Scientism lives on

I was reading through Nate Silver’s blog (which I love) and came across an item about the recent flap about the closing of Chrysler dealerships under their reorganization plan. Some GOP pundits have made the claim that there is a relationship between the decision to close a particular dealership was based on whether or not they were Republican campaign contributors or Democratic campaign contributors.

One blogger on the GOP side even published some data (gasp!). the correlation matrix she published here is presented below together with her conclusions.

Singer’s blog presents a wonderful case of obscure statistics which she then goes on to completely ignore when they show a result with which she doesn’t agree. Why publish it in the first place? Obviously to give her piece, which is basically innuendo, the aura of scientific credibility.

Her conclusion number one:

“Why would there be a highly positive correlation between dealer survival and Clinton donors?”

There isn’t. The correlation is .59 with a sample size of 53. There is no correlation there at all much less a “highly positive” one.

Her conclusion number two:

“Nevertheless, it seems clear that something is going on here.”

Yes there is something going on, but it has nothing to do with this data. What is going on is political innuendo and a ‘guilt by association’ attack. Nothing new about that in politics, but throwing these data around (which few will read much less understand) is a giant smoke screen to mask the real intent.