Pick a poll... any poll...

[elcorazon]

I pick cnn/usatoday.   haha.  I wish.  That one is most likely an outlier.  There is some day to day variation, plus there could be sampling issues with some but really only the bottom 2 are way out of line.  You could explain it to us, vorlon, but I believe even a well-done poll with a small margin of error could be "wrong" in the sense the margin of error is usually described as +/- 3% to a 95% certainty, meaning 5% of the time it is off by more than 3%.  Anyway it seems clear to me that Kerry has a very slight lead right now, which may be no lead at all in the EC, depending on how it breaks down.  With 8 months to go, it is way to close to worry about the nuances, although I'm sure I will.

[elcorazon]

I have forgotten the formula, but the point remains that when a poll has a margin of error, it does not mean it is impossible for a result to be outside of that margin.  This is especially true when there is a new poll every day/week for a long time.  At some point a poll will be anomolous.  Of course, we often don't know that until later.  As Vorlon has pointed out, the results get most accurate right around the election, so we don't know if late changes indicate actual changes or outlier poll results.