Profiling students for remediation using latent class analysis

While clinical exams using SPs are used extensively across the medical schools for summative purposes and high-stakes decisions, the method of identifying students for remediation varies widely and there is a lack of consensus on the best methodological approach. The purpose of this study is to provide an alternative approach to identification of students for remediation using the latent class analysis (LCA) technique. 147 third year medical students participating in the Clinical Performance Examination (CPX) are included in the study. We used LCA to identify students who potentially need remediation based on their performance on CPX. Three distinct clusters of students with different performance profiles were identified. The identification of two rather than one low performing group has significant implications for identifying cut-points as well as for remediation programs. The two low performing groups in our study had low scores on contrasting sets of cases. LCA presents an alternative approach to identification of borderline or low performing groups. This method provides advantages over traditional statistical techniques such as cluster analysis used for grouping students. Based on the flexibility of the model specification, within the LCA framework, we were able to identify more than one group that may need remediation or instruction support.