Comparing the borderline group and borderline regression approaches to setting Objective Structured Clinical Examination cut scores

Abstract

Katharine Jane Reid, Agnes Dodds

The aim of the study was to compare the Borderline Group and Borderline Regression approaches to setting standards on nine Objective Structured Clinical Examinations (OSCEs) undertaken in the final year of medical school. For each of nine OSCEs completed by 301 students we obtained a total score (maximum 40 marks) and a global score on a 5-point rating scale which examiners completed after scoring the OSCE using a checklist. We calculated cut scores for each OSCE using the Borderline Group method by computing the mean and the median score of the group of students whose performance was rated by examiners rated as borderline. We calculated cut scores for each OSCE using the Borderline Regression method by predicting total OSCE scores from global ratings using linear regression, and calculating the cut score by substituting the score of borderline candidates (2) into the regression equation for each OSCE. Both methods established higher standards than an arbitrary 50 per cent criterion. There was also a high degree of concordance between methods. Establishing conceptual standards on OSCEs requires ongoing consideration of appropriate methods and research modelling the effects on cut scores of applying different techniques.