Useful Mind / did not hold up practise daily →
Did not hold up all 15 findings

The implicit bias test

Does the implicit bias test work? Not well enough to judge a person by.

A short online test reveals your hidden prejudice, and training it away makes you fairer.

What actually happened

The IAT is steady enough to study groups and too unsteady to judge a person: test–retest reliability sits around r = 0.5, so the same person scores differently from week to week. Oswald et al.’s meta-analysis (2013) found it predicts discriminatory behaviour weakly. Forscher et al. (2019), pooling 492 studies, found that procedures which do move implicit measures do not then move behaviour. Greenwald, who built the test, has said it should not be used to diagnose individuals.

Check it yourself: doi.org/10.1037/a0032734

What to do instead

Change the process, not the person. Blind auditions, structured interviews with fixed questions and fixed scoring, and decisions written against stated criteria all cut biased outcomes without anyone’s inner state having to change.

The laws that do this job properly

Checked and written by Devendra K Jha. Published 2026-09-16. How a grade is decided.

Every law on Useful Mind tells you how strong the evidence behind it is. The ones that failed are kept here rather than quietly deleted, because knowing which advice is wrong saves you as much as knowing which is right.

See the ones that held up