GMS Journal for Medical Education (Apr 2020)
Predictive validity of a tool to resolve borderline grades in OSCEs
Abstract
There is inconclusive evidence suggesting which standard setting method yields the highest validity for pass/fail decisions in examinations. The Objective Borderline Method 2 (OBM2) is a decision-making tool for reclassification of borderline grades to clear pass or clear fail grades to resolve examiner uncertainty for high-stakes pass/fail decisions. This study evaluated the predictive validity of OBM2 pass/fail decisions, using consecutive years’ Objective Structured Clinical Examination (OSCE) results within a medical cohort (n=271) at the University of New South Wales, Australia. OBM2 decisions in one OSCE (n=687) were compared to marks obtained in a subsequent OSCE via independent samples T-tests and analysis of variance (ANOVA). The extent of the relationship between these two variables determines the predictive validity of OBM2 decisions, given that past student grades are capable of predicting future performance. OBM2 decisions in an initial OSCE were found to have a statistically significant predictive nature for subsequent OSCE marks (p=.005). For initial decisions which reclassified to a pass grade, subsequent OSCE marks were significantly higher than for the cases where initial decisions were reclassified to a fail grade. Stronger associations were identified between related assessment domains/criteria compared to unrelated domains/criteria (Cohen’s d=.469 vs Cohen’s d=.388 respectively). Through demonstrating the OBM2 decisions’ predictive association across exams there is support for the OBM2’s predictive validity, deeming it a promising method to be used for resolving examiner uncertainty when making pass/fail decisions within OSCEs.
Keywords