Wednesday, November 22, 2006

Back to Tinfoil World

I never intended to desert it. That the Democrats won in the last election does NOT mean that we don't now need election reform. We do need it, every bit as much as if the result had been reversed, and the reason is here:

Randy Wooten, owner of Randy's Karaoke Bar out on Highway 14 and one of three mayoral candidates, received no votes. Not even his own.

''He voted for himself. I watched him,'' said Roxanne Wooten, the candidate's wife. ''I was standing right behind him. And then I voted for him,'' she told me Monday.

Neither vote registered on the ES&S machine. Poinsett County election official insisted that their touch-screen machines were working just fine, that Waldenburg's improbable undervote was due to ``operator error.''

Not hardly, said Roxanne Wooten. She said, ``I noticed that the machine was acting jumpy.''

But she made sure she voted for her husband. It was the one vote that mattered.

Randy's two opponents split 36 votes. He came up empty.

It's not the machine error that bothers me. It's the question what can be done now to verify or falsify the minimum of two votes Randy should have gotten, and, as far as I can learn, nothing can be done. This is no way to run transparent elections.

Then there is the whole exit polls comedy. In other countries exit polls are used to see how fair the elections might be. In this country some people try to suppress exit polls altogether, and very few in the media point out this paradox. I'm also still reeling after trying to understand why exit polls are so rapidly adjusted to match final voting percentages of various types, without us also still seeing the raw data. Very hasty, those adjustments are.

And this brings me to the tinfoil land. The following quote is by a representative of the Election Defense Alliance:

Simon, surprised that unadjusted polling data was publicly revealed, given the concerns after the 2004 election about the use of exit polls, downloaded as much of the data as he could in real time. Scheduled and planned revisions on the CNN site took place throughout the evening and by the following morning, the unadjusted exit poll data had been replaced with data that conformed with the reported, official vote totals. This was the planned procedure as indicated by the NEP's methodology.

Adjusting the exit poll data is, by itself, not a troublesome act. Simon explained, "Their advertised reason to do the exit polls is to enable analysis of the results by academic researchers--they study the election dynamics and demographics so they can understand which demographic groups voted what ways. As an analytic tool, the exit poll is considered more serviceable if it matches the vote count. Since the vote count is assumed to be gospel, congruence with that count is therefore assumed to give the most accurate picture of the behavior of the electorate and its subgroups.

"In 2004 they had to weight it very heavily, to the point that the party turnout was 37% Democrat and 37% Republican, which has never been the case--leading to the claim that Rove turned out the Republican vote. This was nowhere witnessed, no lines in Republican voting places were reported. As ridiculous as that was, the distortion of actual turnout was even greater in 2006. The adjusted poll's sample, to match the vote count, had to consist of 49% 2004 Bush voters and only 43% 2004 Kerry voters, more than twice the actual margin of 2.8%. This may not seem like that much, but it translates into more than a 3,000,000 vote shift nationwide, which, depending on targeting, was enough to have altered the outcome of dozens of federal races.

"It should be very clear that weighting by a variety of carefully selected demographic categories, which yields the pre-adjustment exit polls, presents a truly representative electorate by every available standard except the vote count in the present election. So you have a choice: you can believe in an electorate composed of the correct proportions of men and women, young and old, rural and urban, ethnic and income groups, Democrats, Republicans, and Independents--or you can believe the machines. Anyone who has ever wondered what is really in a hot dog should be aware that the machines are designed, programmed, deployed, and serviced by avowedly partisan vendors, and can easily be set up to generate entirely false counts with no one the wiser, least of all the voters."

Simon concluded, "These machines are completely and utterly black box. The idea that we have this enormous burden of proof that they are miscounting, and there's no burden of proof that they are counting accurately--that, first and foremost, has to change."

I bolded out the part that matters for tinfoilery. What Simon talks about there is the fact that the exit polls in 2006 had a question about the voter's 2004 presidential election choices (assuming that the voter had voted in 2004), and this question allows us to see something very odd about the whole charade:

The forcing process in this instance reveals a great deal. The Party affiliation of the respondents in the original 7:07 p.m. election night Exit Poll closely reflected the 2004 Bush-Kerry election margin. After the forcing process, 49-percent of respondents reported voting for Republican George W. Bush in 2004, while only 43-percent reported voting for Democrat John Kerry. This 6-percent gap is more than twice the size of the actual 2004 Bush margin of 2.8 percent, and a clear distortion of the 2006 electorate.

There is a significant over-sampling of Republican voters in the adjusted 2006 Exit Poll. It simply does not reflect the actual turnout on Election Day 2006.

Think about this a little. It could be that for some odd reason the exit polls have, once again, somehow failed to interview enough Republicans all over the place. Or it could be that Republicans lie in exit polls in an effort to look like Democrats. But why would either of these be true? Nobody has yet been able to explain this to my satisfaction. And nobody has explained how this odd problem could be fixed.

There is a psychological argument that people tend to misremember their previous election choices in a way which makes them look as if they voted for the winner in that election, and that might explain the adjusted results, assuming that the recorded votes are correct. But it doesn't explain why we wouldn't have gotten the same misremembering in the exit polls.

Then the possibility that the election results showed a much greater Republican turnout than expected. Well, this would have worked if the Republicans had won the elections, but it doesn't really explain what actually happened at all.

Now, there might be some obvious and simple explanation for all this, and I hope there is, even though it will make me look like an idiot. But better that than the alternatives.