[comebacknewt]

This is not necessarily a sign of polling bias. Newsweak never weights their samples. They simply call at random and record whoever answers.

After the GOP convention, Republicans were excited. They were more likely to pick up the phone and talk with a pollster, and so Newsweak included many more of them in the post-GOP convention polls. After the debate on Thursday, it was the RATs who were fired up and willing to respond, and this poll reflects that.

This race has moved back into a tie. Fn's debate performance helped shore up his support in his base and won him more undecided voters than Bush did.

This was likely to happen anyway as election day approached. The debate just hurried it along. Neither side is going to get a blow out.

[jwalsh07]

This is not necessarily a sign of polling bias.

I agree with this statement. I covered it above. However, if no other pollsters pick up a movement away from pubbie self ID, I would say the poll is suspect. We'll see.

[COEXERJ145]

Actually Newsweak does weight their sample for party affiliation and "adjusts" the numbers to make sure they conform to "census numbers" and "voter turnout model" which most likely is from 2000 and out of date.