[Previous entry: "Market interference, color perception, and the "Ig Noble Prizes""] [Main Index] [Next entry: "We're Back!!!"]

11/06/2004 Archived Entry: "Did Diebold "deliver the vote" as promised ???"

Gordon P. discusses two reports suggesting that Diebold Inc. did exactly what its CEO Walden O'Dell promised he would do in a 2004-Aug-14 fund-raising letter: "Deliver" the election to GeeDubya Bush...

He writes: Now, here's an interesting claim: "Exit polling accurately predicted the results in most states with very little error. Where there were discrepancies, they were significant in the +5 percent range, and always favored Bush."

Furthermore, all the discrepancies appear to be in states using Diebold's "black box" ballotless, non-auditable, non-recountable, backdoor-infested voting-machines [on which Brad has blogged before]. Almost needless to say, the probability of this happening entirely by random chance in just the ~1/3 of the states controlled by Diebold is essentially negligible: if one naively "Back of the Envelope" estimates the odds against this outcome by noting that the election in most states was a near coin-toss, and assuming similar results in the Diebold-controlled states, the odds against this happening only in the Diebold-controlled states by chance alone are on the order of one chance in 100,000!

A similar result obtains from a "meta-analysis" of the telephone-polling data, which, by combining the results of all of the roughly 1,800 telephone-polls available, estimated a more that 80% likelihood that Kerry should have won the electoral vote!

Now granted, statistical "meta-analysis" is fraught with possibilities for error when misused or presented with biased data --- and the telephone-polls are certainly biased (for example, they don't include the (mostly-younger) people who only have a cellphone and don't have a landline, or persons like me who use Caller-ID, and who refuse to pick up for either unknown numbers or for calls that display "UNAVAILABLE" or "TOLL-FREE CALL") --- but combined with those exit-poll results that suggest a systematic +5% offset in the Diebold-controlled states, it's hard not to suspect that there might indeed be a fire behind all this smoke ... :-(