Demographics: William Frey of the Brookings Institution has an interesting post on geographic clustering, and it doesn’t go the way you might think it would go. It turns out that if you use counties as your level of analysis, then it may be the Republicans who are the ones who are clustered, in the nation’s rural counties. There are 177 million people living in the counties that went for Hillary Clinton in the 2016 election, and only 146 in counties that went for Donald Trump. That’s a very anomalous result, one that didn’t even happen the last time we had an Electoral College mismatch: in 2000, 149 million lived in George Bush counties and 133 million lived in Al Gore counties.
(If you're familiar with the recent work of Daily Kos Elections alumnus Xenocrypt, you actually wouldn’t be that surprised, though. He made a similar argument back in January, pointing out that a disproportionately large number of people lived in Clinton counties subsumed within Trump states.)
What’s helpful in Frey’s post is when he starts breaking the Clinton-county and Trump-county populations down along various other demographic lines. You can see how Clinton counties lean more toward the higher-income, the never-married, the foreign-born, the non-white, and the college-educated. In particular, what’s cool (and what should be terrifying to Republicans regarding future elections)are the graphs at the end of the article that break the populations down by both age bracket and race. Trump counties are disproportionately not just white but elderly, and the single-largest bucket by far is people who are under age 15, who are non-white, and who live in current Clinton counties. (In other words, that’s the people who’ll be entering and reshaping the electorate in the next decade.)