The New Statesman's US election prediction: Joe Biden has a 90 per cent chance of victory

A second victory for Donald Trump would require a much bigger polling error than that seen in 2016.

Sign Up

Get the New Statesman's Morning Call email.

As voters in the US go to the polls today, the Democratic challenger Joe Biden is tipped by the New Statesman’s election model as very likely to overthrow Republican incumbent Donald Trump and win the 2020 presidential election.

Our model gives Biden a 90.4 per cent chance of winning the White House. And Trump just a 9.3 per cent chance of retaining it.

On electoral college numbers, the most likely outcome is Biden getting 339 electoral votes, and Trump 199. The probability of a tie is half a percentage point. 

 

Trump’s 10 per cent chance should not be understated  nor too should it be overstated. The president’s very narrow route to the White House rests on a uniform repeat in poll error. His hopes rest on the stars aligning in all the correct states, as they did in 2016. But even then, Trump is behind in almost every state he needs, and a poll error on a scale similar to last time will not win them back. 

The New Statesman’s election model, had it been put to use in 2016, would have given Hillary Clinton a 71 per cent chance of victory. The reality at the time was that polls understated Trump’s core vote in a uniform way, and no hand-wringing or clever calculations four years after the fact will rewrite that chapter of poll history.

What the New Statesman's US election model says today - and what it would have said in 2016
Win probabilities for the Democratic candidate in key states

Survey makers say they have corrected for the mistakes of 2016, and in the midterm elections of two years ago they performed largely admirably. If there is to be a poll error, it is more likely to err on the side of understating Biden than it is of making the same mistake twice. 

And yet, there are still uncertainties: for example, language barriers and problems engaging the traditionally disengaged, mean pollsters have had a hard time getting a representative sample of Hispanic voters in their surveys. That means great difficulty forming clear expectations in states where Hispanics are the deciding demographic. What type of Hispanic voter is being undersampled? Spanish-speaking Americans, traditionally loyal to both their Latino heritage and the Democratic Party? Or disengaged and unenthused conservatives, tempted by Trumpian rhetoric on issues like abortion? 

[See also: Hispanics won’t be loyal Democrats forever]

To be plain, I do not yet know the answer. We could either be looking at Democrats falling short in the sunny states once again, or surging in a blue wave that will make electoral history. I can only say that this uncertainty means I am not convinced the Sun Belt states of Arizona, Texas, Florida and North Carolina are as likely to swing to Biden as the Midwest states of Trumpland.

And so we turn to the Rust Belt.

Donald Trump’s 2016 win in the Midwestern states  commonly referred to as the Rust Belt for their former reliance on manufacturing and industry  came from energising a bloc of votes that have, since the days of Ronald Reagan, been reliable Republicans: white voters without a college education. In the 1980s they represented nearly two-thirds of all Americans who turned out to vote, but today they make up just one in three.

Their influence is concentrated in the competitive states of Ohio, Iowa, Michigan, Wisconsin and Pennsylvania. In 2016, Trump enthused them to turn out for the Republicans in greater numbers than anything we’ve seen in recent memory. 

Polling over the past few weeks and months, however, suggests that the Republican grip on America’s white working class has been waning. Whites without a college degree, particularly white women, have been drifting to Biden since the summer months, with that drift turning to a trickle and an eventual torrent following the first debate back in September.

The size of the shift to Biden could be a case of bad sampling by pollsters, or something a lot more meaningful and real. By the end of this week we will know for sure, but what we do know right now is this: Trump’s grip on his white base is not what it was in 2016. By how much is for the analysts to answer in the days and weeks to come.

[See also: The US 2020 election swing states]

This campaign may have taken place in volatile and unprecedented times, yet Biden’s poll lead has been remarkably stable. Not once since the spring months has that lead narrowed to anything near the error margin. As Republican pollster Bill McInturff put it: “This election is probably the most competitive ten-point race I’ve ever seen.”

Speaking personally: I have faith in the New Statesman model. Joe Biden has had a good campaign. Months back, his character was of little relevance to voters. Since the debates, however, his favourability has ticked up, and electors have seemingly warmed to him as a person. They feel, as one voter in a focus group put it, that Joe Biden is “good enough”; that’s likely all he needs to beat Trump in 2020. 

[See also: Can Joe Biden defeat Donald Trump?]

 Ben Walker is a data journalist at the New Statesman

Free trial CSS