lotstodo said:
There is a blog by a statistician that puts the likelihood that the polls are skewed enough in Obama's favor to miss a Romney win at about 30%. But his assumption is that the "average" of 23 polls is correct. This was not the case in 2008 where 17 of the polls overstated the Obama advantage and only 3 overstated Romney's tally, and he admits that he hasn't studied the "Homer" effect directly in each poll. He based his number on his own assumptions. There is absolutely no poll that doesn't make some assumption about turnout and sample population.
I still do believe that the only way Romney is going to win is if the polls are wrong, and they will have to be wrong by about 2%. That would be a huge miss, but we can hope. It really does point out that this miss is extremely important in the toss-up states far more than the rest of the country. Lets hope that Ohio, Florida, and the rest of the close calls have been fooling them all along and Republicans swamp the polling places.
The latest CNN poll shows the race tied at 49%, but there was an oversampling of Democrats of 11%.
The latest Pew poll shows Obama leading 48% to 45%, but an oversampling of Democrats of 6%.
All the other national polls show the race tied.