Rethinking Low Completion Rates in MOOCs

Completion rates is a pretty poor measure of what I think we really care about the most: educational uptake and learning.

On a side note, I'm a little disappointed in ed-tech. I was hoping for a revolution in pedagogical approaches. But it seems we are still working on it! Bring on the brain-plug learning a la Neuromancer, et al.

Why is a low completion rate causing concern? I never completed the machine learning course by Andrew Ng - my schedule didn't always wind up aligning with the times allotted for homework, and to be honest, there was some laziness mixed in there later on as well. I still got a lot out of the course - I've unexpectedly found myself having to do a lot of machine learning work at my new job, and the stuff I was exposed to in the Coursera course has helped to orient me.

A couple comments on that page stood out for me - one, that even with a low completion rate, more people completed a single online electronics course at MITx than have ever done so in person in the entire history of MIT. Second, the comment likening MOOCs to books - even if I don't finish a book, I still get value from reading it.

It's kinda funny somebody takes this numbers so seriously. My completion rate must be something about zero, but it would be nonsense to say I'm not learning. I hardly ever follow the course "in the real time", which means even if I do all my homework it almost never is full-score (most of the time even doesn't count as completed at all: for example on "archived" courses on edx) and by no means I do all my homework. In fact, right now I can remember at least 2 courses which are extremely hight in my personal "amount of practical knowledge gained from it" toplist, for which I completed exactly zero homeworks and never even watched video on the coursera, because it was easier for me to get it from torrents. For courses with programming it often is downloading assignments, but never submitting them. Why? For various reasons. Sometimes I want to try to do something in other language than suggested one. Sometimes I can think of something more interesting or practical or challenging to do, based on topic I just learned. And if it is the case it actually means I just gained much more from the material than if the best I could do would be just completing what is expected from me. Sometimes, of course, it's the opposite: the material is to easy or just silly for I would care completing assignments or even watching all the videos. And, yes, sometimes (rarely, though) the homework seems to be too hard and time consuming than I'm ready to be doing. Sometimes all I do on the course is using forums, sometimes I never even open the forums once. Sometimes all I get from the course is accidentally getting myself interested in some another topic, very slightly related to the course itself, and I'll be investigating on this topic reading various stuff for a very long time after that. In short, I never even tried to be "completing courses" on MOOC, I just tried to learn what I'm interested in maybe using the course material, maybe not, doing things on my own pace, deciding what I should spend time on by myself, and my observable online behavior hardly tells anything about how much I actually invested in learning something.

So this attitude of "You see these low completion rates?! Not like in real universities at all!" is just silly. Because, yeah, right, MOOC aren't like "real" (old-fashion) universities: it's entirely different and ultimately much better thing. It's about learning, not about completing courses, even if creators of these courses don't think so. Test scores and graduating rates are notoriously bad measure even for "real life" universities, and thousand times more so for MOOC.

I think the low completion rates of MOOCs are, while not necessarily a good thing, also not a bad thing, and basically symptomatic of a good thing. In terms of committed costs, MOOCs for the most part are more like textbooks than traditional college classes, even on an a-la-carte menu as one might have available through a community college or extension program. The cash cost is fairly low, the effort expended in enrolling is low, and the time/effort cost is just the time and effort expended directly in online course activities, not driving to a fixed class location at a fixed time that forces giving up other activities with time conflicts (you still have to give up other activities, but you are more flexible as to which ones), etc.

So they are cheap to experiment with and don't have a lot of sunk costs that people feel they need to justify that stops people from abandoning them if once they get in they expectations aren't met, circumstances change, etc.

So there's a lot lower risk, and people can try a lot more to find what they are looking for. I've finished maybe a third to a half of the MOOCs I've enrolled in, and I certainly wouldn't do that with college classes -- but that doesn't mean that I'm less well served with MOOCs -- the cost of college classes is high enough that I wouldn't take most of the MOOCs I've completed -- much less all those I've taken and often still gotten something out of without completing -- as traditional classes, because the risk of something with a higher priority in my life changing so I couldn't complete them, combined with the cost that goes into a traditional college class, would be too high.

I just recently launched a startup trying to take on this challenge of low completion rates in MOOCs: www.madrassa.co (you can read more in the about section concerning this insane dropout rate of MOOCs)