Associations of sense of belonging and academic self-efficacy with college persistence through student engagement: MASEM with moderation by learning environment.

Previous research, including meta-analytic evidence, has established positive associations among sense of belonging, academic self-efficacy, engagement, and persistence. However, it remains unclear how these factors interact through student engagement and whether these associations differ across learning environments. This study applied meta-analytic structural equation modelling (MASEM) to synthesize findings from 20 independent studies and examine the direct and indirect associations among these variables. A partial mediation model was initially tested, in which sense of belonging and academic self-efficacy were specified as both directly and indirectly associated with persistence through engagement. Although the initial partial mediation model demonstrated acceptable fit, the direct association between academic self-efficacy and persistence was not statistically significant. Following this analysis, a revised partial mediation model was examined as a post-hoc exploratory specification and showed favorable fit indices, with both academic self-efficacy and sense of belonging significantly associated with engagement, which, in turn, was positively associated with persistence. Notably, sense of belonging also retained a significant direct association with persistence, whereas academic self-efficacy was related to persistence primarily through its association with engagement. A full mediation model was also examined and showed less favorable fit indices. Moderator analyses indicated that the associations involving sense of belonging varied across learning environments, whereas the other examined paths did not significantly differ by learning environment. However, sensitivity analysis indicated limited stability for the sense of belonging-engagement pathway in the online subgroup. Overall, the findings highlight the central role of engagement in understanding student persistence. Given the predominantly observational nature of the included studies, the indirect effects should be interpreted as statistical associations rather than evidence of causal mediation.