Script concordance test issues, the trail of expert calibration

Abstract

Abstract Background The Script Concordance Test (SCT) is an assessment tool for clinical reasoning that incorporates uncertainty and depends on expert judgment to identify valid responses. Accurate calibration of expert judgment is important for maintaining validity and reliability; however, the literature has rarely addressed this issue and only through statistical methods. This study aimed to compare calibration strategies using statistical moderation and qualitative inspection. Methods Sixteen experts (n = 16) were recruited to complete 21 clinical vignettes, providing justification for each response. Seven calibration strategies—quantitative, qualitative, and mixed—were then analyzed using the Rasch Facet Model, with particular attention to expert homogeneity, data–model fit, and the quality of expert responses. Results None of the strategies improved expert homogeneity. However, mixed strategies enhanced data–model fit and response quality, and helped address issues related to response process and content validity. Conclusions Calibrating expert judgment using a mixed strategy appears valuable for improving the quality of expert-generated data within an SCT framework. This calibration may address specific psychometric limitations of SCTs and enhance training quality through Learning by Concordance methods.