RCP Avg Strikes Again with Misleading Polls - This Time in Ohio

According to the Real Clear Politics Average in Ohio, the President is leading Governor Romney by 1.7% in the so-called "poll of polls." As I've said before, I am willing to at least give RCP the benefit of the doubt, but still, this average includes four polls that either by accident or design use highly Democrat skewed polls:

RCP is using four polls with skewed samples:

Survey USA, uses a +7% Democrat sample to get Obama to +3%. If we unskew that data to a +3% Democrat Sample, using Survey USA's own internal data of how Republicans, Democrats and Independents are polling, we find Romney with a slight lead of 44.7% to Obama's 44.0%

CBS/Quinnipiac, uses a +9% Democrat Sample to get Obama to +5%. If we unskew that to a +3% Democrat sample, and using CBS/Quinnipiac's own data, we find Obama's lead has cut to one tenth of the poll's prediction to 0.5%: Obama 45.8% to Romney 45.3%

Gravis Marketing used a sample of +9% Democrat, and that showed a tie. If we unskew it to a +3% Democrat sample and use Gravis' own internal data,

The PPP and Fox News polls could not be accurately unskewed because they did not publish their internal data did not provide the necessary breakdowns of how Republicans, Democrats and Independents intended to vote.

So apparently Obama is in fact losing Ohio, very slightly. This makes a difference. Lets look at the RCP "No Tossup" Map, shall we? This is the map using the skewed Ohio sample:

Now let's look at the map using our new data that unskews the samples to +3% Democrat instead of +7-9% Democrat:

One other issue for the President is he's at an average of 46.3%. Well below 50%. Historically you see undecided voters this late (12 days before the election) swing hard to the challenger. I'll even be generous and call it 2:1 for Romney. When that happens it's 51.7% Romney, 48.3% Obama.

Bottom line, exactly what I've been telling you is true: the RCP average, when a Drive-By Media source throws a wrench into the mix and skews the average, it spoils the entire sample. Here in reality, Romney is winning. And that's all that matters.

Swing States Swing to Romney Post Debate

Mitt Romney's debate performance last week was one of the finest debate performances I've seen since I started watching debates twenty years ago. Many predicted he would perform well in the debates, but many also said that debates rarely change elections. Welcome to rarely.

Over the weekend, a lot of things started swinging to Mitt. In Gallup, for example, Governor Romney went from losing by 5% before the debate to tied with the President. This is huge, especially since Gallup polls Registered Voters not Likely Voters. In Rasmussen, Governor Romney went from 2% down on Friday to 2% ahead on Saturday.

But what about the swing states? What happened post debate there?  Let's look at recent polls (regardless of organization) before and after the debate. (We'll use statewide party registration data as our baseline where appropriate.)


Colorado (Party Registration +2% Republican):

- Before the debate, We Ask America had Obama leading by +3%.

- After the debate, Gravis Marketing had Romney winning by 3%. (Using an accurate +2% Republican sample)

Florida (Party Registration +5% Democrat):

- Before the debate, Suffolk had Obama leading by 3% (Sample Democrat +6%) and Gravis had Obama +1%.

- After the debate, We Ask America had Romney leading by 3% and Rasmussen had Romney winning by 2%.

Nevada (Party Registration +5% Democrat)

- Before the debate, We Ask America had Obama leading by 11%.

- After the debate, Gravis Marketing had Obama's lead down to 1%.


Now I recognize we're using lots of polls here, but that's because we need to as these organizations are not being consistent in their polling during the campaign. However, one thing is clear. After the debate, Mitt Romney is making a huge comeback. Game on.

Ohio (Party Registration +1% Republican)

- Before the debate, Public Policy Polling had Obama leading by 4% (using a +5% Democrat sample) and NBC, using a tiny sample and a huge margin of error (using a +8% Democrat sample) had Obama leading by 8%.

- After the debate, We Ask America had Romney leading by 1% and Rasmussen had Obama leading by only 1%.

Virginia (Party Registration +3% Republican)

- Before the debate NBC, using a tiny sample and a huge margin of error (using a +2% Democrat sample) had Obama leading by 2%.

- After the Debate, We Ask America had Romney leading by +3% and Rasmussen had Romney leading by +1%.

Wisconsin (Party Registration +4% Democrat):

- Before the debate, the Marquette University claimed Obama had a lead of 11% over governor Romney. (Sample +7% Democrat).

- After the debate, Democrat Public Policy Polling

Swing State Polls are as Cooked as the National Polls (Part 1)

The Drive-By Media spent last week telling us it was absolutely over. Obama has a commanding lead in the Swing States. He's going to win, they tell us. The current Real Clear Politics average shows Obama with 269 electoral votes either strongly, likely or leaning his way. So they say.

There was a day they would get away with this. We talked about a few of these cooked Swing State polls last week:

Biblical Conservatism: CBS is Now Cooking Swing State Polls

So let's look at where other polls are being cooked, shall we?



Ohio - RCP Avg: Obama +5.5% - Avg Democrat Oversample D+ 6.6%Ohio Voter Registration: +1% Republican

- For Public Policy Polling to get Obama to +5%, they had to oversample Democrats by +4%.

- For the Columbus Dispatch to get Obama to +9%, they also had to oversample Democrats by +9%.

- For Gravis Marketing to get Obama to +1%, they had to oversample Democrats by 10%.

(Other polls were included in this average, however they did not publish their samples.)

Florida - RCP Avg: Obama +3%. - Avg Democrat Oversample: 8%
Florida Voter Registration: +5% Democrat                                                                                              

- For Gravis Marketing to get Obama to +1%, they had to oversample Democrats by 7%.

- For Public Policy Polling to get Obama to +3%, they had to oversample Democrats by +9%

- For the Washington Post to get Obama to +4%, they had to use a sample of only 161 Likely Voters and a margin of error of 11%.

(Other polls were included in this average, however they did not publish their samples.)

I hope I've painted a nice, clear picture for you. In Ohio, to get Obama to an average of +5.5% the polls had to oversample Democrats and average of 6.6%. Unskewed and assuming that 90% of the Republicans would vote for Romney, we're looking at Mitt having a lead just of about .5%.

In Florida, to get Obama to an average of +3% the polls had to oversample Democrats by an average of 8%. Adjust that to an even turnout (which is a very reasonable guess this election) and assume 90% of the Republicans properly weighted back in, you see Romney leading by 2.5% in Florida.

This post will be continued Monday with a look at Wisconsin and Virginia.