Here’s why we were all so shocked last Election Night: a heavily Republican demographic was dramatically undercounted.

So it’s D-Day in Virginia. The last polls before the voting gave Democrat Ralph Northam a narrow lead, right around margin-of-error territory. A month ago, the lead was more substantial, around seven points. But Northam has run a drab race, and Republican Ed Gillespie cut into the lead through the tried and true Republican method—by warning white Virginians that if Northam wins, the state is just going to become one big sanctuary city where undocumented immigrants are given free rein to rob and murder and rape, while liberals take advantage of the ensuing confusion by sneaking around the state tearing down statues of Jefferson Davis.

The experts say Northam should win. But if you want something to worry about today, I implore you to read this report issued last week by Ruy Teixeira, John Halpin, and Rob Griffin of the Center for American Progress. The three authors set about to check the accuracy of last fall’s exit polls.

They looked at “a multitude of publicly available data sources” to try to find out whether last year’s exit polls told the story of the election accurately. They found that in a few respects, the exit polls were right. But boy did they miss some stuff, and one fact in particular.

The exit polls, they found, dramatically understated the percent of the total vote cast by non-college whites—that is, the category that best correlates to the famous white working class. Exit polls had non-college whites casting 34 percent of all 2016 ballots, and college-educated whites casting 37 percent. The actual numbers, according to the new study? College-educated whites were in fact just 29 percent of the total vote, and non-college whites were a whopping 45 percent of the vote.

I know what you’re thinking: There was a massive Trumpian white working class surge that the exit polls missed. But even that isn’t really right. Because they studied the 2012 results too, and found that almost exactly the same thing happened then. The 2012 exit polls had both groups of white voters at about 36 percent. But Teixeira, Halpin, and Griffin found that college-educated whites accounted for 28 percent of the 2012 vote, and non-college whites 45 percent. So apparently, exit polls just consistently undercount the white working class.

The way in which one can speak of a Trumpian surge is that of course these white working-class voters gave Donald Trump a bigger share of their vote in 2016 than they gave to Mitt Romney four years before. The authors note, in fact, that if Clinton had equaled Obama’s 2012 levels of support among non-college whites, “she would have carried, with more robust margins, the states of Michigan, Wisconsin, Pennsylvania, Florida and Iowa (and just missed carrying Ohio). In fact, if she could merely have reduced the shift toward Trump among these voters by one quarter, she would have won the election.”

So there’s your explanation as to why we were all so shocked last election night. A heavily Republican demographic was dramatically undercounted.

Now let’s go back to Virginia. Clinton ended up winning the state comfortably—by around 200,000 votes, which was 5.4 percent, out of nearly 4 million cast. She racked up massive margins in the four Washington, D.C.-area counties that together were home to a little more than one quarter of all votes cast: Arlington, Fairfax, Loudon, and Prince William. These are the places where Virginia’s sanctuary cities would be, if the state actually had any! And they’re the places where Northam will roll, along with Richmond, Norfolk, Newport News (though not Virginia Beach—remember, that’s Pat Robertson country), and a few other cities.

He’ll need a big black and Latino turnout to hold Clinton’s margins in those places. The lieutenant governor candidate (they run separately, not as a ticket, in Virginia) is an African American, Justin Fairfax, which could help black turnout. But there was one dustup last month in which Fairfax was left off some Northam literature printed by a union that hadn’t endorsed Fairfax because he opposed two pipeline projects the union supports. This turned into “Northam slights Fairfax.” Hard to know how many people paid attention.

As for the Latino vote, it’s small, just 5 percent of the vote. But in a close election, that’s a lot. Northam reversed what we might call his early opposition to the opposition to sanctuary cities, so that might hurt him. On the other hand, Gillespie’s constant immigrant-bashing over the past month may jack up Latino turnout.

Huge margins among those groups usually mean the Democrat can lose big among whites and still win. In 2013, Terry McAuliffe beat Republican Ken Cuccinelli by 4 percentage points even though Cuccinelli won the white vote 56-36. But, and here’s the warning: Cuccinelli came closer than the final polls, which had McAuliffe ahead by around six points. Some of the polling under-sampled white voters.

So that’s the meta point here. If Gillespie pulls an upset, it will be because, well, because Northam ran an uninspired race, and because certain stupid sanctimonious lefties caused a little last-minute trouble and decided that making a statement about their purity was more important than defeated a candidate who ran a Trumpist, racist campaign. But it’ll also be because the white voters of Virginia turned out in numbers that surprised the experts—again—and showed that they’re going fight demographic inevitability tooth and nail.

Remember that the counties of northern Virginia tend to report late. It was called for Clinton around 10:30 p.m., but for a while there after the 7 p.m. poll closing, Trump held a narrow lead until that last surge. And even if Northam holds on, we’re still in for a long fight. But at least Trumpism will have been dealt an important blow.

This article was published by the Daily Beast