Another case for skepticism.
Update: Lots of readers — while skeptical of the underlying story – are also skeptical of 538’s math. For instance, from a reader:
I’ve not followed this closely, but from what I gather, about 88% of Chrysler dealers are Republican donors and about 92% of closures are Republican donors.
That doesn’t sound too bad, until you realize that with a number as large as 1,000 closures, the numbers should be much closer.
I’m using round numbers skimmed from the blogosphere, but if you expect 880 out of 1000 to be Republican and the actual observation is 920, then the p-value is 0.001.
The numbers look worse if you parse it by Obama/not-Obama donors as opposed to Rep/Dem.
More importantly, the large numbers could hide a small number of truly egregious instances which have been thoroughly documented, with odds closer to 1 in a Million that the lucky Obama donor got to keep several dealerships while everyone else was sent packing.
The accusation is not obviously wrong, and more data is showing more support for it, not less.
This all reads like witchcraft to me, so I will bow out of the math discussion.