The study also looks at time indicators, such as length of time since receiving a degree, number of years at UNC and number of years in that particular rank. Rank is nearly always a significant predictor of salary, Williford said.
Salary for all faculty is converted to a nine month salary to serve as a base for comparison.
"It is important to note that there is no way to include a productivity measure in this study," Williford said. “That is something that is taken into consideration at the department level and at the school level, but it's not in a form at this point in time that we can grab that, quantify it and put it in a model like this.”
There are three methods used to analyze the data. In the population method, all faculty are included and become part of a model that is developed to predict salaries, Williford said.
The white male model has been used in past studies to show how different qualifications are enabling males to get a salary versus females.
“We create a model that simulates, 'Here’s what a white male would receive in salaries with these characteristics,' then we apply that model to other groups — besides white males — and we look to see if that model fits or not,” Williford said.
The third and final method showcases the differences between salaries that can be discussed in terms of percentage.
“From the regression model, we calculate the difference between the actual salary for each person and the salary that was predicted by the model that we built —that we refer to as the residual,” Williford said.
A negative residual would indicate that the individual would be expected to make more money than their actual salary. Conversely, a positive residual would indicate that the salary a person is making is more than what would be predicted.
A roster for each department is sent out along with the model, allowing the department members to look at that residual and identify particular people who are significantly below the mean.
“I like for them to look at the whole range, because someone may look like they are in the middle and their salary is what is should be, but they may be a very high performer,” Williford said.
Williford said that although data may appear normal for a particular person, there may be other factors not captured in the statistical model that would suggest that the person should be paid more.
“I don’t know what the provost plans to say or do when these data go out," Williford said. "I don’t know whether he intends for there to be a specific response back from department chairs or deans. In the past, the deans did provide feedback about why a salary was particularly low."
Clare Counihan from the Carolina Women's Center said the market value of departments and classes reproduces the disciplinary gaps.
“If econ is econ, and it's always going to be paid more than women’s studies, or it's always going to be paid more than English, that’s one of the huge places where the gendering of academic labor happens,” Counihan said.
Professors Misha Becker and Elizabeth Dickinson are meeting with the provost on Friday to discuss salary disparities.
“He has asked for our help," Dickinson said. "We’re trying to figure out what do we do with this concept not just of gender equity but of race, ethnicity, etc."