Aug 11, 2023·edited Aug 11, 2023

Re Perry Meko and Uhrmacher

Now regress the small town two party vote share versus %Black among VAP. I'd be quite fascinated by that regression, actually. Then even more, a principal components analysis: e.g. just KL transform the raw votes against whatever other demographic data the Census easily provides for those small towns.

Especially if you try to norm the small town reaction (i.e. two party margin) versus the local metro's two party vote.

Expand full comment

I should have the data to do this. Is there a state you had in mind?

Expand full comment

I think the two most interesting would probably be GA and/or NC. Both combine a reasonable amount of rural Black population with some pretty vibrantly growing metros, and of course they are at the state level at least arguably competitive between D and R, as opposed to the other former Confederate states.

NC is probably more polycentric than GA - certainly it is in terms of urbanization: no primate city like Atlanta.

I'm not sure what other data is easily available from the Census for those small towns. One vector in the data set that might be able to be standardized into a scalar would be something having to do with the age pyramid in the small town - proxy for whether the town is producing employment (raw population growth/shrinkage per census for each town should I assume be part of the basic Census dataset - maybe run through a zero phase low pass filter with a 10-20 year time constant to catch longterm trends).

Expand full comment