Reading and Writing (but no 'rithmetic)

I want to do a social research project.  Anyone who's been a librarian, or a teacher, or even a parent has heard the notion that "people who read a lot are better writers than those who eschew reading".  That just sounds completely true, right?  It's a no-brainer!  But there have been very few empirical studies done to prove that fact.  And really, it would be nearly impossible to do - you'd have to get permission from the students you surveyed not only for the survey about their reading habits, but you'd need to get other data about them - from their professors - on how well they write.  That's the only way you can get reliable data.  But research projects these days are very, very sensitive to the participants' identities, so I just don't know how you could do it so that the results are actually meaningful.  You absolutely HAVE to be able to compare the two, and that can only be done if someone working with the study knows the identity of the participants.

Technically, it would be very simple - ask a series of questions like "How often do you read material that is not course-related?" and "What kinds of materials do you like to read {books, newspapers, magazines, blogs, etc)?" and "Now that you are in college, do you find you read more for pleasure or less?"  Then you take those students' answers, and compare them to their writing ability (preferably judged by someone who has experience grading and evaluating writing).  Then see if there's a correlation between those who read more for pleasure and those who are better writers.

Still, I can't figure out how to get around the privacy issues.  Maybe if I partnered with English or Composition faculty who worked that in somehow as part of a course?  What if you took blind surveys of students taking one of the many "fun" classes universities offer now (like mini-semester courses on Tolkien, or other popular authors) and compared that with a blind survey of the same level of student in a regular comp class.  You could make the assumption that the students interested in Tolkien are probably "readers", but I still don't think that would give you reliable data in the long run.

If you have any good ideas, let me know.  In the meanwhile, off and on as I have time, I'm going to continue to look for work others have already done on the subject. I've seen several on reading and general academic level, but I want a more granular study targeting writing ability!