An Analysis of the EEOCC "Four-Fifths" Rule

The U.S. Equal Employment Opportunity Coordinating Council is responsible for the elimination of inconsistencies among the operation of the agencies and departments that enforce Federal equal employment opportunity laws. It has recently promulgated quantitative guidelines to determine if employee selection and promotion rules have a disporportionate impact on a racial, ethnic, or sex group. These guidelines state that if the observed promotion or selection rate for any group is less than four-fifths of the rate for the group with the highest rate, then disproportionate impact will be concluded. In this paper a quantitative analysis of the guidelines is performed which demonstrates that their application in many situations can lead to large statistical errors of both kinds: type I errors (concluding that there is disproportionate impact on a group when in fact all groups have the same probability of passing) and type II errors (concluding that there is no disproportionate impact on a group when in fact there is). Thus, a large number of actual cases of discrimination will go undetected and truly nondiscriminatory practice will be erroneously deemed discriminatory a substantial fraction of the time. Type I and type II errors are presented for a range of actual passing rates for samples from two groups and three groups. Also presented, for up to ten groups, are the numbers of applicants per group required such that the expected minimum observed passing rate is at least four-fifths of the expected minimum observed rate, given that all groups have the same actual passing rate.