’s stereotypes? This question has also intrigued many social psychologists. Their research efforts have given rise to another question, namely, Why is it so difficult to bring about stereotype change? These issues are discussed next.
Early intuitions suggested that if we come to meet members of groups that are unfairly stereotyped in negative terms, we’ll soon recognize and correct our errors. If non-Blacks encounter enough intelligent, hard-working, and mild mannered African Americans, surely they will realize that it is inappropriate to view this group as unintelligent, lazy, and aggressive. These ideas gave rise to the contact hypothesis, which held that social contact between members of majority and minority groups will reduce prejudice. From the start, it was recognized that not any contact will do. Walking through an urban inner city or visiting Chinatown seems highly unlikely to change anyone’s view of the ethnic minorities that reside there. Having an African American janitor seems unlikely to do the trick either. Rather, Gordon Allport suggested in 1954, to reduce prejudice through contact, the contact must be among people of equal status in pursuit of common goals. If, for example, a prejudiced White person has to cooperate at work with an African American teammate, this would eventually lead to liking and respect for this teammate which would then generalize into a more positive view of African Americans.
Although the contact hypothesis seemed intuitively appealing, it proved very difficult to change stereotypes through contact. In their classic Robber’s Cave experiments, Muzafer Sherif and his colleagues created a sense of rivalry and hostility among two groups of boys, and then attempted to change their negative views of each other and eliminate their mutual dislike (Sherif et al. 1961). Sherif and his colleagues recruited well-adjusted middle-class White boys for a summer camp. Initially the boys did not know each other, and were unaware that they were participating in an experiment. After living together in the same cabin for a few days, the boys were separated into two groups. The investigators asked each boy who his best friend was, and made sure that the two would end up in different groups. The two groups were kept completely apart for a while, as each engaged in various activities designed to create group spirit. Once each group had acquired strong group spirit, the investigators initiated a series of activities intended to produce friction among them—a sports tournament, a competitive treasure hunt, and so on. Pretty soon, the boys were referring to members of the other groups as “stinkers,” “sneaks,” and “cheats.” They began showing their mutual animosity by making threatening posters, planning raids, and burning the other group’s banner after a game.
Having successfully created two hostile groups, the investigators began addressing the true purpose of the study—getting rid of the hostility. They began by creating opportunities for pleasant social contact such as watching movies together and eating in the same dining room. This proved to be a total failure. The social events only provided an opportunity for the rival groups to berate and attack each other. They shoved each other in the line for food, threw paper and food at each other, and made rude comments. The investigators then turned to a more elaborate tactic. They created a series of urgent situations that forced the boys to work together. These included a breakdown in the water supply that required the boys to search the water line for the trouble spot, a failure of their truck to start that required the boys to all pull it together with a rope, and an opportunity to rent a movie that the two groups could only afford if they pooled their resources. Although the two groups were able to work together harmoniously on these tasks, their joint efforts did not lead to an immediate reduction in hostility. Initially, as soon as the task was completed, the boys returned to bickering and name calling. But gradually the friction and hostility between the groups was reduced, until, by the time the camp ended, they were seeking opportunities to mingle, entertain, and treat each other.
Recall that the rival groups in this study were made up of essentially similar boys with no prior history of hostility. Even for such groups, intergroup hostility was not reduced through sheer contact on an equal footing in a pleasant situation. And even cooperative contact did not lead to an immediate reduction in hostility. The outlook for groups with a long history of dislike and well-entrenched negative stereotypes of each other seemed even less promising. Indeed, extensive research on the contact hypothesis gives little reason for optimism. For contact to reduce prejudice, the members of the two groups must have equal status, an opportunity to get to know each other, exposure to evidence that disconfirms the stereotypes, shared goals, and active cooperation. And even then, contact is not always effective at reducing prejudice. Negative stereotypes often resist change even in the face of intense manipulations involving cooperation with members of the stereotyped group over long periods of time (for a review, see Stephan 1985).
Why do our group stereotypes so often remain unchanged even in the face of contact with group members whose behavior and attributes are nothing like our stereotypic expectations? One reason is that we may simply not realize that such individuals disconfirm our stereotypes because our interpretation of their behavior can be biased by our stereotypes; we may view their neutral and innocent behaviors as consistent with our negative stereotypes. But even when we recognize that individuals clearly disconfirm the stereotype of their group, we may still fail to generalize from them to the group as a whole because we consider them atypical of their group, as discussed next.
When you encounter a wealthy African American, an aggressive housewife, a respectful teenager, or a member of any other social group who disconfirms your stereotype of that group, you may hang onto your stereotype by “fencing off” this individual (Allport 1954; Rothbart and John 1985; Weber and Crocker 1983). Yes, a White American might say to himself, my African American neighbor Marcus is wealthy and successful, but you can’t really learn anything from him about what African Americans in general are like because he is so atypical of African Americans; he belongs to an unusual subgroup of African Americans, African American executives. Much like one views ostriches as an unusual kind of bird that teaches little about the size and flying capabilities of birds in general, one may view African American executives as an unusual kind of African American that teaches little about the behavior and attributes of African Americans in general. By allocating counterstereotypic individuals like Marcus to a subtype that is considered atypical and unrepresentative of the group as a whole, one may be able to maintain one’s global stereotype of the group even though one knows that some group members do not fit the bill.
Through this mechanism, one may be able to maintain a negative stereotype of African Americans and a strong dislike for this group even though one’s favorite entertainer is Bill Cosby, one’s favorite singer is Michael Jackson, and one’s favorite athlete is Michael Jordan (all of whom, one knows quite well, are African Americans). These liked and respected individuals can be dismissed as irrelevant to the stereotype of the group as a whole if they are viewed as belonging to atypical subtypes such as African American entertainers, African American singers, or African American athletes. In the same vein, the notion of subtyping explains how one can be a prejudiced bigot despite proclaiming that “Some of my best friends are Jews [or Blacks, or members of any other stigmatized group].” The “best friends” are simply fenced off as atypical of their still disliked group.
A landmark article by Renee Weber and Jennifer Crocker suggested that we may be less likely to change our stereotypes when we are confronted with individuals who disconfirm them if we can readily subtype these counterstereotypic individuals (Weber and Crocker 1983). In one study, these investigators attempted to change participants’ stereotype of lawyers as well-dressed, industrious, and intelligent. In one set of conditions, participants read descriptions of 30 lawyers, each described by three sentences that implied different attributes. Participants in all conditions saw the same set of 90 sentences, which included 30 that implied attributes inconsistent with the stereotype of lawyers (e.g., “Larry has difficulty analyzing problems and developing logical solutions” implies unintelligent), 15 sentences that implied attributes consistent with the stereotype (e.g., “Ken often skips lunch and works overtime to finish projects” implies industrious), and 45 sentences that implied attributes irrelevant to the stereotype of lawyers (e.g., religious). In different conditions this identical set of sentences was distributed differently among the 30 lawyers. Most important, in the dispersed condition, the 30 counterstereotypic sentences were dispersed across all lawyers, so each had one disconfirming attribute. In contrast, in the concentrated condition, the 30 counterstereotypic attributes were all concentrated in 10 lawyers, such that this third of the lawyers each had 3 disconfirming attributes, whereas the remaining two-thirds had no disconfirming attributes.
Concentrating the disconfirming attributes within a small subgroup should make it easier to designate this group to an atypical subtype and, thereby, to dismiss its relevance to the overall stereotype. Indeed, participants in the concentrated condition appeared to show less stereotype change than did participants in the dispersed condition, who could not readily subtype the disconfirming individuals. Put differently, participants generalized less from the same examples of unlawyerly behavior when these examples were concentrated in a small subgroup of the observed lawyers than when they were dispersed across the entire sample. Other researchers have since replicated this finding (e.g., Johnston and Hewstone 1992). When we can, we subtype counterstereotypic individuals, and this enables us to maintain our global stereotypes.
It is easier to subtype and dismiss counterstereotypic individuals if they all share another common attribute that provides a good reason for viewing them as atypical. If, for example, the poorly dressed lawyers we encounter are all Black, it is easy to explain why they should differ from lawyers in general—perhaps they are pro bono lawyers, perhaps they come from impoverished backgrounds. With this explanation in mind, it seems reasonable to subtype them as atypical of lawyers as a whole, and it is then unnecessary to generalize from them to other lawyers. Indeed, several studies have shown that people are less likely to generalize from group members who disconfirm the stereotype on one dimension to the group as a whole when all of these counterstereotypic individuals also violate the stereotype on another dimension (Rothbart and Lewis 1988; Weber and Crocker 1983; Wilder, Simon, and Faith 1996). Put differently, individuals who disconfirm the stereotype of their group on one dimension are most likely to lead us to modify our stereotype if they are typical of their group’s stereotype in all other ways. An assertive woman is most likely to provoke people into changing their stereotype of women as unassertive if she seems feminine in all other ways—wears makeup and feminine clothes, is warm and caring, and so on. An assertive woman who, in addition, dresses and behaves in a manner that violates the stereotype of women may provoke less stereotype change because she can so readily be subtyped as atypical and dismissed as irrelevant.
The subtyping of counterstereotypic individuals often results from an active attempt to maintain our stereotypes. We may be motivated to preserve our stereotypes because these help us to justify our social order, our own discriminatory behavior, or our sense of superiority to others (Allport 1954; Fein and Spencer 1997). But we may attempt to dismiss the relevance of counterstereotypic individuals to our stereotypes even in the absence of such motives. An individual who violates a well-entrenched stereotype may seem so surprising and improbable that we may attempt to explain this individual away, much like we do when we come across information that challenges any other expectancy that we harbor (Hastie 1984; Snyder 1984; Wong and Weiner 1981). For these reasons, when we encounter a member of a stereotyped group who violates our stereotype, we may ask ourselves, in effect, “Do I have any good reason for believing that this person is atypical of the group as a whole?” If the answer is positive, we may feel justified in not generalizing from the individual to the group. Recall that when we ask ourselves such one-sided questions, we tend to engage in an equally one-sided search for answers, which often biases us toward confirming our hypotheses (Klayman and Ha 1987; see chapter 3). Therefore, we will often find good reason for dismissing counterstereotypic individuals as irrelevant.
As discussed above, if the counterstereotypic individuals possess another atypical attribute, this provides a good reason for subtyping and dismissing them as unrepresentative of their group. More recent research has shown that the additional attribute characterizing a stereotype-disconfirming individual can facilitate subtyping even if it is not atypical of the stereotype to being with. We may be able to use even neutral information about a counterstereotypic person as grounds for viewing this person as atypical of his or her group (Kunda and Oleson 1995). For example, if we come across an outspoken, assertive woman who violates our stereotype of women as compliant and unassertive, we may feel compelled to revise our stereotype if we have no additional information about her. But we may be able to use any additional information about her as grounds for subtyping and dismissing her as atypical of women in general.
If we know, for example, that this assertive woman had (or did not have) brothers, that her parents were supportive (or unsupportive), that she was attractive (or unattractive), or if we have any other knowledge about her background, we may attempt to use this knowledge to explain why women with this particular attribute are, unlike most women, assertive. We are so good at generating explanations relating just about any attribute to just about any outcome that we should have little trouble coming up with a good explanation (e.g., Andersen and Sechler 1986; see chapter 3). We may theorize, for example, that women with brothers are likely to be assertive because they had the benefit of growing up with assertive role models. Or we may theorize that women without brothers are likely to be assertive because they had the benefit of growing up without being oppressed at home by domineering males. In either case, having successfully assigned the woman to an atypical subtype, we need not generalize from her to other women.
In one study designed to test these ideas, participants read an interview with a lawyer in which he came across as quite introverted, which challenged their stereotype of lawyers as outgoing (Kunda and Oleson 1995). Participants given no additional information about this lawyer did generalize from him to lawyers in general; they rated lawyers, on average, as more introverted than did control participants who had read no interview. A different pattern emerged for two additional groups that were given additional information about this introverted lawyer. One group was told that he worked for a small law firm, the other that he worked for a large law firm. Pretests had shown that lawyers who worked for small or large firms were not expected to differ in their introversion from other lawyers. In other words, participants in these two were given one of two opposite attributes that were both neutral in their implications for lawyers’ introversion.
Although both these attributes were initially neutral, it is not difficult to construct an explanation for why either might make lawyers especially likely to be introverted. One may reason, for example, that either kind of lawyer may be able to get by despite being introverted—small-firm lawyers because they don’t need to deal with many people and large-firm lawyers because in large firms some lawyers may be able to specialize in tasks that require few interpersonal skills. Either attribute, then, may be used as grounds for viewing the introverted lawyer as belonging to an atypical subtype. Indeed, unlike participants given no additional information about the introverted lawyer, those who were also told that he worked for a small or large firm did not feel compelled to generalize from him; their stereotype of lawyers remained unchanged. In sum, participants generalized from a counterstereotypic individual when given no additional information about him. But providing them with an additional neutral attribute sufficed to undercut this generalization. In the process, the neutral attributes also lost their neutrality; they came to be viewed as associated with introversion. Together, these findings suggest that the neutral attributes undermined generalization from the counterstereotypic individual because they served as grounds for subtyping this individual as atypical.
Even if we have no further information that can be used as grounds for subtyping a counterstereotypic person, we may still be able to dismiss this person as an irrelevant exception if the person’s deviation from our stereotype is extreme enough (Kunda and Oleson 1997). Consider Colin Powell, the African American general who orchestrated the successful execution of the Persian Gulf War, or Margaret Thatcher who ruled Britain with an iron fist. These individuals may violate our stereotypes of their groups so extremely that we may feel justified in viewing them as exceptions that don’t prove the rule. We may believe that the sheer extremity of their deviance from our stereotypes provides us with a good enough reason for dismissing them as atypical and irrelevant.
In a series of studies that examined these ideas, participants read about a person who violated their stereotype either extremely or only moderately (Kunda and Oleson 1997). In one study, participants’ stereotype of public relations (PR) agents as extraverted was challenged by a PR agent who was either extremely or only moderately introverted. In another study, participants’ stereotypes of feminists as assertive were challenged by a feminist who was either extremely or only moderately unassertive. Stereotype change was assessed by comparing ratings of the stereotyped group made by these participants to those made by control participants who had not been exposed to any group member. In both cases, the person who deviated from the stereotype only moderately provoked greater stereotype change than did the person who deviated extremely. Indeed, the extremely unassertive feminist provoked no stereotype change at all, whereas the moderately unassertive one did lead participants to view feminists as less assertive. If we want to change people’s stereotypes of African Americans, we are more likely to succeed by introducing them to moderately successful African American individuals than by introducing them to extremely successful ones such as Colin Powell or Bill Cosby. These famous individuals deviate so extremely from their group’s stereotype that people may view the sheer extremity of this deviation as a good enough reason for dismissing their relevance to the stereotype.
It is ironic that the more individuals deviate from the stereotype of their group, the less likely they are to bring about stereotype change. One disturbing implication is that the more inaccurate our stereotype of a group, the less likely it is to change spontaneously following encounters with group members. This is because the more inaccurate our stereotype, the more discrepant it will be from the typical group member. Put differently, the typical group member will deviate more extremely from more inaccurate stereotypes, and so will be dismissed more readily as an exception. This may be one reason why it is so difficult to change racial stereotypes through contact with individuals who disconfirm them (Stephan 1985). These stereotypes may be so inaccurate that average group members are perceived as extreme exceptions to the rule. It also follows that the individuals whose stereotypes one would like most to change, namely those holding extremely inaccurate negative stereotypes, will be the least likely to reduce the negativity of their stereotypes following exposure to group members who disconfirm them. For such extreme perceivers, the average group member will seem like an extreme and irrelevant deviant. Indeed, in these studies, extreme individuals became, if anything, even more extreme in their stereotypes following exposure to individuals who disconfirmed them.
On the positive side, the finding that people who moderately disconfirm perceivers’ stereotypes of their group can provoke stereotype change is encouraging. In everyday life we may be more likely to encounter such moderate violations of our stereotypes than we are to encounter extreme violations. There are far more moderately assertive women than there are Margaret Thatchers. This may be why stereotypes can and do evolve over time.
Despite the difficulty of changing stereotypes through short-term interventions, examination of historical records and of the findings of longitudinal research suggest that over more extended time periods, dramatic changes in stereotypes can occur. Consider a group widely viewed as dirty, drunken, incompetent, brawling, slum-dwellers. You may be surprised to hear that this was the stereotype of the Irish in the United States in the mid-eighteen hundreds, when job ads specifying “No Irish need apply” were common (Sowell 1981). The stereotype of Irish Americans has clearly undergone dramatic change.
Other stereotypes have changed substantially as well. In a series of studies termed “The Princeton trilogy,” different generations of Princeton students were asked to rate several ethnic groups on the same set of adjectives in 1933, 1951, and 1969 (Katz and Braly 1933; Gilbert 1951; Karlins, Coffman, and Waters 1969). Many of the stereotypes appeared to have changed considerably over this period. For example, there was a substantial decline in the extent to which Jews were seen as shrewd and mercenary, and an increase in the extent to which they were seen as ambitious and intelligent. And there was a dramatic decline in the extent to which Blacks were seen as superstitious and lazy, and an increase in the extent to which they were seen as musical. Admittedly, these changes may reflect not only actual changes in stereotypes but also changes in norms toward sanctioning open expression of prejudice (in itself, an encouraging development). Yet, when considered along with historical reports of stereotype change over the long run and experimental demonstrations that stereotype can be changed, these findings point to an optimistic outlook. Groups that are currently stereotyped inappropriately in negative terms may look forward with hope to a future in which they are subject to less prejudice and discrimination. As they gain greater insight into how people use their stereotypes and how they may come to change them, social psychologists may help bring this future about.
Excerpted from Social Cognition: Making Sense of People, Chapter 8, pp. 380-391. © 1999 by The MIT Press, Cambridge, Mass. Reprinted by permission.