Abstract

Data literacy has become a critical core competency for university students. Research has indicated that in digital environments, learners’ digital self-efficacy (DSE) not only influences their learning motivation but is also closely linked to their learning outcomes. Additionally, self-reflection could help students evaluate their learning process and deepen their understanding of the content. However, without appropriate instructional scaffolding, self-reflection may become a mere formality, failing to effectively enhance both the depth and quality of their reflection, as well as their learning motivation. Therefore, this study proposed an ARCS (Attention, Relevance, Confidence, and Satisfaction)-based self-reflective online learning model and integrated it into a 12-week data literacy course, with the intervention implemented over a 10-week period (Weeks 2–11). The aim was to explore data literacy achievement and approaches to learning in data literacy (ALDL) among university students with different levels of DSE, and the quality of self-reflection at different stages. Participants were 52 first-year university students, including 27 males and 25 females. Results showed that the proposed model effectively fostered a positive motivation cycle among students. While students with high DSE (HDSE) outperformed those with low DSE (LDSE) in terms of data literacy achievement and ALDL, the majority of students began their reflective process with technical reflection (88.46%). To further explore the model’s influence on self-reflection, Epistemic Network Analysis (ENA) 3D was employed to analyze the coded results of students’ reflective diaries. The findings indicated that the model effectively promoted multidimensional self-reflection, broadened and deepened reflective focus across both LDSE and HDSE groups, and reduced the quality gap in self-reflection between the two groups. Additionally, the LDSE group enhanced practical application and critical thinking through conceptual understanding, relying on hands-on experience to construct knowledge. In contrast, the HDSE group focused on deep reflection through logical explanation, self-validation, and critical thinking.

Keywords: ARCS model; data literacy; epistemic network analysis (ENA); ENA 3D; self-reflection; digital self-efficacy; ARCS-based self-reflective online learning model

---
*Corresponding author: E-mail: gjhwang.academic@gmail.com (Gwo-Jen Hwang)

Tu, Y. F., Hwang, G. J., & Hu, D. (2025). Effects on the learning achievement, approaches to learning, and multi-stage reflection quality of students with different levels of digital self-efficacy in a data literacy course: An ARCS-based self-reflective online learning model. Computers & Education. https://doi.org/10.1016/j.compedu.2025.105397

Acknowledgments

We are extremely grateful to Ms. Xiao-Ge Guo for her assistance in data collection, and to the students who participated in this study.

Author Introduction

Professor Gwo-Jen Hwang is a distinguished scholar in the field of digital learning, currently serving as a Chair Professor and Vice President at Taichung University of Education. He has made significant contributions to Al in education, mobile learning, game-based learning, and associated topics, with extensive publications in top-tier journals and leadership roles in renowned academic conferences. He has achieved over 100 research projects and published more than 500 articles in SSCI0indexed journals. His work continues to influence how Al-empowered feedback, learning analytics, and adaptive learning strategies transform teaching and learning in higher education.

Dr. Yun-Fang Tu is an Assistant Professor in the Department of Information Science at Soochow University, Taiwan, specialising in Artificial Intelligence in(AIED), technology-enhanced learning, and educational data analytics. She has published extensively in SSCI/SCI-indexed journals, with research focusing on AI-powered educational technologies, mobile and ubiquitous learning environments, and learning behavior analysis. Her scholarly work contributes to evidence-based approaches in adaptive learning systems and digital pedagogy, particularly examining the educational applications of generative AI and learning analytics.

Dr. Dongpin Hu (Peter) is a postdoctoral fellow at Education University of Hong Kong. He received his PhD in Educational Technology from the University of Hong Kong, TESOL from Arizona State University, MEd in Educational Studies from the Education University of Hong Kong, and BSc in Computer Science (Machine Learning and Artificial Intelligence pathway) from the University of London. Before joining EdUHK, he was working as a postdoctoral researcher at the Chinese University of Hong Kong and a part-time lecturer at the University of Hong Kong. Dr Hu’s research interests are AI in Education, epistemic network analysis, content and language integrated learning (CLIL), adaptive learning, and technology-enhanced language learning. He has been published in top-tier journals such as Educational Research Review and Computer Assisted Language Learning.