Confronting the Backlash against Implicit Bias
Posted by The Situationist Staff on May 22, 2009
Below you will find some excerpts from an important paper by Situationist Contributor John T. Jost and six distinguished co-authors (Laurie A. Rudman, Irene V. Blair, Dana R. Carney, Nilanjana Dasgupta, Jack Glaser, Curtis D. Hardin). The paper is titled “The Existence of Implicit Bias is Beyond Reasonable Doubt: A Refutation of Ideological and Methodological Objections and Executive Summary of Ten Studies that No Manager Should Ignore.” The paper will be published in Research in Organizational Behavior. Many thanks to Julian Darwal for putting this post together.
* * *
In this chapter, we respond to recent critiques of research on implicit bias, especially studies using the Implicit Association Test (IAT).
Philip Tetlock and Gregory Mitchell argue that measures of implicit bias including the IAT fail to predict organizationally relevant behavioral outcomes. They claim “there is no evidence that the IAT reliably predicts class-wide discrimination on tangible outcomes in any setting,” accuse their colleagues of violating “the injunction to separate factual from value judgments,” adhering blindly to a “statist-interventionist” ideology, and of conducting a witch-hunt against implicit racists, sexists, and others. These and other charges are specious. Far from making “extraordinary claims” that “require extraordinary evidence,” researchers have identified the existence and consequences of implicit bias through well-established methods based upon principles of cognitive psychology that have been developed in nearly a century’s worth of work. We challenge the blanket skepticism and organizational complacency advocated by Tetlock and Mitchell and summarize ten recent studies that no manager (or managerial researcher) should ignore. These studies reveal that students, nurses, doctors, police officers, employment recruiters, and many others exhibit implicit biases with respect to race, ethnicity, nationality, gender, social status, and other distinctions. Furthermore—and contrary to the emphatic assertions of the critics—participants’ implicit associations do predict socially and organizationally significant behaviors, including employment, medical, and voting decisions on the part of working adults.
THIRTY YEARS (OR MORE) OF RESEARCH ON IMPLICIT BIAS: A PRIMER
The IAT is is merely one methodological innovation in a long stretch of empirical contributions documenting the existence of implicit bias. Implicit prejudice results followed naturally from a century’s worth of research on perception, memory, and learning. The evidence fits the contemporary consensus that an enormous amount of cognition occurs automatically, effortlessly, and outside of conscious awareness.
Cognitive Accessibility as a Method of Demonstrating Implicit Bias
In Devine’s experiment, participants evaluated “Donald” as more hostile after they had been subliminally exposed to a relatively large (vs. small) proportion of words related to common stereotypes of African Americans in a previous, ostensibly unrelated task. Not only were participants’ social judgments affected by stereotypes without their awareness once these had been rendered accessible through subliminal exposure, but this effect on social judgment occurred regardless of how participants had voluntarily reported their personal racial attitudes.
Semantic Priming as a Method of Demonstrating Implicit Bias
According to this paradigm, prejudice and stereotypes are empirically captured by the degree to which they are linked through speed and efficiency to semantically related concepts. Exposure to a word related to women (e.g., lady, nurse) hastens the speed with which people identify female pronouns that appear subsequently; similarly, exposure to a word linked to men (e.g., gentleman, doctor) facilitates the identification of male pronouns.
Evaluative Priming as a Method of Demonstrating (and Measuring) Implicit Bias
Fazio et al. (1995)’s evaluative priming tasks exposed participants to photographs of either white or black faces and asked them to categorize subsequent words as either positive or negative, as quickly as possible. The presentation speed of the faces and words was short enough in duration to ensure that their influence on responses would be the result of automatic rather than controlled processes. When white participants classified positively valenced words, their responses were faster when they had been exposed to white (vs. black) faces, but when they classified negatively valenced words, their responses were faster when they had been exposed to black (vs. white) faces.
Individual Differences in the Motivation to Control Prejudice
Fazio also identified an important moderator of the relationship between implicit and explicit biases, namely the “motivation to control prejudice.” Specifically, they developed a questionnaire measure that included items such as, “It’s never acceptable to express one’s prejudices.” Not surprisingly, participants who were higher in motivation to control prejudice scored lower on an explicit measure of prejudice (the Modern Racism Scale). By contrast, the motivation to control prejudice did not predict implicit bias.
Recent Brain Imaging Work
In 2000, researchers provided physiological evidence for the construct validity of implicit attitudes in general and the predictive validity of the IAT in particular when they demonstrated that IAT scores were correlated with the magnitude of amygdala activation when white participants were exposed to photographs of unfamiliar black (vs. white) faces. The amygdala is involved in emotional responses to threat, such as fear. Self-reported racial attitudes failed to predict amygdala activation under the same circumstances.
Literally hundreds of studies in cognitive accessibility, semantic priming, evaluative priming, and other areas in social cognition provide conclusive evidence that mental processes can and do operate nonconsciously and can be measured implicitly. To take Tetlock and Mitchell’s critique seriously, one would need to set aside so much of social and cognitive psychology that both disciplines would be rendered unrecognizable to contemporary students and scholars.
TEN STUDIES NO MANAGER SHOULD IGNORE
The case for implicit bias in no way depends upon any single methodological innovation (such as the IAT), nor is it restricted to associations about race or ethnicity or gender. Rather than duplicate a number of metanalytical studies have found that measures of implicit bias predict relevant behavioral outcomes, the authors focus on a small subset of recent studies with a diversity of methodological strengths that no manager should ignore.
(1) Employment recruiters who favored native Swedes over Arabs on an implicit stereotyping task were significantly less likely to offer (equally qualified) Arab (vs. Swedish) applicants job interview opportunities. Overall, Swedes were 3 times more likely to receive callback interviews.
(2) Despite the fact that participants regarded female (vs. male) managerial applicants who presented themselves as confident, competitive, and ambitious as highly qualified, they also disliked them and were therefore less likely to recommend hiring them. Participants’ implicit gender stereotyping predicted the extent of disliking.
(3) White student participants who scored higher on measures of implicit bias against various racial/ethnic outgroups were significantly more likely to report engaging in verbal slurs, social exclusion, and physical harm against members of minority groups. They also were more likely to recommend budget cuts disproportionately against Jewish, Asian, and black student associations.
(4) In the context of a video simulation program, police officers were significantly more likely to “shoot” an unarmed suspect when he was black (vs. white) on early trials, but they were able to overcome this bias with practice.
(5) Physicians’ degree of implicit (but not explicit) bias predicted racial disparities in simulated treatment recommendations. Specifically, greater bias was associated with (a) decreased likelihood of recommending thrombolysis for black patients suffering from coronary heard disease, and (b) increased likelihood of recommending it for comparable white patients.
(6) Nurses working in a drug and alcohol treatment and rehabilitation facility who scored higher in implicit bias against intravenous drug users experienced more occupational stress, less job satisfaction, and were more likely to express intentions to leave their jobs.
(7) Undecided voters’ implicit candidate preferences, obtained one month prior to the election, significantly predicted their eventual voting decisions.
(8) Hazardous drinkers’ implicit attitudes toward alcohol predicted heightened appetitive responses in the presence of alcohol and self-reported binge drinking.
(9) Convicted male pedophiles were found to exhibit an implicit association between children and sex, whereas no such effect was observed for other sex offenders (who were not pedophiles).
(10) Adolescents’ self-injury implicit associations were correlated with whether or not they had attempted suicide as well as suicidal ideation up to 6 months after their initial assessment, even after adjusting for demographic and psychological risk factors.
THE PAPER ALSO EXPLAINS . . .
- Why, contrary to what Tetlock and Mitchell claim, IAT research does provide reason to suspect Jesse Jackson and Jesse Helms would perform differently on an IAT test.
- Why IAT test results cannot just to be attributed to the “benign causes” of familiarity (people tend to prefer familiar over unfamiliar stimuli); cultural awareness (implicit bias measures may tap “shared cultural stereotypes rather than personal animus”); sympathy for the disadvantaged (the IAT could be driven by sympathy, rather than antipathy, toward outgroup members); fear of being labeled a bigot (rather than actual bigotry); or differences in IAT participants’ cognitive dexterity.
- Why Tetlock’s “organizational accountability” solution to workplace prejudice is inadequate.
* * *
For a sample of related Situationist posts, see “Legal Academic Backlash – Abstract,” “Naïve Cynicism in Election 2008: Dispositionism v. Situationism?,” and “Implicit Bias and Strawmen.”
This entry was posted on May 22, 2009 at 12:01 am and is filed under Abstracts, Choice Myth, Implicit Associations, Social Psychology. Tagged: Implicit Associations, implicit associations test. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.