A new Decyphered experiment shows the size and scope of polling effects.
The post ‘I’m not influenced by polls, I just vote that way’: Florida voters can spot a bandwagon, just not the one they’re on appeared first on Florida Politics – Campaigns & Elections. Lobbying & Government..
—
Nearly 9 in 10 Florida voters will tell you polls have no real effect on how they cast a ballot. Six in 10 will tell you, in the very next breath, that polls absolutely sway their neighbors. Both things cannot be true. Karen Cyphers found out which one is.
The Sachs Media partner and Chief Research Officer, who holds a political science Ph.D. from Florida State and teaches this stuff to undergrads, dropped a piece on her Decyphered Substack this week that ought to be required reading.
The setup is a 1,000-voter Florida survey about polling itself. Some 41% follow polls at least somewhat closely before an election. A quarter say they ignore them entirely.
Six in 10 (62%) believe polls influence other Floridians, while just 13%, about 1 in 8, cop to being influenced themselves.
That’s nearly a 5-to-1 ratio between what Florida voters suspect about everybody else and what they’ll admit about themselves.
Cyphers, to her credit, doesn’t pretend to buy it.
“This fails every sniff — and statistical — test,” Cyphers said. “There’s a few phenomena that cause this gap, such as the third-person effect, social desirability bias, and the introspection illusion, to name a few.”
To get around the self-reporting problem, Cyphers ran an experiment.
Respondents got a made-up School Board race between two candidates nobody has ever heard of, Hayes (a retired teacher) and Marks (an accountant and parent). There were no party labels, just names and job descriptions. If you’ve ever looked at a down-ballot race, this is roughly the information environment voters actually operate in.
Each respondent was randomly assigned one of four treatments: no polling data, a dead heat, Hayes comfortably ahead, or Marks comfortably ahead.
With no poll attached, 55% said they’d be very likely to bother voting in the race. Shown a dead heat, that shot to 81%. That’s a 26-point jump generated from one sentence of invented numbers about two invented people.
Vote choice moved too. Hayes pulled 18% cold and 34% when shown leading. Marks pulled 22% cold and 32% when the lead was hers.
“That’s a bandwagon effect, and it showed up symmetrically,” Cyphers said. “Which is how you know it’s the polling information doing the work rather than anything about the candidates themselves.”
This is a clean experimental design, and it produces a finding no amount of asking outright would have surfaced.
Just 1 in 8 Florida voters will admit a poll has moved them. One line of fiction about a School Board race moved roughly twice that many.
The polling industry just had a month.
Median Strategies (remember them?) published fabricated results in the Los Angeles Mayor’s race and the Wisconsin and Nevada Governor’s races, then folded and announced the whole thing was a “social experiment.” The fake numbers cleared newsroom filters and landed on Los Angeles Mayor Karen Bass’ own campaign feed before anyone checked. The mastermind was Rahil Prakash, a 21-year-old fresh out of college.
Closer to home, FiftyPlusOne permanently banned The Public Sentiment Institute over polls the James Fishback campaign paid for, a sponsorship the firm never disclosed. The firm also admitted, in its own post-mortem, that it reassigned some respondents’ stated vote choices toward Fishback. Its explanation was that Lt. Gov. Jay Collins’ numbers “did not align with our read of the race,” so it “leaned” some of his supporters toward Fishback. A separate firm, Patriot Polling, got suspended in the same Primary over its own undisclosed Fishback money.
Voters clocked it, as the kids say. More than 1 in 5 had heard something about Median Strategies, and 15% about the Fishback business.
On Primary night, the actual voters gave Collins nearly two and a half times what Fishback got. Turns out the electorate is harder to reweight than a crosstab.
Cyphers also ran a three-question polling literacy quiz. Just 22% of Florida voters got all three right. Only 46% could say what it means for a sample to be weighted.
And her warning about “likely voters” is the kind of thing that should be tattooed on the inside of every political reporter’s eyelids. The phrase sounds like a standard. It is one only if somebody tells you how likelihood was determined.
“If you rely on self-reporting for being likely to vote, you run into the same biases. Everyone says they’re voting when a pollster calls,” Cyphers said, noting that only about half of Florida’s registered voters turned out in the 2022 Midterms. “Better to use voter file data to sample only those with a rigorous voting record.”
Last week, a spate of results from St. Pete Polls and others showed dead heats up and down Florida’s statewide ballot. Byron Donalds and David Jolly were a point apart for Governor, and Ashley Moody and Angie Nixon were tied for U.S. Senate. James Uthmeier and José Javier Rodríguez were too close to call for Attorney General.
Cyphers’ point is that a dead-heat poll is an input to the race as well as a measurement of it. Voters shown a tied contest reported a higher likelihood of voting than voters shown a blowout or shown nothing at all.
“The close polls coming out of Florida now will help produce the electorate that shows up in November,” Cyphers said. She left open which side stands to gain more from that.
Read the whole Substack, and then read the next poll that hits your inbox, which, given the date on the calendar, should be in about 20 minutes.
___
Ed. note: This story was drafted with assistance from AI. Editorial judgment, sourcing, and final review were performed by Peter Schorsch and the Florida Politics editorial team.
The post ‘I’m not influenced by polls, I just vote that way’: Florida voters can spot a bandwagon, just not the one they’re on appeared first on Florida Politics – Campaigns & Elections. Lobbying & Government..
