Regression analysis: Legal applications in institutional research

Abstract

Courts have decided dozens of discrimination cases within higher education over the past twenty years. During this time, increasing attention has been given to inferential statistical investigations, such as regression analyses, to determine liability in these lawsuits, particularly with respect to an institution' s pay, hiring, and promotion decisions of its faculty. In this case, regression analysis is used to evaluate the relationship between a set of independent (explanatory) variables with a single, dependent variable. Discrete (dichotomous) decisions within academia, such as hiring and granting tenure, can be analyzed using logistic regression methods; faculty salary, a continuous variable, can be investigated using multiple regression analysis.This chapter reviews multiple regression analysis, describes how its results should be interpreted, and instructs institutional researchers on how to conduct such analyses using an example focused on faculty pay equity between men and women. The use of multiple regression analysis will be presented as a method with which to compare salaries of male and female faculty in institutional studies or to establish or contest a prima facie case of discrimination in faculty salary equity cases and as a procedure for determining compensation and computing relief during settlement proceedings of faculty pay disparity cases. A brief review of significant case law and court decisions that have defined the role and scope of regression analysis in gender pay disparity cases is also provided. From this review, it will be clear that while statistical analyses are being used by the courts, the various interpretations of these 85 6 NEW DIRECTIONS FOR INSTITUTIONAL RESEARCH, no. 138, Summer 2008

Is One Health a Viable Strategy in Animal Health Litigation: Evidence from Civil Lawsuits in China

“…To test our hypotheses, we used a baseline and the variables of interest. We adopted the Logit and ordinary least squares (OLS) models accordingly [ 119 , 120 ], and for the dummy variables, we used a Logit model [ 121 , 122 ]. The models are specified in Equations (1) and (2): where Win represents a judge’s case decision; Win20 represents cases with a win ratio higher than 20%; is the constant; , , , and are the regression coefficients; and and are the stochastic terms.…”

“…To test our hypotheses, we used a baseline and the variables of interest. We adopted the Logit and ordinary least squares (OLS) models accordingly [119,120], and for the dummy variables, we used a Logit model [121,122]. The models are specified in Equations ( 1) and (2):…”