WK2

CHAPTER11.doc

CHAPTER 11 STEREOTYPING, PREJUDICE, AND DISCRIMINATION: SOCIAL BIASES

In Freedomland, a white woman claims that an African American man stole her car with her four-year-old son in the backseat. A manhunt focuses on the public housing project near the scene of the crime. Media converge on the neighborhood. Cops surround the project. … Lorenzo Council is a black cop assigned to the case. Living in the community, he knows and cares about the people there. He spontaneously warns some students in the local public school:

“Excuse me,” Lorenzo stepped into the room and exploded, paralyzing his audience. “I got to go, but I would like to give you kids some tips on the next few nights out here. I'd like to give you some tips on survival. No theories, no speeches. Facts.” Prowling the room now, Lorenzo glared at them. “Facts. The police, is angry. The police, is scared. And the ones you're gonna see around here tonight? Tomorrow night? They don't know you. They don't know what's in your head, who your mother is, if you're a good kid, bad kid, all they know is they're living on the edge of their own nerves just like you.” (Price, 1998, p. 484) 1

Here, a black cop is trying to protect his own neighborhood kids (he knows what's in their heads, who their mothers are, who's a good kid or a bad kid). He is trying to protect them from police officers he knows to be “living on the edge of their own nerves” and likely to act on racial biases. Treating black kids as a category—as if they all were criminal and dangerous—constitutes stereotyping, prejudice, and discrimination. Moreover, in this case, the accusation is a fiction; there was no black male car-jacker, but everyone was all too ready to believe the white woman's (desperate but racist) story.

In this chapter, we address both blatant and subtle misunderstandings between groups of people, the effects of this bias on people's lives, and strategies for change. Just as aggression research reveals the basic social psychology of processes that can prove deadly, the research on bias reveals processes that are demonstrably unhealthy and sometimes deadly for both agents and targets of bias.

Research on bias is a hot topic, in more ways than one. People have strong opinions about the fact of bias: Some think bias is a thing of the past—not really a problem any more—and others think it is a real and present danger to targeted social groups. As a result of these differences of opinion, people worry about discussing bias for fear of being misunderstood or from irritation about having to listen to the same old insulting, half-baked arguments. Bias is a taboo topic in many campus contexts. People are personally affected by bias, no matter what their category, so feelings run high.

Bias is a hot topic among researchers as well. In the last years, about a third of the talks at social psychology's national and international conferences have addressed bias and intergroup relations. Submissions to one of the field's leading journals were dominated by work on stereotyping, prejudice, and intergroup relations (Suls, 2001). Clearly this topic matters to people. Fortunately, social psychologists have studied it for decades (80 years) and have learned quite a lot about both subtle and blatant biases, their effects on targets, and strategies for change. The core social motives that move through all kinds of intergroup bias include belonging, understanding, controlling, and self-enhancing, as the research indicates. But first, some definitions.

WHAT ARE PREJUDICE, STEREOTYPING, AND DISCRIMINATION?

Conceptual Definitions

As social psychologists often separate cognition, affect, and behavior, so, too, intergroup bias researchers find it useful to distinguish among the legs of this tripod (Fiske, 1998b). All forms of bias involve category-based responses, reacting to another person as an interchangeable member of a social group. People have the clearest category-based responses to members of outgroups, that is, groups to which they do not personally belong. Category-based responses to outgroups typically are more negative than responses to the ingroup, one's own group. Category-based responses include stereotyping, prejudice, and discrimination.

Stereotyping entails applying to an individual one's cognitive expectancies and associations about the group. As such, stereotypes represent one kind of expectation (covered in Chapter 4 , Social Cognition; also see Dovidio, Brigham, Johnson, & Gaertner, 1996; Stangor & Lange, 1994). Recall that an expectation is a coherent concept or naïve theory that renders the world manageable. A cognitive structure, it comprises the attributes of a concept and the relationships among the attributes. As a specific kind of expectation, stereotypes are beliefs about the characteristics of group members and theories about why those attributes go together (Hilton & von Hippel, 1996). Stereotypes are fixed ideas that accompany a category (Allport, 1954b). From a functional perspective, stereotypes justify (or rationalize) our affective and behavioral reactions to the category (Jost & Major, 2001). From a cultural perspective, they embody a societal consensus, a collective belief system (Stangor & Schaller, 1996). Modern definitions do not assume that stereotypes are either accurate or inaccurate, which is an empirical issue we will address.

Prejudice entails reacting emotionally to an individual on the basis of one's feeling about the group as a whole. “The net effect of prejudice … is to place the object of prejudice at some disadvantage not merited by his [or her] own misconduct” (Allport, 1954b, p. 9). One wag defined prejudice as “A vagrant opinion without visible means of support” (Bierce, 1911, p. 264). Many researchers define prejudice as an overall attitude (including affect, cognition, and behavioral correlates; e.g., Dovidio et al., 1996). However, the core component of an attitude, as Chapter 6 showed, is the evaluation of the attitude object, in this case, outgroup members, so it is useful here to define prejudice by emphasizing the feelings component. Prejudices theoretically include both positive and negative affective reactions, but most research focuses on the negative. Nevertheless, even positive prejudices can place the target at a disadvantage, as we will see.

Discrimination entails acting on the basis of one's stereotypes and prejudices, denying equality of treatment that people wish to have. Allport (1954b) defined steps in the rejection of outgroups: verbal rejection, avoidance, segregation, physical attack, and extermination. Although verbal discrimination may not seem serious, stereotypical slurs, jokes, and put-downs create a hostile climate that enables even more serious forms of hostility to be enacted, not to mention offending the targets of such insults. Even if the people making the biased comments would not themselves dream of going further, the verbal disrespect creates norms that enable others with poorer judgment to feel that their own more overt forms of discrimination would be condoned. Avoidance also may not seem serious, but when people stay with their comfortable ingroup, they exclude members of other groups. And they certainly do not learn anything new. Segregation, attack, and group extermination are, without question, forms of discrimination. Later sections will address these complex issues in detail.

Conceptually separating stereotyping, prejudice, and discrimination proves useful because they are correlated, but not redundant, forms of bias. Consider all these plausible combinations: People can discriminate because of either hot prejudices or cold stereotypes. People can hold mental stereotypes but not act on them. People can have strong feelings but few supporting beliefs. Thoughts and feelings are not damaging until they are enacted.

One analysis across studies found the relationships shown in Figure 11.1 , indicating that individual differences in emotional prejudice correlate with discrimination better than stereotypes do (Dovidio et al., 1996). A meta-analysis of experimental studies also found that affective prejudices predict racial discrimination far better than stereotypic beliefs do (Talaska, Fiske, & Chaiken, 2009). Although the relationship between prejudice and discrimination is moderate, its size is comparable to the general attitude-behavior relationship. As in that domain, so too in the intergroup area: Many factors determine when people act on their thoughts and feelings (Mackie & Smith, 1998). For example, attitudes have to be activated to predict behavior, and, likewise, stereotypes and prejudice have to be activated to predict discrimination. Similarly, various motivations determine when people do and do not enact their stereotypes and prejudices, as we saw in the attitudes chapter. In any event, because they are not redundant, it is useful to separate them conceptually.

Figure 11.1 Relationships among Individual Differences in Stereotyping, Prejudice, and Discrimination

Source: From data reported by Dovidio et al., 1996.

Collectively, stereotyping, prejudice, and discrimination all are biases because treating the individual as an exact representation of the group is never accurate. The individual is never identical in all respects to one's image of the group, and no individual exactly represents in all respects the average of the group. Even if the judgment were treated as probabilistic, people rarely put odds on their statements about outgroup members' characteristics. A later section returns to issues of accuracy and error, but stereotypes, prejudice, and discrimination all are biases, in at least the sense of anchoring one's perceptions of an individual based on perceived group averages. That being said, group membership is relevant to behavior, as when it defines particular roles (e.g., police versus gang members) or helps distinguish people (Wegener & Klauer, 2004). Some theories, such as self-categorization theory, covered later, take this perspective.

In addition to biases against people as group members, society designates certain people as marked by an individual stigma, discredited and not fully human (Crocker, Major, & Steele, 1998; Goffman, 1963; Heatherton, Kleck, Hebl, & Hull, 2000; Jones et al., 1984). Physical and mental disabilities, for example, are stigmatizing. Stigma and category-based biases both entail prejudice, stereotyping, and discrimination.

Operational Definitions

As this chapter shows, researchers operationalize bias in a number of ways (Fiske, 1998b). Laboratory measures of discrimination (e.g., in studies of interracial aggression) begin with varying combinations of verbal hostility: written ratings that disparage an outgroup individual or category, negative outgroup attitudes on questionnaires. At the subtle behavioral level, laboratory studies measure nonverbal indicators of hostility, such as seating distance or tone of voice. Related nonverbal measures include coding overt facial expressions, as well as measuring minute, nonvisible movements in the facial muscles that constitute the precursors of a frown (vs. a smile). Measurements of brain activity in emotion centers (such as the amygdala, for vigilance) are being used as well (Amodio, 2008). Equally subtle measures include speed of response to stereotypic associations, memory biases for stereotype-confirming information, stereotypic interpretations of ambiguous information, and stereotypic distortion in judgments.

Moving up a level, measures of discriminatory avoidance include participants' choice to associate or work with an outgroup member, volunteer to help an organization, or directly aid an outgroup member who requests it. In a laboratory setting, segregation can be measured by how people constitute small groups or choose leaders for small groups. Finally, aggression against outgroups can be measured in laboratory settings by competitive games or teacher-learner scenarios in which one person is allowed to punish the other, an outgroup member, with low levels of shock, blasts of noise, or other aversive experiences (that is, measuring sanctioned forms of aggression, Chapter 10 ).

Surveys outside the laboratory measure racial attitudes by distinguishing old-fashioned bigotry from modern, subtler forms of bias. Table 11.1 shows one such scale; the items focusing on prejudice are typically hidden among a variety of other political attitudes. Other scales assess specific motivations regarding prejudice, as well as many specific kinds of racism, sexism, ageism, heterosexism, and more.

TABLE 11.1 Old-fashioned and Modern (Symbolic) Racism Scales (Selected Items)

Old-Fashioned Racism Scale

It is a bad idea for blacks and whites to marry one another. Black people are generally not as smart as whites. If a black family with about the same income and education as I have moved next door, I would mind it a great deal. It was wrong for the United States Supreme Court to outlaw segregation in its 1954 decision.

Symbolic Racism Scale

Irish, Italian, Jewish, and many other minorities overcame prejudice and worked their way up. Blacks should do the same. How much of the racial tension that exists in the United States today do you think blacks are responsible for creating? (1–4: all of it to not much at all) How much discrimination against blacks do you feel there is in the United States today, limiting their chances to get ahead? (1–4: a lot to none at all) Over the past few years, blacks have gotten more economically than they deserve.

(1–4: strongly agree to strongly disagree), except as shown

Core Social Motives

UNDERSTANDING

People maintain biases toward other people partly in order to make sense of intergroup encounters. As Chapter 4 (Social Cognition) indicates, people understand other people in large part by reference to familiar categories, expectations, and concepts. Social categories defined by (e.g.) age, gender, ethnicity, class, and disability carry well-worn baggage that guides people's interactions. Because people are complex stimuli and because outgroups are less familiar than ingroups, people rely heavily on prior beliefs in intergroup encounters. This cognitive miser perspective (Fiske & Taylor, 2013; Chapter 4 ) emphasizes the role of stereotypes as resource-saving devices.

Research supports the understanding motive as one factor in stereotype usage. For example, people code other people by social categories, such as gender, race, and age, and this process can be useful for identifying them. However, it can go awry when people confuse two people within the same category (e.g., confusing two women with each other or two black people with each other). A paradigm that came to be known as “who-said-what” (Taylor, Fiske, Etcoff, & Ruderman, 1978) showed that people make more confusions within category than between categories. Although errors embarrass the perceiver and humiliate the target, this tendency to treat category members as equivalent appears across a range of social categories, including even religion (Weeks & Vincent, 2007; see Fiske, 1998b for a review and Klauer & Wegener, 1998, for methodological refinements); the range shows how useful it is to understanding other people.

Using stereotypes does simplify the processing of information about other people. In one study (Macrae, Hewstone, & Griffiths, 1993), people watched a videotape of a young woman discussing her lifestyle. Some observers learned that she was a hairdresser and some that she was a doctor, and then the video depicted her as engaging in a mix of behavior. Some behavior fit the hairdresser stereotype (enjoys clubs; wears mini-skirts; has a chauvinistic boyfriend) but not the doctor stereotype, and some fit the doctor stereotype (interested in politics; attends opera; drives a fast car) but not the hairdresser stereotype. When people concentrated on the video without distraction, they recalled the inconsistent information, presumably because it was puzzling and they had thought about it (see Chapter 4 ). However, when people had to rehearse an eight-digit number while they watched the videotape—that is, when they were slightly distracted—they showed the opposite pattern of recall. When they were busy, they recalled more stereotype-consistent than inconsistent information. Thus, under the busy conditions of ordinary interaction, people can save cognitive resources by using stereotype-consistent information. Meta-analysis indicates that this stereotypic memory bias is a general phenomenon under the complex circumstances of normal interactions (Stangor & McMillan, 1992). This supports the role of the understanding motive for stereotype usage.

Several other studies indicate that people typically rely on stereotypes more when they are mentally busy or overloaded, making judgments on that basis (e.g., Bodenhausen & Wyer, 1985; Bodenhausen & Lichtenstein, 1987; van Knippenberg, Dijksterhuis, & Vermeulin, 1999). People demonstrably do have more online mental resources when they use stereotypes than when they do not (Macrae, Milne, & Bodenhausen, 1994).

What's more, some people have a chronic orientation to the social world as full of fixed entities, for instance, believing in people with essential traits (as opposed to malleable personalities that develop incrementally). Entity theorists focus not only on fixed personality traits (as described in Chapter 3 ), but also on stereotype-consistent information (Plaks, Stroessner, Dweck, & Sherman, 2001). Stereotypes can form when entity theorists view the social world as amenable to a fixed, stable kind of understanding. Cultures differ in how much they encourage entity theories and thus intergroup stereotyping (Levy, Plaks, Hong, Chiu, & Dweck, 2001) as a strategy for understanding.

BELONGING

People maintain biases toward outgroups partly to cement ties within their ingroups. As we will see, much of prejudice concerns ingroup preference more than outgroup derogation. For example, students often self-segregate in dining halls and cafeterias (Tatum, 1997); this most likely reflects ingroup comfort and favoritism, at least as much as outgroup discomfort, and not at all necessarily outgroup derogation (Brewer, 1979; Mummendey, 1995). This ingroup bias is especially strong among those who value their ingroups (Branscombe & Wann, 1994; Crocker & Luhtanen, 1990), which fits the role of the belonging motive in intergroup bias.

Even more in keeping with our analysis of core social motives, people apparently stigmatize others who threaten the effective functioning of their group (Neuberg, Smith, & Asher, 2000; Schaller & Neuberg, 2012; see also Kurzban & Leary, 2001). That is, just as we have argued that the primary evolutionary context is social—the ingroup—so too should people both try to survive within the group and reject those who undermine that group. From this follows the idea that certain categories of people will be universally stigmatized: The nonreciprocator includes anyone who fails to exchange fully or share cooperatively within the ingroup; thieves and people with disabilities, respectively, do not and may not reciprocate. The treacherous are those who break the bonds of ingroup trust: liars, cheaters, and traitors all thrive on ingroup deception. The countersocializers are those who undermine ingroup values; deviants threaten the mainstream socialization processes that transform newcomers into cooperative group members. The diseased threaten group survival, so people are disgusted by potential contaminators. Finally, of course, the outgroup lies beyond ingroup boundaries by definition. All of these threats to the ingroup are derogated, which fits with the importance of the belonging motive.

Groups are importantly defined by shared goals (see Chapter 12 ), so outgroups presumably have goals that differ from the ingroup's goals. If they do not share the ingroup's goals, the outgroup's goals may actively interfere with the ingroup's goals. At a minimum, they do not facilitate the ingroup's goals. Goal blockage leads to negative emotions (as we saw in Chapter 8 , close relationships). By this argument (Fiske & Ruscher, 1993), then, outgroups, precisely because they are not the ingroup and do not share its goals, will elicit prejudice. Belonging—or not—lies at the core of anti-outgroup biases.

CONTROLLING

Avoiding threat also underlies responses to outgroups. Assuming that people typically are motivated to avoid danger and to have a sense of control over their outcomes, outgroups pose a problem in at least two respects. First, because they are novel, unfamiliar, and unpredictable, relative to the ingroup, they undermine one's sense of control (Fiske & Ruscher, 1993). Loss of control generates anger and resentment.

Second, people tend to pin danger on people who are different: People outside the ingroup represent a variety of tangible (concrete) or symbolic (abstract) threats (Schaller & Neuberg, 2012; Stangor & Crandall, 2000). These include (a) resource conflict (with outgroups or with ingroup traitors), (b) dangers to physical health (from someone who is contagious), (c) symbolic threat to bodily integrity or mortality (from someone physically deformed or dying), (d) violations of a just and fair world (from apparent innocents who suffer), and (e) moral undermining (from deviants). Common kinds of stigma fit these descriptions: people with leprosy, facial damage, terminal illness, accidental disabilities, and homosexual orientations all face rejection, perhaps on the basis of tangible or symbolic threat. People want to preserve the ingroup or at least have the illusion of protecting themselves from these perceived threats (Fritsche, Jonas, & Kessler, 2011).

Ingroups thus seem to control danger by differentiating outgroups from the ingroup and then communicating those differences to each other. This social sharing of stereotypes creates a sense of ingroup prediction and possibly control over external threats. Controlling via stereotypes also protects the ingroup's superior position, maintaining status and power; system justification makes up another important feature of this motive, as we will see.

ENHANCING SELF

The simple idea is that people make themselves feel better by derogating people from outgroups. The simple idea is wrong. First, people are more likely to favor the ingroup than disfavor the outgroup (Brewer & Brown, 1998), so the direction of the most frequent bias is ingroup favoritism, not outgroup derogation.

Second, even ingroup favoritism does not necessarily make people feel better (Brown, 1995; Fiske & Taylor, 2013). What appears true is nevertheless relevant to self-enhancing motives: Short-term threat to self does matter. In a given situation, people who have just experienced threat indeed do derogate outgroup members (Wills, 1981). Being insecure or anxious enhances stereotyping (Wilder & Shapiro, 1989). Threats to self-image facilitate even the automatic activation of stereotypes (Spencer et al., 1998).

One class of threats to self-concept includes reminders of one's own ultimate death, that is, mortality salience. When people confront their own death, they experience existential anxiety (Solomon, Greenberg, & Pyszczynski, 1991). One can bolster one's vulnerable self-image by subscribing to enduring cultural worldviews that will survive one's own limited existence. Recall from Chapters 9 and 10 that mortality salience bolsters prosocial and antisocial behaviors that fit enduring values. The social order, consisting of some groups dominating other groups, can also serve this function. And as we will see, mortality salience indeed intensifies both ingroup favoritism and outgroup derogation. But this process actually may not thereby enhance the self.

Finally, the direction of the hypothesized enhancement process may differ from people's intuitions. Rather than people with low self-esteem favoring the ingroup in order to feel better, it is people with already high (but threatened) self-esteem who favor the ingroup (Crocker et al., 1998). The relationship between self-enhancement and prejudice is not as simple as it seems, so we will come back to it later. The bottom line is that short-term threats to self matter in combination with chronic tendencies to self-enhancement.

Summary of Definitions and Motives

Category-based biases include reactions in favor of one's own ingroup and against outgroups (members of groups not one's own). They take the form of stereotypes (cognitive biases), prejudice (affective biases), and discrimination (behavioral biases). All are biases because—at a minimum—the perceiver treats the other individual as an interchangeable member of the category. Stigma is a more specific term that addresses people who are individually marked for disfavor by society.

Operationally, biases have been measured by subtle means in the laboratory: verbal and nonverbal behavior, neural patterns, cognitive associations, and behavioral choices about whom to help, hurt, and befriend. Surveys tend to operationalize bias by a variety of measures getting at both old-fashioned bigotry and modern, subtle forms.

Several core social motives enter into bias. Understanding motives encourage simplified, efficient stereotypic understandings of others. Belonging emphasizes loyalty to the ingroup and, by extension, at least relative neglect of the outgroup. Belonging also singles out those who undermine ingroup cohesion, from outside or inside. Controlling focuses even more specifically on protecting the ingroup from a variety of threats. Enhancement increases the biases of those who have just experienced immediate threat to the self, especially if they have high self-esteem.

SUBTLE BIAS: (MIS)UNDERSTANDING OTHERS BUT ENHANCING SELF

People's endorsement of stereotypes has decreased dramatically over the last 80 years. For example, students rating ten national and ethnic groups four times over the 20th century are producing ratings that are now more moderate (Bergsieker, Leslie, Constantine, & Fiske, 2012; see Figure 11.2 ); the most negatively rated groups in 1933 (non-Europeans, including “Negroes,” Turks, Japanese, and Chinese) are now neutral and even slightly higher than the most positively rated group in 1933 (Americans), whose positive evaluation has also moderated; European outgroups on average have remained relatively neutral. What's more, the students have increasingly refused to report any stereotypes at all. Whether due to changing cohorts or contexts, overt endorsement has decreased.

Figure 11.2 Moderation of Extreme Stereotypes in the Princeton Quartet

Whites' overall understandings of racial issues, as reported on national surveys over 60 years, have also moved toward tolerance (Bobo, 2001). On items concerning principles of equality and integration, respondents have gone from 54% believing that public transportation should be segregated to virtually no one endorsing that view; similar trends occur for respondents believing that whites should receive preferences over blacks for jobs. So few people endorsed the intolerant sides of these items that they have been dropped altogether from national surveys. On the issue of school integration, the trend has been slower but clearly improving, with fully 68% endorsing school segregation in the early 1940s, down to a mere (but still shocking) 4% in 1995. The more public the arena and the more abstract the principle, the more marked the improvement. For example, in the early 1960s, only a third of whites believed that blacks and whites should be allowed by law to marry each other; by 1995, four of every five whites believed in marital choice. But that means that a full fifth of whites still believed that blacks and whites should be actually prohibited by law from marrying each other.

On another question, however, only two-thirds of whites actively favored intermarriage. This begins to suggest that when racial matters get up close and personal, people are less egalitarian. Other survey evidence supports this white resistance to endorsing all the consequences of full implementation of equality as compared with abstract principles. For example, virtually all whites now report being willing to live next door to a black family, but roughly 70% report that they would move away if blacks came to their neighborhood in “great numbers.” Fifty percent of whites say they would object if their children went to a school where more than half of the children were black. Only about 15% believe that the government should help blacks improve their living standards, because of past discrimination. In short, whites appear more supportive of equal rights in principle than of equal rights in practice, when they require commitment to specific steps involving their own lives and the status of their own group. This suggests layers of support and resistance.

Moreover, the disjunction in survey results fits people's individual behavior. When people have excuses, they will discriminate racially, but only in subtle ways. About the same time as the improved survey results, more unobtrusive and subtle measures of discrimination and prejudice showed that white bias was more prevalent than surveys indicated (Crosby, Bromley, & Saxe, 1980). At that time, a review of interracial helping, aggression, and nonverbal behavior indicated that (a) whites tended to help whites more often than they helped blacks, especially when they did not have to face directly the person in need of help; (b) under sanctioned conditions (e.g., in competitive games or administering punishment), whites aggressed against blacks more than against whites, but only when the consequences were low (under conditions of no retaliation, no censure, and anonymity); (c) studies of white nonverbal behavior indicated a discrepancy between verbal nondiscrimination and nonverbal hostility or discomfort, betrayed in tone of voice, seating distance, and the like. This early review sparked the realization that modern forms of discrimination have taken subtle, covert, and possibly unconscious forms. Current data support this insight.

Most estimates place 70% to 80% of the white population as relatively high on modern, subtle forms of racism. Most people hold prejudices that are subtler than the kinds of blatant prejudices we usually imagine. As the ensuing sections show, subtle prejudice is cool and indirect. It's automatic, unconscious, and unintentional; it's ambiguous; and it's ambivalent (Fiske, 2002). All of these factors make subtle prejudice hard to notice, but its effects are clear, as we shall see. The culture and its norms promote subtle prejudice, which is more cognitively cool and indirect than blatant bias, which is more emotionally hot and direct, covered later.

Cool and Indirect Biases: Modern Racism and Subtle Prejudice

The possibility of something new, subtle, and covert appears in the latest results from scales designed to assess modern racial attitudes (see Table 11.1 ). If you have not already done so, take the scale before reading on. White students' scores on this modern racism scale are not related to direct, personal, negative experiences with blacks (Monteith & Spicer, 2000). Their attitudes do correlate with general anti-egalitarian sentiments, as well as with negative themes in open-ended essays about their racial attitudes (e.g., negative stereotypes, denial of own personal responsibility for racism, and believing blacks are getting more than they deserve). Among adults, modern racism correlates with voting for white over black mayoral candidates, being opposed to busing of school children to achieve integration, having antiblack feelings, lacking sympathy for underdogs in general, and lacking education (McConahay, 1986). People high on modern racism construct intergroup relations as a zero-sum competition. This belief system fulfills an understanding motive.

Subtle prejudice is not a uniquely American, white-on-black phenomenon (Pettigrew, 1998b; Pettigrew & Meertens, 1995). In Europe, subtle forms of prejudice operate against various kinds of outgroups: French against North Africans, British against South Asians, Germans against Turks. In both Europe and the United States, people score higher on subtle prejudice, in general, than they score on blatant prejudice, so far more people would be classified as holding subtle (modern) than blatant (old-fashioned) prejudices. Let's examine subtle prejudice in some detail, precisely because it is so common, yet so unknown to most people.

Subtle prejudice is, first and foremost, indirect. When people hold subtle prejudices, they do not come out and say they are biased. The items measuring subtle bias all blame the outgroup. For example (Pettigrew & Meertens, 1995), some questionnaire items say that outgroups (e.g., for whites: African Americans, Asians) should not push themselves where they are not wanted, but at the same time they should try harder. (Exactly how is an outgroup member supposed to do that—not be too pushy but, on the other hand, try harder? Contradictions are inherent in prejudice.) As another example of being indirect, another scale of discrimination and diversity issues (Wittenbrink, Judd, & Park, 1997, 2001a) includes items claiming that blacks do not take responsibility and that they blame the system too much, as well as faulting them for maintaining their own cultural identity ( Table 11.2 ). Such responses tend to correlate with stereotyping.

TABLE 11.2 Discrimination and Diversity Scales (Illustrative Items)

Discrimination Items

Members of ethnic minorities have a tendency to blame whites too much for problems that are of their own doing. These days, reverse discrimination against whites is as much a problem as discrimination against blacks. A primary reason that ethnic minorities tend to stay in lower-paying jobs is that they lack the motivation required for moving up. In the United States, people are no longer judged by their skin color.

Diversity Items

There is a real danger that too much emphasis on cultural diversity will tear the United States apart. The desire of many ethnic minorities to maintain their cultural traditions impedes the achievement of racial equality. The establishment and maintenance of all-black groups and coalitions prevent successful racial integration.

Items are used with 5-point Likert scales (endpoints are strongly agree and strongly disagree).

Subtle forms of racism also correlate with covert forms of aggression. In one study, white and black participants played a competitive game. The white participants delivered lower-intensity noise bursts to black losers than white losers, presumably for fear of appearing overly racist. But white people who scored high on modern racism delivered longer noise bursts to their black opponents than to comparable white opponents, a more covert form of aggression (Beal, O'Neal, Ong, & Ruscher, 2000).

Subtle prejudice also exaggerates cultural differences, viewing the outgroup's linguistic, religious, and sexual practices as completely and utterly different (Pettigrew & Meertens, 1995). Subtle prejudice tends to exaggerate differences between groups and compress differences within groups, as we will see. These kinds of distinctions come in the context of “us” and “them,” ingroup and outgroup, however, the culture constructs those dimensions: “The outgroup members are all the same and different from us.” So prejudice against the outgroup is justified because they are not like us, and they are to blame for their problems.

The perception that the outgroup differs fundamentally from the ingroup tends to reify (treat as if real) what are actually social constructions. That is, intergroup stereotypes depend on local custom. For example, different cultures and time periods define race and ethnicity differently, which belies perceptions that race and ethnicity are somehow objective. As noted, in the United States, the history of slavery gave rise to laws in many states that defined anyone with as little as 1/16 African heritage as black (Banks & Eberhardt, 1998). In Hannah Craft's (2002) The Bondswoman's Narrative, one unscrupulous character becomes wealthy by exposing the distant African ancestors of apparently white women, who are then sold into slavery, abandoned by their humiliated (presumably purely white) husbands. Clearly, the racial definition is biologically arbitrary (why is someone with half-African and half-European heritage typically viewed as black, not white?).

Nevertheless, once people categorize other people into separate groups, they perceive the groups to possess entitativity, the property of being a coherent, unified, meaningful object (Campbell, 1958). As the groups chapter ( Chapter 12 ) indicates, aggregates of people are perceived as groups to the extent that they seem similar and share a common fate, which makes them seem like a social entity to themselves and other people. When groups are viewed as entities, people believe in their essentialism, an underlying, deep core that makes their group membership seem real (Rothbart & Taylor, 1992; Yzerbyt, Corneille, & Estrada, 2001). This perceived essence often takes the form of a belief in unambiguous biological properties (genetics, “blood,” or nature) that define the group. People do not typically recognize the extent to which group boundaries are not natural categories (like species) but instead social constructions such as race.

One implication of people's belief in a group essence is that they look for similarities among group members, overinterpreting them. Another implication is that they interpret the group's outcomes in terms of the group's own intrinsic characteristics, instead of, for example, their situation. This group-level analog to the fundamental attribution error or correspondence bias ( Chapter 3 ) has been dubbed the ultimate attribution error (Pettigrew, 1979). People tend to view good actions as intrinsic to the ingroup's essence (or disposition) and bad actions as fundamental to the outgroup's essence; conversely, the ingroup's bad actions and the outgroup's good actions are viewed as meaningless products of the situation. For example, women's success at traditionally male tasks is often explained away as due to luck or unusual effort, whereas men's comparable success is often attributed to their intrinsic dispositions, such as ability or ambition (Deaux & Emswiller, 1974; Swim & Sanna, 1996). In interethnic attributions, people protect the ingroup, attributing success and high status to internal factors (our group's stable essence) but any failures to unstable or external factors (Hewstone, 1990). These attributional patterns of credit and blame do indirect damage to the outgroup, in the service of protecting the ingroup. Only the ingroup is given the benefit of the doubt.

If one blames a person or a group for their own bad outcomes, one tends to feel angry and resentful, but if one views the bad outcomes as beyond their control, one feels sympathy and pity (as Chapter 9 discussed). This results in another way that subtle prejudice is cool and indirect: It withholds positive emotions from outgroup members—not being sympathetic, not admiring them. The ingroup/outgroup distinction means that one has positive associations to the ingroup and less-positive associations to the outgroup. The perceptions of the outgroup as negative are less clear, but perceptions of the outgroup are relatively negative because they are less good. These kinds of judgments tend to favor the ingroup so that one rewards the ingroup but not so much the outgroup. For example, one offers a job more readily to an ingroup member. So, subtle prejudice is largely a matter of indirectly neglecting outgroup members.

Explicit measures of subtle racism on attitude surveys all correlate with each other and with overt forms of discrimination (Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997). However, assessing indirect, cool, subtle forms of prejudice still presents a challenge. Although prejudices are often assessed with attitude scales such as those in Tables 11.1 and 11.2 whites' responses to these scales can sometimes be reactive, that is, reflect people's motives to answer in socially desirable ways. For example, people score lower on the scale when administered by a black experimenter in a one-on-one setting, as opposed to a white experimenter or a mass-testing session (Fazio et al., 1995). Ruling out such potential problems—by going beyond self-report—affords an opportunity to study even more subtle forms of subtle bias.

Automatic Biases: Categorization and Associations

Automatic biases reflect what Richard Price, the author of Freedomland, calls “the American flu”:

I wanted to create a woman who—whose credentials as a “liberal” are four-star. … She works in the projects. She works with the children in the projects. She's got … a history of interracial dating. As much as it's possible, for somebody from her town to embrace the life of the other, she has done this. Her sense of being an outcast has probably made this very appealing for her.

And yet again, she—when it hits the fan—that, you know, that kneejerk reaction “a black guy did it” just pops right out of her mouth. I mean, she's an American. It's an acculturated panic button. She knows people are going to buy it. And I went out of my way to create somebody whose got sort of superlative credentials, so to speak … as much as anybody could be, she's an honorary sister as far as everybody's concerned.

Doesn't make a difference. It's the American flu. 2

Automatic biases are not just the American flu; they are a universal virus. Indeed, even bacteria distinguish “us” from “them” (Gibbs, Urbanowski, & Greenberg, 2008)! Subtle prejudice relies on automatic categorization, activation, and application of associations. As Chapter 4 noted, responses are wholly automatic when they occur without intention, awareness, effort, or control. In the first moment of encountering someone, one can see, automatically, some categories about that person. Subtle prejudice builds on this kind of automatic perception process. People are extremely good at categorizing, within a fraction of a second, people's race, sex, and age, based on visual cues that are highly practiced (Fiske, 1998b). People do not always categorize on these dimensions (Quinn & Macrae, 2005), but when they do, it can happen faster than an instant (Ito & Urland, 2003). These categories often activate stereotypic associations, which may then be applied to the individual. As Chapter 6 noted, many attitudes operate automatically or at least implicitly, especially in socially sensitive domains such as prejudice.

This kind of automatic bias appears for subliminal cues, that is, stimuli presented so rapidly that people are not aware of what they have seen. Nevertheless, these subliminal stimuli, presented at an unconscious level, affect people's perceptions, feelings, thoughts, and behavior. When people see a subliminal outgroup photograph, or a subliminal ethnic group name, or even “us” versus “them,” people have distinctly positive associations to the ingroup. Their reactions occur within milliseconds (also see Chapter 6 ).

To demonstrate automaticity, psychologists frequently use priming methods, described earlier (Fazio & Olson, 2003). For example, some of the earliest studies (Dovidio, Evans, & Tyler, 1986; Gaertner & McLaughlin, 1983; Perdue, Dovidio, Gurtman, & Tyler, 1990) showed that white people read and identify positive attributes (such as “smart”) faster just after seeing the word “whites” than after the word “blacks.” The positive associations to “whites” primed them for the positive words that followed. Similar effects occur for “us” and “them” as primes. The speed of these cognitive associations is a reliable measure over time (Kawakami & Dovidio, 2001).

Neuroimaging data also suggest automatic race-based reactions (Hart et al., 2000; Lieberman, Hariri, Jarcho, Eisenberger, & Bookheimer, 2005; Phelps et al., 2000; Wheeler & Fiske, 2005). In the United States, when white people look at faces of unfamiliar black Americans—but also sometimes when blacks look at faces of unfamiliar white Americans— researchers observe differential activation in the amygdala area of the brain, especially for people who score high on subtle prejudice. The amygdala is, among other things, associated with vigilance. This race effect also occurs for dark-skinned white targets (Ronquillo et al., 2007), a result consistent with greater bias against darker-skinned blacks (Maddox, 2004) but also whites with Afrocentric features (Blair, Judd, & Fallman, 2004). This bias operates automatically (Blair, Judd, Sadler, & Jenkins, 2002). More generally, the tight perceptual link between prototypic black male faces and crime (Eberhardt, Goff, Purdie, & Davies, 2004) helps explain the amygdala (vigilance) bias.

Race effects on early perceptual and neural activity have important consequences. Recall the shooter bias, whereby people, even police, fire faster on armed black than white male targets and are slower to avoid firing on unarmed blacks than whites (Correll, Park, Judd, Wittenbrink, Sadler, & Keesee, 2007; Payne, 2001, 2006); the shooter bias results directly from split-second neural responses (event-related brain potentials or ERPs; Correll, Urland, & Ito, 2006). Perhaps these automatic biases explain why having more Afrocentric features predicts death sentences, even controlling for features of the crime, for both black (Eberhardt, Davies, Purdie-Vaughns, & Johnson, 2006) and white defendants (Blair, Judd, & Chapleau, 2004). The automatic perceptual biases correspond to centuries of cultural images linking Africans to dehumanizing images such as apes (Goff, Eberhardt, Williams, & Jackson, 2008). Other severely derogated outgroups—for example, homeless or injection-drug-using people, regardless of race—also receive dehumanizing perceptions and neural responses (Harris & Fiske, 2006).

However, even these automatic, neural-level biases are amenable to early-onset efforts at control. For example, the race-amygdala effect depends on speed of exposure; at the briefest exposures (30 msec.), the amygdala activates, but at longer exposures (half a second), control-related neural components come into play (Cunningham et al., 2004). Similarly, instructions asking perceivers to individuate—that is, to attend to the individual preferences of the pictured person—also eliminate the race-amygdala effect (Wheeler & Fiske, 2005) and rehumanize the homeless and drug-addicted (Harris & Fiske, 2007). And perceivers habituate quickly, dampening the amygdala response to familiar black faces (Phelps et al., 2000). The point is that many, perhaps most, people have relatively automatic category-based responses that are unbelievably rapid, but of course, rapid does not mean inevitable, as Chapter 6 shows.

So far, we have seen automatic biases revealed by categorization and association, along with subliminal cues, conscious priming, and neural responses, all of which avoid the self-report issues of deliberate responding. Like it or not, as the attitude chapter indicates, people make simultaneous associations between outgroup cues and positive or negative words (Fazio & Olson, 2003). As the Implicit Associations Test (IAT) shows ( Chapter 6 ), most people show automatic reactions on the basis of common social categories such as race, gender, and age (others include religion, ethnicity, and nationality; Rudman, Greenwald, Mellott, & Schwartz, 1999). After doing the IAT, students report that they can experience just how automatic subtle prejudice can be. If people have specifically positive associations to their own ingroup (and therefore less positive associations to outgroups), this makes a certain kind of psycho-logic: Balance principles (attitudes, Chapter 6 ) suggest that if people generally like themselves and identify with a particular group, they will also tend to favor that group (Greenwald, Banaji, Rudman, Farnham, Nosek, & Mellott, 2002). Even minimal, arbitrary ingroups, created within an experiment, can elicit automatically favorable ingroup biases and negative outgroup biases (Ashburn-Nardo et al., 2001).

The patterns of automatic associations include four types (Fiske, 1998b):

· Ingroup advantage, whereby the ingroup is more rapidly associated with positive attributes and feelings than the outgroup, and the ingroup is recognized and categorized faster than the outgroup.

· Stereotype-matching advantage, whereby outgroups are rapidly associated with stereotypic attributes, and those stereotypes rapidly prime other stereotypes for the same outgroup.

· Marked disadvantage, whereby groups that deviate from the cultural default (i.e., not white, male, 30–40 years of age, heterosexual, Protestant, able-bodied, etc.) acquire marked status; people are categorized most rapidly along whatever dimensions distinguish them from the default, so that black men are categorized as black faster than as male, and white women are categorized faster as women than as white.

· Categorization advantage, whereby people respond to other people more rapidly when they can categorize them according to race, gender, and age, than when they cannot. Rapid categorization and automatic associations are cognitively useful, saving mental resources in reactions to other people.

One might well question the meaning of all these automatic reactions and whether they reflect merely cultural knowledge or actual personal beliefs (Devine, 1989). That is, if all people, regardless of overt bias, show automatic stereotypic associations to outgroups, perhaps it is because all people know the cultural stereotypes, but that does not inevitably translate to agreeing with them. People do develop implicit associations from exposure to evaluative pairings in the environment, but they may not necessarily endorse them (Karpinski & Hilton, 2001). Children develop implicit evaluative associations to race from as early as age six (Dunham, Baron, & Banaji, 2008), in Anglo American (Baron & Banaji, 2006), Hispanic American (Dunham, Baron, & Banaji, 2007), and Japanese samples (Dunham, Baron, & Banaji, 2006). Children's implicit racial attitudes correlate with parental attitudes (Sinclair, Dunn, & Lowery, 2005). Even infants as young as three months show a visual preference for own-race faces, but newborns do not, and the preference results from exposure to predominantly own-race faces (Bar-Haim, Ziv, Lamy, & Hodes, 2006; Kelly et al., 2005). The familiar face overgeneralization hypothesis can account for ingroup favoritism in adults (Zebrowitz, Bronstad, & Lee, 2007). Hence, even if the culture does create these relatively automatic associations and preferences, it may do so indirectly (through de facto racial segregation of daily experience), quite early (infants!), and as a relatively rapid learning process (e.g., as short as three months).

As we have seen ( Chapter 6 ), people have some kinds of control over the effects of automatic processes. People's responses in fact go beyond the initial, automatic culturally based reactions. Nevertheless, these automatic or implicit responses form the starting point for our interactions with others, which at a minimum makes them into a millstone that handicaps spontaneous interaction. Even if people do not explicitly endorse ingroup preferences, they may still possess them as a foundation underlying more mature and explicit attitudes.

Ambiguous Biases: Excuses

Subtle bias is not only indirect, cool, and automatic but also ambiguous, that is, concealed behind responses that are open to various interpretations. Subtle bias often occurs in nonverbal behavior and in the speed of responses, both of which are ambiguous—hard for ordinary people to decipher. What the work on ambiguous bias pinpoints is that subtle prejudice occurs in settings where people have other excuses for their discrimination. “It's not that he's black; it's just that the other candidate was better on what we value,” where “what we value” turns out to be a moving target (Norton, Vandello, & Darley, 2003).

Consider some puzzling findings that demonstrate the ambiguity of whites' responses: When whites are not angry, they are actually more aggressive toward white than black targets, but when angry (providing an excuse), they are more aggressive toward blacks (Rogers & Prentice-Dunn, 1981). When whites witness an accident befalling a black victim, if they think they are the only bystander (so they have no excuse), they help the black person slightly more often than a white victim, but when other bystanders are present (providing an excuse for inaction), they help the black victim only half as often as the white one (Gaertner & Dovidio, 1977). When white college students are prejudiced, they are actually more likely to hire a well-qualified black than a well-qualified white, but if they have the excuse of poor qualifications, the prejudiced students are instead less likely to hire a black than a white applicant (McConahay, 1983).

One crucial feature in these ambiguity studies and dozens of others is that most white people find the thought of their own possible racism unpleasant, so they avoid behaving in overtly racist ways when the interpretation of their behavior would be obvious to them and to others. Aversive racism (Dovidio & Gaertner, 2004; Gaertner & Dovidio, 1986) refers to forms of racism that are unpleasant to the people who hold them. The research illustrates well-intentioned white people behaving in racist ways mainly when they have a nonracist excuse for their behavior, making it ambiguous. Indeed, when they cannot avoid the conflict between their stated attitudes and their own actual behavior, their prejudice reduces (Son Hing & Zanna, 2002). Implicit and explicit prejudice can operate independently, creating four combinations of low and high scores on each dimension (Son Hing, Chung-Yan, Hamilton, & Zanna, 2008). Aversive racists are high on implicit but low on explicit racism and express prejudice when the attribution for their negative evaluation is ambiguous.

A final, recently identified ambiguous bias against outgroups highlights the importance of emotions in prejudice. Infrahumanization views the outgroup as less human than the ingroup. Specifically, it views the outgroup as less capable of experiencing complex, uniquely human emotions that result from secondary appraisals (Demoulin et al., 2005; Leyens et al., 2003). In this view, outgroups and animals can show relatively simple and automatic fear, but only ingroups can show the more complex, thought-induced emotion of dread. People seem most motivated to view the ingroup as uniquely human in this way, independent of intergroup conflict (Demoulin et al., 2005). Infrahumanization operates automatically (Boccato, Cortes, Demoulin, & Leyens, 2007), and the link between uniquely human emotions and perceived humanity operates in both directions (Vaes, Paladino, & Leyens, 2006). Denying uniquely human emotions (e.g., anguish, mourning, remorse) to the outgroup makes them seem not to suffer as much when, for example, a natural disaster kills their family members; this undermines empathy and aid because although they suffer, they suffer like animals, not like humans (Cuddy, Rock, & Norton, 2007). Like other forms of ambiguous bias, infrahumanizing the outgroup by denying them uniquely human emotions seems harmless only on the surface.

Ambivalent Biases: Mixed Feelings

A final feature of subtle bias is ambivalence. For example, well-intentioned whites sometimes respond more extremely to blacks than they would to comparable whites. Response amplification cuts both ways: responding more positively to a black person in a positive situation but responding more negatively in a negative situation. Other studies have shown these kinds of amplification effects: A positive outgroup member is liked over a comparably positive ingroup member, but a negative outgroup member is disliked more than a negative one. This type of response fits not only aversive racism but also ambivalent racism (Katz & Hass, 1988; Katz, Wackenhut, & Hass, 1986).

Ambivalent racism captures the idea that most whites possess both positive and negative feelings toward blacks ( Table 11.3 ). Reminiscent of modern racism, their antiblack attitudes blame black Americans for their group's relative disadvantage because they supposedly violate the Protestant work ethic (alleged lack of family responsibility, community leadership, personal values, drive). Their problack attitudes (which are “pro” in the paternalistic sense of pity) stress the obstacles black Americans face, ongoing discrimination, and other situational barriers, which correlate with whites' own belief in humanitarianism and egalitarianism. Precisely because ambivalent racism comprises simultaneously positive and negative attitudes, it is unstable, tipping in one direction or the other, overreacting to small cues in the situation. Ambivalent racism predicts precisely this kind of response amplification.

TABLE 11.3 Racial Ambivalence Scale (Sample Items)

Pro-Black

Black people do not have the same employment opportunities that whites do. It's surprising that black people do as well as they do, considering all the obstacles they face. Too many blacks still lose out on jobs and promotions because of their skin color.

Anti-Black

Although there are exceptions, black urban neighborhoods don't seem to have strong community organization or leadership. On the whole, black people don't stress education and training. Many black teenagers don't respect themselves or anyone else.

Protestant Ethic

Our society would have fewer problems if people had less leisure time. Money acquired easily is usually spent unwisely. Most people who don't succeed in life are just plain lazy.

Humanitarianism-Egalitarianism

One should be kind to all people. One should find ways to help others less fortunate than oneself. There should be equality for everyone—because we are all human beings.

Other kinds of subtle prejudice also are ambivalent. Attitudes toward women reflect not only the hostile sexism normally considered but also benevolent sexism, a subjectively positive but paternalizing form of prejudice that belittles women (Glick & Fiske, 1996; 2001a). Hostile sexism is directed toward nontraditional women who violate narrow gender roles (female professionals, lesbians, feminists, and female athletes), and it tolerates sexual harassment (Begany & Milburn, 2002; Weiner, Hurt, Russell, Mannen, & Gasper, 1997) as well as spouse abuse (Glick, Sakalli-Ugurlu, Ferreira, & Aguiar de Souza, 2002). Benevolent sexism is directed toward traditional women who adhere to narrow gender roles: housewives, in particular (Glick, Diebold, Bailey-Warner, & Zhu, 1997). The Ambivalent Sexism Inventory (ASI; Table 11.4 ) reflects two dimensions that together support the status quo for women, by rewarding subordinate women with male protection and appreciation for their alleged purity but threatening uppity women with dislike and exclusion (Glick & Fiske, 2001b). This dynamic—in traditional gender relationships—does not result from political correctness or liberal good intentions. Instead, it reflects fairly universal gender dynamics across cultures (Glick et al., 2000; Glick & Fiske, 2001c). Societal power and intimate interdependence shape gender relations (Rudman & Glick, 2008).

TABLE 11.4 The Ambivalent Sexism Inventory (ASI)

Relationships between Men and Women

Below is a series of statements concerning men and women and their relationships in contemporary society. Please indicate the degree to which you agree or disagree with each statement, using the following scale: 0 = disagree strongly; 1 = disagree somewhat; 2 = disagree slightly; 3 = agree slightly; 4 = agree somewhat; 5 = agree strongly.

Benevolent Sexism (sample items)

No matter how accomplished he is, a man is not truly complete as a person unless he has the love of a woman. Men should be willing to sacrifice their own well-being in order to provide financially for the women in their lives. Many women have a quality of purity that few men possess. A good woman should be set on a pedestal by her man. Most women fail to appreciate fully all that men do for them.

Hostile Sexism (sample items)

Women exaggerate problems they have at work. Once a woman gets a man to commit to her, she usually tries to put him on a tight leash. When women lose to men in a fair competition, they typically complain about being discriminated against. Women seek to gain power by getting control over men. Most women fail to appreciate fully all that men do for them.

Overall, then, prejudice is not just pure hatred; its dimensions include disrespecting groups for their perceived incompetence and disliking groups for their perceived lack of warmth. Logically, then, a bigot could dislike but respect certain outgroups (Asians) and disrespect but like others (old people). The stereotype content model (Cuddy, Fiske, & Glick, 2008; Fiske et al., 2002) hypothesizes that people want to know two things about outgroups (and in parallel, about other individuals in general):

a. Are their intentions good or ill toward me and my group (i.e., are they warm or not)?

b. Can they enact their intentions (i.e., are they competent or not)?

These turn out to be (so far) universal dimensions of social perception, accounting for most of the variance in impressions (Fiske et al., 2007). Groups then may be threatening, for example, when respected as competent but disliked as not warm (e.g., rich people; Cikara & Fiske, 2012). Successful immigrants, such as Asian Americans in the United States, fit this category (Lin, Kwan, Cheung, & Fiske, 2005). Or groups may seem less threatening, for example, when they are liked as warm but disrespected as incompetent (e.g., older people, Cuddy, Norton, & Fiske, 2005; North & Fiske, 2012, 2013). Many groups fall into these mixed quadrants of envy or pity, reflecting a kind of ambivalence consistent with the earlier analyses of ambivalent racism and ambivalent sexism. Of course, some outgroups are flat-out disliked and disrespected: undocumented immigrants (Lee & Fiske, 2006), as well as drug addicts and homeless people (Harris & Fiske, 2006). The warmth and competence attributions predict specific intergroup emotions, such as pride for the ingroup, envy for the competent but cold outgroups, pity for the warm but incompetent outgroups, and contempt for the low-low outgroups; distinct forms of discrimination follow from these prejudiced emotions (Cuddy, Fiske, & Glick, 2007).

In research (Fiske, Cuddy, Glick, & Xu, 2002), the model reveals different kinds of outgroups, many of which reflect this kind of ambivalent prejudice. In the United States, outgroups are differentiated by perceivers on the basis of their perceived competence, the horizontal axis, and their perceived warmth, the vertical axis in Figure 11.3 . Take the figure counterclockwise from the top left-hand corner. Elderly people, people with mental disabilities, and people with physical disabilities all are seen as relatively warm and nice but not competent; they are disrespected but can be liked; in short, they are pitied. Then, in the bottom left-hand corner, poor people, welfare recipients, and homeless people are neither liked nor respected; they're seen as incompetent and not warm. That is what people consider traditional, unambivalent prejudice, but it applies only to those groups, who receive contempt from prejudiced people. Then, the cluster to the lower right consists of rich people, men, Jews, Asian Americans, professionals, and educated people; basically these are groups of people who are disliked but respected—in short, envied. They are seen as highly competent, having achieved a lot, but they're not nice—they allegedly gave up their humanity to get there. And consistent with ambivalent sexism, two subtypes of women, professional women and feminists, appear in that category, too.

Figure 11.3 Stereotype Content Model Showing Different Clusters of United States Outgroups

So who's allowed to be both warm and competent? “Us.” The cultural default. And in this particular sample, that included Christians, women in general, middle-class people, whites, and students. Societal reference groups are up in that top right quadrant. Finally, the groups who are in the middle can be in the middle for different reasons. They can be in the middle because people do not have clear stereotypes, as was probably the case for Native Americans in this sample from the northeastern United States. Groups can also be in the middle because of conflicting stereotypes about them. For example, American blacks end up in the middle, but as soon as they are subtyped into poor blacks and black professionals, they move to the two opposite corners of contempt and pride (Fiske, Bergsiecker, Russell, & Williams, 2009).

This model of stereotype content holds up in other countries, for groups that are important in their societies (in dozens of other nations, Cuddy et al., 2009; Durante et al., 2013). Figure 11.4 shows German views of the European Union (EU) community. Great Britain and Germany end up as envied, highly competent but supposedly not warm. Belgium and Austria are the poor cousins (perhaps due to economic and political scandals at the time). And then in the top left are the southern Europeans—the warm Mediterraneans—plus the Irish. These are places people want to go on vacation, in part because the people are stereotypically so nice. In the multination EU, no single nation ranks the most-admired EU ingroup or shared-reference group, because of a strong EU ideology of equal status, though Sweden, Denmark, and Finland come closest across countries. The ambivalent competence-by-warmth structure of stereotype content is not just something about politically correct Americans.

Figure 11.4 Stereotype Content Model, Showing Different Clusters of EU Nations

A final comment about the ambivalence of stereotype contents: The specific contents, though not the ambivalence, of stereotypes change over time and place. In the stereotype content model, for example, the content of group stereotypes depends on the group's perceived status (which predicts perceived competence) and perceived competitiveness (which predicts perceived lack of warmth), at a particular historical point (Bergsieker et al., 2012; Durante, Volpato, & Fiske, 2009). For example, the first Chinese immigrants in the United States arrived in large numbers to build the railroads, and they were not perceived to be technically competent professionals, as are the current generations of Asian immigrants. Nevertheless, every time and place has its pitied, envied, and contemptible outgroups.

Across ambivalent prejudices, the point holds: Many prejudices involve mixed feelings, and not just because of political correctness or antiprejudice norms. The middle 70% to 80% of the population simply just does not hate outgroups. Many outgroups are given a little something; “they're nice, but they're basically kind of stupid,” or “they're smart, but you know, they're not so nice.” These typical ambivalent reactions maintain stereotypes because people can always find something nice to say, conveniently omitting the negative (Bergsieker et al., 2012). Listeners infer the omitted negative dimension, when they hear only the positive dimension (Kervyn, Bergsieker, & Fiske, 2012).

All these cool and indirect aspects of subtle bias—being automatic, ambiguous, and ambivalent—reflect its origins in people's core social motive to understand their social worlds. Where does subtle prejudice come from? It comes from internal conflict in people's efforts to build an understanding that complies with contradictory information. People want to comply with their own antiprejudice ideals, their understanding that everyone is equal. But their understanding is also influenced at least unconsciously by information coming from the culture all around them. So a tension arises from wanting to comply with norms against bias, and even having personal values against bias, but not wanting to admit to oneself that one ever discriminates, prejudges, or stereotypes. Many people have automatic associations that come from the media, from other cultural influences, and from sheer inexperience with outgroups, just not having enough contact with people who are somehow different from themselves, and so they are left with associations that are relatively stereotypic. These automatic stereotypic thoughts conflict with antiprejudice norms and values, leading to an internal conflict that is resolved by the more subtle, ambiguous, ambivalent forms of bias.

Summary of Subtle Biases: Automatic, Ambiguous, Ambivalent

As overt expressions of bias improved over the course of the 20th century, the concept of subtle bias emerged. The Modern Racism Scale and subsequent scales showed that people might have cool and indirect forms of bias, blaming outgroups for their situations, viewing outgroup disadvantage as an intrinsic feature of their essential make-up, and withholding sympathy and respect. An important and surprising set of insights came from work demonstrating that people's biased associations can be automatic; priming and simultaneous associations both reveal how rapidly people link group cues to thoughts, feelings, and behavior. Many of the associations reflect ingroup favoritism at least as much as outgroup derogation. Subtle bias is also ambiguous, in the sense that people discriminate when they have nonprejudiced excuses for their behavior, which fits the idea of aversive bias that is unacceptable to the people who possess it. And subtle bias is also ambivalent, involving a mix of hostile and subjectively positive feelings that combine, both to create extreme responses and to mask bias in paternalistic or envious guises. Various outgroups are viewed with emotions more complex than simple antipathy. Subtle forms of bias reflect both a motive for coherent (though often misguided) understanding and self-enhancing (protecting one's image as unbiased).

SUBTLE BIAS IS SOCIALLY USEFUL: BELONGING AND CONTROLLING

Subtle forms of bias persist because they are socially pragmatic, whether or not morally acceptable. This section shows that subtle biases guide the perceiver's behavior, shape the reciprocal responses of outgroup members, guide ingroup communications about outgroup members, and create a circumscribed experience of accuracy. The social utility of bias serves motives of socially shared understanding, in part by smoothing interactions in which both people play according to stereotypic roles.

Subtle Bias Predicts Deniable Discrimination

Recent, provocative evidence indicates that stereotype-related behavior is spontaneously primed by outgroup categories (Dijksterhuis & Bargh, 2001; Wheeler & Petty, 2001). Recall from Chapter 2 the Trivial Pursuits study (Dijksterhuis & van Knippenberg, 1998) in which students primed with “professor” performed better on a factual knowledge test than students not so primed. Recall from Chapter 4 the automaticity study in which students primed with elderly stereotypes then walked to the elevator more slowly, and white people primed with black faces responded with more hostility when provoked (Bargh et al., 1996). Automatic evaluations result in predispositions to behave consistently with that evaluation (Chen & Bargh, 1999). Recall embodied cognition ( Chapter 6 ), wherein people respond faster to negative cues when simultaneously pushing a lever away from themselves (enacting avoidance) and to positive cues when pulling a lever toward themselves (enacting approach).

Automatic associations most clearly predict subtle forms of behavior outside people's ordinary conscious control (Dovidio, Kawakami, & Gaertner, 2002; Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997). For example, implicit prejudice predicts spontaneous nonverbal behavior. Subtle forms of nonverbal bias reveal discomfort, avoidance, and anxiety: sitting farther away, ending interviews sooner, making less eye contact, blinking nervously, or speaking awkwardly (Dovidio et al., 1997; Word et al., 1974). The dissociation between automatic processes and controlled processes is clear; whites may even bend over backward to be more pleasant to blacks than whites, but their subtle nonverbal cues indicate bias nonetheless (Vanman, Paul, Ito, & Miller, 1997). Relatively automatic forms of bias predict relatively spontaneous forms of discrimination.

In contrast, more explicit measures of prejudice predict more controlled, deliberate forms of behavior. As noted, scores on the Modern Racism Scale predict strategic but covert forms of interracial aggression (Beal et al., 2000). Perhaps because they feel uncomfortable together, people's subtle forms of bias also predict avoidance when people have discretion about spending time together, as at meals or parties.

Subtle forms of discrimination are socially useful precisely because they operate at the edge of awareness for both parties, so the target is unlikely to comprehend fully what is going on, and the perceiver can deny—to self and other alike—that the interaction was biased. Social utility is not moral justification, of course. Indeed, such hidden biases challenge the justice system's assumption that discrimination has to be consciously intended; legal scholarship is responding to the evidence of automatic, ambiguous, and ambivalent biases (Krieger & Fiske, 2006; Lane, Kang, & Banai, 2007).

Self-fulfilling Prophecies Create Confirming Behavior

Subtle forms of discrimination affect the target. Besides creating negative outcomes (avoidance, neglect) or a hostile atmosphere, an even more insidious effect is bringing about the very behavior that the biased perceiver expects. For example, automatic priming effects have direct impact on the behavior of people's interaction partners. In one study already mentioned (Chen & Bargh, 1997), white participants subliminally primed with black male faces acted in a stereotype-consistent fashion (hostile). But the point here is that they also elicited hostile behavior from their white interaction partner, thereby causing the original participants to see their interaction partner as a hostile person. This implies that whites interacting with a black male (and therefore primed) would act hostilely, evoking hostile behavior in return. The process by which the perceiver's biases affect target behavior, in turn supporting the perceiver's original biases, is termed a self-fulfilling prophecy (Merton, 1948), expectancy confirmation, or behavioral confirmation (e.g., Darley & Fazio, 1980; Klein & Snyder, 2003; Miller & Turnbull, 1986; Snyder & Stukas, 1998).

The process works by way of nonverbal communication and indirect verbal cues. That is, nonverbal behaviors such as smiling, nodding, eye contact, close distance, and greater length or frequency of interactions all communicate friendly attention versus hostile neglect (Harris & Rosenthal, 1985). In addition, verbal cues such as praise and criticism communicate expectancies (Harris & Rosenthal, 1985). Leading questions can confirm expectancies, and open-ended questions can allow targets to disconfirm expectancies (Leyens, 1989; Neuberg, 1989). People convey positive or negative climates, which lead targets to reciprocate, feeling more or less comfortable, resulting in the predicted behavior (Snyder et al., 1977; Word et al., 1974). In a study Chapter 7 noted (Snyder et al., 1977), men speaking with women over the phone saw a photograph of an attractive or unattractive woman as their partner. Although the women were unaware of the randomly assigned photograph, they behaved in a more friendly, sociable way when their partners believed them to be attractive. Similarly (Word et al., 1974), when white interviewers treated naïve white interviewees with nonverbal behaviors typical of white behavior toward blacks (e.g., greater seating distance, more speech errors, shorter interview), the naïve white interviewees saw the interviewer as unfriendly and inadequate, and themselves performed less well in the interview.

Ironically, when perceivers and targets cooperatively confirm stereotypic expectancies, both may feel the interaction went smoothly (Leyens, Dardenne, & Fiske, 1998), but this depends on both subscribing to the same biases. Expectancy-confirming perceivers may behave in a warm, relaxed manner, even as they confirm a negative expectancy (Judice & Neuberg, 1998). Targets may even self-stereotype in expectancy-confirming ways, especially when they believe that close others think the stereotype applies to them (Sinclair, Hardin, & Lowery, 2006) or when affiliative (belonging) motives are high (Sinclair, Huntsinger, Skorinko, & Hardin, 2005). This kind of social tuning can affect even implicit responses (Lun, Sinclair, Whitchurch, & Glenn, 2007; Sinclair, Lowery, Hardin, & Colangelo, 2005). Overall, behavioral confirmation is socially useful because it makes interaction superficially easier for everyone, even if the longer-term consequences are disastrous.

Socially Communicated Biases Build Ingroup Cohesion

Subtle bias also is socially useful because it fulfills the core social motive to belong: Shared stereotypes are easily communicated and enhance group cohesion (Ruscher, 2001). Prejudiced speech facilitates group interaction by being simple-minded and efficient, speaking in shorthand references to outgroup caricatures. Prejudiced speech facilitates group belonging motives, because it enhances ingroup superiority, dominance, and separation from the outgroup ( Table 11.5 ). For all these reasons, group interactions tend to focus on shared stereotypes. When people are motivated to form a consensus, their conversations focus on stereotypic information and encourage stereotypic impressions (Ruscher, Hammer, & Hammer, 1996). In general, discussion polarizes outgroup stereotypes, although one person with a strong supply of counterstereotypic information—an ally—can counteract the consensus (Brauer, Judd, & Jacquelin, 2001; Thompson, Judd, & Park, 2000).

TABLE 11.5 Functions of Prejudiced Speech

Function	Examples	Elaboration
Economy of expression	Group labels such as “white trash” or “Jap”	Shorthand that evokes stereotype is efficient
Group enhancement	Hostile humor	Points out superiority of ingroup
Social functions	Delegitimization, rationalization	Keeps outgroup separate for reduced contact
Ingroup dominance	Control of media, controlling nonverbals	Perpetuates stereotypes and social structure to retain ingroup status
Impression management	Bifurcation, illusion of universality	Presents nonprejudiced image to the world

Social sharing among ingroup members—gossip, rumor, opinions, stories, and media—contributes to consensus in stereotyped beliefs and related prejudices. The most easily communicated traits constitute the core of most stable stereotypes (Schaller & Conway, 2001). For example, people communicate some traits more frequently (intelligent, athletic, sensitive, quick-tempered) and others less frequently (slovenly, pugnacious, stolid, mercenary, grasping). The traits most often appearing in social discourse constitute the most frequent contents of stereotypes (Schaller, Conway, & Tanchuk, 2002). When people actually communicate stereotypes, they are more likely to endorse them than if they merely think and write about them (Schaller & Conway, 1999). This suggests the group-belonging aspect of stereotypic communication.

When people communicate a story containing both stereotypic and counterstereotypic information, the story becomes more and more stereotypic, after several people tell it (Kashima, 2000). Like other forms of rumor transmission (Allport & Postman, 1946), the story simplifies and sharpens the stereotypes as it passes from person to person. Moreover, as people infer other people's prejudices and stereotypes, that gives them social permission to respond in equally biased ways, reinforcing their own and other people's stereotypes and prejudice (Wittenbrink & Henly, 1996). Finding out about the stereotypes held by peers influences prejudiced people in particular, as a validation of their stereotypic beliefs.

Not all stereotype communication occurs by direct transmission of stereotype content. Indirect endorsement of stereotypes comes through the use of abstract terms—implying generality—instead of concrete terms—implying a single specific observation. For example, describing an interaction by labeling one person as “aggressive” differs from saying that person “pushed the other person away.” The linguistic intergroup bias reflects people's tendency to describe expectancy-consistent events (positive ingroup behavior and negative outgroup behavior) more abstractly, because it is presumably stable and typical; they describe expectancy-inconsistent events (negative ingroup behavior and positive outgroup behavior) more concretely, as if it is an isolated incident (Maass, Salvi, Arcuri, & Semin, 1989). This makes cognitive sense: Expectancy-consistent information is more readily attributed to abstract dispositions (Maass, Milesi, Zabbini, & Stahlberg, 1995; Wigboldus, Semin, & Spears, 2000), so it serves an understanding motive. Recall that the ultimate attribution error noted earlier shows the same pattern. More prejudiced people describe stereotypic behaviors in more abstract terms, regardless of valence (positivity or negativity), so this linguistic bias subtly reveals their prejudice and reinforces everyone's stereotypes (Schnake & Ruscher, 1998). People can conversely undermine stereotypes by communicating expectancy-inconsistent information in abstract terms (Ruscher & Duval, 1998). Finally, people use this bias more when the ingroup is threatened (Maass, Ceccarelli, & Rudin, 1996). Protection of the cohesive ingroup serves multiple motives (belonging, controlling, and self-enhancing).

Accuracy Would Make Stereotypes Useful

Most stereotyping research remains agnostic about the truth of stereotypes. Assessing how people perceive other people on the basis of perceived group membership remains an intrinsically important goal. Even if a stereotype were, in some sense, accurate, it would ignore the range of people within the group, including those for whom the stereotype is not at all true. More important, assessing stereotype accuracy poses far more considerable challenges than the casual observer imagines (e.g., Judd & Park, 1993).

First, most stereotypes consist of traits: hostile, dishonest, lazy, harmless, stupid, and so on. If stereotypes consisted solely of easily measured attributes, such as height and weight, the assessment issue would be simpler. However, even with objective physical measurements, stereotypes can distort laypeople's estimates (Biernat, Manis, & Nelson, 1991), and scientists' measurements are not immune (Gould, 1981). According to the shifting standards model, what laypeople mean by even physical descriptions is conditioned on group membership (tall for a woman is not tall for a man; Biernat & Manis, 1994; Biernat et al., 1991). When trait evaluations enter in, the situation becomes even more biased because the same trait carries different meanings for different groups: Different behavior is expected from a woman who is a competent parent than from a man who is; different performance is expected of a black person who is competent at math than of a white person who is (Kobrynowicz & Biernat, 1997). Lower performance is evaluated as competent for the group not typically expected to do well on the task (“competent parent, for a man”; “competent at math, for a black person”). This has implications for people's outcomes in situations where judges must choose between two candidates. A person from the group devalued on that dimension (e.g., women on athletics) may be told consistently that his or her performance is great (for a member of that outgroup, the message implies, though never says), but when forced to choose, judges will instead choose the member of the group stereotyped to be better (Biernat & Vescio, 2002).

Second, even if one can surmount the shifting standards issue, the critical question is, whose point of view constitutes the criterion for accuracy? Do we measure perceptions held by the stereotyped group? They are likely to be motivated to maintain a positive group identity, although they may sometimes adopt the stigmatized identity offered by society as a way to differentiate themselves from others (Ellemers, Spears, & Doosje, 2002). Do we measure the consensus of other groups about the targeted group? If they are the ones holding the stereotype, then we have hardly distinguished between the stereotype and the reality. Indeed, for people who believe that a stereotype validly applies to an individual, increased accuracy motivation and added attention merely strengthen their stereotyping (Madon, Guyll, Hilbert, Kyriakatos, & Vogel, 2006). Do we measure the consensus of experts? Even supposedly neutral parties are regarded with suspicion by groups on conflicting sides of the same question (Vallone, Ross, & Lepper, 1985), so who is truly qualified to judge?

What about objective measures? What one side views as objective, the other side may not; people's judgments of the quality of evidence are affected by their attitudes toward the relevant issue. People on one side of a social dispute tend to discount evidence for the other side of the dispute (Lord, Ross, & Lepper, 1979). And people tend to see their own perspective as less biased than other people's (Pronin et al., 2002). How can we all be less biased than the average person?

Nevertheless, suppose that we could find a criterion (test or measurement) that both ingroup and outgroup could accept as fair. Suppose one group scores noticeably above zero on this criterion measure. What degree of the trait would mean that one group really does possess the trait? Would they have to score higher than the midpoint? Most tests do not have an absolute score that labels a person categorically as possessing the trait or not. All scores are relative to other people's scores.

Absolute scores aside, how much higher does the group have to be than a comparison group (and which comparison group)? How much of a difference matters? The standard statistical answer is to compare the average between-group difference, relative to the variability within each group. The distributions of group scores will matter here. In many instances of group differences, the actual differences are small, and the distributions overlap so substantially that drawing inferences about individuals is meaningless; differences are small relative to individual variation (e.g., Costa, Terracciano, & McCrae, 2001). As Figure 11.5 (A) indicates, overlapping normal distributions acknowledge the possibility of average group differences but even greater variability within the group. In this case, would one really want to judge the individual by the group mean?

Figure 11.5 Hypothetical Distributions of Actual Group Differences

Other issues of distribution also matter. Supposing one set a cut-point for possessing the trait or not, how many of the group would have to have a particular attribute? Some characteristics or behaviors are prescribed as vital for virtually all members of the group (college professors having a Ph.D.). Allport (1954b) described these as the J-curve of conformity behavior (see Figure 11.5[B] ); he uses workers' arrival times at a factory, but one could easily use students' arrival at class, Americans' ability to speak English, or little girls who play with dolls and little boys who play with trucks. The height of the peak (how many people conform exactly) and the length of the tail determine how characteristic the trait really is. In this kind of distribution, most people in the group behave as prescribed, but the pattern still ignores how much this conformity differentiates the group from others. (All people in one group might speak English, but so might all people in the other group.)

But sometimes the exceptional behavior of unusual group members can define perceptions of the group. If a few people's standings on the criterion tend to be extreme, but most group members do not possess the trait at all, and absolutely no one outside the group possesses the trait, Allport (1954b) calls this a rare-zero differential (see Figure 11.5[C] ). For example, some Asian Indian women wear saris in public, and essentially no non-Indian women do so.

The point of examining criteria and distributions is that no single characteristic identifies every last member of a particular group. But even suppose one group is overrepresented as geniuses or murderers; few members of the group are geniuses or murderers, but their group has more than other groups do, relative to their numbers. Does that make the group as a whole stereotypically geniuses or murderers?

Although some have argued that stereotypes generally are accurate (Jussim, McCauley, & Lee, 1995; Ottati & Lee, 1995), or at least that some are accurate (Hall & Carter, 1999), others have argued that this viewpoint is premature, pointless, and damaging (Allport, 1954b; Fiske, 1998b; Stangor, 1995).

A compromise position holds that stereotypes result from the uneven distribution of group members into particular social roles, which leads people to infer that the behavior—which is role-based—actually characterizes the group's inherent dispositions. For example, stereotypes of men and women depend on their current social context. According to social role theory, the group's behavior in its most common social roles causes inferences about the group's enduring dispositions. Applied to men and women, this suggests that men will be stereotyped as agentic, because they are overrepresented in roles that require instrumental activity outside the home; women will be stereotyped as communal because they are overrepresented in roles that require nurturing behavior inside the home (Eagly, 1987; Eagly, Wood, & Diekman, 2000). This social role account suggests that as gender roles change, stereotypes will change. People do think sex differences are diminishing as social roles change, and this particularly holds true for women's roles, perceived to be changing more than men's are (Diekman & Eagly, 2000).

A parallel theory describes the evolution of enemy images in international relations (Alexander, Brewer, & Herrmann, 1999; Alexander, Brewer, & Livingston, 2005). Other countries are assessed on three dimensions: relative power, relative status, and degree of cooperation or competition. Varying combinations of these dimensions result in common stereotypes of other nations as allies, enemies, dependents, and barbarians. These images presumably function to justify one's own actions (as a country or as an individual). The content of the social stereotypes derives from the intergroup context in which they form. In that sense, they could be viewed as apparently accurate from the perspective of a particular side of the situation.

Summary and Conclusion Regarding Functions of Subtle Bias

Subtle forms of bias not only serve cognitive functions of providing a biased understanding while protecting the self from knowledge of that bias but also serve social functions. They allow prejudiced people (a) to discriminate, when they have nonprejudiced excuses for doing so; (b) to shape the behavior of others to confirm their own biases; and (c) to communicate with each other, reinforcing group cohesion. Also, (d) they allow certain kinds of constrained accuracy, in the context of perceiving social role behavior to reflect intrinsic properties of the roles' occupants.

BLATANT BIAS: BELONGING WITH THE INGROUP, CONTROLLING OUTGROUP THREATS, AND ENHANCING THE SELF

If subtle forms of bias result primarily from understanding motives, with a little self-enhancement thrown in, the more overt, explicit forms of bias operate somewhat differently: Blatant bias emphasizes belonging and controlling more than the other motives. Belonging surfaces via ingroup identity and status concerns, whereas controlling surfaces via concerns about resources held by the ingroup. In all of this, people strive to feel good about themselves as group members.

Realistic Group Conflict Theory: Threat to Resources

The most obvious explanation for group prejudices, stereotypes, and discrimination is that group interests conflict, so of course groups harbor biases against each other. Realistic group conflict theory holds that threats to ingroup advantage result in negative intergroup reactions. Threats may be economic, political, military, or prestige-related, but the key idea is that they are, in some sense, real (Campbell, 1965). This obvious explanation is dramatically incomplete, however.

Early research on intergroup relations seemed to support the idea. The classic Sherif and Sherif (1953) Robbers Cave study took place in a boys' summer camp ( Chapter 1 cited it as evidence for belonging). When researchers arranged for the boys to compete as two separate teams (Eagles versus Rattlers), they each formed ingroup friendships, favored their own group, and were hostile to the other group. When the researchers later forced the two teams to cooperate in their shared common interest, they became more amicable.

Some other research also seemed to fit (for reviews, see Brewer & Brown, 1998; Jackson, 1993). Intergroup competition increases hostility, and intergroup cooperation decreases hostility, not only for children, but also for adolescents (Rabbie & Horowitz, 1969) and adults (Blake & Mouton, 1961). White voters living in areas experiencing or about to experience busing (which could be construed as realistic group conflict) show less racial tolerance (Bobo, 1983). Corroborating the field research, some laboratory studies have found that manipulated conflict increases hostility (Worchel, Axsom, Ferris, Samaha, & Schweitzer, 1978). Some of the work testing the frustration-aggression hypothesis ( Chapter 10 ) at a societal level also seemed to fit. When people's own economic outcomes suffer, they turn hostile toward competing groups (Hepworth & West, 1988; Hovland & Sears, 1940).

Despite deserving to be true, realistic group conflict theory receives only limited and uneven support. Many other studies fail to support the core hypothesis and instead find that other variables matter more than sheer group self-interest. For example, links between macroeconomic fate and intergroup hostility prove elusive (Green, Glaser, & Rich, 1998; Green, Strolovitch, & Wong, 1998). Consistent with modern racism, the perceived symbolic threat to one's group matters more than the personal, tangible consequences (Kinder & Sears, 1981; Sears & Kinder, 1985).

What apparently matters most are people's perceptions of group relations: group threat, group identity, and symbolic threats, as we will see. Because subjective perceptions of group threat do fit some version of realistic group conflict theory (e.g., Bobo, 1983), one might as well drop “realistic” from the theory's name and shift toward “perceived group conflict theory.” Perceived conflict does predict negative attitudes toward the outgroup (Brown, Maras, Masser, Vivian, & Hewstone, 2001; Hennessy & West, 1999). For black South Africans, under competitive and threatening intergroup conditions, degree of ethnic group identification was associated with negative attitudes about nonblacks (Duckitt & Mphuthing, 1998). For white Anglos, perceived conflict with Latinos predicted opposition to bilingual education programs (Huddy & Sears, 1995). Under realistic group threat, European Americans especially resent Asian Americans (Maddux, Galinsky, Cuddy, & Polifroni, 2008).

Competition matters mainly when people identify with their ingroups. More important, ingroup identification by itself can account for intergroup hostility, even without competition (Brewer & Brown, 1998). Not only perceived threats but also another subjective perception, group identity, predicts intergroup responses. For African Americans, the part of racial identity that includes perceived oppression does predict positive attitudes toward affirmative action, over and above personal benefit (Schmerund, Sellers, & Mueller, 2001). In short, social identity and perceived threat matter more than realistic conflict of interest. Social psychologists therefore have energetically pursued these variables.

Social Identity, Self-Categorization, and Related Theories: Threat to Group Identity

As we have just seen, groups fight over more than mere material outcomes; they fight over symbolic and identity issues. Intangible outcomes—such as group recognition, status, and prestige—create conflict far more often than do tangible resources. Even when the conflicts appear to concern resources, frequently the real payoffs are pride in one's identity in a valued group able to win such resources. Group conflict is an inherently social competition that goes beyond concrete self-interest. One result of the struggle for positive identity is bias against the outgroup.

This type of explanation contrasts with the flavor of the more purely cognitive explanations launched in the first several sections of this chapter. In this identity-oriented view, bias results from more social processes than just the inevitable cognitive processes, which the first part of this chapter documents. Conflict also emerges as a normal part of intergroup interaction (e.g., Oakes, 2001). People align themselves with a particular social perspective, their self-definition as a group member. According to social identity and self-categorization theories, as we will see, self-definition in a social context provides ingroup perspectives that naturally contrast with the perspectives of outgroup members.

BASIC THEORIES

Social identity theory (SIT) locates interactions along a continuum from interpersonal (based on individual characteristics) to intergroup (based on salient group memberships) (Tajfel, 1981; Tajfel & Turner, 1979). Social identity is the part of one's self-concept that derives from group membership; it includes evaluations of group attributes, as well as prescriptions about ideal group attributes. For example, people from California may view themselves as more informal and fit, whereas people from New York may view themselves as more driven and stylish. Social identity aims to base self-esteem on a positive evaluation of one's group in comparison with other groups. Thus, social identity requires at least one other group for contrast.

The relevant core social motives are belonging and self-enhancing. Social identity theorists argue that categorizing people into “us” and “them”—creating ingroups and outgroups—differentiates social categorization from other types of categorization. Because the self is directly implicated, self-enhancing motives are inherent to social identity theory. In this view, people need a positive social identity. Belonging motives determine the effort to find a positive and distinct group identity.

Categorization of any kind exaggerates differences between groups and minimizes differences within groups. This makes group members seem less variable than they would be as individuals. And, theorists argue, they may actually behave in less variable ways as self-identified group members because of conformity pressures ( Chapters 12 – 13 ).

Self-categorization theory (SCT) builds on social identity theory, abandoning the self-esteem hypothesis and focusing on what categories people actually use in a given intergroup setting (Turner, 1985). The self is not fixed but depends on fit to categories salient in a given social context (e.g., police and protesters, not men and women). Self-categorization depends on two kinds of fit. Comparative fit specifies that perceivers will compare between-group differences with within-group differences and use those distinctions that best differentiate the groups (who is shouting, who is facing which way, who is wearing what). This so-called meta-contrast determines the relevant social categories. In this view, self-categorization rests on actual group differences.

Normative fit specifies that the direction of the group differences must match the socially shared meaning of the two social categories (who believes what, who protests and who supports the state). The emphasis on recognizing socially defined group differences as important aspects of people's self-concept emerges from the European context of social psychology, some of which has a more sociological and political orientation. Also Europeans, more than many North Americans, have contact with people of other national identities, many of whom value their own group membership. Thus, European group identities are more clearly acknowledged as socially accepted group differences. This contrasts with the American melting-pot ideology that traditionally seeks to assimilate all group differences into a single color-blind identity.

Self-categorization theory is less a theory of ethnocentrism and bias than a theory of psychological group membership (Turner & Reynolds, 2001). As such, we see in the next chapter that it illuminates various processes of psychological groups and social influence.

Theories related to SIT and SCT address specific aspects of social identity. For example, the balance between a unique identity and a group identity appears in optimal distinctiveness theory (Brewer, 1991). In this view, bias affirms the satisfaction of belonging to the right groups and differentiates the ingroup from outgroups at the optimal level of difference. People balance individual autonomy against group identity, and versions of this occur even in non-Western cultures (Vignoles, Chryssochoou, & Breakwell, 2000). Belonging motives underlie ingroup favoritism, but a need for distinctiveness (self-enhancement) predicts outgroup derogation (Vignoles & Moncaster, 2007).

As another example, subjective uncertainty reduction theory (Hogg & Abrams, 1993) proposes that people identify with groups in order to find group norms that reduce uncertainty. Ingroup favoritism results from attraction to a group that allays the tension of uncertainty (see the next chapter).

The core theory, SIT, sheds light on some critical processes related to intergroup bias: perceived homogeneity, ingroup favoritism, and group-based self-esteem. As we will see, all reinforce the positive distinctiveness of people's social identity with their ingroup (Brewer, 2007a).

GROUP HOMOGENEITY

Categorizing people into a group implies seeing them as similar to each other. SIT suggests that people will accentuate the homogeneity within groups and the differences between groups. People often see the outgroup as more homogeneous than the ingroup (“they are all alike; we are more varied”). Although this perception sometimes occurs (Linville, Fischer, & Salovey, 1989; Messick & Mackie, 1989; Mullen & Hu, 1989), the picture is more complicated than intuition suggests. First, group variability might appear in three different components (Ostrom & Sedikides, 1992; Park & Judd, 1990):

· Perceived stereotypicality of each group (the degree to which they receive high ratings on stereotypic attributes and low ratings on counterstereotypic attributes)

· Perceived dispersion of each group (how much or how little group members seem to vary from others in the same group)

· Perceived similarity within each group (whether group members seem to resemble each other)

All three—stereotypicality, dispersion, and similarity—have appeared as higher for outgroup than ingroup, especially in natural settings.

The perceived homogeneity effect is common, although far from universal (Brewer & Brown, 1998; Brown, 2000; Devos, Comby, & Deschamps, 1996). Outgroup homogeneity is stronger for (a) groups that are unfamiliar (Linville, Fischer, & Salovey, 1989), (b) real, as opposed to laboratory groups (Ostrom & Sedikides, 1992), and (c) organized in people's minds at more abstract, superordinate levels, often true for outgroups (Park, Ryan, & Judd, 1992). However, the effect weakens or reverses for (d) numerical minority perceivers (Mullen & Hu, 1989; Simon, 1992) and (e) dimensions important to ingroup identity (Kelly, 1989; Simon, 1992); under those conditions, an ingroup homogeneity bias occurs. Overall, when the intergroup context is salient—under group threat or competition—people highly identified with their group see both ingroups and outgroups as more homogeneous (Ellemers et al., 1997).

Perceived group homogeneity serves to justify and reify the designation of a group. Sometimes, the context emphasizes the social reality of the outgroup, as when a dominant group stereotypes a subordinate group. Sometimes, the context emphasizes the social reality of the ingroup, as when a minority pulls together for mutual support along common dimensions of identity. Sometimes, the context emphasizes the social reality of both groups, as when they compete. Social identity theory describes how the intergroup context emphasizes interpersonal (more variable) or intergroup (more homogeneous) perceptions.

INGROUP FAVORITISM

Prejudice consists largely in liking “us,” more than disliking “them.” People tend to think of prejudice against outgroups as the core phenomenon, but SIT suggests—and the data show—instead that prejudice favoring the ingroup is the foundation (Brewer, 1999, 2007b; Hewstone, Rubin, & Willis, 2002; Mullen, Brown & Smith, 1992). A positive social identity suggests benefiting the ingroup because they are part of the self and better than the outgroups (Otten & Wentura, 2001). Of course, ingroup favoritism in a zero-sum competition means depriving the outgroup. Therefore, much intergroup discrimination occurs by neglect rather than deliberate derogation.

Evidence of ingroup favoritism over outgroup derogation takes several forms (Hewstone et al., 2002). We saw earlier that ingroup favoritism occurs automatically, prior to awareness, and that the effect selectively associates positive attributes with the ingroup, rather than negative attributes with the outgroup (Dovidio et al., 1986; Gaertner & McLaughlin, 1983; Perdue et al., 1990). We also saw earlier that aversive and subtle forms of racism consist in the absence of positive reactions to racial outgroups, more than the presence of virulent negative reactions (Gaertner & Dovidio, 1986; Pettigrew & Meertens, 1995). People benefit the ingroup by giving rewards rather than withholding punishments (Mummendey & Otten, 1998). As a parallel, patriotism (ingroup pride) does not mean nationalism (belligerence toward other nations) (Feshbach, 1994).

Ingroup favoritism is encouraged by several moderators (Hewstone et al., 2002) that all protect positive social identity: Strong ingroup identification (Branscombe & Wann, 1994; Perreault & Bourhis, 1999) provides the most obvious case for ingroup bias protecting social identity. But minority status (Mullen et al., 1992; Otten, Mummendey, & Blanz, 1996) and moderate distinctiveness (Jetten, Spears, & Manstead, 1998) also suggest cases of vulnerable social identity in need of protection. Here, too, ingroup favoritism bolsters a positive identity. In the case of high power or status, which also enhances ingroup favoritism (Brown, 2000; Mullen et al., 1992), people still discriminate when they feel vulnerable (e.g., because their high status is perceived to be unstable, illegitimate, and permeable; Bettencourt, Charlton, Dorr, & Hume, 2001). Perhaps high-status people also discriminate sometimes because they feel entitled (Fiske, 1993, 2010; Richeson & Ambady, 2003; Sachdev & Bourhis, 1987, 1991).

Ingroup favoritism does not promote personal advantage, nor does it even promote absolute group advantage. That is, people will favor the ingroup over the outgroup, maximizing the relative difference, even at the expense of the ingroup's absolute outcome (Rabbie & Horowitz, 1969; Tajfel, Flament, Billing, & Bundy, 1971). The minimal group paradigm especially shows this by placing people into arbitrary, temporary groups in the laboratory and giving them points to allocate to ingroup and outgroup members (excluding themselves). SIT fits findings that intergroup differentiation (a) occurs only on dimensions that favor the ingroup and (b) increases when people need to differentiate the groups (given conflict, boundary breakdown, ingroup importance) (Hogg, 1995b). When group distinctiveness is under threat, high identifiers in particular will differentiate their group by ingroup favoritism, according to meta-analysis (Jetten, Spears, & Postmes, 2004). Ingroup favoritism can occur even on negative dimensions that help define the group and differentiate it from other groups; according to SCT, fit should matter most (Reynolds, Turner, & Haslam, 2000). Overall, ingroup favoritism is a key result of SIT and SCT.

SELF-ESTEEM

SIT makes two predictions about self-esteem and discrimination. As part of the search for a distinct, positive social identity, (a) intergroup differentiation should elevate self-esteem and, conversely, (b) low self-esteem should motivate people to discriminate, in order to raise their self-esteem (Hogg & Abrams, 1990). Regarding the first prediction: When people do show ingroup favoritism, they feel better about themselves afterward, in direct support of SIT (Lemyre & Smith, 1985; Oakes & Turner, 1980; for a review, see Rubin & Hewstone, 1998). Consistent with SIT, the kind of self-esteem that improves is state self-esteem, which responds to changes in the person's circumstances. Because SIT posits that identity will change, depending on context, so would self-esteem. Trait self-esteem, which is more stable, does not change as a function of discrimination. Higher state self-esteem thus is an outcome of discrimination.

The second prediction, that low self-esteem should motivate discrimination, has not received clear support. Some studies have shown support, but an equal number have failed to find support (Aberson, Healy, & Romero, 2000; Brewer, 2007a; Hewstone et al., 2002; Rubin & Hewstone, 1998). Instead, research needs to focus on temporary reductions in state self-esteem among people highly identified with their group and whether low state self-esteem can cause discrimination in the service of social change that aids group identity (Turner & Reynolds, 2001). What's more, people with high self-esteem actually may be those most likely to discriminate, at least in the most direct and self-serving way of discriminating in favor of a group in which they are active participants (Aberson et al., 2000). Although everyone tends to discriminate in favor of the ingroup, people with high self-esteem may feel less constrained about displaying their self-interested behavior.

SUMMARY OF IDENTITY THEORIES

Partly in reaction to the emphasis of realistic group conflict theory on material, tangible resources as the basis for intergroup conflict, social identity theory and related approaches focus on people's need for a distinct, positive self-definition as a group member and the ensuing impact on intergroup reactions. Perceived group homogeneity follows from categorizing people into groups, and because the self is a member of one of those groups, ingroup favoritism follows. In the short run, ingroup favoritism increases state self-esteem.

Authoritarianism: Threat to Values

Yes or no?

· Obedience and respect for authority are the most important virtues children should learn.

· A person who has bad manners, habits, and breeding can hardly expect to get along with decent people.

· People can be divided into two distinct classes: the weak and the strong.

· Human nature being what it is, there will always be war and conflict.

· Homosexuals are hardly better than criminals and ought to be severely punished.

A whole series of “yes” responses to these and similar items would indicate conventionality, submission to authority, toughness, cynicism, and aggression on behalf of those values, according to the theory of authoritarian personality (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950). A slightly more bizarre set of items was also thought to correlate with this personality syndrome:

· The wild sex life of the Greeks and Romans was tame compared to some of the “goings-on” in this country, even in places where people might least expect it.

· Some day it will probably be shown that astrology can explain a lot of things.

· Most people don't realize how much our lives are controlled by plots hatched in secret places.

· Nowadays more and more people are prying into matters that should remain personal and private.

The authoritarian personality theory proposed a constellation of personality traits, brought about by strict and punitive child-rearing practices. This socialization theoretically resulted in idealizing one's parents and displacing aggressive impulses onto outgroups. The hypothesized result was widespread ethnocentrism.

The ethnocentrism part of the empirical foundation was solid: People prejudiced against one group tend to be prejudiced against other groups, a reliable observation. The authors of the original theory were grounded in the Holocaust, trying to explain Nazi genocide and to understand anti-Semitism elsewhere as well. Yet much of their theoretical background relied on Freudian interpretations of child-rearing and personality, assumptions about socialization and personality structure that have not been well supported (e.g., Brown, 1965; Duckitt, 1992). Plus, some well-known methodological problems included unrepresentative samples, interviewer bias, failure to control for education or social class, and focusing on authoritarianism of the political right. Moreover, the scale failed to word some items in reverse, so that an acquiescence response bias (agreeability; see Chapter 2 ) would tend to be confounded with agreeing to items indicating authoritarianism (Hyman & Sheatsley, 1954). These critiques of both method and theory undermined the credibility of the scale.

However, some relationship between authoritarianism and prejudice remains. A modern version of this insight, the Right-Wing Authoritarianism (RWA) Scale (Altemeyer, 1981, 1988) proves to be more viable. This version of the scale focuses on three themes: conventionality (conformity to traditional values), authoritarian submission (obeying powerful leaders), and authoritarian aggression (inflicting harm on those who deviate from conventional beliefs, especially when harming is sanctioned by one's leaders). RWA consistently correlates with prejudice against a variety of outgroups (e.g., Duckitt, 1993; Meloen, Van der Linden, & De Witte, 1996). But why do these responses tend to correlate?

Intense but insecure identification with the ingroup may underlie all four aspects of RWA (conventionality, submission, aggression, prejudice) (Duckitt, 1992). Valuing the ingroup and perceiving it as threatened combine to demand cohesion for the ingroup and therefore conventional conformity to norms, obedient submission to authorities, and aggressive intolerance of deviance. RWA's underlying psychology rests on perceiving the world as a dangerous place, which arouses fear and hostility, as well as moral superiority, based on authority sanctions, which justifies aggression against perceived threats, namely outgroups. People high on RWA endorse harsh, punitive policies on a variety of social problems: AIDS, abortion, child abuse, drugs, homelessness, trade deficits, as well as higher education (Peterson, Doty, & Winter, 1993), although they lack knowledge about these issues (Peterson, Duncan, & Pang, 2002). They also deny the reality of the Holocaust (Yelland & Stone, 1996). And they dislike feminists (Haddock & Zanna, 1994) and homosexuals (Haddock, Zanna, & Esses, 1993). All these RWA responses indicate a refusal to negotiate constructively with perceived threat to cherished conventional values, which is then one important predictor of prejudice.

A related perspective relies not on individual differences in personality but on situational factors. Terror management theory focuses on perceived threat not to the group but to the continuity of the self, as when people confront their own mortality (see Chapter 9 with regard to mortality salience and prosocial behavior, as well as Chapter 10 with regard to mortality salience and aggression). However, in contrast to RWA, mortality salience depends on context, not personality (Pyszczynski, Greenberg, & Solomon, 1997, 1999; Solomon, Greenberg, & Pyszczynski, 2000). The main idea is that people instinctively desire continued life and that death-related thoughts are threatening. People avoid death-related thoughts by trying to suppress them when they are conscious. When death-related thoughts are unconscious, people try to maintain self-esteem (see Chapter 5 ) and to defend their cultural worldview, including cherished values that will live beyond their own lifetimes.

Most relevant here are findings that when mortality is salient, people respond especially harshly to those who violate or criticize accepted values and those who are from outgroups (Greenberg et al., 1990; McGregor, Zanna, Holmes, & Spencer, 2001; Rosenblatt, Greenberg, Solomon, Pyszczynski, & Lyon, 1989; Schimel et al., 1999). Some of these findings hold especially for people high on RWA, when it is measured. The link between reactions to mortality salience and authoritarianism makes sense: Authoritarianism, like mortality salience, sees the world as threatening and dangerous and sees conformity to cherished values as the first line of defense. Context determines how values are enacted; for example, liberals, when tolerance is salient, become more tolerant under mortality salience (Greenberg, Simon, Pyszczynski, Solomon, & Chatel, 1992). Similarly, authoritarians express prejudice as a function of their currently salient self-categorized social identity (Reynolds, Turner, Haslam, & Ryan, 2001). But all these variations on authoritarian responses depend on perceived threat to cherished values.

More generally, meta-analysis across cultures finds relationships between political conservatism and acknowledging threat: concern about threat, awareness of death, concern about system instability, avoidance of ambiguity, and motivation for order. Liberalism correlates with seeking experience, tolerating uncertainty, and integrating complexity (Jost, Glaser, Kruglanski, & Sulloway, 2003). Consistent with liberals' responsiveness to novelty, ambiguity, and complexity, they show greater activity in the brain's conflict-related anterior cingulate cortex; conservatives' more structured and persistent style makes them less responsive in this discrepancy-monitoring area (Amodio, Jost, Master, & Yee, 2007).

Social Dominance Orientation: Threat to Group Status

Another personality variable related to prejudice is social dominance orientation (SDO) (Pratto, Sidanius, Stallworth, & Malle, 1994; Sidanius & Pratto, 1999), the degree to which one endorses a hierarchy in which some groups dominate other groups. SDO differs from authoritarianism and interpersonal dominance, but like them it does predict prejudice. People high on SDO support policies that maintain inequality (for example, opposing affirmative action; Sidanius, Pratto, & Bobo, 1996), and they choose careers that enhance existing hierarchies. For example, they are more likely to go into business, police work, or public prosecution (enhancing current hierarchies), rather than teaching, social work, or public defense (attenuating current hierarchies). High SDO people do well in hierarchy-enhancing roles: for example, students majoring in business and economics. Low SDO people excel in hierarchy-attenuating roles: for example, students majoring in anthropology, ethnic studies, sociology, social work, or nursing (Pratto, Stallworth, Sidanius, & Siers, 1997; van Laar, Sidanius, Rabinowitz, & Sinclair, 1999).

Men are more likely than women to be high on SDO, and it correlates with being tough-minded and viewing the world as a competitive place (Sidanius, Pratto, & Bobo, 1994). High SDO favors the powerful. For high-status groups, high SDO exaggerates ingroup favoritism. Even for low-status groups, if they view the hierarchy as legitimate, high SDO also predicts favoring the high-status groups (Levin, Federico, Sidanius, & Rabinowitz, 2002). According to both lab and field studies, having a high-status position leads to SDO, which in turn causes prejudice (Guimond, Dambrun, Michinov, & Duarte, 2003).

How does SDO relate to RWA? Although both relate to an inflexible defense of some group interests over others, their psychology differs. One way to understand the relationship between various individual differences and prejudice is according to the kind of threat posed to the ingroup. Authoritarianism focuses on perceived threat to ingroup values in a dangerous world, whereas social dominance orientation focuses on perceived threat to ingroup status in a competitive world (Duckitt, 2001; see Figure 11.6 ). In this dual-process theory of ideology and prejudice, one path goes from punitive socialization, to social conformity, to viewing the world as dangerous, to RWA, and thence to ingroup favoritism and outgroup prejudice. The other path goes from unaffectionate socialization, to tough-mindedness, to viewing the world as competitive, to SDO, and thence to ingroup favoritism and outgroup prejudice. The links among socialization, personality, worldview, and ideology receive strong research support in the United States, Europe, South Africa, and New Zealand (Duckitt, 2001). Different kinds of blatant prejudice result from different kinds of motivational dynamics.

Figure 11.6 Dual-Process Model of Right-Wing Authoritarianism and Social Dominance Orientation

System Justification: Threats to the Status Quo

Blatant forms of bias operate to justify the current system of inequality. Several theories make related arguments. Among the first to make this argument in social psychology, Jost and Banaji (1994; Jost, Banaji, & Nosek, 2004; Jost, Burgess, & Mosso, 2001) proposed that stereotypes operate in the service of system justification, the psychology of legitimizing current social arrangements, even at personal and group cost. For high-status groups, supporting the self (ego justification), the group (group justification), and the system all are compatible processes. But for low-status groups, system justification operates at the expense of the other two. So, for example, women may buy into benevolent sexism as an ideology because it legitimates the stability of current arrangements between men and women (Glick & Fiske, 2001b); sexist women receive some benefits from believing that women are pure and need to be protected and cherished by men, in return for women being submissive. Women are prized for being nice, while men are prized for being competent, in this view. More generally, society's status beliefs (Ridgeway, 2001) organize cooperation across unequal groups, such as traditional gender roles or black-white relations in the Old South. The legitimating myths of SDO (Sidanius, Levin, Federico, & Pratto, 2001) are ideologies that intellectually and morally justify the superiority of high-status groups in the existing social structure. For example, beliefs in the genetic superiority of whites or men or heterosexuals would be a legitimating myth; pseudoscience has often been used to justify the status quo.

Summary of Blatant Bias

Blatant, overt forms of bias sometimes result from direct conflict over tangible resources, as suggested by realistic group conflict theory, but not as often as people suppose. More often, bias against outgroups results from favoring the ingroup, but seeing it as threatened in less tangible ways. Social identity theory, self-categorization theory, and optimal distinctiveness theory all address people's attempts to maintain a positive and distinct social identity. This tends to result in seeing group members as relatively homogeneous, typically favoring the ingroup, and boosting state self-esteem. Individual differences in authoritarianism, viewing the world as a dangerous place that threatens ingroup values, operate in parallel with social dominance orientation, viewing the world as a competitive hierarchy that threatens ingroup status. Both sets of beliefs work to justify the existing social arrangements, with some groups on top and others subordinate. In all these theories, people's core social motives appear in people's need to belong to a group (identity); to control resources, status, and values (threats); and to enhance themselves (feeling good as a group member).

EFFECTS OF BIAS ON TARGETS: BELONGING, CONTROLLING, AND SELF-ENHANCING

At first glance, Anglo whites often assume that members of racial and ethnic minorities in the United States must suffer from low self-esteem. Quite the contrary is true; blacks, for example, have self-esteem higher than whites (Crocker, Major, & Steele, 1998), at least in recent years (Twenge & Crocker, 2002). People from most groups subject to stereotyping, prejudice, and discrimination can still feel good about themselves and their groups, precisely as social identity theory would argue. In doing so, they face the challenge of affirming their identity as belonging to a group devalued by society. Stigma(devaluation due to a marked identity) adheres to their group and to themselves as a member of it. Nevertheless, a positive social identity buffers the effects of stigma. The processes are various and not simple, involving specific identity issues, stereotype threat, and attributional ambiguity, as we shall see.

Collective Identity and Well-being

A person's collective self-esteem (CSE) (Crocker & Luhtanen, 1990; Luhtanen & Crocker, 1992) is the value one places on one's social groups, as a function of social identity, or perceived memberships in various social groups, including, for example, race, ethnicity, gender, age, weight, college, geographic region, and so on. As Table 11.6 indicates, CSE has four components. Two elements of CSE—(a) feelings about one's worth as a group member and (b) one's personal, private regard for one's group—predict psychological well-being (i.e., depression, life satisfaction, hopelessness), even controlling for personal self-esteem. Another component, (c) public regard for one's group, does not correlate with private regard for one's group among blacks, but the correlation is high among Asians (with whites in the middle; Crocker, Luhtanen, Blaine, & Broadnax, 1994). A fourth component, (d) strength of identity, does not relate directly to well-being.

TABLE 11.6 Collective Self-Esteem Scale

Subscale and Sample Items

Membership

I am a worthy member of the social groups I belong to. I am a cooperative participant in the social groups I belong to.

Private

In general, I'm glad to be a member of the social groups I belong to. I feel good about the social groups I belong to.

Public

Overall, my social groups are considered good by others. In general, others respect the social groups that I am a member of.

Identity

The social groups I belong to are an important reflection of who I am. In general, belonging to social groups is an important part of my self-image.

The collective self-esteem of people from stigmatized groups depends on the situation, on which identity is salient, and on which social representation of that identity is salient (Crocker, 1999). In this view, self-worth is constructed, depending on the social context. Various dimensions of social identity vary with the situation. For African Americans, for example, the centrality (importance) of their own racial group identity increases when race is salient (e.g., the participant is the only racial ingroup member among others watching a video of black-on-white violence) (Shelton & Sellers, 2000). Momentary salience of racial identity is an important component, according to the multidimensional model of racial identity (Sellers, Smith, Shelton, Rowley, & Chavous, 1998).

Many aspects of collective identity function as relatively stable individual differences. For example, the Multidimensional Inventory of Black Identity(MIBI) (Sellers, Rowley, Chavous, Shelton, & Smith, 1997) suggests that identity consists of centrality (sense of belonging), ideology (from assimilationist, to humanist, to oppressed, to nationalist), and private regard (own esteem for blacks and self as black). Individual differences on the MIBI predict frequency of interracial interactions and choosing black studies courses. Some individuals' identity configurations are notable—those for whom race is central to identity, ideology is (black) nationalist, and beliefs maintain that other groups have negative attitudes toward blacks. This identity configuration predicts reported experiences of discrimination but also buffers against their mental health consequences, such as depression, anxiety, and stress. Otherwise, people are bothered by experiences of discrimination, even at the level of micro-aggressions (hassles such as being followed around a store or being overlooked for service) (Sellers & Shelton, 2003).

The dilemma of belonging to a stigmatized group is real. Race- and ethnicity-based stressors can undermine physical and mental health (Allison, 1998; Clark, Anderson, Clark, & Williams, 1999; Contrada et al., 2000). Gender discrimination also undermines women's psychological well-being (Swim, Hyers, Cohen, & Ferguson, 2001). Nevertheless, stigmatized group members are not passive targets of discrimination; instead, they cope actively (Heatherton et al., 2000; Swim & Stangor, 1998). When blacks interact with whites whom they expect to be prejudiced, they can counteract prejudice, making both people enjoy the interaction more, though the black participant may feel less true to self (Shelton, 2003).

Problems of racial bias require researchers to examine all sides of the interaction (Shelton, 2000). Ingroup dynamics, as well as outgroup dynamics, matter to intergroup encounters (Prentice & Miller, 2002). For example, ethnic identity makes people more aware of discrimination, especially when it is subtle or ambiguous (Operario & Fiske, 2001b; Shelton & Sellers, 2000). Perceiving discrimination cuts both ways: It adds stress, but it also allows coping.

As one negative example, exaggerated stigma consciousness (vigilance when interacting with outgroup members) can make one forego opportunities to invalidate stereotypes or even create a negative loop of expecting prejudice, reacting negatively, and eliciting negative behavior (Pinel, 1999, 2002). Even a momentary, role-played stigma can create a paranoid stance (Santuzzi & Ruscher, 2002). Nevertheless, being an oblivious victim is hardly healthy. For example, women believe it's best to respond to sexist jokes, though only a few do respond because of the social costs (Swim & Hyers, 1999). Women do not report sexual harassment nearly as often as people think they do; indeed, the most common response is to avoid the perpetrator (Fitzgerald, Swan, & Fischer, 1995). Perceiving prejudice has both costs and benefits; people's vigilance and overt responses depend on those factors (Barrett & Swim, 1998).

Attributional Ambiguity

All negative life events cause people to appraise the degree of threat, which then motivates coping. One of the ways that people can cope with bad outcomes, if they have a stigmatized identity, is by attribution to the other person's prejudice (Major, Quinton, & McCoy, 2002). When individuals from stigmatized groups have a negative encounter, they could attribute its cause either to something about themselves personally or to prejudice against their group membership; that dilemma is termed attributional ambiguity (Crocker & Major, 1989). In one study, women received negative feedback from a male evaluator whom they knew to be either prejudiced or not. When he was prejudiced, they attributed his feedback to that prejudice and reported feeling less depressed than women evaluated by an unprejudiced man. In a related study, black students received either negative or positive feedback from a white evaluator who could see them (and therefore their race) or not. Attributions to prejudice protected their self-esteem from the negative feedback but injured their self-esteem after positive feedback because they could not take credit for their own success (Crocker, Voelkl, Testa, & Major, 1991). According to correlational studies, people who consistently attribute their negative outcomes to discrimination have lower self-esteem than those who do not (Major, Quinton, & McCoy, 2002). Experimental studies show that when clues to prejudice are clear, attributions to it do enhance self-esteem. In real life, clues to prejudice are often more ambiguous, and a negative outlook explains both perceived discrimination and low self-esteem.

People differ in the extent to which they use attributions to prejudice. For example, consider black conservatives such as Supreme Court Justice Clarence Thomas, who minimize the role of prejudice in the lives of black Americans. People who believe in the ideology of individual opportunity are less likely to attribute any bad outcomes to discrimination (or any good ones to affirmative action) (Major, Gramzow, McCoy, Levin, Schmader, & Sidanius, 2002). Stigmatized groups also differ. For example, women who feel overweight blame negative evaluations on their weight, not on the evaluator's prejudice, and their self-esteem suffers (Crocker, Cornwell, & Major, 1993). This is especially true if they subscribe to the Protestant ethic notion that their weight is due to moral failings of self-indulgence and laziness (Quinn & Crocker, 1999). But otherwise, when people from most stigmatized groups blame negative outcomes on prejudice, it buffers them against taking it personally.

Of course, one cannot take credit for positive feedback, if one attributes it to prejudice of the overcompensating, positivity bias sort. Affirmative action can cause this kind of dilemma if people think they were hired only because of their membership in a protected category. When women think they were picked solely on the basis of gender, they attribute their own outcome as less due to merit and see it as less important (Major, Feinstein, & Crocker, 1994).

By law, affirmative action requires that people meet a certain merit standard and then adds gender (or race) as an additional criterion (e.g., Plous, 1996). This framing tends to make affirmative action seem more reasonable to people: All else equal, pick the person from the underrepresented group, because diversity helps the institution's effective functioning. (For more on the social psychology of affirmative action, see Skedsvold & Mann, 1996; Turner & Pratkanis, 1994.)

Two other mechanisms protect the stigmatized (Crocker & Major, 1989). One involves selective ingroup comparisons: comparing one's own outcomes with those of other ingroup members, thereby ignoring the outgroup (especially effective when they are doing better). People whose outcomes compare unfavorably with outgroup members (e.g., the dominant group) are unaffected emotionally, presumably because the advantaged outgroupers are seen as irrelevant. But people whose outcomes compare unfavorably with ingroup members feel depressed and attribute less ability to themselves (Major, Sciacchitano, & Crocker, 1993).

Another protective mechanism involves selectively devaluing domains that disadvantage one's own group and valuing domains that advantage one's group (Schmader & Major, 1999). People apparently do this when they view group status differences as legitimate and the other group ranks lower than theirs (Schmader, Major, Eccleston, & McCoy, 2001). This is presumably the case in which it is easiest to justify devaluing the domain. However, the devaluation hypothesis is most related to its cousin, recent research on stereotype threat.

Stereotype Threat

Stereotypes impute a variety of ability differences to a variety of groups. Members of groups described as low ability face a predicament in any relevant achievement setting. If they fail, their failure reflects not only on themselves but also on their group, confirming the worst stereotypes about it. This predicament—called stereotype threat (Steele, 1997; Steele, Spencer, & Aronson, 2002)—actually lowers performance when group membership is salient in the testing situation, presumably because the threat directly interferes with the performance of stigmatized group members. Over time, stereotype threat may cause affected individuals to disidentify with the domain, that is, to dismiss it as unimportant and direct their energies elsewhere. In one study, for example, black students were unaffected by (bogus) failure on an intelligence test, due to both disengagement of self-esteem from being contingent on that domain (Major, Spencer, Schmader, Wolfe, & Crocker, 1998).

The classic stereotype threat studies show the stigmatized group underperforming only when their group membership is made salient, for example, by requiring them to list their race at the beginning of a test (typical test format), by telling them that the test is diagnostic of ability (the normal condition), or by making them the only member of their group present (all too frequent for minorities). When these conditions are removed, the stigmatized group performs as well as the control group. Stereotype threat occurs for everyone under relevant circumstances: blacks on intellectual tests, Latinos and women on math tests, men on emotional sensitivity tests, whites on athletic performance, and people of lower socioeconomic status on intellectual ability tests (Croizet & Claire, 1998; Gonzales, Blanton, & Williams, 2002; Inzlicht & Ben-Zeev, 2000; Leyens, Désert, Croizet, & Darcis, 2000; Spencer, Steele, & Quinn, 1999; Steele & Aronson, 1995; Stone, Lynch, Sjomeling, & Darley, 1999). Members of any group can suffer stereotype threat, even if they are normally stereotyped as adept. Thus, under stereotype threat, white men outperform white women, but the white men underperform when tested alongside Asian men (Aronson et al., 1999). Similarly, Asian women underperform when their gender is salient but outperform when their ethnicity is salient (Shih, Pittinsky, & Ambady, 1999), and children as young as kindergarten show these effects (Ambady, Shih, & Pittinsky, 2001). Stereotype threat occurs outside the laboratory in thousands of college students (Massey & Fischer, 2006). The effects of stereotype threat are not trivial, even presenting a health risk. For blacks under stereotype threat, blood pressure reliably increases (Blascovich, Spencer, Quinn, & Steele, 2001), suggesting a stress response that, if chronic, could contribute to hypertension. And, of course, ability tests have widespread significance in people's lives.

Why should stereotype threat cause people to underperform when the evidence indicates stereotype-threatened people are trying even harder, not simply failing because it is expected (Steele et al., 2002)? Stereotype threat is not the self-fulfilling prophecy, described earlier, in which someone encounters a prejudiced person and subtle pressures to conform to a negative stereotype. Instead, the person must only be a member of a group stigmatized with regard to the particular achievement domain, care about the domain, and attempt to perform well but be undermined by the threat in the air. Attempts to uncover mediation of stereotype threat suggest both motivation and cognition (Steele et al., 2002; Wheeler & Petty, 2001). Some evidence, but not all, suggests anxiety. Some evidence, but not all, suggests stereotype activation. Some evidence, but not all, suggests lower expectations for performance (Kray, Galinsky, & Thompson, 2002; Stangor, Carr, & Kiang, 1998; but see Steele et al., 2002). Some suggestive evidence proposes that efforts to suppress the stereotype may have the ironic effect of keeping it activated and so interfering with performance (Steele et al., 2002). Some combination of vigilance and stress seems likely.

Remedies for stereotype threat occur at the level of affected individuals, contexts, and institutions (Steele et al., 2002). For individuals, an understanding that intelligence is not a fixed entity, but malleable, may undermine stereotype threat. Just as entity theorists stereotype other people more (Plaks et al., 2001), so they themselves may be more vulnerable to stereotype threat (Aronson, Fried, & Good, 2002). Individuals could also try to identify less with their own group (Schmader, 2002), but such a strategy is unlikely to have wide appeal. On the other hand, realizing that race is a social construction, as multiracial people do, can buffer against stereotype threat (Shih, Bonam, Sanchez, & Peck, 2007). Affirming the self, identity, and values works well (e.g., Sherman et al., 2013).

At the level of immediate context, presenting a test as insensitive to group differences or as fair to all groups can prevent stereotype threat. Similarly, presenting the setting's actual diversity or commitment to diverse groups can increase trust, which averts stereotype threat (Steele et al., 2002). Also, a temporary psychological disengagement from the immediate situation can bolster persistence and performance (Nussbaum & Steele, 2007).

At the level of longer-term institutional factors, mentors can advocate both high standards and the student's ability to meet them, thereby motivating effort (Cohen, Steele, & Ross, 1999). Interventions that mitigate doubts about minority belonging in the college can raise academic achievement (Walton & Cohen, 2007). The institution can also encourage opportunities for cross-race friendships, which the next section will show can prove helpful to all concerned.

Summary of Target Responses to Bias

Belonging, controlling, and self-enhancing each motivate bias's effects on members of stigmatized groups. Collective self-esteem is the value one places on one's collective identity, belonging to a social group. Collective identity can make people more aware of subtle forms of prejudice but can also buffer against its ill effects, protecting well-being and enhancing the self. People's particular collective identity depends both on momentary situational factors and on stable individual differences. People's degree of vigilance depends on the costs and benefits of perceiving discrimination, including their ability to control it or not.

Attributional ambiguity, interpreting one's outcomes as due to prejudice or personal factors, presents a dilemma for targets of prejudice. Attributing outcomes to prejudice can enhance the self in the case of negative outcomes but deprive the self of credit for positive outcomes. People and groups differ in the extent to which they use this strategy, and the extent to which it works when they do.

Stereotype threat describes people's response to a threat in the air, in performance situations, when failure not only could be personal but also could confirm stereotypes about one's group. People deal with this predicament by increased vigilance, which ironically undermines their performance. However, understanding the malleability of intelligence, perceiving a test as unbiased, and trusting the setting all can avert stereotype threat.

STRATEGIES FOR CHANGE: CONSTRUCTIVE INTERGROUP CONTACT CAN CONTROL BIAS

Friendship cures more than stereotype threat. Intergroup contact—that is, interactions between individual members of socially distinct groups—has proved tried and true, but only under specific circumstances. Gordon Allport (1954b) formalized the conditions of constructive intergroup contact:

· Equal status of the groups within the context

· Common goals

· Cooperation, or at least no competition

· Authority sanction for the contact

These conditions make complete sense, but they are not easy to meet; we are still learning what works (Paluck & Green, 2009).

In one well-known application, children in integrated classrooms are made to be fully interdependent (combining common goals and cooperation): In small groups, they need each other because each is responsible for teaching the others part of the day's lesson. This resulting jigsaw classroom (Aronson, 1990; Aronson & Bridgeman, 1979; Aronson & Osherow, 1980) improves their liking for each other across group lines and improves self-esteem, morale, and the academic performance of minorities, without undermining the performance of majority students.

Generally, if the groups do perceive that they are coming together as equal partners, on common ground, that they need each other, and that those in power want them to get along, contact works well to reduce prejudice. In a meta-analysis of 376 studies and more than 156,000 participants, Pettigrew and Tropp (2006) found clear support for the beneficial effects of intergroup contact and for the importance of meeting as many of the constructive conditions as possible. The listed conditions are not each necessary, but instead each facilitates people's overcoming their hostility and discomfort.

Causal direction matters here: Do prejudiced people shun contact, and tolerant people seek it, so of course contact correlates with tolerance? Or does intergroup contact itself lower people's prejudice? No doubt both could be true, but the contact hypothesis needs to show the latter in isolation from the former. And, indeed, the more rigorously the study tests contact actually causing tolerance, the stronger the effect (experiments' effects are stronger than correlational studies). The more the study removes reverse causality, by allowing participants no choice about the contact, the stronger the effect. Nevertheless, both randomly assigned and voluntary contact generally reduces prejudice (Van Laar, Levin, Sinclair, & Sidanius, 2005).

Emotional prejudices show bigger improvement as a result of contact than stereotypic beliefs do. As in other research on bias, emotions drive the process more than cognitions do (Dovidio et al., 1996; Talaska, Fiske, & Chaiken, 2008; Tropp & Pettigrew, 2005). People feel anxious and threatened in intergroup encounters, as attested both by self-reports (Stephan & Stephan, 2000) and by physiological indicators such as cardiovascular reactivity (Blascovich, Mendes, Hunter, Lickel, & Kowai-Bell, 2001). The emotional strain has consequences for people's ability to self-regulate during and after such encounters: Neuroimaging and cognitive performance both show that interracial contact depletes executive function, for minority and majority alike (Richeson et al., 2003; Richeson, & Shelton, 2003; Richeson, & Trawalter, 2005; Richeson, Trawalter, & Shelton, 2005). Overcoming automatic reactions—a necessary feature of constructive contact—requires executive function (Payne, 2005).

Much of the contact effect lies in the opportunity for intergroup friendships (Pettigrew, 1997, 1998a). The effects are the best for outgroups that present the best opportunity for friendship and the least opportunity for threat once comfort is established (Pettigrew & Tropp, 2006). For example, intergroup contact with homosexuals improves attitudes the most; contact with ethnic and racial minorities shows average effects; and contact with physically and mentally disabled people shows the least. In contact settings that provide the most opportunity for friendship—such as workplaces over the long term—effects are stronger than in short-term recreational or travel contact. Even just knowing about other ingroup members' friendships with outgroup members can improve intergroup attitudes (Wright, Aron, McLaughlin-Volpe, & Ropp, 1997).

Moreover, contact's beneficial effects generalize to other members of the outgroup and to other outgroups (Pettigrew & Tropp, 2006). Contact diversifies people's friendships, which in turn generalizes to more positive feelings about a variety of outgroups (Pettigrew, 1997). Multicultural and multiracial friendships represent the intersection of interpersonal and intergroup relations (Gaines & Liu, 2000). Intergroup friendships must develop freely, through voluntary mutual disclosure and responsiveness (Shelton, Trail, West, & Bergsieker, 2010).

How does such contact work? When people are interdependent, when they need each other to complete a joint task for a shared reward, they go to more trouble to learn about each other (Erber & Fiske, 1984; Fiske, 2000a; Neuberg & Fiske, 1987; Ruscher & Fiske, 1990). They selectively attend to the most diagnostic cues: information that disputes their stereotypes about the other person's group. They then use this counterstereotypic information to make personalized dispositional inferences about the other person. As a result, they form more idiosyncratic, individuating impressions (based less on the other person's category membership and relatively more on their own interpretations of the other person's characteristics). All of this undercuts stereotypes and allows the possibility of individual friendship.

In intergroup contact, people do not necessarily ignore each other's group membership; indeed, if people remember each other's categories while getting to know them, they may be honoring that person's social identity. For example, immigrants may value their culture of origin as well as the culture of their host country (Berry, 2001). Thus, for the positive effects of intergroup contact to generalize beyond the immediate groups of people directly in contact, mutual differentiation may be important for two reasons. People need to retain some awareness of each other's social identity in the contact situation. Also, group members need to seem typical, in the sense of being representative of their group (Brown, 2000; Brown, Vivian, & Hewstone, 1999; Greenland & Brown, 1999; Hewstone & Brown, 1986; Hewstone, Rubin, & Willis, 2002).

This social identity perspective contrasts with various alternative ideas about optimal outcomes of intergroup contact. One perspective argues that people should decategorize altogether, minimizing their prior social categories (Miller & Brewer, 1986), but another perspective argues that people should recategorize, using a common ingroup identity, a new more-inclusive group that supercedes prior separate group memberships (Dovidio, Kawakami, & Gaertner, 2000). When groups are temporary, decategorization works well. But the common ingroup identity also works well, mediating changes for both temporary and longer-term groups (Dovidio, Gaertner, & Validzic, 1998; Dovidio et al., 1997; Terry, Carey, & Callan, 2001). In the end, “we're different groups, but all on the same team,” seems to be the key (Dovidio et al., 2000; Gaertner & Dovidio, 2000). A variant, the cross-categorization approach (Brewer, 2000; Marcus-Newhall, Miller, Holtz, & Brewer, 1993), maintains at least two separate but cross-cutting category systems (such as occupation and ethnicity, or gender and team), in which people represent all combinations. People who are outgroup on one dimension but ingroup on another fare better than people who are double outgroup members, though not as well as double ingroup members (Migdal, Hewstone, & Mullen, 1998; Mullen, Migdal, & Hewstone, 2001; Urban & Miller, 1998). Evidence supports a balance between maintaining one's various identities and finding identities in common with other individuals, under as many as possible of Allport's main criteria for constructive contact (Park & Judd, 2005).

Ultimately, contact works well when people do not merely tolerate differences but are enthusiastic about meeting people who differ from them on some dimensions. Approaching the outgroup member, rather than merely not avoiding, improves contact experiences (Kawakami, Phills, Steele, & Dovidio, 2007). Outgroup orientation includes liking to meet, get to know, and spend time with people from other ethnic groups (Wittig & Molina, 2000)—and it mediates the positive effects of intergroup climate on prejudice reduction.

CHAPTER SUMMARY

Biases operate against both group members and individuals who are stigmatized. Such category-based responses to outgroups include emotional prejudices, cognitive stereotypes, and discriminatory behaviors. Discrimination ranges from verbal hostility and avoidance all the way to segregation, physical attack, and group extermination. Operationally, social psychologists study explicit, overt forms of bias on self-report questionnaires and blatant forms of aggression. The implicit, subtle forms of bias appear in millisecond stereotypic and prejudicial associations, as well as in nonverbal behavior, both of which are less controllable than verbal and other overt behaviors.

Various core social motives operate to sustain bias. Most important is belonging to one's own group, at the expense of the outgroup. Biases then further our efficient shared social understandings of those outgroups. Controlling the perceived unpredictability of the outgroup and controlling perceived threats to the ingroup both contribute to biases. Enhancing the self as having a positive, distinct ingroup identity also enters into the motivational mix.

Subtle forms of bias build understandings that protect the self. Whites' racial attitudes have changed over the century, but mostly in overt ways that protect the self-image as egalitarian. In more subtle ways, such as nonverbal behavior, bias persists. The new, implicit forms of bias include the cool and indirect forms reflected in modern racism and subtle prejudice scales that show patterns of policy preferences that all blame the outgroup, oppose support for minorities, exaggerate intergroup differences, and make ingroup-serving attributions. Automatic forms of bias appear in responses to subliminal intergroup cues and millisecond priming, all automatic associations that advantage the ingroup, match the stereotype, reinforce categorization, and disadvantage outgroups. The subtle forms of bias—besides being cool, indirect, automatic—also are ambiguous, providing excuses for discriminatory behavior by people who find the thought of their own racism to be aversive. Subtle bias is also ambivalent, with prime examples being racism and sexism, which include subjectively benevolent paternalistic responses as well as the more obvious hostile ones. Many outgroups are pitied (liked but disrespected) or envied (respected but disliked).

Subtle forms of bias are socially useful, facilitating both ingroup belonging and control over perceived threats. Automatic forms of subtle bias produce deniable discrimination in the unconscious nonverbal priming of stereotype-consistent behavior (e.g., walking more slowly after being primed with elderly stereotypes). Self-fulfilling prophecies elicit from targets the very behavior that confirms stereotypes, again operating largely via nonverbal cues and, ironically, smoothing the intergroup interactions. Ingroup communication about outgroup stereotypes helps to build ingroup cohesion, as people trade stories about us versus them, using easily communicated ideas and linguistic biases that make the outgroup's negative behaviors seem especially diagnostic of stable traits. People use the same term in different ways, depending on the group (“good parent” for a man implies different standards than “good parent” for a woman). Such issues complicate the question of how accurate stereotypes may or may not be. Some challenges to this line of work would include picking fair and unbiased criteria, dimensions, observers, cut-offs, and distributions. All this has to take account of the unequal distribution of social roles, whereby groups overrepresented in particular roles (housewife, welfare recipient, enemy) are attributed traits consistent with those roles. Subtle bias is controllable to the extent that people have the motivation and capacity to understand outgroups differently.

Blatant forms of bias focus even more on cementing ingroup belonging in the face of perceived outgroup threat, thereby enhancing the self. Realistic group conflict theory addresses actual conflicts over resources, although these matter less than perceived conflict and symbolic resources. Social identity theory and related efforts focus on the psychology of belonging to a group, in contrast with acting as an individual. People seek positive, distinct social identities, and they minimize within-group differences, while maximizing between-group differences, in direct comparisons and normative interpretations. Self-categorization, optimal distinctiveness, and uncertainty reduction all are served by ingroup belonging and favoritism, to the exclusion of the outgroup. State self-esteem is served by differentiating between ingroup and outgroup.

In addition to variations in social identity, other important personality differences are authoritarianism and social dominance orientation. Authoritarianism, in its revised form, right-wing authoritarianism, emerges from punitive socialization, which encourages social conformity, in the face of a world perceived to be dangerous; the resulting RWA endorses authoritarian submission to the powerful and aggression on their behalf, favors the threatened ingroup and harbors prejudice toward the dangerous outgroups. In parallel, social dominance orientation emerges from unaffectionate socialization, which encourages tough-mindedness, in the face of a world perceived to be relentlessly competitive; the resulting SDO endorses social hierarchy, ingroup favoritism, and outgroup prejudice. Both RWA and SDO result in some of the most virulent forms of blatant prejudice.

Both subtle and blatant biases affect their stigmatized targets, but not in the most obvious ways. Collective self-esteem can buffer individuals from the consequences of bias, depending on identity salience in the situation. Individual differences in racial identity's centrality, ideology, and regard similarly can buffer people against discrimination, although highly identified individuals are more aware of the discrimination in the first place. Hypervigilant stigma consciousness can be detrimental to people's well-being. Likewise, attributing negative outcomes to prejudice can buffer people's self-esteem, but it can also undermine the self-affirming effects of positive outcomes that are attributed to race. Other buffering strategies include comparing self to ingroup others, rather than the potentially threatening outgroup, and devaluing domains where the ingroup fares poorly.

Stereotype threat, the focus of a particularly active area of research, describes how members of any group can experience a threat in the air, based on the stereotype that their group performs poorly on some dimension. The mere reputation suffices to make people underperform; unlike a self-fulfilling prophecy, one need not meet a prejudiced individual. Identity salience, again, depends on context, and therefore so does stereotype threat. The most likely mechanisms of stereotype threat are probably vigilance and stress. Stereotype threat is remedied by knowing that intelligence is malleable and improvable, not fixed; by fair-minded contexts that minimize group differences; and by larger settings that demonstrate a commitment to diversity and trust.

Constructive intergroup contact can mitigate bias and its effects on both perceivers and targets. It requires conditions that provide the opportunity for intergroup friendship: equal status, common goals, cooperation, and authority sanctions. The jigsaw classroom exemplifies intergroup contact, but the concept has endured for decades, and the effects are strong. The mechanisms involve people's mutual attention to each other in the service of furthering their interdependent goals, which allows them to go beyond group membership, while not ignoring it. The salutary effects of contact generalize when people discover cross-cutting categories or a common ingroup identity. In the end, enthusiastic openness to people from other groups paves the way for improving intergroup relations.