Purpose: To test the hypothesis that using explicit behavioral criteria for evaluation can improve the accuracy of students' self-evaluation in problem-based learning (PBL). Methods: Second to fourth-year medical students of a seven-year curriculum were studied. PBL performance of students was assessed by students using a student version of the Tutotest (S-Tutotest(superscript C)) for self-evaluation, and by facilitating tutors using the tutor version of the Tutotest (T-Tutotest(superscript C)) and a global rating score for evaluation and feedback. A multiple-choice achievement test was taken by students at the end of the PBL course. Results: 315 students' evaluations were obtained with a 78.9% completion rate. Students' self-ratings were poorly correlated to tutors' ratings. Tutors' Tutotest ratings were moderately correlated to their global ratings. Students' achievement test scores were poorly correlated to tutors' Tutotest ratings and global ratings, and uncorrelated with students' self-ratings. Students of poor achievers tended to overestimate and students of good achievers tended to underestimate their performance. Conclusions: Our study did not fully support Arthur's argument that explicit criteria of performance are important for the validity improvement of self-evaluation. Results of student evaluation varied with methods of evaluation. Explicit behavior criteria are useful for average grade students to use self-evaluation for PBL learning but not so for students with poor or good academic achievement.
Purpose: To test the hypothesis that using explicit behavioral criteria for evaluation can improve the accuracy of students' self-evaluation in problem-based learning (PBL). Methods: Second to fourth-year medical students of a seven-year curriculum were studied. PBL performance of students was assessed by students using a student version of the Tutotest (S-Tutotest(superscript C)) for self-evaluation, and by facilitating tutors using the tutor version of the Tutotest (T-Tutotest(superscript C)) and a global rating score for evaluation and feedback. A multiple-choice achievement test was taken by students at the end of the PBL course. Results: 315 students' evaluations were obtained with a 78.9% completion rate. Students' self-ratings were poorly correlated to tutors' ratings. Tutors' Tutotest ratings were moderately correlated to their global ratings. Students' achievement test scores were poorly correlated to tutors' Tutotest ratings and global ratings, and uncorrelated with students' self-ratings. Students of poor achievers tended to overestimate and students of good achievers tended to underestimate their performance. Conclusions: Our study did not fully support Arthur's argument that explicit criteria of performance are important for the validity improvement of self-evaluation. Results of student evaluation varied with methods of evaluation. Explicit behavior criteria are useful for average grade students to use self-evaluation for PBL learning but not so for students with poor or good academic achievement.