Reply #73: it's sort of ironic [View All]

As I said, it's hard to be sure what Freeman is claiming, but I don't think he has ever bluntly conceded any likelihood of non-sampling error in the exit poll. In October he said that "the numbers indicate... a discrepancy on the order of 10 million votes." Hey, a million here, a million there....

The irony is that one encounters fraudsters who claim that E/M has tried to conceal and to downplay the exit poll discrepancy. But if E/M had never released these mean WPEs by state, the fraudsters would still be working with much smaller discrepancy figures. (Not only would they be using model estimates -- well, actually, crude approximations of the model estimates -- but those model estimates would incorporate pre-election polling or vote counts. Just referring to the math on exitpollz.org and taking the average margin, it would be Kerry +4.1 in Ohio instead of +8.8. New Hampshire, Kerry +11.5 instead of +15.0. In California, I think Kerry +11.4 instead of +20.8.)

Or think of the trees that might have been saved if E/M had just fudged the completion rates in the precinct-by-partisanship table, to spare themselves the "higher rates in Bush strongholds" shtick. But really, I'm surprised that anyone ever took that argument seriously, and amazed that anyone still does. I guess out of 300 million people, you can find someone to argue just about anything.