Abstract

A clear understanding of what comprises successful student performance and the individual difference determinants of performance can improve our ability to select and train successful students. The first objective of this thesis was to specify a general multidimensional model of graduate student performance. The second objective was to determine the predictive validity of four of the major predictors of graduate student performance. The first objective was accomplished by a review and synthesis of the existing literature. The resulting 10 dimension taxonomy should be useful for future research developing measures of student performance and evaluating predictors of student performance. The second objective was accomplished by conducting metaanalysis of the predictive validity of the Graduate Record Exam General tests (GRE), the Graduate Record Examination Subject tests, the Miller Analogies Test (MAT), and undergraduate grade point averages (UGPA). The validity of the GRE and UGPA were estimated across 7 different criteria including degree attainment, grades, faculty ratings, publication record, and comprehensive examination scores. For the MAT meta-analyses, the validity of the MAT was estimated across 18 different criteria including all of the criteria examined in the GRE meta-analysis as well as measures of job performance and creativity which were included to establish the relevance of these measures for both academic and work performance. Overall, this study examined 6,743 correlations across samples totaling to 100,832 subjects. Results indicate several important findings. First, the GRE and MAT are valid predictors of most measures of academic performance including faculty ratings, degree attainment, research productivity, grades, and comprehensive examination scores. Second, across all academic criteria, the GRE Subject tests are the single best predictor of student performance. Third, the MAT, despite being an admissions measure, is a valid predictor of job performance and creativity suggesting that selecting students using standardized tests results is admitting students who will be successful in school and at work. This thesis represents the largest quantitative evaluation of these predictors ever conducted and provides compelling evidence that the GRE, MAT, and UGPA are all valid predictors of academic performance.