Development and validity evidence for the probabilistic reasoning test in undergraduate medical students

Script Concordance Tests (SCT) are widely used to assess clinical reasoning in medical education, but their use in high-stakes examinations raises concerns related to ambiguity, reliance on expert panels, and limited acceptability among students. To address these issues, we developed the Probabilistic Reasoning Test (PRT), a simplified format designed to assess normative probabilistic reasoning within clinical scenarios using a single predetermined correct answer and a three-point response scale.

We conducted a study including fifth-year medical students, pediatric residents, and pediatricians. Participants completed matched triads of SCT, PRT, and multiple-choice questions (MCQs) based on identical clinical scenarios. Student Objective Structured Clinical Examination (OSCE) scores were collected. Outcomes included concordance between SCT and PRT responses, internal consistency, relations with other assessment modalities, score interpretation across experience levels, and participant acceptability.

A total of 575 participants were analyzed (538 students, 19 residents, 18 pediatricians). Concordance between SCT and PRT responses was high (agreement 91.4%, weighted κ = 0.91). Internal consistency was modest and similar for both formats (Cronbach's α = 0.63 for PRT; 0.62 for SCT). PRT scores correlated strongly with SCT, moderately with MCQs, and weakly with OSCE scores (p < 0.05). Unlike SCT, PRT did not significantly discriminate between levels of clinical experience. Satisfaction was consistently higher for PRT, with 90% of participants preferring it over SCT for examinations.

The PRT showed strong directional concordance with SCT and higher acceptability in undergraduate settings but limited differentiation across experience levels. It may represent a pragmatic candidate format for undergraduate summative assessment of normative probabilistic reasoning.