An old “PBS NewsHour” clip, in all its dull, steady glory, pretty much pinpoints the moment that famed Iowa pollster Ann Selzer got hot.
Shot on New Year’s Day 2008, the video features her sitting face to face with reporter Judy Woodruff in what looks like a living room; both women are wearing turtlenecks, Selzer’s underneath a sweater, Woodruff’s a blazer, and while the atmosphere appears cozy, the interview isn’t particularly.
“She’s standing a little too close to me, and I’m a little bit backed up against the wall,” Selzer told me, laughing. We were sitting in her office in West Des Moines in December, only weeks before Iowa’s Feb. 1 first-in-the-nation caucuses, and Selzer had just asked if I wanted to hear about the time she predicted that an upstart senator from Illinois named Barack Obama would win the Iowa caucuses with unprecedented voter turnout, sent the American political universe into an epic tizzy, and became one of Washington’s insider celebrities.
I did want to hear.
The interview starts with Woodruff questioning the methodology of Selzer’s Jan. 1 poll for the Des Moines Register, as Mark Penn, chief strategist for Hillary Clinton, had in a widely circulated email memo.
Woodruff: Today’s Iowa poll, it’s out. You’re assuming that 60 percent of the voters in the Democratic caucuses will be first-time caucus attendees. How did you assume that? Why did you assume it?
Selzer: Well, actually, I assumed nothing. That’s what my data told me. We put our method in place, and we let the voters speak to us.
As Selzer speaks, the camera zooms in close, about the most dramatic shot employed by public television news, until you’re staring directly into the auburn-framed face challenging the conventional wisdom of a Clinton dynasty.
Despite a pleasant affect — she nods and smiles as Woodruff speaks — Selzer betrays a hint of testiness during the exchange about the poll that she now says “pretty much made my career.” (We all know how that story ends — Obama wins Iowa, emerges from the nation’s primary contests victorious, becomes president.)
“We were just doing what we do, minding our own business,” Selzer said to me eight years later.
By the time Selzer puts out her next poll from this small office on a street that boasts not one but two quilting shops, the poll-reading public will be more slavishly devoted to the opinions of the people of Iowa than ever. Selzer, who has overseen nearly every one of the Register’s Iowa Polls since 1987, is almost universally thought to hold the keys to their secret, fickle hearts.
When Homer wrote of his hero Odysseus, he was “that godlike man.” The D.C. poets use the same laudatory epithet-style when they sing of Selzer; she is uniformly “the great” or “most respected” or, as I saw on a book jacket recently while walking through the Des Moines airport, “Iowa’s ‘polling queen.’”
It’s a reputation that is not unearned: She has on more than one occasion foretold an outcome that no one else saw coming, been pilloried for it, and not budged — the polling Cassandra of Des Moines. And her rise has come at a show-me-a-hero time in polling, as the industry rapidly loses its grip on what has made it tick for so many years: Landline calls are fewer and farther between, and Internet polls might be the wave of the future. Which is why this one question seems so pressing: What makes Ann Selzer so good?
Iowa politics is not Selzer’s birthright. She was born a couple of states away in Topeka, Kansas, the middle child in a still-close family of five. Her father was a Mayo Clinic-trained surgeon. Her mother had a master’s degree in creative writing and had been a reporter at a daily newspaper in Kansas City, “so we were a family about language and grammar,” Selzer said, not numbers. She stayed close to home for her undergraduate degree, attending the University of Kansas and starting off pre-med — her father died when she was in high school, and she worked for his old partner for a time, changing bandages and prepping patients for procedures. She found it too rote. “I’m interested in original ideas,” she said.
Selzer found them in her first semester of college in 1974 when she did a survey project in a communications class that unexpectedly allowed her to combine twin interests in language and data. “I was probably born to be a data girl,” she said. (The first poll she did was of neighborhood moms when she was 5, on the question of whether a nickname her family had given her was unflattering. She now accedes that the poll might have been skewed on account of the leading nature of the inquiry.) She finished the semester thinking how fascinating the class was, that there couldn’t possibly be a department in it.
But there was. And there was grad school for it, too — a communications theory and research program at the University of Iowa. She nearly went to the University of Wisconsin, but during a visit to the school, Selzer had trouble “getting two grad students in a room” to talk to her, which if you spend any amount of time with her, you understand would be a problem. She’s simpatico and a joiner; in Des Moines, she belongs to a discussion club, a choral group (she made sure to tell me that it’s an auditioned ensemble), and an executive women’s network, which was having its Christmas party the day I visited her. That’s why “I wore a little sparkly something,” she told me.
Before settling back in Iowa permanently, Selzer did an academic fellowship for a year in England, a place sometimes incompatible with her casual Midwestern demeanor — she once, “charmingly!” she insists, called her residence hall warden an “old goat” — and a year-long congressional fellowship in Washington working for Barbara Mikulski of Maryland, then a representative now a senator. Selzer was taken with her experience — that “a kid from Iowa can go to Washington and write legislation,” that the cozy office was filled with clacking typewriters, that the other young people were ambitious too. “The thing that was so alive about working on the Hill is that we knew stuff before the world knew it,” she said. “I liked that. That gives me a kick.”
After a couple of jobs in polling and a turn as a jury consultant — “it was so ugly” — Selzer landed a position at the Register running its in-house polling during the halcyon days of print. It was December 1987, and the caucuses were just around the corner.
Polling is a relatively new thing in the 240-year existence of the republic. George Gallup only busted open the field of public opinion surveying in the 1930s (to help his mother-in-law win a race for Iowa secretary of state, in fact). And the attention paid to the Iowa caucuses is even more recent, although to listen to the hype, you would think that the Founding Fathers had glad-handed with every mugwump from Sioux City to Belle Plaine. While the state has been holding caucuses since 1846, it’s only had its first-in-the-nation status since 1972. And not until 1980 did the caucuses truly claim their reputation as the Olympics of pandering, the place where an underdog candidate could become a made man.
And it’s not just politicians whose careers can be made off the caucuses.
Selzer was just weeks into her job at the Register when she spotted a problem with the paper’s poll, which showed George H.W. Bush winning the upcoming Republican caucuses. She smelled a rat. The paper’s in-house polling operation had “figured out some creative ways so they could turn polls around faster,” she said. They were re-contacting previously polled people, a no-no in her book. When she asked for a crosstab — a deep dive into the numbers — Selzer found that when isolated out, newly contacted voters showed Bob Dole winning instead: “So I went to the editor and I said, ‘I believe the Register is publishing that George Bush will win the caucus, and I don’t think that’s true.’”
The polling was redone to Selzer’s standards, and Dole did indeed win. Bush came in third. Selzer’s advice for early career advancement (she was 30 when she went to the Register) is to “find something that is a problem and fix it and make the boss look good.” Of course, Selzer has been wrong, too. She pulls a face when the 2004 general election comes up. Selzer wrongly had John Kerry beating George W. Bush in Iowa. “I was at a watch-the-returns party, and I just had to slink out of there when I realized what counties were not yet counted,” she said. “I thought, ‘Ohhh, I’m losing this one. I mean, I don’t care who wins and loses so long as it’s the person who I have winning my poll.” She ended up writing a self-flagellating piece for the Register a few days after the election headlined “Iowa Poll was a miss, and I don’t like it.”
Although Selzer’s firm does polls for private companies and public opinion surveys outside of Iowa, her home state remains her bread and butter. The caucuses are a beast unto themselves. They “have nothing directly to do with presidential politics” but rather “are private business meetings of the political parties,” said Dennis Goldford, a political scientist at Drake University. A Jacksonian-era innovation, the caucuses were meetings in which, theoretically, any party member could participate.
In practice, however, party bosses tried to keep them exclusive. In 1948, Emory English, who headed a historical magazine in the state, wrote about a town in northern Iowa where an old shed happened to catch on fire at exactly the scheduled time of a caucus, attracting “nine-tenths of the people in the village, including members of the volunteer fire department.” While the townspeople gawked, party insiders convened, selected a slate of delegates without opposition and then went on their merry ways.
As the years went by, the caucus process became less arson-prone but more entwined with polls. The combination of a 1975 survey and a New York Times article might explain why we all care so much about Iowa nowadays.
In October of that year, the Register held a straw poll at the Jefferson-Jackson dinner, where a number of the Democratic candidates were to speak. When the results came back, according to a report by the storied Times political reporter R.W. Apple Jr., the party apparatchiks were stunned. “Mr. Carter, whose Presidential aspirations have been considered laughable by many Washington experts, won 23 percent of the total,” the article said. “His nearest rival was a write-in candidate, Senator Hubert H. Humphrey of Minnesota, with 12 percent.”
National papers took note of the Times item, which appeared at the bottom right corner of the front page,1 and soon set their own reporters out on the Iowa trail to document the Carter phenomenon. The goober-farmer-turned-gubernatorial-officeholder won the state’s contest and went on to the White House. By the time the 1980 caucuses rolled around, candidates were looking to replicate Carter’s success — and America’s greatest media/polling echo chamber was born.
Selzer’s deep hatred for the way high level math was taught in an undergraduate class helps form the philosophical basis of her polling practice. “They would say, ‘Well, make a guess as to what the answer’s going to be,’” she said. “I’m not good at guessing about math. I just think if you’re thinking about a future event, how do you know? And do you want a method that would blind you from it changing?”
This unwillingness to guess lies at the heart of her thinking about polling; Selzer doesn’t want to prejudice herself. “I like to say, “Keep your dirty hands off your data,’” she said. “That’s the making assumptions of what is or isn’t going to happen and then deciding you’re going to weight down the minority vote because you don’t think they’re going to show up.”
Selzer says that how a sample is drawn is “the heart of the science” of polling. That makes pollsters’ models — the mathematical calculations that help them determine this group — the arithmetic aortas of the discipline. George Gallup compared public opinion polling to sampling soup: “As long as it was a well-stirred pot, you only need a single sip to determine the taste.”
The argument that polling is not only a science but also an art is common, and echoes what doctors often say about their profession’s virtuoso elements. “It’s not necessarily guesswork, but there is a large dose of judgment that goes into ultimately picking a model and also risk involved when there is difference across those viable alternatives,” said Jon Cohen, who in 2008 headed the ABC News/Washington Post poll, one of the few surveys other than Selzer’s to show Obama up on Clinton in Iowa.
Most pollsters’ models involve set cut-offs of how many of this or that group is likely to show up and vote, relying on criteria like past attendance and enthusiasm to try to predict whether a person will caucus. If the neatly delineated poll numbers that we see in news articles are akin to the disembodied head of The Great and Powerful Oz, then the mathematical model that produced them is the little man pulling levers behind a curtain.
Cohen, who is now at SurveyMonkey, said more than one model is typically at play, so pollsters test different variables in order to gain confidence in their final model — pull the red lever and previous caucus attendance becomes more important; pull the blue one and voter enthusiasm takes precedence.
To Selzer, though, there is great hazard in hazarding a guess, and her method for determining who might actually caucus reflects this: Her polls live in the moment, taking into account no history. Polls for 2016, for example, won’t be weighted to reflect the voter turnout of 2012 or 2008 based on a hunch that the elections might see similar demographics come to caucus. She wipes the slate clean every time.
For all the questions about what makes her so good, there is perhaps something to be said for the relative simplicity of Selzer’s methodology. “When Politico says, oh, you know, she’s got a secret sauce, I go, well the saucy part, I’m good with that, but secret, not really,” she said. If you know what to look for, you can find Selzer’s numerical recipe at the top right hand corner of her polling reports. For what it’s worth, David Yepsen, a former political columnist for the Register, thinks the saucy part comes from Selzer’s being “closer to the news on a daily basis than someone in Washington, D.C., or New York” and knowing what questions to ask of Iowa voters. Selzer took to the state with the zeal of a convert: “She’s very bordering on boosterism at times,” said her older brother, Tad. “It’s fun to go down there and visit her — she has a vast network of acquaintances in Des Moines, she has a cocktail lounge where she’s immediately recognized, she’ll have new places where we’ll eat. She’s very much an Iowan.”
While it might not be a trade secret, Selzer does have a distinctive method. Her work for the Iowa Poll begins with calls to a list of registered voters that she gets from the Iowa secretary of state’s office. Mark Blumenthal, head of election polling for SurveyMonkey, points out that there has been a split among media pollsters — as opposed to internal campaign pollsters — about whether to seek out respondents from a list such as this or through a process known as random digit dialing — randomly generating phone numbers to call in the hopes of casting nets far and wide to find voters. Media pollsters “up until the last few years virtually all favored RDD,” he wrote in an email. “Mostly because of worries about non-coverage of listed voters for whom telephone numbers are not available.”
In other words, pollsters wanted to ensure that they wouldn’t be missing registered voters — especially in general elections — who decided not to make their phone numbers public information, or the newly registered, often young people. Polls done from lists are less expensive, though, and some pollsters, like Amy Simon, writing in 2006, have suggested that they are better at homing in on actual voters. “One hypothesis offered is that samples drawn from voter registration lists by definition consist of actual voters, while RDD studies rely entirely on respondents’ self-reporting about whether they are in fact registered to vote,” she wrote.
The next step in the Selzer process is that respondents are asked a series of questions to determine whether they’re likely to attend a caucus — only about 20 percent of Iowa’s registered voters do. They’re also asked who they’ll caucus for. Selzer collects data on all the people she contacts, though, not just the ones who say they’re going to participate. When the calls are done, then the data for all those called is weighted based on the “known population parameters” of the voter list — age, sex and congressional district. The list is not weighted by factors such as past caucus and general election voting activity.
“We are not like many other polling firms that say, ‘Well, we think’ — they begin with a guess about what they think the future is going to look like,” she said. “I don’t think that’s science.”
The “we” of Selzer’s firm — she has been in private practice since 1992 — is the pollster herself and Michelle Yeoman, Selzer’s research assistant of over 10 years. There are no cubicle bays filled with number-crunchers at Selzer & Co., just the two women in a single-story “Little House on the Prairie” structure that was built to Selzer’s exacting standards a couple of years ago. It’s an airy space that still smells new, with unmarred blond wood floors and a thriving crop of houseplants. The focal point of the space is a hallway that cuts horizontally through the building. Selzer and Yeoman have offices at either end of it and will call across the 20 or so feet that separates them to answer a question or finish one of the other’s thoughts.
“We built this on purpose so that there’s constant communication,” Selzer said. Their office doors, though rarely closed, were a pet project — she’s a fan of the show “Damages” (“there’s a lot of killing!”) and asked her builder to replicate the sliding barn doors featured in the Glenn Close character’s office.
“Part of our culture in this office is we have the phrase, ‘I just want to say this out loud,’” Selzer said. “It’s a culture of overcommunication, and some day I will write a book called ‘Just Say It Out Loud.’” It’s a nicety that seems in keeping with Selzer’s attention to detail in modes of communications. Her phone bankers receive manners training — “Iowa nice,” she says, grinning. And the power of politeness so inspired her, she says, that she stopped swearing years ago.
As Iowa nice and easygoing as Selzer might be, she is also hard-driving. Once, she sued her landlord — Selzer brags about having kept him on the stand for six hours (she is not a lawyer) — and won back her $2,100 deposit; another time, she translated “Amazing Grace” into Vietnamese for a choir performance (she does not speak Vietnamese).
But most of all, Selzer is a fundamentally earnest person operating in one of the more cynical spheres possible. She sent me an email one day with the subject line, “My check list of things I’d like to do.” She’d ticked off a few items already: winning a blue ribbon at the Iowa State Fair (for cheeseless pizza) and a Scrabble tournament, singing with the Mormon Tabernacle Choir. The items that remained were by turn ambitious, fame-curious and whimsical: write a funny book, meet Oprah Winfrey and David Brooks, have tea with the Queen of England.
Her sincerity makes the fact that she’s become a Washington in-crowd “it girl” all the more paradoxical, since her celebrity is based on something fundamentally disingenuous; Washington perversely inserts Selzer into the news cycle by calling into doubt her polls and then just as quickly embraces her again after she’s proved the political elite wrong.
That’s what happened during the 2014 midterm elections. It was a tough cycle for polls generally, but not for Selzer. “Once again it is Ann Selzer’s polling world in Iowa, we’re just lucky to live in it,” Chuck Todd, host of “Meet the Press,” tweeted after Republican Joni Ernst’s surprise defeat of Democrat Bruce Braley by more than 8 percentage points in the state’s Senate race. Three days earlier, Selzer had anticipated that outcome in a Register poll, while others were showing the race neck and neck.
But while Selzer has so far thrived in an industry in crisis, it’s safe to say that she’s not immune from the problems accompanying its rapidly shifting landscape — polling calls to cell phones are a more complicated business, and online polls are gaining in popularity but remain untested. She declined to say where she sees her practice in five years, but it’s not difficult to guess that it will look radically different given these changes.
What will not change is that for Selzer, politics is a solo sport. It’s ironic, given that polls are devoted to tracking the electoral movements of history’s neediest megalomaniacs, that the pollster’s gratification comes from sitting alone in a room and finding that her numbers have correctly predicted waves of thought and movement, that the opinions shared over the phone that are then tabulated and aggregated have come together to predict perhaps even a moment in history.
The night of the 2008 Iowa caucuses, Selzer wanted to see whether the massive turnout she had predicted would bear out on the ground. So she drove to the nearest caucus site.
“And it’s lit up like a Norman Rockwell painting, there’s snow on the ground,” she said. “It’s a middle school, so there’s not a big parking lot, and it’s on a part of Grand Avenue where there really aren’t side streets. So in order for people to get there — and both Republicans and Democrats are there, there’s 1,000 people coming into this school — they are parking six, seven, eight, blocks away and walking on icy sidewalks to get there. And there they are, their arms linked, kind of two by two, three by three.” It was then, without the numbers in front of her, that Selzer knew her poll would be OK.
“It just honestly — it kind of made my heart melt a little bit because here’s democracy, here they come.”