.
Learning analytics has the capacity to provide potential benefit to a wide range of stakeholders within a range of educational contexts. It can provide prompt support to students, facilitate effective teaching, highlight aspects of course content that might be adapted, and predict a range of possible outcomes, such as students registering for more appropriate courses, supporting students’ self-efficacy, or redesigning a course’s pedagogical strategy. It will do all these things based on the assumptions and rules that learning analytics developers set out. As such, learning analytics can exacerbate existing inequalities such as unequal access to support or opportunities based on (any combination of) race, gender, culture, age, socioeconomic status, etc., or work to overcome the impact of such inequalities on realizing student potential. In this editorial, we introduce several selected articles that explore the principles of fairness, equity, and responsibility in the context of learning analytics. We discuss existing research and summarize the papers within this special section to outline what is known, and what remains to be explored. This editorial concludes by celebrating the breadth of work set out here, but also by suggesting that there are no simple answers to ensuring fairness, trust, transparency, equity, and responsibility in learning analytics. More needs to be done to ensure that our mutual understanding of responsible learning analytics continues to be embedded in the learning analytics research and design practice.
The datafication of learning has created vast amounts of digital data which may contribute to enhancing teaching and learning. While researchers have successfully used learning analytics, for instance, to improve student retention and learning design, the topic of privacy in learning analytics from students' perspectives requires further investigation. Specifically, there are mixed results in the literature as to whether students are concerned about privacy in learning analytics. Understanding students' privacy concern, or lack of privacy concern, can contribute to successful implementation of learning analytics applications in higher education institutions. This paper reports on a study carried out to understand whether students are concerned about the collection, use and sharing of their data for learning analytics, and what contributes to their perspectives. Students in a laboratory session (n = 111) were shown vignettes describing data use in a university and an e-commerce company. The aim was to determine students' concern about their data being collected, used and shared with third parties, and whether their concern differed between the two contexts. Students' general privacy concerns and behaviours were also examined and compared to their privacy concern specific to learning analytics. We found that students in the study were more comfortable with the collection, use and sharing of their data in the university context than in the e-commerce context. Furthermore, these students were more concerned about their data being shared with third parties in the e-commerce context than in the university context. Thus, the study findings contribute to deepening our understanding about what raises students’ privacy concern in the collection, use and sharing of their data for learning analytics. We discuss the implications of these findings for research on and the practice of ethical learning analytics.
This article seeks to explore different combinations of human and Artificial Intelligence (AI) decision-making in the context of distributed learning. Distributed learning institutions face specific challenges such as high levels of student attrition and ensuring quality, cost-effective student support at scale using a range of technologies, such as AI. While there is an expanding body of research on AI in education (AIEd), this conceptual article proposes that combinations of human-algorithmic decision-making systems need careful and critical consideration, not only for their potential, but also for their appropriateness and ethical considerations. We operationalize a framework designed to consider robot autonomy at four key events in students' learning journeys, namely (1) admission and registration; (2) student advising and support; (3) augmenting pedagogy; and (4) formative and summative assessment. We conclude the article by providing pointers for operationalizing options in human-algorithmic decision-making in distributed teaming contexts.
Since its inception in 2011, Learning Analytics has matured and expanded in terms of reach (e.g., primary and K-12 education) and in having access to a greater variety, volume and velocity of data (e.g. collecting and analyzing multimodal data). Its roots in multiple disciplines yield a range and richness of theoretical influences resulting in an inherent theoretical pluralism. Such multi-and interdisciplinary origins and influences raise questions around which learning theories inform learning analytics research, and the implications for the field should a particular theory dominate. In establishing the theoretical influences in learning analytics, this scoping review focused on the Learning Analytics and Knowledge Conference (LAK) Proceedings (2011–2020) and the Journal of Learning Analytics (JLA) (2014–2020) as data sources. While learning analytics research is published across a range of scholarly journals, at the time of this study, a significant part of research into learning analytics had been published under the auspices of the Society of Learning Analytics (SoLAR), in the proceedings of the annual LAK conference and the field’s official journal, and as such, provides particular insight into its theoretical underpinnings. The analysis found evidence of a range of theoretical influences. While some learning theories have waned since 2011, others, such as Self-Regulated Learning (SRL), are in the ascendency. We discuss the implications of the use of learning theory in learning analytics research and conclude that this theoretical pluralism is something to be treasured and protected.
Evidence shows that appropriate use of technology in education has the potential to increase the effectiveness of, eg, teaching, learning and student support. There is also evidence that technology can introduce new problems and ethical issues, eg, student privacy. This article maps some limitations of technological approaches that ensure student data privacy in learning analytics from a critical data studies (CDS) perspective. In this conceptual article, we map the claims, grounds and warrants of technological solutions to maintaining student data privacy in learning analytics. Our findings suggest that many technological solutions are based on assumptions, such as that individuals have control over their data ('data as commodity'), which can be exchanged under agreed conditions, or that individuals embrace their personal data privacy as a human right to be respected and protected. Regulating student data privacy in the context of learning analytics through technology mostly depends on institutional data governance, consent, data security and accountability. We consider alternative approaches to viewing (student) data privacy, such as contextual integrity; data privacy as ontological; group privacy; and indigenous understandings of privacy. Such perspectives destabilise many assumptions informing technological solutions, including privacy enhancing technology (PET).
