Friday, June 8
Next up: visualize more than one question. A “select all that apply” question is basically just a set of yes/no questions. I have a way to visualize that now, but what about a set of multiple choice questions? After some brainstorming, this ended up as the rough goal:
I spent the next few hours working on this new challenge. It requires a new Shadron program for graphics and a new C++ program for reading and counting data. It turns out that counting second degree relationships is kind of a pain! I ended up needing 5 nested for-loops to count through each pair of nodes: question A, option A, question B, option B, and survey responder. “Options” are the choices that the participant has (T/F, abc, abcdef, etc.) for each question. In that picture above, options are grouped into questions by each straight line. At this point I hadn’t settled on how to show the counts of how many people chose each single option, but you can see how many people chose each pair of options by the thickness of the line connecting them.
I ended up working way past 5:00 and nearly to midnight since I was having so much fun putting this together. Here’s the final product comparing 2 questions, then another comparing 3 and then 4 questions:
As I was writing this program, it felt natural to represent each individual option as a circle of varying diameter like in the last visualization. As always, everything is adjustable. I didn’t go through the trouble to label everything (since this is just another proof-of-concept), but it still came from real survey data. Just like the last one, this is designed for data scientists and not the public.