The other trend, of course, is that the density of America's non-Hispanic white population is decreasing. Nearly half of the children born in the United States since 2000 — 49 percent — are Hispanic or nonwhite. If I asked you which state had the most heavily non-Hispanic white population, you might need a few more seconds, but might end up with the correct response: Vermont.
While there's only one majority-senior county (Sumter), there are an increasing number of majority-minority counties in the United States. They're geographically clustered, with a number in the Southwest (thanks to large Hispanic populations) and in the South (thanks to large black populations). Age is much more homogeneously distributed throughout the country; with the exception of Sumter County, home to a huge retirement complex called the Villages, there are a lot fewer places with high densities of older people than there are places with high densities of white people.
Which, at last, brings us to politics. You might guess that Sumter County voted for Donald Trump in the primary, because Trump won every county in Florida, save Marco Rubio's home county of Miami-Dade. You might also guess that Hillary Clinton won Sumter, since she did better with older voters. But if I asked you how Trump or Clinton did in those majority-minority counties, it might be harder to predict. (Save the majority-minority county of Miami-Dade, which I already told you.)
We took the Census Bureau's new 2015 population estimates and compared them to county results from the primaries to see how the presidential candidates fared in places that had a larger or smaller density of nonwhite residents. (We included Bernie Sanders in this analysis.)
We'll reiterate first that most counties are majority-majority — they are mostly white and mostly non-Hispanic. In fact, most American counties are very white. But many of the more diverse counties are also more populous. A county isn't defined by population size, so there are a lot of big, fairly empty, very white places in the United States.
Here's how the candidates did in those counties:
Donald Trump did a bit better in many of the more diverse counties, for reasons we'll explain below. But there was a much more distinct trend on the Democratic side: Clinton got more than 50 percent of the vote on average in counties that had nonwhite populations (here meaning: the part of the population that isn't non-Hispanic white) that made up at least a fifth of the population.
As the diversity of counties increased, Sanders did worse. In counties that were 20 percent nonwhite or less, Sanders averaged 50 percent of the vote to Clinton's 46.6 percent. In counties with populations that are 20 to 50 percent nonwhite, Clinton got 59.6 percent of the vote to Sanders's 36.8. Sanders's best state was his home state — the heavily white state of Vermont.
Since there are so few counties at the right end of that spectrum, we can look at it in another way, too. Plotting the density of the nonwhite population versus the vote share, the trends for Clinton and Sanders are more obvious.
Clinton's results curve up as the minority population grows. Sanders's curve down.
Trump's don't really change.
That's because Trump's results were linked not to the population but to the point in the primary campaign that contests were held. The majority-minority counties where he did best were in the Northeast and in California — because he was steamrolling the competition by then, if any existed. He did worst in Texas, since Texas Sen. Ted Cruz won the state.
Clinton's strongest majority-minority counties were in the South — a reflection of her strength with African American voters. (She also led big with Hispanic voters, but not quite as dramatically.)
Sanders's best places were Washington and Hawaii, where he was helped in large part by the contests being caucuses, where he consistently did better.
This tells us a great deal about the nature of the primary contests, and, therefore, the general. There were clear preferences among Democratic voters based on racial identity; the presumptive Republican nominee did about as well regardless of the composition of the population, in part because most of the nonwhite voters in those counties were voting in the other party's primary.
It's the green chart above that tells us the most about the general election: lots of places with increasingly dense and growing nonwhite populations. Trump may well win Sumter County in the general. But there are far more majority-minority counties than majority-senior ones, and in those we can anticipate the Democrats to build a big lead.