Opinion: Polls Can't Be Trusted, and Three Elections in a Row Prove It

I'll say what a lot of people think and few say out loud: polls can't be trusted. Not the national ones, not the state ones, and not the ones that will flood your feed before the next election.

People don't understand how much stigma there is around saying you voted for Trump. Say it at work, at a dinner, on a college campus, and see the room change. So when a stranger calls and asks who you're voting for, a lot of people would rather say "undecided," or hang up. Then the pollsters act surprised on election night. They have been surprised three presidential elections in a row.

The Record: Three Misses in a Row

In 2016, according to the American Association for Public Opinion Research (AAPOR), the pollsters' own professional body, polls "fueled high-profile predictions" that Hillary Clinton's chance of winning "was about 90 percent." Trump won. AAPOR's 2017 review found that "some Trump voters who participated in pre-election polls did not reveal themselves as Trump voters until after the election, and they outnumbered late-revealing Clinton voters." It said that could be late deciding or "the so-called Shy Trump effect."

In 2020 it got worse. AAPOR's task force wrote that the polling error "was the highest in 40 years for the national popular vote and the highest in at least 20 years for state-level estimates." Polls in the final two weeks were too favorable to Joe Biden by 3.9 points nationally and 4.3 points in state presidential polls. The task force found "a systemic overstatement of the Democratic-Republican margin in nearly every contest." And it admitted that figuring out why "appears to be impossible with the available data."

In 2024, AAPOR called it "a good year for public polling." Read a little further and you find this: "For the third straight presidential cycle, pre-election polls underestimated Republican vote shares relative to Democrats'." Three for three, all in the same direction. If a weatherman missed the same way three times running, would you trust his forecast?

Then there's Iowa. Three days before the 2024 election, the Des Moines Register published J. Ann Selzer's Iowa Poll showing Kamala Harris ahead of Trump 47% to 44%, as the Iowa Capital Dispatch reported. Trump won Iowa by 13 points. That is a 16-point miss on the margin, in the state's best-known poll.

Why Polls Can't Be Trusted

Start with the shy voter. Just before the 2016 election, researchers at USC's Schaeffer Center asked voters how comfortable they were sharing their choice. Trump voters were less comfortable telling telephone pollsters than Clinton voters were, 54 to 62 on a 100-point scale, the center reported in 2017. AAPOR's 2020 task force wrote that "it is plausible that Trump supporters were less likely to participate in polls overall," though it found no evidence that the people who did answer were lying. Whether people fib or simply hang up, it comes from the same place, and the poll misses them either way.

Next, almost nobody answers. Pew Research Center reported that typical response rates for its telephone polls fell to 6% in 2018. When 94 out of 100 people won't talk to you, you have to guess what the missing people think. That guessing is called weighting.

That's where my word "manipulated" comes in, and I mean it. From what I've seen, left-leaning outlets manipulate what you hear, and they do it on purpose. Not by accident and not by bad luck. Remember the riots in Paris, or the Black Lives Matter protests, described as "mostly peaceful" while buildings burned on camera behind the reporter? That wasn't a mistake. That was a choice. Polls get the same treatment. The numbers you see are built by people making choices, and too often those choices lean one way. AAPOR's own 2020 report says it: "The results of pre-election polling are increasingly as much a function of statistical adjustments, such as assumptions about likely voters, as they are a function of actual data." Change the assumptions, change the headline.

And I'm not saying this because I'm a Republican. I'm saying it because I see it. That's why The Digital Free Press aims to be as unbiased as possible, and that goes for the right too. A story about a politician getting a parking ticket isn't news, no matter which party they're in.

In 2024, Nate Silver accused pollsters of "herding," writing that some swing-state polls were "a little too close." AAPOR says its tests found no herding and credits "a broader reliance on political variables for sampling and weighting." Either way, the pollster's choices shaped the result, not just the voters' answers.

Then there are polls paid for by campaigns, parties and interest groups that release what helps them. And there are "push polls," which AAPOR itself calls "political telemarketing under the guise of research" and condemns.

Trump sued the Register and Selzer, alleging the Iowa poll was an attempt to "create a false narrative of inevitability for Harris." On Sept. 23, 2026, Iowa District Court Judge Scott Beattie dismissed the case with prejudice, writing that it sought to stretch Iowa law by "turning speech that enjoys the highest category of First Amendment protection into a liability." Selzer said, "Polling is a social science, not clairvoyance." Trump is appealing, the Register reported Sept. 26. So no court has found that poll was fraud. My point doesn't depend on one. A poll doesn't have to be a crime to be wrong, and to shape the story in the final days.

Ignore the Polls. Go Vote.

This isn't about who you vote for. It's about who gets to tell you what the country thinks. Polls move money, coverage and turnout, and they can tell voters the race is already over. That's the real danger: people who stay home because a poll told them it's decided.

So here's my ask. Ignore the polls. Don't let a number on a screen tell you your vote doesn't matter, and don't let it tell you the race is in the bag either. Get out and vote. Bring your family, bring your neighbor, bring a friend. The only poll that counts is the one on Election Day, and it only counts if you show up.

Hold Everyone Accountable.

Yossi Schmidt is the founder of The Digital Free Press.

Sources: AAPOR Ad Hoc Committee on 2016 Election Polling, "An Evaluation of 2016 Election Polls in the United States" (2017): https://aapor.org/wp-content/uploads/2022/11/AAPOR-2016-Election-Polling-Report.pdf ; AAPOR Task Force on 2020 Pre-Election Polling, report (2021): https://aapor.org/wp-content/uploads/2022/11/AAPOR-Task-Force-on-2020-Pre-Election-Polling_Report-FNL.pdf ; AAPOR Task Force on 2024 Pre-Election Polling, executive summary: https://aapor.org/wp-content/uploads/2025/10/Task-Force-on-2024-Pre-Election-Polling-Report-Executive-Summary.pdf ; USC Schaeffer Center, "Could 'Shy' Trump Voters' Discomfort With Disclosing Candidate Choice Have Skewed Telephone Polls?" (Feb. 2, 2017): https://schaeffer.usc.edu/research/could-shy-trump-voters-discomfort-with-disclosing-candidate-choice-skew-telephone-polls-evidence-from-the-usc-election-poll/ ; Pew Research Center, "Response rates in telephone surveys have resumed their decline" (Feb. 27, 2019): https://www.pewresearch.org/short-reads/2019/02/27/response-rates-in-telephone-surveys-have-resumed-their-decline/ ; Nate Silver, Silver Bulletin, "There's more herding in swing state polls than at a sheep farm in the Scottish Highlands" (Nov. 1, 2024): https://www.natesilver.net/p/theres-more-herding-in-swing-state ; AAPOR, election polling resources and "What is a 'Push' Poll": https://aapor.org/election-polling-resources/ ; Iowa Capital Dispatch, "Judge dismisses President Donald Trump's 2024 Iowa Poll lawsuit" (Sept. 23, 2026): https://iowacapitaldispatch.com/2026/09/23/judge-dismisses-president-donald-trumps-2024-iowa-poll-lawsuit/ ; Des Moines Register, "Trump appeals dismissal of his lawsuit against Des Moines Register" (Sept. 26, 2026): https://www.desmoinesregister.com/story/news/politics/2026/09/26/trump-lawsuit-des-moines-register-appeal-dismissal-iowa-poll-j-ann-selzer/91938045007/