As the results rolled in on Tuesday night, so did a strong sense of déjà vu. Pre-election polls, it appeared, had been misleading once again.
While the nation awaits final results from Pennsylvania, Arizona and other key states, it is already clear — no matter who ends up winning — that the industry failed to fully account for the missteps that led it to underestimate Donald J. Trump’s support four years ago. And it raises the question of whether the polling industry, which has become a national fixation in an era of data journalism and statistical forecasting, can survive yet another crisis of confidence.
“I want to see all the results in, I want to see where those deviations are from pre-election polls and final margins,” Christopher Borick, the director of polling at Muhlenberg College in Pennsylvania, said in an interview. “But there’s ample evidence that there were major issues again. Just how deep they are, we’ll see.”
In some states where polls had projected President Trump losing narrowly — like Ohio, Iowa and Florida — he had already been declared the winner by a comfortable margin by late Tuesday evening. And in states that had seemed more than likely to go for Joseph R. Biden Jr., like Michigan and Nevada, the result was too close to call last night.
To some degree, it was clear this process would be unwieldy. With large numbers of mail-in ballots and early in-person votes still uncounted, the earliest returns in most states gave an inflated sense of Mr. Trump’s strength, since voters in Republican areas turned out in higher numbers on Election Day — and those ballots were often the first to be tabulated.
It is also possible, said Patrick Murray, the polling director at Monmouth University, that Republicans’ efforts to prevent certain populations from voting easily had a sizable impact — a factor that pollsters knew would be immeasurable in their surveys.
“We need to know how many votes were rejected,” he said. “I won’t know, until somebody actually gives me some data, what happened. And it’s possible that we will never know.”
He added, “We will never know how many ballots were not delivered by the post office.”
But what is now clear based on the ballots that have been counted (and in almost all states, a majority have been) is that there was an overestimation of Mr. Biden’s support across the board — particularly with white voters and with men, preliminary exit polls indicate.
While polling had presaged a swing away from Mr. Trump among white voters 65 and over, that never fully took shape.
Partly as a result, Mr. Biden underperformed not only in polyglot states like Florida but also in heavily white, suburban areas like Macomb County, Mich., where he had been widely expected to do well.
Dr. Borick pointed out that while state-level polls had widely misfired in 2016, they held steady in the 2018 midterms. This led him to conclude that people’s views on Mr. Trump may be particularly difficult to measure.
“In the end, like so many Trump-related things, there may be different rules when polling an election with him on the ballot,” Dr. Borick said. “I’m a quantifiable type of human being; I want to see evidence. And I only have two elections with Donald Trump in them — but both seem to be behaving in ways that others don’t behave.”
Analyzing pre-election polls alongside exit polls is like comparing apples to apples — if one batch is rotten, the other probably is, too. But the exit polls can still provide a few clues as to what pre-election polls might have missed.
At the top of that list is Mr. Trump’s strength among college-educated white voters, particularly men. According to the exit polls, the candidates split white college graduates evenly — after an election season in which almost every major poll of the country and of battleground states had shown Mr. Biden ahead with white degreeholders.
And if there is a tendency for polls to underrepresent Mr. Trump’s support, it does not only affect college-educated voters, as “shy Trump” theorists have often suggested. Some studies had posited that highly educated Trump supporters might be more likely to say they preferred his opponent because of social pressure. In many high-quality phone polls before the election, Mr. Trump’s support ran in the mid-to-high 50s among white voters without degrees. But the results of the exit polls put his support with this group firmly in the mid-60s, about on par with his totals in 2016.
There is also no certainty about how much of the electorate these voters comprised. Pollsters puzzled over this question in the wake of 2016, and came to varying conclusions; this year’s results are likely to reignite that discussion.
On the subject of the coronavirus pandemic, it is also notable that compared with most pre-election surveys, the exit polls showed a smaller share of respondents favoring caution over a quick reopening. As of Wednesday afternoon, with final adjustments still anticipated to the data, there was only a nine-percentage-point split between the voters saying it was more important to contain the virus and those saying they cared more about hastening to rebuild the economy, according to the exit polls. In pre-election surveys, the split had typically been well into the double digits, with a considerable majority of voters nationwide saying they preferred caution and containment.
It appears that the virus was also less of a motivating factor for voters than many polls had appeared to convey. This year, the exit polls — conducted as usual by Edison Research on behalf of a consortium of news organizations — had direct competition from a new, probability-based voter survey: VoteCast, collected via an online panel assembled for The Associated Press by NORC, a research group at the University of Chicago. By looking at the divergence between the exit polls’ numbers and the responses to the VoteCast canvass, we can see that there were far more voters who considered the coronavirus a big-deal issue in their lives than people who said it was the issue they were voting on.
The VoteCast survey found that upward of four in 10 voters said the pandemic was the No. 1 issue facing the country when presented with a list of nine choices. But in the exit polls, when asked which issue had the biggest impact on their voting decision, respondents were less than half as likely to indicate it was the pandemic. Far more likely was the economy; behind that was the issue of racial inequality.
Not every pollster fared poorly. Ann Selzer, long considered one of the top pollsters in the country, released a poll with The Des Moines Register days before the election showing Mr. Trump opening up a seven-point lead in Iowa; that appears to be in line with the actual result thus far.
In an interview, Ms. Selzer said that this election season she had stuck to her usual process, which involves avoiding assumptions that one year’s electorate will resemble those of previous years. “Our method is designed for our data to reveal to us what is happening with the electorate,” she said. “There are some that will weight their data taking into account many things — past election voting, what the turnout was, things from the past in order to project into the future. I call that polling backwards, and I don’t do it.”
Inevitably, Robert Cahaly and his mysterious Trafalgar Group — which projected a number of close races in the battlegrounds — will also get another look from curious commentators wondering why his polls have been so close to accurate, both in 2016 and this year.
The firm was among the only pollsters to show Mr. Trump’s strength in the Midwest and Pennsylvania four years ago, and while its polls this fall may end up being a little on the rosy-red side, they appear to have gotten closer to the final horse-race results in states like Michigan, Wisconsin and Nevada than did other pollsters, by not giving short shrift to Mr. Trump’s strengths.