Compensation Committee Releases Data on Faculty Salary Equity

In a report carrying several caveats, the Faculty Compensation Committee presented results Thursday showing there was no evidence of a systematic salary gap between male and female faculty at Duke.

The results showed that while an expected salary at all three faculty ranks for a female faculty member would be marginally lower than for a comparable male faculty member – a result that had been found in previous years studies – the gap was small enough to fall into the realm of statistical variation.

The conclusion of the study presented by statistics faculty members Jerry Reiter, Merlise Clyde and Fan Li was that accounting for variables such as department, time in faculty rank and faculty rank at hire, "we can't really pin down the estimates accurately enough to feel confident in claiming there are systematic differences for gender," Reiter said.

"Nonetheless, I think the fact that the intervals for gender tend to include mostly negative values across all three ranks is something to note," Reiter added.  

The Duke committee did a regression analysis that adjusted the salary figures for a number of background characteristics, such as department, time in rank and rank at hire, since each also influence market salaries.

Using regression modeling is standard practice in other industries and at other universities to determine salary inequities, they added. But they also cautioned that other factors that influence salaries – such as productivity and awards – were not available in the data used in the study. In short, Reiter said, it's the best model Duke has to study a complicated issue.

The finding on gender salary equity was of particular interest to some Duke faculty because of a recent Chronicle of Higher Education (CHE) report stating that the average salary for female assistant professors was 80 percent of the average salary for male assistant professors.

Reiter and Clyde said the Duke study was more rigorous. Simply comparing averages, as in the CHE study, fails to take into account a series of other factors, such as market differences in salaries for different departments, they said.  For example, at Duke several departments that pay comparatively high salaries have more male than female assistant professors.  These across-department differences aren't reflected in the marginal rate of 80 percent, but they are accounted for in the regression analysis.    

"There may be a lack of data, but looking at unadjusted statistics just because one can compare them to other universities (as in the CHE study) is not a solution to the problem," Reiter said.

The Faculty Compensation Committee is charged with examining for systematic racial and gender differences in faculty salaries every two years. For the first time, the study broke out Asian faculty into a separate group and explored salaries for Caucasians, Asians and a third group called Underrepresented, which included African-Americans, Hispanics and Native-Americans.

The committee presented three findings:

  • There was no sufficient evidence at any rank to conclude that average salaries differ systematically for men and women, after adjusting for available background characteristics.
  • At the associate and full professor rank, average salaries are higher for Underrepresented faculty members than for Caucasian faculty members, after adjusting for available background characteristics.
  • At the assistant professor rank, average salaries are higher for Asian faculty members than for Caucasian and Underrepresented faculty members, after adjusting for available background characteristics.

None of the gaps were large, and small populations in several of the study groups could influence some results. For example, while Underrepresented Distinguished Professors had higher salaries, there are only nine of them.

During faculty questions, Clyde added that one result that did came out of the study is a persistent gap in expected salaries for new hires and long-time hires. The data indicate the market favors the former, Clyde said, and has in previous studies.

Reiter added that the study also was limited to the 969 people in tenure track positions, exempting the clinical sciences.  As has been historically the case and per the committee’s charge, the study didn't cover non-tenure track faculty such as professors of the practice and lecturers.  Several faculty members, including the committee, felt this would be a good addition to the analysis.

"As a member of a department that represents a larger and growing number of non-tenure track faculty, I'm curious and would support an equity study that looks at them as well," said Helen Solterer, professor in the Department of Romance Studies.