clock menu more-arrow no yes mobile

Filed under:

Donald Trump has collapsed in general election polls

Tim Hiatt/MSNBC/NBCU Photo Bank via Getty
Andrew Prokop is a senior politics correspondent at Vox, covering the White House, elections, and political scandals and investigations. He’s worked at Vox since the site’s launch in 2014, and before that, he worked as a research assistant at the New Yorker’s Washington, DC, bureau.

Donald Trump loves to brag that he's ahead of Hillary Clinton in the polls. "I beat Hillary Clinton in many polls," he repeatedly insisted at a debate earlier this month.

Here on planet Earth, that isn't true or even close to true. In 33 general election polls tracked by HuffPost Pollster over the past two months, Trump has led Clinton in just one.


Trump now trails Clinton by 9 points in the HuffPost Pollster polling average and by 11.2 points in RealClearPolitics' average — and he's behind Bernie Sanders by even more in both. Any of these showings from Trump would be the weakest performance from a major party nominee in the past 20 years.

The trendlines for the billionaire are terrible too. Back during the waning months of 2015, he regularly came within a few points of both Clinton and Sanders. But historically, polls conducted so far in advance have been essentially meaningless.

Crucially, Trump's decline has happened just when these polls actually start to mean something. In newer polls, Trump almost never comes close to either Clinton or Sanders anymore. And in past races, changes in general election polling that have occurred during this period of the campaign have often ended up sticking.

"Were this a few months ago, I'd say, 'What's the big deal?'" says Christopher Wlezien, a political science professor at the University of Texas. "But polls today are much more meaningful than they were 90 days ago. And the polls today are much less favorable for Trump."

General election polls usually start telling us something around this time of year

Wlezien is the co-author, with Robert Erikson of Columbia, of The Timeline of Presidential Elections, a political science book that provides an invaluable guide for anyone trying to make sense of polling data.

The two crunched general election polling numbers for the past 60 years of presidential contests. And the below graph shows, essentially, how closely the polls tend to be related to the eventual outcome at various points in the campaign. The higher a point is on the y-axis, the more predictive the polls for that time in the campaign:

Adjusted R-squared predicting the Democratic presidential vote division from the Democratic division in the polls by date, starting 300 days before Election Day, in 12 elections from 1956 to 2012 (certain elections in that period are excluded due to lack of early polling data).
Robert Erikson and Christopher Wlezien, The 2012 Campaign and the Timeline of Presidential Elections, Figure 6

Unsurprisingly, Wlezien and Erikson found that as the campaign goes on, the polls start looking more like the eventual outcome.

But they also found that the rate of this improvement isn't linear. There are some volatile periods of the campaign in which the polls' predictive value surges quite quickly. And there are other, more stagnant periods in which the polls might change, but those changes don't tend to have any lasting impact.

One of those volatile and consequential phases is the one we're in right now. The authors found that around 300 days before the election (mid-January), general election polls are essentially meaningless — their predictive value is close to zero. But by the time we get to mid-April of the election year, polls explain about half the variance in the eventual vote split. And mid-April polls have correctly "called" the winner in about two-thirds of the cases since 1952.

Things still change quite a bit afterward, of course, but in that three-month period (which encompasses most or sometimes all of the contested primary voting), we've gone from polls telling us basically nothing about what will happen to polls telling us "about half the story" of the election, says Wlezien.

That means that poll changes between January and April have often told us a great deal. "A meaningful portion of changes in preferences during this period tend to stand the test of time and impact the election result," Erikson and Wlezien write.

In contrast, poll changes in the following three months or so — between April and the conventions (which this year are in July) — tend not to "stick" as much. The polls may move around, but their predictive value doesn't get much better. The next volatile period in which poll predictability usually surges is the convention season.

Why poll changes around this time often stick

The most likely reason polls around this time start telling us more is pretty simple: With the primaries winding down, voters now know more about the candidates and have spent more time thinking about them in an electoral context. "By the time we get to this point, we pretty much know the candidates," says Wlezien.

Indeed, much of the early poll movement is just because "people flirt," Wlezien says. "These are just survey responses, not signed contracts. Especially early on, some people tend to flirt around and say, 'Yeah, I could vote for so-and-so.'"

But as the election year goes on, voters get more serious. One of the key dynamics Wlezien and Erikson describe in their book is that "the campaign brings the fundamentals [of the election] to the voters." Rather than simply deciding between two names on a list, many voters begin to incorporate how they feel about things like the incumbent's job performance and the state of the economy into their decision-making.

Meanwhile, perceptions of the candidates themselves harden. This is what seems to be hurting Trump at the moment, since a sizable majority of voters view him unfavorably, and his GOP rivals aren't similarly declining in the polls — Ted Cruz's performance against Clinton has remained constant, and John Kasich's has even improved. So the poll change seems to indicate Trump's personal weakness, not any strength from Democrats in general.

However, another pattern Wlezien and Erikson identify could theoretically help Trump later on — maybe. Around this time of year, they write, it's common for some partisan voters to flirt with supporting the other party. But "it's not clear that the decision is fully factored in yet," Wlezien says. Historically, many of these voters have "come home" to their usual party after the conventions. "The conventions really, really bring it home; this person is a Democrat, this person is a Republican," he adds.

So, he suggests, it's possible that many Republicans who say they wouldn't vote for Trump now would in fact do so later on, which could lead to a tightening in the polls. "When it becomes clear that it’s a choice between the Democrat Hillary Clinton and the Republican Donald Trump, they might vote for the Republican," he says. This is what's tended to happen in the past.

But Trump is very different from the typical Republican, and perhaps his extreme personality could prevent these wayward partisans from coming home. Furthermore, the GOP's convention could end up being quite bitter and chaotic — depending on how it goes, it could divide the party's voters further rather than uniting them.

So we don't know how the polls will move in the future, and this election has often defied historical trends so far. What we do know, though, is that we're right around the time previous polls have started to become predictive — and this year's polls are saying Donald Trump is a loser.

Donald Trump put on a master class in victim-blaming