Constrained regression median for measuring possible salary discrimination

This study proposes a new analytical method for empirically examining possible salary discrimination against a protected (e.g. female or minority) group. The proposed approach is referred to as ‘Constrained Regression Median’ (CRM). The method can be used for regressing salary with job qualifications as dependent variables. The result of CRM is transformed into a 2 × 2 contingency table and the degree of salary discrimination is determined by the result of the contingency table. The approach can be easily explained to and understood by lawyers, judges and other individuals involved in a discrimination case. The CRM method is particularly useful and practical when a data set maintains the problems of outlier and/or multicollinearity between job qualifications and gender (or race).