PRRAC Poverty & Race Research Action Council
Home About PRRAC Current Projects Publications Newsletters Resources Contact Us Support PRRAC Join Our Email List

"Implicit Bias, Racial Inequality, and Our Multivariate World,"

by Andrew Grant-Thomas September/October 2011 issue of Poverty & Race

Richard Banks and Richard Thompson Ford make a number of potentially important arguments. I focus here on two: first, their assertion that the Implicit Association Test may measure conscious-but-concealed bias rather than implicit bias; and, second, their claim that attention to unconscious or implicit bias deflects attention from “substantive inequalities” and the policies needed to remedy them. Like Banks and Ford, I refer here almost exclusively to IAT-based work, but note that evidence for the prevalence and impact of implicit bias extends well beyond results garnered through use of the IAT and well beyond the domain of racial attitudes.

What Does the IAT Measure?

In addition to the possibility that the IAT taps concealed-but-conscious bias, some research psychologists have argued that the IAT may tap other kinds of mental content as well, including the subject’s awareness of biases in the culture, anxiety about being labeled a racist, and sympathy with, or guilt regarding, disadvantaged populations. Some critics also protest the inference, drawn largely from IAT test results reported at the Project Implicit demonstration site, that most Americans harbor “racist” attitudes against Black people. Both criticisms underline the need for greater clarity about the meaning of implicit, and wider appreciation of the contingency of our racial attitudes and related behaviors. I take these points in turn.

On the one hand, Banks and Ford are doubtless right to note that some testers will deliberately misreport their explicit attitudes. On the other, they are wrong to believe that that fact poses a problem for the IAT. The main purpose of the IAT, after all, is to probe attitudes people may be unable or unwilling to report. Myriad studies offer strong support for the notion that implicit attitudes, as gauged by the IAT, and explicit attitudes, as inventoried through self-reports, are related but distinct. Self-reported attitudes, and those probed by the IAT, have been found to be associated with different kinds of brain activity.

Leaving aside the details of the highly technical, largely methodological grounds on which researchers in the cognitive sciences wage their wars of interpretation, one would have to be awfully cynical to suppose that most people who express surprise at their IAT results, including the lead researchers behind Project Implicit, are simply being disingenuous. We have very little reason to believe this. The likelier explanation is that self-reports reflect attitudes of which subjects are aware, IAT results reflect attitudes of which they are not, and sometimes there is a dismaying difference between the two.

In any case, for those of us concerned with the role that implicit racial biases may play in the world, their critical feature is not that they operate outside our awareness, but that they operate automatically—without need for intentionality or reflection. Someone taking the IAT, knowing it is meant to reveal “hidden bias,” may well try to manage the expression of those biases. (Whether they are able to do so is another matter.) The same person looking to hire a new employee, sit next to one person or another on the bus ride home, decide whether to call 911 about a late-night scuffle outside her home is apt to be less vigilant. In these cases, automatic biases may well influence her actions and help trigger the consequences that flow from them.

What about the broader criticism, that IAT results may reflect mental factors other than personal bias? John Jost, Laurie Rudman and their co-authors offer a compelling response in their 2009 review: “If IAT scores were [sic] measured nothing more than familiarity or sympathy (or any of the other artifacts proposed by critics), then there is no way that such scores would predict discriminatory attitudes and behaviors in the manner and to the extent that they do.” Here we get to a question arguably more fundamental than the one about precisely what the IAT assesses: Does it provide information that reliably helps us anticipate behaviors we care about? Yes, it does.

A Nation of Racists?

Researchers have accumulated significant evidence that implicit bias, as measured on the IAT or in other ways, correlates with discriminatory behavior. Employment recruiters with large implicit biases in favor of native Swedes were much more likely to invite applicants with male Swedish names for interviews than they were to invite equally qualified applicants with male Arab names. White students with high implicit bias scores were more likely to report having directed verbal abuse or physical violence against racial others. Many studies have shown that police officers and civilians alike are more likely to shoot unarmed Blacks than unarmed Whites, and to shoot armed Blacks but not armed Whites in video simulations.

More broadly, Anthony Greenwald and his collaborators found in their meta-analysis of relevant research studies that IAT results did much better than self-reported attitudes in predicting Black-White and other intergroup behaviors—including hiring and salary decisions, sentencing decisions and intention to vote for John McCain in 2008. The reverse was true in the seven other behavioral domains examined. The researchers also found that IAT and self-report measures offered the best behavioral predictions when used in tandem than either did when used alone.

While the power of the IAT to predict interracial behavior has often been impressive by the standards of behavioral science, its predictive capacity nonetheless must be considered modest by real-world standards. As a rule, cultural information, social setting, recent experience, explicit attitudes and other factors together influence individual behaviors much more than implicit attitudes alone do. And, again, having implicit bias is not the same as embracing that bias, and people can be differently alert about whether, when and how they express their biases. As a result, “low-bias” people will act in discriminatory ways sometimes and “high-bias” people will often refrain from doing so.

Where does this leave us? On the one hand, according to the Project Implicit web site, “75-80% of self-identified Whites and Asians show an implicit preference for racial White relative to Black” and a large and growing body of empirical work indicates that such preferences help predict many race-related behaviors and judgments, doing so better than self-reported data on racial attitudes. On the other hand, implicit biases usually account for modest amounts of the variation in such behaviors and, as Jost, Rudman and their colleagues note, implicit bias researchers warn repeatedly against using the IAT to diagnose individual prejudice.

This is shades-of-gray stuff, and as such very much in tension with the American inclination to reduce matters of race to stark, either-or binaries. Thus, in the United States, a person is either Black or not-Black. The degree to which many of us are invested in the distinction, in particular, is evident in the back-and-forth about Barack Obama’s racial identity. Either George W. Bush’s leaden response to Hurricane Katrina betrayed his racism or his diverse cabinet showed that he was not-racist. Either Obama’s election confirmed what the Wall Street Journal called the “myth of racism” or it is completely anomalous. When it comes to race, we are often blind to shades of gray.

Racial Bias and Inequality

Suppose we suspected that many people in the United States, especially members of its White-identified majority, harbored readily activated biases, implicit or explicit, against people of color and especially against Black Americans. (In 2009, public opinion scholars Lawrence Bobo and Camille Charles concluded that “between half and three-quarters of whites in the United States still express some degree of negative stereotyping of blacks and Latinos.”) Suppose we knew that these biases sometimes manifest in discriminatory behaviors. Suppose we recognized the substantial role that human discretion plays in the distribution of societal benefits, burdens and resources in such opportunity arenas as housing, education, employment and criminal justice. And what if we also recognized that the power to distribute benefits and burdens was vested overwhelmingly in the hands of White Americans?

Under this set of assumptions, that three in four African Americans are confined to 16% of the nation’s census block groups would not surprise us. The gross overrepresentation of Latinos and African Americans in our country’s prisons would not shock us. It might not even shock us to learn that the only two states that allow prisoners to vote, Maine and Vermont, are also the two “whitest” states in the country. An argument like that proposed by Alberto Alesina and Edward Glaeser, that the United States’ greater racial and ethnic diversity accounts for half the difference between this country’s public welfare spending and Europe’s more generous support for its poor, would seem quite plausible on its face.

The point, of course, is that interpersonal bias has very practical implications for our work on race. Let’s consider the case of racial segregation in some detail. With reference to the pronounced residential segregation of African Americans, I suggest that racial preferences might enforce segregation in the United States today in at least four ways.

Historical sediment. Many people have elaborated on the ways that White racial attitudes, especially in the North, fueled a range of “fight” and “flight” responses directed mainly against African Americans through much of the twentieth century. I propose two ongoing effects. First, history has bequeathed us patterns of segregation in many metro areas of the Northeast and Midwest that would require time to disrupt even if racial attitudes, policies and housing market practices today presented no further obstacles to doing so. Absent strong remedial action, segregation in motion tends to stay in motion. Second, in some areas, historical antagonisms and discriminatory public policies have entrenched entitlements to “racialized space” that residents regard as invariable.

Policy preferences. We know that public support for policies depends substantially on the explicit racial preferences people bring to their considerations. We know much less about how implicit attitudes affect policy choices, though one recent study concluded that the IAT captures automatic attitudes that shape individual preferences for immigration policy.
Current policy struggles with implications for racial segregation abound. In Milwaukee, then-County Executive Scott Walker (same guy) successfully championed a fight against developing public transportation that would have connected mostly-Black city residents to jobs in mostly-White suburban areas. To similar effect, Westchester County’s (NY) Executive so far has defied federal orders to dismantle exclusionary zoning ordinances that have limited the availability of affordable housing throughout much of the county. It is quite likely that racial attitudes drive much of the dynamic in these cases and many comparable ones across the country.

Private actions. Any hope we have to generate much greater neighborhood integration will depend largely on modifying people’s automatic associations about race. A video experiment by Maria Krysan and three collaborators found that Whites in Chicago and Detroit regard all-Black and racially-mixed neighborhoods as much less attractive than literally identical neighborhoods with White residents alone. We see the corresponding dynamics in places like Cincinnati, St. Louis and Philadelphia, where African-American and, in some cases, Latino movement to older suburbs have been echoed by the movement of Whites to the exurbs.

One of the most notable findings of the literature on residential segregation has been the status of African Americans as both the least-favored neighbors and the group most disposed toward integration. Bobo and Charles report that “[a]ctive racial prejudice—negative racial stereotypes, feelings of social distance, and perceptions of racial group competition—is the primary factor driving preferences for neighborhood racial integration, and prejudice is therefore implicated in the persistence of racially segregated communities.” A more recent trend finds more middle-class African Americans wanting to settle in predominantly “Black” neighborhoods, possibly presenting yet another attitudinal barrier to greater integration.

System justification. System justification theory highlights the tendency, shared by advantaged and subordinate groups alike, to legitimate the status quo. The professed beliefs of many Whites that residential segregation is fed mainly by the wish of African Americans and Latinos to “be with their own,” by their reluctance to do the hard work required to succeed, or simply by (legitimate) socioeconomic differences rather than (illegitimate) racial aversions can all be construed as supportive of the theory.

So too, arguably, do results from a 2007 Pew Research Center survey showing majorities of African Americans agreeing that Blacks were mainly “responsible for their own condition” (53%) and that the “values held by middle class blacks and poor blacks have become more different” (61%). Almost 4 in 10 respondents believe that “because of diversity within their community, blacks can no longer be thought of as a single race.” How same-group racial attitudes, inflected by notions of deservingness, condition the wish of Black Americans to live among same-race peers across class lines, and our willingness to remain invested in the broader struggle to upturn the racial status quo—these are issues that would reward additional study.

Our Multivariate World

Of course, the persistence of segregation and most other features of racial inequity cannot be due entirely to the persistence of implicit or explicit biases. For one thing, unless we believe that racial biases have actually worsened over time, even pervasive bias cannot account for the rapidly increasing resort to incarceration over the last three decades, the resegregation of public schools in the South, the recent widening of the racial wealth divide, or the bifurcation of fortunes within the African-American population, among other trends.

For another, we know that personal biases are not required to maintain some inequalities, though pervasive biases surely exacerbate them. An appallingly high number of Black and Latino children attend high-poverty schools with too few qualified teachers, crumbling buildings and classes that prepare them poorly for college. In 2000, one in four Black children and one in eight Latino children (but only one in 100 White children) lived in a severely distressed neighborhood. These kinds of institutional and structural inequalities have terrible, self-reinforcing consequences for the people of color who suffer them, regardless of the play of biases within them.

We are complex creatures living in a multivariate social world largely of our creation. Making substantial progress in remedying racial injustice and inequality will require a multi-pronged, insistently integrated approach that engages issues of bias, culture, ideology, institutional and structural inequities, and power. The lag with respect to progressive policy reform that Banks and Ford lament is about the ascension of color-blindness as a norm in public life; about who controls the policy levers, and who does not; about the cultural models to which those decision-makers and most of their constituents subscribe with respect to racial inequality; and so on. An unhealthy preoccupation that racially progressive people have with anti-discrimination law has very little to do with it.

Implicit Bias and Racial Justice – Next Steps

In terms of research, and even more in terms of vision and strategy, we have much to learn and much to do. With respect to the implicit bias agenda in particular, we have a range of pressing needs.

We need a deeper understanding of the factors that shape the initial development of implicit bias in young people and on a community-wide scale; better processes and tools for demystifying the idea of implicit bias with lay audiences; wider acceptance of the need to engage implicit bias among racial progressives and moderates; better tools for measuring implicit bias and its behavioral effects across contexts; much greater headway in fashioning policy and practical remedies to bias; and more insights into the ways implicit bias shapes our social structures and vice versa. By what mechanisms do our biases shape the institutional structures we create and allow to persist? How do we reconcile that premise with the systems perspective that draws attention to unintended consequences? How do the structures we create impact the way we think about people?

Time to get back to work.

Andrew Grant-Thomas is Deputy Director, Kirwan Institute for the Study of Race and Ethnicity at Ohio State Univ.

Join Our Email List
Search for:             
Join Our Email List