For the third straight election for the US presidential race, it turns out that polls underestimated the vote for Donald Trump, which remains difficult to gauge, although some improvement in the polls was recorded this time.

“They failed to highlight the key piece of information: the overall Donald Trump push,” summarized Michael Bailey, a political science professor at Georgetown University.

In more than 90 percent of U.S. counties, the vote for the Republican real estate mogul was higher than in 2020, according to a New York Times report.

Most spectacular: the former president likely won all seven key states in the election, while the most recent polls had him in a threaded battle with Democrat Kamala Harris.

However, in five of those states where the results are now known, Donald Trump’s victory was achieved by one, two or three percentage points – within the margin of statistical error.

"We're going to help our country heal, we have a country that needs help, and it needs help very badly"

Donald Trump addresses his supporters as he stands on the cusp of election victory

Follow US election results live: https://t.co/cqzSQFdDUk pic.twitter.com/tNNMOBUwwU

— BBC Breaking News (@BBCBreaking) November 6, 2024

Donald Trump “may have been underestimated a little bit, but I think the polls, overall, ended up being rather accurate,” Kyle Condick, a political scientist at the University of Virginia, assessed for his part.

The polls “suggested that Trump had a good chance of winning, and he won,” he added.

The polling firms’ performance was expected to come under the microscope this year, after two consecutive and stunning failures: they failed to predict Donald Trump’s victory in 2016; and they overestimated the margin of victory for Joe Biden in 2020.

“Trump this time was underestimated by about two points” in key states, calculates Peter Acevedo, an Atlas executive.

In Pennsylvania, based on the most recent average of polls compiled by the website RealClearPolitics, the Republican was reportedly ahead but by just 0.4%. By this stage of the count, he was winning by a 2% margin.

In North Carolina, polls were giving Mr. Trump a 1.2% lead. He beat Ms. Harris by three points.

In Wisconsin, the Democrat was reportedly ahead by 0.4%. But it was the Republican who won by 0.9%.

The heart of the problem has not changed since Donald Trump’s emergence on the American political scene. And that is that a portion of his voters don’t even participate in the polls, they deny it.

In the most recent surveys conducted by the NYT and Siena College, “white Democrats were 16 percent more likely to respond (to the people conducting the survey) than white Republicans,” a phenomenon that intensified during the campaign, the paper noted two days before the election.

JUST IN – Trump soon to address nation and his joyous supporters pic.twitter.com/v2m8qOW5Gt

— Insider Paper (@TheInsiderPaper) November 6, 2024

Researchers tried to address the problem with solutions such as statistical adjustments, but apparently, this proved to be insufficient.

“The polls underestimated Trump’s progress” in the share of the “Hispanic electorate,” according to Pedro Acevedo, who pointed to the much larger-than-predicted victories the Republican racked up in Nevada and Florida.

This, he said, is also true in the case of white men, especially in the counties, he said, citing Iowa as an example. Polling on Saturday showed Ms. Harris leading in that state by three points, even though the state is solidly in favor of Republicans.

Donald Trump won the state by more than ten points.

“Those who decided at the last minute probably chose Trump (…) after the interviews were over,” J. Ann Sheltzer, in charge of this stunningly flawed poll, tried to explain to the Des Moines Register.

Hispanic and white men “were probably underrepresented in the sample,” Acevedo estimated for his part.