Beyond confirmatory approaches: integrating quasi-experimental design, clustering and machine learning for metaverse adoption profiling

Purpose This study aims to propose a hybrid methodological framework that integrates quasi-experimental design, unsupervised clustering and supervised machine learning algorithms to identify distinct adoption profiles in metaverse contexts, advancing beyond traditional confirmatory statistical approaches. Design/methodology/approach A quasi-experimental study involving 310 undergraduates who were exposed to three metaverse applications on Meta Quest devices was conducted. K-means clustering, based on seven constructs, identified segments that were validated using random forest and support vector machine, supported by factor analysis and comparative tests. Findings Two adoption profiles emerged: Skeptics/pragmatists (38.1%) and enthusiasts (61.9%). Perceived value, perceived usefulness and trust were the strongest predictors, with Social Influence and Enjoyment as supporting factors. Both models showed excellent predictive performance (AUC-ROC = 1.000; 0.994) and large effect sizes, confirming the validity of the segments. Research limitations/implications This study is limited to a single-university sample, thereby restricting generalizability across diverse organizational contexts. Future research should replicate this methodology across various cultural and organizational settings to validate its generalizability. Practical implications The results offer actionable insights for leaders and educators on tailoring immersive experiences by emphasizing perceived usefulness, trust and value creation to enhance engagement and participation in Metaverse environments. Social implications This study highlights the opportunities and risks of immersive technologies, emphasizing gamification and attention management mechanisms to reduce cognitive distraction and enhance focus in educational and organizational contexts. Originality/value This study combines quasi-experimental design, unsupervised clustering and supervised machine learning as an alternative to confirmatory-only adoption models. This study advances metaverse adoption research by linking segment-based evidence to behavioral interpretation grounded in usefulness, value, trust, presence and immersive engagement.