Stoa
Research/Evidence Ledger

The Implicit Association Test measures hidden prejudice that predicts discrimination.

Verdict: contested

Contested

The test is unstable over time and only weakly predicts real behaviour.

What the evidence shows

The Implicit Association Test (Greenwald et al., 1998) measures how quickly people pair concepts (e.g. faces with 'good'/'bad' words) and was promoted as revealing unconscious bias underlying real-world discrimination. Millions have taken it, and it underpins many diversity trainings.

A meta-analysis of its predictive validity (Oswald et al., 2013) found IAT scores correlated only weakly with discriminatory behaviour — and no better than explicit self-report measures. Separate work shows individual IAT scores are unstable, changing substantially when the same person retakes it. The test may capture something about cultural associations in the aggregate, but using an individual's score to diagnose personal prejudice, or to predict how they will act, is not supported by the evidence.

Sources

  1. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The implicit association test.

    Journal of Personality and Social Psychology, 74(6), 1464–1480

    Introduced the IAT as a reaction-time measure of automatic associations, proposed as an index of implicit attitudes.

    DOI: 10.1037/0022-3514.74.6.1464 →

  2. Oswald, F. L., Mitchell, G., Blanton, H., Jaccard, J., & Tetlock, P. E. (2013). Predicting ethnic and racial discrimination: A meta-analysis of IAT criterion studies.

    Journal of Personality and Social Psychology, 105(2), 171–192

    IAT scores were only weakly related to discriminatory behaviour and did not outperform explicit measures.

    DOI: 10.1037/a0032734 →

Related