Huge discrepancies between exit polls and actual votes in Florida.

Exit polls showed Kerry ahead by as much as 11% but actual votes yielded 52% Bush, 47% Kerry. So, these polls were off by 16%.

Exit polls are usually the most accurate of all polls. This is the first time such a discrepancy has ever been registered. Statisticians have got to get back to work to explain how their sampling techniques could have been that wrong, that one time and in that State.

Now, is it possible that people thought they voted for one candidate while the machines attached their votes to a different candidate? But there seems to be a confirm page that comes up before a voter submits the vote. Is it possible that some people didn't check that last page and committed too quickly? Can anyone think of any other scenarios, short of mischief?

According to one report here exit polls showed fewer hispanic voters in florida than in 2000 when there were more and in another state they reported that 63% of the electors were female which any 5 year old could tell you was odd. Exit polls are always unreliable if people don;t want to say they voted for somene (eg Bush) and frankly all US polls look pretty poor given the way they are all over the place. Look at the UK election next year - they usually get it right though our exit polls also miss a lot of conservative voters.