Tuesday, May 24, 2011

Lee's Correlation of Everything with Everything

If you haven't found it already, walk-don't-run over to Overcoming Bias for a preview of the best dataporn datamine I've seen in quite a while.  That would be the Ken Lee's  (George Mason) thesis,* inadequately captioned as "How States Vary," not much less inadequately characterized as "the correlation of everything."  As described by his supervisor, Robin Hanson:

Ken collected 81 features of states, 56 cultural rankings and 25 demographic variables ...), and did a factor an analysis on them. 
One--word answer: health, and Robin is doing some followup analysis of the health findings right now (three-word advice: skip cancer screening (link; cf. link)).   The whole project seems almost like a spinoff of  a Hanson blog post of a few months back, where he divided us all into farmers and foragers. The earlier post kicked off a lovely comment thread, including those who think he is totally bollox and and those who just want to tweak the model (maybe three classes?  Farmers, foragers and nomadic herdsmen?) (good comments on the new item too).  One thing you can't easily sidestep: Lee's maps, suggesting that something is going on. 
*Idle curiosity: why does GMU call it a "thesis" rather than a "dissertation?"

No comments: