From Prompting to Understanding: Factorial Structure and Network Psychometrics of a Deep-Learning Scale in Generative AI Context

Abstract

This study aimed to develop and validate a scale for assessing deep learning approaches in their own learning among university students in the context of generative AI tools. Most existing measures were originally designed for traditional educational settings and do not reflect the substantial changes introduced by AI in learning practices. To address this gap, a 16-item scale was designed across four dimensions—search for meaning, relating ideas, analysing evidence, and attention to concepts—after adapting the items to AI-supported environments. The scale was administered to 765 Damanhour University students with prior experience using generative AI. Data were examined through exploratory and confirmatory factor analyses, convergent and discriminant validity tests, and network analysis. Results indicated strong psychometric properties, with validity and reliability at high levels. The network analysis revealed clear interconnections among items and highlighted central nodes representing core aspects of deep learning, which shaped other learning behaviours and reinforced the adoption of deep learning approaches. Overall, findings confirm the scale as a modern, reliable tool for capturing students’ deep learning strategies in AI-enhanced contexts, offering implications for future research and for improving higher education practices in light of ongoing digital transformations.