>>I was called by a pollster last night. The questions surprised me. Not once was I asked who I would vote for or if I had voted. Yet the purpose of the poll was determine who I would vote for. At the end of the poll, I asked the fellow how could he tell who I would vote when that question was never asked. He informed me that they could tell by how I answered the questions. I wonder...
>
>Among persons I know personally, this Gallup (http://www.gallup.com/tag/Election%2bTrends%2bby%2bGroup.aspx) has been quite accurate. If someone I know is older, churchier, less educated, whiter and less educated, maler, whiter maler, more conservative and Republicany, whiter, Souther they are usually for the same one of the candidates. If someone I know is younger, less churchy, more educated, less white and more educated, femaler, less white and femaler, more liberal and Democraticy, less white, less Southern they are usually for the same other candidate.
>
>References to polling result appear to cut two ways: 1) supporting your position when your candidate is ahead 2) totally meaningless and fraught with big margins of error and bias when your candidate is behind. Maybe we should do polling about polling to resolve this?

The current polls seem extremely wise and accurate ;-)