Few academic ideas have been as eagerly absorbed into public discourse lately as “implicit bias.” Embraced by Barack Obama, Hillary Clinton and most of the press, implicit bias has spawned a multimillion-dollar consulting industry, along with a movement to remove the concept of individual agency from the law. Yet its scientific basis is crumbling.
Implicit-bias theory burst onto the academic scene in 1998 with the rollout of an instrument called the implicit association test, the brainchild of social psychologists Anthony Greenwald and Mahzarin Banaji. A press release trumpeted the IAT as a breakthrough in prejudice studies: “The pervasiveness of prejudice, affecting 90 to 95 percent of people, was demonstrated today . . . by psychologists who developed a new tool that measures the unconscious roots of prejudice.”
In the race IAT (there also versions for everything from gender to disability to weight), test-takers at a computer are asked to press two keys to sort a series of black and white faces and a set of “good” and “bad” words. For part of the exercise, the test-taker presses one key for white faces and words like “happy,” and the other key for black faces and words like “death.” Then the protocol is reversed, pairing white faces with “bad” words and black faces with “good” words. (The order is randomized, so some test-takers sort black faces with “good” words first.)
A majority of test-takers—including about 50% of blacks, according to some accounts—are faster at the sorting game when white faces are paired with good words. This difference is said to represent an “implicit bias” in favor of whites that can explain racial disparities in society.
Mr. Greenwald and Ms. Banaji did not pioneer response-time studies; psychologists already used the methodology to measure how closely concepts are associated in memory. And it’s widely accepted in psychology that automatic cognitive processes and associations help people navigate daily life. But Mr. Greenwald and Ms. Banaji, now at the University of Washington and Harvard, respectively, pushed the technique into charged political territory. Not only did they confidently assert that any differences in sorting times for black and white faces flowed from unconscious prejudice, they claimed that the implicit bias allegedly measured by the IAT could predict discriminatory behavior. In the final link of their causal chain, they argued that this unconscious and pervasive predilection to discriminate is a powerful cause of racial disparities.
As they wrote in “Blindspot,” their 2013 best seller: “Given the relatively small proportion of people who are overtly prejudiced and how clearly it is established that automatic race preference predicts discrimination, 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.”
If these sweeping claims were correct, every personnel decision could be challenged as the product of implicit bias. The pressure to guarantee equality of outcome through quotas would grow stronger. But the politics of the IAT had leapfrogged the science behind it. Core aspects of implicit-bias doctrine are now under methodological challenge.
A person’s IAT score can vary significantly each time he takes the test, undercutting its reliability as a psychological instrument. Test scores have almost no connection to what IAT research ludicrously counts as “discriminatory behavior”—trivial nuances of body language during a mock interview, say, or a hypothetical choice to donate to children in Colombian slums rather than South African ones.
Mr. Greenwald and Ms. Banaji now admit that the IAT does not predict “biased behavior” in the lab. (No one has even begun to test its connection to real-world behavior.) The psychometric problems associated with the race IAT make it “problematic to use to classify persons as likely to engage in discrimination,” they wrote, along with a third co-author, in 2015.