Back to top

​University of Western Ontario

Available Studies

Western undertook a total of five salary equity studies (1975, 1989-91, 1995, 2005, and 2009). The most recent study in 2009 found no statistically significant difference between salaries of male and female faculty members, but found that gender-related salary differentials had arisen in new hires since the last equity study and subsequent adjustment in 2005. The university made $500,000 available in awards for gender-based below-the line adjustments, to be distributed through a Salary Anomaly Committee, established under the Collective Agreement. Faculty members were not required to apply, or to be nominated, to receive a gender-based adjustment.  

The 2005 study was conducted in order to make an analysis and description of the degree to which gender differences in compensation might persist at Western following the adjustments made by a previous pay equity exercise and the disbursement of Anomaly Fund that occurred in 1995. The 2006 study is an attempt to slightly alter the methodology used in the 2005 report, but to further create an estimate that will be used in a salary adjustment at UWO. The purpose of this report is to outline the process by which a gender pay gap would be addressed. The 2009 report builds on the analytical approach used in 2005, the general approach taken in this study involved first replicating the 2005 regression analysis, using the same variables but with data for 2009. The purpose of the replication was to test whether gender disparity had persisted.

The data used in the 2005 study comprise 918 Full Time Probationary and Tenured faculty Members and all Full Time Limited-Term Members (including "Permanents") who are eligible for faculty association membership. The 2006 study used the same data. In 2009, an updated data set was used  

A multiple regression analysis was conducted in the 2005 study. The model used for analysis regressed Salary on gender, years since highest degree, years since first degree, years at Western, rank, years at current rank, relative performance (from performance reviews), department average salary, and faculty. The study also looked at the distribution of men and women who fell above or below the predicted salary line. Further, a full regression was run on the male population only, and compared to female data, similar to a Blinder-Oaxaca method, only no estimate was created from this process.  

The 2006 study’s methodology is described as an algorithm. A regression is estimated for men and women. Then if a woman’s predicted salary in the female model is less than her predicted salary in the men’s model, the difference in the two predictions would constitute the salary adjustment necessary. The regression model used in the 2006 study was identical to the 2005 study, with the exception of the use of a hybrid variable of faculty and department was used instead of faculty alone. Further, the 2006 study opted to use Blinder-Oaxaca methodology explicitly to determine salary adjustments. The rationale for Blinder-Oaxaca methods was that men likely earn extra returns on variables such as rank and experience that would not be properly captured by a gender dummy variable in a single regression equation.     

The 2009 study estimated the same base regression model as the 2005 study, however the log of salary was used in one treatment, and some variables were given non-linear forms. Again, unit was used rather than faculty. There was some discussion to the interpretation of the regression results as a full population or as a sample, but ultimately the sample-view was chosen for this report. The Blinder-Oaxaca decomposition was not used in the 2009 study. In addition, the 2009 study included some graphical analysis of rank using data from “Bovey 6” universities (Guelph, McMaster, Queen's, Toronto, Waterloo, and Western).

The 2005 found a gender-based salary differential best estimated as $2,162, however this finding only applied to long-term faculty, as limited-term faculty exhibited no significant difference in gender pay. In the graphical analysis of the predicted vs. the actual pay, a larger share of female faculty were found to be below the predicted regression line than male faculty.

The main finding of the 2009 report was that observed characteristics explained the variation in male and female pay. In other words, there was no statistically significant salary difference between male and female faculty at Western. Graphical analysis suggested that the share of female faculty in higher-ranked positions was lower than the comparable Bovey 6 universities. This finding could suggest that gender discrimination is operating within the observed characteristics.

The 2005 study noted that there remains unaccounted salary variation among individuals (both male and female); this indicates that there are further dimensions to salary determination that remain to be analysed. The 2006 study used a formula that compared individual to predicted salaries, thus the suggested corrections are summarized as:  

  1. 91% of women Assistant Professors would receive a correction. For those receiving a correction, the average value is $ 3986 (average correction of 5.5% of salary) and the individual corrections range from $49 to $10,145.
  2. 57% of women Associate Professors would receive a correction with an average correction of $2355 (average correction of 2.5% of salary). Individual corrections range from $86 to $6610.
  3. 72% of women Professors would receive a correction with an average correction of $2759 (average correction of 2.5% of salary). Individual corrections range from $261 to $7617.

The study found that there were instances where male faculty would be eligible for salary adjustments (although far fewer such cases were present) using the same methodology, however the scope of the study was solely focused on women.  

The 2009 report suggested that observed that persistent efforts by UWO to address gender equity gaps likely contributed to the lack of statistical evidence of bias. However, the study also suggested that observed characteristics, which were controlled for, may be partially caused by gender bias. Further study may yet reveal a gender pay gap. For example, performance evaluations tended to rank women lower than men. Controlling for performance would then tend to mask a source of gender pay inequity. Among the recommendations of the report, future regression analysis should pay special attention to performance ratings.