Developing and validating the student learning agency scale in generative artificial intelligence (AI)-supported contexts

While generative artificial intelligence (AI) empowers students in their learning, it may also have the potential to undermine their learning agency. Understanding student learning agency in the generative AI-supported contexts (SLA-GAI) has become critical for educators. However, student learning agency is a domain-specific construct, current scales neglect the specific generative AI situations. Therefore, current scales are considered inapplicable for measuring SLA-GAI. This study aims to conceptualize the student learning agency and develop and validate the SLA-GAI scale. We conducted an exploratory sequential design with two stages. In the first stage, a literature review and group interview with experts were conducted to conceptualize the student learning agency and generate initial items. 268 valid samples from university students were then collected and exploratory factor analysis (EFA) was employed to adjust the initial items. In the second stage, 425 valid samples were collected and randomly split into two subsamples (n1 = 268, n2 = 245) to conduct EFA and confirmatory factor analysis (CFA), respectively. The findings indicate that the SLA-GAI scale includes ten factors with 34 items. Four of these - self-cognition, goal setting, self-adjustment and self-reflection - represent the key ability dimension of the SLA-GAI scale. The other three factors - selective action, responsible action and participative action represent the active action dimension of the SLA-GAI scale. Finally, self-efficacy and volition present the essential mental characteristics of the SLA-GAI scale. The developed SLA-GAI scale can reflect the latest and abundant meaning of the student learning agency in generative AI-supported contexts. It can help educators assess and develop student learning agency in the generative AI-supported contexts and explore related theories.