A new poll showing President Barack Obama with substantial leads over Mitt Romney in Florida and two other swing states is likely to intensify a national debate over the accuracy of political polls.
The poll -- by Quinnipiac University, CBS and the New York Times -- shows
Obama with a 9-point lead among likely Florida voters, 53 percent to 44
percent.
That's troubling for Romney, considering a consensus among political
experts that he can't win the election without winning Florida.
If it's accurate, said University of Central Florida political scientist
Aubrey Jewett, "It would mean Romney was in a world of hurt. It probably would
mean his chances of winning the presidency would be nearly zero."
But Jewett, who's politically neutral, thinks it's not accurate -- and so
do the campaigns of Romney and Republican Senate candidate Connie Mack IV. The
poll showed Mack trailing Democratic Sen. Bill Nelson 53-39 percent.
It showed Obama leading Romney 53-43 percent in Ohio, probably the most
important swing state in the election besides Florida, and up 54-42 percent in
Pennsylvania.
Romney Florida strategist Brett Doster called it "almost malpractice from
a polling standpoint."
Their complaint is the same one conservatives have been making nationwide
about polls since they began showing Obama edging ahead of Romney in the
aftermath of the nominating conventions.
These critics contend the polls are based on "turnout models," or
predictions of voter turnout on Election Day, that include too many Democrats
and not enough Republicans.
Spokesmen for several pollsters including Quinnipiac deny that, saying
they aren't using turnout models at all.
Their polls show fewer Republican respondents, said Quinnipiac spokesman
Peter Brown, simply because the party has lost popularity compared to the 2010
and 2004 election years. Respondents, they say, are less likely to claim GOP
affiliation.
Jewett and a couple of other neutral experts said they accept the
pollsters' explanation, but still don't accept the results of the new
Quinnipiac poll.
"There's a natural tendency to question the accuracy of polls when you're
losing," said University of Virginia political scientist Kyle Kondik. Kondik
noted Democrats have voiced the same criticism of political polls in past
years when Republicans dominated.
"But if Obama's ahead by 9 points in Florida, that means he's winning by
more than he did last time, which seems odd," he said.
Kondik pointed out that polling averages posted on political news
websites show a much narrower race, with Obama just 2-3 points ahead in
Florida. This jibes with predictions by political scientists who study the
effect of the economy and the power of incumbency on elections.
Mack told reporters during a stop in Tampa today that polls including the
new Quinnipiac survey are oversampling Democrats.
"Most people recognize that these polls are out of whack," Mack said.
In the Mack race, however, other recent polls have also shown Nelson with
double-digit leads; the average is about a 9-point edge for Nelson.
Conservative political analyst Dick Morris, in a blog posting a week ago,
said pollsters nationwide are basing their polling samples on "2008 models of
voter turnout," when excitement over electing the nation's first black
president altered the electorate.
Black voters swelled to 13 percent of national voters from 11 percent in
2004, and Hispanics and under-30 voters also hit "historic highs," Morris
said. That, he said, isn't likely to happen again in November.
Conservative blogger Dean Chambers of Virginia has even created a
website, "Unskewed Polls," that adjusts the results of published polls to
match what he considers a more realistic voter turnout. His findings give
Romney a significant lead.
In the new Quinnipiac poll, 27 percent of the respondents told pollsters
they were Republicans, 36 percent said they were Democrats and 37 percent gave
no-party, minor party or "don't know" answers.
That's a 9-point advantage for Democrats over Republicans, much bigger
than Democrats' 40-36 percent edge in actual Florida voter registration.
Democrats also aren't likely to have a 9-point advantage in turnout on
Election Day, Doster said.
In 2004, he said, exit polls showed 41 percent of voters were Republicans
and only 37 percent Democrats. In 2008, "a super high-water mark for
Democrats," they beat Republicans, but only by 37 percent to 34 percent.
Still, most agencies that publish national political polls say they don't
seek to achieve any partisan breakdown in their polling samples.
"That's exactly what we're trying not to do," said Michael Dimock of the
Pew Research Center, whose polls have also generated criticism from
conservatives.
Like most pollsters, Pew and Quinnipiac adjust the samples in their polls
to reflect the population's breakdown by age, gender and race -- a process
called "weighting" -- and then count likely voters in the sample.
They ask respondents what party they associate themselves with, but don't
seek to adjust the sample to create any partisan balance.
"He's wrong if he thinks we're going to weight the poll to some idea of
previous elections," Dimock said, referring to Morris's critique.
If the polls show what appear to be excessive numbers of people saying
they belong to one party, said Brown, it's because, "The Republican brand is
clearly having problems."
"Party self-identification changes based on the times and what's going
on," he said.
He pointed out that Democrats in Florida, Ohio and elsewhere previously
made exactly the same criticism -- in 2004 and in 2010, election years
dominated by Republicans.
