Most Americans Don’t Trust Polls But Results Still Influence Whether They Vote, Survey Shows

Americans are approaching the midterm elections with deep skepticism about the surveys measuring the race, according to a new POLITICO poll.

The findings show that only 25% believe political polls correctly estimate votes, wxposing a credibility problem for an industry whose numbers shape campaign strategy, fundraising and public expectations. Yet distrust does not necessarily mean indifference, as many respondents said an unfavorable poll would make them more determined to vote.

Conducted by PF Global, the online survey of 2,083 U.S. adults ran October 24-27 and has an overall margin of error of plus or minus 2 percentage points. About one-quarter of respondents said polls suggesting a Democratic wave in November were wrong and expected Republicans to win key races.

Meanwhile, 41% said surveys tend to underestimate either Democrats or Republicans, revealing doubts that cross party lines. Republicans who supported President Donald Trump in 2024 were particularly skeptical. Roughly two in five said political polls underestimate their party.

That suspicion was stronger among Trump voters who identified as "MAGA Republicans." Three in five said polls indicating Democrats would win key races were incorrect, compared with about one-third of Trump voters outside that group.

Republican leaders have reinforced those doubts as closely watched contests tighten. House Speaker Mike Johnson said on Fox News that polls were "grossly underestimating Republican enthusiasm around the country," according to POLITICO.

Trump has labeled polls "fake" and "rigged," while Vice President JD Vance has argued that unfavorable surveys can discourage Republican voters. However, the skepticism also reaches Democrats. Just under one-third of respondents who voted for former Vice President Kamala Harris said polls tend to underestimate Democratic support.

Polling mistakes and fabricated surveys have added to the industry's credibility challenges. POLITICO cited incorrect expectations in Democratic primaries, including surveys that overstated Abdul El-Sayed's eventual winning margin in Michigan's Senate primary.

Separately, Wedgewood Polls admitted using fake data in several Democratic primary surveys after scrutiny from FiftyPlusOne, according to the report. Another organization published bogus polling about the Los Angeles mayoral race to test how unverified numbers would spread.

Another complication is that respondents can change their minds. One in five Americans surveyed said they had given a polling answer about their voting intentions that differed from how they ultimately voted.

Polls measure sentiment during a particular period, but election forecasts use polling and sometimes other information to estimate outcomes. Public confidence in those forecasts was also limited. Only 18% said they trusted poll-based forecasts more than prediction markets, compared with 12% who favored the betting markets. One-third trusted neither.

Nevertheless, polling can still motivate participation. Roughly 55% said they would "definitely" vote if surveys showed their preferred candidate far behind. Among Republican midterm voters who believed polls underestimate their party, roughly 75% gave that answer.