Validation of the Graduate Record Examination and the Miller Analogies Test in a doctoral program

In view of previous inconclusive studies and often crude criteria, the authors studied the predictive validity of the Graduate Record Examinations (GRE) Aptitude Test and the Miller Analogies Test (MAT) in a doctoral program in Education. Four rather discrimina-ting criterion variables derived from an analytic exercise and course grades in statistics and research were used. Twenty-five classes in a core course over a nine-year span provided subjects (N = 348) for the data analyses. The highest first-order correlations between GRE scores and criterion measures ran about.5 whereas those for the MAT were very low. Considering both the restriction in range of scores and the fact that a combination of predictors through multiple regression procedures yielded still higher correlations than those afforded by single prediction, the investigators concluded that the GRE Aptitude Test was useful in this program, but not the MAT. EVIDENCE on the predictive validity of the various Graduate Record Examinations (GRE) of the Educational Testing Service (1949-1977, 1946-1977) has indeed been mixed, if one may judge from reviews by Lannholm (1968) and by Thacker and Williams (1974). The latter two reviewers have described the results as inconclusive and have called for further studies. Much of the so-called inconclusiveness has come from low correla-tions and these, in turn, from persisting conditions, especially restric-tion in range owing to selective admissions and attrition. But not to be overlooked is the crude nature of the criterion. Thus, however appeal