CEMEDEPU-Grade-AGI Dataset — Preclinical Alignment Study

This dataset supports the first convergent parallel mixed-methods empirical test of the assessment filter effect — a structural condition in which summative examination architecture intercepts and neutralises pedagogical innovation before it can manifest as measurable academic outcomes — in a preclinical medical curriculum in Latin America. It encompasses all quantitative and mixed-methods data generated during the study, structured to enable independent replication of the Alignment Gap Index (AGI) diagnostic instrument, verification of the inter-rater reliability statistics reported in the manuscript, and adaptation of the assessment flexibility rating procedure for use in analogous preclinical settings.
The dataset is a companion to, not a substitute for, the qualitative codebook deposited separately as Online Resource 1 (Codebook_S1.docx). Together, the two deposits constitute the complete open data package for the study. The qualitative codebook (Codebook_S1.docx) contains all methodological documentation for the qualitative and mixed-methods strands: operational definitions, inclusion and exclusion criteria, anchor quotations, inter-rater reliability matrices, assessment flexibility rating sheets, and the prospective analytical audit trail. The present dataset description documents the data files themselves: their formats, variable structures, values, and relationships.
What this dataset enables. (1) Independent replication of all descriptive statistics, the Pearson correlation, and the AGI values reported in the manuscript. (2) Verification of inter-rater reliability statistics (κ = 0.79; κ = 0.73; κw = 0.74) against the raw per-analyst coding records. (3) Adaptation of the Assessment Flexibility Rating Scale for use in other preclinical curricula. (4) Meta-analytic use of course-level AGI scores as a benchmark for future cross-institutional comparisons.