A paper in the new issue of Psychology of Women Quarterly examines gender-based pay gaps among U.S. faculty using two methodologies. The multiple regression and resampling simulation approaches are different, yet they lead to the same conclusion - a gender-based pay gap exists.

The paper's three authors are faculty at the University of Tennessee-Knoxville. The study focuses on quantitative data gathered from the university's 14 colleges. Dr. Cheryl B. Travis, who is associated with the psychology and women's studies departments, says Tennessee has conducted salary studies for many years. This paper includes a new statistical methodology conceptualized by co-author Dr. Louis J. Gross of the Departments of Ecology and Evolutionary Biology, and Mathematics.

"We wanted to do a salary study using his new methodology, resampling, and have a comparable standard multiple regression study to see would they find similar outcomes," Travis says.

"When you have different methods and you're still finding the same outcomes in pretty much the same size, degree and direction that gives you an increased level of validation and validity to the finding that there is, in fact, a real and persistent gender gap."

The database for the study consisted of full-time faculty -assistant, associate and full professors - but did not include campus administrators. It also excluded the salaries for academic department heads. Pay based on 12-month appointments was adjusted to nine months. There was no ethnic data utilized.

Multiple regression provides information on strength of association, specific dollar estimates and the option to identify outliers by gender. Resampling simulation allows for analysis at the department level and is beneficial where distributions depart substantially from normal, particularly where there are unequal error variances.