Questions: Implicit Association Test and Implicit Bias Measurement

Question 1

A job applicant scores high on a race IAT. Human Resources concludes: 'This person will likely discriminate in hiring decisions.' Based on the research evidence, this conclusion:

• A. Is strongly supported — the IAT was designed to predict discriminatory behavior
• B. Overstates the evidence — individual IAT scores have moderate test-retest reliability (~.40–.50) and modest predictive validity for actual discriminatory behavior (r ≈ .15–.25)
• C. Is justified because IAT scores correlate perfectly with explicit prejudice measures
• D. Is justified because IAT responses cannot be faked or controlled

Answer: (B)

Drawing strong behavioral predictions from an individual's IAT score overstates what the measure can support. While the IAT reliably detects associations at the group level, its individual-level predictive validity for discriminatory behavior is modest (meta-analytic r ≈ .15–.25) and its test-retest reliability is around .40–.50 — meaning the score fluctuates substantially across sessions. The IAT is more useful for describing patterns in large samples than for predicting any individual's behavior.

Question 2

A person raised in a society with consistently negative media portrayals of a social group scores high on an IAT for that group, yet sincerely endorses egalitarian values and has never acted discriminatorily. The most accurate interpretation is:

• A. Their IAT score reveals hidden prejudice they are unwilling to acknowledge
• B. Their IAT score likely reflects cultural exposure to societal associations — which nearly everyone acquires — rather than personal endorsement of bias or a prediction that they will discriminate
• C. Their IAT score proves they will eventually discriminate under the right conditions
• D. Their self-report is more valid than the IAT in this case, so the IAT result should be discarded

Answer: (B)

Current consensus is that IAT scores reflect cultural exposure to stereotypes as much as personal prejudice. Living in a society with consistent stereotypic associations causes nearly everyone to acquire those associations implicitly. What varies is whether the associations are consciously endorsed, controlled, and allowed to translate into behavior. The IAT cannot distinguish cultural acquisition from personal prejudice — which is why individual-level behavioral predictions from IAT scores are unreliable.

Question 3

A person's IAT score measured today will reliably predict their IAT score if they take the test again in two weeks.

• T. True
• F. False

Answer: False

Test-retest reliability for the race IAT is around .40–.50 — substantially lower than the .70–.90 range typical of personality scales used as stable individual differences. An individual's score fluctuates across sessions due to mood, priming, context, and other transient factors. This instability limits the IAT's use as a trait-level measure of an individual's implicit bias and makes repeated individual measurement unreliable.

Question 4

Because the IAT bypasses conscious control, a high IAT score indicates that a person will discriminate regardless of their stated values or structural constraints in their environment.

• T. True
• F. False

Answer: False

The dissociation between implicit association and deliberate discrimination is central to this literature. Predictive validity for actual discriminatory behavior is modest (r ≈ .15–.25) and highly context-dependent. Top-down control, accountability structures, and clear behavioral norms substantially moderate whether implicit associations translate into behavior. The presence of an implicit association is not a behavioral destiny — people actively regulate the influence of their associations on their actions.

Question 5

Why do researchers argue that structural interventions (blind review, standardized hiring rubrics) are often more effective at reducing discriminatory outcomes than trying to lower people's IAT scores directly?

Model answer: IAT scores reflect broad cultural associations that nearly everyone acquires — directly reducing them at scale is difficult and effect sizes are small. Structural interventions change the decision-making environment so that implicit associations have less opportunity to influence outcomes, regardless of whether individual IAT scores change. The system can produce less discriminatory results without requiring every individual to first achieve a low IAT score.

This is the key distinction between person-centered and systems-centered approaches to bias. If the pathway from implicit association to discriminatory behavior runs through unstructured judgment, removing unstructured judgment removes the pathway — even if the association remains. This is why blind auditions increased female musicians in orchestras without any intervention on judges' implicit biases.