[aristeides]

That means that Kerry could be at 30%, and Dean at 15%.

That one candidate is at the edge of the margin of error has 5% probability. It's possible, but not very likely.

That two candidates are at the edge of their margins of error has a much lower probability than that. I don't know how to calculate this probability (you can't just square 5%, because the fact that one of the candidates is much higher or lower than he should be means that any other candidate should move in the opposite direction, but it's very unlikely one single candidate should move by the whole difference.) But since it is much lower than 5%, and I would also guess lower than 1%, in statistics it would ordinarily be discounted.

[cousair]

Newsweek has Kerry at 30%, Edwards at 13% and Dean and Clark at 12% even. The other poll says 52% of folks asked do not want Bush reelected with 44% saying they do....does anyone else come up with this and how accurate is Newsweek???? The last time I looked it was totally pushing left voters . Remember who Zogby is..and how he askes his poll questions/