Comprehensive assessment of students' academic performance plays an important role in educational planning. The aim of this study was to investigate variables that influence student's performance in a retrospective sample including all undergraduate students who entered in a Brazilian dental school, in a 20-year period between 1984 and 2003 (n=1182). Demographic and educational variables were used to predict performance in the overall curriculum and course groups. Cluster analysis (K-means algorithm) categorized students into groups of higher, moderate or lower performance. Clusters of overall performance showed external validity, demonstrated by Chi-square test and ANOVA. Lower performance groups had the smallest number of students in overall performance and course groups clusters, ranging from 11.8% (clinical courses) to 19.2% (basic courses). Students' performance was more satisfactory in dental and clinical courses, rather than basic and non-clinical courses (p<0.001). Better student's performance was predicted by lower time elapsed between completion of high school and dental school admission, female gender, better rank in admission test, class attendance rate and student workload hours in teaching, research and extension (R²=0.491). Findings give evidence about predictors of undergraduate students' performance and reinforce the need for curricular reformulation focused on with improvement of integration among courses.
dental education; dental students; curriculum; cluster analysis
