Welcome back, dear reader, for more gripping data from my Metadata MOOC. This time let’s look at activity on the discussion forums.

I’m moving the raw data and the definitions of the stats down, into what would be an appendix, if blog posts had appendices.

Posts + Comments, in the figures above, is my own addition. Yes, Posts and Comments are different: if I understand correctly how Coursera’s discussion forum works, a post either starts a new thread, or is a new “top level” contribution to an existing thread, while a comment is a reply to an existing post. But me, I think it makes sense to lump all contributions to a thread together… I hypothesize that many students don’t pay attention to or care whether they’re posting or commenting. So for this post, at least, I’ll just count everything as equal, and refer to all posts and comments as “Contributions,” and all participants who posted or commented as a “Discussant.”

The Total Activity figures, above, look more or less like you’d expect: rapid rise in the first few weeks, then more or less a plateau. If you look at new activity per week (difference between one week and the previous week), however, instead of cumulative activity, you get a different picture: the number of new contributions, and the number of discussants, declines steadily. Which again, you’d expect, given that the number of active students per week declines.

For me, one of the Great Unanswered Questions about MOOCs is: How many students does it take for a discussion forum to be self-sustaining? By which I mean, questions posted by students are likely to be answered by other students, before the instructor or TA has a chance to reply. I remember seeing the figure of 1,000 as an estimate (though for the life of me I can’t remember where), but I think this figure disproves that: Week 1 saw 400 discussants, and that was an active week. Later weeks saw 100 discussants, and even fewer. I’d have to revisit the threads for those later weeks to see how successful they were in getting students answers, but I remember the forums being pretty active throughout the entire course.

What’s interesting here is that we see an uptick in contributions and discussants in week 8. Now, I haven’t done any content analysis of the discussion forums yet (though a colleague and I are planning to do that), but I hypothesize that this uptick will be explained by the burst of “thank you” posts and comments near the end of the course.

Week 1 saw the largest number of threads (108), and things stabilized at around 25 during weeks 5-8. The number of contributions per thread ranged between about 3 to about 9; the number of discussants per thread ranged between about 1-4. I’m actually a little surprised at how large these numbers are. Given how prevalent power law distributions are in information-related phenomena, I’d have expected more threads with only 1 post in them, and I’d have expected those to drag the average down. Once we get the data dump from Coursera of the discussion forums, I plan to see how many threads there are with different numbers of contributions.

There are a lot more analyses that I want to run on this data. (My old stats professor would call this a “fishing expedition.”) But, I’ll save those for another day, just so I can bring this post to a close.


Raw data:
[table “” not found /]

[table “” not found /]

Definitions of the stats:

  • Total Views: The total number of times that any student has viewed a thread.
  • ​Total Threads: The total number of threads (“topics”) in the forums across all sub-forums.
  • Total Posts: The total number of posts across all threads.
  • Total Comments: The total number of comments across all posts.
  • Total Votes: The total number of votes (both up and down) cast in the forums.
  • Total Reputation Points: The total number of reputation points currently held by the students. Students obtain reputation points when their posts are voted up (or down) in the forums by other students. For each student, his/her reputation is the sum of the square-root of the number of votes for each post/comment that he/she has made.
  • Number of Participants (posting, commenting, voting): The number of (unique) users who have performed said action–posting, commenting or voting–at least once since the start of the session.