When Fit Meets Risk: Survey Data and Codebook on Task–Technology Fit, Trust Calibration, Perceived Risk and Reliance on Domestic Generative AI Assistants in China

This dataset accompanies the research manuscript entitled “When Fit Meets Risk: Task–Technology Fit, Self-Perceived Trust Calibration and Reliance on Domestic Generative AI Assistants in China.” The research investigates how task–technology fit, perceived risk, self-perceived trust calibration, AI literacy, and prior experience relate to reliance intentions among users of domestic generative artificial intelligence assistants in China. The dataset originates from a quantitative cross-sectional survey. The source study reports 501 collected responses, of which 490 were retained for analysis after applying documented eligibility and response-quality exclusions. The deposited files provide cleaned survey responses, variable definitions, data-processing documentation, and a record of excluded observations. The measurement framework covers output quality fit, functional and adaptability fit, interaction fit, timeliness fit, performance risk, privacy risk, misinformation/hallucination risk, trust calibration, AI reliance intention, AI literacy, and prior experience. The dataset includes item-level survey responses, coded respondent characteristics, and calculated composite variables. Questionnaire items are measured on seven-point Likert scales where documented in the codebook. The analytical treatment of trust calibration excludes TC4 from its composite score while retaining the original item in the data for transparency. The dataset is intended to support independent examination of the research measures, verification of the reported empirical analyses, and reproducibility of the associated study, subject to applicable confidentiality and data-access conditions. The accompanying codebook and processing documentation provide information about variable definitions, response coding, study eligibility rules, and dataset preparation. Research design: Quantitative cross-sectional survey. Population: Users of domestic generative AI assistants in China. Reported analytical sample: 490 respondents. Reported collection date: 18 September 2026. Institution: School of Management Engineering, Xuzhou University of Technology, China. Ethics approval: BXD202609170053, School of Management Engineering, Xuzhou University of Technology. Associated manuscript: When Fit Meets Risk: Task–Technology Fit, Self-Perceived Trust Calibration and Reliance on Domestic Generative AI Assistants in China.