Can We Share Multiple Choice Questions across Borders? Validation of the Dutch Knowledge Assessment in Family Medicine in Flanders

Abstract

Background. One of the methods to test knowledge of Family Medicine trainees is a written exam composed of multiple choice questions. Creating high-quality multiple choice questions requires a lot of experience, knowledge, and time. This study explores the opportunity to run the Dutch knowledge assessment in Flanders as well, the use of this test for formative purposes. Methods. The study test was performed in a Flemish sample of postgraduate Family Medicine (FM) trainees and FM trainers. The Dutch test, adjusted to the Flemish context, was analyzed according to the classical test theory: difficulty factor and discriminating power of the items and reliability of the test. Results. 82 of the 154 items well divided the group into two equal parts of correct and incorrect responders. The distribution of the discrimination index, of the items with an acceptable difficulty factor, was [−0.012–0.530]. The item-test-correlation shows that 52 items do not fit, and 87 items need revision in varying degrees. The test reliability was 0.917. Conclusion. The test was highly reliable, but many MC questions appeared to be too easy and poorly discriminative. Therefore, we question the test validity and recommend reconsideration of the items based on difficulty before it is applied and used as a mandatory formative test.