Assessing fairness in selection toward applicants who request accommodations in higher education admissions tests

This study assesses differential prediction and differential validity in higher-education admissions policy for individuals with various disabilities, who were granted or denied (on either technical grounds or professional considerations) test accommodations on the Israeli Psychometric Entrance Test (PET). The sample comprised 124,501 records of first-year students in 2,036 academic departments at six universities We applied a common-line regression model using the standard high-stakes PET score, the high-school Matriculation Exams score (ME), and their composite score (CS), as predictors, and the first year GPA as the criterion. Pearson correlations between predictors and criterion were also computed. Findings showed that accommodation policy was generally fair towards individuals with disabilities, with minimal over- or under-prediction. One important practical implication was that applicants should be cautioned that failure to provide adequate documentation of their disabilities could result in technical rejection, which may result in the under-prediction of their academic performance.