While the first Learning Analytics Conference in 2011 introduced learning analytics as a distinct research focus and practice, the collection, analysis, and use of student data has been part of education and specifically higher education for a very long-time preceding LAK’11. Likewise, concerns regarding ethical issues data have always been an integral part of discourses surrounding and research into the collection, analysis, and use of student data. As higher education become increasingly digitised and datafied, higher education institutions have access to not only more data, but also greater varieties and velocity of data, often in real-time, deepening concerns about the ethical implications but also raising different ethical issues than before.
Since LAK’11 many institutions developed codes of practice or policies to address various concerns and to ensure ethical practice. Learning analytics is, however, not an island, and emerging evidence suggests that discourses in the broader context of data science and the deployment of human-algorithmic decision-making systems have moved ‘beyond’ ethics, as a wide range of issues such as fairness, equity, social justice and more recently, calls to decolonise data and learning analytics, are claiming attention.
In this talk, I will tentatively map the trajectory of ethics in learning analytics and conclude with discussing some implications of this move ‘beyond’ ethics for policy, research and learning analytics practice.
Paul Prinsloo is a Research Professor in Open and Distance Learning (ODL) in the College of Economic and Management Sciences, University of South Africa (Unisa). His academic background includes fields as diverse as theology, art history, business management, online learning, and religious studies. Paul is an established researcher and has published numerous articles in the fields of teaching and learning, student success in distance education contexts, learning analytics, and curriculum development. His current research focuses on the collection, analysis and use of student data in learning analytics, graduate supervision and digital identity.
While self-regulation of learning (SRL) has generally been recognised as an important factor of learning success, it is especially important in open and flexible learning contexts - such as massive open online courses (MOOCs) or courses with active learning design - where learners are assumed to be active agents capable of managing their own learning processes. Learning tactics and learning strategies are considered important elements of SRL and ability to regulate one's tactics and strategies is considered a key SRL skill. Over the recent years, we have explored different computational methods for identifying and understanding students' learning tactics and strategies from trace data - moving from action counts, to action sequences, to micro SRL phases; from sequence mining to process mining to network analysis; and from a single course to multiple courses to different learning modes (fully online, blended, MOOCs). Many relevant insights have been made along the way. At the same time new questions have been opened that require additional / complementary analytics approaches. In this talk, I will give an overview of the research we have done so far on identifying and understanding learning strategies from traces of student learning behaviour, and also outline potential directions for moving this research ahead.
Jelena Jovanovic is a Professor at the Department of Software Engineering, University of Belgrade, Belgrade, Serbia. She received her B.Sc., M.Sc., and PhD degrees in Informatics and Software Engineering from the University of Belgrade in 2003, 2005, and 2007, respectively. She is teaching undergraduate and postgraduate courses in programming and applied artificial intelligence. As a researcher, for years, she was primarily focused on semantic technologies, particularly, Semantic Web / Linked Data technologies, and their application in the educational domain. In the last few years she is more into statistical / machine learning methods and techniques, social network analysis, and other computational approaches that allow for data analysis and extraction of meaningful information from unstructured data. She is particularly interested in combining these methods and techniques (machine intelligence) with large-scale knowledge bases (collected human intelligence) for better understanding of the learning process. Her interests also include how the insights obtained from the data can be effectively communicated to teachers and/or learners, as a meaningful feedback. She is a member of the GOOD OLD AI research network, LS3 research lab at Ryerson University, Toronto, Canada, and C3L research centre at University of South Australia, Adelaide, Australia.