This will be a short diary as I am no Nate Silver or David Nir and may be totally unqualified to write a critique of electoral polling. That said, I'm gonna do it anyway. Two PPP polls that came out today were disappointing enough to me that I dug around a little in the polling samples and was surprised to see that they do not seem representative of the actual electorate. I am referring to the Missouri poll of Akin v. McCaskill and the Wisconsin presidential poll, each showing the red team up by one point, but each using a polling sample that appears to be significantly redder than the electorate in the respective states.
First, the Missouri poll uses a polling sample that voted 49-44 for McCain over Obama in 2008, when the actual results were 49.4-49.3 in favor of McCain. They also used a sample that identified by party as 30% Democrat, 39% Republican, and 32% Independent when the exit polls from 2010, a good year for Republicans, show an electorate that was 34% Democrat, 37% Republican, and 28% Independent.
Likewise, in the disappointing Wisconsin poll, the polling sample voted 51-43 for Obama over McCain, but the actual results were 56-42 in favor of Obama. The sample in the recent poll identifies as 32% Democrat, 34% Republican, and 34% Independent, when the exit polls from 2010, again a good year for Republicans, show an electorate that was 37% Democrat, 36% Republican, and 28% Independent.
If both of these polls were adjusted, even pessimistically, to reflect the 2010 electorate, they would both show a solid victory for the blue team. So, what's up with PPP?
Pre-emptive apology: I don't diary much and couldn't figure out how to link to any of the polls. I am also open to being shown the error of my analysis, as I am an admitted amateur.