Posts Tagged ‘oversampling’

Who Won The Obama-Romney Debate Last Night: A Closer Look At The Post-Debate Poll Shows It Wasn’t Even CLOSE.

October 17, 2012

Here’s an interesting article which has the virtue of laying out the most important facts in a very few words:

CNN Poll: Obama wins overall, Romney wins many issues
Oct 16, 2012
by Carla Marinucci

The major post debate polls are showing most debate watchers think President Obama was the winner — and a typical finding comes in the CNN post-debate poll of registered voters who actually watched the debate. They were 33 percent Democratic and 33 percent Republican — which means it’s weighted a bit to include about 8 percent more Republicans than the voting population as a whole.

Bottom line: by 46-39 percent, the poll respondents called Obama the winner. But asked who did the debate make you more likely to vote for, the respondents were tied between Romney and Obama, 25-25 percent.

Still, Romney won in other key areas:
*On who would better handle the economy: 58 percent Romney; 40 percent Obama.
*On who would better handle health care: 49 percent Romney; 46 percent Obama
*On taxes: 51 percent Romney; 44 percent Obama
*On who is a stronger leader: 49 percent Romney; 46 percent Obama
*On who is more likeable: 47 percent Obama: 41 percent Romney
*On who cares more about your life: 44 percent Obama; 40 percent Romney
*On who answered more directly: 45 percent Romney; 43 percent Obama

Okay, first the sampling of Republicans versus Democrats that Marinucci mentions.  It is incredibly interesting to me that CNN is giving Democrats an eight-point advantage in their sampling formula.  Do you know what the balance between Democrat and Republican voting was in the historic Obama wave election of 2008?  It was +7 Democrat.  CNN says it’s going to be +8 this time.  It’s really amazing when you stop and think about it: CNN and the other polling organizations that have such an oversampling formula for Democrats are literally predicting that this election will be even BIGGER for Obama than it was in 2008 when he was the messiah and everybody was blaming Bush and nobody could blame Obama because he promised hope and change.  Is that what you think?  Seriously?  Do you believe that the masses are even crazier for Obama now than they were four years ago?  I sure as hell don’t.

Okay, so the CNN poll shows that by a seven-point margin that Romney blew the doors off in the last debate, “Obama won the debate.”  You might remember that first debate – when Romney destroyed Obama by 50 damn points.

Well, here’s the question, when it’s time to vote, do you believe that voters will ask:

Who won that second debate?  The poll says maybe Obama did.  And the first debate that Romney won by fifty points is meaningless but that second debate was huge.  I’ll vote for Obama…

Or do you think they will ask:

Who will do a better job handling the economy?

If it is the second thing, like every sentient being (sorry Democrats, I know that rules you out) thinks it is, Mitt Romney won last night’s debate HUGE.  Who would better handle the economy?  After watching the debate, people by an 18-point margin chose Romney over Obama.

[Update, 10/17/12: Oh, my.  It’s even worse than I thought it was for Obama.  I found out that the CBS post-debate poll was even WORSE in proving that Obama has no answer for the economy and people overwhelmingly know he doesn’t.  Check this CBS article points out: the American people think that Mitt Romney will do a better job handling the economy than Obama by a 31-point margin (65% to 34%).

The Gallup poll among likely voters out today has Mitt Romney up by six points nationally.  I can’t imagine that Obama even slowed Romney’s momentum down last night.

Reweighting Polls In Real Clear Politics Average To (Most Accurate In 2008) Rasmussen Assumptions Gives Romney Massive 7.8 Point Advantage

September 29, 2012

Rather interesting.  If you take the mainstream polls that are popping up like weeds in spring, and reweight the assumptions to reflect the most-accurate-in-2008 Rasmussen model, you get Romney winning in a landslide.

The truth about 2012 polls
By Douglas E. Schoen
Published September 27, 2012

In the 2012 race for the White House President Obama is ahead, but the polls  are misleading.
It seems that each new poll brings good news for  Obama. He’s up six points nationally according to the latest Bloomberg numbers.  Gallup’s weekly tracker has the president up six as well. And it looks like  crucial swing states are going for Obama in a big way: the latest Quinnipiac  poll gives Obama a nine point edge in Florida, a 10 point advantage in Ohio and  a 12 point lead in Pennsylvania.

To be sure, Obama is ahead in this race. But by how much has become a serious  point of contention and one that deserves further examination.

Republicans and Democrats alike have honed in on the fact that recent media  polls are oversampling Democrats. Indeed, we have seen many polls that are  heavily skewed. There was the Washington Post/ABC poll that had a +9 Democrat  skew in late August. There was the Marquette poll for Wisconsin from two weeks  ago with a D+8 sample. And the newest swing state poll from Quinnipiac gave  Obama a spread between Democrats and Republicans that was even greater than the  historic Democrat advantage in 2008, a seven point spread between voters  identifying themselves as Democrats or Republicans at 39 percent to 32 percent,  in each state they polled.

In a recent interview, Romney pollster Neil Newhouse made the argument that  these mainstream polls are skewed in favor of Obama. “I don’t think [the polls]  reflect the composition of what 2012 is going to look like,” he said.

In order to address this, some conservative outlets have taken matters into  their own hands. One website,,  has begun reweighting mainstream polls to more closely track the demographic  assumptions that the conservative leaning Rasmussem Reports uses. The results  have been staggering: the re-weighted polls all put Romney ahead of Obama with  margins of between 3 and 11 points. If one looks at the Real Clear Politics  average Obama is currently up four percent over Romney. But according to,  Romney has a 7.8 percent edge on Obama.

The expectation by mainstream pollsters is that the turnout in 2012 will be  the same, if not better, for Democrats as it was in 2008.

There was a seven-point increase in the party identification gap between 2004  and 2008 nationally, which was the largest shift in a generation. If you look at  individual swing state results from 2008 you see a different story in certain  cases like Florida where exit polls show a Democratic skew of just D+3.  Pennsylvania and Ohio were similar to the national result at D+7 and D+8, but  these figures are still less than pollsters are sampling. The Quinnipiac poll  skew was D+9 in Florida and D+11 in Ohio – a substantive difference from using  2008 as a predictor.

The assumption the pollsters are making that turnout in 2008 will be the same  and even better for Obama than in 2008 is flawed. Not only are we looking at a  terrible economic situation, but there will be key differences in turnout from  2008 that will affect the results and the accuracy of these polls.

Democratic  registration may be overstated. It is my belief that many weak  Obama voters are saying that they are Democrats when they really aren’t  partisans at all: they are disillusioned with American politics. What this means  is that these people aren’t even certain to vote in November and if they stay  home, Obama’s numbers will surely be affected.

Additionally, fewer young people will turn out at the polls this year. As  evidenced by Obama’s push to mobilize the youth vote, a group that he won  handily in 2008, demonstrates, this is a key group that is becoming increasingly  apathetic and is apt to turn out in fewer numbers.

Even so, these differences do not mean that the Democrat skew in the  electorate is a total misrepresentation of the electorate today. While the  seven-point bulge from 2008 is too large for this election, there is most likely  to be a three to five point skew to the Democrats.

If you look at the results of these polls and account for the average three  point margin of error and take another 2-3 points from Obama’s number to account  for the overstated Democrat skew, you still have Obama ahead by anywhere between  two and four points.

What’s more, Obama’s job approval rating has been 50 percent or higher in  each of the last four Gallup daily tracking figures and now 50 percent of  registered voters prefer him in the election. Crossing the 50 percent threshold  is a crucial indicator of an incumbent’s chances at reelection and shows that  things are shifting in Obama’s favor.

To be sure, Obama is in the lead. The polls are oversampling Democrats, but  he remains ahead. But it is also a lead that should not be exaggerated. The race  is not over yet and the debates beginning next week will be a decisive test for  both the president and Governor Romney.

Douglas E. Schoen has served as a pollster for President Bill Clinton and  is currently working with New York City Mayor Michael Bloomberg. He has more  than 30 years experience as a pollster and political consultant. He is also a  Fox News contributor and co-host of “Fox News Insiders” Sundays on Fox News  Channel and Mondays at 10:30 am ET on Live. He is the author of ten  books including,“Hopelessly Divided: The New Crisis in American Politics and  What it Means for 2012 and Beyond” (Rowman and Littlefield 2012). Follow  Doug on Twitter @DouglasESchoen.

I’m not quite sure how Schoen (who clearly is a lifelong Democrat) goes from the lines that I have boldfaced to stating that “To be sure, Obama is in the lead.”  He doesn’t explain the mental gymnastics that led to that point in his routine.  That said, I do recognize that the man is an expert in his field of polling and that he is able to both access and understand the internals of polls in a way that I cannot.  That said, there are other polling experts, such as Dick Morris, who is looking at the same data and drawing the opposite conclusion.

What puzzles me is that somehow after presenting Unskewed Polls in a credible light, Schoen completely dismisses their findings without even bothering to wave his hand at a reason.  As an example, Schoen cites Obama’s job approval rating as over 50%; but Unskewed Polls shows Obama SERIOUSLY under water by more than 8 points:

It’s kind of incoherent for Schoen to present the Unskewed Poll and then completely dismiss it without bothering to explain why we should similarly dismiss it.

When you look at the mess of assumptions and oversampling and whatnot, it very much seems that “polling” is like unto a witch doctor peering into the “internals” of chickens to try to predict the future.  It just aint very wise to put too much credence in either one of them.