Pro-Democrat samples fuel Harris surge; unfavorable vice president closes gap with Trump in polls

Vice President Kamala Harris’ growing lead in the polls may be exaggerated by oversampling Democratic voters, and some critics say it’s an intentional bid to bolster her momentum.

Democrats are celebrating after Ms. Harris closed the poll gap with former President Donald Trump and even surpassed him nationally and in some battleground states.

A few weeks ago, Mr. Trump held a consistent lead in nearly every critical poll in a matchup with President Biden. Ms. Harris has more than caught up with Mr. Trump since replacing Mr. Biden on the ticket on July 21.

She is ahead of Mr. Trump nationally by nearly 2 percentage points and leading slightly or tying Mr. Trump in all seven battleground states.

Republican Party analysts say polling methodology gives Ms. Harris a phantom advantage because many polls sample a far smaller share of Republicans than exit poll showings from the 2020 presidential election.

In other words, critics say, the polls oversample Democrats, perhaps purposely, to generate enthusiasm and boost fundraising for Ms. Harris. Last week, the Harris campaign announced it had raised $540 million in July, more than four times the amount Mr. Trump raised in the same period.

Mr. Trump, who is in a dead heat with Ms. Harris after leading Mr. Biden for weeks, mocked the oversampling of Democrats.

“It’s fake news,” Mr. Trump said during a rally in Michigan. “They can make those polls sing.”

Ms. Harris’ advancement in the polls has undoubtedly fueled excitement and donations.

She opened a 5-point lead over Mr. Trump in a Suffolk University/USA Today poll released Thursday. The 48% to 43% results in favor of Ms. Harris represent an 8-point turnaround from June when Mr. Trump was 4 points ahead of Mr. Biden.

Like several other major polls showing a Harris lead, the survey included more answers from those identifying as Democrats, 37.10%, than Republicans, 33.8%, according to the data. Among those polled, 48% said they voted for Mr. Biden in 2020 and 41.6% said they chose Mr. Trump.

Democrats also outnumbered Republican respondents in a poll released on Aug. 27 by Yahoo News and YouGov. This poll found Ms. Harris ahead of Mr. Trump by 47% to 46%.

Among those who participated in the poll, 33% identified as Democrats and 29% as Republicans. The percentage of Republican participants was significantly below the 2020 exit polling, in which 36% of voters identified as Republican.

In the same 2020 exit poll, 37% of voters identified as Democratic, a 1-point advantage over Republicans.

The Yahoo/YouGov poll showed Mr. Trump winning independents by 44% to 35% over Ms. Harris. The former president had a 17-point advantage with seniors in the poll and picked up 13% of Black voters and 39% of Hispanic voters.

Mr. Trump’s poll numbers “are all significantly better than 2020,” said Jim McLaughlin, who polls for the former president. “How can they have Donald Trump losing? It’s simple. They undersample Republicans.”

The skewed sample, Mr. McLaughlin said, aims to “tamp down support and donations for Trump.”

Skeptics of Ms. Harris’ poll bounce question how she can climb in the election polls without significantly raising the lowest job approval ratings for a vice president in modern U.S. history. Those approval ratings have slightly improved since Ms. Harris began running for president on July 21.

The polling analysis site FiveThirtyEight calculates a 42.3% approval rating for Ms. Harris, up from 37.1% on July 7.

The latest polls are compared with voter surveys taken ahead of the 2016 and 2020 elections, which showed far more support for the Democrats than the final results. The 2016 polls predicted a win for Democrat Hillary Clinton, and the 2020 polls forecast a much bigger win for Mr. Biden, whose victory was narrow.

Suffolk Poll Director David Paleologos defended the proportion of Democrats and Republicans sampled in the latest Suffolk/USA Today poll. The Democratic voter advantage, he said, is nearly identical to the advantage recorded in the 2016 presidential election exit polls, which he said more likely mirrors the 2024 race because of the pandemic’s impact on the 2020 election.

As for the poll’s inclusion of more Biden voters, Mr. Paleologos said it is inaccurate to directly compare those numbers with 2020 because the sample includes people who didn’t vote, picked a third-party candidate or refused to say how they voted.

The Suffolk Poll is ranked as one of the most accurate polls in the U.S.

Those who conducted the Yahoo/YouGov poll said party identification was based on “the respondent’s most recent answer given prior to Nov. 1, 2022 … and weighted to the estimated distribution at that time, which was 33% Democratic, 27% Republican.”

The poll selected respondents from YouGov’s “opt-in panel” to represent all U.S. adults, and the poll’s margin of error is approximately 2.7%, which is considered very reliable.

Other prominent polls surveyed a smaller share of 2020 Trump voters than those recorded in 2020 presidential election exit polls.

The share of Trump voters in the New York Times/Siena Poll sample last month was several percentage points lower than 2020 results in the critical battleground states of Pennsylvania, Wisconsin and Michigan.

In Wisconsin, for example, 44% of voters surveyed said they voted for Mr. Trump in 2020, nearly 5% down from the 49% he won in the state nearly four years ago. Mr. Biden won less than 50% of Wisconsin voters in 2020, but 52% of poll respondents said they voted for him.

The poll found Ms. Harris leading Mr. Trump by 4 percentage points in all three states.

“They have dramatically understated President Trump’s support both among all registered voters and in their likely-voter model. In each state, the gap between the survey’s recalled 2020 vote and the reported 2020 election results is more than the margin between Kamala Harris and President Trump,” Mr. Trump’s chief pollster, Tony Fabrizio, said in an internal memo. “Once again, we see a series of public surveys released with the clear intent and purpose of depressing support for President Trump.”

Siena Poll Director Don Levy said asking respondents whom they supported in the prior election has been “historically unreliable” and does not indicate that Democratic voters were oversampled for the survey. He acknowledged that pollsters may be grappling with “response bias,” or the over-participation of Democrats who are more willing to take part in the polls and the under-participation of Trump supporters who hang up.

Despite sampling issues, Mr. Levy said, Ms. Harris has closed the gap with Mr. Trump since she entered the race.

“You can criticize the sampling methodology, but if you look at the trend lines, the current movement in the race is undeniable,” he said.

Mr. Paleologos played down the oversampling influence on recent polls. He noted that nearly 20 states do not require voters to register by party.

“There is no fixed amount of registered Democrats and Republicans in the United States,” Mr. Paleologos said. “Party Identification is a function of many factors, including excitement for candidates, other races or ballot questions in the state. In short, it is a fluid, moving target.”

Mr. Paleologos pointed out that Democrats polled 3 points ahead of Republicans in the exit polls taken during the 2016 presidential election, which Mr. Trump won.

YouGov and New York Times/Siena polls are ranked among the nation’s most accurate.

Other analysts have questioned polls showing Ms. Harris in the lead.

Ryan James Girdusky, a political and polling analyst who writes on substack, pointed out that two new polls by Fairleigh Dickinson University and NPR/Marist College show Ms. Harris leading Mr. Trump among senior voters by 16 points and 11 points, respectively.

Mr. Girdusky said the advantage for Ms. Harris among seniors is an example of response bias, in this case, concentrated participation by older White liberal voters.

The same bias contributed to polling that inaccurately undercounted senior support for Mr. Trump in 2016 and 2020.

“With less than 70 days to go, Kamala Harris’ marginal lead in polls is built off the back of a group she will lose, and probably somewhere in the margin Biden and Hillary lost: four to nine points,” Mr. Girdusky wrote in The American Conservative.

The FDU poll showed Mr. Trump trailing Ms. Harris nationally by 7 points. Democrats and voters leaning Democratic comprised 42.3% of respondents, and 38% identified as Republican voters and leaners. Independents made up 13.1%.

The poll is not necessarily overselling support for Ms. Harris, FDU Poll Director Dan Cassino said.

“The question is only meaningful if we assume that partisanship is set, that some people are Republicans and some people are Democrats, and they never change,” Mr. Cassino said. “In fact, people move around — especially between the independent and leaner categories — all the time. All this tells me is that, when we were in the field, more independents were leaning toward the Democrats. Next week, those same people could be independents or leaning towards the Republicans.”

As for the poll’s senior support for Ms. Harris, Mr. Cassino said subgroups, including seniors, have a higher degree of sampling error and pollsters weight their calculations to try to eliminate those errors.

“We don’t have any reason to believe that older Trump voters, in particular, would be less likely to take a survey or that older Harris voters would be more likely, but we’d have no way of knowing if that were happening. Weighing can solve some of these issues, but not all of them,” Mr. Cassino said. “In general, is it possible that Democrats in that group were more likely to pick up the phone and do a survey? Sure. Much of our field time was during the Democratic National Convention, and so while we didn’t pick up on higher response rates among Democrats while we were in the field — we’ve never seen a problem with shy Trump voters — it’s certainly possible.”