Psychometric properties of the Dundee Ready Educational Environment Measure (DREEM) applied to medical residents

Abstract

The Dundee Ready Educational Environment Measure (DREEM) is a measure of students' perceptions of the educational environment, but its performance in evaluating the educational environment in the residency setting has not yet been described. This study aimed at describing the psychometric performance of DREEM applied to medical residents. DREEM was applied to 97 residents from 12 training programs on four specialties in six institutions in three Brazilian cities. Psychometric measures included factor analysis, Cronbach's alpha coefficients, item-to-total correlations, t-test comparisons of scores between genders, institutions, specialties, and programs, correlations with the global score of the Quality of School Life Scale (concurrent validity), and test-retest reliability. Generalizability theory procedures were applied to a random subset of data. Programs (8) were the objects of measure, while institutions (6), specialties (4), raters-within-programs (40), and items-on-the-scale (50) were facets. Variance components, generalizability (G) and dependability (D) coefficients were calculated. Cronbach's alpha was 0.93. DREEM showed high discriminant and concurrent validities. Test-retest reliability was moderate. Interactions between programs, raters and items accounted for 68% of the total variance. G and D coefficients were 0.95 and 0.67, respectively. The instrument proved to be useful for relative comparisons at both resident and program level.