Posted by swamphawk22 on 11/5/2012 4:06:00 PM (view original):
How about Gallup did about +2% in the 2008 and 2004 elections for Democrats.
You said "most." You do know what "most" means right?
One example is not even close. And Gallup has been leaning the other direction this time around. Which would support my argument that a poll is only as good as the voter model they use. They can be wrong for reasons other than trying to "skew" the results. Non-partisan pollsters don't have much interest in doing that as it damages their credibility.