The contribution of regression analysis to the elimination of gender based wage discrimination in academia: A simulation

Abstract Potential responses of a university administration confronted by the existence of statistical evidence of sex discrimination in their salary structure are explored. A description of an appropriate use of regression analysis in eliminating sex discrimination in an academic institution is followed by an examination of regression models generally accepted by the courts in discrimination cases. Salary regressions are then estimated for a large midwestern university (LMU) in 1983–1984. The LMU data base is used in a simulation exercise to explore alternative methods of eliminating the initial statistical evidence of sex discrimination. The simulation results indicate that at LMU, the statistical evidence can be eliminated within a reasonable period of time, with a relatively small expenditure, and in a manner consistent with the long-run goal of establishing a non-discriminatory wage structure. If LMU is typical, then the standards of proof being accepted by the courts give potential violators guidance and incentives which should lead toward the minimization of sex discrimination in academia.