WASHINGTON (Reuters) - It has become the new battle cry for Republicans: All the polls showing Mitt Romney trailing by big margins are just wrong because pollsters are interviewing too many Democrats.
Surveys showing President Barack Obama leading nationally by 5 to 7 points, and even more in swing states, have come under fire from the Romney campaign and conservatives who accuse polling companies of misjudging their data at best, and deliberately skewing it against Romney at worst.
While pollsters say the Republicans are griping because they are losing, the kernel of the conservatives' complaints - that pollsters frequently survey more Democrats than Republicans - is true.
But poll companies do not go out of their way to find Democrats - it is just that there are more of them on voter registers than there are Republicans and independents.
Thirty-five percent of registered voters identify with Democrats, 28 percent with Republicans, and 33 percent are independents, according to a Pew study in August.
That make-up of the electorate is reflected in many of the recent polls that show Obama well ahead in the race for the November 6 election.
Even then, Democrats do appear to be over-represented in some of the surveys being criticized by Republicans. The percentage of Democrats interviewed in some polls is a few points higher than the 35 percent found in the Pew study.
"I'm a little uncomfortable at some of the samples, which strike me as surprisingly Democratic," said Stuart Rothenberg, publisher of the non-partisan Rothenberg Political Report, which analyzes political races.
But given the number of public - and internal campaign polls - with similar results, Rothenberg said it was clear that Obama had a healthy lead over Romney, 40 days before Election Day.
"If I don't focus on an individual poll here or there and look at the dynamic of the race, and the broad array of polls, it tells me that the president has a significant lead at this point," he said.
The debate over polls intensified on Wednesday, when a trend of improving numbers for Obama solidified. A Quinnipiac/New York Times/CBS poll in particular drew protests for giving Obama big leads in the swing states of Ohio, Florida and Pennsylvania.
Peter Brown, assistant director of the Quinnipiac University polling institute, said Quinnipiac's samples were random.
Quinnipiac, like most pollsters, does not choose who it will interview based on party affiliation.
If a certain percentage of respondents are Democrats, then that is just because it has turned out that way, Brown said.
"Our numbers are based on a random sample," he said. "We get what we get."
Some conservatives agree reluctantly that, overall, the polls are not going in Romney's favor.
"I've been in politics long enough to know that the louder one side gets complaining about the polls, the more likely it is that this is the side that, in reality, actually is losing," conservative commentator Erick Erickson, who runs the RedState blog, wrote on Thursday.
Pollsters point to history, noting that surveys often get the races right.
"I don't want to be making a claim that the polling is infallible, or even that the average of the poll is infallible," said Charles Franklin, director of the Marquette University poll in Wisconsin. "But the more polls we have across more races, meaning different states and nationally across different years ... on average they perform quite well."
In 2010, when Republicans won huge victories in the midterm elections, Democrats accused polling firms of oversampling Republicans, but the results proved them wrong.
The last presidential election when pollsters were off track was George W. Bush's Electoral College victory in 2000, when opinion surveys forecast that the Republican would win the popular vote, but he ended up trailing Al Gore.
This year, conservatives began grumbling about alleged Democratic oversampling after the two parties' conventions, when Obama's poll numbers began to tick upward.
The complaints got louder as the Democrat's post-convention "bounce" became a consistent lead in many national polls. Virtually every poll, the critics said, except for the one run by conservative commentator Scott Rasmussen, oversamples Democrats to boost Obama's election prospects.
Conservative talk-radio host Rush Limbaugh told his listeners the new polls were part of a conspiracy to convince Republicans to give up on winning the White House.
"They are designed to do exactly what I have warned you to be vigilant about, and that is to depress you and suppress your vote," Limbaugh said on his popular show.
Obama is leading Romney by 7 percentage points in the Reuters/Ipsos daily tracking survey, roughly in line with many other polls. Rasmussen has the race tied at 46 percent.
Some conservative outlets, like the website unskewedpolls.com, recast polls to reflect what they say is the true Republican share of the electorate. The website has Romney leading by an average of 7.8 percent nationally.
Top Romney adviser Eric Fehrnstrom said some surveys were assuming a higher Democratic turnout in 2012 than was likely.
"They contain some flawed methodology that's been pointed out by other people that assume a higher Democratic turnout in 2012 than we actually experienced in 2008. I don't know any political operative or political scientist who believes Democrats are going to turn out in the same numbers that they did four years ago," he said.
The stakes are high for candidates when polling gets bad. Donors are unlikely to open their wallets for races they see as lost causes, and supporters are less likely to turn out to volunteer or vote for a candidate who seems headed for defeat.
There are also risks to looking too good. A candidate who seems to be cruising to victory might lose donors who decide he does not need their money. Complacent supporters are also more likely to stay home on Election Day.
But pollsters note their stakes are equally high, with solid business reasons to get their calls right.
"Our reputation is based entirely on our ability to accurately call the election," noted Cliff Young, managing director of Ipsos Public Affairs, which is conducting polls this year for Reuters.
"At the end of the day, calling the election correctly is the key proof point that you are a competent pollster."
Additional reporting by Steve Holland; Editing by Alistair Bell and Peter Cooney