Massive open online courses (MOOCs) are all the rage these days, so I decided to try one out of curiosity. My choice was Coursera’s Statistics One, a six week course taught by Andrew Conway, a psychology professor from Princeton University. Here’s a short review of the first version of the course from fall 2012.
My background: As a computer scientist I’ve had lots of maths exposure at university, including statistics and probability theory. But that was a long time ago and my knowledge has become quite rusty due to lack of practice. I generally enjoy learning new stuff, but usually I stay within my comfort zone, which is distributed systems and software engineering.
Content: Statistics One was very much unlike the mathematical statistics treatment I was used to. It was announced as a course suitable for everyone and it covered much of the basics, but there were parts that were mostly relevant to students from the social sciences. While it’s interesting to see how psychologists set up their experiments (my little sister is a psychologist), I would have preferred a bit more mathematical depth and references to literature or online resources. This would have helped to gain a deeper understanding or to get explanations from a different angle, especially since the suggested text book didn’t cover the entire course. Anyway, all in all this was by far the easiest maths related course I’ve ever had.
Lecture videos: Having watched lots of conference videos and also some video lectures before, I’ve seen good and bad. Without any doubt the videos for this course were the best so far; a very professional, TV quality production with excellent audio. Conway’s English was clear and easy to understand and I didn’t need subtitles at all, it was like watching CNN news. My most favorite thing with online lectures is that you can hit the pause button to think about concepts, take notes, or look things up. I would have loved that feature back at university.
Quizzes, assignments, exams: Quizzes and assignments were a crucial element that kept me motivated, otherwise I would have dropped the course after week 4 when all interesting topics (at least for me) had been covered. They were fairly easy (some students struggled a lot with R though) and didn’t take a lot of time.
The use of R: The choice of using R in the course seemed to be controversial among students. Several dropped the course with great drama because they "had to write code". Actually, R was used for plotting and as a scientific calculator. No control structures were used, just variable assignments and function calls. Learning the tools of the trade is important in my opinion, I definitely didn’t want to calculate regression coefficients by hand.
Community: No matter what you do on the Internet, especially when it’s for free, there’s always the bottom one percent that will raise hell on the forums. I was so utterly disgusted by the way some people behaved that I stayed away from the forums after week one. I guess things improved after a while when the trolls dropped out, but I did fine on my own.
Conclusion: I really enjoyed Statistics One and recommend it to anyone who is serious about learning statistics or refreshing their knowledge. This is a university level course, not entertainment, it takes time to watch the lectures and to work through the quizzes and assignments. Even though you can tell that a lot of time went into preparing and running the course, there were some minor issues in the organizational area that caused irritation among students. But I’m sure that will be fixed in Statistics One, Version 2.0.