ANALYSIS OF GENDER EQUITY
IN 1997 FACULTY SALARIES
AT UW-MADISON
By Margaret
Harrigan,
Office of Budget, Planning and Analysis
Distributed to the Academic Deans for the Fall 1999 Gender Pay Equity Review
Table of Contents Introduction Methodology Rank as an Explanatory Variable
Results Results by College (Table 1. Analysis of faculty salaries by School/College)
Medical School Results (Table 2. Analysis of faculty salaries in the Medical School)
Results by Divisional Committee (Table 3. Analysis of faculty salaries by Divisional Committee)
Results by Cohort of Hire (Table 4. Analysis of faculty salaries in by Cohort of Hire)
Limitations of the Analysis Conclusions Appendix A. Summary Statistics for Women and Men Faculty Appendix B. Techical details about the analyses and key to variables used in the regression analyses Appendix C. Detailed regression results of the 1997 analysis of gender equity in faculty pay.
(Appendix C is posted here as a 32 page MS-Word document)The June 1992 Gender Equity Study of Faculty Pay at the University of Wisconsin - Madison found a significant gap in salaries between men and women. As a result of the study, the Faculty Senate adopted a plan to provide salary adjustments to women faculty. Salary increases were provided to about 370 women (86% of all women faculty) in June 1993. In addition, the Senate called for a follow-up study that would make further recommendations to ensure gender pay equity on a continuing basis.
In December 1995 the Faculty Senate adopted as guidelines the recommendations included in the March 1995 Report of the Follow-up Committee to the Faculty Gender Pay Equity Exercise. As part of these recommendations, three committees are asked to monitor faculty salaries on a routine basis: The Advisory Committee to the Equity and Diversity Resource Center, the Committee on Women in the University, and the Faculty Commission on Compensation and Economic Benefits. The report advocates that statistical analyses such as those contained in the follow-up report and the original study should be repeated on a periodic basis.
This document represents an update of the pay equity study, using faculty salary information from November 1997. UW Madison contracted with an outside consultant, Delores Conway (Associate Professor of Business and Statistics at the University of Southern California), to advise us with this analysis. We adopted Dr. Conway’s recommendations and incorporated them in the analysis described below. The regression results for the university as a whole, and for each of the university’s schools and colleges, find no statistically significant gap in salaries between men and women faculty in 1997.
With multiple regression analysis, we attempt to explain differences in salaries by taking into account factors known to affect salary, such as education and experience. Gender and race/ethnicity are also included as variables in the regression equation, so that we can measure differences in salaries between men and women and between minorities and non-minorities after compensable factors have been taken into account. Compensable factors used in these regression analyses include years on the faculty at UW, highest degree earned, rank, outside market value of the person's discipline, years since highest degree, hiring cohort, and department within the university. However, no measure of merit other than rank is included in the regression equation. Although the number of publications authored, awards received, etc. are not included and are important factors in the determination of salaries, this omission should not have an impact on the results of this analysis unless there is a significant difference between women and men in the quality of their work.
As in the earlier reports, the UW- Madison faculty is divided into two groups for this analysis: faculty outside of the Medical School and faculty within the Medical School. Medical School faculty are treated separately because they have a considerably different structure for their pay.
Rank as an Explanatory Variable
While it is accepted that rank should be a compensable factor, it has been argued that the assignment of rank may be biased against women. This could result, for example, if women were hired at a lower rank or promoted more slowly than were comparable men. If this were the case, including rank as a predictor in the regression equation would cause us to understate the true salary gap. Because of these concerns about rank, the 1992 faculty pay equity study reported two sets of analyses: one that included rank as a predictor and one that excluded rank.
Delores Conway, our consultant for this report, argues that rank is an important predictor of salaries and must be included in any faculty salary analysis. This university competes in the academic marketplace for faculty and rank is an important factor in determining salaries at all universities. If women faculty are promoted more slowly or hired at a lower rank than men are, this should be analyzed and addressed independently of a salary equity study. Therefore, this report focuses on the results of regression analysis that includes rank as an explanatory variable.
Table 1 shows the results of the regression analysis for the campus as a whole (excluding the Medical School) and for each school and college separately. The estimated coefficient on gender, the percentage by which salaries of men exceed salaries of women (calculated from the gender coefficient), and the standard error of the estimate are reported. Detailed results of the regression analyses are provided in Appendix C.
As indicated in the table, the gap in 1997 salaries between men and women is -0.5 percent for UW faculty as a whole (95 percent confidence interval is –2.3% to 1.3%). That is, women in 1997 are paid approximately one-half percent more than men are, on average, after controlling for differences in rank, experience, field, and other factors in the equation. This difference is not statistically significant from zero at the 95 percent confidence level.
In order to draw accurate conclusions from the data, it is important that the individuals whose salaries are being analyzed are fairly homogeneous (for example, the employes should have similar job responsibilities, performance criteria, market, etc.). Therefore, rather than look at the university as a whole, each school or college should be examined independently. In addition, analyses by cohort of hire or by separate discipline areas (such as by divisional committee affiliation) may be illuminating. However, in small subgroups (such as schools with less than 40 faculty), or where there are few men or few women in a subgroup, the data for one or two individuals may strongly influence the results of the regression. Thus results of these regressions must be interpreted with caution. Table 1 includes results by school or college other than the Medical School; Medical School results are contained in Table 2. The analysis by divisional committee is reported in Table 3. Table 4 presents the analysis by cohort of hire.
The gender gap in salaries varies somewhat by college, although for none of the schools or colleges are the differences statistically significant at the 95 percent confidence level. Two units show a negative salary gap (women are paid more than men): Pharmacy and Letters and Science. The other schools and colleges show a salary gap favoring men, ranging from less than one percent in the Business School to about five percent in the School of Law.
Table 1. Analysis of Faculty Salaries by School/College
UW-Madison School/College Total Faculty
Women Faculty
Gender Coefficient
Standard Error
% Gender Pay Gap
University Wide (excludes Medicine) 1718
406
-0.005
0.009
-0.5%
Agricultural and Life Sciences 291
31
0.021
0.027
2.2%
Business 80
13
0.007
0.055
0.7%
Education 140
52
0.013
0.024
1.3%
Engineering 172
12
0.032
0.041
3.3%
Human Ecology 39
28
0.017
0.029
1.7%
Law 37
10
0.053
0.059
5.5%
Letters & Science 836
210
-0.021
0.013
-2.1%
Nursing and Pharmacy 51
27
0.021
0.076
2.2%
Pharmacy alone 29
5
-0.067
0.110
-6.5%
Veterinary Medicine 50
10
0.024
0.038
2.4%
Continuing Studies 22
13
0.028
0.130
2.8%
Notes: Percent gender pay gap is the amount by which the men's salaries exceed women's salaries, after controlling for compensable factors in the regression equation. Faculty are reported in the school or college of their primary tenure department. Nursing is included in Pharmacy because there are currently no male faculty in Nursing. Five faculty with a tenure home in the Institute for Environmental Studies are reported in the college of their secondary tenure department or in Agricultural and Life Sciences, if they have no secondary tenure department.
Source: UW-Madison Office of Budget Planning and Analysis November 1997 payroll and tenure file information.
Table 2 includes the results of the examination of the Medical School faculty. As mentioned above, the Medical School is not included in the analysis of the university as a whole. A substantial number of Medical School faculty are in clinical departments and earn clinical practice income through the UW Medical Foundation as well as a salary paid by the university. Because of the clinical pay, Medical School faculty salaries are analyzed separately from other UW faculty. In this study, faculty are divided into two groups: those who receive clinical practice pay (CPP) and those who do not. Two definitions of income are analyzed: base salary paid through the university and total income (including base salary and clinical practice income paid through the UW Medical Foundation). Note that only tenured and tenure-track faculty are included in this analysis -- CHS faculty are not examined here.
Table 2. Analysis of Faculty Salaries in the Medical School
Medial School Faculty Total Faculty
Women Faculty
Gender Coefficient
Standard Error
% Gender Pay Gap
All Medical School Faculty Base Salary only
367
57
-0.009
0.026
-0.9%
Total Compensation (including clinical pay)
367
57
0.016
0.039
1.6%
Medical School Faculty not receiving clinical practice pay Base Salary Only
192
45
-0.016
0.027
-1.6%
Medical School Faculty receiving clinical practice pay Base Salary Only
175
12
-0.007
0.058
-0.7%
Total Compensation (including clinical pay)
175
12
0.090
0.079
9.4%
Notes: Percent gender pay gap is the amount by which the men's salaries exceed women's salaries, after controlling for compensable factors in the regression equation.
Source: UW-Madison Office of Budget Planning and Analysis November 1997 payroll and tenure file information and UW Medical Foundation.
As shown in the table, there is no statistically significant difference between the salaries of men and women in the Medical School, after controlling for compensable factors such as rank and experience. Compared to men in the Medical School, women receive slightly more base salary and slightly less total income, on average. Since women are more likely to be in basic science departments and less likely to earn CPP than men are, it is important to divide the faculty by whether or not they earn CPP. Of those faculty who receive a base salary only, women on average earn 1.6 percent more than men. Only twelve women faculty currently receive CPP at the university. Compared to men who earn CPP, these women receive about 1 percent more in base salary and 9 percent less in total income. None of these differences are statistically significant at the 95 percent confidence level.
Results by Divisional Committee
At UW-Madison, each faculty member chooses one divisional committee (Humanities, Social Studies, Physical Sciences, or Biological Sciences) for purposes of tenure review. Generally, faculty members in a given tenure department are associated with only one divisional committee, although some departments are affiliated with two or more divisions (for example History with Humanities and Social Studies, Biochemistry with Biological Sciences and Physical Sciences). For this analysis, current faculty members who have not yet chosen a divisional committee are assigned to a divisional committee based on their tenure department.
In Table 3 the results of separate regression analyses are reported for faculty in each divisional committee. Men are paid slightly more than women, on average, in two divisions: Biological Sciences and Social Studies. In Humanities and Physical Sciences, women earn slightly more than men, after controlling for differences in rank, experience, and other compensable factors. None of the differences in salaries between men and women are statistically significant for any of the four divisions.
Table 3. Analysis of Faculty Salaries by Divisional Committee
UW-Madison Divisional Committees Total Faculty
Women Faculty
Gender Coefficient
Standard Error
% Gender Pay Gap
Biological Sciences 310
54
0.013
0.022
1.3%
Humanities 375
140
-0.026
0.014
-2.6%
Physical Sciences 478
37
-0.002
0.029
0.2%
Social Studies 555
175
0.003
0.014
0.3%
Notes: Excludes Medical School faculty. Percent gender pay gap is the amount by which the men's salaries exceed women's salaries, after controlling for compensable factors in the regression equation. Faculty who have not yet chosen a divisional committee affiliation are assigned to one based on their tenure home. In a few cases, individuals who are the only person in a department with a particular affiliation are reassigned to the division used by other members of the department.
Source: UW-Madison Office of Budget Planning and Analysis November 1997 payroll and tenure file information.
In addition to examining salary differences by college, analyses were also undertaken to compare the gender gap for different hiring cohorts. This allows a comparison of the gender differences in salaries between those hired recently and those who have been on the faculty for many years. Faculty were divided into four hiring cohorts: those hired prior to July 1979, hired between July 1979 and June 1988, hired between July 1988 and June 1993, and hired since June 1993. The last cohort corresponds to faculty members who were hired after the 1993 faculty pay equity exercise was completed. A gap in salaries for the last cohort could signal a problem with starting salaries. Again, separate equations were used to analyze the salary differences by cohort. Results of these regressions are presented in Table 4. For two of the hiring cohorts analyzed, women are paid more than men, on average, after controlling for factors such as rank and discipline. The opposite is true for those hired since 1993 and hired between 1979 and 1988. None of these differences are statistically significant at the 95 percent confidence level.
Table 4. Analysis of Faculty Salaries by Cohort of Hire
Cohort of Hire Total Faculty
Women Faculty
Gender Coefficient
Standard Error
% Gender Pay Gap
All Cohorts Combined 1718
406
-0.005
0.009
-0.5%
Hired before July 1979 578
63
-0.054
0.028
-5.3%
July 1979 - June 1988 459
109
0.010
0.017
1.1%
July 1988 - June 1993 428
140
-0.008
0.013
-0.8%
Hired since July 1993 253
94
0.009
0.014
0.9%
Notes: Analysis is based on faculty positions and salaries as of November 1997. Percent gender pay gap is the amount by which the men's salaries exceed women's salaries, after controlling for compensable factors in the regression equation. Excludes faculty in the Medical School.
Source: UW-Madison Office of Budget Planning and Analysis November 1997 payroll and tenure file information.
It is important to note the limitations of regression analysis. First, the accuracy and reliability of the multiple regression approach depends on whether the model includes all the relevant factors. Quantitative measures of merit in teaching, research, and service are not included in these regression models. Thus, this analysis assumes that men and women are on average equally meritorious, given the other variables in the analysis. If this assumption is untrue, the results of the regressions are suspect. Second, statistical analyses are useful for providing data about faculty salaries in the aggregate. Although the results of these analyses show no significant gender gap in salaries for the university as a whole or for any of the schools or colleges, it is possible that a pay gap exists for an individual faculty member. Third, an analysis with only a small number of observations may be unduly influenced by the characteristics of one or two individuals. The number of individuals in the school or college or other grouping should be considered when interpreting the regression results. However, an analysis of large groups --such as a large college-- may obscure some gender differences within the group, such as at the department level.
This study contains many multiple regression analyses designed to measure the difference in salaries between men and women faculty after controlling for legitimate compensable factors such as rank and experience. Analyses were performed for the university as a whole, for each school and college, by divisional committee, and by cohort of hire. In none of the regression equations were statistically significant gender salary gaps observed.
We hired a statistical consultant, Delores Conway, to advise us on our methodology. In following her recommendations, we simplified the analysis. For the 1992 and 1995 studies, we included analyses both with rank and without rank as a predictor. Conway contends that rank is an important determinant of salaries and that if we exclude rank we throw away valuable information. If there is concern about gender bias in assigning faculty rank, it should be the subject of a separate study. Therefore, this report includes rank as a predictor of faculty salaries.
Our consultant also stated that graphs could be useful in gaining a better understanding of salary distributions. In particular, they may be helpful in identifying "outliers" paid substantially less (or more) than others in a unit with the same rank and similar years since degree. Although not included in this report, graphs showing men and women faculty, their salaries, rank, and years since highest degree were prepared for each tenure-granting department. These graphs were distributed to the deans of each school and college to assist them in their on-going efforts to maintain equity in faculty pay.
[Top of Page] [Appendix A] [Appendix B] [Appendix C] [ Fall 1999 Gender Pay Equity Review]
[Academic Planning and Analysis Home]posted September 15, 1999. UW-Madison Office of Budget Planning and Analysis.