Ok, I'm gonna restate what I know to make sure I'm clear. Sorry if this just feels like a dupe of everything that's been said.
You want to see how people treat information and how they use it to build their models to support their theorems. You're going to study as much the actual data and it's comparison to models as you're going to study people and their approaches to using the data. You will probably even study the specific technology choices they make to examine any underlying trends (altho I predict that your datasets will be too small in those cases to make terribly accurate models).
Because it's being done for academic reasons and because there is a clear history of this on SO/MSO, I don't see where this will be a problem if executed in moderation and if you keep in line with the other guidelines of the SE2.0/SOIS system.