by on January 12, 2009 · 6 comments

in Uncategorized

Since everybody loves maps, I thought I would preview two from the Forum article on the 2008 election that I co-authored with Tom Schaller. The first is a map of counties where McCain improved on George W. Bush’s 2004 vote percentage. The second is a map of counties where native southern whites make up 65 percent or more of the population.

From Monkey Cage

From Monkey Cage


King Politics January 12, 2009 at 12:51 pm

It doesn’t look good for Texas, Tennessee, Kentucky or W. VA does it? In Tennessee’s defense, not that I’m from there, but there was a tightly fought, and well publicized, race for partisan control of the state legislature that likely produced higher turnout.

Chris January 12, 2009 at 3:10 pm

These maps are fun, but we can read too much into them. Remember the ecological fallacy. The map makes it appear that the state of Tennessee is more Republican this year than 4 years ago. The exit polls (individual-level data), reveal that Obama improved in Tennessee with urban and suburban whites, and did worse with rural whites (than did Kerry). Statewide, both received the same % of the vote in ’04 and ’08.

Beware the ecological fallacy, even if it comes with cool maps.

Phil Klinkner January 12, 2009 at 3:20 pm

The ecological fallacy is always there, but your evidence for TN seems to conform with our observations. My guess is that native southern whites are most likely to be found in rural TN, exactly where you say Obama did worse than Kerry. Also, we did a regression model in the article and controlled for the percent of native southern whites. The variable was significant.

Finally, where did you find exit poll data with race and location?

Chris January 13, 2009 at 10:30 am

My initial comment was mostly in response to the previous poster, not the original post and article.

However, your abstract overstates what the data show. In your abstract, you state that Obama was elected with a coalition of upscale whites and minorities. This is probably more correct for the primary than the general. The evidence suggests a number of low-income whites, very high-income whites, and minorities supported Obama.

In regards to the South, I think you can overstate your argument as well. Obama did better there, as you point out, especially in Rim South states. But you focus on the racial divisions in the South, when it is really in Deep South states. The racial component exists in Deep South states like AL, based on the exit polls. But not in NC and VA (where Obama improves among whites relative to Kerry) and even in places like TN (where there is no meaningful change among white voters between ’04 and ’08).

By focusing on the county change in small counties in states like TN and WV, it is the equivalent of treating a 1000-person county the same as Memphis or Charleston, WV.

I think it is an interesting project, though and we need to assess what happened in ’08. But county level changes in counties with a total voter turnout of 1000 people only can add so much. Individual-level exit polls are key.

Phil Klinkner January 14, 2009 at 9:37 am

Yes, the maps don’t weight by county size, but in the regression models, we weighted for size of county.

Comments on this entry are closed.

{ 1 trackback }

Previous post:

Next post: