Appellate Advocacy Blog

Editor: Tessa L. Dysart
The University of Arizona
James E. Rogers College of Law

Sunday, March 14, 2021

Revisiting – and Reconsidering – Implicit Bias

Many academic institutions, professional organizations, and private corporations have embraced implicit bias training as a method by which to combat discrimination. The concept of implicit bias states that all individuals harbor unconscious biases that lead to, among other things, discrimination and the unequal treatment of individuals based on race. Although certainly well-intentioned (eradicating discrimination is a moral imperative), empirical studies suggest that: (1) the Implicit Association Test (IAT), which is used to detect individuals’ implicit biases, is flawed; (2) there is a weak correlation between implicit biases and biased behavior; and (3) few, if any, attempts have been made to quantify the degree to which implicit bias, particularly in light of explicit biases, impacts behavior.

 1.    The Implicit Association Test is Flawed

Some scholars and commentators have relied on the Implicit Association Test (IAT) to diagnose an individual’s implicit biases. The problem is that the IAT is flawed in many respects.

To begin with, the IAT sets arbitrary cutoff scores to determine whether an individual’s responses reveal implicit biases, yet fails to provide any assessments of the differences, if any, between the many individuals who score above or below those cutoffs.[1] Additionally, IAT scores are arguably context-dependent, as the IAT produces different results for individuals when they complete the test multiple times.[2] Furthermore, the IAT fails to meaningfully distinguish between implicit and explicit bias. As one scholar explains, “the IAT provides little insight into who will discriminate against whom, and provides no more insight than explicit measures of bias.”[3] One commentator states as follows:

The IAT is impacted by explicit attitudes, not just implicit attitudes … It is impacted by people’s ability to process information quickly on a general level. It is impacted by desires to want to create a good impression. It is impacted by the mood people are in. If the measure is an amalgamation of many things (one of which is purportedly implicit bias), how can we know which of those things is responsible for a (weak) correlation with behavior?[4]

To be sure, one scholar acknowledged that “what we don’t know is whether the IAT and measures like the IAT can predict behavior over and above corresponding questionnaires of what we could call explicit measures or explicit attitudes.[5]

2.    Neither the Implicit Association Test Nor The Presence of Implicit Bias Reliably Predicts Biased Behavior

Empirical studies suggest that implicit biases do not predict biased behavior. Indeed, one researcher acknowledged that the IAT “cannot predict behavior at the level of an individual.”[6]  In fact, the evidence shows precisely the opposite:

Researchers from the University of Wisconsin at Madison, Harvard, and the University of Virginia examined 499 studies over 20 years involving 80,859 participants that used the IAT and other, similar measures. They discovered two things: One is that the correlation between implicit bias and discriminatory behavior appears weaker than previously thought. They also conclude that there is very little evidence that changes in implicit bias have anything to do with changes in a person’s behavior. These findings, they write, "produce a challenge for this area of research.[7]

Additionally, researchers recently “examined 63 studies that explicitly considered a link between changes in bias and changes in actions … [but] they found no evidence of a causal relationship."[8] Put simply, very few, if any, sociological or psychological studies have established with any degree of reliability that implicit bias directly or proximately caused biased, or discriminatory, behavior. As one social psychologist explains:

Almost everything about implicit bias is controversial in scientific circles. It is not clear, for instance, what most implicit bias methods actually measure; their ability to predict discrimination is modest at best; their reliability is low; early claims about their power and immutability have proven unjustified.[9]

This is not to say, of course, that implicit bias does not exist, or that it does not have a material impact on biased behavior. It is to say, however, that the IAT – and evidence supporting a connection between implicit bias and biased behavior – is, at best, premature and, at worst, untenable. As two prominent scholars explain:

[M]uch murkiness surrounds (a) the proper causal explanation for alleged IAT effects, (b) the psychological meaning of IAT scores, (c) the statistical generality and potency of alleged relations between IAT scores and actual behavior, and (d) boundary conditions on alleged IAT effects.[10]

What’s more, even where researchers have claimed to reduce implicit biases, they found no concomitant reduction in biased behavior. That fact alone should cause scholars who have championed implicit bias to think that, just maybe, they have jumped the proverbial gun.

3.    Few, If Any, Attempts Have Been Made to Quantify Implicit Bias’s Impact on Biased Behavior

Assuming arguendo that implicit bias impacts biased behavior, scholars have made little, if any, attempt to quantify implicit bias’s impact on biased behavior. For example, is implicit bias responsible for 5%, 10%, 20%, or 50% (or more) of biased behaviors? This is particularly problematic given that the presence of other factors, such as explicit biases and prejudices, directly impact biased decision-making. This flaw should not be surprising. After all, if implicit bias is the product of unconscious – and thus involuntary – actions, it would appear difficult for researchers to credibly claim that they possess the ability to reliably measure and quantify a phenomenon that resides outside of their conscious awareness. But without attempting to do so, reliance on implicit bias as a predictor of biased conduct raises more questions than answers.

The research cited above is merely a sample of the articles that have cast doubt on the nexus between implicit bias and biased behavior. To be sure, the point of this article is not to say that implicit bias bears no relationship to biased behavior. It is to say, however, that the evidence for such a relationship is inconclusive, contested, and, quite frankly unpersuasive. As such, the adoption of programs in universities and corporations that strive to educate students and employees on the allegedly negative effects of implicit bias is, at best, premature and, at worst, misguided. What’s more, relevant research has produced “little evidence that implicit bias can be changed long term, and even less evidence that such changes lead to changes in behavior.”[11]

Ultimately, eradicating discrimination, addressing inequality, and ensuring equal opportunity are moral imperatives. The question, however, is how best to do that.


[1] See Azar, B. (2008). IAT: Fad or Fabulous. American Psychological Association. Retrieved from:

[2] See id.

[3] Bartlett, T. (2017). Can We Really Measure Implicit Bias? Maybe Not. Retrieved from:

[4] Lopez, G. (2017). For Years This Popular Test Measured Anyone’s Racial Bias. But It Might Not Work After All. Retrieved from: racism.

[5] Id.

[6] Lee Jussim, Mandatory Implicit Bias Training is a Bad Idea (2017), available at:

[7] Bartlett, supra note 3, retrieved from:

[8] Brandie Jefferson, Change the Bias, Change the Behavior? Maybe Bot (Aug. 2019), available at:

[9] Jussim, supra note 6, available at:

[10] Gregory Mitchell & Philip Tetlock, Antidiscrimination Law and the Perils of Mindreading. 67 Ohio St. L. J. 1023- 1121 (2006).

[11] University of Arkansas, Research Questions Link Between Unconscious Bias and Behavior (July 2019), available at:

Current Affairs, Law School, Legal Ethics, Legal Profession, Science | Permalink


Post a comment