This year's political polls in Florida accurately predicted President-elect Donald Trump winning the White House but by tighter margins than he actually did. For other races here, polls were less reliable, including some incorrectly predicting passage of amendments for recreational marijuana and abortion rights.
This year marked at least the third consecutive presidential election when polls underestimated the strength of Republican candidates, including Trump.
The postmortem analysis of this year's pre-election polling reflects the challenges of calculating public opinion, experts said. They underestimated Republican voter turnout in Florida. Democratic voters remained more willing to participate in polls, possibly skewing results. Polls that are taken weeks or months before Election Day fail to capture the sentiments of late-deciders.
Opinion polls are influential. They can help politicians understand what the public wants, and dictate strategies on political spending and messaging. Publicized results can also influence perceptions about which candidates or issues might win on Election Day, affecting turnout.
READ MORE: County-by-county look at how Florida voted in the presidential election
This election's polls in Florida included some good, bad and ugly:
Nearly all the mainstream polls projected Trump easily winning Florida over Vice President Kamala Harris, by an average of 8%. He actually won by just over 13%. The poll in mid-October by the Marist Institute for Public Opinion in Poughkeepsie, New York, which surveyed 1,510 Florida adults, showed Trump ahead by just 4 percentage points, with a 3.6% margin of error. It still showed Trump winning, but it far overstated support for Harris among Democrats.
The polls were much worse predicting how tight the race would be for Republican Sen. Rick Scott over Debbie Mucarsel-Powell. Florida Atlantic University's polling lab in late October showed Scott winning 50% to 46% with a 3.2% margin of error. Marist's poll in October was even closer, projecting Scott winning 50% to 48%, with a 3.6% margin of error. Scott actually won by nearly 13 points.
- FAU's polling lab in late October showed Harris slightly ahead of Trump nationally. Trump actually won both the electoral vote 312-226 and the popular vote by about 2.3 million votes. Its polling also showed Florida's recreational marijuana amendment with barely enough support to pass (it didn't). The University of North Florida's opinion lab in late October also showed Florida's abortion rights amendment with enough support to pass (it also failed).
Michael Binder, faculty director of the Public Opinion Research Lab at the University of North Florida, said this year's polls slightly underestimated Republican turnout in Florida – a challenging problem in every election cycle. That tilted the outcome of some polls, he said.
“It looks bad when all the directionality is the same, but that's something that we struggle with,” he said.
Political polls in 2020 were especially bad, with the highest errors in 40 years for the national popular vote and the highest in at least 20 years for state-level estimates of the vote in presidential, senatorial and gubernatorial races. Experts blamed Republicans' apparent unwillingness to participate in polls, likely following Trump's lead when he said polls were fake and intended to suppress votes.
The last time the country's top polling scientists studied the issue in detail, they couldn't identify any obvious answers about why polls were so bad: "Conclusive statements are impossible," the American Association for Public Opinion Research wrote in its November 2022 report. The same group said it will review 2024 political polls at its next national conference in May in St. Louis.
Kevin Wagner, co-executive director of FAU’s Political Communication and Public Opinion Research Lab, said polls showed overperformance by Democratic senators across states, including Florida, Michigan, Pennsylvania, Wisconsin, Arizona and Nevada. By Election Day, polls seemed to show congressional and Senate races looking more like the presidential race.
READ MORE: How Florida voted for U.S. Senate, House, and constitutional amendments
“The expectation is there would be less ticket splitting. But I mean, there still was more than you would normally expect, because you had Democratic Senate candidates winning in states that Donald Trump won,” Wagner said.
He said the poll's failure to predict the margin of Scott’s victory in Florida was especially worrying.
"That's one we're going to have to study a little bit more and understand why Senator Scott numbers were not more representative of where he ended up," he said. "Some of that may have been late deciders, some of that may have been just the nature of how these samples broke down."
The polling experts said the Florida polls might have missed the mark on the state's controversial amendments to legalize recreational marijuana and enshrine abortion rights because organized opposition to those measures – in each case led by popular Gov. Ron DeSantis – didn't heat up until closer to Election Day. The DeSantis administration employed state resources to fight both efforts.
"We did pretty well," Binder said. "If we had waited two weeks and polled, you know, a couple days before the election, I think we would have done much better on (amendments) three and four,” he said.
This story was produced by Fresh Take Florida, a news service of the University of Florida College of Journalism and Communications. The reporter can be reached at julialejnar@ufl.edu.
