Videotape and simulated patients were used for evaluating third year medical students' performances in solving familiar and unfamiliar clinical problems. To assist the doctor to rate the students' performances by the review of videotapes, criteria were set up for students' behaviours that could be construed as evidence that the students recognized cues obtained by interviewing the patient and the information content of such cues. Criteria for data use by the students were also set. A mean of 90% of the pertinent medical history data was collected by the students. The amount of pertinent historical data collected was found to correlate very closely with student-patient interaction analysis score for the familiar and unfamiliar problems (r = 0.89 and r = 0.92, respectively). Significant differences in mean scores for data generation (P less than 0.001) and for data use (P less than 0.01) were found between those students who solved and those who failed to solve the familiar problem. With the unfamiliar problem a significant difference was found between the mean scores for data generation (P less than 0.01), but not for data use. The implications of these findings in medical education are discussed.
