Very good point. Of course, for a completely fair comparison you need to ask the same question (before the course starts) to the traditional students. I expect

Very good point. Of course, for a completely fair comparison you need to ask the same question (before the course starts) to the traditional students. I expect it can't be much more than a percent or so, but it depends on whether the university penalizes students for signing up for a course they don't finish (in my years of college, the CS department didn't seem to care very much. It was very different for other departments, though). Of course, without penalties, you will get students that start out a semester following more courses than they can finish, only to decide later on which one to drop. I'm not entirely sure how that translates to MOOCs, though.