Biostatistics Journal Club
October 4, 2016
12:00p.m. – 1:00p.m.
COPH 2280
Amanda L. Golbeck, Ph.D.
Professor, UAMS Department of Biostatistics
Early Statistical Methods for Evaluating Faculty Equity Issues
Abstract
The American academic women’s movement began in the late 1960’s. It was noted early on that there were women faculty with equivalent training, experience, responsibilities, and productivity to men who were being paid less. Statisticians and others began to document and estimate the extent of these differences. In 1970, a study reported results of simple additive linear models of salary; this study included sex among the explanatory variables and presented correlations within a statistical testing framework. In 1972, a second study used these same methods with data on full time faculty members that came from a large national faculty survey. In 1972 and 1973, several more studies used these same data but improved the statistical methods in a number of ways that will be discussed in this talk. These improved studies promoted use of the statistical methods for purposes of advocacy: Build a regression model for male salaries, use it to predict women’s salaries, and advocate for individual increases when actual salaries for women were lower than predicted. The American Association of University Professors became involved and catalyzed many local studies around the country that used regression models to advocate for salary increases. These studies uncovered a number of problems with data; there have been many misuses of linear models in affirmative action; and there have been criticisms of using individual rather than statistical remedies for discrimination indicated by linear models.
- Golbeck AL. Statistical methods in academic salary equity studies: The early 1970s. Proceedings of the 1st International Statistics Conference in Croatia, 2016.
- Golbeck AL. Equivalence: Elizabeth L. Scott at Berkeley. Chapman and Hall/CRC, Boca Raton, forthcoming.