Cyc, the Fox News Poll has not shown Bush to consistently have a lead, nor has it, when showing a Bush lead, shown as large a lead as say Gallup, Washington Post, or Newsweek. Note too if you will that the Fox News/Opinion Dynamics Poll Final for the 2000 election was the one that got both the spread - 0 points, and the percentage of popular vote - 48% ea - the closest of all the major pollsters. For the 2002 midterms, they outperformed Gallup, Zogby, Pew, and Harris. For the 2003 special elections, they called 4 out of 4 correctly. During this year's primary season, their track record was very good - aqmong other things, they were the first to indicate Dean was not a real player. They got laughed at for that call ... particulary garnering dismissive scorn from the Deaniacs.
Oh, and electoralvote.com hardly can be deemed non-partisan. As to the methodology there, they go with the most recent poll for each state, having found a brief experiment with averaging the 3 most recent polls rendered results which distressed their visitors. A much better picture, IMO, are offered by
RealClearPolitics[/i][/u] and
Dave Leip's Election Atlas. Over the past dozen years, among the best overall prediction performance - in terms of winner and vote share alike, primaries, general, and presidential elections, has been that of the
Iowa Electronic Market.