Koenig, J.A.; Wiley, A. The Validity of the Medical College Admission Test for Predicting Performance in the First Two Years of Medical School. Academic Medicine, 71;S83-S85, 1996.

PURPOSE: To assess the validity of the Medical College Admission Test for predicting performance in the first two years of Medical School.

METHODS: Sixteen medical schools, representative of the 125 U.S. medical schools, participated in the predictive validity study. Data obtained from each school included basic science grades, USMLE Step 1 scores, MCAT scores, undergraduate GPA (UGPA), undergraduate institution selectivity, medical school grades, and Step 1 scores. Relationships among the variables were analyzed using multiple regression analysis. Regressions were run separately for the entrants to each school. Two student cohort groups were studied, those who entered medical school in the falls of 1992 and 1993. Performance was summarized by computing end-of-year grade point averages for each year and a two-year cumulative GPA, which was the simple average of the two end-of-year averages. Multiple correlation coefficients were obtained for five unique predictor sets: Undergraduate science GPA and undergraduate nonscience GPA (UGPAs); MCAT scores; UGPAs and selectivity; UGPAs and MCATs; and UGPAs, selectivity, and MCAT scores. For each regression, all variables were entered simultaneously.

RESULTS: MCAT scores had slightly higher correlations with medical school grades (median correlations ranging from .615 to .67) than did UGPA data (median correlations ranging from .54 to .58). Prediction of performance was improved when the two sets of predictors were considered together (median correlations of .70 to .76). A considerable difference was noted when median correlations were compared for UGPA predicting Step 1 (median R=.48) versus MCAT predicting Step 1 (median R=.72). When MCAT and GPA data were considered jointly, the median correlation coefficient increased. MCAT's utility in the admission process became evident when values obtained for predictor set 3 (UGPA and selectivity) were compared with those obtained for predictor set 5 (UGPA, selectivity, and MCAT). The increases in median values were .15 for year 1 GPA, .11 for year 2 GPA, .12 for cumulative GPA, and .22 for USMLE Step 1.

CONCLUSIONS: The authors noted that establishing the validity of the MCAT for predicting performance in the first two basic science years of medical school is a crucial step in the overall validity studies planned for the MCAT. In addition to the variables examined in their study, the authors suggested that other criterion variables need to be evaluated such as USMLE Step 2 and 3 scores as they become available. The authors stated that a study is underway that will examine the extent to which MCAT predicts performance differently for individuals grouped by gender and racial/ethnic status.

Contact Us    © 1995-2008 AAMC    Terms and Conditions    Privacy Statement