Anthropomorphism’s impact on chatbot adoption: A meta-analytic structural equation modeling approach

Abstract

Customers’ attitudes toward chatbots become crucial in the context of customer service frontlines transitioning from human-human interaction to human-robot interaction. However, there is still significant inconsistency regarding the impact of chatbot anthropomorphism on the user adoption process, which poses challenges for designers and managers. This meta-analysis examines 32 studies involving 12,502 participants and 196 effect sizes through the lens of the Stimulus-Organism-Response (SOR) theory, categorizing anthropomorphism in chatbots into linguistic and visual types. It investigates their influence on chatbot adoption through the mediating effect of cognitive and affective states. Using a meta-analytic structural equation model (MASEM), our study finds that the impact of anthropomorphism on adoption is fully mediated by cognitive and affective states, while the mediating effect of the affective states is greater than that of the cognitive states. Within the cognitive states, anthropomorphism shows significant positive correlations with cognitive attitude, cognitive trust, cognitive satisfaction, and perceived value, while exhibiting no significant relationship with risk. In the affective states, both forms of anthropomorphism have substantial positive influences on affective attitude, affective trust, social presence, and perceived warmth. Notably, only linguistic anthropomorphism enhances affective satisfaction. Furthermore, this study reveals how different sample features, chatbot characteristics, and usage industries moderate the effect of anthropomorphism on chatbot adoption. The findings provide a crucial foundation for future research and offer guidance for designing and deploying chatbots across various service settings.