Fisher's Exact Test

Fisher's exact test is a statistical significance test used in bias auditing when sample sizes are small (typically n ≤ 40 or when expected cell counts fall below 5) and the chi-squared approximation would be unreliable. Unlike the chi-squared test, which uses an asymptotic approximation, Fisher's exact test calculates the exact probability of observing the given distribution of selections across groups under the null hypothesis of no discrimination. This makes it particularly valuable for auditing AI hiring tools in organizations with small applicant pools or when examining intersectional categories where subgroup sizes are limited. Fisher's exact test is computed from a 2×2 contingency table of group membership versus selection outcome. OnHirely automatically selects Fisher's exact test when sample sizes warrant, ensuring statistical conclusions remain valid regardless of data volume.