On the occasion of Veterans Day, Facing South last week had a post up about veterans in the South and veteran care. Part of the post was a list of the “10 Cities with Highest Percentage of Veterans”. Nine turn out to be in the South. 

It made me curious: Southern cities with a high percentage of veterans, those can’t have been the most promising locales for the Obama surge, can they? The lone non-Southern city was the conservative redoubt of Colorado Springs, after all.

Looking up the results for the counties in question yielded an unexpected mish-mash of votes, however.

First, here is the list of the top 10 cities and the counties they are in – note that in Virginia, the cities are their own counties. (For a methodological note, see footnote 1).)

Table 1: Top 10 cities with highest percentage of veterans in 2000

Now for the election results from 2004 and this year in those top 10 cities that had the highest share of veterans in 2000 (respectively the counties they are in). As said, it’s a very mixed picture:

Table 2: Top 10 cities with highest percentage of veterans in 2000 (resp. the county they are in): how did they vote in 2004 and 2008?

If you’re an election geek like us, you’ll have seen this electoral map from the NYT. It shows which counties in the US actually shifted toward McCain, in comparison with how they voted in 2004. (The map showing which counties shifted by how much to Obama is interesting too.)

Since the country as a whole saw a 9% swing to the Democrat, it’s just a small part of the country that moved toward McCain, obviously. Just 22% of counties, as the Times helpfully notes. But their geographical concentration is noteworthy, as apart from obvious bits in Arizona and Alaska, the candidates’ home states, most of the counties in question form a perfect arc in the Highland South, from Oklahoma eastwards to Tennessee and then upwards through the Appalachians.

Striking as the pattern is, however, it’s become fodder for some misinterpretation as it did the rounds on the blogs. Some of it may just be a matter of emphasis. Some of it, however, has to do with the way the differing racial demographic balances in red states cloak the true concentration of McCain switch voters.

In terms of general emphasis, I’d be a bit wary about impressions when these counties become dubbed “the McCain belt” — you’d almost think that these were the best counties for McCain, rather than just the ones that moved toward him most. For example, McCain won Alabama and Louisiana by about 20 points, a more ample margin than he got in Tennessee, Kentucky or West-Virginia. So what’s the real McCain Belt?

The more interesting point is about race. The NYT map showing the electoral shifts to McCain obviously does not take into account the role of race, it just maps the overall results. One thing, however, that distinguishes the Appalachians is that they have a very small black population. In the Deep South, on the other hand, you have some of the largest black minorities around. Those black populations turned out en masse for Obama — and so their extra votes for Obama effectively canceled out the shift to McCain among whites there.

Do Southern whites constitute the real McCain Belt?

Compare the Electoral Shifts map above, with its “McCain belt” stretching from the Oklahoma to the Appalachians, with this one:

How has the white vote shifted between 2004 and 2008? In this map, McCain getting 25% more of the white vote in a state than Bush got in '04 would colour the state a fiery red; McCain getting 25% less would make it the coolest blue. The map shows that whites in much of the Deep South swung to McCain, while whites in the Mountain and Pacific West, the Midwest and the Atlantic South swung strongly to Obama.

This map shows, state by state, how much the white vote, taken separately, changed since 2004. It looks very different, doesn’t it?

Continuing on the previous post, which covered basic demographic categories of gender, race, age, income, education and party ID, here are several other side-by-side comparisons between the exit poll data on the 2000, 2004 and 2008 presidential elections.

Among which groups has Obama done better or worse, and by how much, than Kerry and Gore did? A look at first-time voters, religious groups, married versus unmarried voters, union households and gun-owning households, urban, suburban and rural voters, and voters from the different regions of the country.

When looking at these charts, keep the overall, national data in mind. Gore got 48.4% of the vote, Kerry 48.3% and Obama 52.6% – so that’s the standard. If Obama gained 5% or more in a demographic group compared to Kerry and Gore, it means he made bigger advances in this group than on average; if he gained 3% or less, it means he “underperformed” in comparison with other demographic groups.

FIRST TIME VOTERS

Share of voters: 9% in 2000; 11% in 2004; 11% in 2008.

Yes, that’s one huge blue victory in 2008 – the contrast with previous cycles, in which the Democratic candidate already had the advantage, is enormous. It’s an advance that dwarfs all others in this overview.

PROTESTANTS

Share of voters: 54% in 2000; 54% in 2004; 54% in 2008.

Note that the increased turnout that Obama inspired among African-Americans (and, presumably, a corresponding decreased turnout among the white evangelical vote Bush mobilised so successfully in 2004) should have helped amplify Obama’s gains among Protestants.

AP and NBC yesterday belatedly called North Carolina for Obama, making the state’s result the second last to come in. Only Missouri hadn’t been called yet. So how did it become so close? Facing South has a good summary up of the main strategical and political reasons. But I would like to look more specifically at the geography and demographics of the race.

For Obama to win the state required a 12.4% swing (that being the margin by which Bush was elected in 2004). He got a 12.6% swing. Which parts of the state pushed Obama over the line? Where did his efforts of persuasion fall short? What demographics were at play? An in-depth look.

• An Observer’s Guide to Election Night, by our own JoefromChicago. There are 435 congressional races, 35 senate races, and 11 gubernatorial contests today, Joe points out, and “these races merit attention in their own right, but they also may be early indicators of the way the presidential race will turn out”. A convenient list of the races you should be paying attention to as you watch the presidential results come in, ordered by time of night.

   

• Swing State Project: Poll closing times & Key races, by DavidNYC. Very handy country map with the poll closing times, and a list of key House, Senate, gubernatorial and state legislative races, arranged by poll closing times (linked in is also a list of key ballot measures).

   

• The American Prospect 2008 Election Night Guide. Comprehensive guide, encompassing six sections, among which an overview of key swing state counties, a list of Senate races that would pave the way to a utopian 60 Dems, a review of bellwether House races, and a number of ballot initiatives to watch.

   

• US Election Atlas, by Dave Leip. A long-standing, invaluable elections resource. Browse the results of previous elections going back to 1789 (no, really). Not just by state – results by individual county are available back to 1960.

   

• Google Maps Historical Election Results, going back to 1980. Click or zoom into a state or county – the map will show you the winner and moreover, with a click of the mouse you get both the electoral breakdown and basic demographic data (income, age, race/ethnicity). Click on the National Almanac map, which should show up on the right, and you can find additional demographic info on language and occupation by state or group of states. The Google Maps historical election results tool should also be available on Google Earth (h/t Marc Ambinder).

   

• Census 2000 Interactive map. Zoom in on states and counties to find demographical data on population density, racial composition and black or hispanic population.

   

• Ancestries by state, tables derives from the US Census Bureau’s American FactFinder. OK, so only very tangentially relevant, but very interesting. (Keep an eye on those Hillary-loving people of “United States American” – i.e., non-ethnic – ancestry in the Appalachians and the Border South, which are very unlikely to go Obama.)

   

• Cleveland Plain Dealer Data Central, for all your political data from the state of Ohio. Aside from the interactive map of county-level results back to 1960, which overlaps with the above resources, there’s an interactive map that breaks down the voter registration and demographics on both congressional district, county and ZIPcode level. And a map and table showing how voter registration has gone up or down since 2004 by county. And a useful explanation, with map, of how Ohio may be a swing state, but inside Ohio there really are but a dozen of swing counties.

   

• Counties to watch according to Marc Ambinder in Pennsylvania, North Carolina, Nevada and Colorado and in Florida, Virginia, Ohio.
   

While the reporters assigned to presidential candidates are condemned to a mix of grind and hype, reporters who get the chance to survey the country often come up with the best stories. Interviewing voters, sampling local opinion, sketching the political geography, they write the field reports, a ubiquitous genre of its own. No self-respecting election-time newspaper is complete without one.

A lot of them, of course, end up being cookie-cutter stuff: a little bit of this, a little bit of that, and done is the day’s work. Here’s a few from the last couple of days I thought more interesting (h/t to the Electoral Map, where I think I found most of these on the “Morning Reading Lists”). The common thread: Obama’s chances to win over red states or counties.

Battling on the Other Side’s Turf
Washington Post
1 November

Best of the bunch: in-deph local flavour from Southside Virginia.

Heartwarming? Yes; moving stories, a hopeful narrative, characters who feel real and alive. Feeling that the reporter understands the local scene? Pretty high. Strategic survey? Not so much; demographic analysis doesnt take much space here. Evocation of historical legacies, racial and/or industrial? Present. Nuanced? Yes. Topical? Yes – US Rep. Virgil Goode, portrayed here as a well-established incumbent, is unexpectedly facing a tight race, according to the latest polls.

Why the New Virginia Is Leaning Toward Obama
Time
27 October

Timely dispatch from exurban Prince William County.

Heartwarming? Not so much. More of an analytical take, and what anecdotes are there are fairly depressing. Feeling that the reporter understands the local scene? High enough; he studied the numbers and knows where to look. Strategic survey? Yes. The choice of location itself is an attempt to pinpoint the very frontier zone where the elections will be decided. Evocation of historical legacies, racial and/or industrial? Not so much. Nuanced? Yes. Topical? There’s the account of a fearmongering McCain coordinator, but you might have seen it already.

Obamalina
The Nation
22 October

Long review of how the Obama campaign made North Carolina into a toss-up state.

Heartwarming? In a combative way. Fuelled more by awe at the campaign’s success than touching personal anecdotes. Feeling that the reporter understands the local scene? High. He’s from there and he’s got the political scars to prove it. Strategic survey? Yes: it’s all about pinning down the overall Obama strategy and why it’s successful (“it’s the economy, stupid”). Evocation of historical legacies, racial and/or industrial? Yes, but of the weary rather than wistful type. Nuanced? Not so much. Topical? No immediate hook beyond the electoral fate of the state itself.

Westmoreland County up for grabs
Pittsburgh Post-Gazette
28 October

Gritty impressions from a Pennsylvanian county that went for Bush and loved Hillary.

Heartwarming? Not really. The news is depressing – meet the working class McCain Democrats. But the people are real. Feeling that the reporter understands the local scene? Mwah. You keep wanting her to dig a little deeper. Strategic survey? Not so much. Though the description of how the polarisation between high-income, Republican subdivisions along Route 30 and low-income, Democratic riverside settlements are all muddled up this year should rouse the political geographer in you. Evocation of historical legacies, racial and/or industrial? Some wistful type of the latter. Nuanced? Fairly. Topical? Considering McCain’s decision to stake his fate on Pennsylvania and the racial/cultural resentments there, yes.

A propos of nothing in particular, an electoral map of sorts.

I selected all the states where McCain is currently leading, if even by the narrowest of margins, and painted them red (using the gadget at 270towin.com). To be generous and cautious, I actually used the pollster.com numbers from a week ago, when he was still leading in Missouri and North Dakota as well. And I selected the states that are absolutely foolproof safe for Obama and painted them blue.

These two selections count for almost exactly the same share of the Electoral College. These two selections represent roughly the same proportions of the US population.

A useful map, then, perhaps, to have at hand for two occasions:

a) Whenever someone harangues you about “the real America”, “heartland America” or “flyover country, where Joe Sixpack lives”.

Check: the Bos-Wash corridor with upstate NY and Vermont; Illinois; and the Pacific coast with Hawaii – together those states have as many Americans as all the even marginally red states together.

b) When you want to wonder at how unfavourable the underlying fundamentals of this race are for McCain.

Normally you start from a base level of reliable support, and then contest as many of the few remaining battleground states as you need to win. But this base level is just precariously low for McCain this year. Mostly because of a few givens: Bush’s unpopularity, the economic crisis and the loss of trust in the Republican brand on the economy, the unpopularity of the Iraq war. (And that’s not just “headwind”, as you’ll now find some conservatives describing it; it’s the result of Republican mismanagement.) But it’s also because of the Obama campaign’s willingness to reach far into red-state America and its access to the resources to do so, a testament of Democratic enthusiasm.

Either way, North Carolina, Virginia, Colorado and Florida are not part of the base level of support that Republicans can build on like they were in 2004. Which means that the Republican base level is as low as 185 Electoral College votes, rather than 249. And just 185 EVs? That’s so little that it barely counterweighs even the safest of safest blue states.

Again, nothing particularly newsworthy about any of this, but I still found it a pretty telling map.

On his blog, Brian Beutler remarked upon the difference between the popular perception of California as a bastion of liberal group think and the reality:

California’s a much different kind of “blue” state than is, say, Massachusetts. The dense population centers outside of San Diego and Orange counties are liberal enough to give California’s electoral votes to the Democrats every four years. But for the most part the rest of the state is bright red.

He emphasised the stark contrast between blue and red counties and concluded that in this sense, aside from the San Diego and Orange counties, California “rightfully belongs” in the same category as Oregon and Washington.

While praising Beutler’s post, Ezra Klein offers a somewhat different take. There may be a real contrast between the blue coast and the red inlands, but what it’s informed by is primarily ethnic demography:

The state’s political transformation in recent years has been somewhat ideological, but it’s been much more demographic. Namely, it’s been driven by Latino immigration. Folks think of California and conflate its politics with San Francisco and Hollywood. White, affluent, cultural liberals. But that’s not why California is reliably blue. In 2004, Bush had a five percent margin among white voters.

This sets California apart from a state like Washington, he continues:

In the aggregate, whites everywhere are somewhat conservative. But in other liberal states, they really do swing left. In Washington, Kerry had a six percent advantage among whites. In Vermont, he had an 18 percent advantage. [..] California, by contrast, is a very Democratic state, but somewhat less coherently liberal. It’s solid blue because Latinos are solid blue, not because the place is packed with liberals.

This had me thinking. On a national electoral map, when placed on a scale from clear blue to bright red, California and Washington are the same pale blue. But if the white vote in those states differs so clearly, does it look different elsewhere too? How different would the map of red and blue states look when only showing the white vote?

The 2004 Presidential election – national vote (all groups)

This map of the 2004 election results is not the usual type: the country is not artificially divided up between "blue" and "red" states. Instead, it shows the degrees in between. A state where Bush won 100% would be fiery red, a state where Kerry won every vote would be the coolest blue, and a state where the vote was divided equally is white.

Read on and view the map for white voters only beneath the fold.

Trivia absurdism of the day. At Beyond Red and Blue, Robert David Sullivan last month did some serious research into bellwether states:

I calculated the percentage-point differences between each county’s swing and the nationwide swing for each election from 1980 through 2004, then added them all up to find out the places that have deviated the least from the US total over that time. (For example, there was a swing toward the GOP and George W. Bush of 2.86 points in the last election. A county that swing 12.86 points toward Bush and a county that swung 7.14 points away from Bush would each be penalized 10 points for that election.)

The result is a map, list and Excel spreadsheet of the Top 50 Bellwether Counties, 1980-2004.

Which county is #1 — the single most conformist county in matching the country’s overall swing for the last seven elections?

Defiance, OH.

Who voted for the “bailout” bill? (Talking of lousy branding.) More relevantly, who voted against it? Ever since it all fell apart, pundits are dissecting consequences and solutions, but also simply the lay-out of the vote itself. How did the expected majority collapse? Who defected? Are there any patterns?

A couple of basic ones have been presented, beyond the partisan breakdown of Democratic and Republican votes. Since two-thirds of Republicans and 40% of Democrats voted no, the landscape is more interesting than usual.

Breakdown one: those in more or less safe seats versus those in vulnerable seats. Representatives who are facing a tough fight this campaign, or who were elected last time by a narrow margin, were much more likely to vote against. Two: ideology. A rebellion of the rock-ribbed conservatives in the Republican Party, and a lesser one of liberals in the Democratic Party. Three, and mentioned less often: those who are in or close to the party top or House Committees versus the rank and file.

You’ve read all this, though there are more twists to it than you might think. What I was wondering was whether there was a fourth axis: geography. How do the votes from the different regions stack up? A question of combining roll call 674 with Wikipedia’s list of US Representatives by state. Green stands for “yes” votes, red for “noes”:

Republican vote by region

There are significant variations by region — and by state. Republican opposition was strongest in Texas, which delivered 15 “noes”, the Southwest and the Plains. Of the 13 remaining republicans in the Northeast, however, a majority voted in favour.

The Democratic vote varied at least as greatly by region. The Democrats in the Northeast voted in favour by almost 3:1. But more than two-thirds of the Democrats in the Southwest and Mountain states voted against.

Democratic vote by region