Validation of the AI literacy questionnaire for Chinese pre-service teachers: psychometric evidence and profiles for differentiated educational evaluation

This study translated and adapted the AI Literacy Questionnaire for Chinese pre-service teachers, grounded in Expectancy-Value Theory as a pre-specified theoretical framework. Examining its psychometric properties in a sample of 341 pre-service teachers (78% female; 86.5% third-year students) from Guangdong Province, exploratory and confirmatory factor analyses supported a 24-item, five-factor structure: AI Ethics, AI Behavioral Commitment, AI Self-efficacy, AI Cognitive Application, and AI Intrinsic Motivation. The five-factor model showed acceptable fit and stable results across estimation methods, with the separation of self-efficacy and intrinsic motivation consistent with theoretical expectations. Internal consistency, convergent validity, and discriminant validity were supported. Configural invariance across gender was established, though metric and scalar invariance were not, likely attributable to the limited male sample size (n = 75). Criterion-related validity was partially supported, with AI Ethics, Behavioral Commitment, and Cognitive Application showing significant positive associations with AI teaching integration. Exploratory profile analysis identified four distinct profiles interpreted through Expectancy-Value Theory: Overall High-level Type (27.8%), Medium-level Type (41.3%), High Ethics-Low Self-efficacy Type (19.4%), and Overall Low-level Type (11.5%). The profiles showed significant grade development characteristics and differences in AI teaching integration ability, providing a basis for differentiated instructional approaches. Given sample characteristics, results should be interpreted with appropriate generalizability boundaries.