Purpose: To examine the relative efficacies of three types of performance feedback used with computer-based instruction (CBI) in making diagnoses.
Method: In 1990, 75 final-year medical students at The University of Queensland were pretested for domain knowledge and diagnostic skill in the area of acute abdominal pain. The students were also asked to indicate their confidence in each of their diagnoses. Following these pretests, the students were randomly divided into five groups of 15 students each. One group received traditional CBI, using a "question-and-explanation" format, to learn domain knowledge. The other four groups received one of two types of CBI to learn diagnostic accuracy, and these four groups received one of three types of performance feedback (which differed considerably in the amounts of information imparted). Then the students took posttests. One-way analysis of variance, Student's t-test, and the Tukey test were used to compare the performances of the different groups of students.
Results: Although the students in the traditional CBI group significantly improved their performance on multiple-choice questions (by 58%) compared with the other groups of students (6-10% improvement), these students did not improve their diagnostic performance. In contrast, the CBI groups that used simulated-patient cases and structured-performance feedback did improve their diagnostic accuracy (by as much as 16%, compared with a 1% decline for the traditional CBI group). Contrary to expectation, the different types of feedback were equally effective. Except for the students in the traditional CBI group, the students' confidence increased from pretest to posttest regardless of their performances.
Conclusion: CBI that uses simulated-patient cases and structured-performance feedback seems to be efficient and effective compared with traditional CBI methods. However, care should be taken to ensure that students do not become overconfident of their abilities.