The Self-Regulation for AI-Based Learning Scale: Psychometric Properties and Validation

Abstract

Artificial intelligence technologies are transforming university students' learning processes, making self-regulation skills increasingly crucial. However, existing self-regulation scales inadequately reflect AI-assisted learning environments' unique dynamics. This study developed the Self-Regulation for AI-Based Learning Scale (SRAILS) for university students and examined its psychometric properties. The scale comprises four main dimensions and nine sub-dimensions: motivational components (intrinsic/extrinsic motivation, self-efficacy), cognitive/metacognitive strategies, time and task management, resource management, and technological adaptation. The study included 750 university students from various Turkish faculties. Exploratory Factor Analysis supported construct validity, while Confirmatory Factor Analysis confirmed the nine-factor structure, demonstrating good model fit. Convergent and discriminant validity were established. Cronbach's alpha and McDonald's omega reliability coefficients ranged from .853-.913. Criterion-related validity was confirmed through significant positive correlations between all scale dimensions and external criteria: academic GPA, technology interest, and digital literacy levels (r = .20-.60, p < .01). SRAILS provides a comprehensive, reliable assessment of students' self-regulation skills in AI-assisted learning environments. This scale contributes originally to literature by enabling personalized learning experience design and supporting effective instructional strategy development.