MOOCs are an interesting area as well, as they share similar drop off to free mobile games - high initial recruitment and engagement, but immediate drop off. I love a lot of the components that MOOCs use and utilize them in my course design, but the high drop off is why I support traditional class-based environments. The cost of tuition and fear of failure ensure they stick with it. In another world, I think it's one of the reasons why many physical sports like martial arts have high drop offs as well, no risk from failure.

This is one of the reasons I don't particularly care for motivation studies, as I believe it's only a temporary method to generate traction. Once learning becomes difficult, it requires a higher order of motivation I call "discipline" to maintain.

But back to MOOCs, there is work on MOOC forum analysis[1]. In short, forum activity and the people you associate with in an educational forum can have a significant impact on your grade. Note this study was post-hoc analysis, but students that primarily discussed non-course relevant topics were more likely to fail the course.

There is also the fact that MOOC students come from a variety of life experiences. Students of different countries have different cultures and as a result, will behave differently [2]. France, for example, only wanted to view the course in [2], not engage.

I think the beauty of a MOOC is the accessibility and automated feedback feature that from their design. Beyond that they suffer from a lot of human conditions that can't really be controlled without removing or reducing those features. Plus, as I mention, there isn't a lot of cog psych research on "discipline" yet to identify the qualities that encourage its development besides Duckworth's "grit" (if I'm wrong I'd love to be pointed towards them).

[1] http://ceur-ws.org/Vol-1446/GEDM_2015_Submission_2.pdf

[2] https://files.eric.ed.gov/fulltext/ED592695.pdf