Sure, Joe Biden has a healthy, steady lead in the polls. But could the polls be wrong?
It’s a burning question as Sunday morning brought a slew of fresh, and near-final, polls before Election Day. The Washington Post reported “Biden holding a slight lead over President Trump in Pennsylvania and the two candidates in a virtual dead heat in Florida” — two utterly critical swing states. The New York Times showed Biden holding “a clear advantage over President Trump across four of the most important presidential swing states,” including both Pennsylvania and Florida. The Wall Street Journal said Biden led Trump nationally by about 10 percentage points, in line with his margin for months.
For many Democrats who suffer PTSD from 2016 wondering if these numbers could be off, the answer is an anxiety-soaked "yes." Analysts say the polls are better this year, after making changes from last time. But here are two main ways the polls could go awry.
Pollsters could fail in the Midwest again
In 2016 pollsters had a Midwest problem. Despite correctly forecasting a popular vote win for Clinton, most polls showed a solid Democrat advantage in upper-Midwest states like Michigan and Wisconsin. When those states gave their electoral votes to Trump, Hillary Clinton’s hopes came crashing down.
In multiple postmortem analyses, pollsters found that their Achilles' heel was the white voter without a college degree. In past years, the polls were reasonably accurate without taking education into account. But in 2016, voters without a college degree shifted heavily to Trump, throwing off pollsters’ sample populations.
BuzzFeed News has journalists around the US bringing you trustworthy stories on the 2020 Elections. To help keep this news free, become a member.
All pollsters only speak to a sample of people in the US population, and those samples generally aren’t enormous — often around 1,000 voters. That sample should reflect the state’s population, with a proportionate mix of genders, races, and economic backgrounds. When one of those proportions is off, pollsters “weight” the sample by giving more statistical value to the responses of voters with that trait.
As it happens, voters with a college education are more likely to take a survey in the first place. The obvious solution is to weight those respondents less, and this year, some pollsters are doing just that. But many are not. The New York Times reported that this year, about 20% more pollsters are weighting by education, but that’s still fewer than half of the total.
Some of these changes got a test-run in the 2018 midterm elections. When the results came in, CNN’s Harry Enten wrote, “public polling passed this test with flying colors.” FiveThirtyEight’s Nate Silver unenthusiastically agreed, writing that “they’re about as accurate as they’ve always been.” In his view, polls have always been stronger than people give them credit for.
Yet some still worry about polling bias toward Democrats. Real Clear Politics election analyst Sean Trende wrote that “pollsters have consistently underestimated Republican strength.”
In an interview with the New Yorker, Trende said that weighting education is good, but not a silver bullet. “That was the big, easy fix. But maybe it was something else.”
The “shy Trump voter” could reappear
One of the boogeymen of the 2016 polls was the notion that when called by a pollster, some conservative voters might be hesitant to reveal their preference for Trump. The so-called shy Trump voter might actually lie to the pollster and say they’re voting for a Democrat or someone else.
Republican pollster Peter Cahaly told Politico that live-callers — traditionally the gold standard for accurate polls — actually undermine the results. “People avoid awkward conversations. So when a person you don’t know calls and asks how you feel about Donald Trump — and you don’t know how they feel — you tend to give them an answer that you think will make them look at you in the best light.”
If you have a news tip, we’d like to hear from you. Reach out to us via one of our tip line channels.
It’s an attractive explanation because it would seem to explain how polls and the media were taken by surprise in 2016. But trying to measure the proportion of shy voters is hard. Researchers with the University of Southern California’s Dornsife Daybreak Poll try to gauge it by asking respondents who they think their social circle is voting for.
“People might be embarrassed or afraid to admit to pollsters that they will vote for an unpopular candidate,” researcher Mirta Galesic told BuzzFeed News. “But they might feel more comfortable saying who their social contacts will vote for, because after all, the pollster doesn’t know their names and phone numbers.”
The social metric changes everything.
Right now, the Dornsife poll’s social circle question shows a popular-vote lead for Biden, but a Trump victory in the electoral college. There’s another plausible explanation, though, for people thinking their neighbors may vote for Trump, or expecting a Trump win even if they don’t plan to vote for him: There’s a lot of skepticism about polling and belief in Trump magically pulling out a victory after 2016.
How realistic are these scenarios?
Most experts believe neither of these problems are likely. The voter without a college degree was a big blindspot for polls in 2016, but education is just one variable out of many that pollsters measure and weight. The New York Times/Siena College poll uses 10 variables. Pew uses 12.
While Republicans did outperform the polls in a few key states in both 2016 and 2018, Democrats could top expectations too. A New York Times analysis showed that in Nevada and New Mexico, polls in both years underestimated Democrats.
As for the “shy” Trump voter, it’s more theory than observation, and many analysts think it’s wrong. When researchers compared how 2016 poll respondents said they would vote with how they actually did, they found that some people didn’t vote as they told pollsters they would — but at about the same rate as in previous elections. And the Morning Consult couldn’t find a significant difference between the presidential preferences of people who took online polls and respondents to live-callers.
The bottom line is that we are two days from one of the most polarizing elections in modern memory. Little to nothing will ease voter anxiety, especially that of Democrats who still feel burned by 2016.
But Silver had some advice for Democrats on Saturday night: “It's a good night to have a glass of wine or whatever and chill out about the polls. It's pretty unlikely that the overall polling outlook is going to look much different on Tuesday morning than it does now.”
“We've gotten a lot of data, most of it very recent. 91 million people have already voted,” he went on. “There's no October surprise unless you want to count the latest COVID spike, which isn't good news for Trump. Trump can win but there's not much indication of a last-minute surge toward him.”
Silver’s FiveThirtyEight puts Trump’s odds of winning at only 1 in 10.
But is that really such a slim chance? As Silver explained Saturday, a Trump win is still “plausible.”
And, of course, all the polling presupposes that the voter can vote, and the votes are counted. But Republicans are trying to limit voting — and even to toss out votes that have already been cast. That’s a wild card the polls can’t factor in.