Curious Iowa: How do political polls work?

Ahead of Iowa’s first-in-the-nation caucuses, polls seek to tell us which candidate is leading with voters, who’s improving, and who’s falling behind. With so much focus on the state every four years - and so many polls - shouldn’t most Iowans eventually be called to participate in a poll?

That’s what Sandra Hoeg, of Decorah, wants to know. She wrote to Curious Iowa — a Gazette series that answers readers’ questions about our state, its people and the culture - to ask why she’s never been polled.

Hoeg has lived in Iowa for more than 18 years and votes in every election. She has a friend who has been polled, but neither Hoeg nor her husband have been called. Hoeg wrote to Curious Iowa asking, “Why have I never been polled? How can we trust these poll numbers in Iowa for accuracy?”

We spoke to pollsters from the Des Moines Register/NBC News/Mediacom Iowa Poll and the Iowa State University/Civiqs poll about Iowa caucus polling to answer her questions.

  Buy Photo

How are caucus polls conducted?

Not every poll uses the same methodology.

The Des Moines Register/NBC News/Mediacom Iowa Poll, which has been owned by the Des Moines Register since 1943, uses phone calls to survey likely caucusgoers.

“Caucus polling, the way that we conduct it, is the most expensive polling there is because it’s such a tiny group of people who end up going to caucus and we want to cast a net to be as close as we can to anybody who is eligible to go to caucus,” said J. Ann Selzer, President of Selzer & Company, which conducts the Iowa Poll.

  Buy Photo

Any Iowan on the Secretary of State’s list of registered voters who has a phone number is eligible to be called by the Iowa Poll. Previously, the Iowa Poll would call those categorized as active voters, but in 2021 the state changed how it defines active voters. As a result of that change, more than 565,000 Iowa voters were moved to “inactive” status in 2023.

“Whether you’re registered as no party or as Democrat, we’re going to contact you because people shift in terms of their political views and we don’t want to lock people in,” Selzer said of this cycle’s caucus polling. “There are some polls that are only talking to registered Republicans and we cast our net more widely, and as I said, even more widely this year. That is, if you haven’t voted in the 2022 general election, if you didn’t vote in the 2020 presidential election, we don’t care. We’re going to keep you on our list and in previous years, you might not have been on our list.”

The Iowa Poll uses random calling and does not accept volunteer participants. Selzer said cellphones were added to the poll’s methodology in 2009.

“I used to say polling gets more difficult, political polling, every cycle. And then I said every year and now I say every second because there’s a lot of changes in how the electorate is relating to media, what they’re using their phones for.” Selzer said. “You have to remember when we first started calling cellphones, they were literally just a phone. They didn’t do all of these other things.”

The Iowa Poll is released four to five times during an election year, depending on budget. NBC News and Mediacom are sponsors this year. They do not have authority over the workings of the poll, but they do get naming rights.

The ISU/Civiqs Poll uses a web-based survey. The poll, which partners with polling and data analytics firm Civiqs, is funded by the Lucken Professorship in Political Science and the Iowa State University Department of Political Science and College of Liberal Arts and Sciences.

Prior to 2020, caucus polling was done in-house at ISU, using talent from ISU’s political science department and statistics department and the Center for Survey and Statistics Methodology.

  Buy Photo

“And at some point, that just didn’t work anymore.” Lucken Professor of Political Science Dave Peterson said. “The phone calling gets difficult, lots of people don’t answer their phone anymore and so the number of phone calls we would need to make to get one person to take the survey was just increasing over time and it got increasingly expensive.”

The web-based survey methodology is cheaper, easier and allows ISU to produce more polls.

“Civiqs solicits people to take surveys. They sort of select people on the front end to ask them to participate and they build a panel of people nationwide,” Peterson said. “They’ve got a reasonable number of people in every state who have agreed to take polls and then when they get someone like me who comes along and hires them for a specific poll in a specific state, we draft the poll, we write the questions and then we send out the email invitation, when it’s time, to the people that Civiqs chooses.”

The ISU poll is also tracking voter intent over time by resurveying some respondents. This allows the poll to show how people may change favor from one candidate to another.

The ISU/Civiqs poll is operating as a five-part monthly poll and this cycle it has been interviewing Iowa Republicans and Independents who are likely Iowa caucus participants. The first part was released in September 2023.

Why have I not been polled?

It is possible that you received a phone call from the Iowa Poll, but you weren’t surveyed because you missed the call or didn’t answer. The Iowa Poll does not leave messages.

Selzer, whose company has been conducting the Iowa Poll since 1997, received a phone call in December asking her to participate in the poll. It was the first time that’s happened. She declined.

“If you do the math, statistically it’s 2.2 million Iowans are registered to vote,” Selzer said. “You might not be called.”

While the Iowa Poll made the distinction that it will talk to likely caucus goers regardless of political affiliation and voting history, other survey firms may further narrow who they want to talk to.

“You’re going to want to talk to people who are likely to [caucus],” Peterson said about the 2024 Iowa caucus, “and so you’re going to over sample Republicans and under sample Democrats. So if you’re registered as a Democrat right now, you’re less likely to get called … and so it’s a little bit more expensive to talk to you, for frankly what isn’t useful information if we’re trying to understand the caucuses.”

How accurate are caucus polls?

Polls aim to build a sample of people that is representative of the demographics of Iowa according to U.S. Census data

Peterson used the example of a 50-year-old white man with a college degree living in the fourth district.

“If there’s a few too many of us in the sample relative to Iowa, you count us a little less. So I would be 0.95 of a person, essentially. Or if there’s not quite enough of us, we’d weight me a little more, 1.05.” Peterson said. “So when we’re doing the calculations of percentages, instead of just sort of adding up the number of people who support Donald Trump and dividing by the number of people we interview, we add up those weights … And so in the end, what you get then is a representative sample.”

Peterson said though people can be distrustful of surveys, polling is a valuable way to understand the state of a race. He suggested looking at the totality of polling instead of focusing on single polls to get a sense of what is going on.

At the end of the day, polling captures public opinion at a specific point in time. Selzer said that although the polls have to stop surveying in the days ahead of caucus night so they have time to tabulate data, voter intent and opinion can continue to change in the days leading up to caucus day.

She recounted Rick Santorum polling at 5 percent in 2012.

“The first time he ever got double digits in our poll was the first night of interviewing in the final poll and it grew … and when you averaged it all out, it was just 15 percent,” Selzer said. “But if you looked at the trajectory, it looked like he was going to meet Mitt Romney.”

Rick Santorum won the 2012 Iowa Republican caucus, beating Mitt Romney by 34 votes.

“I never say never, anything could happen and it can happen very late,” Selzer said.

Latest poll results

The latest Iowa Poll surveyed likely Republican caucus participants Dec. 2-7. Former President Donald Trump was the first choice of 51 percent of those surveyed. This was up 8 percentage points from Trump’s support in the October Iowa Poll.

Florida Gov. Ron DeSantis polled at 19 percent while former U.N. ambassador Nikki Haley polled at 16 percent. No other candidates polled at more than 5 percent. The poll’s margin of error is plus or minus 4.4 percentage points.

The latest ISU/Civiqs poll surveyed likely Republican caucus participants Dec. 8-13.

Trump was the first choice of 54 percent of likely Republican caucus participants. DeSantis polled at 17 percent and Haley followed at 15 percent.

Nearly three-fourths of those surveyed said their minds are made up on their first choice. The poll’s margin of error is plus or minus 6 percentage points.

Have a question for Curious Iowa?

Tell us what you’d like us to investigate next.

Comments: bailey.cichon@thegazette.com