Trump campaign officials are convinced that polls showing Joseph R. Biden with a big lead are tilted left by undercounting Republican voters and are missing a hidden cache of the president’s supporters who are leery of pollsters.
President Trump and his campaign staff are not under the illusion that he is winning at this stage, but they think the race is much closer than polls suggest, especially in the battlegrounds of Pennsylvania, Michigan and Wisconsin.
“They get some of these people who say they have no opinion or are undecided when they really are Trump supporters,” said a Trump campaign insider who did not want to be identified without authorization from the campaign. “The big thing the polls are doing is just underrepresenting Republicans, and it’s blatant.”
Mr. Biden’s presidential campaign refused to discuss either phenomenon, but they have said they are taking nothing for granted and are doing everything possible to avoid the sort of Election Day that stunned Democrat Hillary Clinton in 2016.
The numbers generally bear out claims of lopsided samplings in 2020 surveys.
A national poll released this week by Economist/YouGov showed Mr. Biden, the presumed Democratic presidential nominee, with a 9 percentage point advantage over Mr. Trump. The unweighted polling sample was 41% independents, 34.7% Democrats and 24.2% Republicans.
In 2016, Republicans made up 31% of the electorate, similar to their share of turnout in every presidential election since Ronald Reagan won the White House. The 2016 electorate was 35% Democrat, similar to the polling sample, and 34% independent, which is a much smaller share than in the poll.
A national poll by ABC News/Washington Post, which gave Mr. Biden a 10-point lead, used a similar sample of 39% independents, 30% Democrats and 24% Republicans.
The same Economist/YouGov poll recorded Republicans giving Mr. Trump an 87% approval rating.
Democratic political operatives and pollsters dismiss discrepancies in the samples as inconsequential or reflective of likely turnout this year.
“Republicans are whistling past the graveyard when they criticize current polling,” said Democratic strategist Brad Bannon. “There are fewer Republicans in the samples, but that’s because the president has turned off so many voters over the last three-plus years.”
Finding hidden Trump voters is more difficult.
Some of these voters emerged in recent focus groups conducted by the Trump campaign in which the participants were grouped according to party affiliation, said the campaign insider.
Members of the Democratic groups would quickly offer their disapproval of the president, but some members of the Republican groups would not offer their positive opinion of the president until they realized they were in a friendly crowd.
Pollsters missing these voters and underestimating Mr. Trump’s support, particularly at the state level, is certainly possible, said Christopher Borick, director of the Institute of Public Opinion at Muhlenberg College in Allentown, Pennsylvania.
“But I think a systematic error on the scale of 2016 is not very likely,” he said. “I think if the president’s campaign is banking on the polls systematically understanding his support, they are likely to be disappointed.”
Mr. Borick noted the strong performance of polling in the 2018 midterm elections and that many pollsters began weighting samples by educational attainment to correct what he saw as one of the primary causes of errors in 2016.
G. Terry Madonna, director of the polling center at Franklin & Marshall College, said the relatively small number of undecided voters in most presidential preference polls, which has been around 14%, would be higher if there were an army of hidden voters in America.
“I don’t see any evidence of a huge hidden vote,” he said. “I’m not going to tell you that there aren’t some people who will [untruthfully] tell you they are not going to vote for him, but there’s not a lot of evidence that I’ve seen that that exists.”
Pollsters also defend their performance in 2016 by citing Mrs. Clinton’s lead in national polls just prior to the election averaged 3%, close to her win the popular vote by 2%.
The state polls in 2016, however, missed the outcome by a wide margin. Most state surveys that year were not conducted in the final 10 days of the race, when many voters made up their minds.
A poll released Wednesday reflected a tightening race in battleground states, including the “Blue Wall” of Pennsylvania, Michigan and Wisconsin that Mr. Trump put in the Republican column in 2016 for the first time in a generation.
The Democratic-aligned Change Research/CNBC poll showed Mr. Biden’s lead shrinking to an average of 3 points — 48% to 45% — among likely voters in Arizona, Florida, Michigan, North Carolina, Pennsylvania and Wisconsin.
His leads ranged from 2-point edges in Arizona and Pennsylvania to a 5-point lead in Wisconsin.
Two weeks ago, Mr. Biden held an average 6-point lead over Mr. Trump in the same battleground states poll.
The poll did not provide a breakdown of the survey sample by party affiliation.
Regardless of the Trump campaign’s dubious view of the polling, the president has shown he is sensitive about his job approval numbers, which have taken a hit as the coronavirus crisis drags on.
Mr. Trump mused this week about why his approval ratings are worse than those of Dr. Anthony Fauci, the administration’s top infectious disease expert.
“So it sort of is curious, a man works for us, and yet they’re highly thought of and nobody likes me,” Mr. Trump quipped at a news conference. “It can only be my personality.”