2012 White Obama Lovers Map by County Project: COMPLETE!

[Adam Griffin]

Very slowly but surely, I am working on building out a map that will display the approximate share of the white vote President Obama received in 2012 in each county. Inevitably and once a PNG gets too unruly in size, it'll be converted into a map layer for online use, resembling this very successful project. You've seen several of these individual state maps posted on various parts of the forum before as I have completed them.

If anyone is interested in helping out with this project, then please let me know. I can help explain how to approximate the white share of the vote in each county (if you are not familiar with how to do this). You'll need at least one of the following (in order from most ideal to least):

• 2012 voter turnout by county, by race
• 2012 voter turnout by race & 2012 voter registration stats by county
• 2012 Census estimates by county

Interactive Google Map (Click Here or on Image Below)

Winner by County

COMPLETED STATES - OBAMA'S SHARE OF THE WHITE VOTE:

DC: 87.3%
Vermont: 66.3%
Hawaii: 65.3%
Rhode Island: 60.2%
Massachusetts: 57.8%
Maine: 55.5%
Connecticut: 52.6%
Oregon: 51.3%
New York: 51.2%
Washington: 51.1%
New Hampshire: 50.3%
Iowa: 50.1%
California: 49.0%
Delaware: 48.4%
Minnesota: 48.4%
Wisconsin: 48.2%
Colorado: 46.7%
Maryland: 45.7%
Illinois: 45.4%
New Jersey: 44.7%
Michigan: 44.3%
Pennsylvania: 44.0%
Ohio: 42.0%
Nevada: 40.7%
New Mexico: 39.1%
Florida: 38.3%
Indiana: 38.0%
Montana: 37.8%
Virginia: 37.3%
Arizona: 36.6%
Missouri: 35.8%
North Dakota: 35.4%
South Dakota: 35.3%
Nebraska: 33.4%
W. Virginia: 32.8%
Alaska: 32.7%
Kansas: 32.5%
Kentucky: 31.5%
Tennessee: 30.8%
North Carolina: 30.6%
Idaho: 28.1%
Arkansas: 27.1%
Oklahoma: 27.1%
Wyoming: 25.4%
South Carolina: 23.4%
Georgia: 19.6%
Texas: 19.6%
Utah: 18.3%
Alabama: 15.8%
Louisiana: 14.4%
Mississippi: 11.4%

[Adam Griffin]

Why are DeKalb County whites so liberal?  Or maybe the methodology used to estimate the white Obama % gets a little wonky in counties that are demographic outliers?  

I am confident in the overall projection for Dekalb County.  Roughly 38% of 2012 voters in the county were white (according to a combination of SoS turnout by race stats and a projected breakdown of "Unknown" and "Other" voters). 55% were black. If we assume 60% of "Other" voters (7% of turnout) voted Obama and 95% of blacks voted Obama, then that suggests ~55% of whites voted for Obama.

I am, however, currently having a running debate over a potential phenomenon that may exist in rural counties that are super-majority black. I came across this in Georgia, as well as the other states (Georgia was the first state I did; keep in mind that Georgia publishes voter turnout in elections by race, unlike the other states here). You'll see several counties that are rural, heavily black and that have high white Obama support. In almost all of these counties, it is virtually black and white population only. As such, there is no possibility that blacks are turning out in higher numbers (because SoS provided this turnout data), nor that blacks are less Democratic (that would make the whites even more Democratic). I considered all of this because I thought my formula had to be wrong for some of these counties.

My hypothesis for these counties is that when whites live in a super-majority black county, their tendency to kneejerk against blacks by voting Republican dissipates. In many Georgia counties where you have 55/45 or 50/50 black splits, whites are 10% Obama or less. However, when you get to around 60-65% black counties, there is a clear rebound in the share of the white vote Obama received. This is just a crude drawing so don't take it as 100% reflective of anything, but this is kinda what I'm seeing:

For Florida, why is Miami-Dade at only 34%, while Osceola is at 57%?  Does it have something to do with how you count Hispanics?

I think that for a number of reasons, Miami-Dade will be less Democratic in terms of white Obama support than Osceola (and Broward, for that matter). I do think my formula is incorrect here. I adjusted Latino support based on FL 2012 exit polling, but it was applied statewide. This means that in many of the northern counties, Latino support might be lower (non-Cubans; not as much impact considering lower Latino populations) in my formula than it is here (and higher in the South, particularly Miami-Dade; Cubans). This will get adjusted at some point, but figuring out which counties get what will take an exponentially larger amount of time than it did to create the overall result. In addition, the formula assumes that half of all registered Latinos turned out in 2012 - this, too, will vary depending on the part of the state and whether we're dealing with Cubans or not.  

[Adam Griffin]

Why are DeKalb County whites so liberal?  Or maybe the methodology used to estimate the white Obama % gets a little wonky in counties that are demographic outliers?  

It's a downtown with a substantial non-Southern population.  That's not particularly surprising to me.   It would be interesting to see if Obama won white voters in Orleans Parish, Mecklenburg, NC or Davidson, TN (probably the least likely).

Fulton County is the downtown, no?  I guess my question is more, why would DeKalb whites be more liberal than Fulton County whites?  

Downtown in a Southern city developed with a lot of post-1950 growth (Atlanta, Houston, Dallas, Austin, etc.) means a cluster of tall buildings within a 20 mile radius region with the density of a Northeastern inner suburb.  It's not particularly obvious where downtown ends and the built up suburbs begin.  The inner suburb to outer suburb divide is much more dramatic culturally.

Yes, a part Fulton is technically downtown, although downtown's broader limits can be considered to extend to the Dekalb County line. You have substantial populations of majority-white, wealthy and progressive communities in a triangle between Little Five Points, North Druid Hills and Decatur. In addition, Dekalb has a higher population density than Fulton (2500 versus 1900, if I recall correctly; which is understandable, given Fulton's rather odd shape). Speaking of Fulton's odd shape, I think that may be an element you're forgetting here: Atlanta proper is less than 1/2 of Fulton's population, and Fulton incorporates some very staunchly conservative, wealthy and super-majority white areas to the north and extreme southwest.

Just to illustrate: here are two contiguous areas - one in Fulton and one in Dekalb - that are roughly equal in population, roughly equal in white population and comparable in latitude (I've purposefully excluded some of the core of downtown ATL because it is both relatively small in population & has too many minorities to make the comparison make sense).

The Dekalb portion is more Democratic than the Fulton portion despite the fact that the Fulton portion has a 22% black population (Dekalb's has 14%). The Dekalb side has a larger Latino population, but turnout as a share of VAP in Georgia in urban and rural areas alike for Latinos is often around 40% or less, so you're talking about in reality 6% of the electorate being Latino versus roughly 3%).

[Adam Griffin]

SF, Portland, Seattle. There's your map. lol

Hey, I don't know if you noticed or not...but this is the Political Geography & Demographics forum. So if you can't discuss issues in an empirical (or at the very least, in a non-ignorant) fashion without being a blatant partisan hack, then kindly show yourself the door and fuck off.

[Adam Griffin]

+100. I can't be a big help here (living abroad and being almost completely Internet-based), but waitng for the result with great interest. BTW - what is so special in that Mississipi county, that gave Obama only 2% of white vote? Even by Deeep South standard that's something

I can't necessarily speak for RGriffin, but when I calculated for Mississippi, I used some constant assumptions across counties, a constant black and white turnout percentage across the state, and constant minority percentages in the breakdown (I think it was 96-4 for blacks, around 75% or so for Hispanics and Native Americans, etc.) Whatever is left would be my calculated white vote.

However, if for some reason a county had a black turnout that was a lot lower than the state average, or for some reason Romney did better with blacks (I don't know why that would be the case in Mississippi, but there had to be some county-to-county variation), it would make the calculation undercount white votes for Obama.

I did 95/70 on mine. I remember you saying you had a lot of negative Obama percentages in MS counties (I had several myself); having such a high minority level of support in the formula would definitely cause that.

[Adam Griffin]

As I map out differences between states, I take pause and may need to find subtle ways to tweak my formulas. As you can see in the map below, there are two abrupt changes in gradient with respect to which types of counties appear and don't appear depending on which side of the state line they find themselves on. In the instance of GA/FL, it actually makes sense: one state was notably contested by Obama, while the other was not. In the AL/MS example (why there are so many <10% Obama white counties in eastern MS & not in western AL), I don't see an immediate reason as to why this should be the case.

All counties in white were counties where whites gave Obama 10-19% of the vote. Red counties are counties where whites gave Obama 20% or more of the vote; blue counties were where whites gave Obama less than 10% of the vote.

 

[Adam Griffin]

SC has been added to the map.

2012 statewide share of the white vote for Obama: 23.4%

[Adam Griffin]

NC has been added.

2012 statewide share of the white vote for Obama: 30.6%

[Adam Griffin]

TN has been added.

2012 statewide share of the white vote for Obama: 30.8%

[Adam Griffin]

Wow, Obama did a lot better in Tennessee then I would have expected. I guess the only thing separating it from North Carolina in terms of competitiveness is the massive black flight from the state.

One big issue with Tennessee is that its black population remains (at least, it appears) vastly under-represented at the ballot box.

I will admit that I am the least confident in my county-by-county projections for Tennessee than for any other state, if only because I had almost nothing with which to work (I think my overall statewide white share of the vote projection is accurate, however). TN doesn't publish voter registration stats by race for each county, nor does it seem to track voter turnout by race and/or by county. I literally had only the 2012 Census numbers and the 2008 exit polling to use for this. The 2008 exit poll said that 84% of TN voters were white and 12% black; 2010 population was 76% white and 17% black. In my 2012 projections, the number worked out to 14.5% black. If the voters had matched the population along black and white lines in 2008, then TN would likely have been 53/46 or so.

[Adam Griffin]

Yeah, TN isn't the Deep South after all. Western and Middle Tennessee used to be quite Democratic until the early 2000s, since then the Democratic party has collapsed here - you saw a similar thing happening in KY. TN is really KY's twin. They are quite similiar politically, even though TN is slightly more Republican and conservative.

Don't forget that even in 2008, Tennessee had a Democratic Governor and 5/9 House seats were won by Democrats.

Adam, nice job on Tennessee, considering the constraints you mentioned. I have calculated the white vote % for Davidson County before and I got the same number as you. Based on my interactions with people it would seem like a decent majority of whites are voting Democrat here but I usually hang around places like Downtown, East Nashville, Germantown and the West End area where whites are voting for the Democrats. That drops precipitously once you get outside the 440/Briley Parkway loop.

Well, it's good to know that things look more or less correct. Someone was sort of busting my balls about Davidson County in chat last night, but it's one of the counties about which I had more confidence. I was actually shocked to see Knox County as low as it was, and in particular, Hamilton County be so low compared to the rural areas to the west of them (in relative terms; it doesn't shock me that places like Warren, Van Buren, Grundy, Marion, Franklin, etc are as Democratic as they are). Even Shelby County was a bit surprising for me.

[Adam Griffin]

I'm a little surprised that North Carolina whites didn't vote for Obama at a higher rate, especially when compared to Tennessee.

I actually think the most interesting part of this is the level of rural white Dem support still left in NC, particularly when comparing the NC mountains to areas just across the border in any direction.  However, I did assume Raleigh and Charlotte white voters were a lot more Dem than the nation, so that part is surprising.  It will be very interesting to see VA for this reason.  I wonder if Obama actually won white voters in Fairfax or Henrico?

I, too, was surprised at the amount of white D support in western NC, though I knew it would be higher than the statewide total. While not necessarily indicative of anything, I recall driving through Cherokee/Clay/Jackson/Macon Counties a couple of years ago and seeing just a smorgasboard of local Democratic yard signs (that proudly declared such, which is odd for this general area) and Obama signs.

What isn't surprising is the abrupt change in support when crossing borders - especially from GA into just about any other state. When you look at presidential results by county, you can see an almost immediate shift in gradient when you cross into NC (and TN/AL in some cases). Also, once you cross into NC from GA or TN, you can really sense a cultural shift just from driving along the highways.

Something I've considered/wondered about is how much dilution there might be with respect to inconsistencies in reporting Census race/ethnicity in the western part of the state. The capital of the Cherokee Reservation is in Cherokee (city), NC, in both Swain and Jackson Counties. Swain, Jackson and Graham Counties have 28%, 9% & 7% Native American populations, respectively. This still leaves 3-4 counties in that broader area where white Ds are strong but that do not have any substantial Nat-Am populations.

Most "Indians" in the area in reality are white these days, but I'm wondering if they're self-reporting as white (but perhaps possess more traditional Native-American voting habits)?

[Adam Griffin]

LA has been added.

2012 statewide share of the white vote for Obama: 14.4%

[Adam Griffin]

AR has been added.

2012 statewide share of the white vote for Obama: 27.1%

[Adam Griffin]

LA has been added.

2012 statewide share of the white vote for Obama: 14.4%

Orleans Parish is quite remarkable.  I thought it would be in the low 40's at best.

It looked higher than I thought it'd be, too. I'm going to go back and check a few counties from these past two states - LA had turnout by race but AR was another one that proved to be difficult.

[Adam Griffin]

Updated/refined projections for a few counties in Florida (Osceola & Dade included).

[Adam Griffin]

KY has been added.

2012 statewide share of the white vote for Obama: 31.5%

[Adam Griffin]

KY has been added.

2012 statewide share of the white vote for Obama: 31.5%

I'm not sure of some of these numbers. I'm sure Campbell County was over 35%, since Obama did a bit better than that overall, and the county is very, very heavily white. (Most nonwhite people who I knew growing up lived in other counties.)

Well, unfortunately I can't see the chart, but I suspect that the white vote for Obama in Campbell county wasn't much greater than that--Obama only got 37% of the vote there and the county is about 3% minority.

I have it at 35%. Is the file too big for you to see?

Also, as I was making Kentucky, I could hear in the back of my head the future chants of Bandit claiming that I didn't have the whites in his state showing up as Obama enough, that the Tea Party is dead, that Obama actually won a majority of whites...

[Adam Griffin]

Virginia next?

Likely, but I'm really dreading those cities.

[Adam Griffin]

Try this.

[Adam Griffin]

VA has been added.

2012 statewide share of the white vote for Obama: 37.3%

[Adam Griffin]

Virginia is fascinating, but I have to ask, how does Roanoke County only have 7% of Obama white vote? Its 87% white and 37% Obama.  

Also, Franklin County is 89% white and gave Obama 34% in 2012, and Bedford county is 92% white and gave Obama 27%.  These two are showing as 9% and 7% on the map.  Further to the east, Appomattox is only 75% white and gave Obama only 30%, so the 13% figure there seems to check out.  Maybe a "1" got cut off in a couple of places?

I ran into issues with VA after I had started - the Census Bureau weirdly spitted out data that was completely transposed and had all sorts of incorrect figures for counties. With that being said, I did start over and managed to acquire the accurate data (or so I thought). I noticed in my spreadsheet that Richmond City and Richmond County were transposed in comparison to the order the Obama percentages had them, so I switched them to correct the problem (or so I thought). Bedford County was actually the one that made me notice the error in my first version; it'd be odd for me to screw it up again with the new data. I'll check this out later. I did think that some areas of VA looked rather odd, but there seemed to be multi-county correlations in almost all of the cases and I don't recall an instance where I had the city as substantially more conservative than the county with the same namesake.

Keep in mind that for states where there is no turnout by race, I have to acquire the Census data, adjust the counties with a formula based on the discrepancy between that and turnout, and then separately paste in the Obama percentages (sometimes, county likes with "Mc" and ones like this where there are "city/county" variations can cause transposition errors but I am nearly 100% sure I fixed all of those for VA). From there, I build another formula to plug and play in the white share of the vote, so there are some possibilities for less than small variance (though in almost all counties by default, everything usually looks accurate on the first go; I do manually adjust some counties that have large prison populations, huge/established Latino populations, etc).

[Adam Griffin]

You should send this map to the Huffington Post. It could make the main page...

We'll see! I think I'll have to make a better rendition of the map before I do that. The problem with this map is that it's kind of ugly close up, and I'm making it by hand. I do plan to do a Google Fusion Tables map (like the 2008 presidential precinct project one) when I'm finished.

[Adam Griffin]

WV has been added.

2012 statewide share of the white vote for Obama: 32.8%


I'll answer questions some of y'all asked in just a bit.

[Adam Griffin]

Excellent! After Oklahoma and (sigh) Texas with it's myriad of counties will be added - all South will be done))) And in some aspects this is the most interesting part of the project...

Personally, I don't consider TX nor OK to be part of the South. I almost don't even care to do the rest of the nation; as you said, the South really is the most interesting element. Speaking of areas that aren't the South...


DC & MD have been added.

DC 2012 statewide share of the white vote for Obama: 87.3%
MD 2012 statewide share of the white vote for Obama: 45.7%