Visualizing discussions is an interesting challenge. In this post I use a simple technique in the context of discussions in an on-line course.

I have been auditing Dr. Anthony C. Robinson's Maps and the Geospatial Revolution on Coursera, a MOOC (Massive Open Online Course). 10% of a student's grade for the course comes from the weekly discussion assignments. If a student makes 20 posts within the 5 week course, they will receive full credit. When venturing into the forum for Week 1, I was struck by the low quality of the "discussion" taking place. It did not seem at all like an exchange of ideas.

My gut told me that the participation credit was driving bad behavior. The data lover in me wanted to prove it with data. The data visualizer wants to see it to believe it.

I found two threads to compare. In the general discussion forum I found a thread, "Where are you located?" and in the assignment forum I found "Where are you from?". More or less the same question (engaging discussions on the topic of identity aside), but with one key difference: the posts in the assignment forum were for credit. Take a look (click for bigger):

Posts go in chronological order from left to right. Every author is a line. An author who posts only once is a dot. An author who posts more than once has his posts connected by a line.

The difference is clear. While this does not constitute proof, it's certainly compelling.

Statistics of note: The general discussion thread has 1.34 posts per author and 96 characters per post, compared to 1.20 posts per author and 81 characters per post in the discussion assignment.

Edit: I have been corrected. Apparently all posts in all forums count for credit, though this is not entirely clear from the syllabus. The key difference between the two posts remains, but must be restated: Participants in the assignment forum were there primarily for credit, while those in the general forum were there primarily to discuss.