DO ADMISSION TESTS PREDICT ACADEMIC PERFORMANCE FOR MEDICAL STUDENTS? A LONGITUDINAL STUDY

BACKGROUND: The National Exam for Admission to Higher Education (EXANI-II) is a tool widely used in Universities in Mexico as a standardized admission test. Nevertheless, there is no published evidence establishing whether this test correlates with or predicts academic performance of medical students. EXANI-II has been used since 2003 to select students admitted to the School of Medicine, University of Yucatan.

METHOD: We studied the 2003-2008 cohort, which was the first generation of students selected with EXANI-II. Pearson test and linear regression were obtained, considering performance (grades) on EXANI-II (Independent variable), basic sciences and clinical sciences (Dependent variables).

RESULTS: A total of 94 students were included. EXANI-II had a medium Pearson correlation, both with basic (r= 0.453, p<0.001) and clinical sciences(r=0.367, p<0.001), accounting for 20.6% and 13.5% of the variance. Among specific areas of EXANI-II, Nature sciences was the only one who showed medium correlation both with basic and clinical sciences (r=0.385, p< 0.001, and r=0.313, p=0.002). English showed small correlation only with clinical sciences (r= 0.267, p=0.009). Other areas evaluated attained r values above 0.10, but p values were not significant. Interpretation of the magnitude of the effect size for linear correlations is based on Cohen's suggestions that an r of 0.10 be considered small, an r of 0.30 as medium, and an r of 0.50 as large.

CONCLUSIONS: EXANI-II as a whole has a medium correlation with academic performance both in basic and clinical sciences. This data is similar to those obtained in studies in Canada and Spain. When specific areas of EXANI-II are considered, only Nature sciences and English correlate significantly with academic performance. A higher correlation with Nature sciences has also been described for the Medical College Admission Test. EXANI-II appears to be a reasonably good predictor for academic performance in our context.