CRI researchers Jake Wintermute & Ariel Linder, along with CRI alum Matthieu Cisel, have published a new study of the dynamics of drop-out in online courses. The manuscript, "A survival model for course-course interactions in a Massive Open Online Course platform," arrived last week in the journal PLoS ONE.
Digital learning, once an emerging trend at the margins of education, is suddenly the new normal. Yet little is known about what drives success in online courses. The authors wanted to create a mathematical description of MOOC certificate rates that might reveal more about why students persist or drop out. They were inspired by "survival models" which have often been used to describe the lifespan and longevity of living organisms.
"This was a collaboration that was very specific to the CRI environment," says Jake Wintermute, lead author of the study. "We had people who study complex systems, people who study aging and people who study education, but only at the CRI do you see all those different ideas working together."
The study drew on a large dataset of more than 1,000,000 registrations to France Université Numérique (FUN), a French MOOC platform. On average only about 8.1% of the registrations resulted in an earned certificate of completion. Wintermute and colleagues identified user-specific and course-specific factors that correlate with a completed course. But the largest single factor appeared to be registation bursts: when users sign up for multiple courses in a short period of time.
"We found that students who register for many courses get about the same number of certificates as students who register for few courses," Wintermute explained. "It is as though students who begin multiple courses eventually choose to focus on just one. This makes sense considering how easy it is to register for these free and open-access courses. There is really no reason not to register for a course that seems interesting and there is no cost to dropping out if you change your mind."
Beyond the easy-come, easy-go nature of multiple registrations, the study revealed several more complex factors that influence completion rates. For example students who chose to register for difficult courses, defined as courses with low completion rates, tended to do better in all of their other courses.
"We hope this study can chart a path to a more integrated understanding of online education," said Wintermute. "Everything you learn influences everything else you learn. If we can move beyond thinking of just single courses in isolation, we might give students the preparation they need to avoid burning out."
Wintermute, E. H., Cisel, M. & Lindner, A. B. A survival model for course-course interactions in a Massive Open Online Course platform. PLoS ONE 16, e0245718 (2021).