Implementing next generation privacy and ethics research in education technology

Abstract

For the developers of next‐generation education technology (EdTech), the use of Learning Analytics (LA) is a key competitive advantage as the use of some form of LA in EdTech is fast becoming ubiquitous. At its core LA involves the use of Artificial Intelligence and Analytics on the data generated by technology‐mediated learning to gain insights into how students learn, especially for large cohorts, which was unthinkable only a few decades ago. This LA growth‐spurt coincides with a growing global “Ethical AI” movement focussed on resolving questions of personal agency, freedoms, and privacy in relation to AI and Analytics. At this time, there is a significant lack of actionable information and supporting technologies, which would enable the goals of these two communities to be aligned. This paper describes a collaborative research project that seeks to overcome the technical and procedural challenges of running a data‐driven collaborative research project within an agreed set of privacy and ethics boundaries. The result is a reference architecture for ethical research collaboration and a framework, or roadmap, for privacy‐preserving analytics which will contribute to the goals of an ethical application of learning analytics methods. What is already known about this topic Privacy Enhancing Technologies, including a range of provable privacy risk reduction techniques (differential privacy) are effective tools for managing data privacy, though currently only pragmatically available to well‐funded early adopters. Learning Analytics is a relatively young but evolving field of research, which is beginning to deliver tangible insights and value to the Education and EdTech industries. A small number of procedural frameworks have been developed in the past two decades to consider data privacy and other ethical aspects of Learning Analytics. What this paper adds This paper describes the mechanisms for integrating Learning Analytics, Data Privacy Technologies and Ethical practices into a unified operational framework for Ethical and Privacy‐Preserving Learning Analytics. It introduces a new standardised measurement of privacy risk as a key mechanism for operationalising and automating data privacy controls within the traditional data pipeline; It describes a repeatable framework for conducting ethical Learning Analytics. Implications for practice and/or policy For the Learning Analytics (LA) and Education Technology communities the approach described here exemplifies a standard of ethical LA practice and data privacy protection which can and should become the norm. The privacy risk measurement and risk reduction tools are a blueprint for how data privacy and ethics can be operationalised and automated. The incorporation of a standardised privacy risk evaluation metric can help to define clear and measurable terms for inter‐ and intra‐organisational data sharing and usage policies and agreements (Author, Ruth Marshall, is an Expert Contributor on ISO/IEC JTC 1/SC 32/WG 6 "Data usage", due for publication in earl...

Understanding privacy and data protection issues in learning analytics using a systematic review

“…Moreover, lacking regulatory guidelines for human tool users. Combined solutions. This is a hybrid solution combining the previous two solutions, few that are documented (eg, Marshall et al, 2022). Such solutions may hold promise to integrate benefits from previous solutions, mutual reinforcement for the future.…”

“…However, there are some difficulties in the proposal and implementation of combined solutions, due to their complexity and workload. This has only been attempted by Marshall et al (2022) who propose a replicable framework for ethical LA that includes both technical solutions and legal‐ and framework‐based solutions, and which also introduces a new standardised privacy risk measure. Furthermore, this privacy risk measurement mechanism helps to automate data privacy operations (Marshall et al, 2022).…”

Technological frameworks on ethical and trustworthy learning analytics

“…The first group of submission focuses on bridging the gap between research and practice to provide a necessary infrastructure to protect student privacy while preserving the utility of LA methods. A good illustration of this is the paper by Marshall et al (2022). This paper addresses the practical challenges in sharing data within a cross-disciplinary research collaboration involving LA researchers and industry specialists in Education Technology and Data Privacy experts.…”

“…The special section provides a starting point for broadening the discussion on how to translate the latest and most innovative research in adjacent fields such as data privacy and security into practical tools and applications for an ethical and privacy aware LA.The first group of submission focuses on bridging the gap between research and practice to provide a necessary infrastructure to protect student privacy while preserving the utility of LA methods. A good illustration of this is the paper by Marshall et al (2022). This paper addresses the practical challenges in sharing data within a cross-disciplinary research collaboration involving LA researchers and industry specialists in Education Technology and Data Privacy experts.…”

“…As indicated by Marshall et al (2022), in addition to the infrastructure, a crucial aspect of developing trustworthy LA is centred around the algorithms and methods that would allow us to measure and mitigate potential risks. In that sense, Vatsalan et al (2022) propose a new re-identification risk measure based on a Markov Model that does not require strong assumptions about the adversary's knowledge of the datasets.…”