Predictive validity of academy mock tests against MIR exam: a retrospective multicohort observational concordance study

Resumo

Introduction: MIR examination is a requirement to access Spain’s public specialist training system. This study evaluates, across five years, the concordance between mock exams administered by a MIR preparatory academy and the official MIR results. Methods: Candidate records were linked; on the matched intersection we computed Pearson’s R and Spearman’s ρ (percentiles, net scores), bias (simulation−official), and RMSE. Annual key performance indicators (KPIs) were pair-weighted and analyses stratified by percentile cohorts (≤27, 28–73, >73); heatmap summarize percentile correlations. Results: Across years, weighted percentile correlations typically fell in the 0.71–0.76 range, with small average bias and moderate RMSE. Cohort analyses indicated larger errors at extreme percentiles in some years. Net‑score bias weighted by pairs was near zero overall, with year‑to‑year variability. Conclusions: Academy mock tests provide a broadly consistent ranking signal relative to official MIR outcomes. Residual bias and dispersion suggest potential gains from calibration—particularly at cohort extremes—and motivate routine monitoring of concordance KPIs in future cohorts.