As dvwjr revealed Here Newsweek oversampled Republicans again.
Advise viewing the sampling and std. dev. error statements before placing much faith in this poll.
Sorry to be a bummer - I am confident (and have been for 2 months) W will win by 5-7%, but I smell the set-up before the attempted "October Surprise".
Potential hit is Troops Mutiny
Whenever newsweak polls more of one party than the other, they apply a weighting formula to even things back out to the voter turnout model of 2000. Gallup does not care what party the person polled belongs to. They use a "random" sample.
LLS