Learning analytics performance improvement design (LAPID) in higher education: Framework and concerns

Abstract

Learning Analytics Dashboards (LAD) promise to disrupt the Higher Education (HE) teaching practice. Current LAD research portrays a near future of e-teaching, empowered with the ability to predict dropouts, to validate timely pedagogical interventions and to close the instructional design loop. These dashboards utilize machine learning, big data technologies, sophisticated artificial intelligence (AI) algorithms, and interactive visualization techniques. However, alongside with the desired impact, research is raising significant ethical concerns, context-specific limitations and difficulties to design multipurpose solutions. We revisit the practice of managing by the numbers and the theoretical origins of dashboards within management as a call to reevaluate the “datafication” of learning environments. More specifically, we highlight potential risks of using predictive dashboards as black boxes to instrumentalize and reduce learning and teaching to what we call “teaching by the numbers”. Instead, we suggest guidelines for teachers’ LAD design, that support the visual description of actual learning, based on teachers’ prescriptive pedagogical intent. We conclude with a new user-driven framework for future LAD research that supports a Learning Analytics Performance Improvement Design (LAPID).