The Graduate Record Examination Aptitude Test as a predictor of graduate student performance
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Today, graduate schools have many more applicants than they have professors or available space. Therefore, the university administrators must select only those students who have a high probability of succeeding in graduate school. Various methods of selection are used by the different graduate schools, but the Graduate Record Examination Aptitude Test (GRE-AT) is becoming a prime predictor variable. However, information on the validity of this test is varied and relatively sparse. Also, since each university has its own standards, local studies should periodically be made to evaluate its requirements. The purpose of this study was to analyze data related to the use of the Graduate Record Examination Aptitude Test (GPE-AT) for admission to the graduate school of a large university of the South. Answers were sought to such questions as: "What is the relationship, if any, between scores on the GRE-AT and grades made in academic courses at the graduate level?" "Of what value are GRE-AT scores as predictors of success in graduate school?" The sample group consisted of 115 students who entered the graduate school in the fall of 1963 and were enrolled in major areas in the College of Arts and Sciences. Other criteria met by the group were: (1) They had GRE-AT scores recorded with the graduate school; (2) Their previous academic averages were available; (3) They had completed at least one graduate course, other than "special problems" courses; (4) They were United States citizens; and (5) They met other 1963-64 graduate school admission requirements. The basic data were obtained from departmental summary sheets and a copy of each student's permanent record card, which were provided by the Dean of the Graduate School. The summary sheets provided: (1) student's name, (2) registration number, (3) major, (4) previous academic average (PAA), (5) GRE-Verbal score, (6) GRE-Quantitative score, (7) GRE-Total score. The copy of each subject's permanent record card provided the student's complete graduate academic record subsequent to his admission to graduate school, and, for students previously enrolled in this same university, the previous academic record, along with date of birth, major area, etc. From this basic data additional data were computed, such as age, number of graduate hours (NGH), ana graduate grade point average (CGPA). Since the primary purpose of this study was to evaluate the use of the GRE-AT for predicting academic success at the graduate level, these scores were the prime predictor variables studied. The criteria of success were: (1) the student maintained a cumulative GGPA of 3.0 or higher, or (2) the student was awarded an advanced degree. Other variables that the GRE-AT was correlated with were the previous academic average (PAA), the number of graduate hours completed (NGH), and the student age at the time of enrollment. Pearson's Product-Moment Correlation was used for finding the relationships between the GRE-AT scores and the variables of GGPA, age and NGH. The Bi-Serial Correlation was used for the variables of PAA and academic success. Expectancy tables and decision theory graphs were, also, constructed for the relationships between the GRE-AT and the two variables of GGPA and academic success. The major conclusions that were drawn from the analysis of the data in this study are as follows: 1. The relationship of the GRE-AT and GGPA was low. However, the GPE-V and GRE-T were significant at the .01 level of confidence. 2. The PAA showed roughlv the same relationship with the GRE-AT as did the GGPA with the GRE. 3. The correlation between PAA and GGPA was .38. However, this was significant at the .01 level and was probably depressed by such factors as preselection of the sample and the small range of graduate grades. 4. The decision theory graphs for the relationship of the GRE and GGPA showed: (a) 98 of the 100 students with GGPA's of 3.0 and above made 400 or above on the GRE-V, (b) 85 of the 100 students with 3.0 or above made 400 or above on the GRE-Q, and (c) 82 of the 100 students with a GGPA of 3.0 or greater made 900 on the GRE-T. The false positives and misses on these graphs show why the relationships of the GRE-AT and GGPA were low. 5. Under the present admission requirements, 18 students or 18 per cent of the group who have since maintained a 3.0 or greater GGPA would have been barred from graduate school if the GRE-AT had been the only admission requirement. This is, also, 15.7 per cent of the sample group. 6. On the basis of this study, students who would not meet the current admission requirements (GRE-T score of 900 or more) did succeed. On the other hand, students who would meet present requirements for admission did not succeed. 7. The GRE-AT seems to be a fairly good supplementary tool for academic prediction. However, it should be used in conjunction with other predictors, such as PAA and/or as a confirmation of ability. The limitations of a study of this type are too numerous to make any absolute statements. It was not possible to control many of the subject and environmental variables which could have biased the results. Also, a relatively small, homogeneous sample group, enrolled in fairly heterogeneous majors could have considerably lowered the relationships of the GRE-AT with the criteria. Therefore, the above conclusions should be considered in view of these limitations.