• Sun. Nov 29th, 2020

The Election May Hinge on Pennsylvania. Can the Polls Get the State Right?

Welcome to Poll Watch, our weekly look at polling data and survey research on the candidates, voters and issues that will shape the 2020 election.

If President Trump pulls off a come-from-behind victory on Nov. 3, it’s likely to run through Pennsylvania — one of the three states he won by less than one percentage point in 2016, and arguably the one that’s still within range for him.

Joseph R. Biden Jr., the Democratic nominee, has built a polling lead in Pennsylvania that is now considerably stronger than Hillary Clinton’s was on the eve of the election four years ago. But the president’s weakness is driving the race more than Mr. Biden’s strength. Mr. Biden has only recently achieved a positive overall favorability rating among voters in Pennsylvania, according to various surveys.

High-quality polls of Pennsylvania conducted this month have put Mr. Biden up by anywhere from five to 13 points among likely voters. But they have also shown 5 percent to 10 percent of those voters declining to express support for either major nominee.

A Trump comeback will probably depend on his winning over a good share of those undecideds while driving down Mr. Biden’s support among demographics that have long since turned against the president — such as older voters and suburbanites — but that have not swung as heavily to the Democrat. It’s a possibility that Democrats and pollsters alike are unwilling to rule out, especially given the lingering shock from Mrs. Clinton’s loss in 2016.

That year, a late swing tilted the race in Mr. Trump’s favor — and laid bare the problems with battleground state polls, most of which had not taken into account the difference in vote preference between white voters with or without college degrees.

This year, views of the president are much firmer, meaning there is less volatility in the race, and pollsters have tried to adjust for their mistakes from four years ago. It is now the industry standard for state polls to take into account education level, ensuring a fair representation of voters without college degrees. But survey researchers have other uncertainties to worry about this year — particularly the potential for widespread voter suppression, which polling has no established method of accounting for.

Keep up with Election 2020

Just over a week before Election Day, The Upshot’s polling average shows Mr. Biden with a six-point lead in Pennsylvania. That’s slightly narrower than his polling leads in Michigan and Wisconsin, the other two states that put Mr. Trump over the top in 2016, but wider than his leads in the Sun Belt battlegrounds of Florida, Arizona and North Carolina. As a result, if the election comes down to one state, there’s a good chance it will be Pennsylvania.

In a Morning Call poll conducted by the widely respected Muhlenberg College Institute of Public Opinion and released last week, likely voters in Pennsylvania were asked whether Mr. Trump deserved to be re-elected; 51 percent said no. That’s right in line with the share of voters who said in a separate question that they would vote for his opponent.

But in the poll a substantial chunk of Mr. Biden’s voters did not express an actively positive view of their candidate. Forty-two percent of likely voters expressed a favorable view of him, and 42 percent unfavorable.

The poll is an outlier in this regard, as a number of other high-quality Pennsylvania polls in recent weeks, including one from The New York Times and Siena College, have shown Mr. Biden’s favorability reaching or surpassing 50 percent. But as recently as mid-September, most of those same polling firms had shown less than half of voters expressing a favorable view of Mr. Biden.

Christopher P. Borick, a political science professor at Muhlenberg who runs its polling institute, said that these results brought back shades of 2016, when both candidates were widely disliked but Mr. Trump succeeded in convincing swing voters that Mrs. Clinton was the worse option. “It’s no surprise, when you look at the president’s campaign and his efforts to drag Biden down, that he thinks that’s the path again,” Dr. Borick said.

The key difference is that Mr. Biden has not faced the same antipathy as Mrs. Clinton, who was disliked by an equal share of Pennsylvania voters as Mr. Trump was. The president’s attempts to drive up negative opinions of Mr. Biden — including his misleading accusation that, if elected, the former vice president will ban fracking in Pennsylvania — have so far not succeeded.

In 2018, Democrats picked up four House seats in Pennsylvania, largely in suburban areas, and Mr. Trump is struggling in those regions. Mr. Biden has made them a primary focus, as he did on Saturday, when he spoke in Dallas Township, a suburb of Wilkes-Barre, hammering Mr. Trump for his handling of the coronavirus crisis.

But polls show that Mr. Biden has yet to fully capitalize on the softening of Mr. Trump’s support. Averaging the results of the most recent Times/Siena and Morning Call/Muhlenberg polls, in the Philadelphia suburbs Mr. Trump is polling nine points behind the share he received there in 2016 exit polls, but Mr. Biden is only three points ahead of Mrs. Clinton’s total among this group.

Troubling for Mr. Biden in a different way is the fact that he has not yet matched Mrs. Clinton’s share of support in Philadelphia proper. Averaging the results of the two recent polls, he has the backing of 73 percent of Philadelphia voters, down from 83 percent for Mrs. Clinton in 2016. According to the Times/Siena poll, Mr. Trump was supported by 24 percent of Philadelphians, nine points ahead of his exit poll numbers in 2016.

At the core of the president’s coalition are conservative and rural voters, and men roughly between the ages of 40 and 65. Beyond that, he has used an argument premised on pre-pandemic economic prosperity in attempts to tamp down Mr. Biden’s support among key Democratic groups, including Black, Latino and other nonwhite voters.

That effort has been only mildly successful, and across the state Mr. Biden has stripped away Mr. Trump’s support from both younger and older voters. In both polls, Mr. Biden beat Mr. Trump by more than two-to-one among voters under 30, far outpacing Mrs. Clinton, who won Pennsylvania’s youngest voters by just nine points, according to exit polls. And Mr. Biden held a double-digit lead among those 65 and up, a group Mr. Trump won by 10 points in 2016.

Image
Credit…Doug Mills/The New York Times

Mr. Trump won four years ago largely on the strength of voters who chose him as the lesser of two unpalatable options. Exit polls in 2016 found that 15 percent of voters in Pennsylvania had decided on their candidate within a week of the election — the highest number in any major swing state that year — and those voters broke heavily for Mr. Trump.

Data from exit polls can be imprecise and should be accompanied by a grain of salt when closely examined. But using the available numbers, a report compiled by the American Association for Public Opinion Research determined that those late deciders had given Mr. Trump a pickup of more than two percentage points, helping to explain the polls’ misfire.

This year, with voters’ views of Mr. Trump more firmly locked in, Dr. Borick said a bigger cause for concern among pollsters was the potential that complications to the voting process could have unexpected effects on turnout and balloting. If restrictions to voting access cause voters in specific demographics to have disproportionate difficulty casting a ballot, or if a sizable number of mail-in ballots are rejected, it may have a real impact on the final vote tallies.

Dr. Borick said that he had been talking with colleagues about how to account for this issue, but that they hadn’t come to a solution. “Trying to capture potential suppression, it’s just hard,” he said. “How do you locate it? How do you build it in? What are the parameters you can look for?”

The Pennsylvania Supreme Court recently issued a series of rulings rejecting lawsuits backed by the Trump administration that sought to restrict which ballots are counted. As a result, mail-in ballots postmarked by Election Day will be counted, even if received late, and it will be harder for those ballots to be thrown out because of discrepancies in voters’ signatures.

Democrats are far more likely than Republicans to say they plan to vote by mail in Pennsylvania this year, and various recent studies in other states have found that Black voters are much more likely to have their mail-in ballots tossed out than white voters.

The court decisions have been encouraging for Democrats, but there’s still the possibility of other complications, including voter intimidation on Election Day or subsequent legal challenges over the validity of ballots. Besides, Pennsylvania is allowing no-excuse absentee balloting for the first time this year, so the process of counting huge numbers of such ballots is going to be an experimental one no matter what.

In areas that are slow to count mail-in votes, the initial results will probably appear disproportionately Republican early on election night, only to have the margin shift in Democrats’ favor as absentee ballots are counted. If the election does hinge on Pennsylvania, it may mean an agonizing wait before a winner is declared.