Educational Data Mining (EDM) and Learning Analytics (LA) have expanded rapidly in recent years, fuelled by the proliferation of digital learning environments and advances in machine learning. Earlier work, including the topic-modelling analysis by [1], has charted the main thematic contours of these areas but also highlighted a notable omission: limited engagement with ethical issues such as privacy, algorithmic bias, and data governance. In this paper, we conduct a systematic review of EDM and LA publications from 2020 to 2024 to examine how ethical concerns have been taken up since that gap was first flagged. Our analysis of peer-reviewed studies shows that, although ethical issues are increasingly acknowledged, this attention is often cursory in empirical work. Only 18% of the studies explicitly implemented or evaluated privacy-preserving methods, and 10% reflected critically on potential biases in their datasets or models. Moreover, fewer than 8% reported involving key stakeholders (e.g., students, teachers) in decisions about data collection or use. Overall, our results suggest that ethics remains peripheral rather than embedded in EDM/LA research practice. We close by outlining an “Ethical-by-Design” framework for educational analytics, aimed at aligning methodological innovation with a sustained commitment to student well-being and equity.
Educational Data Mining (EDM) and Learning Analytics (LA) have expanded rapidly in recent years, fuelled by the proliferation of digital learning environments and advances in machine learning. Earlier work, including the topic-modelling analysis by [1], has charted the main thematic contours of these areas but also highlighted a notable omission: limited engagement with ethical issues such as privacy, algorithmic bias, and data governance. In this paper, we conduct a systematic review of EDM and LA publications from 2020 to 2024 to examine how ethical concerns have been taken up since that gap was first flagged. Our analysis of peer-reviewed studies shows that, although ethical issues are increasingly acknowledged, this attention is often cursory in empirical work. Only 18% of the studies explicitly implemented or evaluated privacy-preserving methods, and 10% reflected critically on potential biases in their datasets or models. Moreover, fewer than 8% reported involving key stakeholders (e.g., students, teachers) in decisions about data collection or use. Overall, our results suggest that ethics remains peripheral rather than embedded in EDM/LA research practice. We close by outlining an “Ethical-by-Design” framework for educational analytics, aimed at aligning methodological innovation with a sustained commitment to student well-being and equity.
