Salary Equity: Similarities and Differences in Outcomes from Two Common Prediction Models

Stepwise multiple regression was used to develop salary prediction equa-tions, one from each of the two faculty samples most frequently utilized in this type of research (males only, all faculty). These equations were developed using faculty information from an American Association of University Professors (AAUP) Category I institution showing typical sex-difference patterns in salary and other professional characteristics. Sex did not enter into the all-faculty equation. However, an analysis model using residuals showed that women, on the average, were paid significantly less than men. This sex difference was larger using the male, rather than the all-faculty, sample. Higher percentages of males than females were both overpaid and underpaid. Efficacy of the analysis model and differences in results from the two salary prediction equations are discussed. Since the early 1970s, projects examining potential sex differences in salaries paid to faculty members employed within postsecondary educa-tional institutions have proliferated. Many of these projects resulted from complaints of discrimination in pay, most frequently made by women faculty. Multiple regression has become the most accepted analysis technique used to examine faculty salaries for sex bias (Hengstler, Muffo, & Hengstler, 1982; Long, 1979; Morse, 1979), with the stepwise technique utilized most frequently. It has been used primarily to determine if a mean sex difference in salary exists once other important professional characteristics are statis-tically equated. The results then often are used to identify individuals who are underpaid. The specific potential predictor variables for salary vary depending on what information is available, usually in a computerized file, at th