Clinton supporters attend a rally in Tempe, Arizona, on November 2. (Daniel Acker/Bloomberg)

Many Democrats have believed that a coalition of minorities, millennials and single women would help create a new Democratic majority for years to come. Hillary Clinton’s presidential campaign was counting on it.

But the “rising American electorate,” as it’s called, failed to carry Clinton across the finish line. It didn’t even come close. According to national exit polls, among Latino voters she fell six points from President Obama’s numbers in 2012; she dropped five points each among 18-to-29-year-olds, unmarried women and African Americans. Together, these groups made up the same percentage of the electorate in 2016 as they had in 2012. Some of the battleground-state figures are even more striking. In Ohio, Clinton was 13 points behind Obama among 18-to-29-year-olds. In New Mexico, she fell 11 points among Latinos.

Why did the Democrats’ strategy fail so miserably? Ultimately, because they overestimated the strength of a coalition based on identity politics.

This strategy goes back, unfortunately, to “The Emerging Democratic Majority,” a book Ruy Teixeira and I wrote in 2002. We argued that by the end of the first decade of the century, a progressive-centrist coalition based on professionals, women, minorities and “close to an even split” of the white working class would create majorities for Democrats.

That’s what happened in 2006 and 2008. But in the wake of the Great Recession, Republicans were able to convince many white working-class voters that Obama administration policies had bypassed them, leading to a Republican wave in 2010.

Liberal Democrats consoled themselves with a modification of our theory. They argued that demographics were trending in their favor, with a growing portion of the electorate represented by unmarried women, millennials and “people of color” (a highly misleading grab-bag term). In their view, Democrats didn’t need to worry so much about the shrinking white working class. Proponents of the “rising American electorate” theory included pollsters Stanley Greenberg and Celinda Lake; Page Gardner, the president of the Voter Participation Center; and senior analysts from the Center for American Progress. Last year Gardner promised that “the Rising American Electorate will be game changers in the upcoming elections.”

Clinton’s campaign put a version of the theory — with the addition of a few other groups, such as people with disabilities — into practice. It ran ads targeted at these voters, including more than 2,500 Spanish-language ads from January through September. Clinton anointed singer Katy Perry and television star Lena Dunham as surrogates to court the millennial vote and called on LeBron James , Beyoncé and Jay Z to help get out the black vote. And the campaign hoped that her historic candidacy would appeal to the key group of unmarried women. At the Democratic convention, Clinton first appeared in a video showing her breaking a glass ceiling.

She sought to unite her coalition with a vision of inclusivity and a promise to advance the fortunes of each group (Latinos, for instance, got immigration reform, and millennials got relief from student debt), along with a warning that Donald Trump — who conveniently managed to insult each of Clinton’s target groups at one point or another — meant them harm.

But Democrats can’t win elections simply by appealing to the identity groups of the rising American electorate. These groups don’t add up to a sure majority unless one assumes the Democrat wins near-unanimity among them and the Republican only bare majorities or less among Republican-trending groups. Besides such traditional GOP constituencies as farmers, small-business people and managers, three groups of voters have become increasingly Republican: the white working class, defined as whites without a four-year college degree; whites with a four-year college degree but not an advanced degree; and seniors. While the proportional numbers of the white working class have been shrinking over the past few decades, they remain formidable, particularly in battleground states, and the numbers of four-year-degree whites and seniors have not been declining.

Here’s a back-of-the-envelope calculation: If you leave out white voters with advanced degrees — who by a small majority have backed Democrats — then whites without college degrees and with four-year degrees made up about 60 percent of the electorate, and much higher shares in states like Ohio, Wisconsin, Pennsylvania and Iowa, on which the election turned. In Ohio and Wisconsin, for instance, these groups composed somewhere between 75 and 80 percent of the electorate. The Republicans got about 60 percent of their vote, which was enough to win these states regardless of the Democrats’ large margins among nonwhites or Clinton’s success among single women and millennials.

What about the future? Won’t America’s turn toward a majority-minority nation — which the census forecasts for 2044 — create an automatic Democratic majority? Much of that reasoning rests on the rising population of Hispanics and on an assumption, implicit in a term like “people of color,” that as Latinos age, settle, and move up the income and education ladder, as other minorities have done, they will remain loyal Democrats.

But there are signs that Hispanics are following a trajectory more similar to that of the Irish than African Americans. In the American National Election Study of the 2012 vote, Hispanic support for Obama was 70 percent among those with only a high school diploma but 55 percent among those with some college. (There is no similar data available yet for 2016.)

There is also a political divide between first-generation immigrants and American-born Hispanics. According to a Gallup poll in August, Clinton enjoyed a far greater edge over Trump among foreign-born Hispanics than among those born in the United States. According to a Pew poll, bilingual Hispanics were far more supportive of Clinton than those who speak only English.

And there is a further complication. As sociologist Richard Alba has contended, when Hispanics intermarry with whites, they often identify their children as white. These, of course, are elusive socio-political categories masquerading as racial or national designations, but the liberals who argue that a majority-minority nation will favor Democrats are basing their claim on how voters identify themselves.

It is equally difficult to make a case that younger voters will inevitably tilt Democratic. In Iowa this year, Trump and Clinton split the 18-to-29-year-old vote that Obama had won easily. In Missouri, where younger votes backed Obama by 58 to 39 percent in 2012, they supported Trump this time by 51 to 40 percent. And in national and several key state surveys, there is some evidence that Democrats are losing their sure grip on 30-something millennials.

None of this suggests the converse: that the Republicans, with their current majorities of the white working class, four-year college graduates and older voters, have a lock on the presidency or Congress. What it suggests is that politics matters.

In the United States, our party coalitions are heterogeneous. The New Deal Democrats grouped together Northern big-city ethnics with Southern whites and union leaders with Texas oilmen. The Reagan coalition brought together Walmart and country-club Republicans. Obama’s coalition included Wall Street and Silicon Valley high rollers along with the inner-city poor. But the key to winning elections has been to define a majority that can include your coalition as well as a significant slice of your rival’s.

Politicians and parties have traditionally done this by purporting to champion the people (defined variously as the middle class, the forgotten Americans, the common man) against an elite of some kind that is blocking the achievement of a goal that the people hold dear. The existence of a common adversary is essential to holding together what might otherwise be feuding groups. Ronald Reagan’s coalition was united by a common opposition to “big government” liberals who were soft on communism. This year, Trump championed the “silent majority” (a Richard Nixon term) and Franklin Roosevelt’s “forgotten America” against an establishment that was standing in the way of “making America great again.”

Politics, of course, is vastly more complicated than this, and so is actual governing. But our successful politicians have adhered to this simple formula and have subsumed their platforms and programs on trade, taxes, immigration and foreign policy under it.

For Clinton, the adversary was Trump, but it was also, as the campaign developed, the “basket of deplorables” who backed him. While Clinton claimed she meant only a subset of Trump’s white working-class supporters, her charge was taken — with some justice — as applying to the group in its entirety. So instead of creating a majority that included her base plus a significant slice of potential Republican voters, Clinton defined her coalition against them. (Many liberal pundits reinforced this perception by repeatedly characterizing Trump’s voters as poorly informed and racist, even though in 2008 and 2012 many of these voters backed Obama.)

Infused with the promise of a rising American electorate, Clinton wrote off the significant slice of voters Democrats need — in states like Ohio, Michigan and Wisconsin — to win majorities. But she also failed, except among single and college-educated women, to sufficiently rouse her target groups. Millennials, it turns out, care about more than the relief of their student debts. Hispanics don’t necessarily rate immigration reform first among their concerns, and many of them are as leery of illegal immigration as one of Trump’s so-called deplorables. They want a larger vision of the future. In this year’s election, Clinton didn’t give it to them, and for that reason her vote fell short even among the groups she relied on.

Twitter: @JohnBJudis

Read more from Outlook:

Carlos Lozada: Who Is Latino?

Follow our updates on Facebook and Twitter.