Are polls "biased" to boost Obama?
(CBS News) On the heels of several new polls showing President Obama opening up big leads over Mitt Romney in key battleground states, some Republicans are accusing pollsters of a Democratic bias -- and suggesting that the recent poll numbers don't tell the real story.
Tuesday, a new Washington Post poll showed Mr. Obama leading Romney by 8 points in Ohio - a spread that indicates a distinct advantage for the president in a state many believe is critical to a Romney victory. A new battleground poll from CBS News/New York Times/Quinnipiac out Wednesday shows Mr. Obama with a 10-point advantage in the state, as well as a 12-point lead over Romney in Pennsylvania and a 9-point lead over his rival in Florida. If accurate, these numbers signal trouble for Romney in finding a path to 270 electoral votes.
In recent days, however, Romney officials and other conservatives have cast doubt on the polls, saying the polling models favor Democratic candidates. The Romney camp has also suggested that internal surveys show a different picture.
Romney political director Rich Beeson told reporters yesterday that the campaign encountered what he cast as a polling flaw in the nominating contest as well, and that it "has confidence" in its own methodology.
"You saw in the primaries... you know, we used a specific set of data for our primaries. Each week we would go in, and you know, we'd be 10 or 11 down...whether it was Ohio or Michigan or Wisconsin," Beeson said. "We relied on our internal data. We knew where each day at any given point. That's the same thing we're doing now. The public polls are what the public polls are. I kinda hope the Obama campaign is basing their campaign on what the public polls say. We don't. We have confidence in our data and our metrics."
On Wednesday, Ed Gillespie, a senior adviser to the Romney campaign, amplified that message.
"I'm struck by a couple of things," he said on Fox News. "Three swing state polls [are] out today and in every single one of them, they have a Democratic voter participation that is higher than the participation in the electorate in 2008. I don't know anyone on the ground in any of these swing states who believes that there will be a higher Democratic percentage of the electorate in 2012 than there was in 2008."
Gillespie argued that pollsters' overstated predictions of Democratic voter turnout would explain why "Governor Romney could be tied or leading with independents in those polls and yet losing the net poll to President Obama."
Critics have also argued that in addition to oversampling Democrats, pollsters are undersampling people who might be sympathetic to Republicans.
"Everyone right now believes that the polls are biased to President Obama because of this kind of a sample," said conservative talk show host Hugh Hewitt in an interview earlier this month with Lee Miringoff, director of the Marist Institute for Public Opinion. "I'll tell you what's on people's minds. I'd really love to hear you answer it, which is that was the height of the President Obama 'Hope and Change' movement. He was extraordinarily strong in November of 2008... Against that backdrop, what level would you become concerned as a professional pollster that you had oversampled Democrats?"
Many nonpartisan pollsters, including Miringoff, say they weigh their surveys using available statistical data - and not by party identification and other characteristics that might change over time.
"We do not weight for party ID," said Peter Brown, assistant director of the Quinnipiac University Polling Institute, which is conducting battleground state polls along with CBS News and the New York Times. "We do not predetermine how many Democrats, Republicans and independents will be in our sample."
The reason, he says, is that "party ID is a changing statistic. People will over time change back and forth in terms of how they view themselves politically."
Instead, Brown says the polls are weighted by "immutable characteristics - race, gender, age." Respondents are then asked their party identification during the interviews.
"We do that because there is a set standard that we can compare ourselves with to make sure we're getting an accurate demographic representative. And that standard is the United States Census Bureau data," said Brown. "What we get is what we get," he added.
Moreover, some pollsters argue that using the 2008 turnout numbers to draw conclusions, as Gillespie and others have done, ignores the fact that state demographics are likely to change over a period of four years.
In an interview with CBSNews.com, Miringoff pointed to the fact that "there are a lot more Latino voters this year than there were four years ago. Exit polls from four years ago aren't going to pick up that change." Similarly, he argues, 2004 exit polls wouldn't have suggested that Mr. Obama could compete in traditionally Republican-friendly states like North Carolina and Virginia - but he won both in 2008.
"The numbers that people are using as the yardstick for whether pollsters like us are wrong are the exit polls from last election," said Miringoff. "They were good for the time they were taken, but that was four years ago."
Still, some Republicans argue that there's a so-called "confirmation bias" among pollsters that's impacting how they conduct their surveys.
Erick Erickson, a conservative blogger who has recently been critical of Romney, said that even though he believes Romney is behind, polling organizations are likely influenced by a Democratic bent.
"I do not believe the polls are all wrong. I do not believe there is some intentional, orchestrated campaign to suppress the GOP vote by showing Mitt Romney losing. I actually believe that Mitt Romney trails Barack Obama," Erickson wrote in a blog post on Wednesday. "But I also believe the polls are reflecting a bigger Democratic strength than is really there."
Erickson points to a so-called "confirmation bias" in mainstream polls, citing a blog post by the National Review's Jim Geraghty. "The polls are confirming what the press thinks and that they have a larger than 2008 Democratic turnout is of no consequence to them," he writes.
Miringoff dismissed the notion that polling agencies would fall prey to such a media bias.
"Even if I could in any way do what pollsters are often accused of doing when others don't like the results, there's no motive," he said. "There is an election and somebody is going to win and when they do we will either look good or we will look bad. I would prefer to choose the right candidate with as a close as a point spread as possible."
Politically, he said, "we don't have any skin in this. We're evaluated on how accurate we are."
