Support for the Validity of the Graduate Record Examination
American Psychological Association
Comments on the article by R. J. Sternberg and W. M. Williams (see record 1997-04591-002) regarding the empirical validity of the Graduate Record Examination (GRE) in predicting graduate student performance in psychology. It is argued that Sternberg and Williams used a misleading approach for analyzing their data. Predicting precise levels of achievement in graduate school from GRE scores alone when all admitted students are judged to be highly capable will necessarily result in low correlations. High achievement in any profession is dependent on a confluence of factors, and any single factor will necessarily be a relatively weak predictor of level of success. It is only by combining data regarding such factors that this type of prediction is strengthened. Given that Sternberg and Williams's study focused on the use of GRE scores in making graduate admissions decisions, a more appropriate methodology for examining their data involves decision theory and selection accuracy.