We’ve discussed the basic process and remedies for implicit bias in other posts, how implicit bias manifests in jury trials. In this post, we take you behind the scenes of academia to explore the conflicting opinions of the researchers. The academic consensus is a complicated minefield that has caused a considerable amount of debate in academia.
The debate centers on the psychometric properties of a computer-administered test called the Implicit Association Test or IAT. The test was developed by Banaji & Greenwald in 1995 with a focus on “unconscious” racial biases. Other iterations of the test assess biases related to gender, sexual orientation, religion, etc. The test measures one’s cognitive association between concepts or feelings and objects or people. Given the pervasiveness of bias, the IAT seems to be valuable in revealing the presence of negative associations with different groups.
However, there has been considerable controversy about the IAT that has spawned an academic brouhaha that has lasted for 20 years without a clear resolution. The controversy pertains to the implications of the IAT: Does it identify biased people? IAT advocates say yes and IAT opponents say no. The other major issue is, Can it predict future behavior? Again, Pro-IAT scholars say yes, while critics of the IAT say no.
The original authors of the test, Banaji & Greenwall created the term, “implicit bias.” They developed the Implicit Association Test to measure the valence of associations that individuals have about others. Certainly, it was groundbreaking research and quickly spread through academia. Banaji & Greenwald were excited about the potential value of the IAT.
Criticism of the IAT springs from concerns about using associations to predict actual biased behavior. The creators of the IAT care concerned about dismissals of test results and what the results mean. They also felt that criticisms allow others to devalue indications of actual biased behavior as just another flawed psychometric endeavor.
This brings us to the perspective of the anti-IAT reserchers. Most are concerned about how employers and lay people might use IAT test results to make decisions about the test taker. Blanton, Jaccard, and colleagues have published responses to Banaji & Greenwald. The strongest criticisms are about the validity of the test. Does the test actually measure bias? Is it any different from what explicit bias tests measure? Additionally, IAT critics cite poor test-retest reliably. That means that a test taker’s score or results might be different each time they take it. Measures like the IAT, they argue, do not provide actionable results. Finally, critics of the IAT are concerned about how test results indicating “bias” could impact the test taker’s self-concept without explanation or understanding.
Fortunately, there is some agreement among academicians. Above all, both IAT advocates and IAT critics agree that the test results should not be used to make important decisions (i.e., employment) about test takers.