To demonstrate the quality assurance of the comprehensive examination of sixth-year students at Nippon Medical School, 4 undergraduate examinations were compared with the national examination for medical practitioners (NEMP) using scatter graphs and Pearson's correlation coefficient. Of the 93 sixth-year students at Nippon Medical School, 57%(n=53) reported their scores on the NEMP in response to a request from the Academic Quality and Development Office. Correlation coefficients of the grade point average (years 1 to 5), average scores on graduation examinations of 24 subjects, scores on the trial examination of NEMP, and scores on the sixth-year comprehensive examination with overall scores on the NEMP were 0.62, 0.46, 0.68, and 0.63, respectively. These results suggest that the sixth-year comprehensive examination is more suitable than are graduation examinations for predicting the NEMP score.

To demonstrate the quality assurance of comprehensive examinations for fourth-year students at Nippon Medical School, scores on comprehensive examinations were compared with those on a trial of computer-based testing (CBT) of a nationwide medical and dental student evaluation system in 2003. Pearson's correlation coefficients between scores of two comprehensive examinations and the CBT score were 0.45 and 0.67, and the correlation coefficient between the average score of the two comprehensive examinations and the CBT score was 0.55. Fourth-year comprehensive examinations are useful tools for summative evaluation and prediction of CBT performance.

The Ministry of Health and Welfare of Japan is planning a pooling system for multiple-choice questions (MCQs) for the national examination for medical practitioners. To clarify possible problems of such a system, a field study was performed by 10 medical schools in Japan using 90 MCQs from previous examinations. Nine hundred twenty-four 6th-year students participated in the field test. For each MCQ, the correct-response rates at the originating school and those obtained in the field test were significantly correlated. Thus, the correct-response rates to questions on the field test could be predicted from the rates at the originating schools. However, for each question the correct-response rate was significantly higher for students of the originating school than for students of other schools. In the national examination, care should be taken to prevent differences in scores on the basis of question sources.