Does candidate race influence simulated patient ratings in standardised assessments of clinical practice? A single-blinded randomised study in UK medical schools.

Abstract

OBJECTIVES: Standardisation of medical examinations involves minimising assessor stereotyping and bias for a fair process. This study aimed to determine whether being a non-white candidate affected scoring by simulated patients, compared with a white candidate, at three different performance grades in the same history-taking station. DESIGN: Single-blinded, video-based, randomised study. PARTICIPANTS: 163 simulated patients watched a randomly allocated set of six videos. Each set consisted of three white and three non-white (South Asian, black and Chinese) candidates performing at either fail, borderline or pass grades. Therefore, each simulated patient assessor observed one white and one non-white candidate at each grade and scored communication and professionalism domains. MAIN OUTCOME MEASURE: The median and interquartile range of the difference between total scores for the white and non-white candidates were compared at all three performance grades. RESULTS: The black fail candidate scored statistically significantly lower than their white fail counterpart. The black borderline and Chinese borderline candidates scored significantly higher than their white counterparts. No other differences were statistically significant at p<0.0057. CONCLUSIONS: Being a black candidate at the fail level of performance was associated with a lower score than being a white candidate at the fail level of performance, thereby indicating a negative stereotype against black students. However, being black or Chinese at a borderline grade was associated with higher scores than being white candidate at the same grade potentially due to self-awareness of potential bias when there is uncertainty regarding the performance.