It’s possible that Donald Trump will win the 2016 presidential election. But to do that, he’d have to do better on Election Day than the polls currently indicate he will.
Usually, most of the uncertainty about whether polling support will translate into vote totals is located on the Democratic side, whose base includes lots of lower-propensity voters who don’t always come out to the polls (or who confound pollsters’ expectations by showing up in higher numbers than expected).
But 2016 is not a typical race, and Donald Trump is not a typical nominee. There are a couple of reasons to believe there are at least some “hidden” Trump voters out there — supporters who, for one reason or another, aren’t showing up in polls.
In order for there to be enough of them to make a difference on Election Day, though, the Trump campaign will have to find them — something it’s not at all clear they’re able to do, and that their candidate might have made harder.
The “shy Trumpers” exist — but there aren’t enough of them to decide the race
Essentially, there are two theories of why Trump voters might be missing from the polls. One is that they simply aren’t being counted by pollsters as likely voters. The other, though, is that they’re lying to pollsters about whom they plan to vote for because they don’t want to admit they’re voting for such a polarizing candidate.
“Shy Trumpers” could make the race look twice as good for Hillary Clinton as it actually does; if they tell pollsters they’re voting for Clinton when they’re really voting for Trump, that’s a net of two pretend votes in Clinton’s favor.
Enough of those, and you could see a solid-looking lead for Clinton open up out of nowhere — if one in every 25 Trump voters is afraid to admit her support, for example, a poll could show a 4-point lead of 52-48 among a group of voters who’ll actually split 50/50 on Election Day.
Luckily, though, there’s an easy way to test this. If the logic of the “shy Trumper” hypothesis is that voters will be afraid to admit to a stranger that they’re voting for Donald Trump — like the Bradley effect, invented to explain why a black candidate underperformed his polling — then they should only feel the need to lie in polls conducted over the phone, not online.
Morning Consult, in conjunction with Politico, tested just this hypothesis at the end of October. It found that there really are a few “shy Trumpers” — but that a majority of voters still say they support Hillary Clinton, even when there’s no one to judge them.
What’s particularly interesting about the “shy Trumpers” isn’t how many of them there are, but who they are. Morning Consult found that the voters most likely to lie to phone pollsters (or at least to tell phone pollsters something different than they’d say online) were educated, affluent voters — the sort of voters most likely to feel uncomfortable admitting that they support a candidate who’s been playing on racial tensions for his entire presidential campaign.
Over the phone, voters with a college degree supported Clinton by a huge 21-point margin; online, they only favored her by 7 points.
Voters making more than $50,000 a year, meanwhile, actually prefer Donald Trump 50-49; ask them over the phone, though, and only 44 percent of them would admit to supporting Trump, while 54 percent would claim to support Clinton.
The problem for Trump is that there simply aren’t enough of those voters to sway an election. As Morning Consult points out, the sort of educated, affluent voters who make up the “shy Trumper” constituency represent a bigger share of the electorate in a Republican primary than they do in the general election.
In the general election, many of the less educated, less affluent voters are also nonwhite — a demographic that overwhelmingly prefers Clinton. In other words, many of the people who tell a phone pollster they’re voting for Hillary Clinton are likely to mean it.
Can Trump turn out voters who’ve been missing since the Ross Perot era?
But what if the “missing Trump voters” are people who aren’t being contacted by pollsters at all, or are getting written off as unlikely to vote?
Trump appeals to a particular group of voters — whites without a college degree — who “don’t fit the bill of the traditional stalwart Republican,” says political scientist Donald Green. They “look, on their face, like low-propensity voters.”
You can already see the effects of this discrepancy in polling. Usually when pollsters shift, a few months before the election, from polling all registered voters to polling only “likely voters” — the people who say they’re certain to vote in the election, or people who have a consistent voting record over the past several cycles — Republicans usually start to look like they’re doing better. But this election, likely voter “screens” don’t actually help Trump all that much.
The Trump campaign argues that these voters will, in fact, turn out — that their sheer enthusiasm for the candidate, and their desire to stick it to the Washington establishment, will drive them to the polls. And there’s actually historical precedent for that.
H. Ross Perot had a gambit of appealing to “people who felt they had no voice,” Green says, that was surprisingly effective. “There were lots of people who did not vote in ’88 who voted in ’92,” he says. “’92 was a very high-turnout year. And Perot was still on the scene in ’96, but he was gone in 2000 — and those people disappeared.”
The problem with this theory is that low-propensity voters are, by definition, unlikely to turn out on their own — and the Trump campaign doesn’t appear to be devoting a ton of resources and expertise to getting them to the polls. They have a surprisingly sophisticated data operation, but they just aren’t using it for get-out-the-vote yet.
And even if they were, they’d have to contend with the message of their own party’s nominee.
“If you say the process is rigged,” says Green, “you’re sort of implicitly saying the process doesn’t really matter” — an especially bad message if you’re trying to reach out to low-propensity voters to begin with. Perot voters stopped showing up in 2000 because they didn’t think their votes made a difference after all; “I wonder,” Green muses, whether Trump has managed to compress 1992 and 2000 into a single cycle, inspiring motivators and then demotivating them.
Trump performs surprisingly well among the voters who say their likelihood of voting is between 0 and 10 percent. Those might be the real hidden Trump voters — the ones who are determined to stay hidden on Election Day.