House Speaker Mike Johnson is challenging polling that shows Democrats with a substantial lead ahead of the Nov. 3 midterm elections, arguing that survey researchers are failing to capture Republican support and may be underestimating the GOP’s chances.

“The polls are wrong,” Johnson told an audience in North Carolina on Thursday, according to the Financial Times. He argued that Republicans have become less willing to participate in surveys because they distrust pollsters, potentially skewing the results in Democrats’ favor.

The latest national numbers show why Johnson is disputing the current outlook. The Silver Bulletin’s generic congressional ballot average put Democrats ahead by 8.9 percentage points as of Oct. 16. The measure asks voters which party’s candidate they would support in a congressional election, providing an indication of the national political environment heading into the midterms.

Republicans point to recent presidential elections as evidence that polling can miss conservative support. A review by the American Association for Public Opinion Research found that polls conducted during the final two weeks of presidential campaigns overstated Democratic margins by an average of 2.7 percentage points in 2024, 4.6 points in 2020 and 3.1 points in 2016.

However, polling errors have not consistently favored Democrats in midterm elections. The same review found that polls overstated Republican margins by an average of 0.6 percentage points in 2022, while the average error in 2018 overstated Democratic margins by just 0.1 points.

Republican strategist Ford O’Connell argued that conservative voters’ reluctance to answer surveys could be affecting the current numbers. He predicted Republican victories in Texas, Iowa and Maine if previous polling patterns repeat themselves.

The Senate races are particularly important because several contests remain close in polling averages. According to figures cited by the Financial Times, Democrats hold leads of three percentage points or less in Texas, Alaska, Maine, Kansas and Iowa. A polling error large enough to erase those narrow margins could change the balance of several competitive races.

Data analyst Harry Enten said that if Republican candidates outperform polling averages by 3.75 percentage points, Republicans would retain Senate control by a 52-48 margin, according to Mediaite. That scenario depends on the size and direction of any polling error and is not a prediction of the final result.

Maine illustrates the uncertainty. A Suffolk University poll of 500 likely voters conducted Oct. 8-11 found Democratic challenger Troy Jackson leading Republican Sen. Susan Collins, 49% to 47%, with 4% undecided. The two-point difference fell within the poll’s 4.4-percentage-point margin of error. The university’s release did not specify how the results were weighted.

Beyond polling methodology, turnout could prove decisive. Republican pollster Sam Kay of OnMessage said GOP voters were “pretty depressed,” citing frustration over the war with Iran, diesel prices and President Donald Trump. Whether that dissatisfaction will cause Republicans to stay home on Election Day, or whether those reluctant to answer surveys will still vote, remains uncertain.

A White House official rejected the suggestion that frustration with Trump would suppress Republican turnout.

Historical trends also present a challenge for the president’s party. Since World War II, the party holding the White House has gained House seats in a midterm election only twice, according to the Financial Times. David Paleologos, director of Suffolk University’s Political Research Center, said the drop in participation between presidential and midterm elections tends to favor Democrats in Iowa because rural and suburban voters are more likely to sit out midterms.

Paleologos cautioned, however, that Maine often follows its own political patterns, making it difficult to apply national trends uniformly across the country.

With Election Day approaching, the central question is whether current surveys accurately reflect the voters who will actually cast ballots. Johnson and other Republicans argue that polling may be missing conservative support, but past midterm results show that survey errors do not consistently benefit either party.