In November, Donald Trump could become the first Republican presidential nominee to lose Orange County, California, since 1936. He could also be the first to lose Virginia Beach, Virginia, since 1964. But he could simultaneously become the first Republican to win an electoral vote from Maine since 1988 and only the second Republican to carry Iowa since 1984.
Hillary Clinton is favored to win the presidency, perhaps by a lot. Republicans are still favored to hold the House. In other words, after all the madness, the balance of power in Washington post-2016 could look surprisingly similar to that after 2012. Yet beneath the surface, the tectonic plates of the American electorate are shifting.
By now, it's clear where the fault lines lie: The 2016 election is poised to be among the most polarized elections ever, not only along gender and generational lines, but especially along lines of race and educational attainment.
In August, Nate Cohn of The New York Times put it well when he wrote: “The simple way to think about Mr. Trump’s strength is in terms of education among white voters. He hopes to do much better than Mitt Romney did in 2012 among white voters without a degree so that he can make up the margin of Mr. Romney’s four-point defeat and overcome the additional losses he’s likely to absorb among well-educated voters and Hispanic voters.”
There’s evidence that Trump is underperforming Romney among Asians and African-Americans, not just Latinos and college-educated whites. Clinton, on the other hand, has been underperforming President Obama among non-college-educated whites.
To get a handle on how these shifts could affect the electoral landscape, we modeled how many of Romney’s votes came from college-educated whites and minorities and how many of Obama’s votes came from non-college-educated whites in each state, county and congressional district. The difference between these two vote totals, shown in the map above, can tell us where Clinton and Trump have the most potential to build on 2012.
Then we went a step further: How would the 2016 map look if one out of every five whites without a college degree who voted for Obama in 2012 defected to Trump and if one out of every five non-whites and college-educated whites who voted for Romney in 2012 switched to Clinton? (Why one out of five? It’s a somewhat arbitrary number but represents a realistic shift of these groups, according to polls released over the past few months.)
Let's call this scenario the “2016 Vote Swap.” In it, Clinton would win the election, and her share of the two-party vote would be 52.7 percent — 0.7 percentage points higher than Obama’s 2012 showing. However, we also estimate she would win 10 fewer electoral votes than Obama did in the Electoral College.
The shift for Clinton among college-educated whites and non-whites would allow her to pick up North Carolina (15 electoral votes). But the shift among non-college-educated whites would cost her Ohio (18), Iowa (6) and Maine’s 2nd District (1). That's not far off from what polls and FiveThirtyEight’s forecast models show.
This model — described in detail below — isn't perfect. The available polling tells us that, for example, whites with a college degree who voted for Romney in heavily Mormon Utah are much likelier than their peers in Texas or Alabama to defect to Clinton (or a third-party candidate). Conversely, non-college-educated whites who voted for Obama in eastern Ohio are probably much likelier than their peers in Washington state or Vermont to defect to Trump.
But it's a decent starting point for understanding how the electoral map is changing and explains why Trump could outperform Romney in states like Rhode Island and Oregon while Clinton could outperform Obama in states like Arizona, California and Georgia.
Do you live in a Clinton or Obama “surge” county?
Just as fascinating as the Electoral College implications of these potential vote shifts is how they could alter the political maps within each state. If Clinton wins greater shares of college-educated whites and non-whites than Obama did but sheds some non-college-educated whites, the Democrats' coalition would continue to grow more urban and suburban, while the Republicans' coalition would continue to get more rural.
In 2012, Obama won 693 of America's 3,100-plus counties (22 percent) on his way to winning 62 percent of all Electoral College votes. Under the “Vote Swap" scenario described above, our model projects that Clinton would win the election with even fewer counties: just 631, or 20 percent, the lowest share for a presidential winner in modern history:
The model suggests that several traditionally Republican suburban locales with diversifying and highly educated electorates could be poised to flip and support the Democratic presidential candidate: Orange County, California; Gwinnett County, Georgia; Chester County, Pennsylvania; Fort Bend County, Texas; and Virginia Beach. The model also suggests that Clinton could make major gains — while still falling short — in Douglas County outside of Denver; Hamilton County outside of Indianapolis; and Delaware County outside Columbus, Ohio.
But Trump could be poised to flip the script in plenty of counties with large populations of non-college-educated white voters and Democratic heritages. These include Macomb County, Michigan; Stark County, Ohio; Aroostook County, Maine; Niagara County, New York; Luzerne County, Pennsylvania; Clark County, Washington; and Racine County, Wisconsin. Trump even has a decent chance to win tiny Elliott County in Kentucky — which hasn't voted Republican since its founding in 1869, according to Sabato's Crystal Ball contributor Robert Wheel.
For Democrats, the two most strategically crucial pro-Clinton shifts could come in Miami-Dade County, Florida, and Wake County, North Carolina. Both could play big roles in offsetting losses elsewhere in those states and effectively put Trump away.
In Miami-Dade alone, Clinton could pick up enough Cuban and other Latino defectors from the GOP to double Obama’s 74,346-vote statewide margin from 2012. And, Clinton could pick up enough Romney-supporting college-educated whites and Latinos in Wake County — the heart of North Carolina's Research Triangle — to cut Romney’s statewide margin of 92,000 votes in half.
Under the “Vote Swap” scenario, Clinton would carry 211 of 435 congressional districts, only two more than Obama carried — underscoring how short her coattails could be in House races. Traditionally Democratic seats in Minnesota's Iron Range and southern Illinois could fall to Trump. Conversely, three districts in California, two in northern New Jersey, one in Northern Virginia and one in southwestern Texas could flip to Clinton.
One casualty of 2016 vote shifts could be the bellwether status of Indiana’s Vigo County, the only county in America that has voted with the winner of every presidential election since 1956. Its knack for picking winners has attracted journalists and even a documentary filmmaker to Terre Haute. But in our “Vote Swap” scenario, Trump would be the favorite to carry Vigo, even as Clinton prevails nationally.
If the streak ends, our sincere condolences to Terre Haute.
How we built our model
To gauge Clinton’s and Trump’s upside potential, we began by reverse-engineering the 2012 electorate in each state, county and congressional district. To do this, we used population data from the Census Bureau’s 2012 American Community Survey and voter turnout data from its 2012 voting and registration report to estimate the demographic breakdown of the electorate within each geographic area by five groups: college-educated whites, non-college-educated whites, African-Americans, Latinos, and Asians/others.
Then, using using data from 2012 and 2008 exit polls, we estimated Obama’s and Romney’s levels of support from each group within each state. We applied those support levels within each geographic area and adjusted each subgroup’s vote totals proportionally to fit the actual reported votes for each state, county and district. For example, our best estimate is that 4,825 of the 47,416 major-party votes cast in Allen County, Ohio, came from African-Americans and that Obama carried them 4,371 to 454.
Finally, we used these estimates to calculate, in each state, county and district, the share of college-educated whites and minorities who voted for Romney and the share of non-college-educated whites who voted for Obama. Our “Vote Swap” scenario depicts what would happen if 20 percent of each of those groups switched parties in 2016.
