Statistical methods for the assessment of students' knowledge as illustrated by the Introduction to Internal Medicine exam

Abstract

Background We wanted to develop substantial and statistical methodology for complex assessment of quality of teaching internal medicine in medical university. Our aim was also to check connection between the results obtained during the midterm and final exam.

Materials and methods We have compared the results obtained by Polish (n=235) and English Divisions (n=81) students achieved during the midterm exam and multiple-choice final exam. The mean scores were calculated with t-Student test. For further evaluation Wilcoxon tests were used with the Bonferroni correction, The Stuart-Maxwell test was carried out to verify the hypothesis about correlations be- tween results in the midterm and final exam.

Results The mean midterm exam score was 84.4% in PD and 72.6% in ED (p <0.0001) and mean final exam score was respectively 72.3% and 55.6% (p <0.0001). Good result of the final exam was obtained by 62% of students who passed well the midterm exam.

Conclusions It is crucial to use appropriate tools to grade the quality of tutorship. To evaluate that one should use advance statistical tests. The fact that ED students achieve less points on the exams might have few reasons like a language barrier. Obtaining a good result during midterm exam does not guarantee passing the final exam.