Ooops, hold the phones:
Zogby International conducted telephone interviews of 1018 likely voters chosen at random nationwide.Still, it might be worth considering Zogby's results since he at least acknowledges the methodological problems -- and had perhaps the most accurate results in 2000:
Of all the pollsters monitoring the 2000 election, Zogby made the most accurate prediction of the down-to-the-wire race for president.Then again, that's from NewsMax, which has a business relationship with Zogby, and Zogby's state polls from 2000 were not nearly as accurate as his final national sample.
In any case, Zogby's latest poll, conducted September 8-9, had Bush/Cheney over Kerry/Edwards by 2 points (47-45) in the head-to-head and 4 points (46-42) in the comprehensive one that included Nader, Badnarik, etc. In each poll, there are 7-8% undecideds, who break mostly for challengers over incumbents when it is time to vote.
August 12-14, Zogby had Kerry up by 7 in the head-to-head (50-43) and by 4 in the broader one (47-43). The margin of error is 3.1%, so this looks like a close race that swung a bit towards Bush in the past month. 53% of those polled disapproved of Bush's job as President. When over half of voters say this, it usually means big trouble for an incumbent.
Moreover, Bush's margin over Kerry is larger in the "red states" (51-38) from 2000 than Kerry's is over Bush in the "blue states" (47-40). Since an electoral vote win is a win, whether it is triggered by 51-49 or 60-40 popular vote margin, this likely means the race remains very close on the state-to-state level. Who cares if Bush is winning Texas and Alabama decisively? Kerry has to retain all the close states, like Iowa, New Mexico...and Florida.