WEST DES MOINES, Iowa — When the Des Moines Register deploys its breaking news banner to report the results of a new Iowa poll, the state’s political operatives hold their collective breath.
Other surveys come and go, but the Register’s pollster, Ann Selzer, stands apart.
Selzer has attained near-oracular status across more than three decades of conducting surveys — a seer who can predict the political winds before Iowa voters caucus every four years. She has predicted upsets and larger-than-expected wins alike, and she is one of just six pollsters to receive an A+ grade from polling site FiveThirtyEight.
{mosads}“She’s most commonly referred to as the gold standard. She’s polled Iowa for so long and she knows it so well, but a lot of it is she uses a very specific methodology and she’s always been very accurate,” said David Kochel, a longtime Republican strategist who guided Iowa campaigns for former presidential candidates Mitt Romney and Jeb Bush. “Polling the caucuses is one of the most difficult feats in the polling business, because it’s so hard to get a handle on who’s going to turn out.”
Selzer’s methodology will be put to the test again in the lead-up to the 2020 caucuses, and this time around she has the added challenge of navigating new caucus rules that are expected to affect voter turnout.
But her consistency is so much a part of Iowa’s political culture that veteran campaign staffers tend to trust her more than their own pollsters.
Days before the 2008 caucuses, Selzer’s final survey showed then-Sen. Barack Obama (D-Ill.) leaping out to a 7-point advantage over then-Sen. Hillary Clinton (D-N.Y.).
On a conference call hours after the poll results came out, Clinton’s pollster, Mark Penn, criticized Selzer’s survey, arguing it included far too many first-time caucusgoers, according to three sources on the call. Plus, the survey was conducted during the holiday week between Christmas and New Year’s, not the best time to capture voters at their homes.
{mossecondads}Three staffers working for Clinton’s campaign who were on the call, and who asked for anonymity now that they lead other Democratic campaigns in Iowa for the 2020 race, felt a sense of dread. A few days later, when Iowa Democrats held their caucuses, Obama won by 7.8 percentage points.
Penn, who is now co-director of the Harvard CAPS/Harris Poll, which releases certain surveys exclusively to The Hill, did not respond to an email seeking comment. He is also an opinion writer for The Hill.
Selzer’s was the only poll that had showed him leading by such a wide margin — and the only poll that accurately predicted that almost 60 percent of caucusgoers that year would be first-time attendees.
“There were so many unknown elements. But what are we going to do, not poll?” Selzer recalled in an interview with The Hill.
Conducting a poll is not an entirely scientific endeavor. Every pollster has to make at least some assumptions about who should be included — which respondents are likely to show up and vote, and which are likely to stay home. The choices a pollster makes impact the final data.
Some pollsters make those assumptions with an eye toward voters’ past behavior: Did they show up at the last caucus? Are they regular voters?
In an interview in her office, set in a small house in a commercial district of West Des Moines, Selzer said her methodology is different: She asks voters whether they will show up, and trusts their responses.
“My method shows me who’s going to do what in a future event. What [other pollsters] do is say, ‘Well, who did it in the past?’ That’s fine, unless there’s change. And change is accelerating,” Selzer said.
“I have to think [other pollsters are] looking back. As I say, I like to poll forward,” she added. “My way is keeping my dirty fingers off the data.”
Her method seems to work: In 2004, she found then-Sens. John Kerry (D-Mass.) and John Edwards (D-N.C.) breaking away from former Vermont Gov. Howard Dean (D) and former House Minority Leader Dick Gephardt (D-Mo.) in the days before the caucuses. In 2014, she forecast Sen. Joni Ernst’s (R) surprisingly big win over Rep. Bruce Braley (D). And in 2016, she foresaw a nail-biter between Clinton and Sen. Bernie Sanders (I-Vt.).
Even the surveys where Selzer’s final numbers don’t mirror the exact result give important clues about the snapshot in time she captures.
Through all of 2012, Selzer found former Sen. Rick Santorum (R-Pa.) struggling at the back of the Republican pack. On the first day of her final poll before that year’s caucuses, Santorum had zoomed into double digits, while former Speaker Newt Gingrich’s (R-Ga.) support ebbed. Santorum’s support kept growing, and he narrowly won the caucuses over Romney and former Rep. Ron Paul (R-Texas) — both of whom registered within half a percentage point of what Selzer had measured.
Selzer caught the political bug in graduate school, after serving a congressional fellowship while working on a doctorate on communication theory. She wrote her dissertation on the gender gap apparent in national elections between 1960 and 1980, a paper she said she wrote with the explicit goal of getting a job in the polling industry.
After a brief stint at the Des Moines Register, she opened her own firm, then won the contract to conduct the Register’s polls in 1997. Her relationship with the Gannett-owned paper led to other gigs conducting surveys for the Detroit Free Press and the Indianapolis Star.
“My business model for a while was editors would leave the Des Moines Register and they’d go to another newspaper and they’d say, we need a polling operation. Call Ann Selzer,” she said.
But Selzer’s sixth caucus cycle comes with new obstacles for her and the Utah-based phone bank she uses. Voters who are already less likely to pick up their phones — either landline or cellphone — are becoming harder to reach. Once a voter is on the phone, a surveyor must read the tedious list of 20-something Democratic candidates who are running.
And this year, for the first time, Iowa Democrats will allow voters to participate in a virtual caucus by phone, rather than requiring them to show up in person — in effect, adding a second universe of voters to poll beforehand. Planning for the virtual caucuses alone has consumed much of Selzer’s time.
“The time we took to figure out how we’re going to do virtual caucuses and how we’re going to marry this all together, I doubt there’s another polling operation that has invested as much think-work into how they’re going to handle all of that,” she said.
But Iowa’s tradition of political engagement may make things easier. In a state where Democratic voters are used to being accosted by nosy reporters asking their political opinions, Selzer says the sky-high enthusiasm for the 2020 presidential contest has made it easier to keep potential respondents on the phone. She foresees a massive turnout on what is likely to be a frigid Feb. 3 caucus night.
“Any time you’ve got two dozen candidates in the state identifying people to show up, you’re going to have a big turnout. That’s just pure logic,” Selzer said.
Selzer has conducted three polls of Iowa Democratic voters in the past year. Her most recent, in June, showed former Vice President Joe Biden leading the field with 24 percent of the vote, trailed by Sanders, Sen. Elizabeth Warren (D-Mass.) and South Bend, Ind., Mayor Pete Buttigieg (D), all in the mid-teens.
She declined to say when the next Iowa poll will go in the field, or how many surveys the Register will commission between now and caucus night, a long-standing tradition of silence intended to block a campaign’s ability to influence the results.
She acknowledged the pressures of maintaining her accurate record, even with the added complexities of a huge field, high turnout and virtual caucuses. But she remains sanguine. Her methods will “work until they don’t,” she said.
“I may be golden, and I may be a goat, and if I’m a goat I’d better have something prepared to say. But part of what I will say is we gave it our best shot,” Selzer said.