GOP, Dems both blame the messenger, allege bias in polling data

Cries of bias are getting louder as first one political party then the other reacts with dismay to polling data moving against it in the excruciatingly tense final weeks of the presidential election.

As Election Day, Nov. 6, approaches, movements even within the margin of error have President Obama’s and challenger Mitt Romney’s campaign teams wringing their hands or exchanging high-fives.

This week it’s Team Obama’s turn to suggest polling methodology is skewing opinion research against its man.

The allegations against reputable pollsters such as Gallup and Pew Research come just weeks after conservatives complained skewed polls were underestimating support for Romney.

ADVERTISEMENT
The charges highlight the obsession partisans on both sides have with the polls, as well as the effort both sides are making to find every advantage they can in a campaign that increasingly appears headed for a photo finish.

“It’s common,” Gallup editor in chief Frank Newport told The Hill. “The campaigns have a war room-type mentality, and both campaigns feel the need to quickly jump on any news of any type that could be viewed as negative for their candidate.”

Neither side wants a disappointing poll to turn off supporters and depress turnout in a key swing state that could be the difference between winning and losing.

“Absolutely, there’s the potential for this snowball effect,” said Dr. Costas Panagopoulos, director of the Center for Electoral Politics and Democracy. “But you can’t have it both ways. You can’t argue the polls are great when they favor your candidate and are flawed when they move against your candidate.”

The latest complaints have come from the Obama campaign, which on Monday issued a statement warning reporters to take with a grain of salt a Gallup poll that found Romney with a 4 percentage-point lead in 12 battleground states.

The poll was “an extreme outlier” that exposed “deep flaws” in how Gallup locates likely voters, according to Democratic pollster Joel Benenson. Gallup only started releasing presidential polls of likely voters this month.

Benenson argued Gallup’s likely-voter screening method “created a bias against groups inclined to support Obama,” and was the reason the candidates were tied among women, traditionally an Obama stronghold.

Newport shot back in a statement saying there was “no evidence” Gallup’s likely-voter model was disproportionately Republican. Panagopoulos said the claim was rubbish, and that the Gallup poll merely reflects the general enthusiasm gap the president has suffered in conjunction with his slide in the polls.

“These likely-voter screens are not flawed. This reflects a weakening enthusiasm among Obama supporters, and that’s how it works,” he said. “When Obama supporters are excited, they’re more likely to say they’re likely to vote. When Romney supporters are more enthusiastic, they’re more likely to say that they’ll vote. So these are compositional changes generating those differences in the polls right now — they’re not necessarily a change in candidate preference.”

Some liberal commentators launched similar complaints against a Pew poll earlier this month that recorded the first shift toward Romney after the first presidential debate.

The Pew poll, which showed Romney with a 4-point lead over Obama with likely voters, included too high a percentage of Republicans in its sample, some Obama supporters said.

Just a few weeks ago, conservatives were up in arms over polls that showed Obama ahead of Romney.

Republicans said the surveys relied on voter sample sizes that gave too much weight to high Democratic turnout. Even Romney’s campaign got in the act, arguing pollsters relying on the 2008 turnout to determine the makeup of the 2012 electorate were making a fatal mistake.

Nate Silver, the author of the 538 blog for The New York Times, wrote on Oct. 9 that he thought little of this argument no matter which side was making it, since party identification can be fluid.

While a number of polls in the last week have shown Romney with a lead, that hasn’t stopped his campaign from complaining about some polls.

On Tuesday, the Romney campaign circulated a memo from Republican pollster Rich Beeson citing “a myriad of public polls in Ohio that show the race within the margin of error.”

“Some are more credible than others, and some sample the race correctly, while others do not,” Beeson cautioned.

Panagopoulos said claims of polling bias were not unique to this election, but that they seem to have escalated in this cycle because of the sheer proliferation of surveys and the increased attention paid to the horserace aspect of the election.

Newport agreed, saying claims of polling bias go “back as far as George Gallup in 1936.” But, he added, it’s not just the campaigns taking issue with the data anymore.

“Now you have so many different sources — some skilled, and some having no idea [what] they’re talking [about],” he said. “They’re there, ready to make quick comments on blogs and opining on different forums.”

Most polling analysts recommend looking at a polling aggregate, or an average of polls, to get an idea of where the race truly stands and to eliminate statistical noise.

According to the RealClearPolitics average of polls, Obama and Romney are tied at 47 percent nationally, and the race appears just as tight in the bulk of battleground states. 

Panagopoulos ran a study of final major polls from 2008, and found most were within 1 or 2 percentage points of the final margin of Obama’s victory over Sen. John McCain (R-Ariz.) — an impressive accomplishment, considering the unique nature of the election and the conventional-wisdom-shattering demographics that turned out to vote.