Donald Trump’s election has thrown into doubt the assumption that Democrats were emerging as America’s natural ruling party from a confluence of demographic and social changes.

I argued here last month that Trump’s candidacy was accelerating trends that could reshape the two-party system: the consolidation of lower-educated white voters in the Republican Party and the flight of college-educated whites and minority voters to the Democrats.

Many — myself included — predicted that these shifts would hand the election to Hillary Clinton.

That obviously didn’t happen. Was the theory wrong?

Not entirely

First, let’s not overstate Clinton’s defeat. She got a million more votes than Trump nationwide and lost some of the states that were crucial to his Electoral College victory by relatively narrow margins. Less than 100,000 votes separated the two in Pennsylvania and Wisconsin. Around 200,000 separated them in Florida and North Carolina. They were only tens of thousands of votes apart in Michigan. If 1 or 2 percent of the electorate had voted the other way, all our prophecies would have come true.

The first part of the theory — the consolidation of white voters without a college degree in the Republican Party — went to plan. It’s how Trump conquered the one-time Democratic strongholds of Michigan, Pennsylvania and Wisconsin.

The third part — the flight of minority voters to the Democrats — also came to pass. As expected, this trend accelerated compared to 2012.

But it wasn’t accelerating as fast as we thought.

Relatively fewer black and Hispanic voters turned out for Clinton than many anticipated.

The reason black turnout fell compared to 2008 and 2012 seems to be that Barack Obama, America’s first black president, drove more African Americans to the polls than anyone else could.

The Hispanic surge that wasn’t

The reason for lower-than-expected Hispanic turnout, argues Sean Trende at RealClearPolitics, may be that liberal and mostly white analysts are far too reductionist about this demographic and have wrongly decided that immigration policy is a make-or-break issue for them.

This ignores an awful lot of contrary evidence, such as the fact that a majority of Hispanic voters told exit pollsters in 2008 that immigration reform wasn’t important to them or voted Republican anyway.

In a normal election year, I would have agreed with Trende. Indeed, as many forward-looking rightwingers have argued, increasingly middle-class and mostly Catholic Hispanics should be a natural fit for the Republican, not the Democratic, Party.

But this year, Republicans nominated for president a man who didn’t just disparage the tens of millions of mostly Hispanic immigrants who live and work in the United States illegally; Trump went out of his way to offend Hispanic Americans, claiming, for instance, that the judge in a federal court case against his now-defunt Trump University could not be fair because his parents were born in Mexico — a claim even Republican House speaker Paul Ryan described as the “textbook definition” of racism.

This year, we had polls (PDF) showing Trump scaring the living daylights out of 82 percent of Latinos.

This seems to contradict Trende’s nothing-unusual thesis. As I wrote here a few days ago, it seems incredible that Trump’s behavior wouldn’t have driven up Latino turnout in Clinton’s favor.

Wary suburbanites

Then again, I also through Trump’s behavior would appall more college-educated, middle-class Americans, who, despite all the attention being paid to the white working class, decided the outcome of the election.

Exit polls (which were a little off in other areas, so take this with a grain of salt) said white college graduates essentially split their support 50-50 between Clinton and Trump.

That’s a huge improvement for Democrats from four years ago, when Mitt Romney won by the same demographic by 14 points.

Except it wasn’t enough and that is also why Clinton lost Florida and North Carolina and Pennsylvania.

I listed various reasons why the middle class might not have supported Clinton more wholeheartedly, from underestimated economic stress to white resentment.

The most convincing thesis to me seems Jonathan Haidt’s, who argued before the election, in The American Interest, that Trump was forming a coalition of nativist whites and “status-quo conservatives”: the sort of suburbanites who voted for Romney in 2012. These voters can drawn into an alliance with authoritarians, Haidt warned, if they believe that progressives have subverted the nation’s traditions and identity so badly that dramatic political action is only way they can stand athwart history anymore yelling “Stop!”

Long term

If this is the case, then Democrats can still take comfort from long-term demographic prospects.

Ruy Teixeira, one of the first proponents of the “emerging Democratic majority” theory, argues at Vox that the 2016 election may come to be seen as “the last stand of America’s white working class.”

Non-college whites are a shrinking demographic. Racial minorities are growing. And those parts of the country that trend Democratic are economically most vibrant.

The Trumpian populism of the 2010s will likely have no more staying power than the agrarian populism of the 1880s and 90s, which was similarly driven by demographic groups on the decline and similarly undercut by ongoing structural change.

That doesn’t mean Democrats can sit back and wait for the country to change in their favor.

Trende cautions that all coalitions eventually fall apart when governing forces parties to choose among its members. That is why the New Deal coalition collapsed in the 1960s, when Democratized prioritized the rights of blacks over the prejudices of Southern whites. It’s why the Reagan coalition collapsed in the 1990s, when blue-collar voters felt betrayed by the Republicans’ embrace of globalization.

Republicans could also turn the tables on Democrats and do even better with white voters, argues Harry Enten at FiveThirtyEight. In some Southern states, he points out, Republicans win close to 90 percent of the white vote. “Who’s to say that won’t happen in the Midwest?”

In politics, little is permanent and nothing is certain. But it still seems to me the big story in American politics are these demographic shifts between the parties and what they mean for policy, polarization and the nation’s collective sense of identity. The fact that these shifts didn’t produce the outcome we expected this year doesn’t mean they aren’t happening and having an impact.