Sacramento, Nov 7 For the third consecutive presidential election, US polling organisations struggled to accurately predict Donald Trump's electoral performance, with some surveys dramatically underestimating his support in crucial battleground states.
The most notable polling miss came from veteran Iowa pollster J. Ann Selzer, whose final poll showed Vice President Kamala Harris leading Trump by 3 percentage points in Iowa, according to a New York Times article published on Monday, one day ahead of Election Day.
According to AP News' election result, the poll missed by 16 points, with Trump winning the state by 13.2 points (55.9 per cent to Harris's 42.7 per cent).
"The poll findings we produced for The Des Moines Register and Mediacom did not match what the Iowa electorate ultimately decided in the voting booth today," Selzer acknowledged in a statement on Tuesday. "I'll be reviewing data from multiple sources to learn why that happened."
Another significant miss came from The New York Times/Siena College poll released on Sunday, two days before the election, which showed Harris clearly leading in Georgia, North Carolina, Nevada, and Wisconsin, with Pennsylvania and Michigan tied.
However, Trump ended up winning or leading in all these states.
Even in reliably Democratic states, polls showed significant errors, Xinhua news agency reported.
In New Jersey, the polling discrepancies were particularly severe, with a Rutgers survey in mid-October missing Trump's final percentage by double digits. In Maryland, Trump overperformed his polling average by 4.1 per cent, while Harris underperformed by 1.2 per cent, according to The Independent.
James Johnson, founder of J.L. Partners, one of the few polling firms to accurately predict Trump's victory, told Newsweek Wednesday that the error stemmed from familiar problems: "The key thing is people made the same mistakes they did in 2016. They understated the Trump voter who is less likely to be engaged politically, and crucially, more likely to be busy, not spending 20 minutes talking to pollsters."
New York Times chief polling analyst Nate Cohn noted that "white Democrats were 16 per cent likelier to respond than white Republicans," suggesting a structural bias in their survey response rates, as reported by Vox.
Some pollsters still argued that the aggregate polling data wasn't entirely wrong despite the misses. Yahoo News reported on Wednesday that major election models showed the race as essentially a toss-up, with FiveThirtyEight and Nate Silver's Silver Bulletin each giving Harris 50 per cent odds of winning. Split Ticket put her chances at 53 per cent, while The Economist estimated 56 per cent.
The final Yahoo News/YouGov survey had Trump and Harris tied at 47 per cent among likely voters, with 6 per cent either backing third-party candidates or undecided. According to FiveThirtyEight's analysis cited by Yahoo News, presidential election polls had typically been off by about 4 points on average since 2000.
Online betting markets appeared to gauge Trump's chances more accurately. Major betting companies including Betfair, Kalshi, Polymarket, PredictIt, and Smarkets all gave Trump better-than-even odds of winning on election eve.
The polling misses have led to renewed criticism of the industry.
Comedy Central host Jon Stewart captured the public frustration during his election night broadcast. "I don't ever want to hear, 'We've corrected for the overcorrection with the voters."
As of Wednesday afternoon, Trump's popular vote margin stood at 3.5 per cent, though that could narrow as more results come in from populous states like California, according to Yahoo News. The former president has won five of the seven key battleground states, with Nevada and Arizona still to be called.
The continued difficulty in accurately polling Trump's support has prompted polling experts to question the industry's ability to adapt to changing voter behaviour and communication patterns, particularly among less politically engaged voters who may be less likely to participate in traditional polling methods.
Disclaimer: This post has been auto-published from an agency feed without any modifications to the text and has not been reviewed by an editor