Meet the anti-pollsters who got the last two elections right – by ignoring public opinion

According to a group of political scientists, Donald Trump was destined to win

The race between Kamala Harris and Donald Trump, if the polls were anything to go by, was extremely close.
Except, it wasn’t: Mr Trump was destined to win – according to one group of political scientists who forecast the exact result, state-by-state, nearly a month ago.
Peter Enns and his team studied every election since 1980 and came to the conclusion that just two issues are truly consequential in determining election results: the economy and the popularity of the incumbent president.
In marked contrast to polling, which largely under-calculated the swing towards Mr Trump this time around, Mr Enns’s model involves no direct contact with Americans on how they intend to vote.
Mr Enns, a Cornell University professor and co-founder of Verasight, told The Telegraph: “We plug in the 2024 numbers from 100 days or more in advance, and then say okay, if these past relationships hold, that lets us predict what will happen in each state, right?”
He was right. In a special forecast published in October, Mr Enns and three other political scientists predicted that Mr Trump would win 312 electoral college votes. Once Arizona officially declares its result, that is exactly how many the president-elect will have.
And this year’s prediction wasn’t an anomaly either: in 2020, he and his team called Joe Biden’s victory, correctly forecasting every state except Georgia – which was won by the Democrat by only 11,000 votes.
The implication is not that other issues were unimportant during the election. In Florida, a state won by Mr Trump three times, a ballot to make abortion a constitutional right received 57 per cent of the vote – just shy of the 60 per cent threshold needed to pass.
Pro-abortion policies were a key plank of the Democrats’ campaign but despite support for the amendment, Mr Trump achieved a resounding victory in the eastern state.
Essentially, Mr Enns says, the economy and voters’ negative views of Mr Biden alone were enough to swing the election in the president-elect’s favour.
Mr Enns’s model considers four economic factors: the number of jobs in the private sector, average hours worked by production workers, the unemployment rate, and the amount people are earning.
The model analyses the performance of each metric over the length of a president’s term – but places more statistical emphasis on the last 100 days as campaigning gets under way.
The reality is, none of the factors were on Mr Biden’s side.
The US economy added just 12,000 jobs in October – well below forecasts of 113,000. It was the lowest job growth since December 2020. September was stronger, adding 223,000, but growth in August was low too – sitting at just 78,000.
Average hours worked by production workers has also been declining since August, falling from 40.10 per week to 39.90.
The unemployment rate had been improving, marginally, dropping from 4.3 percent to 4.1 per cent between July and October. But before May, it had consistently been below 4 per cent since January 2022, signalling a rise in recent months.
Americans’ personal income has also been stagnating for months. Inflation also peaked at 9 per cent on June 22, after rising consistently for several years. While it has now come down, everyday Americans still feel the pinch of price rises.
From the economic point of view, “it’s not that voters are looking at the data,” Mr Enns explains, “It’s just that voters have a sense of economic conditions… voters have a gut sense of how things are going.”
Mr Biden has been one of most unpopular presidents historically, with his approval rating consistently polling below 40 per cent. Barack Obama’s, by contrast, averaged at 59 percent when he left office.
Mr Enns and his team also believe voters adjusted given that Mr Biden was no longer on the ballot. In fact, their model gave the president just a 1 in 10 chance of winning the election, compared with 1 in 4 for Harris.
It raises questions about how useful polls are in predicting the outcome of presidential elections, especially given how they miscalled the 2016 and 2020 elections.
Just two predictions came close to the final results in this election: British firm JL Partners’s forecast and The Telegraph’s own poll.
Other polls predicted a much narrower race. Nate Silver, the widely respected elections analyst, said it was “literally closer than a coin flip” heading into the final day.
The final New York Times/Siena poll also predicted a narrow Harris victory.
Ann Selzer’s poll even suggested a three-point swing to Mr Trump in Iowa, a safe Republican state.
Instead, every state except DC shifted Republican.
But Mr Enns, who appears to have found a winning formula, says the problem is not intrinsically to do with polls but how we use them.
“I think most polls are trying to provide information about what the electorate thinks, and when we try to use polls to predict the outcome, now it has to do a few things…who’s going to vote, how they will vote…all ahead of time. That’s really complicated,” he explained.
He also pointed out that in some states the results were very close. Compared with polling averages, Mr Trump won Wisconsin by one point and Michigan by 1.4 – both of which were in the margin of error (generally three points plus or minus).
As a result, despite miscalculating the swing to Mr Trump, the polling industry will take this year as a win.
“At a glance, in the battleground states, polls ran a little hot for Harris but really not so bad, but when you dig deeper, it’s all a little less impressive,” Michael Bailey, author of the book Polling at a Crossroads, told the BBC.
Mr Bailey explained that pollsters should “move on” from taking random samples and use more sophisticated modelling techniques.
Mr Enns’s highly accurate forecasting model also calls into question whether campaigns in the modern age have much of an impact.
It would be foolish to stop campaigning altogether, said Mr Enns, but he added that he believes such long campaigns and the amount of money spent on them are disproportionate in their material impact on election results.
Ms Harris’s campaign is believed to have spent nearly $1.9 billion, around $300 million more than Trump. The Democrats also had hundreds of thousands more volunteers out on the streets encouraging people to vote, all to no avail.
During the end of Mr Trump’s campaign, there were reports that his advisers were concerned about him going off-topic during long, rambling speeches at rallies.
There was also particular concern about a comment made by comedian Tony Hinchcliffe, who in October at a Madison Square Garden rally in New York described Puerto Rico as a “floating island of garbage”.
Nearly one million Puerto Ricans live in swing states, yet despite the comment – and intense media scrutiny and exposure by the Democrats – Latino voters increased their support for Mr Trump by 13 points, something no polls picked up on.
According to Mr Enns, voters had already made an adjustment for Mr Trump’s erratic behaviour and comments, and made up their minds on Mr Biden and the economy.
“There would have been a penalty for these extreme statements, and there isn’t, because the outcome again aligns exactly with what was forecasted…once people get in the voting booth, it’s Democrat or Republican,” he explained.
The importance of the economy and the popularity of the incumbent in deciding elections raises questions for the Democrats. Ms Harris was widely criticised for being unable to construct a coherent economic narrative and to separate herself from Mr Biden’s legacy.
It also shows the party’s next presidential candidate, to a large extent, how they might be able to win. But if Mr Trump is popular at the end of the next term, and people feel better off, depending on who stands, it may not matter who the Democrats pick.