The influence of e-commerce virtual streamer characteristics on consumers' purchase intention: Based on a hybrid SEM-ANN approach

With the advancement of artificial intelligence technology, virtual streamers have gained significant traction in live streaming e-commerce. Despite their increasing significance, the influence of virtual streamer characteristics on consumers' purchase intention remains underexplored in the extant literature. Based on the SOR theory, this study explores how e-commerce virtual streamer characteristics influence consumers' purchase intention through the mediating roles of flow experience and virtual attachment. This study conducted a questionnaire survey of 327 consumers who had watched live streaming of e-commerce virtual streamers and used partial least squares structural equation modeling (PLS-SEM) for empirical analysis. The results show that the anthropomorphism, expertise, and responsiveness of e-commerce virtual streamers have a positive impact on consumers' flow experience and virtual attachment, thereby enhancing their purchase intention. Flow experience and virtual attachment play parallel and chain mediating roles. Artificial neural network (ANN) analysis further validated the robustness of the research findings and revealed that the responsiveness of e-commerce virtual streamers is a key factor in consumers' flow experience, which in turn significantly promotes the formation of virtual attachment. Additionally, the introduction of the Shapley Additive Explanations (SHAP) model enhances the interpretability of the ANN analysis. Furthermore, virtual attachment plays a crucial role in consumers' purchase intention. This study not only provides a new theoretical perspective for exploring the mechanism by which virtual streamers in e-commerce influence consumers' purchase intention, but also offers empirical evidence for e-commerce platforms to optimize the design of virtual streamers and improve their marketing strategies.