The GRE as a predictor of persistence to a PhD

Abstract

Because dropout from PhD programs is common and is a problem with serious consequences for both students and institutions, identifying indicators of likely dropout would be very valuable. Scores on admissions tests might be useful, but existing data on their utility is contradictory and typically based on highly restricted small samples from just a single institution or a small set of institutions. Programs dropping the GRE as an admissions requirement noted this lack of convincing evidence that the test was useful in predicting the criterion of primary interest–persistence in graduate school PhD programs versus early dropout. HLM and quartile analyses were used to provide that evidence with a sample of 1,672 graduate programs containing 157,924 students. GRE Verbal and Analytical Writing scores, but not Quantitative scores, are shown to predict persistence versus dropout in a variety of majors with especially strong results in business, engineering, and the physical sciences (e.g., in the physical sciences only 40% of the students with low GRE Analytical Writing scores in their programs persist while 78% of the students with high scores do so).