Psychometric validation of the Laval developmental benchmarks scale for family medicine

Abstract

Background With the implementation of competency-based education in family medicine, there is a need for summative end-of-rotation assessments that are criterion-referenced rather than normative. Laval University’s family residency program therefore developed the Laval Developmental Benchmarks Scale for Family Medicine (DBS-FM), based on competency milestones. This psychometric validation study investigates its internal structure and its relation to another variable, two sources of validity evidence. Methods We used assessment data from a cohort of residents (n = 1432 assessments) and the Rasch Rating Scale Model to investigate its reliability, dimensionality, rating scale functioning, targeting of items to residents’ competency levels, biases (differential item functioning), items hierarchy (adequacy of milestones ordering), and score responsiveness. Convergent validity was estimated by its correlation with the clinical rotation decision (pass, in difficulty/fail). Results The DBS-FM can be considered as a unidimensional scale with good reliability for non-extreme scores (.83). The correlation between expected and empirical items hierarchies was of .78, p < .0001.Year 2 residents achieved higher scores than year 1 residents. It was associated with the clinical rotation decision. Conclusion Advancing its validation, this study found that the DBS-FM has a sound internal structure and demonstrates convergent validity.