Hi Thump, WOW! Fascinating data! OK, *this* is new.... Now, I'm a statistics sort of person (live and breathe the stuff), so I put these figures to a quick t-test to see whether they could possibly be due to chance alone. These figures show exit polling results vs. "actual" vote count for 10 paper ballot states and 8 non-paper ballot states. Among the paper ballot states, the differential between exit count and final count averaged 0.2% +/- 1.5% (better outcome than estimated for Kerry), while the differential for the non-paper-ballot states averaged -6.0% +/- 5.0% (worse outcome than estimated for Kerry). Using the most conservative possible statistical assumptions (and I am generally a rather conservative statitician) of 2-tails and a heteroscedastic model, I find that the probability is less than 1% that these results are due to chance alone. Breaking this down further, there is greater than a 50% chance that the differential in paper-balloting states is random and neither helped nor hurt Kerry. However, there is less than a 1% probability that non-paper balloting didn hurt Kerry. Scientists base their conclusions on a convention of 5% or less. So any scientist would conclude that non-paper balloting reliably hurt Kerry and, further, that there would be some reason that this balloting came out hurting him. <insert election fraud theories> As I said, I'm pretty statistically conservative. (You Republicans should like that about me! ;-) So I pose the following questions to you, Thump. They are pretty important, and they represent the last remaining assumptions of a valid test: (1) Obviously not all 50 states are represented in this sample. On the surface, they do not appear "loaded" to support an argument of universal election fraud, as the electronic vote in Colorado came out 1% better for Kerry than exit polling predicted. However, I still wonder. Is there some reason the remaining 32 states were not shown? What criteria were used to determine which states were included in the sample and which weren't? (2) Were the exit polls done the same way for all the states sampled? Also, whose exit polls are these? Oh, in case y'all are interested, I played around with the above data, with the assumption that 10 out of 18 states do paper ballots and 8 out of 18 do non-paper. If the outcome were determined by popular vote, Kerry would have had around a 90% chance of winning, if the little problem with non-paper ballots were corrected (according to these figures, assuming a 1.75% standard error of the mean and therefore a minimum of a 3.4% mean non-paper-ballot error in Bush's favor). If I'm not mistaken, a correction of the vote from a few of the states in the graphs would have resulted in an electoral victory for Kerry. Hmmmmm.... Of course this is only a rough estimate, but my tentative conclusion would be that George Bush is the only 2-term US president never to have been elected. That'll be a great Trivial Pursuit question in the year 2052 (a presidential election year, assuming our government lasts that long). In follow-up, I had a long chat with a very, VERY radically conservative friend, and she is of the opinion that the Democrats have carved out a much closer election outcome than they would have gotten had they not cheated (!!). Of course this assertion makes me chuckle, but it brings one thing to light for me. Neither side trusts the other not to cheat. I think that means this non-paper, non-recountable, electronic mess is a nightmare for both sides. We should ALL be screaming to the government either to FIX it (i.e. with full paper acountability) or TRASH it. Even if you don't buy my conclusions, you should at least be concerned that non-paper balloting creates almost four times the "error," irrespective of who benefits from it. Peace, Sarah
This PTG archive page provided courtesy of Moy Piano Service, LLC