Saturday, February 23, 2008

Pollster.com on Ohio vs Wisconsin

A post on Pollster considers the following puzzle: Ohio and Wisconsin are demographically similar and the differences would tend to militate in favor of Obama (more urban, larger Black population), so why is Clinton doing so much better in Ohio polls than she did in Wisconsin leading up to the primary?

The author, who declares himself an Ohio native, breaks down the demographics and comes up with the provisional half-answer that:

    The answer, for the moment, appears to stem mostly from her continuing strength among Ohio's downscale white Democrats. In Wisconsin, as noted above, Obama ran slightly ahead of Clinton among less-educated white voters. However, in both the Quinnipiac poll conducted two weeks ago and the ABC/Washington Post poll done earlier this week, Clinton continues to hold an enormous lead among less-educated white voters.
Which leaves the obvious follow up: Why is Clinton doing better in that cohort in Ohio than in Wisconsin. My hypothesis -- trotted out in this brief exchange with Redhorse -- is the influence of Appalachian culture in the Ohio electorate. From speaking with people who study Appalachia, verified by my personal observations living around here, is that as a whole1 Appalachians tend to be suspicious of institutions, outsiders and people who are different. Appalachians tend to respond well to populist appeals -- such as those Hillary has been making -- and they have a tendency toward less-than-progressive attitudes about race.

The regional cross-tabs in the Survey USA poll offer equivocal support for my hypothesis. On the one hand, Obama's support is by far the lowest (24%) in SE Ohio -- the part of the state that is physically within Appalachia. On the other hand, Hillary has the strongest support in the Columbus and Toledo areas. I know Columbus has a fairly significant population of Appalachian migrants -- don't know about Toledo.

There's a number of interesting studies one could do with this. I don't have answers, just a couple of suggestions.


1Yes, I am generalizing here. That's what you do when you try to predict the behavior of a group. Everything I've seen says the generalization is valid. The danger comes when one tries to use a generalization to predict the behavior of an individual. When I meet someone whose parent moved up from the hills of Kentucky I try not to assume that he/she harbors anti-Black attitudes. This is dicey business given that a) the former is valid social science, but the latter is bigotry and b) that sort of bigotry can influence the generalized assumptions in the first place, throwing the whole enterprise into a soup of elitist snobbery. Nonetheless, it doesn't do to pretend the issue doesn't exist in the presence of evidence that it does.

3 comments:

Jill said...

One follow-up I would ask, Scott, is: what's the voter turnout data for the Appalachian population in Ohio? Is it significant enough to have an impact?

Scott Piepho said...

What I have done here -- without stopping long enough to realize it's what I was doing -- is propose studying Appalachians as a separate voting cohort. One of the first stumbling blocks to that would be definition. We could keep it regional, in which case note that the SE Ohio region itself only accounts to 3% of SUSA's sample. But that ignores a significant Appalachian diaspora spread throughout the state.

All of which is to say, I don't know, but it's hard to believe that the Appalachian vote writ large is not significant. And in particular, it's likely to be a significant part of that "downscale white" cohort that Mark Blumenthal, the Pollster.com poster, identifies as the source of Hillary's strength.

Anonymous said...

Intersting stuff. So does this explain Ted's unwavering support for Clinton? Also, is there any polling data for Akron and Canton? Both have significant populations that originated from Appalachia. More so as a percent than Columbus me thinks.