Putrid Poll

Pew is out with a new poll showing Obama with a commanding ten point lead. Do not believe it. 

Here are the cross tabs:  http://www.people-press.org/files/2012/08/8-2-12-Topline-for-release.pdf  .

A few takeaways:

Of those interviewed:

Democrats: 33%

Republicans 22%

Independents: 38%

Rasmussen says the current party mix from its surveys of 15,000 interviews each month: is 35% GOP, 34% Democratic, 31% independent, or non-aligned. 

Pew has the approval rating for Obama: 51% to 42%. Today in Gallup: 45-49, Rasmussen: 45-53. 

Here is Pew's track record from prior races versus actual results using their final poll before the election:

2008: Obama by 11%, won by 7%

2004: Kerry by 1%, lost by 2.4%

2000: Gore by 4%, won by 0.5%

1996: Clinton by 19%, won by 8%

1992: Clinton by 10%, won by 5%

They are always favorable to Democratic candidate, by an average of over 5% in their final poll. That is a huge error for a final poll. 

We are seeing a systematic oversampling of Democrats by groups I view as friendly to Democrats: the major networks, Pew, Reuters, AP.  Rasmussen and Gallup, both of whom interview far more people than the once a month pollsters,  continue to show a race that is about even. 

The Pew poll says Obama is ahead by 4% in battleground states. Does that make sense?  That he is ahead by 10% nationally, but 4% in battleground states?  In 2008, Obama won by 7% overall. In battleground states -- he won Michigan by 16%, Pennsylvania by 10%,Wisconsin by 14%, Minnesota by 10%,  Colorado by 9%, Iowa by 10%, New Hampshire by 10%, New Mexico by 15%, Nevada by 12%, Virginia by 6%, Ohio by 5%, Florida by 3%, Indiana by 1%, New Hampshire by 0.3%, and lost Missouri by 0.2%. Overall, average margin in 15 battleground states was about 8%.   My guess is that Pew's battleground number is closer to reality, and their national is simply wrong.