Effects of posting examinations and answer keys on students' study methods and test performance.

A rich database was assembled to test the validity of contradictory claims about the effects of policy regarding posting of code-protected individual student scores and keyed examination questions. Four second-year lecture courses, each in a specialty area and each testing using multiple-choice examinations at the end of quarters, were compared. Two of the courses follow the practice of regularly posting scores and keyed questions; one holds review sessions during subsequent quarters to discuss some questions, while the other allows students to review their tests in a proctored room without access to paper and pencil. Students completed a questionnaire reporting the hours they spent studying for each test using a variety of methods, including old examinations, and estimates of the proportion of repeat questions on each test. Because most of the questionnaires were signed, it was possible to match self-reported study patterns with actual test scores and with overall GPA. Other information, such as test reliability, grading patterns, and National Board performance, was also gathered. A focus group was conducted. Results showed that students customized their study patterns and there was a minor clustering of practices such as late start for studying and use of old exams. Students with higher overall GPAs and those intending to specialize scored higher on tests. There were no associations between hours spent studying using any approach, including old exams, and test performance. There were no differences between course characteristics of the two courses that posted and the two courses that did not post exams, including National Board performance. Potential reasons why there was no association between hours or type of examination preparation and test scores are discussed. The focus group revealed that there are multiple goals among students regarding what they hope to achieve in the testing situation and that perfect performance was not a goal. Because there was no direct effect of posting exams on student test performance and because the practice of not posting exams created a climate of distance between students and faculty members, it is suggested that exams be posted as a regular policy.