Advertisement

The polls point to a Biden victory but can they be trusted this time?

<span>Photograph: Alexander Drago/Reuters</span>
Photograph: Alexander Drago/Reuters

For months, activists and Democratic party officials have been telling Joe Biden supporters that the only answer to the question “can we trust the polls?” is to go out and vote for Biden, and then get others to do the same.

“The polls are a mirage,” one organizer told the Guardian last month.

For partisans on either side of the presidential election, “go vote” remains the only sound advice. Tens of millions of people have acted on that advice and cast their ballots early in record numbers. Others are preparing to vote in person on Tuesday.

Related: US election polls tracker: who is leading in swing states, Trump or Biden?

After the dust has settled on the election, there will be plenty of time to analyze whether the state-level polling delivered a better picture of the race this year than it did in 2016, political organizers say.

But one thing is certain: the polls at the end of the 2020 presidential race are telling a very different story from the polls at the end of the 2016 race, and it’s a rosier picture for Biden than it was for Clinton.

An unchanging polling graphic has emerged in this presidential race: two lines, a blue one above and a red one below, running in parallel, separated by 8 points or so, for the entire year.

Those lines represent the national polling averages over time of a head-to-head match-up between Biden and Donald Trump, and they have never intersected.

In the last presidential election, between Trump and Hillary Clinton, the polling averages intersected every couple of months, weaving their way towards an endpoint that depicted Clinton ahead by three points. She won the popular vote by two points.

The averages this time have Biden up by 7.5 points (Real Clear Politics), 9 points (New York Times/Upshot) and 9 points (FiveThirtyEight) – depicting a lead that is two or three times larger than that depicted for Clinton.

The White House is not won through a popular vote, of course. But opening up a large national lead is impossible without opening up state-level leads. And the state-level polls in 2020 also depict larger leads for Biden in key battlegrounds than they did for Clinton – with some important caveats.

A sign in displayed in a garden of a home in Emmaus, Lehigh county, in the battleground state of Pennsylvania.
A sign in displayed in a garden of a home in Emmaus, Lehigh county, in the battleground state of Pennsylvania. Photograph: Matt Smith/Rex/Shutterstock

We can focus on three states Trump narrowly won in 2016 – Wisconsin (Trump +0.7), Michigan (Trump +0.3) and Pennsylvania (Trump +0.7) – and compare the work, then and now, of three influential outlets: FiveThirtyEight, an elections forecaster; Real Clear Politics, a polling aggregator; and the New York Times’ Upshot, a forecaster-pollster-aggregator.

The Upshot produced an insightful product for 2020 that delivers a polling average – and then delivers the same average “if the polls were as wrong as they were in 2016”. The tool currently depicts Biden with a lead in Wisconsin of 10 points, or 4 points if the polls were equally wrong; in Michigan of 8 points, or 4 points; and in Pennsylvania of 6 points, or 1 point.

The upshot: even if the polls were as wrong as they were in 2016 – which was very wrong indeed – Biden still would look to be ahead in the three key states that Trump won that year.

FiveThirtyEight projects vote share – meaning the margin between the candidates on election day. In each of the three states, the site’s projections are significantly better for Biden than they were for Clinton. In Wisconsin the number has moved from 5.3 points four years ago to 8 points, in Michigan from 4.2 points to 8.1 points, and in Pennsylvania from 3.7 points to 5.1 points.

Real Clear Politics averages polling results. Here, too, the estimated sizes of the Democratic candidate’s leads have changed since 2016, in Wisconsin from 6.5 points to 6.4 points; in Michigan from 3.4 points to 8.2 points; and in Pennsylvania from 1.9 points to 3.5 points.

The Wisconsin number, in which Biden’s lead in the polling averages is depicted as a fraction smaller than what we read about Clinton, is one of the aforementioned caveats – exceptions in the data to the narrative that Biden’s lead is significantly larger than Clinton’s across the board.

Such exceptions point up the fact that Trump continues to have a path to victory, reliant on an extraordinary turnout by his base of voters.

A woman holds a sticker before joining a car caravan supporting Donald Trump in Easton, Pennsylvania. Trump appears to have been losing ground among white women.
A woman about to join a car caravan supporting Donald Trump in Easton, Pennsylvania. Trump appears to have been losing ground among white women. Photograph: Caitlin Ochs/Reuters

But even that base appears to be shrinking, among seniors, white women without a college degree and white people overall. One of Trump’s rare areas of growth appears to be with Black and Latino voters, especially men.

Trump also could have a path to victory that in effect bypasses the numbers, in the potential chaos of a unique election featuring an unprecedented number of mail-in ballots, a lurking public health threat, a potential blizzard of lawsuits and undefined security concerns.

No recent election in the United States has featured such a destabilizing force as a president sowing false accusations of voter fraud, calling for the results to be declared before all the votes are counted and encouraging supporters to make a show of force at polling sites, as Trump has.

But US elections are decentralized affairs in which local officials methodically count and report batches of votes in thousands of jurisdictions, meaning that the voters will be heard – even though some cities expect to take days to count their mail-in ballots.

As many as 150 million might vote for either Biden or Trump, and at least half that many have voted already. The election does not look close in the polls. Before long we’ll know how the polls look through the lens of the results.