Demographic Parity

Demographic parity (also called statistical parity or independence) is a fairness metric that requires an AI system to select candidates at equal rates across all demographic groups. Formally, a system satisfies demographic parity if P(selected | group A) = P(selected | group B) for all groups A and B. In AI hiring, demographic parity would mean that the proportion of candidates advancing from each racial, gender, or age group is the same. Demographic parity is closely related to the four-fifths rule, which is a relaxed version: rather than requiring exact equality, the four-fifths rule allows selection rates to differ as long as no group's rate falls below 80% of the highest rate. While demographic parity is intuitive and easy to measure, it has been criticized because it does not account for differences in qualifications across groups. If one group has more qualified candidates on average (due to historical access to education and opportunity), enforcing strict demographic parity could require selecting less qualified candidates from that group or more qualified candidates from another. Most regulatory frameworks use the four-fifths rule rather than strict demographic parity as their compliance threshold.