The person who did this analysis, Marla Singer, assembled exactly the data set that I think needs to be assembled: a list of all of the Chrysler dealerships, both the ones that are remaining open and the ones that are closing, along with information about their owners' political contributions.

She then did an analysis that I think wasn't really the right one: she performed a multiparametric binary logistic regression, with donations to Hillary Clinton, Obama, McCain, the Democratic Party, the Republican Party, and nobody as independent variables, and likelihood of dealership closure as the dependent variable.

The result that most closely approached statistical significance was a tendency for Clinton donors' dealerships to stay open (p=0.125). However, Obama donors' dealerships were actually slightly (and insignificantly) less likely to stay open (p=0.509). Democratic donors' dealerships were also slightly (and insignificantly) less likely to stay open. And Republican donors' dealerships were more likely to stay open (p=0.676).

The bottom line: Singer reports no statistically significant findings in this preliminary analysis. She notes a tendency toward significance of Clinton donors' dealerships being more likely to stay open, but the findings for Obama, Democrat donors, and Republican donors, are in the opposite direction to that expected.

I still think that some of the anecdotal observations that directorblue has linked to are worth following up on, and I hope that Marla Singer will keep analyzing her data set, and post it, so that the rest of us can have a look at it. But so far, in my opinion, there's no reason to suspect a systematic bias at all.