That wasn't the full extent of Obama's coalition, certainly, but it was the collection that was central to stringing together enough votes to overcome the electorate that had picked Bush over John Kerry in 2004 and fought to a draw between Bush and Al Gore four years earlier. In 2004, the Bush campaign made a strategic decision to improve its performance over 2000 by turning out its base of conservatives: Religious fundamentalists and the white voters who had been trending more strongly toward the GOP even as they made up a smaller percentage of the electorate.
On Tuesday night, the voters who elected Donald Trump as the 45th president of the United States didn't really make up a new coalition of support that overcame what Obama had built. Obama's coalition was built to overcome a Republican-leaning white plurality. Trump's voting base was that plurality, reinvigorated.
In polling leading up to Election Day, Trump almost never topped the 47 percent mark that he ended up with in the final national tally. (This is subject to change, and will probably trend down as votes are counted.)
This suggests a few possibilities. The first is that Trump was the beneficiary of undecided voters breaking his way. (There were some indications that this was happening as the election approached.) The other possibility (which isn't mutually exclusive) is that Hillary Clinton's supporters didn't turn out as heavily as polling expected. There's some evidence of that, too.
Having Clinton supporters stay home made Trump's goose-the-white-vote strategy more effective. Broadly, exit polls, which are both preliminary and notoriously fluky, suggest that Trump's electorate looked a lot like Mitt Romney's in 2012 — with a notable exception.
The columns below take total support for candidates and evaluate how much of that support came from each slice of a demographic group. Clinton's and Obama's columns should be taller than Trump's and Romney's as a general rule, since the Democrats received more raw votes. (Rounding errors mean this isn't always the case.)
First of all, note how broadly similar Trump and Romney's bases of support are.
Gender. More of the support for Trump and Romney came from men than did support for Clinton and Obama. The exit polls suggest that slightly more of Trump's support came from men than Romney's, but the difference is small.
Age. The younger vote, those lighter shades of blue, made up more of Obama's base of support than Clinton's. This fits neatly with the idea that younger voters weren't thrilled with Clinton as an option, but neat explanations can often be overly neat.
Race and ethnicity. Most of the support for both Trump and Romney came from whites. Exit polls suggest that Trump did about as well as Romney with nonwhite voters, which was probably not something we would have expected to see six months ago.
Income. We'll come back to this.
Race and gender. Clinton earned less support from white men than did Obama four years ago. In May, we noted that white male Democrats have consistently been more skeptical of Clinton than other Democratic voters.
Education and race. This is a place we see a big difference between Trump and Romney. Look at the blue bars and the dark red one. Trump saw much more support from white men without college degrees — enough to offset what he lost from white women thanks to lack of interest among those with college degrees. Clinton got more support from white women with college degrees than Obama, but less from those without.
The income question is an interesting one. On Wednesday, a graph of support by income group was popular on social media. It showed that lower-income Americans voted for Clinton more heavily than Trump — running contrary to the general argument that Trump was appealing to low-income voters. But, of course, there's a strong correlation between income and race. If we separate out white voters from the overall numbers, there's a clear trend among whites, with Trump earning more support from lower-income whites than wealthier ones.
This isn't scaled to composition of the electorate. The $50,000 to $99,999 range is the plurality of income brackets both among whites and overall, and is why the income graph in the charts above skews more wealthy than you might expect. We're getting into smaller and smaller sample sizes in the exit polls here (meaning more and more margin of error) but about a third of whites without college degrees fall into that income bracket. About a fifth earn more.
This was the Trump electorate: White voters, skewing more heavily toward those without degrees than Romney four years ago. It was doing just enough better with whites without college degrees to make up for those he drove away, while facing off against an opponent who wasn't able to re-create the Democratic coalition that offset the Republican white vote in the last two presidential cycles.
Trump's triumph wasn't building something new. It was taking a somewhat dilapidated property and rebranding it.