Experts try to make sense of gap between polls and vote in Georgia


Governor’s Race

Election Results: Nathan Deal-R 53 percent; Jason Carter-D 45 percent

Landmark Communications Poll: Deal 51 percent; Carter 45 percent

SRBI Inc-AJC: Deal 43 percent: Carter 42 percent

Survey USA: Deal 47 percent; Carter 42 percent

U.S. Senate Race:

Election Results: David Perdue-R 53 percent; Michelle Nunn-D 45 percent

Landmark Communications Poll: Perdue 50 percent; Nunn 46 percent

SRBI Inc-AJC: Perdue 45 percent; Nunn 41 percent

Survey USA: Perdue 47 percent; Nunn 44 percent

There was one thing most polls seemed to agree on in the weeks leading up to Tuesday’s election — the two top races in Georgia were so excruciatingly close that both might have to be decided by runoffs.

But then the voters stepped in. Republicans won in a rout. It wasn’t a horse race — it was a political blowout.

What happened?

Experts chalk up the stunning outcome to an 11th-hour Republican surge, hinted at only in a few late polls. Those statistical snapshots showed the electorate increasingly flocking to the GOP as the election neared. Democratic momentum was fading a bit as GOP steam increased in the final days.

Some of those predicting runoffs didn’t take into account caveats, like margins of error and undecided voters, that swung the numbers.

Meanwhile, some earlier surveys were simply imprecise. They relied on automated calling and Internet surveys, cheaper methods scorned by more established pollsters.

“We have major polling problems (in Georgia),” said Kerwin Swint, chairman of the political science department at Kennesaw State University.

“No one here knows how to model turnout based on voting patterns, population, and issues.”

Across the nation, some polls underestimated Republican groundswell on Tuesday. But in Georgia the results were among the most heavily skewed. Survey after survey suggested that Republicans Gov. Nathan Deal and U.S. Senator-elect David Perdue might not surpass the 50 percent benchmarks needed to avoid long, costly and unpredictable runoffs.

The state attracted national attention as pundits speculated that the outcome in Georgia could decide whether Republicans won control of the U.S. Senate.

Instead, Deal and Perdue won election Tuesday with 53 percent of the vote, a full 8 percentage points ahead of their Democratic opponents.

A morning-after analysis by polling guru Nate Silver listed Georgia as one of the states with polling that was the most heavily biased toward Democrats — off by 6.4 percentage points in the race for the U.S. Senate and 4.6 points in the governor’s contest.

Buoyed by demographic changes and waves of newly registered voters, out-of-power Democrats were hopeful for a resurgence. Republicans have dominated state politics ever since underdog Sonny Perdue in 2002 was elected the first Republican governor since Reconstruction.

Topping the Democratic ticket were two young, articulate candidates from well-known Georgia political lineages. But former state Sen. Jason Carter, candidate for governor, did only slightly better than Democrat Roy Barnes four years ago in a failed comeback bid for governor against Deal. Carter pulled just 45 percent of the vote. The story was much the same for U,S. Senate candidate Michelle Nunn, daughter of former Democratic U.S. Sen. Sam Nunn.

A key problem for Democrats was their lack of support from white voters. Exit polls said that Deal and David Perdue won 73 and 74 percent of white voters respectively.

Landmark Communications, based in Alpharetta, surveyed Georgia voters in the final days before the election and placed both Deal and Perdue with  leads.

“We identified the Republican surge that took place in the closing days,” Landmark president Mark Rountree said.

“And in the end Georgia had the same surge for Republicans that the rest of the country saw, so the GOP candidates scored a few more percentage points than our, or anyone’s, poll reflected.”

The Atlanta Journal-Constitution commissioned New York-based Abt SRBI Inc. That survey, which used a mix of live calls to land lines and cellphones, took place Oct. 16-23. It showed the governor’s race in a dead heat and Perdue holding a slim lead in the Senate race. In it, the Libertarian candidates had 6 and 5 percent of the vote respectively. Ultimately that support was pegged at just 2 percent Tuesday night.

SRBI founder and chief research officer Mark Schulman said there were signs of a Republican wave in Georgia and elsewhere but the size of it “has befuddled the pollsters.”

“It’s something we need to cogitate on,” Schulman said.

“Clearly the electorate was polarized,” said Schulman, who said the climate of the midterms favored the GOP. “But in the past, the signs of a wave were more clear.”

But not everyone was surprised. Operatives with the Perdue and Deal camps said the final results reflected what their internal polls had been showing for some time.

“There are a lot of pretend pollsters in Georgia, that’s all I’m gonna say,” said Paul Bennecke, a consultant for the Perdue campaign. “They don’t know what they’re doing. They get a phone list. They do a bunch of random phone calls. They try to weight it based on what they think is right, then they throw it up on paper and say, ‘Here’s what it is.’ That’s not how our polling operation worked.”

Deal spokesman Brian Robinson also said that many polls in the race were flawed. Robinson said that those interpreting the polls failed to take into account that so called “undecided voters” would eventually make up their minds.

“Those voters largely broke for us,” Robinson said.

But as the head scratching continues, some are suggesting that polling problems point to a need for a wholesale overhaul.

“Clearly, the pollsters have not been able to solve the problems bedeviling their field,” Larry Sabato, director of the Center for Politics at the University of Virginia.

“It may call for a whole new model built around the internet.”

Experts say the technique used by pollsters is significant. Live calling to homes and cell phones is considered the gold standard. Most of the public polls are done through automated calls to homes that under federal law cannot be made to cell phones. About 30 percent or more of registered voters only have cell phones so they are excluded.

Sabato said Georgia’s demographics are gradually changing, and the state is following the Virginia/North Carolina model.

“Within the not-to-distant future, Georgia will be a purple competitive state. But it obviously isn’t there yet.”