The Nature of Contemporary Prejudice: Insights from Aversive Racism



Within the United States, declines in the overt expression of racial prejudice over several decades have given way to near universal endorsement of the principles of racial equality as a core cultural value. Yet, evidence of persistent and substantial disparities between Blacks and Whites remain. Here, we review research that demonstrates how the actions of even well-intentioned and ostensibly non-prejudiced individuals can inadvertently contribute to these disparities through subtle biases in decision making and social interactions. We argue that current racial attitudes of Whites toward Blacks in the United States are fundamentally ambivalent, characterized by a widespread contemporary form of racial prejudice, aversive racism, that is manifested in subtle and indirect ways, and illustrate its operation across a wide range of settings, from employment and legal decisions, to group problem-solving and everyday helping behavior. We conclude by describing research aimed at combating these biases and identify new avenues for future research.

More than four decades after the Civil Rights Act was signed into US law granting Blacks and Whites equal access to public settings and institutions, racial and ethnic divisions continue to permeate American society. At the heart of these divisions is a fundamental paradox of the American identity. It is a society at once founded on the principles of justice and equality on the one hand, and built upon racist traditions and marred by a legacy of slavery on the other. Although this paradox may in part reflect a unique cultural and political history (an ‘American dilemma’, as first described in 1944 by Swedish economist Gunnar Myrdal), the psychological forces that shape this conflict are pervasive and enduring. Both within and outside the United States, unprecedented global immigration creates new challenges psychologically as well as socially and economically for host societies. As recent waves of ethnic conflict in Europe and East Asia attest, the perception of differences in values, beliefs, and customs among immigrants and citizens can trigger open conflict and violence (Stephan, Renfro, Esses, Stephan, & Martin, 2005). However, in countries that have historical traditions or contemporary norms of egalitarianism that discourage not only the expression but also the personal acknowledgement of bias, prejudice may not be expressed blatantly but often in more subtle, yet equally pernicious, ways (see Pettigrew & Meertens, 1995).

Here, we describe one psychological legacy of the American dilemma –‘aversive racism’– a form of prejudice characterizing the thoughts, feelings, and behaviors of the majority of well-intentioned and ostensibly non-prejudiced White Americans. Although we focus on race relations within the United States, we note that similar processes have been observed within members of dominant groups in other nations, such as Canada (Son Hing, Chung-Yan, Hamilton, & Zanna, 2008), England (Hodson, Hooper, Dovidio, & Gaertner, 2005), and the Netherlands (Kleinpenning & Hagendoorn, 1993), in which overt forms of prejudice are similarly recognized as inappropriate. We begin by describing the historical origins and psychological underpinnings of aversive racism and consider its consequences for Whites treatment of Blacks, the quality and nature of interracial interactions, and race relations more generally. We then discuss research aimed at combating aversive racism and conclude by describing recent developments and new directions for future research.

Racial Attitudes: Historical Trends and Group Perspectives

Over the past 40 years, public opinion polls have revealed substantial declines in Whites’ endorsement of prejudiced views toward minority groups, and Blacks in particular, in the United States (Devine & Elliot, 1995; Gaertner & Dovidio, 1986; Madon et al., 2001; Schuman, Steeh, Bobo, & Krysan, 1997). In part due to the civil rights legislation of the 1960s, Whites increasingly support integration in schools, housing, jobs, and public transportation, as well as interracial marriage (Dovidio & Gaertner, 2004). Consistent with these trends toward greater endorsement of racial equality, national observances promoting greater cultural awareness and acknowledging historic social and political struggles of racial and ethnic minority groups have become a ubiquitous part of American culture. Despite these declines in overt prejudice, however, evidence of substantial racial disparities and discrimination remain – from important health indices, such as infant mortality, to disparities in medical treatment, earned wages, and access to and quality of a range of basic services, from health care and job training, to employment, housing, and education (Dovidio, Penner, Albrecht, Norton, Gaertner, & Shelton, 2008; Elvira & Zatzick, 2002; Geiger, 2003; Rosenfeld, 1998; Smedley, Stith, & Nelson, 2003; see also Smelser, Wilson, & Mitchell, 2001). Moreover, the different experiences of Blacks and Whites shape differing perceptions of the importance of race in US society. In recent national surveys, nearly three fourths of Blacks and only one third of Whites reported that racial discrimination is a major factor accounting for disparities in income and education levels (USA Today/Gallup, 2008). Whereas a vast majority of Whites (71%) reported that they were satisfied with the way Blacks are treated in society, a nearly equivalent proportion of Blacks (68%) reported that they were dissatisfied with the way Blacks are treated in the United States (Gallup Minority Rights and Relations Survey, 2007).

What might account for the discrepancy between the decline in Whites’ negative racial attitudes and the persistence of social, political, and economic disparities and discrimination experienced by Blacks and other minorities? One possibility is that explicit prejudice still exists but that Whites have become more aware of social norms against it and are therefore more guarded about public expressions of bias. Another possibility is that new forms of prejudice have emerged that are less deliberate (e.g., non-conscious and unintentional) and less direct than their historical counterparts. Both perspectives suggest that racial biases are now less blatant than in the past, and that new perspectives and techniques are needed to understand the depth and scope of contemporary racism.

The Nature of Aversive Racism

Research from the 1920s through the 1950s typically portrayed prejudice as a psychopathology (Dovidio, 2001). However, stimulated by developments in the area of social cognition, by the mid-1960s and early 1970s, much more attention was devoted to examining how normal, often adaptive, cognitive (e.g., social categorization), motivational (e.g., needs for status), and sociocultural (e.g., social transmission of beliefs) processes can contribute to the development of Whites’ biases toward Blacks (see Dovidio, 2001; Dovidio & Gaertner, 2004). Because of important historical roots, social categorization by race within the United States is largely automatic, where the actual or imagined presence of a Black person is often enough to automatically activate racial categories without conscious effort or control (Dovidio & Gaertner, 2004). Social categorization spontaneously activates more positive feelings and beliefs about in-group members (‘we's’) than out-group members (‘they's’; see Gaertner & Dovidio, 2000, for a review). In addition, Whites automatically activate stereotypes of Whites as intelligent, successful, and educated, and of Blacks as aggressive, impulsive, and lazy (Blair, 2001). Intergroup processes, such as system-justifying ideologies as well as perceived competition over material resources, can also form a basis for negative racial attitudes (see Sidanius & Pratto, 1999).

In 1970, Kovel first distinguished between dominative and aversive racism. Dominative racism is said to reflect the traditional, blatant form. According to Kovel, the dominative racist is the ‘type who acts out bigoted beliefs – he represents the open flame of racial hatred’ (p. 54). Aversive racists, in contrast, sympathize with victims of past injustice, support principles of racial equality, and genuinely regard themselves as non-prejudiced, but at the same time possess conflicting, often non-conscious, negative feelings and beliefs about Blacks that are rooted in basic psychological processes that promote racial bias (Gaertner & Dovidio, 1986). The negative feelings that aversive racists have toward Blacks typically do not reflect open antipathy, but rather consist of more avoidant reactions of discomfort, anxiety, or fear. That is, they find Blacks ‘aversive’, while at the same time find any suggestion that they might be prejudiced ‘aversive’ as well. Although other frameworks for understanding contemporary racial bias have been described, such as modern racism (McConahay, 1986), and symbolic racism (Sears, Henry, & Kosterman, 2000), all hypothesize a fundamental conflict between the denial of personal prejudice and underlying unconscious negative feelings and beliefs. Whereas modern and symbolic racism characterize the attitudes of political conservatives, aversive racism characterizes the biases of those who are politically liberal and openly endorse non-prejudiced views, but whose unconscious negative feelings and beliefs get expressed in subtle, indirect, and often rationalizable ways (Dovidio & Gaertner, 2004; Gaertner & Dovidio, 1986; Nail, Harton, & Decker, 2003).

Aversive racists are characterized as having egalitarian conscious, or explicit, attitudes but negative unconscious, or implicit, racial attitudes (Dovidio & Gaertner, 2004). Explicit attitudes operate in a conscious mode and are exemplified by traditional, self-report measures. Implicit attitudes, in contrast, are evaluations or beliefs that are activated by the mere presence (actual or symbolic) of the attitude object and often reside outside of awareness. Implicit attitudes and stereotypes are typically assessed using response latency procedures (e.g., see Gaertner & McLaughlin, 1983), memory tasks, physiological measures (e.g., heart rate and galvanic skin response), and indirect self-report measures (e.g., biases in behavioral and trait attributions). The Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998; see also Greenwald, Poehlman, Uhlman, & Banaji, forthcoming), for example, relies on the basic finding that people are typically faster at categorizing groups of stimuli stored in memory that are similar in valence compared to those that differ in valence, and measures the automatic stereotypic or evaluative (e.g., good–bad) associations that people have with different racial groups for which they may lack introspective awareness.

Methodological techniques for assessing implicit attitudes have become increasingly useful for differentiating aversive racists (those who endorse egalitarian values but harbor implicit racial biases) from individuals who are truly non-prejudiced (those who also endorse egalitarian ideals but do not harbor negative implicit biases; see Son Hing, Chung-Yan, Grunfeld, Robichaud, & Zanna, 2005). Consistent with the aversive racism framework, whereas the majority of Whites in the United States appear non-prejudiced on self-report (explicit) measures of prejudice, a similar percentage of Whites typically show evidence of racial biases on implicit measures that are largely dissociated from their explicit views (Dovidio, Kawakami, & Beach, 2001; Hofmann, Gawronski, Gschwendner, Le, & Schmitt, 2005). Thus, a substantial proportion of Whites in the United States can be characterized as exhibiting reactions toward Blacks consistent with aversive racism.

Consequences of Aversive Racism

In contrast to more traditional forms of racism, which are blatant and expressed openly and directly, aversive racism operates in more subtle and indirect ways. Specifically, whereas blatant racists exhibit a direct and overt pattern of discrimination, aversive racists’ actions appear more variable and inconsistent. Sometimes, they discriminate (manifesting their negative feelings), and sometimes, they do not (reflecting their egalitarian beliefs). Nevertheless, their discriminatory behavior is predictable.

Because aversive racists consciously recognize and endorse egalitarian values and truly aspire to be non-prejudiced, they will not act inappropriately in situations with strong social norms when discrimination would be obvious to others and to themselves. Specifically, when they are presented with a situation in which the normative response is clear (e.g., right and wrong are clearly defined), aversive racists will not discriminate against Blacks. In these contexts, aversive racists will be especially motivated to avoid feelings, beliefs, and behaviors that could be associated with racist intent. However, the non-conscious feelings and beliefs that aversive racists also possess will produce discrimination in situations in which normative structure is weak, when the guidelines for appropriate behavior are unclear, when the basis for social judgment is vague, or when one's actions can be justified or rationalized on the basis of some factor other than race. Under these circumstances, aversive racists may engage in behaviors that ultimately harm Blacks but in ways that allow Whites to maintain a non-prejudiced self-image and insulate them from recognizing that their behavior is not colorblind. Thus, although the processes through which contemporary biases emerge can often be subtle, the consequences can be severe.

Support for the aversive racism framework has been obtained across a broad range of experimental paradigms and participant populations, including emergency and nonemergency helping behavior inside and outside of the laboratory, selection decisions in employment and college admission, interpersonal judgments, and policy and legal decisions (see Dovidio & Gaertner, 2004). In addition, a recent study by Pearson, Dovidio, and Pratto (2007) found, as hypothesized within the aversive racism framework, that whereas blatant prejudice is characterized by feelings of antipathy and hate, the subtle bias associated with aversive racism is often ‘cooler,’ reflecting the recruitment of cognitive rationalizing processes in decision making contexts (e.g., legal judgments). Because the subtle bias associated with aversive racism occurs without personal awareness and the actions can be attributed, even by observers, to factors other than race, the influence of aversive racism commonly goes unrecognized by Whites. As a consequence, whereas blatant expressions of prejudice, such as hate crimes, are readily identified and inhibited by social sanctions, aversive racism is likely to persist relatively unchallenged over time. Here, we illustrate this persistence with research in three different areas: helping behavior, selection decisions in employment and university admissions, and legal decisions.

Helping behavior. Early tests of the aversive racism framework focused on prosocial behavior for both theoretical and practical reasons. Theoretically, because aversive racists are hypothesized to be particularly effective at censoring negative behavior toward Blacks, the biases associated with aversive racism may often manifest as differential prosocial responses toward Blacks and Whites in need. From a practical perspective, the Kerner Commission (Report of the National Advisory Commission on Civil Disorders, 1968), charged with investigating the causes of the 1967 race riots in the United States, cited White America's failure to assist Blacks in need, rather than actively trying to harm Blacks, as a primary cause of racial disparities and, ultimately, civil unrest. Indeed, it was research on the differential behavior of Whites toward Black and White motorists who were stranded on a highway that represented the first empirical work on aversive racism (Gaertner, 1973). In another early test of the aversive racism framework, Gaertner and Dovidio (1977) showed that Whites did not discriminate against a Black, relative to a White, victim in an emergency when they were the only witness to an emergency event and bore sole responsibility for helping. However, when White bystanders believed that others also witnessed the emergency, and could therefore justify not helping on the basis of an ostensibly non-racial factor (the belief that responsibility was shared and that someone else could intervene), they were substantially less likely to help a Black, than a White victim. Although bias was expressed subtly, the consequences were severe.

Importantly, the results of a recent meta-analysis of 31 experiments on Whites’ interracial helping behavior conducted over the past 40 years evidence a stable pattern of discrimination reflective of aversive racism that has not subsided over time (Saucier, Miller, & Doucet, 2005). Based on these findings, the authors concluded that racism and discrimination against Blacks ‘can and will exist as long as individuals harbor negativity toward Blacks at the implicit level’ (p. 14).

Selection decisions. Another domain in which aversive racism was hypothesized to emerge was in Whites’ education and employment selection and evaluation decisions. In particular, based on the aversive racism framework, we expected that Whites’ biases against Black job applicants and employees, such as in the cases of hiring and promotion, would likely not surface when candidates have impeccably strong qualifications but, rather, would manifest in the more common grey area when a candidate's qualifications are more marginal and could be argued either way on grounds that are ostensibly unrelated to race.

Dovidio and Gaertner (2000) examined this hypothesis in a study of White college students’ hiring recommendations for Black and White applicants for a selective campus position within the same college in the years 1989 and 1999. Consistent with the aversive racism framework, when the candidates’ credentials clearly qualified or disqualified them for the position (strong and weak qualification conditions), there was no discrimination against the Black candidate. However, when candidates’ qualifications for the position were less obvious and the appropriate decision was more ambiguous (moderate qualifications), White participants recommended the Black candidate significantly less often than the White candidate with exactly the same credentials. Consistent with the meta-analytic findings of Saucier et al. (2005), when the responses of participants in 1989 and 1999 were compared, whereas overt expressions of prejudice (measured by items on a self-report scale) declined over this 10-year period, the pattern of subtle discrimination in selection decisions remained essentially unchanged. More recently, Otero and Dovidio (2005) conceptually replicated this research with human resource professionals in Puerto Rico, focusing on the moderate and strong qualification conditions. The findings illustrate the generalizability and persistence of these effects even among experienced professionals in the field.

Additional research has offered further insight into processes that underlie these effects. Hodson, Dovidio, and Gaertner (2002) investigated how Whites who are high or low in prejudice (based on explicit self-report measures) justify biased decisions against Blacks in selection procedures. Analogous to the Dovidio and Gaertner (2000) experiment, participants evaluated applicants for college admission who had consistently strong records of accomplishment (very strong college board scores and strong academic records) or had more mixed qualifications (strong on one dimension and modest on the other). As expected, no anti-Black bias was found among higher and lower prejudiced college participants when applicants had uniformly strong college board scores and records of high school achievement. However, when applicants were strong on one dimension (e.g., on college board scores) and relatively weak on another (e.g., high school grades), Black applicants tended to be recommended less strongly than were White applicants by more highly prejudiced Whites. Moreover, to justify their decisions, these participants systematically changed the importance they ascribed to various admissions factors as a function of applicants’ race. For Black applicants, higher prejudice participants assigned the weaker dimension (either college board scores or grades) greater weight in their decisions, whereas for White applicants, the stronger of the qualifications received greater weight (see also Brief, Dietz, Cohen, Pugh, & Vaslow, 2000, for employment bias against Blacks; and Uhlmann & Cohen, 2005, for evidence of similar biases against women).

Son Hing et al. (2008) investigated another form of racial bias, discrimination against Asian job applicants in Canada, and directly explored the role of unconscious (implicit) biases in this process. Paralleling the findings of subtle bias against Blacks in the United States, these researchers found that when assessing candidates with more moderate qualifications, evaluators recommended White candidates more strongly for the position than Asian candidates with identical credentials. However, when evaluating candidates with exceptionally strong qualifications, no such selection bias emerged. Moreover, Son Hing et al. obtained direct evidence of the role of implicit attitudes in the selection process. Implicit bias against Asians (as measured by an IAT), but not explicit prejudice, predicted weaker support for hiring Asian candidates who had moderate qualifications. However, when the Asian candidate had distinctively strong qualifications (and a failure to hire the applicant was not justifiable on a factor other than race), neither implicit nor explicit prejudice predicted the hiring decision. Support for the Asian applicant was generally strong in this case.

Legal decisions. Finally, another important line of evidence of the persistence and prevalence of aversive racism comes from experiments of simulated juror decision making. In one such study, Johnson, Whitestone, Jackson and Gatto (1995) examined the impact of introducing inadmissable (and non race-related) evidence that was damaging to a Black or White defendant's case on Whites’ judgments of the defendant's guilt. Replicating and extending earlier work (Faranda & Gaertner, 1979), the authors found no evidence of bias as a function of defendant race when exposed to only admissible evidence. However, consistent with aversive racism, exposure to potentially incriminating evidence deemed inadmissible by the court increased perceptions that the Black, but not White, defendant was guilty of the crime. Furthermore, when probed about their decisions, participants’ reported that they believed that the inadmissible evidence had less effect on their decisions when the defendant was Black than when the defendant was White, suggesting the unconscious and unintentional nature of their bias. Hodson et al. (2005) recently replicated these findings in the United Kingdom using a similar paradigm, demonstrating the cross-national pervasiveness of this phenomenon.

Several other studies on legal decision making (e.g., Dovidio, Smith, Donnella, & Gaertner, 1997a; Knight, Guiliano, & Sanchez-Ross, 2001) have yielded similar evidence of this subtle but persistent pattern of discrimination when Whites are provided with ostensibly non race-related bases for their decisions. However, consistent with the aversive racism framework, when testimony is included that suggests that criminal allegations against a defendant may be racially motivated (thus triggering bias concerns), Whites no longer racially discriminate (Sommers & Ellsworth, 2000).

Overall, these studies show that, in contrast to the dramatic decline in overt expressions of prejudice, subtle forms of discrimination continue to exist, apparently largely unabated. As we noted earlier, one reason for the persistence of these types of biases is that they are largely shaped by non-conscious processes that can result in behaviors that are often unintended and difficult to control. Because of the subtle and varied nature of these biases, contemporary prejudice not only systematically influences decision making but can also fundamentally impact everyday social relations in ways that contribute substantially to misunderstandings and mistrust in intergroup relations.

Bias and Interracial Interaction

An important implication of the aversive racism framework is that the dissociation between the positive conscious (explicit) attitudes and negative unconscious (implicit) attitudes of aversive racists fundamentally influences the ways Whites interact with Blacks. Considerable past research has shown that implicit and explicit attitudes influence behavior in different ways and under different conditions (see Dovidio, Kawakami, & Gaertner, 2002b; Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997b; Dovidio, Kawakami, Smoak, & Gaertner, 2009; Fazio, Jackson, Dunton, & Williams, 1995; Fazio & Olson, 2003; Wilson, Lindsey, & Schooler, 2000). Whereas explicit attitudes typically shape deliberative, well-considered responses for which people have the motivation and opportunity to weigh the costs and benefits of various courses of action, implicit attitudes typically influence responses that are more difficult to monitor or control (e.g., some nonverbal behaviors; see Chen & Bargh, 1997; McConnell & Leibold, 2001) or responses that people do not view as diagnostic of their attitude and thus do not try to control.

For instance, Dovidio et al. (1997b) found that Whites’ negative implicit attitudes predict nonverbal cues of discomfort (increased rate of blinking) and aversion (decreased eye contact) toward Blacks (see also Word, Zanna, & Cooper, 1974), whereas Whites self-reported, explicit attitudes predict open evaluations and liking of Blacks. Aversive racists, who have favorable conscious views of Blacks but also harbor unconscious negative attitudes or associations (see Karpinski & Hilton, 2001), are, thus, likely to convey mixed-messages in interracial interactions.

Biases in interpersonal relations. One fundamental implication of these processes is that Whites and Blacks are likely to form very different perceptions of race relations, with Blacks developing a general sense of distrust of Whites (Dovidio, Gaertner, Kawakami, & Hodson, 2002a). In particular, Whites’ perceptions about how they are behaving or how they are perceived by others are based more on their explicit attitudes and overt behaviors, such as the verbal content of their interaction with Blacks, and less on their implicit attitudes or less deliberative behaviors. In contrast, the perspective of Black interaction partners in these interracial interactions allows them to attend to both the spontaneous (e.g., nonverbal) and deliberative (e.g., verbal) behaviors of Whites. To the extent that the Black partners attend to Whites’ nonverbal behaviors, which may signal more negativity than their verbal behaviors, Blacks are likely to form more negative impressions of their partners and be less satisfied with the interaction compared with Whites.

Research on egocentric biases in social perception suggests a basic social psychological mechanism for the formation and maintenance of the differing perspectives of Blacks and Whites in social interactions. Because people often have greater access to their own internal mental states (e.g., motivations, intentions) than the mental states of others, they often utilize and weigh introspective information more heavily when making self-judgments than when making judgments of others (‘introspection illusion’; see Pronin, 2009). In part because of the prominence of one's own mental states, within social interactions actors often fail to recognize that their internal states are not readily visible to their partners, who instead base their interpersonal judgments more on the behaviors of the individuals they are interacting with. Consistent with this reasoning, Dovidio, Kawakami, and Gaertner (2002b) found direct evidence of Blacks’ and Whites’ divergent perspectives in intergroup interactions based on access to different sources of information (e.g., verbal versus nonverbal behavior). White participants, whose implicit and explicit racial attitudes were previously assessed, engaged in separate videotaped conversations with a White and a Black partner that were race-neutral in content. Supportive of hypotheses derived from the aversive racism framework, Blacks’ perceptions of their White partners’ friendliness were predicted by assessments of Whites’ nonverbal (but not verbal) behavior and their partner's implicit (but not explicit) racial attitudes. In contrast, consistent with an introspective bias, White participants’ judgments of their own friendliness were associated with their explicit (but not implicit) attitudes toward Blacks and were predicted by independent raters’ assessments of their verbal friendliness. Thus, because of their very different perspectives and reliance on different information (verbal versus nonverbal behavior), Whites and Blacks left the same interaction with very different impressions.

Other studies further suggest that the nonverbal behaviors emitted by the White participants in interpersonal contexts, which may be exacerbated by concerns about appearing racist (Goff, Steele, & Davies, 2008), may primarily reflect underlying negative affective responses that are often automatic and difficult to control (Amodio, Harmon-Jones, & Devine, 2003; Dovidio et al., 1997). To the extent that Blacks attribute the subtle nonverbal avoidant behaviors of Whites to explicit, rather than implicit, prejudice, they may misperceive these behaviors as intentional. Results of recent studies suggest that such an attribution may influence Blacks’ affective experiences in a way that produces a self-fulfilling prophecy. Shelton, Richeson, and Salvatore (2005) found that the more ethnic minorities expected to be the target of prejudice, the more negative affect they experienced during the interracial interaction, even when engaging in more positive compensatory behaviors to avoid rejection by their partner. These negative perceptions and experiences can fuel tensions in social interactions and lessen Blacks’ and Whites’ interests in initiating and sustaining cross-group contact (Pearson et al., 2008). In addition, perceptions of bias in one's partner can reinforce the common belief that members of other racial and ethnic groups are generally less interested in engaging intergroup contact than are members of one's own racial or ethnic group (Shelton & Richeson, 2005). Together, these findings suggest that the subtle and complex nature of contemporary prejudice can shape the everyday perceptions of White and Black Americans in ways that interfere with interpersonal trust and communication that are critical to establishing positive and effective cross-group interactions.

Biases and team performance. Besides shaping different impressions and perceptions, contemporary biases can also influence interpersonal relations in ways that unintentionally but adversely affect the performance of Whites and racial and ethnic minorities working in teams. To the extent explicit bias may impact perceptions of friendliness and support, overt biases may be expected to directly impact group productivity. To the extent that implicit racial attitudes may be detected through more subtle behaviors, these unconscious biases may erode trust between group members and indirectly hinder team performance.

To test this reasoning, Dovidio (2001) examined the interpersonal impressions and performance of Blacks who interacted with non-prejudiced Whites (i.e., those low in both explicit and implicit prejudice), prejudiced Whites (i.e., those high in both explicit and implicit prejudice), and White aversive racists (low in explicit prejudice but high in implicit prejudice) on a joint problem-solving task. The results for perceptions of friendliness were comparable to those of Dovidio et al. (2002b). Whites who scored low in explicit prejudice (i.e., non-prejudiced Whites and aversive racists) reported that they behaved more friendly than those who scored high in prejudice (overtly prejudiced Whites). However, Black participants showed sensitivity to both their partners’ explicit and implicit attitudes: They perceived Whites who were unbiased on the implicit measure (non-prejudiced Whites) to be more friendly than those who showed implicit biases (aversive racists and prejudiced Whites). They were also less trustful of aversive racists and overtly prejudiced Whites than of non-prejudiced Whites. Results of the groups’ efficiency in problem-solving showed a similar pattern. Interracial teams consisting of a Black participant and a non-prejudiced White participant performed the best, interracial teams with an overtly prejudiced White participant were the next most efficient group, and those with a White aversive racist performed the worst. Presumably, the conflicting messages displayed by aversive racists (Dovidio et al., 2002) and the divergent impressions of the team members’ interaction reduced the overall effectiveness of the team.

These results suggest that, to the extent Blacks are a racial minority in an organization and are often dependent on high-prejudiced Whites or aversive racists for work-related tasks, their performance is likely to be objectively poorer than the performance of the majority of their White counterparts who primarily work with other Whites. Thus, within the workplace, even unconscious and unintentional biases can have consequences that may be detrimental to the job performance and, ultimately, the well-being of racial and ethnic minorities, and the success of professional organizations more broadly. Indeed, in a recent national survey of 1,700 corporate managers and professionals, subtle bias among coworkers was cited as a leading reason for job change and voluntary layoff, particularly among people of color, affecting an estimated 2 million workers in the United States annually at an estimated cost of $64 billion in wages (Corporate Leavers Survey, 2007). Among those who reported experiencing bias at work, nearly one third (27%) indicated that their experience strongly discouraged them from recommending their employer to prospective employees, and 13% reported that their experience discouraged them from recommending their employer's products or services to potential clients. The report concluded that, ‘overt and illegal discrimination is no longer the largest threat to recruiting and retaining the “best and the brightest.” Unfairness, in the form of every-day inappropriate behaviors ... is a very real, prevalent and damaging part of the work environment’ (p. 2, Executive Summary).

Combating Aversive Racism

Prejudice-reduction techniques have traditionally been concerned with changing conscious attitudes (overt racism) and obvious expressions of bias, and have commonly utilized educational programs and campaigns aimed at combating such views and behaviors (Stephan & Stephan, 2001). However, because of its pervasiveness, subtlety, and complexity, conventional interventions and legal practices for eliminating racial bias are often ineffective for combating aversive racism. Aversive racists already recognize prejudice as harmful, but they do not recognize that they are prejudiced. Other techniques are thus required.

Redirecting in-group bias. One basic argument we have made in our analysis of contemporary social biases is that the negative feelings that typically develop toward other groups are often rooted in basic socio-cognitive processes. One such process is the categorization of people into in-groups and out-groups. Because categorization is a basic process fundamental to intergroup bias, this process has been targeted in efforts to combat the negative effects of aversive racism.

The Common Ingroup Identity Model (Gaertner & Dovidio, 2000) is one such intervention approach that harnesses social categorization as a means to reduce intergroup bias and has received strong empirical support in interventions with both child and adult populations (Gaertner et al., 2008). Specifically, if members of different groups are induced to think of themselves as a single superordinate in-group rather than as two separate groups, attitudes toward former out-group members will become more positive by reaping the benefits of in-group status. Enhancing the salience of a common ingroup identity has been shown to inhibit the activation of both implicit (Van Bavel & Cunningham, 2009 and explicit (Gaertner & Dovidio, 2000) biases. Thus, by changing the basis of categorization from race to an alternative, inclusive dimension, one can alter who ‘we’ is and who ‘they’ are, undermining a potent contributing force to contemporary racism. The formation of a common identity, however, need not require groups to forsake their subgroup identities. It is possible for members to conceive of themselves as holding a ‘dual identity’ in which other identities and a superordinate identity are salient simultaneously (Crisp, Stone, & Hall, 2006; Gaertner, Mann, Murrell & Dovidio, 1989).

Acknowledging and addressing unconscious bias. As described earlier, aversive racism is characterized by conscious (explicit) egalitarian attitudes and negative unconscious (implicit) attitudes and beliefs. Simply because implicit attitudes can be unconscious and automatically activated, however, does not mean that they are inevitable or immutable to change. To the extent that implicit attitudes and stereotypes are learned through socialization (Karpinski & Hilton, 2001; but see Fazio & Olson, 2003), they can also be unlearned or inhibited by well-learned countervailing influences. For example, extended practice in associating counter-stereotypic characteristics with racial and ethnic minority groups has been shown to inhibit the automatic activation of cultural stereotypes (Kawakami, Dovidio, Moll, Hermsen, & Russin, 2000). Implicit motivations to control prejudice can similarly inhibit the activation of spontaneous racial biases even when cognitive resources are depleted (Park, Glaser, & Knowles, 2008).

The problem, in practice, is that Whites are typically motivated to avoid seeing themselves as racially biased and often adopt a colorblind strategy when engaging in interracial interactions, particularly when they anticipate racial tension. However, efforts to be colorblind can sometimes produce ‘rebound effects,’ causing biases to become activated even more. Indeed, Uhlmann and Cohen (2005) found that participants who were more confident in the objectivity of their judgments were also more likely to discriminate against equally qualified female candidates for a stereotypically male job (chief of police), inflating criteria that favored male over female candidates. Ironically, the act of affirming a non-prejudiced self-image can further increase the likelihood that even ostensibly non-prejudiced individuals will discriminate. Monin and Miller (2001) found that, when given the opportunity to disagree with a prejudicial statement (and, thus, affirm a non-prejudiced self-image), individuals were more likely to discriminate against women or a racial minority group when making a subsequent hiring decision. The authors reasoned that the opportunity to reinforce one's egalitarian image (even when done privately) gave participants a ‘license’ to act in a discriminatory manner (see also Effron, 2009).

Whites’ attempts to be colorblind can also alienate minority group members, who generally seek acknowledgement of their racial identity, and further contribute to interracial distrust. Consistent with this reasoning, Apfelbaum, Sommers, and Norton (2008) found that although avoidance of race was seen as a favorable strategy by Whites for promoting more positive interracial interactions, in practice, failure to acknowledge race actually predicted decrements in Whites’ nonverbal friendliness and resulted in greater perceptions of racial prejudice by Black interaction partners. Clearly, Whites’ intuitions about processes that enhance or attenuate racial bias may not always be supported empirically.

Despite these challenges, it may nevertheless be possible to capitalize on aversive racists’ good intentions and induce self-motivated efforts to reduce the impact of unconscious biases by making them aware of these biases. Work by Monteith and Voils (1998) indicates that when low-prejudiced people recognize discrepancies between their behavior (i.e., what they would do) and their personal standards (i.e., what they should do) toward minorities, they feel guilt and compunction, which subsequently produces motivations to respond without prejudice in the future. With practice, these individuals learn to reduce prejudicial responses and respond in ways that are consistent with their non-prejudiced personal standards. When extended over time, this process of self-regulation can produce sustained changes in even automatic negative responses. Dovidio, Kawakami, and Gaertner (2000) found that greater discrepancies between what one would do and should do produced higher levels of guilt among Whites in an initial experimental session, and this relationship occurred primarily for low-prejudiced participants. These findings indicate the potential recruitment of self-regulatory processes for low- but not high-prejudiced participants. When participants returned three weeks later, there was generally greater alignment (i.e., smaller discrepancy) between what one would and should do – an indication that both high- and low-prejudiced participants showed a decrease in overt expressions of bias. However, low- and high-prejudiced Whites differed in terms of the extent to which they internalized these changes. Low-prejudiced Whites who had larger initial discrepancies showed greater reductions in implicit stereotyping; in contrast, for high-prejudiced Whites, the relationship was weaker and nonsignificant. These findings demonstrate that the good intentions of aversive racists can be harnessed to promote self-initiated change in even unconscious biases with sufficient awareness, effort, and practice.

Son Hing, Li, and Zanna (2002) extended work along these lines by examining responses of people identified as non-prejudiced (low in both explicit and implicit prejudice) and aversive racists (low in explicit prejudice but high in implicit prejudice) to self-awareness of one's own hypocrisy. In a study conducted in Canada with Asians as the target minority group, participants were assigned to either a hypocrisy condition, in which they reflected on situations in which they had reacted negatively or unfairly toward an Asian person, or to a control condition in which they were not asked to write about such situations. The researchers predicted that making people aware of violations of their egalitarian principles would arouse guilt among aversive racists (who harbor negative feelings toward Asians) and thus produce compensatory behavior when recommending funding for Asian student groups among aversive racists but not among non-prejudiced participants. The results supported the predictions. Aversive racists in the hypocrisy condition experienced uniquely high levels of guilt and displayed the most generous funding recommendations for the Asian Students’ Association. The funding recommendations of truly low-prejudiced participants, however, were not affected by the hypocrisy manipulation. Son Hing et al. (2002) concluded that making people aware of their biases is particularly effective at reducing bias among people who explicitly endorse egalitarian principles while also possessing implicit biases – the factors that characterize aversive racists.

Additional support for this conclusion was obtained from a study that examined physicians’ responses to descriptions of patients showing signs of coronary artery disease. Green et al. (2007) found that physicians higher in implicit (but not explicit) racial bias were less likely to recommend aggressive treatment (thrombolysis) for Black relative to White patients. However, physicians who were aware that their recommendations could be influenced by non-conscious racial biases did not show the same relationship between implicit bias and treatment recommendations: Among those who were aware of this potential influence at the outset (approximately 24% of respondents), those with stronger implicit pro-white bias were, in fact, more likely to recommend Black patients for thrombolytic treatment than those with weaker implicit biases. Thus, these individuals were able to consciously ‘correct’ for their implicit biases when making clinical recommendations.

Controlling implicit bias through non-conscious goals. Although bias control has traditionally been conceptualized in terms of conscious efforts to inhibit negative attitudes and stereotypes that become activated in one's mind, recent studies suggest that biases may also be combated at the implicit level through non-conscious processes that inhibit their activation in the first place. In particular, research on non-conscious self-regulation (Bargh, 1990) suggests that goals, such as efforts to be egalitarian, need not be consciously pursued in order to exert influence over one's thoughts and behavior. In a series of studies, Moskowitz, Salomon, and Taylor (2000) found that individuals with chronic (explicit) race-related egalitarian goals, but not those with non-chronic egalitarian goals, were faster to respond to egalitarian-relevant words when primed with African-American compared to White faces. In addition, whereas individuals without chronic egalitarian goals responded more quickly to stereotype-relevant (versus stereotype irrelevant) words following African-American faces, individuals with chronic egalitarian goals showed no such evidence of stereotype activation in response to African-American faces (see also Lepore & Brown, 1997; Kawakami, Dion, & Dovidio, 1998).

Building on these findings, Moskowitz and colleagues (see Moskowitz & Ignarri, forthcoming, for a review) have found that interventions that enhance motivations to be egalitarian (e.g., having participants describe a personal incident in which they failed to be egalitarian towards African Americans) can not only attenuate, but actively inhibit non-conscious stereotyping. In their research, Whites who were primed in this way were slower to respond to stereotype-relevant, compared to stereotype irrelevant, words after being primed with African-American faces, indicating stereotype inhibition rather than simple non-activation. Importantly, these effects occurred at speeds too fast to implicate conscious control and occurred despite participants being unaware of the relevance of the cognitive reaction time task to stereotyping, implicating an implicit form of bias control operating outside of conscious awareness. Together, these findings suggest that one need not chronically or even consciously pursue egalitarian goals to inhibit the activation of stereotypic thoughts; even temporarily activating egalitarian goals was sufficient to reduce implicit stereotyping.

Current research on non-conscious goal pursuit is exploring social and cognitive extensions of these effects. For example, work by Aarts, Gollwitzer, and Hassin (2004) on goal contagion reveals that goals such as motivations to not be prejudiced may become automatically activated simply in the presence of egalitarian-minded others, suggesting the importance of observing others’ egalitarian behavior and egalitarian social norms more generally for controlling automatic biases. Other research reveals that goals need not even relate to egalitarianism in or to be effective in combating implicit bias. Sassenberg and Moskowitz (2005), for example, revealed that priming creativity (a goal that conflicts with the energy-saving and simplifying features of stereotyping) similarly reduced stereotype activation, suggesting that any goal that is incompatible with stereotyping (in this case, the goal to form atypical associations) may contribute to successful bias control. Finally, Spencer, Fein, Wolfe, Fong, and Dunn (1998) identified important boundary conditions for these effects. Whereas some goals may inhibit implicit biases, other, seemingly unrelated, goals (such as the goal to maintain one's self-esteem) can actually enhance the activation of stereotypes (Spencer et al., 1998). Thus, understanding both the nature and the personal and social functions of non-conscious goal pursuits has important implications for understanding the conditions under which efforts to control biases are likely to succeed or fail.

In summary, whereas conscious efforts to avoid stereotyping may often fail or even exacerbate bias because individuals lack insight into the processes that promote and regulate it, passive implicit goals to not stereotype may succeed by co-opting the very psychological mechanisms that sustain it, replacing stereotypic associations with egalitarian or atypical associations when perceiving or interacting with members of other racial and ethnic groups.

Reducing implicit biases through intergroup contact. Recent studies have also begun to explore the effects of intergroup contact on implicit forms of bias. For example, Tam, Hewstone, Harwood, Voci, and Kenworthy (2006) and Turner, Hewstone, and Voci (2007) assessed the effects of contact on implicit and explicit attitudes toward elderly persons and Whites’ attitudes toward South Asians, respectively. In general, consistent with the idea that implicit attitudes reflect conditioned associations (Fazio & Olson, 2003) that may be dissociated from explicit attitudes, measures of the overall amount of intergroup contact (e.g., proportion of neighbors who are out-group members) were generally found to be better predictors of lower implicit prejudice than were measures of the quality of contact (e.g., self-disclosure and emotional closeness), which better predicted explicit attitudes. However, Aberson and Haag (2007) found that among those who experienced little intergroup contact, the quality of contact did predict their level of implicit bias. Interestingly, in all of these studies, the effects of contact on implicit attitudes were not mediated by factors that typically mediate explicit attitudes (e.g., reduced intergroup anxiety, greater perspective taking), but, rather, showed a direct, positive impact on implicit attitudes, suggesting the potential value of simple exposure to out-groups for reducing unconscious biases (for a similar argument, see Karpinski & Hilton, 2001). More generally, these findings suggest that, as hypothesized within the aversive racism framework, implicit and explicit attitudes are qualitatively distinct, as opposed to merely reflecting different components of the same attitude, and need to be considered in tandem when developing interventions aimed at combating contemporary prejudice.

Future Directions in Aversive Racism Research

Although research on aversive racism has seen considerable progress over the past three decades as new methodological tools for assessing unconscious bias (e.g., response latency measures, physiological measures) have become increasingly available, a number of important questions remain. For example, how do aversive biases develop within individuals over time? Are there specific biological, social, and environmental factors that may influence individuals’ susceptibility to acquiring implicit biases? Understanding the biological (e.g., physiological, neural, endocrine, and genetic) underpinnings of implicit bias may provide a clue. For instance, some individuals are genetically predisposed to show heightened sensitivity to interpersonal threat and social rejection (for reviews, see Fox, Hane, & Pine, 2007; McEwan, 2007), and it remains to be seen whether genetic factors may also influence the acquisition and extinction of automatic social biases (see Olsson, Ebert, Banaji, & Phelps, 2005). There is some behavioral evidence to suggest this may be the case. The results of a recent study by Livingston and Drwecki (2007) suggest that the differentiation of implicitly biased from ‘truly’ non-prejudiced individuals may be rooted, in part, in the relative ease with which the former group is capable of acquiring automatic negative associations in general.

Similarly, understanding the cognitive- and social-developmental trajectory of implicit bias over the lifespan may offer new insights into the nature and dynamics of contemporary racism. New evidence from studies of implicit and explicit attitudes in children reveals an onset time of late childhood for the emergence of aversive biases, strengthening into early adulthood. In an initial study examining White children's and adults’ attitudes toward Blacks, Baron and Banaji (2006) found that children as young as 6 years of age showed evidence of strong implicit pro-White evaluations (based on race IAT scores) and that implicit and explicit attitudes about race diverge by age 10, with implicit attitudes remaining stable but self-reported attitudes growing considerably more favorable over time. These findings have been replicated in both US and Japanese samples (Dunham, Baron, & Banaji, 2006), suggesting surprising consistency and durability of these processes across cultures and contact experiences.

Additional evidence indicates that children become strategic in their race-related responses around age 10, exerting conscious effort to appear non-prejudiced. Specifically, Apfelbaum, Pauker, Ambady, Sommers, and Norton (2008) found that, as expected, 10- and 11-year-olds generally outperformed 8- and 9-year-olds on a task assessing efficiency in categorization of simple colors. However, 10- and 11-year-olds performed more poorly than the younger children on a similar task that required racial categorization. The authors reasoned that older children's efforts to follow norms to appear non-prejudiced prompted them to forgo the use of racial category information and adopt a colorblind strategy, even at the expense of task performance. This rationale was further supported by the finding that the children's reported awareness of racial stereotypes was negatively associated with their performance on the racial categorization task.

Further support for this developmental time-course was obtained by McGillicuddy-De Lisi, Daly, and Neal (2006), who found evidence of aversive racism in the way that 10-year-olds distributed resources to Black and White story characters. Whereas younger children (7- to 8-year-olds) distributed money equivalently, regardless of race, age, performance, or need, older children (9- to 10-year-olds) showed evidence of differential allocation by race, consistent with aversive racism. Older children allocated less money to needy Black than to needy White children, but allocated more money to Black than to White children who performed unusually well on the designated task. Furthermore, whereas the younger children justified their allocations based on principles of equality for both Black and White targets, the older children made different justifications for the allocations for White versus Black targets, appealing to principles of equality for the former and equity for the latter.

Together, these findings implicate a window of mid-to-late childhood for the emergence of cognitive (e.g., shifting standards, Hodson et al., 2002) and behavioral strategies (e.g., strategic colorblindness in social interactions, Apfelbaum, Sommers, & Norton, 2008) that aversive racists may employ to establish and continually reinforce a non-prejudiced image while engaging in discriminatory practices.

Other research questions relate to the qualitative nature of contemporary prejudice and the potential roles that different types of implicit and explicit biases might play in helping to establish more nuanced understandings of prejudice. Research by Son Hing et al. (2008), for example, demonstrated how people with various ‘profiles’ of implicit and explicit prejudice (e.g., self-identified racists versus principled conservatives) respond differently in different race-related situations. Future research might also consider the potentially distinct roles of pro-in-group and anti-out-group implicit attitudes as predictors of different affective, cognitive, and behavioral responses to in-group and out-group members. For instance, aversive racists may show either positive or neutral implicit attitudes toward the in-group coupled with neutral or negative implicit attitudes toward the out-group, consistent with exposure to either more positive, neutral, or more negative exemplars of these groups over time. Similarly, ‘truly’ low/non-prejudiced responses (low implicit and low explicit bias) may reflect either apathy or a convergence in attitudes toward in-group and out-group members, with potentially differing implications for social relations and behavior. Future research might thus seek to incorporate implicit measures that assess in-group and out-group attitudes independently, such as the Go/No Go Association Task (Nosek & Banaji, 2001), Single Category IATs (Karpinski & Steinman, 2006), or priming procedures (see Cunningham, Preacher, & Banaji, 2001), rather than rely on the relative group comparisons of traditional IATs, to capture their potentially distinct contributions to intergroup behavior.


The influence of aversive racism is pervasive, and it persists because it remains largely unrecognized and thus unaddressed. In essence, the challenge of aversive racism is that it represents a fundamental discrepancy between mind and action. In mind, aversive racists truly believe that they are non-prejudiced, but in action they discriminate in subtle but consequential ways. Without sufficient recognition of the subtle nature of contemporary biases and without the appropriate tools for combating these particular biases, significant progress toward a truly just society will be difficult to achieve. Good intentions alone are not good enough.

Nevertheless, aversive racism can be combated with new approaches and strategies that are uniquely targeted at unconscious racial prejudice. For example, because aversive racists are truly motivated to be non-prejudiced, making them aware of their unconscious biases (in a nonthreatening way) can arouse powerful motivations for change. Increasing sensitivity to the discrepancy between their genuine commitment to egalitarian principles and the nature of their biased actions produces self-regulatory responses that can help aversive racists control their bias in the short run and, with practice and effort over time, reduce their unconscious biases in the long run. By creating truly egalitarian habits of mind, good intentions will more directly translate into consistent, socially responsible, and just action.


Preparation of this article was supported by NSF Grant # BCS-0613218 awarded to the second and third authors.

Short Biographies

Adam R. Pearson is a doctoral student in social psychology at Yale University. He received a BS in Biology from Cornell University and an MA in Psychology from the University of Connecticut. His research has appeared in Psychological Science, Social Cognition, Personality and Social Psychology Bulletin, Group Processes and Intergroup Relations, Social Justice Research, and in edited volumes including Advances in Group Processes, The Psychology of Hate, Explaining the Breakdown of Ethnic Relations, and The Social Psychology of Intergroup Reconciliation. His research explores how conscious and non-conscious biases impact everyday perceptions, judgments, communication, and behavior.

John F. Dovidio's research interests are in stereotyping, prejudice, and discrimination; social power and nonverbal communication; and altruism and helping. Much of his scholarship has focused on ‘aversive racism,’ a subtle form of contemporary racism. In addition to his articles and chapters, he is co-author of several books, including Emergency Intervention; The Psychology of Helping and Altruism; The Social Psychology of Prosocial Behavior; and Reducing Intergroup Bias: The Common Ingroup Identity Model; as well as co-editor of Prejudice, Discrimination, and Racism; Power, Dominance, and Nonverbal Behavior; On the Nature of Prejudice: 50 Years After Allport; and, Intergroup Misunderstandings: Impact of Divergent Social Realities. He has been president of the Society for Personality and Social Psychology and of the Society for the Psychological Study of Social Issues, as well as chair of the Executive Committee of the Society for Experimental Social Psychology. Dr. Dovidio has been Editor of the Journal of Personality and Social Psychology – Interpersonal Relations and Group Processes, Editor of Personality and Social Psychology Bulletin, and Associate Editor of Group Processes and Intergroup Relations. He is currently Co-Editor of Social Issues and Policy Review. He is currently Professor of Psychology at Yale University; he previously taught at Colgate University and at the University of Connecticut. Dr. Dovidio holds a BA from Dartmouth College and an MA and PhD from the University of Delaware.

Samuel L. Gaertner (BA, 1964, Brooklyn College; PhD, 1970, City University of New York: Graduate Center) is Professor of Psychology at the University of Delaware. His research interests involve intergroup relations with a focus on understanding and reducing prejudice, discrimination and racism. He has served on the editorial boards of the Journal of Personality and Social Psychology, Personality and Social Psychology Bulletin, and Group Processes and Intergroup Relations. Professor Gaertner's research, together with John F. Dovidio, has been supported by grants from the Office of Naval Research, the National Institutes of Mental Health and currently, the National Science Foundation. Together with John Dovidio, he shared the Gordon Allport Intergroup Relations Prize in 1985 and 1998, as well as the Kurt Lewin Memorial Award (a career award) from the Society for the Psychological Study of Social Issues, Division 9 of the American Psychological Association.


Correspondence address: Department of Psychology, Yale University, 2 Hillhouse Ave., New Haven, CT 06520, USA. Email: or