As Raw Story originally reported on Wednesday, the Exit Polling on the day of the Election showed Kerry ahead by robust margins in many states. By the time the day had ended Kerry's margin had been eclipsed by Bush's. But, notably, that only happened in states that do not have fully voter-verifiable paper trails (e.g. E-Voting machines that don't offer printed/verifiable paper receipts for the voter to confirm that their vote was recorded correctly).
Here are a couple of rudimentary graphics now to show the difference.
In the 10 states polled where there were voter-verifiable paper trails, the Final Results barely differed from the Exit Polls (which, until this year, have been widely regarded as highly accurate since they are taken as folks actually leave the polling place). The average change from Exit Poll to Final Results in these states was a tiny .8% gain for Bush. (Exit Numbers displayed in BLUE, Final Numbers in RED):
And now look at the data from states where there are no paper trails. Note the huge difference in Final Results that changed radically from the Exit Polling results. The average gain for Bush --- in these states only --- was 6.25% over the Exit Polling numbers!
What could possibly account for the huge differences there between the Exit Polls and the Final Results? And only in the Non-Paper Ballot States?
The numbers shown in these graphs seem to be taken from the "4pm Exit Polls" as shown on this chart. That chart is not particular user-friendly, but with a bit of close study, it makes sense.
Those "4pm Exit Poll" numbers however, match up with the "Final 6pm Exit Polls" as reported by Raw Story on the day of. So I'm not sure if the person who compiled the chart and the above graphics was in a different time-zone to account for the 4pm/6pm discrepancy. It doesn't really matter to be frank, as long as the numbers were correctly reported. Which they seem to have been.
It's my understanding that a major news organization is investigating this part of the story now. So we'll see what can be learned in the days ahead.
UPDATE: 11/9/04 Raw Story asked a former MIT "Associate Professor of Mathmatics" to look at these results, and his conclusion was that the probabilities of the Exit Polls all closing for Bush the way they did was approximately "1 in 50,000 or .002 percent". Nonetheless, that same Professor has looked closely at the available Exit Polling data and has "ruled out any significance of a variance between electronic balloting and paper ballot states" in the matter. That story is here.
I have spoken to Raw Story editor, John Byrne about this, and he stands by the report though he adds that to get any conclusive information on the matter we'd have to now take a look at the Exit Polling data on a county-by-county level. At this time, that information has not been released to the public, though apparently it is for sale somewhere. Not sure at this time how much it would cost to purchase for further analysis (and not sure why, at this point, it isn't simply made available to the public given all the troubling reports about this election!)