The Impact of Virtual Laboratories on Student Motivation and Academic Performance: An Integrated Fuzzy-Sem and Machine Learning Study

Abstract

This study explored the impact of virtual laboratories (VLs) on university learning and seeks to fill a gap in the literature: most VL research reports positive outcomes, but rarely explains why they occur or whether psychological mechanisms generalize predictively. The solution comes from a synthetic model combining Fuzzy-SEM, which is great for modelling uncertainty within Likert-based motivation and engagement constructs, with supervised machine learning models that provide causal explanation combined with predictive validation. We analyzed data from 432 undergraduates combining VL usage logs, motivation–engagement surveys, and official academic records. Fuzzy-SEM confirmed a mediated motivation–engagement–performance pathway, which confirms that VLs significantly boost performance primarily by converting motivational activation to sustained engagement. Predictively, the 1D CNN better fitted the classical ML models (AUC-ROC = 0.94) suggesting the possibility of early identification of at-risk students through behavioural and affective proxies. Practical implications should be to apply VLs as complementary motivational approaches to training practice and to monitor prediction weekly for intervention. In theory, the study bolsters engagement frameworks by elucidating how VLs exert their effect. Methodologically, it presents an integrated Fuzzy-SEM + ML pipeline that facilitates both explanatory context and potentially deployable prediction, although it recognizes the limitation of single-institution and self-report.