Wednesday, May 25, 2005

Lampley Won't Throw in the Towel
Argh. This guy just won't quit .

I could go on at length here about the curious disconnect between the actual data in the report and its guesswork conclusion, how Edison/Mitofsky systematically validate all their sampling choices and their methodology, in effect eliminating any logical underpinnings for their ultimate summation, all the while selectively ignoring the lopsided skewing of pro-Bush discrepancies in the most critical swing states. I could spend some time dissecting what I believe is an obvious whitewash, a delicate sidestep away from the potential public relations disaster of being tied forever to the most notorious election theft in history.

Lampley clearly doesn't understand what he's talking about. The Mitofsky report doesn't undermine the conclusions they draw because the bias they suggest occurred wasn't a result of their methodology, Jim. It was *respondent* induced bias. As for the error occuring more in the swing-states, well, that doesn't contradict the Mitofsky thesis but rather fits in quite nicely. It hardly requires a stretch of the imagination to conclude that Bush voters might have been more relectant to participate in voting sites where there were significant numbers of Kerry voters. But this is giving this suggestion more credence than it deserves. There are a host of possibile problems that could lead to more significant error in the 'swing' states that do not involve fraud.

But none of that is necessary, because the entire Edison/Mitofsky report is irrelevant to the argument, given that it is based on the assumption the final official vote tally is accurate.

I'll leave aside, for the moment, the fact that Lampley is now calling Mitofsky irrelevant when he was citing him in the first place. Just to get this straight: Lampley thinks that Mitofsky got his exit polls right. But rather than claim vindication, Mitofsky has decided to disavow his own exit poll results to 'whitewash' things for the Bush administration. Anyone who knows Mitofsky's political leanings is probably laughing their arse off at the moment. For the rest of us, we can only marvel at Lampley's confidence that someone would trash their own accurate work for the benefit of being ridiculed for having got the results wrong.

Make no mistake: my argument is that the final official vote tally is anything but accurate, that it is the product of massive vote fraud carried out through the programing of Diebold voting machines and various other machinations aimed at suppressing, destroying or losing Kerry votes.

The problem with this is that the allegations have been investigated by both Congress and the media and they have found nothing. For a good round up of the various failed allegations of vote fraud, see here (must sit through 1 ad to view if not Salon subscriber). And, as John Cole notes no Diebold voting machines were used in Ohio as can be seen here. Whoops.

My argument is that what were accurate were the exit polls. As one Ivy League research methodologist has noted, "Apparently the pollsters at Mitofsky and Edison have found it more expedient to provide an explanation unsupported by theory, data or precedent than to impugn the machinery of American democracy."

Lampley cites a couple of partisan's who have made a rather poor effort to refute the Mitofsky thesis on the problems with the exit polling. In their article, they make wildly false claims. They claim that a 5% error is a 'statistical impossibility.' This is false. It is certainly improbable ASSUMING that the sample is free from bias. Error estimates are made regarding *random* variation...not systematic variation as a result of some bias. These same fellows offer the same false dichotomy Lampley does (Lampley probably got it reading them). And it is just as wrong comming from them as it was from Jim. While they chastise Mitsofsky & Edison for unsubstantiated speculation, these authors engage in orders of magnitute worse speculation (i.e. massive election fraud) without any corroborating evidence. Indeed, they just make a circular argument. We know the exit polls were accurate because the election was cleary stolen (wrong in swing states), and we know the election was stolen because the exit polls were accurate (their alleged corrobrating evidence). The problem is that the exit polls (properly scrubbed post-election) PREDICTED A BUSH ELECTORAL COLLEGE VICTORY. While Bush earns larger tallies above that of the margin of error reported for the exit poll, the exit poll results still indicate that Bush would have won.

This statistician waves his hands at the thousands of 'reported irregularies' as if that counts as substantiation. None of these allegations have been supported with any credible hard evidence. In reality, these authors provide no compelling refutation of the Mitofsky thesis. They provide no credible evidence of fraud. And their statistical argument fails utterly.

Various statisticians have reported that the odds on the occurrence of variances from exit polls to actual results such as were produced in this election range up to 959 000 to 1. Sounds like DNA. As US Count Votes notes in a statistical abstract, "No matter how one calculates it, the discrepancy cannot be attributed to chance."

I don't know who the various statisticans are. Lampley only cites one. And that guy plainly makes misleading statements. Rare events *do* occur. An accurate statement would argue that ceterus paribus, the exit polls results were unlikely to be due to *random* error. But Mitofsky doesn't attribute the problem to random error. Hence this point is entirely beside the point. The vulnerabilities of any survey to systematic error are significant. We don't toss actual vote tabulations in favor of a multi-state massive vote fraud conspiracy because the survey results were off. Nuff said. D.GOOCH


Post a Comment

<< Home