[xm177e2]

It's not necessarily bias that led to these bad results. There are really three major groups in the electorate: Liberal Base, Conservative Base, and Swing Voters. In most previous elections, the liberals have turned out their base in large numbers, the larger but less excitable conservative base turns out in similar numbers (but as a smaller percentage of the whole), and swing voters end up deciding the election.

So the exit pollster decided to save money by measuring only the third group, swing voters (by doing exit polling outside of polling places with lots of swing voters). And he found that they were not breaking for Bush enough for Bush to win. He completely missed the GOTV effort and the conservative turnout.

It's conceivable that flawed mechanics, and not bias, led to the failure of the exit pollsters.

[hoyaloya]

Yes, that is conceivable. I do note my experience with exit pollsters in 2000. It goes something like this:

I live in a heavily Republican suburb of a major city. The population is probably 80% white, 10% asian, 8% hispanic, and 2% black. I stood in line to vote for about an hour and had a chance to watch the exit pollster (that's what he claimed he was) work. In that hour, he interviewed 4 people.

2 black men, 1 hispanic woman, and a dude with a pony tail.

[Wonder Warthog]

"So the exit pollster decided to save money by measuring only the third group, swing voters (by doing exit polling outside of polling places with lots of swing voters). And he found that they were not breaking for Bush enough for Bush to win. He completely missed the GOTV effort and the conservative turnout."

Sorry, but not possible. Voters exiting the polling places don't wear signs saying "Conservative", "Liberal", or "Swing". How is the pollster supposed to know which group they are (suppposedly) in? They certainly wouldn't use any such labels "on themselves".