In this diary series, we've cycled through just about every ethnic group with available data. But there's one left: "Americans" - and the quotation marks are there for a reason.
In practice I am defining this group as whites who did not list an ancestry or listed American or United States as ancestry on the census form. Very few non-whites list American as an ancestry. There is also a correlation between the percent whites who list American ancestry and whites who do not list any ancestry, which is why I lumped them all together.
Who are these "Americans"? We can't tell, but there's lots of possible reasons to fall into this category: too many ancestries to list, unknown ancestry, patriotism, annoyance with labels, privacy concerns, getting bored filling out the census form, and sheer contrariness come to mind.
No matter, one thing is clear: nationwide, on average, the more whites describe themselves as "Americans" in a county, the less support for Obama among whites.
Ten Second Summary
Those who identify as "American" generally did not support Obama very strongly, and typically less so than other whites in their region; this is true to varying degrees in different parts of the country. There are two exceptions. First, the Northeast, where "Americans" supported Obama. Second, Appalachia, where support was split, and greater than support from other whites.
Support Among Whites
This analysis relies on location-specific estimates of Obama's support among non-white racial categories to estimate Obama's support among whites in each county. This estimate gets shaky when the percent white is below around 75%, and pretty lousy below 50%. The map below shows all the results - negative values (I told you the estimate can get shaky!) are plotted as zero. Counties less than 50% white will not be considered further for this analysis.
When we plot percent support for Obama among whites against percent of whites who are "American" in individual states, most states show graphs that are rather dull, partly because many states have a small range of variation in the percent "American." Here's an exception - the state with the best regression, Florida:
Florida shows a very wide range in values on the x-axis that could be attributed to a rural/urban, born-and-raised/retiree cultural split.
First, where are these "Americans" located?
There's not a uniform distribution, nor is their uniform behavior, as one might expect from such a catch-all category.
In fact, we can loosely group states and regions according to where on the original plot their data fall, although some areas could easily fall into several categories. When we do this, we find remarkable differences across the country, although the regressions are a little rough. Here are the results of these groupings, just showing the regressions:
You may notice something very strange: only one line is going steeply up. In Central Appalachia, support for Obama increases as the population becomes more "American." Elsewhere it decreases or is flat (the Northeast). Yes, folks, this part of Appalachia is, indeed, different from the rest of the country - because they were far more supportive of Obama than we might have expected based on this variable. The Northeast also stands out, in that support for Obama among whites is essentially independent of percent "American" - and high.
We can use these graphs to generate support levels for different groups of "Americans" based on their (rather poorly-defined) region. These regions are not the same as what is plotted above, but related. For example, both Old South and New South have at one endpoint of their regression very high levels of white 'Americans' with very low support for Obama, dubbed here Confederacy 'Americans'. The difference between Old South and New South is at the other end of the regression; Old South has low support among whites who identify an ancestry, but New South has high support among those who identify an ancestry.
Defining the Regions
Let's take a closer look at the regions. Each map below shows how well one of the lines above can predict Obama's performance. We're looking for yellow. If counties on the Yankee map are yellow, it means they sit pretty close to the Yankee regression line. Green means a pretty poor prediction, and blue means lousy. The numbers refer to the residual - the difference between what the lines predict, and what the election results were.
Starting in the Northeast, many counties from Maryland on up to Maine sit on the line. There's a concentration of counties in New Jersey, Pennsylvania, and New York, however. If we look at New England - essentially a more pro-Obama subset of Yankee - we see, well, New England. An area more or less on the West of the Connecticut River (Vermont and Western Massachusetts) is mostly green/blue - this area is even more pro-Obama.
Moving South and West, the Heartland region has some pretty sharp boundaries on the West. There's low support for Obama among "Americans" here, but not abysmally low. We see some old friends show up as blue spots here and there - the counties with high concentrations of those with Dutch ancestry, for instance. Note especially how state borders are not always followed. The border states on the Southern side are mostly green - meaning they don't fit all that well on the Heartland line.
So we move on South to Real America. This regression describes mainly areas of Appalachia and the Upland South on over to the Ozarks (and parts of Florida?). Support for Obama among "Americans" here was higher than in the Heartland. But we see some blue patches in Central Appalachia...
In Central Appalachia, support for Obama was split among "Americans," and far higher than everywhere else except the Northeast. The map shows the counties that are described well by this regression stretching from Pennsylvania down into Southern Ohio and Northern Kentucky, and some in Tennessee as well. There's a good deal of yellow in the region where two regressions meet, and a county could be predicted by either the Central Appalachia line or the Real American line. What is unique about Central Appalachia is the central region (in Central Appalachia) of high percent "American" and about 50% support for Obama.
Finally we come to the Confederacy and Old West. Looking at the maps, there's no clear demarcation, and not much difference, just a fade from one to another around the Southern Plains. There's actually a little lower support among "Americans" in the Old West. We can gain a few more insights from the Old West map, however. First, there appears to be another region - the Pacific Coast, essentially the Northern California coast and Oregon and Washington west of the Cascades - where support among "Americans" is about 10 percentage points higher than the rest of the Old West. Then we have the Mormon Corridor, where support is lower. And finally, the New West - blue spots here and there, including many cities.
All these maps put together - and they more or less look like this. (For more maps of cultural regions, go here.) In other words, politics among "Americans" more or less follows previously described cultural boundaries in the United States.
This diary is the twelfth in a series taking a close look at the 2008 electorate and exploring three themes: diversity within demographics, progressive feedback loops, and demographic change.
Tomorrow: The Appalachian Electorate: Surprisingly Democratic
Cross posted at Open Left.
Diaries in this series (updated list):
Why Republicans Should Be Really Scared
African-Americans – We Are Not All of Us Alike
East and South Asian Americans – Diverse and Growing
West Asian Americans – Rapid Change
Native Americans – Increasing Participation
Islander Americans – In Need of More Representation
Native Alaskans – An Economic Factor?
Latino Electorate – Increasing Influence
European-Americans – Tribal Politics Persist
“Americans” – You Might Be Surprised
Appalachia – Surprisingly Democratic
Why Republicans Should Be Really, Really Scared
Why Republicans Should Be Really, Really, Really Scared
A Few More Tidbits