Adverse action in the selection process: beyond the uniform guidelines
The goal of this thesis is to supplement the analyses of a previously administered state-level multiple-choice examination using the 80 Percent Rule with Differential Item Functioning (DIF) analyses. DIF analysis addresses group differences when candidates are equated for level of ability (a more direct way to address test fairness) in an attempt to identify potentially biased exam items. By using a hierarchal logistic regression approach, the study evaluates Non-Uniform and Uniform DIF alongside the 80 Percent Rule assessment in an effort to address limitations identified in the 80 percent rule statistic as introduced in Section 4D of the Uniform Guidelines. The archival data utilized in this study were provided by a State Agency located on the West Coast of the United States. Job applicants (N = 1,517) across the state participated in the selection measure for a Cashier-based Classification and 8 demographics were utilized in the assessments (Male, Female, Over 40, Under 40, Caucasian, Hispanic, African-American, and Asian). It was concluded that the use of hierarchal logistic regression analysis in this study provided (1) clarity on violations identified through the 80 Percent Rule assessment and (2) evidence of item bias for several protected groups.