Motivation, engagement, and perceived competence gains in competition-based medical education: a theory-informed structural equation modeling study

Abstract

Competition-based learning is increasingly used in medical education to promote active participation, innovation, and collaborative problem solving. However, less is known about how students’ participation motivation, engagement, perceived competence gains, and satisfaction are related within such learning environments. This study examined a theory-informed model of these associations in a national medical student innovation competition. This secondary cross-sectional study analyzed post-competition survey data from 1,162 medical students who participated in the 2025 National College Student Medical Innovation Competition in China. Four latent constructs were examined: participation motivation, engagement, perceived competence gains, and satisfaction. Confirmatory factor analysis and structural equation modeling were conducted using maximum likelihood estimation. Discriminant validity was evaluated using latent correlations, the Fornell–Larcker criterion, heterotrait–monotrait ratios, and comparison with an alternative model combining engagement and perceived competence gains. Indirect associations were examined using bias-corrected bootstrapping with 5,000 resamples. Robust maximum likelihood and weighted least squares mean- and variance-adjusted sensitivity analyses were also conducted. The four-factor structural model showed acceptable fit: χ 2 (165) = 1155.123, comparative fit index = 0.949, Tucker–Lewis index = 0.941, root mean square error of approximation = 0.072, and root mean square residual = 0.031. Participation motivation was positively associated with engagement (β = 0.657), engagement was strongly associated with perceived competence gains (β = 0.914), and perceived competence gains (β = 0.234) and participation motivation (β = 0.381) were positively associated with satisfaction (all p < .001). The engagement–perceived competence gains latent correlation was high ( r = 0.909), the Fornell–Larcker criterion was not met, and the heterotrait–monotrait estimate was 0.913, indicating limited discriminant validity. Nevertheless, the four-factor model fitted significantly better than the model combining engagement and perceived competence gains, Δχ 2 (3) = 543.909, p < .001. Sensitivity analyses showed a similar pattern, although ordinal-model fit was mixed. Participation motivation, engagement, perceived competence gains, and satisfaction were closely associated within this competition-based learning environment. Comparative model testing supported retaining engagement and perceived competence gains as separate factors, but the constructs substantially overlapped and evidence of discriminant validity was limited. Because the data were cross-sectional and self-reported, the findings should be interpreted as model-consistent associations rather than causal or temporal relationships.