>That may say something about polling in elections.
>It also might be that ~10% of the people are changing their minds
>on a week-by-week basis. After all, that's what the final barrage
>of advertisements is all about.
Still, when you tease out demographics from the various polls, the contradictions are even more stark. For example, while Zogby-Reuters and Gallup now both have Gore and Bush within a couple of points, Gallup's numbers show that young people are supporting Bush overwhelmingly (54 to 39) with the race essentially a dead-heat for everyone age 30 and above. Zogby claims that Gore's strongest support is among young voters, a flat out reversal of Gallup's age demographics - although his exact numbers on this point were not posted.
Part of the explanation may be the model of who is a likely voter, since different groups of young people may be eliminated from polls based on the model of whether they will actually vote. Which is the key to most of the bizarre poll behavior since turnout is the great unknown- the 1994 GOP takeover of Congress was based not on some great tide of conservative voting but on Dem-leaning voters deciding to stay home.
Given the pretty underwhelming enthusiasm for this race, the weather prediction for each state on election day may end up being a far bigger factor than most of these polls.
-- Nathan