Why Donald Trump’s Latest Theories on the Election Don’t Hold Up

7 minute read

Trailing by every traditional campaign metric as the race for the White House enters its final fortnight, Donald Trump is leaning on a pair of unproven theories to peddle hope to his loyal supporters.

The first is the idea that a substantial percentage of Trump’s support isn’t reflected in voter surveys because respondents are wary of the social cost of revealing their preference for the Republican presidential nominee. Political scientists call this the “shy voter” theory. The second is that Trump’s unconventional candidacy will drive to the polls a silent crop of new voters—largely white, uneducated and male—who have declined to participate in past elections. Call it the “hidden majority” theory.

In the waning days of the race, Trump and his advisers have trotted out these theories in large part because the candidate’s favorite metric—polls—no longer looks so favorable. The candidate that used to gleefully tout polls showing him ahead during the Republican primary is now lagging Democrat Hillary Clinton. He’s cast reputable surveys as “crooked” and “phoney,” while blaming an international conspiracy for his floundering candidacy. His campaign is being outspent and out-organized in the field. Facing the challenge of keeping his core supporters motivated, the GOP nominee is deploying arguments designed to be difficult to debunk.

But while it is difficult to disprove the idea that Trump’s support simply isn’t reflected in the polls, political science, early voting figures and voter registration data illustrate why both of Trump’s scenarios seem implausible.

Take the “shy Trump voter” theory. Trump’s campaign has been arguing for months that some portion of the candidate’s supporters have been declining to state their preference for the GOP nominee because they don’t want to be associated with his positions or behavior. “It’s become socially desirable, especially if you’re a college-educated person in the United States of America, to say that you’re against Donald Trump,” Trump campaign manager Kellyanne Conway said in August.

The idea might sound farfetched—what person would be ashamed to reveal their political leanings to a total stranger over the phone?—but the notion that Americans mislead pollsters has often been an obsession of pre-election poll truthers. In 2008, for example, observers cast doubt on Barack Obama’s polling lead over John McCain by invoking the so-called “Bradley Effect,” a theory that argued that white voters had a tendency to lie to pollsters when they plan to vote for a white candidate over a black one. The theory is named after Tom Bradley, the black former Los Angeles mayor whose lead over a white candidate in the 1982 candidate’s race evaporated in polling booths. The idea had fretting Democrats in a tizzy in 2008, but it wasn’t borne out by election results.

And neither available data nor past performance suggest that Trump is likely to outperform expectations because voters are misleading pollsters. Despite winning the Republican nomination, Trump actually underperformed his vote expectations in the primaries, notes David Rothschild, an economist with Microsoft Research. And while Trump did better in anonymous online surveys than in live telephone calls, in the general election Clinton leads in both metrics by similar amounts. That’s a sign, Rothschild wrote for the Huffington Post, that “social desirability bias is not a meaningful factor in any possible polling bias for the 2016 presidential election.”

Alexander Coppock, a political scientist at Yale University, tried to further test the “shy Trump voter” theory. He conducted an experiment in which a representative national survey of thousands of adults were split into two groups. The first were asked a series of basic policy questions. The second also were asked a question about the presidential race in a way that allowed them to maintain privacy about their preference.

The result, Coppock wrote on the Washington Post’s Monkey Cage blog, showed no sign of a silent Trump majority. “When I used a survey tool specifically designed to measure such an undercurrent, I was unable to find it,” he wrote. “The study provides no hint of a silent majority that withholds its opinions from pollsters but will nevertheless turn out to vote for Donald Trump on Election Day.”

For his part, Trump contends that a silent crop of largely white new voters will emerge on Election Day. That theory was also employed to explain the consensus-defying outcome in the June vote for the United Kingdom to leave the European Union, a comparison Trump frequently makes on the campaign trail. “There’ll be a lot of Brexit happening in about two weeks. A lot of Brexit,” he said in Sanford, Fla. on Tuesday. Conway explained it more plainly in August: “We think there’s a big, hidden Trump vote in this country.”

But early vote metrics, along with voter registration data from key states, provides little evidence to support the notion of a massive shift in the electorate. First, take voter registration. The long and exciting Republican primary undoubtedly drew more GOP registrations, helping the party chip away at historical Democratic advantages in some swing states. Yet in some battlegrounds, much of the shift was the result of conservative voters changing their registration to reflect their existing voting habits. In other words, these new Republicans had already been voting Republican, and simply formalized their preference to participate in the primaries. Truly new registrants, on the other hand, may have favored Democrats. Since the GOP primary ended, Democrats have made substantial gains in registration in nearly every state.

Two weeks from Election Day, more than 8 million people have already voted early or by absentee ballot. Neither party is seeing an unexpected surge in turnout for the GOP. That’s significant, because both Republicans and Democrats expect about 40 percent of the overall vote to be cast before Nov. 8 in those battleground states that permit early voting.

It’s hard to know for sure how people have voted so far. But campaigns look for the demographic and partisan makeup of those who have voted early or requested and returned absentee or mail ballots. Comparing those numbers to past performance can provide important clues. Both parties cherry pick data that suggest they have an edge, but there is no indication of a surprising shift in the composition of the electorate.

Part of Clinton’s plan for victory has included running up the score early. On a conference call with reporters in early October, Clinton campaign manager Robby Mook said the Democrats could run up an “insurmountable lead” in Nevada, Florida and North Carolina during the early voting period, making the Election Day outcome in those states a given. And while it is still too soon to determine whether they have those states in the bag, there is also no data to show increased GOP turnout at levels that would indicate the Trump campaign’s hypothesis is correct.

More Must-Reads from TIME

Write to Alex Altman at alex_altman@timemagazine.com