This past May, 78 per cent of female faculty received increments to their annual salary, ranging from $50 to over $10,000. These raises followed from the Faculty Pay Equity Committee Report (August, 2005) and from the Implementation Committee Report (March, 2006) that examined gender-based differences in salary at Western.
The widespread presumption underlying these reports is that Western has been discriminating against women in terms of compensation. Neither committee was mandated to examine unfairness in men's salaries.
To examine the question of gender-based differences, the committees carried out a multiple regression analysis in an attempt to determine what factors explain individuals' actual salaries.
Factors like number of years at Western, faculty membership, years since highest degree, gender, and so on were used in the analysis and the result was an equation that tried to explain why different people make different salaries (e.g., people in Faculty X make more than people in Faculty Y which partially explains why Chris, who is a member of X, makes more than Pat, who is a member of Y). Both committees concluded that gender is a factor in Western salaries; even when the other factors were taken into account, men earned, on average, about $2,200 more than women.
We would like to comment on some interesting aspects of these reports.
One is that the committees concluded that assistant professors and, more specifically, those most recently hired were suffering the most from gender-based pay inequity. This is a rather surprising, and disturbing, conclusion. However, it also seems implausible for a couple of reasons.
- if the presumption about Western discriminating against women is correct, it would be senior female faculty who would show the strongest impact; over time, their salaries would have fallen increasingly further behind their male counterparts.
- the committees' conclusion implies that the main source of male-female salary discrepancies is due to starting salary differences. That is, those women who have been hired very recently must have received much smaller salaries than their male counterparts. Could this possibly be true? Interestingly, starting salary was not included as a factor in the analysis and, thus, it was not investigated as an explanation. However, there are at least two arguments suggesting that starting salaries is not the reason for the male-female differences.
First, there has been a public commitment by this university for over 10 years to ensure that women are paid equitably compared to men. Is it likely that during this time Western's deans have been allowing women faculty to be shortchanged in starting salaries?
Indeed, in our faculty, starting salaries are negotiated by the dean himself.
According to the multiple regression analyses, the estimated male-female starting salary difference in Social Science for a new PhD is $5,535. If this estimated discrepancy reflects bias, rather than a flaw in the regression analyses, then our dean must be a misogynist. We don't believe that. Nor does it seem likely that the deans in the two faculties in which the estimate of the discrepancy was more than $10,000 (Law and Dentistry) could have acted so egregiously.
Second, and equally importantly, the data in Figure 3 of the August, 2005 report suggest that there is virtually no male-female salary difference among faculty making up to about $80,000, which presumably includes the assistant professors who were hired recently.
There is, of course, a straightforward way to determine whether a male-female salary difference exists among newer faculty. Analyze those data directly, including doing multiple regression analyses only on assistant professors. We have been told that these analyses have not been done.
Another noteworthy aspect of the reports is that there are about a dozen or more male faculty members making more than the highest paid female faculty member. More importantly, according to the multiple regression analysis, they are all making substantially more than predicted by the equation (see Figure 3 in the August, 2005 report). Why? Merit was included as a factor in the analysis, although the only merit indicator used was the most recently available PAI rating (used for yearly pay raises), rather than any career-based measure. (Male and female faculty received virtually identical PAI ratings, relative to others in their departments.)
Could these large salaries be explained by merit factors that weren't included in the regression analyses, for example, career achievement, competitive market adjustments used to retain these individuals or CRC status? Or, are the salaries a result of having served in a high administrative office (e.g., dean or above). Or, are they due to genuine sex-based discrimination? It's impossible to know from the reported analyses.
However, if these salaries are due to legitimate merit factors omitted from the regression analyses, both those analyses and any conclusions about the effect of gender would be seriously compromised.
Importantly, inclusion of these very highly paid individuals in the analysis may be the reason that the committees came to the conclusion that men make significantly more than women. That is, these highly paid male faculty (and not the other 700 male faculty) may be the source of the significant gender effect found in the analyses. This hypothesis could also be addressed. We could include some of the missing variables, e.g., market adjustment, in another analysis. We could also remove any salaries over some large value (e.g., $150,000) and redo the analyses. We have been told that none of these analyses have been done.
In conclusion, we have suggested that improbable results and the omission of important variables make the regression analyses and the subsequent pay raises suspect. We have pointed out that data either exist in the current data base or could be gathered without too much trouble to test some of the hypotheses we outlined. Unfortunately there is no will in the university administration to do so.
One of us was told directly by a senior official who is ultimately responsible for pay equity that no further analyses would be done on the data reported in the two equity reports. A private appeal to the university administration's publicly stated commitment to transparent and accountable decision-making was not persuasive in changing this official's mind.
We now make public our appeal for further data analyses.