OKLAHOMA CITY — Heading into Election Day, polls indicated the Oklahoma gubernatorial race was a virtual dead heat.
And the Cook Political Report even declared the race a toss-up, saying that Democratic candidate Drew Edmondson possibly could upset the prohibitive favorite, Republican Kevin Stitt.
But when voters had their say, Stitt defied the pollsters' expectations, beating Edmondson by a 12-point margin.
Pollster Bill Shapard said as it turned out, the state's rural areas broke late for Stitt, and Republican voters in those areas turned out in large numbers to support him.
"We did not see any catalyst for them turning out, but they did," said Shapard, founder of SoonerPoll.com, an Oklahoma independent, nonpartisan polling firm. His firm's final poll released ahead of the election showed Stitt ahead of Edmondson by just 3 percentage points with a margin of error of five.
While it's nice to be right and have the results mirror the eventual outcome, Shapard said a pollster's job ultimately is to be more explanative than predictive.
"I don't convince myself that I have some sort of crystal ball," he said.
Shapard said polls merely measure public sentiment at a single point in time and are more useful to identify polling trends, how certain constituencies feel about issues and how those impact decision-making.
Republicans, for instance, are struggling to gain the support of predominantly white, suburban women, while Democrats are failing to connect with the workingman, particularly in rural areas, he said.
"(Polling) gives you a good sense of what preferences different types of people have," said Keith Gaddie, a University of Oklahoma professor of political science and journalism. "It gives you a decent sense of what people are thinking about when they are voting."
But pollsters consistently work under uncertainty because they can't know turnout numbers, he said.
Also, members of the media often ask polling to do what it's not supposed to do -- predict an outcome. It predicts nothing, he said.
"People misuse our product and then blame us for it," Gaddie said. "It's like making guns."
David Flaherty, CEO and founder of Magellan Strategies, a Colorado Republican-leaning polling firm, said Stitt won with undecided voters.
Magellan, which polled Oklahoma's likely gubernatorial voters about two weeks out, showed Stitt winning by about 7 points with a margin of error of 4.
"We have to make judgment calls on what percentage of the turnout is going to be Democrat, Republican and independent," Flaherty said. "You need to get the demographics right of turnout."
A lot of pollsters won't release their results publicly, he said.
"The damage to your reputation can be devastating if you really get it wrong," Flaherty said.
At the same time, though, it's getting harder to conduct polling, Flaherty said.
"It is getting more difficult to forecast it and measure voter opinion accurately," he said.
For his poll, Flaherty said he started with 20,000 random, unique phone numbers, just to get 500 participants. At some point, pollsters have to consider whether non-responsive bias is slipping into the poll results when so many likely voters can't be reached or refuse to participate, he said.
"[Campaign polling] is definitely running its course," he said.
The cost is also increasing. On average, it cost about $12 per landline call and $20 per cell phone, Flaherty said.
Fewer and fewer Oklahomans under age 65 own landlines, he said.
"The landlines are going to eventually be gone," he said. "[Eventually], the cost is going to be exorbitant."
Stecklein covers the Oklahoma Statehouse for CNHI's newspapers and websites. Reach her at email@example.com.