The development and validation of a multi-dimensional medical students’ learning self-efficacy questionnaire for clinical education

Learning self-efficacy (SE) assesses how learners understand and evaluate their ability to polish their learning process. Learning clinical medicine requires prolonged training, traditionally premised on longitudinal immersion in patient care. Such a process is domain-specific, whereas learning SE for clinical education remains under-explored. Unidimensional assessment is insufficient for capturing the inherent capabilities upon which well-trained physicians provide care. We aimed to establish a multi-dimensional learning SE questionnaire for clinical education among undergraduate medical students, evaluating the structure validity, followed by assessing the dimensionality of different models. Medical students of 2^nd to 4^th grades from Taiwan in 2022-2023 completed a multi-dimensional medical learning SE (MLSE) questionnaire, including four factors for basic science learning (conceptual understanding (CU), higher-order cognitive skills (HC), practical work (PW), and everyday application (EA)), and three for clinical mastery performance (medical communication (MC), evidence-based medicine (EBM), and Professionalism)). We tested factors' intercorrelation, used exploratory and confirmatory factor analysis (EFA/CFA) for structure and validity assessment, and compared the fitness and dimensionality between models. Twenty-four items grouped into seven independent factors (3, 3, 4, 3, 5, 3, and 3 items in CU, HC, PW, EA, MC, EBM, and Professionalism, respectively) were established and finalized, with sufficient fitness, good convergent and construct validities. All MLSE factors significantly correlated (0.49-0.87; <i>p</i> < 0.001), demonstrating good convergent and discriminant validity. We established six models (first-order uncorrelated or correlated construct, one to three second-order dimensions ('basic medical SE', 'clinical medical SE', 'Cognition', or 'Application' of different structures), and a final model 7 containing four second-order dimensions (Cognition, Application, MC, and clinical medical SE) exhibiting adequate model fitness and measured learning SE satisfactorily. Our MLSE model structure disclosed vital SE factors with intercorrelations associated with medical students' learning processes during clinical education. Polishing these dimensions may help promote their learning SE.