Democrats are from cities, Republicans are from exurbs

Net vote change since '88 election

Net vote change since '88 election

The first reaction many people may have when looking at a standard red/blue map of county-level election results is often "Holy crap, look at all that red!" While that huge red bulls-eye in the nation's middle is easily explained away—trees and tumbleweeds don't vote; only people do—when it's presented without disclaimers, it can help reinforce the perception that this is a "center-right nation." One of the best ways around that is with a cartogram (a map that distorts boundaries to account for another variable, like population) ... but how do we express that in a way that still looks like the United States, and not a piece of purple roadkill?

Princeton's Robert Vanderbei put together a fantastic 3-d map in 2012, which I'd recommend that you take some time to explore, that visualizes the country with blue skyscrapers towering over pink plains, reflecting the heavy concentration of votes in urban areas. However, I wanted to try going in a different direction, with a flat map, but one where the color varies by intensity according to the change in the number of votes over the years.

Counties where there has been a lot of growth in the last few decades show up as either bright red or bright blue, depending on whether those additional votes tended to be Democratic or Republican. Counties that have a small population, and where there has been little or no growth, stay grey. (In a very few cases, counties where there has been a lot of growth, but where the gains in Democratic and Republican votes balance each other out, also stay grey.) This way, that terrifying red swath of counties across the nation's middle where few people actually live fades from view, while you get a clearer sense of where the Democrats and Republicans are actually running up the score: along the coasts and in the Midwest's cities for the Dems; in Texas, the Mountain West, and all along the spine of the Appalachians and Ozarks for the GOP.

In a way, this map is a sequel to the maps from the piece I wrote last week on when the nation's counties hit their population peak. There were some striking similarities between the map showing when each county peaked, and the map showing the trend in county-level presidential results between 1988 and 2012. The counties where the population peaked long ago, and where the greatest movement toward the Republican Party, were both located all across the Great Plains and Appalachia.

As I pointed out last week, though, even though that describes a majority of the nation's counties, those counties only contain a small minority of the nation's population. The counties where the most of the people are, and where most of the nation's growth occurs, are where we really need to focus ... and they, too, are moving in different directions. More and more Democratic votes are showing up in the cities, especially on the coasts, while more and more Republican votes are showing up in exurban counties, especially in the interior south, and that's what this map tries to show.

An interactive version of the map, which you can zoom in on and pan around, is over the fold; so too is more explanation of how it works and what it means.

As you mouse around the interactive map, you'll see a mysterious-looking number attached to each county, positive for blue counties and negative for red counties. That's the net change in the number of Democratic and Republican votes over the period from 1988 to 2012. In other words, it finds the disparity between the number of Democratic votes in each county in 1988 and 2012, the disparity between the number of Republican votes in each county in those same years ... and then finds the difference between those two figures. If the increase in Democratic votes outstripped the increase in GOP votes, you have a positive number.

Here's an illustration. Consider King County, Texas, which you'll find in the parched lands halfway between Lubbock and Wichita Falls. If you wanted to advance a "Dems in disarray" narrative, here's where you would start. On a map of 2012 presidential results, this would be one of the most blazing-red counties anywhere: Barack Obama got a whopping 3 percent of the vote here, while Mitt Romney got 96 percent. Even on a map of electoral trends, this would still be one of the reddest counties: Michael Dukakis still got trounced here, but got a respectable 37 percent of the vote. That's a Partisan Voting Index shift from R+10 to R+48 in the space of 24 years!

At the end of the day, though ... so what? If you crack open the underlying data, Obama got 5 votes here, while Dukakis got 64, for a difference of -59. Meanwhile, Romney got 139 votes here, while Bush the Elder got 111, for a gain of 28. That's a net Republican gain of 87 votes (even though there were 30 fewer votes cast in 2012 than in 1988). King County occupies more than twice the square mileage of New York City, so, to the eye, it's a perceptible pixel on the nation's map ... even though it contains slightly more than one one-millionth of the nation's votes.

Contrast that with a different King County, the one in Washington. Obama got 668,004 votes here in 2012, while Dukakis got 349,663: a gain of 318,341. Romney, on the other hand, got 275,700 votes here, while Bush got 290,574; despite there being over 300k more votes here in 2012, Romney managed to get fewer votes than Bush did, 14,784 fewer to be exact! The net change is 333,215, enough to almost single-handedly move Washington—which used to be a swing state in the 70s and 80s—into the realm of safely blue states, despite (as you'll notice in the map above) Republican gains in much of the rest of the state.

Bearing in mind how that works, here are the top 25 counties for the largest Democratic net gain. You'll notice that, for the most part, this list mostly overlaps with the largest counties by population, period; Maricopa Co., Arizona (Phoenix), Harris Co., Texas (Houston), and California's Inland Empire (San Bernardino and Riverside Cos.) are the only multi-million-person counties that don't make the list. While they all had net Democratic gains, it wasn't as lopsided; for instance, Maricopa Co. was the single largest gainer in total votes (at 699,441, even more than Los Angeles Co.), but 371,366 of those were Democratic and 307,548 were Republican, for a net of "only" 63,788. (If you're wondering why the Dem and GOP votes don't add up precisely to the number of total votes, that's because it also incorporates 3rd-party and write-in votes.)

County	Descrip.	Total change	Dem change	GOP change	Net change
Los Angeles, CA	Los Angeles	+ 536,396	+ 844,551	- 354,383	+ 1,198,934
Cook, IL	Chicago	- 14,546	+ 358,564	- 383,040	+ 741,604
King, WA	Seattle	+ 323,064	+ 318,341	- 14,874	+ 333,215
Kings, NY	Brooklyn	+ 181,515	+ 235,018	- 55,272	+ 290,290
San Diego, CA	San Diego	+ 323,087	+ 293,693	+ 13,583	+ 280,110
Broward, FL	Ft. Lauderdale	+ 316,749	+ 290,038	+ 23,785	+ 266,553
Prince George's, MD	DC area	+ 164,863	+ 214,122	- 50,811	+ 264,933
Miami-Dade, FL	Miami	+ 388,914	+ 324,470	+ 62,044	+ 262,426
Philadelphia, PA	Philadelphia	+ 15,733	+ 139,240	- 122,586	+ 261,826
Santa Clara, CA	San Jose	+ 101,563	+ 173,008	- 79,599	+ 252,607
Orange, CA	LA area	+ 257,357	+ 243,427	- 3,898	+ 247,325
Queens, NY	Queens	+ 48,344	+ 145,341	- 98,478	+ 243,909
Wayne, MI	Detroit	+ 69,980	+ 145,624	- 78,182	+ 223,806
Alameda, CA	Oakland	+ 116,645	+ 159,401	- 54,633	+ 214,034
Dallas, TX	Dallas	+ 115,579	+ 159,401	- 54,633	+ 213,654
Franklin, OH	Columbus	+ 193,626	+ 198,751	- 10,728	+ 209,029
Montgomery, MD	DC area	+ 135,096	+ 158,213	- 30,838	+ 189,051
Fairfax, VA	DC area	+ 200,922	+ 189,562	+ 6,132	+ 183,430
De Kalb, GA	Atlanta area	+ 122,015	+ 145,703	- 25,787	+ 171,490
Bronx, NY	Bronx	+ 72,813	+ 120,924	- 46,078	+ 167,002
Oakland, MI	Detroit area	+ 191,083	+ 174,257	+ 13,155	+ 161,102
Orange, FL	Orlando	+ 294,533	+ 219,642	+ 71,352	+ 148,290
Cuyahoga, OH	Cleveland	+ 43,164	+ 93,831	- 51,788	+ 145,619
Contra Costa, CA	SF area	+ 106,715	+ 121,413	- 22,135	+ 143,548
New York, NY	Manhattan	+ 90,676	+ 114,484	- 26,808	+ 141,292

One thing you'll notice as you look over the list is that these are almost entirely in blue states (with the exception of Dallas Co., TX, and DeKalb Co., GA, though those are the two big states that observers expect to move into swing state territory in, say, the 2020s, thanks to demographic change). But that wasn't always the case: Illinois, California, even Maryland used to be swing states in the 70s and 80s, and Dukakis managed to narrowly lose all three of them. The huge Democratic gains in the counties above is exactly what pushed those states into safely blue territory. And Florida may be an even more extreme case. While it's a swing state now, it was one of the Duke's worst states in 1988; he fared worse there than in Alabama or Nebraska! Gradual gains in the Miami and Orlando metro areas are what put the Sunshine State into play.

Most of these counties have grown significantly over the decades; however, a few have stayed almost level in terms of the number of votes and still changed dramatically. Cook Co. (Chicago) and Philadelphia Co. may be the most dramatic examples, where there were as many or more lost Republican votes as gained Democratic votes. It's impossible to parse out from this data how much of that is actual changed minds, and how much is the replacement of Republicans who died or white-flighted to somewhere else, with Democrats who moved from elsewhere or aged into the electorate ... but in some cases, the change in composition is just as much from Republican subtraction as it is from Democratic addition.

Now let's look at the counties with the largest net gains by Republicans. The first thing you'll notice is that while the Democratic net gains mostly came in the counties that contain the nation's largest cities, the Republican net gains come, for the most part, from peripheral counties, growing areas that are one (or more) county lines away from the center of a metropolitan area.

County	Descrip.	Total change	Dem change	GOP change	Net change
Utah, UT	Provo	+ 89,486	- 1,252	+ 88,816	- 90,068
Montgomery, TX	Houston area	+ 113,967	+ 14,526	+ 97,609	- 83,083
Westmoreland, PA	Pittsburgh area	+ 30,282	- 12,988	+ 42,460	- 55,448
Waukesha, WI	Milwaukee area	+ 94,963	+ 21,181	+ 72,331	- 51,150
Collin, TX	Dallas area	+ 211,591	+ 78,481	+ 129,112	- 50,631
Cherokee, GA	Atlanta area	+ 79,132	+ 15,463	+ 61,921	- 46,458
Forsyth, GA	Atlanta area	+ 71,387	+ 12,224	+ 57,961	- 45,737
Denton, TX	Dallas area	+ 158,571	+ 54,774	+ 100,135	- 45,361
Davis, UT	SLC area	+ 52,685	+ 5,021	+ 46,392	- 41,371
St. Tammany, LA	NoLa area	+ 58,079	+ 10,090	+ 46,389	- 36,299
Galveston, TX	Houston area	+ 36,036	+ 878	+ 34,146	- 33,268
Douglas, CO	Denver area	+ 143,431	+ 54,060	+ 87,169	- 33,109
St. Johns, FL	St. Augustine	+ 87,620	+ 27,161	+ 59,285	- 32,124
Livingston, LA	Baton Rouge area	+ 28,319	- 2,208	+ 29,734	- 31,942
Washington, PA	Pittsburgh area	+ 18,416	- 7,182	+ 24,579	- 31,761
Williamson, TN	Nashville area	+ 67,402	+ 17,278	+ 49,003	- 31,725
Shelby, AL	Birmingham area	+ 58,428	+ 12,913	+ 44,384	- 31,471
Baldwin, AL	Mobile area	+ 49,893	+ 9,153	+ 40,083	- 30,930
Beaver, PA	Pittsburgh area	+ 4,324	- 13,272	+ 16,580	- 29,852
Calcasieu, LA	Lake Charles	+ 17,626	- 5,573	+ 22,201	- 27,774
Williamson, TX	Austin area	+ 116,041	+ 42, 286	+ 69,684	- 27,398
Allegheny, PA	Pittsburgh	+ 37,676	+ 3,873	+ 30,902	- 27,029
Santa Rosa, FL	Pensacola area	+ 52,690	+ 12,514	+ 39,213	- 26,699
Washington, UT	St. George	+ 37,447	+ 5,283	+ 31,392	- 26,109
Clay, FL	Jacksonville area	+ 62,964	+ 17,986	+ 44,080	- 26,094

You can call these areas the "exurbs," though there really isn't a hard-and-fast line between the exurbs and the suburbs; the exurbs look pretty much the same as the suburbs, just further out with bigger lots, though they might contain some older, once-standalone towns that got engulfed by suburban growth. Nobody has really adequately defined the difference, not even David Brooks, erstwhile chronicler of the virtues of the exurbs, though he claims they "have broken free of the gravitational pull of the cities and now exist in their own world far beyond." That's not really true, though; would all those houses in, say, Forsyth Co., Georgia, be there if they weren't within driving distance of Atlanta? Even if its residents aren't all commuting all the way in to Atlanta, they're still commuting to suburban office parks that are part of the same economic ecosystem.

There are a few exceptions to the largely exurban character of these counties, though. Some are big towns/small cities that stand on their own, some of which are tourist or retirement destinations that seem to particularly attract Republicans (St. George, in southern Utah, or Santa Rosa Co., adjacent to Pensacola in Florida's Panhandle). Lake Charles, Louisiana, hasn't seen much growth, but, like most of the rest of the Cajun-flavored parts of that state, has simply swung hard from the Democrats to the Republicans, over culture-war and energy-sector issues. And Provo, Utah, sort of falls in that standalone category, too; while Utah Co. is next door to Salt Lake Co. and has its fair share of commuters, it has its own (fervently conservative) identity, thanks to Brigham Young University.

The other exception is Allegheny County, the location of Pittsburgh, and the only county with more than a million residents that finds its way onto the pro-Republican list. It's initially hard to imagine, since Pittsburgh itself is a city with a strong labor tradition and remains pretty blue. However, Pittsburgh makes up only about a quarter of the county's population; there's also a lot of suburbs, and also once-vibrant mill towns along the rivers that have largely emptied out.

In addition, it's a much whiter and older population than most other urban counties; with organized labor playing less and less of a role here, and with elderly ex-unionists—the same people who made this one of the bluest areas anywhere in the country in the 80s—either dying or retiring elsewhere, it's only natural that it would move toward swingier territory. That's even more noticeable in the counties surrounding Pittsburgh (like Westmoreland and Washington), which used to be dotted with coal mines and steel mills but now essentially function as exurbs. For instance, Beaver Co., west of Pittsburgh, gave 46 percent to Obama (R+5) but 66 pecent to Dukakis (D+20)! If western Pennsylvania were its own state, it'd be on much the same trajectory as once-blue West Virginia ... but instead, Pennsylvania's presidential vote has stayed pretty stable over the decades, thanks to an equally-rapid blue trend in Philadelphia and its suburbs.

One other thing you've probably noticed, looking at the two charts, is that the net gains in the Democratic-trending counties are much, much larger than the ones in the GOP-trending counties. In fact, you have to sort through 61 Democratic counties before you find a net gain that's smaller than Utah County's 90k pro-GOP gain. So, why then, you might be asking, haven't the Democrats completely wiped out the GOP at a national level? Well, that's because those gains are so heavily concentrated in so few counties. There are far, far fewer counties that have a net gain for the Democrats (under 1,000) than ones with a net gain for the Republicans (more than 2,000 of them). It's the same tendency I discussed last week: almost all of those empty counties in the middle of the country have trended against the Democrats.

But, as I said earlier ... so what? Those thousands of red-trending counties don't have as many people in them. Among the 900-odd counties with a net trend for the Dems, the average net gain was over 20,000 votes. Among those more than 2,000 counties that went the other way, the average net gain for the GOP was only a little more than 3,000 votes. You can see that difference when you perform the same analysis from above on the nation as a whole. Obama got almost 66 million votes in 2012, a gain of around 24 million votes over Dukakis's 42 million. Meanwhile, Romney got about 61 million votes, an increase of only 12 million over G.H.W. Bush's 49 million: it's a Democratic net gain of 12 million, if you add it all up.

Unfortunately, though, as much as this huge growth in Democratic fortunes in the nation's urban areas makes it easier to win the presidency, the corresponding decline in the rural areas makes it harder to win the House (without which, of course, you don't accomplish much). Democratic votes were distributed more efficiently throughout the country before the dynamics of the Big Sort really accelerated during the 1990s; Dukakis was competitive in places as far-flung as Missouri, Louisiana, the Dakotas, even Oklahoma, and those coattails extended downballot. Even without considering the pernicious effects of gerrymandering, the clustering of more and more Democrats in fewer and fewer places creates a lopsided map of fewer dark-blue districts and more light-red districts. It's a classic case of two steps forward, one step back.

If you want to see the entire dataset, it's available through Google Drive.