So I suppose it’s a sign that I’ve managed to get myself knee deep in graduate school: I’m starting to do my own research.  It’s funny, despite the fact I teach a class on research methods I’m quickly finding out how little I really know about real research process.  Two methods courses and a statistics class notwithstanding none of my experience has been literal real world graduate caliber research.  I learn so much by application – I wish I could see how professors do their own research.  So as it stands I’m doing a lot of mimicking what books tell me and what I can discern are proper methods.  In less classier words, fake it until you [hope to] make it.  Thank God I found some help in an unexpected place: Survey Methods instructor Jane Burris.
I’ve begun to gather my own data for the Facebook project, with a formal DMI 1100 student random sampling and IRB’s blessing and all that jazz, and like any good researcher I of course find a couple hundred million questions I want to ask after my survey is released into the wild.  I made an ultimate newbie mistake too – turns out the ambiguous category on the form builder labeled “number of responses allowed” doesn’t pertain to individual respondents but how many people can respond to your survey period.  Cleared that one up this past week, I’m praying it doesn’t mar my response rate too badly.  The UIUC form builder is insufficient to perform real survey functions – it doesn’t support skip logic, use of visual aids, or partial response records.  In order to perform research at UIUC we have to use it, though.  The 20$ subscription to Survey Monkey for my convenience sample my senior year was leagues better, you’d think the university could afford something superior.  I know, I know, it’s probably like one LAS social science IT guy who gets tasked with that and 1800 other things.  So make it an assignment in a CS class, problem solved.
Here’s the interesting part to me, though.  I wouldn’t have been able to do any of this research if I didn’t happen to take a flexible methods course this semester with an amazingly helpful instructor: Jane Burris.  Not only is Jane not officially in sociology, but she doesn’t have any formal professorship standing – and she’s the best methods instructor I’ve had to date.  Why?  She guides the advanced students through whatever research they wish to do.  No contrived artificial projects on far away unrelated countries or pretend ethnographies on environments we all know too well – I was told to design my own survey, collect my own data, and work to analyze it for an eventual publication!  No other class in Sociology (that I know of) offers that kind of opportunity.  Thanks to Jane I’ve been able to not only start the Facebook Project, but start it with a little confidence about my methods.
There are drawbacks though.  I never really learned statistics. I mean sure, I can tell you about how to use and interpret a few statistical significance tests or even a little bit about linear or logarithmic regression.  But I have absolutely no idea how to employ which statistical tests to my own data.  In class we were taught the mechanics of a test, how to interpret the results – but not how and when to use it!  I’ve picked up some basics from the crosstabs and Pearon’s Chi-square material I teach in 380, but I want to know when I should isolate specific variables to determine a causal factor.  If I’m studying social capital and have all of these substantively defined concepts and conceptualized variables to represent this, how then do I take a statistics test and say something about them?  I’m not just talking oh look the median number of friends on Facebook – I’m talking about controlling for race and year in school to identify if gender alone significantly impacts the ways Facebook is used as a supplement to social capital!  We’re talking many variables that all intersect that I don’t know how to relate to one another with statistics in meaningful ways.  So enough complaining, I just wanted to give examples.  I’m hoping I can find a class or an individual who can tell me that Cronbach’s alpha would best illustrate the connections between my matrix of Facebook usage variables to say, perceptions on digital privacy.  So that is, designing a plan of analysis for my data.  I can coax a computer into doing the thuckethead statistics for me and check with a book to see what the results mean in technical terms.
A happy conclusion?  Jane has offered to help me sort through my data this summer.  I’m looking forward to seeing what results!