RESUMO
Background: The study of predictors of academic performance is relevant for medical education. Most studies of academic performance use global ratings as outcome measure, and do not evaluate the influence of the assessment methods. Aim: To model by multivariate analysis, the academic performance of medical considering, besides academic and demographic variables, the methods used to assess students' learning and their preferred modes of information processing. Material and methods: Two hundred seventy two students admitted to the medical school of the Pontificia Universidad Católica de Chile from 2000 to 2003. Six groups of variables were studied to model the students' performance in five basic science courses (Anatomy, Biology, Calculus, Chemistry and Physics) and two pre-clinical courses (Integrated Medical Clinic I and IT). The assessment methods examined were multiple choice question tests, Objective Structured Clinical Examination and tutor appraisal. Results: The results of the university admission tests (high school grades, mathematics and biology tests), the assessment methods used, the curricular year and previous application to medical school, were predictors of academic performance. The information processing modes influenced academic performance, but only in interaction with other variables. Perception (abstract or concrete) interacted with the assessment methods, and information use (active or reflexive), with sex. The correlation between the real and predicted grades was 0.7. Conclusions: In addition to the academic results obtained prior to university entrance, the methods of assessment used in the university and the information processing modes influence the academic performance of medical students in basic and preclinical courses.