Andrew McGill / The Atlantic

Remember “Soccer Moms?” They were all the rage in 1996, representing the slender slice of the suburban electorate Bill Clinton supposedly needed to win over to keep the presidency. Like the Macarena, the Soccer Mom turns 20 this year, but she doesn’t have the clout she once did. Now popular is the “White Working Class,” a catch-all label for a group of voters whose fears and anxieties have defined the 2016 campaign, or at least dominated media coverage.

But I misspeak. A good election-year Proper Noun Name describes a narrow set of swing voters who will determine the outcome of the race. The White Working Class, on the other hand, is neither narrow nor particularly swing-y. Most analysts say non-Hispanic white voters with a high-school education—and particularly those who are men—deeply favor Donald Trump, and have so for months. That demographic doesn’t favor Democrats; as my colleague Ron Brownstein noted earlier this year, no Democratic candidate besides Bill Clinton has gotten more than 40 percent of their vote in nearly 40 years. And there’s little doubt Trump’s America-First message has won him fans among folks who work with their hands for a living.

But how much space does the White Working Class take up in Trump’s coalition? A quarter of his votes? More than half? This is a bit tougher to figure out. Most polls describe the view of the population—say, that 81 percent of black Americans support Hillary Clinton, or that 12 percent of men want Libertarian Party nominee Gary Johnson to be president. But that doesn’t tell us what percentage blacks represent within Hillary Clinton’s vote, or if more women support Johnson than men.

So with the help of Morning Consult, a media and survey-research company, we flipped the polls around—to try to construct a profile of both candidate’s core supporters. Morning Consult ran this analysis using nearly 50,000 online interviews collected from June through August.

The first chart is pretty simple. You’ll note that we’ve normalized the two samples according to the candidates’ overall standing, so it’s easier to compare them numerically against each other. But let’s start drilling down.

Already, we’re seeing huge differences between the two candidates. Only about one in 10 Trump voters is black, Asian, or Latino, compared with nearly four in 10 Clinton voters.

So more than 60 percent of Trump’s coalition is white and doesn’t have a college degree—a category of people who could easily fit into the White Working Class subgroup. (The Morning Consult data counts anyone who has a bachelor’s degree or higher as having “graduated college.”)

This is often where the analysis stops and the stereotypes about Trump voters set in—they’re laid-off workers, fellas in hard hats looking for mining jobs, young high-school graduates still living with their parents. We already know some of these tropes don’t hold up, and that Trump-leaning towns are more likely to be economically prosperous and demographically insulated. But the Morning Consult data has a bit more to say:

Contrary to the popular narrative, men aren’t responsible for Trump’s support among non-college-educated white people, despite his testosterone-heavy campaign—women are. In fact, make that older women. This slice isn’t pictured here, but white women over 45 who didn’t go to college make up a full quarter of Trump’s base. Heck, they are Trump’s base. And they’re not poor, either: Only 23 percent of Trump’s white, non-college-educated voters make less than $50,000 a year.

For what it’s worth, here’s a complete picture of Trump’s voter base.

Clinton’s demographic map is relatively smooth—her supporters come from various backgrounds. But Trump’s don’t: In November, he’ll be hoping for high turnout from less-educated whites and low turnout from everyone else.

In some ways, we’re back to moms—though they're older and less educated than those who were fashionable in 1996. Trump needs them in November, but Hillary Clinton could make a run for them, too. White, older women without a college degree account for a fifth of all undecided voters, Morning Consult found, the highest chunk of unclaimed demography still left. We’ve seen plenty of television interviews with an Ohio guy in coveralls during this election. Perhaps we should be focusing on the woman standing next to him.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.