Polls in 2000 wrong by 5 to 8 points

[Zuni]

on average. Do not listen to them. We all know who won in 2000, and that is not who the polls said.

[ciaobox]

1. Exactly what I have been saying for a week...

Add +5 to +8 to Kerry on every poll you see. Subtract -5 to -8 from Bush. Those are the ACTUALS.

[Proud2BAmurkin]

2. And remember, all were 5 to 8 points in AWOL's favor.

!

[Village Idiot]

3. Here's an accounting:

Electoral College Update 32: Kerry 277, Bush 261


Even so, that Gallup result, like previous Gallup polls this year, is so postive for Bush it's off the charts. Ruy Teixeira has reasonable critiques of the pollster's oversampling of Republican and white voters here and here. But I have to say that Tom Burka has the most satisfying explanation:
The Gallup Organization, responding to charges that it greatly over-sampled Republicans in calculating Bush's and Kerry's support among likely voters in its latest poll, said that it did so in order to take into account the massive voter fraud which is expected to take place between now and election day.
According to Gallup, among registered voters, Bush leads 46%-43%, a virtual tie within the margin of error. Among Gallup's likely voters -- by which they mean, "voters whose votes are likely to be counted" -- Bush will win easily, 52%-44%.

"We're quite surprised that other polling organizations have failed to artificially jack up the Republican sample in the likely voter mix given the track record of Sproul and Associates and their ilk," said Jack Gallup. "Why aren't other pollsters counting the soon-to-be uncounted?"…

"We expect thousands upon tens of thousands of democratic voters to be methodically disenfrachised -- and that's just in Texas. In the swing states, we expect democratic voters to be -- well, rounded up and taken to Texas."