Chinese University Students' Continuance Intention to Use AI Chatbot Learning Support Tools: An Integrated Analysis of the Theory of Planned Behavior, AI Trust, and AI Anxiety
Year: 2026
Type: article
Study design: Observational Study
Abstract: This study examined Chinese university students’ continuance intention to use conversational AI learning support tools by integrating the Theory of Planned Behavior, AI trust, and AI anxiety. Its objectives were to examine (1) associations of attitude, subjective norm, and perceived behavioral control with continuance intention; (2) AI trust’s direct and statistical indirect relationships through attitude; and (3) AI anxiety’s relationships with attitude, perceived behavioral control, and continuance intention, including their stability across model specifications. We yielded 428 complete responses from experienced student users and measured six constructs using 23 items. Analyses included descriptive statistics, reliability tests, Pearson correlations, HTMT diagnostics, hierarchical regression, and indirect-effect analysis using 5,000 bootstrap samples and bias-corrected 95% confidence intervals. Primary models controlled for educational level, discipline, usage frequency, and usage experience. The results uncover that attitude, subjective norm, and perceived behavioral control were positively associated with continuance intention. AI trust was positively associated with attitude and continuance intention, with a statistical indirect effect through attitude of 0.098 (95% CI [0.022, 0.181]). The proposed negative anxiety mechanism lacked robust support: positive zero-order correlations became weakly negative under some controls. Its indirect relationship through attitude was model-sensitive, while that through perceived behavioral control included zero. Cross-sectional data and limited discriminant validity restrict these findings to conditional associations, without establishing causality or temporal mediation. Substantial overlap among the positive constructs was evident. The study highlights construct overlap and model sensitivity in AI continuance research. Universities may support responsible use by strengthening students’ controllable use capabilities, clarifying norms, promoting calibrated trust, and encouraging output verification and disclosure. Future research should refine measurement and use independent samples, confirmatory factor analysis, and longitudinal designs.
Authors: Yi Xiaoyu, Songyu Jiang
Institution: Rajamangala University of Technology Rattanakosin
Language: English
Subfield: Information Systems and Management
Field: Decision Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7225338973