Are Polls Undercounting Trump’s Support Again?

Republicans are hopeful that polls showing a dead-even presidential race don’t reflect the real level of support Donald Trump will receive in November.
The former president overperformed his polling numbers on Election Day in 2016 and 2020, but his chances of doing that again this year are lower because polls have gotten better at capturing likely Trump supporters who were undercounted in the past, according to pollsters tracking the 2024 race for the White House.
“Many pollsters today are using past vote [history] to correct for the Trump undercount,” said Cliff Young, the president of Ipsos polling.
Young and other pollsters said they’ve adjusted their methodology in numerous other ways to make sure they include an accurate mix of Democratic and Republican voters.
Polls now increasingly draw their samples from voter registration databases instead of relying on randomized lists of telephone numbers.
Registration-based sampling systems provide more detailed information on where respondents live. The practice makes it easier to weight the data to ensure that rural and other right-leaning areas that favor Trump aren’t underrepresented in the results, said Charles Franklin, the director of the Marquette Law School Poll in Wisconsin.
In another important polling change, pollsters for the past two election cycles have also typically weighted their results for education to make sure they accurately capture non-college educated voters who lean toward Trump.
Trump led President Joe Biden in polls for most of the year before Biden dropped out of the race in July and was replaced by Vice President Kamala Harris. Trump is now tied or trailing Harris in many national and battleground state polls.
Trump campaign spokesperson Steven Cheung said in response to a request for comment that the former president was “dominating” in polls with two months to go in the election.
“Polling shows President Trump is dominating both nationally and in the battleground states because voters want a return to pro-America policies that actually work,” Cheung told Newsweek.
Newsweek requested comment from the Harris campaign for this article.
Much of the debate around Trump’s effect on polling dates back to the 2016 election. That year pollsters routinely undercounted voters without college degrees, an oversight that led most polls to conclude Trump was trailing his Democratic opponent Hillary Clinton by several percentage points.
The biggest Trump undercount in 2016 occurred in the Blue Wall battleground states of Pennsylvania, Wisconsin and Michigan, which showed Clinton with a comfortable lead all the way up to Election Day.
Trump led in only one of the 133 polls that were taken in the three Blue Wall states between the start of September and the last day of the race, according to a Marquette Law School Poll study.
On the night of the election, Trump carried all three states and the surprise wins helped seal his Electoral College victory. He won a majority of the vote among people without college degrees, according to national exit polls.
In 2020, the political polling industry’s focus on education helped ensure most polls were more accurate than they had been four years earlier.
But Trump still received more votes in 2020 than polls suggested he would, further fueling the popular theory that there are millions of so-called “shy Trump voters” across the country who don’t admit in surveys that they plan to vote for him for president.
The theory is largely dismissed by polling experts who argue it’s difficult to determine from polls whether respondents who say they’re not backing Trump are not telling the truth, or are simply undecided about who they plan to vote for.
“This is a problem on the margins,” Young said, adding that the larger problem is figuring out how to capture the sentiment of likely Trump supporters who don’t typically vote.
in 2024, the challenge remains less about getting honest responses from potential Trump voters, and has more to do with getting any kind of a response at all, Franklin said.
“People who support Trump are mostly quite distrustful of the news media and other institutions, and that includes pollsters,” Franklin said. “It’s not that they do the [poll] interview and say they’re not voting for Trump but then do. It’s that they won’t do the interview to begin with.”
Pollsters said the issue appears directly linked to Trump, who drew voters to the polls in 2016 and 2020 who hadn’t voted regularly in prior presidential elections. Polls were more accurate in the 2018 and 2022 midterms when Trump wasn’t on the ballot.
It remains to be seen if pollsters can get it right with Trump back on the ballot.
“Pollsters certainly have made changes to how we do polls to try to reduce this problem,” Franklin said. “Whether they work, we won’t know till early November.”
Sam Chen, a Pennsylvania-based Republican strategist whose firm conducts some polling, said Republicans shouldn’t discount the polls this time around or assume that the shy Trump voter effect will sway the election.
“One of the mistakes the Republican Party made in 2020 was assuming that the silent [Trump] majority would stay silent. The reality was they already spoke up,” Chen said. “I don’t think the Trump vote is being undercounted as much as it used to be.”