Abstract:

Past research in Learning Analytics has assumed a linear relationship between LMS activity and learning outcomes. It is often seen as a manifestation of effort, motivation, or self-regulated learning. In fact, several studies have used statistical techniques that assume linearity to scrutinize the contribution of LMS activity to the variance of academic performance or the prediction of educational success or failure, failing to find a significant contribution. We tested this assumption in a cohort of first-year business students (N=582) in a blended learning program. Our statistical analysis shows that a logarithmic curve best describes the relationship between the total number of clicks in the LMS during an entire semester before exams and the average grade during the same period. These results imply that, at least in this population, and presumably in other blended-learning environments, LMS activity does not reflect dispositions that behave linearly regarding academic performance. Instead, it either stands for a more complex set of variables or reflects some student’s internal factors that can improve academic achievement only up to a certain point, after which they become somewhat irrelevant to outcomes. We advocate for qualitative research to answer this puzzle in forthcoming work.

Keywords:

Learning Analytics, LMS activity, academic achievement, U-shaped relationship.