Actions Matter: The IAT’s Fallacious Arguments

Mahzarin Banaji and Frank Dobbin, a Harvard psychologist and sociologist, explain in a recent Wall Street Journal article, “Why DEI Training Doesn’t Work—and How to Fix It.”  The article is based on something called the Implicit Association Test (IAT), invented in 1998 and available to all on a Harvard University website. The IAT is a major topic of Banaji’s research.

Diversity, equity, and inclusion (DEI) is a major industry in the U.S., so it makes sense for the business editor of the Journal to cover it. What makes less sense is an article that essentially ignores the scientific problems with the Implicit Association Test upon which DEI “trainings” are largely based. Despite its popularity, the IAT is flawed both empirically and philosophically. DEI, which depends on the IAT, doesn’t work because the IAT is based on at least two fallacies:

(a) Philosophical: It is a fallacy to suggest that implicit bias, if it exists at all, can have any real effect without some explicit bias. What you think (consciously or unconsciously) has no societal implications unless it results in something measurable. Implicit bias, even supposing it exists, is irrelevant; only what people actually do is societally relevant—unless we want to bring back a ridiculously amplified version of Orwellian ‘thought crime’ where even unconscious thoughts are punishable. If implicit bias predicts explicitly biased action, as the authors claim, why do we need the implicit measure? Conversely if implicit bias is compatible with the absence of explicit bias, it is irrelevant. Either way, the concept of implicit bias is redundant.

(b) Empirical: Banaji and Greenwald in their book Blindspot comment:

[G]iven the relatively small proportion of people who are overtly prejudiced and how clearly it is established that automatic race preference [as measured by the IAT] predicts discrimination…

Au contraire, Google reports “[T]he IAT cannot indicate whether a person is or is not prejudiced. . . .” Vox, a generally sympathetic source, said as long ago as 2017, “For years, this popular test measured anyone’s racial bias. But it might not work after all.” A 2021 academic article concludes: “IATs are widely used without psychometric evidence of construct or predictive validity.” The IAT is not scientifically valid and cannot predict anything.

The next sentence in the Banaji and Greenwald book shows why the IAT is so popular despite its conceptual and empirical deficiencies:

. . . it is reasonable to conclude not only that implicit bias is a cause of Black disadvantage but also that it plausibly plays a greater role than does explicit bias in explaining the discrimination that contributes to Black disadvantage.

“Black disadvantage” exists and social pressures demand an exogenous explanation, such as racial bias. Another possibility, an endogenous explanation, such as interest and ability differences between self-identified racial groups, has become a social taboo even though a mountain of evidence supports it. But this possibility is unacceptable to many opinion makers, so an alternative that seems to prove prejudice, no matter what people say, lives on long after it should have been abandoned.

If group differences exist, people will detect them and form stereotypes. The psychological process is a basic one called stimulus generalization, shown by animals from people to pigeons. Even if stereotypes are accurate, in the current political climate they will be labeled “prejudicial” and provide fuel for the DEI industry—an industry that survives partly because of society’s refusal to face the facts about human difference.

People, and groups, are not all the same. “Equal opportunity” does not mean (as Banaji & Dobbin seem to believe) proportionately equal group results. Even in a perfect society, there would be more male than female engineers and more female than male nurses. So what! We must discard the idea that people are identical in ability and ambition—group differences are not always due to outside prejudice—and instead treat people as individuals not identity markers. Until we end this contemporary obsession with group disparities, the perpetrators of the IAT and DEI fallacies will have a steady income.

Image: Adobe Stock