Research Question: What early engagement patterns in virtual learning environments predict negative course outcomes?

Context: It is well known that performance on assessments and in-class attendance are predictive of final course results. Yet grades often come too late in a class term for early interventions and attendance is difficult to measure in online learning environments. To address this gap, we developed a model for identifying early risk factors in online courses based on student interaction patterns in a virtual learning environment (VLE).

Data source: Open University Learning Analysis Dataset (OULAD), which includes daily logs of UK student VLE interactions and grades in 7 science and social science online courses occurring in 2013-14.

Goal: Develop a model for identifying early risk factors based on student interaction patterns that predict negative course outcomes (i.e., failure or withdrawal) in a VLE.