Building Digital Therapeutic Alliance with Smoking Cessation Chatbots: An Experimental Examination of Communication Style and User Personality

Digital Therapeutic Alliance (DTA), the collaborative bond between users and therapeutic AI tools, may determine whether people engage with health chatbots, yet factors shaping this alliance remain unclear. We examined whether autonomy-supportive versus controlling communication styles influence DTA formation and whether personality traits moderate this relationship. In a preregistered experiment, 750 daily smokers viewed vignettes of a smoking-cessation chatbot using either communication style. Structural equation modeling revealed that communication style did not significantly affect DTA, despite successful manipulation checks. However, DTA strongly predicted chatbot usage intentions. Among personality traits, only agreeableness predicted DTA; other Big Five dimensions and style-by-personality interactions were non-significant. Technological innovativeness and baseline quit intention also predicted usage intentions. These findings highlight that relational dynamics matter for chatbot acceptance, but subtle conversational variations alone may not shape alliance formation. We discuss implications for human-centered chatbot design and future research directions.