Saturday, August 17, 2013

Mathematical Biostatistics Boot Camp 1 (Johns Hopkins, Coursera) - how NOT to teach math online

I apologize in advance for how angry this course makes me. It is teaching like this that makes people believe that math is inherently difficult or that you have to have a certain mindset to do math. That simply isn't the case at all. I have a degree in math, teach statistics and have considered becoming an actuary. I know my statistics! Still I found myself being confused as to what exactly Professor Caffo was trying to say at times. I have no doubt that he is able to teach this stuff well at Johns Hopkins but Prof. Caffo seems oblivious to online learning.

Whereas the UPenn calculus course was a great example of how you can teach advanced math over the internet this is the exact opposite. Sure you can assume that your audience has more mathematical maturity than the average audience since they need to know basic calculus in order to understand anything but as will be described below the course is a confusing muddle of poorly designed slides, lengthy videos and poor assessments.

Professor Caffo uses the LaTeX Beamer package to design his slides which I like. I use the same thing for my calculus and statistics lecture slides. Unfortunately he doesn't seem to have designed the slides in a matter that fits his lecture. The biggest mistake that he makes is that he tends to present the entire slide all at once rather than in parts. To his credit he does sometimes use highlighting to let you know where he is on a slide but there is just too much information presented at once. Sometimes it is hard to figure out where he is on a slide because his lecture style is more conversational (not that this is a problem in and of itself) so he often goes off on mini-tangents about something causing you to lose your place within the slide. There are a few videos where the lecture itself gets confusing and these should probably be re-recorded. Of course that is made difficult by my next problem with the lectures...their length.

I believe that most mooc video lectures should be under 15 minutes of length and deal with a very specific topic. This makes it easier for the students to go back and re-watch the video or find specific parts of the lecture that they want to review. I think this is even more important when dealing with technical material and to me is one of the strongest selling points of a mooc. In a traditional classroom setting you have to have lengthy lectures in order to get all of the material taught. You can't have students showing up for 8-10 lectures each week for 15 minutes or so.

One of the things I liked about the UPenn calculus course was that you had that short video followed by a homework assignment pertaining to just that video. The videos were also set up so that if you did one per day you would finish everything on schedule. This makes it easier for the student to absorb the material because they can focus on one specific thing at a time.

Now to be fair, many of Prof. Caffo's videos are under the 15 minute mark but there are also quite a few that are way over. In the first three weeks of videos there are four videos over 20 minutes and one of those is over a half an hour! The strange thing is that in many of these videos there are times when Prof. Caffo makes it seem as though the lecture is about to end but the video continues. This might have just been an editing error on their part, i.e. they intended for the video to end there, but this is something like the third iteration of the course so it could have been fixed by now.

Sometimes there is a significant amount of time spent on a new topic before any examples are given. The examples do help to clarify the material but after so long many students are struggling far more than they need to be. The examples also seem hurried at times when they are actually the most important part of the lecture itself. Since students don't have a section to go to and work out problems with a graduate students it is more important that they are able to see the theory applied in detail in the lecture.

This might seem like a minor point, after all you can still go back and find specific parts easily and you can also pause and continue to watch whenever you like. I think that people tend to watch the lectures through in their entirety the first time through; at least I know I do. I will concede that a course can still be great with longer videos and if that were my only criticism of this course I would probably only mention it in passing but combined with everything else it makes for a worse experience.

My absolute biggest problem with this course is not the length of the videos, or the poorly designed slides. It is the lack of problems that really makes me mad. The only way to learn math is to do math. That is it. There is no other way. You can watch all the video lectures you want and read all of the math books in the world. If you do not do problems you will not master the material.

In the third week of the course there were three video lectures, two of them over 20 minutes long and one of those was the 32 minute one. None of them had in-video quizzes. There were quite a few the first week but after that they seem to be absent. This is unfortunate especially with the length of the videos. The in-video quizzes are natural break points. Beyond that there is an optional homework assignment followed by a non-optional quiz that is generally very similar to the homework. These tend to be about 8-10 questions long. That simply is not enough for a course like this. There should be that many questions per each video. Some students will only need to do a couple in order to understand, some may need all 10. The way that Professor Caffo designed the course you get only a couple of questions per topic at best. In week two there were lectures on covariance and correlation but neither of those were addressed in the assessments.

I should point out that some students have complained about a lack of problems with practical applications. This was never really a sore point for me because I enjoy math for math's sake. However this is supposed to be a bio-statistics course so perhaps the complaint is not without merit.

The problem with moocs is that you can not assume students have access to any material outside of what is provided. I have a few mathematical statistics books but most students will not. These books tend to be on the expensive side since they are textbooks and the whole point of a mooc is to be as accessible as possible. I truly believe that Prof. Caffo's course in person teaches students a solid foundation of mathematical statistics. They are able to ask questions during lecture of sections, they have faculty members that they can approach with questions. They have a textbook. This allows them to not only read the material but also gives them access to more problems and worked solutions.

When all the students have are the lecture videos (although the forums can be helpful when it comes to specific problems) there needs to be additional support provided throughout the structure of the course. My biggest suggestion is to add a lot more problems, ranging from the trivial to the more advanced.

There are two things I would like to conclude with. The first are the videos that Prof. Caffo gave after the first week where he went over the quiz questions in detail. This is a step in the right direction because at least it gives students a way to work through the problems with someone. Unfortunately this is given after they are due and when they are working on new material. Providing similar videos of worked problems that are similar to the homework problems might be a nice option. These could be released before the quiz deadlines so students have a place to go when they are stuck other than the forums.

Finally this was one of the earlier courses offered on Coursera, it has been offered a few times since. There was still a lot to learn about designing moocs. There still is! Professor Caffo is offering a second course that is described as a continuation of this one. Perhaps that course was built on the feedback from this one and will be less problematic. If so, hopefully Prof. Caffo will go back and make changes to this course down the road.

















1 comment:

  1. Thanks for sharing this valuable and useful information. You must have researched a lot on this topic. Great job done keep posting more
    Biostatistics

    ReplyDelete