The Effect of Disaggregation on Measures of Wage Discrimination in Academia

Statistical tests for the presence of sex, race, or ethnic based wage discrimination within a large organization can obscure discrimination within individual sectors of that organization, especially if the sectors have relatively few members of the minority group. As a result, if a large organization has units that operate at least semiautonomously, testing for discrimination in the organization as a whole may be neither appropriate nor sufficient. Yet this is what is generally done. A case study of a university is used to illustrate the potential sensitivity of measures of wage discrimination to the level of aggregation chosen for study.