Interrater Reliability in the American Board of Physical Medicine and Rehabilitation Part II Certification Examination: Impact of a New Assessment Design.

Abstract

OBJECTIVE\nThe design of medical board certification examinations continues to evolve with advances in testing innovations and psychometric analysis. The potential for subjectivity is inherent in the design of oral board examinations, making improvements in reliability and validity especially important. The purpose of this quality improvement study is to analyze the impact of using two examiners on the overall reliability of the oral certification examination in Physical Medicine and Rehabilitation (ABPMR).\n\n\nDESIGN\nRetrospective quality improvement study of 422 candidates for the ABPMR Part II Examination in 2020. Candidates were examined by examiner pairs, each of whom submitted independent scores. Training for all 116 examiners included examination case review, scoring guidelines, and bias mitigation. Examiner performance was analyzed for both internal consistency (intrarater reliability) and agreement with their paired examiner (interrater reliability).\n\n\nRESULTS\nThe reliability of the Part II Examination was high, ranging from 0.93 to 0.94 over three administrations. The analysis also demonstrated high interrater agreement and examiner internal consistency.\n\n\nCONCLUSIONS\nA high degree of interrater agreement was found using a new, two-examiner format. Comprehensive examiner training is likely the most significant factor for this finding. The two-examiner format improved the overall reliability and validity of the Part II Examination.