On a sunny afternoon in mid-April—just weeks before Pennsylvania’s presidential primary—the polling center at F&M’s Center for Opinion Research hums with 20, one-sided phone conversations. Interviewers are seeking data for the latest Franklin & Marshall College Poll, producing what will most likely be the final snapshot of voter attitudes before the primary election on April 26.
Retirees, members of the Lancaster community, young professionals and F&M students sit in cubicles facing their computer monitors, listening to responses from sample respondents. To prevent any off-the-books conversation that might bias an answer, each interviewer reads poll questions verbatim into their telephone headset, logging results in the Center’s polling software.
“If the Republican primary election for president were being held today, would you vote for Donald Trump, Ted Cruz, John Kasich, or aren’t you sure how you would vote?’” Susan Kugel, a retired grade school teacher from Chicago, asks a young man from Montgomery County.
He wasn’t sure—and he’s not alone. Even this close to the primary, 15 percent of Republican respondents polled by F&M remain undecided about which name they would tick on the ballot.
According to Berwood Yost, director of the Center for Opinion Research, indecision is a common thread in primary polling.
“We know historically that primary polling is much less accurate than pre-election polling during a general election,” he says. “Typically, turnout in a primary election tends to be lower. It’s harder to figure out who’s going to vote. People aren’t deciding until really, really late in the election cycle, so you see a lot of undecided voters, right until the end.”
With so many variables in play, keeping a finger on the pulse of public opinion is no easy task. Together with F&M Poll Director G. Terry Madonna, Yost has made a 20-year career out of asking Pennsylvania voters their opinions on topics ranging from health to consumer product safety.
When the two pollsters started working together at Millersville University in the early 1990s, they borrowed the phone banks from the alumni development offices to make calls in the evenings.
“It was all low-tech,” recalls Madonna, who trained students in government classes to conduct the polls. They recorded voters’ answers with pencil and paper.
The duo brought the poll to F&M in 2004 and steadily built the Center for Opinion Research into a regional powerhouse. In addition to operating the longest-running statewide poll in Pennsylvania, the Center serves numerous non-profit and government clients, including the School District of Lancaster and the Federal Consumer Product Safety Commission.
Today, the era of polling with pencil and paper is long gone. At the Center, a complex computer program loads questions and records answers—the program even shuffles the order of questions for each call.
Asked to describe the changes in his field over the last 20 years, Yost laughs. “Everything. Can that be an answer?”
As more Americans abandoned landlines for cell phones in the early 2000s, Yost explains, the landscape of public opinion research shifted dramatically.
Since cell phone and online use have changed the game, pollsters work hard to stay ahead of the curve and adopt new methods for data collection. In this respect, the F&M Poll is at the forefront of its field.
After a year of careful testing, the Center rolled out a new methodology for collecting polling data this winter. First, voters in the random sample for the F&M Poll received a letter describing three options for participation: respondents could complete the poll online, call the center to speak to an interviewer, or wait for an interviewer to call them.
Yost, along with his Center colleagues Jacqueline Redman and Scottie Thompson, worked on designing this new approach. They hoped these new collection methods would give people more ways than ever to participate in the poll—and give the Center a more accurate picture of the presidential and Senate races in Pennsylvania.
“We think we give more people the opportunity to respond. Our samples are more representative now than they were because of people’s willingness to answer in different ways,” Yost says.
Four years ago, Nate Silver—author of the bestseller “The Signal and the Noise” and something of a polling gadfly—predicted precisely this sort of methodological shift in The New York Times.
“[There] will always be an important place for high-quality telephone polls, such as those conducted by…major news organizations, which make an effort to reach as representative a sample of voters as possible and which place calls to cellphones,” Silver wrote in 2012. Online polls have their place, too, he added—especially for reaching younger demographics.
The mixed-mode approach newly adopted by the Center for Opinion Research is precisely “where the industry is headed,” Yost confirms.
The Center’s historical commitment to examining—and re-examining—polling methodology has played no small role in helping the F&M Poll rocket to the top of the pack. During the 2012 presidential election, Silver even recognized the poll as one of the most accurate in the country.
A Snapshot in Time
When you’re talking about Pennsylvania politics—especially during this loud, fractious presidential primary—accuracy is a tricky subject, no matter which polling method is on the table. Even though mixed-mode polling might help F&M develop a better overall picture of how voters in Pennsylvania feel about presidential candidates, public opinion is notoriously shifty.
In March, for example, the Center published its findings on the presidential primary race—and the results for Republican voters floored everyone. Coming off of a big win in his home state of Ohio, Gov. John Kasich surged past Ted Cruz in the polls and trailed frontrunner Donald Trump by only three percentage points.
According to Stephen Medvic, The Honorable & Mrs. John C. Kunkel Professor of Government, likely voter models—the way pollsters determine which of their respondents are most likely to vote—have everything to do with this surprising result. “We’re not really predicting election outcomes,” Medvic cautions. An expert on political campaigns and voter behavior, he regularly works with the polling center on his own research projects.
“The election’s in April—if you do a poll in March, you’re not predicting it. It’s a snapshot in time, but you want it to be an accurate snapshot,” Medvic says of the March poll. “Depending on how you model what the electorate will look like, you might get different results.”
The March poll happened to use two different likely voter models: one with high voter turnout, which favored Trump, and one with lower voter turnout, which favored Kasich.
“If a lot of independents change their registration and vote Republican, or if a lot of Republicans who don’t typically turn out in primaries turn out in this primary for Trump, he’s going to do better,” Medvic says. “In closed primaries, where it’s mostly regular Republicans, [Trump] doesn’t do as well.”
Of course, if Trump picks up momentum during his campaign in New York, public opinion could shift yet again before Pennsylvania voters head to the polls. “People always wonder, how can 600 people really represent the entire state of Pennsylvania?” says Medvic. “But if the poll is done correctly, if the sample is drawn properly, polls really are an accurate reflection of what people are thinking at that moment.”
This window into the Pennsylvania voter is why Sarah Niebler, an assistant professor of political science at nearby Dickinson College, follows F&M Poll results. She believes the poll is especially valuable for studying public opinions about state-level government, which has a direct impact on how people lead their daily lives.
“If you live in a community and your staff and interviewers live in the state, they have a much better sense of how to identify whether the people that they’ve talked to are truly representative of the state than outside polling firms do,” she says. “What’s going on in state legislatures, what’s going on in gubernatorial offices across the country, particularly with budgets—this matters. This matters for funding of local schools, it matters for funding of higher education.”
The Media Storm
During election season, it can be hard to escape the buzz that comes along with political polling. Once a poll is released to the press, Madonna can field as many as 20 calls a day from reporters. He also makes appearances on CNN, MSNBC, Fox News, and local networks to discuss the results.
“My job is to interpret the poll for the media and the public,” says Madonna. “Every poll has elements in it that are more timely, more salient, more interesting to the public. This year, it would be the presidential election in our state, but we regularly do things that are more research-oriented.”
Yost is also quick to remind people that the Center for Opinion Research remains busy long after the hoopla of election season is over. The staff conducts significant research about attitudes toward health and wellness in Lancaster County and in F&M student populations, too.
Perhaps most exciting for Medvic, the Center for Opinion Research helps social scientists advance our understanding of why people develop certain beliefs. “It’s not just a poll to make a splash to get a headline,” the professor says. “It’s much more serious than that. Good survey research can really help us learn a lot about what our fellow citizens believe.”
And believe they do. All afternoon and into the evening, interviewers at the polling center will keep dialing and recording the answers that help Yost, Madonna, and their team paint an ever-shifting picture of democracy in action.
“When it’s time to make the last call, I flip the lights or play music from my iPhone,” says Karen Glover, leader of the late shift at the calling center. “We try to have fun,” she adds, before picking up a headset and getting back to the serious business of polling.
Read more about primary polling and the use of the F&M Poll in academic research:
- There Are Limits to Polling in Primaries
Philly.com (May 19, 2016)
Berwood Yost explains why primary polling is fluid, has always been fluid, and will always be fluid.
- The Effects of Issue Salience, Elite Influence, and Policy Content on Public Opinion
Political Communication (Vol. 2, No. 16)
F&M’s David Cuik and Berwood Yost explore the conditions that motivate people to use content over cues—and vice versa—in forming political attitudes.
The Pundit in Election Season
If you follow regional or national politics closely, chances are you’ve heard or read commentary from Terry Madonna, director of the Center for Politics and Public Affairs and director of the Franklin & Marshall College Poll. In March and April alone, he represented the College on CBS News, CNN, C-SPAN, NPR, Fox Business Network, Australian Public Radio and the Canadian Broadcasting Corporation—in addition to his dozens of regular appearances in Pennsylvania media.
With a U.S. Senate seat at stake in the commonwealth and a presidential election looming, Madonna has been especially busy. The day before an F&M Poll release, he typically speaks with representatives from each of the poll’s nine media partners. And in the hours after the poll hits the news, he provides analysis for up to 30 additional media outlets.
His most memorable day came in April 2012, when Rick Santorum criticized the F&M Poll on national television after it showed him—accurately—losing ground in the presidential primary. “I received 39 phone calls in five hours asking me to defend the poll,” Madonna says. “He was beating me up.” Santorum pulled out of the race days later. “That was the craziest week,” Madonna says.