A Simple Swing in the Vote--with One Regional Exception

More from Andrew Gelman:

Comments

[Nathan Myers]

Huh, even Texas and Utah are on the trend line. But why isn't MS circled?

[dbt]

Don't lump Louisiana into that. The changes there are demographic, not electoral.

[Nathan Myers]

And aren't Alaska and Arizona both outside the region, and accounted for by the "native offspring" effect rather than entrenched racism? Curiously: compare to Illinois and Hawaii. Hawaii is way prouder of Barry than Illinois is.

The real oddity is Arkansas. Can somebody please explain Arkansas?

[ogmb]

If there were one regional exception, FL, NC, SC, GA and TX would also be circled. This graph requires a more complex explanation.

[bad Jim]

Hawaii is (HI) rather high on this graph, but I gather that the discussion concerns the state that fell below the line. The NY Times has a graph showing counties that voted more Republican this time than last: southern Appalachia, the Ozarks, coastal Louisiana, Texas and Oklahoma, Florida's panhandle.

It's not easy to find a link to the graph. Try this:

http://elections.nytimes.com/2008/results/president/map.html

[joel]

OGMB: I think you're looking at the wrong region.

The region that saw a flat or red-ward shift was the Appalachian/Ozark swath. Looking at county-level results as well (http://www.dailykos.com/images/user/28416/map2.jpg) makes this even more apparent.

In the Professor's graph, AZ, AK, and HI should probably be marked in a different color due to home-state advantage. TX shifted blue-ward for the reverse reason (Bush has been recipient of home-state advantage the last two cycles which masked some of the underlying demographic changes).

LA is its own kind of regional shift, with a high number of lean-blue voters being displaced into other locations -- Dallas and Houston were big recipients of this shift, which is part (but not all) of their shift left this cycle.

If I were advising President-Elect Obama, I would suggest a new TVA style green energy project in the heart of the Appalachia/Ozark corridor -- after passing a labor-friendly EFCA. This is a heavily economically depressed region where low-skill whites and blacks are in fierce competition for the next-to-the-worst paying jobs. Creating new good paying, labor friendly jobs in a progressive industry will go a long way toward repairing this region both economically and culturally.

[mike]

What's the with Arkansas? Is that a Clinton effect? Bill & Hill need to work on that...

[koganbot]

Also notice that six of the ten that shifted most toward Dem were in the West: Utah, Nevada, Nebraska, Montana, North Dakota, Idaho. Three of those still vote strongly Republican in presidential races (Utah, Nebraska, Idaho), but still there's a general shift in that part of the country, and it's not all demographic.

[ogmb]

Joel: The Appalachians include NC/SC and don't extend into Oklahoma.

That NY Times map bad Jim linked to is an eye-opener. If you compare it with a map of high African American population ratios what you get is that the red shift occurred in the area of the South with few or no blacks.* IOW, the Cracker Crescent:

http://www.census.gov/geo/www/mapGallery/images/black.jpg

*As noted, Louisiana is an aberration due to black emigration after Katrina.

[Chris]

ogmb's explanation is good, except for one thing: southeast Arkansas has quite a few blacks, but still shifted heavily away from Obama. Also, there is a sharp discontinuity at the AR-LA border although demographics are similar on both sides.

IOW, southeastern AR (with significant black population) shifted sharply away from Obama while very similar northeastern LA didn't.

Perhaps this has something to do with a certain famous ex-governor of Arkansas? McCain sought the PUMAs in vain in most places, but maybe in Arkansas there really were some?

[Nathan Myers]

I am very impressed at joel's note about the reverse home state effect on Texas. So, Texas appearing on the upper trend line is as illusory as Arizona and Alaska appearing on the lower one. In this case the racism effect is very nearly exactly commensurate with the home-state effect. (Hawaii excepted.)

Perhaps this is because the psychological mechanism is actually the same for both?

Hawaii deserves investigation. Hawaii is not heavily populated with transplanted Texans (although I knew a few). Most of the population is at least part non-white, increasingly so, though there are very few black people -- lots of Japanese, lots of Filipino, lots of Portuguese (who do not consider themselves white), lots of Chinese. Education and eloquence aren't well respected, but if they watched the debates, they would have been weirded out by McCain's tics. People in Hawaii are very unforgiving of spazziness. It would be very interesting to see how the debates affected polling in Hawaii.

[ogmb]

SE Arkansas might be part Clinton, part black migration, part disenfranchisement.

[ogmb]

TX likely shifted blue because of Latino voters:

http://www.census.gov/geo/www/mapGallery/images/hispanic.jpg

Hawaii is Obama home turf. Boston is red because it was Kerry's home turf.

[Kolohe]

Nathan Meyers-

Hawaii is full of pro-(department of)defense culturally consersative Democrats (hence the rather narrow 53-47 Kerry over Bush win) who have been totally alienated by Bush incompentence and, even more than that, love a home town boy (or girl) making good on a national stage (see also Jasmine Trias of American Idol a few years ago)

[jonp72]

Obama's cool-as-a-cucumber personality also dovetails nicely with Hawaii's "hang loose" ethos.

[Joshua Herring]

I notice that MA is not circled and should be. It's closer to the trend line than both AZ and AL, both of which are circled.

Most of the discussion here seems to want to conclude racism. I would suggest that in the two major cases there are alternate explanations. In Arkansas, as another commenter suggests, it may well be a lingering Hillary effect. In West Virginia (and indeed the Appalachian regions of Kentucky as well) it is highly likely that this is a coal vote. One need not be employeed in coal directly to worry about the policies of a president who has strongly implied he wants coal out of business for environmental reasons. In an election that was generally characterized by voters worrying about the future of their jobs, it should surprise anyone that West Virginians would be worried about the future of their jobs.

As for the racism explanation in general, Alabama, which has a lot of blacks, is close to the trend line. Georgia, which has a lot of blacks, is not. Missouri, which clearly patterns with the circled group, is not circled. And no one has explained Massachussetts.

All told, racism may have something to do with it, but this case is far from being made. As usual, the explanation is probably a lot more complicated and would involve a lot more work to extract - but hey, playing with R is fun, and it's always uplifting to think the worst about our neighbors, so why not, right?

[ogmb]

MA was Kerry's home state. He received supernormal support there, which returned to normal liberal support plus the nationwide swing. We're down to three explanations:

1. White South (OK, AR, TN, NE TX, N AL, E KY, FL panh, SW WV, W PA -- but not VA, NC, SC, MS, rest of AL and TX; GA is ambiguous)
2. Black migration (S LA and MS)
3. Hometown effects (IL, DE, HI, AZ, AK?, MA, TX?)

A fourth might be disenfranchisement, when voters realized their vote would not sway the outcome, which might have affected the West Coast, but the pattern is unclear. Also, "race" is not the same as "racism".

[ogmb]

Oh, and you should look at the NY Times map, not the Gelman chart, which is too coarse-grained, and compare it to the minority maps from census.gov.

[Chris Kromm]

You're exactly right -- the Great Plains are solid GOP.

[JonF]

Re: The Appalachians include NC/SC and don't extend into Oklahoma.

In demographic terms, much of the population of eastern OK (AKA "Little Dixie") is Appalachia/Ozark derived. Certainly the cultural attitudes have carried over.
Very little of SC is in Appalachia proper. A large portion of NC is, but NC, like VA and of course FL, has benefitted from major in-migration from other part of the country.

Re: lots of Portuguese (who do not consider themselves white)

??
If you mean "Brazilians" then that would be sensible I suppose. But if you mean European Portuguese, that's just bizarre.

Re: White South (OK, AR, TN, NE TX, N AL, E KY, FL panh, SW WV, W PA -- but not VA, NC, SC, MS, rest of AL and TX; GA is ambiguous)

Nebraska is not in any sense a "Southern" state. Also, I don't think western Pennsylvania patterns well with the South (white or otherwise) either. I would however stick the far western parts of Virginia and North Carolina (Asheville is the exception) in that cultural pot. Northern Georgia may once have belonged there too, but has seen a large influx of people from parts distant, diluting its historical character.

[johne]

I think Nathan was talking about modern, Portuguese-descended Hawaiians, JonF. Their forebears would have had it pressed on them, by Hawaii's white, American elite, that they were not white, though perhaps a cut above the Mexican charros, who had also been brought to the Islands.

[ogmb]

"In demographic terms, much of the population of eastern OK (AKA "Little Dixie") is Appalachia/Ozark derived"

Geologically, there are five areas of the Appalachians: Northern, Piedmont, Blue Ridge, Ridge & Valley, and Plateau. The only area for which an argument that Appalachia = redshift can be made is the Plateau (Western slope). Northern, Piedmont, Blue Ridge and Ridge & Valley have mostly gone blue (roughly a N/S split, with the North going blue and ambiguous shifts in the South), only the Plateau has mostly shifted towards red, but by no means uniformly so. "Appalachia" as an explanation for the redshift is a non-starter. If you're trying to find cultural pockets with Appalachian heritage to make the explanation stick you're simply reaching.

"Nebraska is not in any sense a "Southern" state."

NE TX = Northeastern Texas. Nebraska has shifted towards blue. There are about three counties in SW PA that have shifted slightly red, that's simply negligible. Other than the redshift in Southern Arkansas I have not seen a serious challenge to my proposition.

[ogmb]

"much of the population of eastern OK (AKA "Little Dixie") is Appalachia/Ozark derived"

Also, a plurality of E OK is Native American, a demographic McCain heavily courted. Most of the remaining red shifts in the West occurred in Native American counties.

[JonF]

Re: If you're trying to find cultural pockets with Appalachian heritage to make the explanation stick you're simply reaching.

Voting is far more about culture than it is about geology. I agree that the poster who said "Appalachia" was using the wrong terminology, a geological/geographical term when what he wanted was a cultural one. That term is best given as "Scots-Irish" (see "Albion's Seed"). The red shifted region in the upper/mountain South is a territory that was heavily settled by Calvinistic Scots-Irish, who then migrated west through Tennessee and Kentucky, into Arkansas and S. Missouri, thence into NE Texas and E Oklahoma. And this migration has been largely undiluted by later population fluxes except in certain locations (e.g, Asheville NC). This particular ethnic group is vulnerable to ethnic hatreds (back in the Old Country too: the Ulstermen of Ireland are notorious for their anti-Catholic hatred long after everyone else in Europe forgave and forgot the Reformation-era battles), as well as hellfire-and brimstone religiosity (the Calvinism again, unmodified by Enlightenment rationality which tamed and softened the Calvinists of England, Scotland proper and New England).
In short, this is what unifies the whole of that red-shifted region, from Wheeling down to Tulsa.

[ogmb]

Of course Appalachia is a geological term, but it's perfectly applicable to cultural phenomena as long as those are confined to the geological area. Geology often defines culture. In this case it's just incorrect.

As about your Scots-Irish theory, I can't parse this. Self-reported Scots-Irish ancestry is centered around North Carolina, Virginia and Tennessee. Other than TN those went blue, along with the multitude of non-Southern Scots-Irish centers. On the other hand, self-reported "American" ancestry matches quite closely with the red-shift area. Whether it is the Scots-Irish element within that hodgepodge I can't ascertain without DNA analysis.

http://en.wikipedia.org/wiki/Image:Scotch_irish1346.gif
http://en.wikipedia.org/wiki/Image:American1346.gif
http://en.wikipedia.org/wiki/Image:Census-2000-Data-Top-US-Ancestries-by-County.jpg

[Carly Corday]

"If I were advising President-Elect Obama, I would suggest a new TVA style green energy project in the heart of the Appalachia/Ozark corridor -- after passing a labor-friendly EFCA. This is a heavily economically depressed region where low-skill whites and blacks are in fierce competition for the next-to-the-worst paying jobs."

Reporting from the Ozarks, North Central Arkansas: You got that right! One son (Southern Missouri) with 3 years college and a skill certification that once ensured good pay has been cut back to 3-4 days a week, 8 hour days instead of 10 or 12. He's gone from working himself ragged to worrying himself sick. Another son (right here) is now on 3-day weeks, with worse to come--he will be off work for the entire month of December.

We are dying here, rather suddenly.

We all 3, white and Southern, voted for Obama. He is the answer IF there is one. We also want the war ended in Iraq, where a third son, the youngest, has already served and will have to serve again in 2009.