Quinnipiac University released three new "swing state" polls for Florida, Ohio, and Pennsylvania and all show troubling signs for Obama. Most worrisome for Obama (and promising for conservatives) is that he is polling under 50% in all three states: 42% in Florida, 44% in Ohio, and 47% in Pennsylvania.
Despite multiple "official" visits, the power of incumbency, being against an opponent supposedly nobody likes, and having not been campaigned against yet, those numbers are incredibly weak. But dig into the numbers, and you find that it gets even worse for Obama as the pollsters weight the data to favor Obama in every possible way.
Florida:The poll shows Romney leads Obama by 44-43%, after figures are weighted. (In weighting, pollsters can alter the data to reflect what they believe the actual election turnout will be, so as to minimize the effect of skewed polling samples. This estimate can be quite subjective). Before weighting, Romney's lead was larger. So, how did the pollster weight the Florida data? The Republican sample was weighted to be only 28% despite being 36% in 2010 (a "Republican year") and 34% in 2008 (a "Democrat year").
The weighting also reflects the pollsters opinion that Democrat turnout will be 3% higher than Republican turnout, 3% points higher than it was in 2010 and equal to 2008. So even by weighting to Democrat-favorable 2008 turnout figures (unlikely to be repeated in 2012), Obama is still trailing.
Ohio:In Ohio, it's more of the same, but even a lot worse. While the poll shows Obama winning by just 44-42%, those figures also reflect very favorable Obama assumptions. Once again, the polling sample is weighted to drop Republican turnout to a shockingly low 24% while it was 37% in 2010 and 31% in 2008. Meanwhile, the pollster kept Democrat turnout at 36%, exactly where it was in 2010 and only 3 points less than 2010.
So while turnout in 2010 saw Republicans at 37% to 36% for Democrats, this poll "weights" the turnout to be 36% for Democrats and 24% for Republicans. And even then, Obama's lead is just 2%. If you weight the Ohio numbers to actual 2010 turnout, Romney leads 48-42%. So, a lot of these polls reflect only how pollsters prefer to "weight them." But in any case, a weighted sample of 24% for Republicans is extremely low.
PennsylvaniaPennsylvania is much of the same. While Republicans made up 37% of the electorate in both 2008 and 2010, the pollster once again weights the information so that Republican turnout is just 29%. The Democrat have a 7 point margin advantage in the weighted sample, equal to 2008 (and more than double the 3 point advantage of 2010). Here, Obama is still at only 47%, even though he scored 55% of the PA vote in 2008 (and even though a large chunk of the GOP has apparently moved out of the state according to the pollster).
Bottom LinePolling is as much subjective as it is objective. After collecting scientific data (polls) pollsters can then alter that data to reflect whatever turnout they expect. In all three swing state polls, the samples were weighted to favor Obama, and were very similar or even more generous than 2008 turnout. Even the weighted numbers show very good signs for the tea party, opponents of Obamacare, freedom lovers, and Romney, and very troubling signs for Obama.
Comments
Votes, not polls are what counts. So far the GOP has alienated women, Hispanic, and LGTB voters and It’s far too late to do anything about it. Good luck with that “freedom lover.”
Hi Dustin –
Until a likely voter model is determined, almost any polling, let alone weighting, is more fortune telling than science.
Woody, nice talking points, but pandering, belittling, insulting the intelligence of, and patronizing people by obsessing over their race and gender has worked wonders for liberals for along time. At some point, it will start to wear thin. But refusing to go to such demeaning lengths to get a vote is not “alienating.”
RealTime, I agree it definitely is not science. But you can also apply logic to the polls to get a sense of where things are. (And if you choose to weight, weighting by dropping Republican respondents far lower than anything seen while leaving Dems at normal or above average levels is purely dream-state and fortune-telling malpractice.
Let me get this straight. Republicans cry foul when the Dems try to shove their leftist ideology down the throats of the Catholic Church. And somehow the religious “freedom” issue gets fabricated into a anti-women message. Say what? I must have missed something here. Yeah. What I missed was a perverted, twisted, slanted media trying desperately to frame every Republican issue as anti-gay, anti-women, anti-minority, etc.
Gee, did I leave anything out? This media has long ago lost any sense of neutrality or impartiality. They might as well be on the payroll of the DNC. Nauseating. Vile. Disgusting.
Hi Dustin –
“In weighting, pollsters can alter the data to reflect what they believe the actual election turnout will be, so as to minimize the effect of skewed polling samples.”
Not quite. Weighting has less to do with election turnout, and more to do with sample size, and the proportions within that sample.
For example, say that your sample is 20% Republican, 20% Democratic, and 60% Independent. Your likely voter turnout will be 40% Democratic, 40% Republican, and 20% Independent. You would weight your sample accordingly, to get useful data.
“Before weighting, Romney’s lead was larger.”
Republicans were oversampled in the original data.
“So, how did the pollster weight the Florida data? The Republican sample was weighted to be only 28% despite being 36% in 2010 (a “Republican year”) and 34% in 200.”
Democrats were also undersampled and weighed at 31%. What is the actual percentage amongst registered voters?
http://www.palmbeachpost.com/news/state/gop-closes-in-on-democrats-in-florida-voter-2107649.html
“Figures released by the Division of Elections today show Florida has 11.2 million voters, with 40.5 percent registered as Democrats and about 36.2 percent as Republicans. The gap of 4.3 percentage points between Democratic and Republican registrations compares to a 5.8-point gap that favored Democrats heading into the 2008 presidential election.”
Republicans had 366 respondents in the survey, Democrats had 336. Democrats hold a slight edge in registered voters. A three point differential in the weighting is not unconscionable.