There’s less than a week until Election Day. Polls show Donald Trump trailing nationally and in key battlegrounds, including Pennsylvania. And behind the scenes there are whispers about Republicans bracing for an electoral wipeout.
On the surface, there’s a lot about the 2020 presidential race that looks like the 2016 contest, when Trump shocked expectations, pollsters, and the media. He might yet do it again.
But there are some key differences this time. Trump is facing an even higher degree of difficulty to pull off another stunner. In fact, there are realistic scenarios in which Democrat Joe Biden wins running away.
“I would not be surprised if after all the votes are counted we have anywhere from a Trump squeaker to Biden winning it by double digits, and anything in between would not surprise,” said Patrick Murray, director of the Monmouth University Polling Institute.
With just days left in the campaign, we spoke to two of the region’s most venerable pollsters to understand how this race compares with 2016, and how to consider the final surveys.
What different about this year’s polling?
Polling averages — which pull together the results of numerous polls so as not to give any one survey too much weight — show Biden leading Pennsylvania by roughly 4 to 6 percentage points, according to separate calculations by Real Clear Politics, FiveThirtyEight, and the New York Times.
Several historically reliable pollsters have found even bigger leads for Biden (more on that below).
But we all remember that Hillary Clinton also seemed to have a solid lead in 2016. Six days before that election, the Real Clear Politics average had her ahead by 5 percentage points — even bigger than it shows Biden’s lead right now.
But Biden’s edge appears more steady.
There are a few reasons why. First, far more people have made up their minds. That leaves fewer undecided voters to swing the race in the final days, as happened in 2016. Biden is consistently racking up support in the 48 to 51 percent range, better than Clinton managed for most of her campaign.
Support for Trump is more solid, too, since Republicans have coalesced around him.
And far fewer voters are choosing third-party candidates.
At this point in 2016, roughly 1 in 5 voters was “volatile” — either undecided or only leaning toward a candidate, leaving room for big shifts, Murray said. The vast majority of those undecided voters broke for Trump.
The polling averages this year show that about 95% of voters now support either Trump or Biden. At a comparable time in 2016, only about 88% had picked one of the two major-party candidates, according to Real Clear Politics.
“The number of undecideds in this race is significantly lower than it was four years ago,” said Chris Borick, a pollster at Muhlenberg College. “Biden’s been more stable at a higher number than Clinton ever was in 2016.”
That points to another factor: Biden is more well-liked, or at least less loathed, than Clinton was.
Days before the 2016 election, only 36% of Pennsylvanians had a favorable view of her, Monmouth found. For Biden, 50% had a favorable view as of early October. That makes it harder to peel voters away from Biden — and for controversies and last-minute attacks to harm him the way they did Clinton.
“That’s very big and one of the reasons why I think Trump never wanted Biden as his opponent,” Borick said.
A significant chunk of people who told pollsters they were planning to vote for Clinton didn’t, particularly in rural Pennsylvania, Murray said, probably because they just didn’t like her that much.
But polls said Trump would lose in 2016. What about that?
It’s true that most pollsters, journalists, and political professionals in both parties expected Trump to lose.
The misses came in key swing states, like Pennsylvania, for two major reasons.
The first was that pollsters did not account for the vast differences in voting preferences among white people with and without college degrees. That had never been a major factor before, but turned out to be a stark dividing line. Key states like Pennsylvania, Michigan, and Wisconsin have far more white voters without college degrees, who make up the core of Trump’s support, and even now give him a chance to win reelection while again losing the national popular vote.
In the years since, most reputable pollsters have adjusted their methods to account for educational attainment. Major polls in 2018 were largely accurate, Murray noted.
The second factor is that the world is unpredictable and polls are not predictions, they are a snapshot of a moment time. Events can change things. In 2016, late news — the renewed FBI investigation into Clinton’s emails — put a damaging issue back into the spotlight just before Election Day.
Something like that could happen again, though so far this year even seismic events have failed to produce much change in the polls. Most high-quality polls show Trump’s support stuck in the 43% to 45% range. People’s opinions of Trump are largely fixed.
“Take the polls for what they’re worth: They’ll give you a sense of where the race is, but that final little point or two are often too hard to capture," Borick said.
How could Trump still win?
Trump is aiming to again score an upset by drawing out people who don’t usually vote. His campaign boasts that it has a powerful voter-turnout operation, and Pennsylvania polls are still close enough that even some small late movement in his direction could make for a close finish.
The polls, despite some adjustments, could be off again. Trump supporters claim there are people who lie to pollsters or won’t answer, though there’s little statistical evidence of this. If the polls are as off as they were last election, Pennsylvania would essentially be a toss-up, according to New York Times calculations.
And since far more Democrats than Republicans are voting by mail, problems with mail voting — voter error or court cases that disqualify some votes — could erase some number of largely Democratic votes.
Add those factors together, and there’s enough room to see a possible (but narrow) Trump win.
Could Biden win big?
There’s another possibility that few Democrats will speak out loud: Biden could actually win this pretty easily.
Remember, a lot broke well for Trump late in 2016 — the renewed FBI investigation, a deeply unpopular opponent, Democratic infighting, Russian interference aimed at discouraging Democratic voters — and his margin in Pennsylvania was still less than 1 percentage point.
Those things aren’t true this year, and several widely respected polls show Biden leading by 7 to 10 percentage points. Late movement in his favor, or even a polling error that underestimates Democratic support (as happened in 2012) could turn a solid lead into a big win.
Meanwhile, as the race closes, Trump is facing a spike in coronavirus cases nationally and within his administration.
Lastly, if early voting goes mostly smoothly, Democrats could have an edge in the final days. They’ll have a chunk of votes in the bank and be able to focus more attention on their remaining voters. Republicans will face the uncertainty of drawing their voters out to the polls amid a pandemic.
If all that happens, Biden might even approach the biggest victory in recent Pennsylvania presidential races: Barack Obama, who won by 10 percentage points in 2008.