Saturday, November 06, 2004


A brace of important items from our venerated colleague Avedon Carol, who's a one-woman goldmine of information today. The first, from Scoop (New Zealand), is Faun Otter's state-by-state analysis of exit polls vs. "corrected" exit polls vs. official ballot counts:
Method: Grab one site which lists the exit polls before they were "corrected.”

(Correction is the procedure by which the exit polls are retrofitted to match the figures provided by the vote counting machines. It is easily done by changing the exit poll results, such as the 2.00 a.m. flip-flop of the Nevada exit poll scores which was done without any change to the sample size. A slightly less obvious sleight of hand is to alter the weighting. Weighting is the name for a multiplier used to correct sample subgroups to match the proportions in the whole of a state population. Thus an exit poll can be ‘corrected’ by saying something to the effect [of], “Oh well, the vote results show we must have under sampled Republicans and therefore we’ll multiply that subgroup of the exit poll sample by 1.5 to make our results fit the figures the ballot counting machines are spitting out.”)

The Bush people argue that the exit polls are skewed by the methodology employed. It is odd that they don’t say what that error producing part of the methodology might be. A skew means a systematic error is introduced by the test protocol and causes a consistent shift in one direction.
IF this was true, then all the exit polls would show the same sort of shift from 'actual' results.

The GOP offer an alternative argument that the exit polls are not large enough samples and therefore the results are off by a large random error.

IF this was true, then the exit polls should scatter on either side of the actual result, especially if the final result is close to 50/50.

So what do we actually see when comparing exit polls with actual results?

There is skew - but ONLY in states which the Republicans had previously stated to be target states in play. The skew is in the same direction every time; that is to say in favor of Bush.

The exit poll results are not scattered about the mean as the alternative theory predicts.

They are all on the Kerry side of the vote counts as issued by the states except for a hand full of states which hit amazingly close to the exit poll figures.
Go have a look at the figures. Then examine the startling graphic below, which comes from Enemy of the People:
It wasn't the touchscreen, paperless voting that produced the anomolous results. It was the optical scan voting in very small Florida counties that produced an amazing number of huge Bush majorities in counties in which the overwhelming majority of registered voters are democrats.

The [table above] is a list of all the counties in Florida in which there were approximately 20,000 or fewer votes cast. What you see over and over again is that counties with huge majorities of registered democrats voted overwhelmingly republican. If you focus on the percentages in the leftmost column, the "percent change" is the amount by which actual votes cast exceed the vote that would have been projected by assuming that each voter in the county voted in accordance with their party affiliation. A "100%" or higher figure for republicans indicates a huge Bush majority in a county that was overwhelmingly comprised of registered democrats. This phenomenon is simply not observed in the larger counties . . . yet every single county in Florida in which 20,000 or fewer votes were cast exhibited a 90% pro-Bush variance relative to projections based on party affiliation . . . .

Bottom line: since the Bushies have taken control of Florida, we've seen a tidal shift in the small counties to almost Saddam-like republican majorities.
UPDATE (via Scrutiny Hooligans): You will be shocked, shocked by Thom Hartmann's assertion that Florida was gamed:
When I spoke with Jeff Fisher this morning (Saturday, November 06, 2004), the Democratic candidate for the U.S. House of Representatives from Florida's 16th District said he was waiting for the FBI to show up. Fisher has evidence, he says, not only that the Florida election was hacked, but of who hacked it and how. And not just this year, he said, but that these same people had previously hacked the Democratic primary race in 2002 so that Jeb Bush would not have to run against Janet Reno, who presented a real threat to Jeb, but instead against Bill McBride, who Jeb beat.

"It was practice for a national effort," Fisher told me . . . .

[A] few hours after the exit polls were showing a clear Kerry sweep, as the computerized vote numbers began to come in from the various states the election was called for Bush.

How could this happen?

On the CNBC TV show "Topic A With Tina Brown," several months ago, Howard Dean had filled in for Tina Brown as guest host. His guest was Bev Harris, the Seattle grandmother who started www.blackboxvoting.org from her living room. Bev pointed out that regardless of how votes were tabulated (other than hand counts, only done in odd places like small towns in Vermont), the real "counting" is done by computers. Be they Diebold Opti-Scan machines, which read paper ballots filled in by pencil or ink in the voter's hand, or the scanners that read punch cards, or the machines that simply record a touch of the screen, in all cases the final tally is sent to a "central tabulator" machine.

That central tabulator computer is a Windows-based PC.

"In a voting system," Harris explained to Dean on national television, "you have all the different voting machines at all the different polling places, sometimes, as in a county like mine, there's a thousand polling places in a single county. All those machines feed into the one machine so it can add up all the votes. So, of course, if you were going to do something you shouldn't to a voting machine, would it be more convenient to do it to each of the 4000 machines, or just come in here and deal with all of them at once?"

Dean nodded in rhetorical agreement, and Harris continued. "What surprises people is that the central tabulator is just a PC, like what you and I use. It's just a regular computer."

"So," Dean said, "anybody who can hack into a PC can hack into a central tabulator?"

Harris nodded affirmation, and pointed out how Diebold uses a program called GEMS, which fills the screen of the PC and effectively turns it into the central tabulator system. "This is the official program that the County Supervisor sees," she said, pointing to a PC that was sitting between them loaded with Diebold's software.

Bev then had Dean open the GEMS program to see the results of a test election. They went to the screen titled "Election Summary Report" and waited a moment while the PC "adds up all the votes from all the various precincts," and then saw that in this faux election Howard Dean had 1000 votes, Lex Luthor had 500, and Tiger Woods had none. Dean was winning.

"Of course, you can't tamper with this software," Harris noted. Diebold wrote a pretty good program.

But, it's running on a Windows PC.

So Harris had Dean close the Diebold GEMS software, go back to the normal Windows PC desktop, click on the "My Computer" icon, choose "Local Disk C:," open the folder titled GEMS, and open the sub-folder "LocalDB" which, Harris noted, "stands for local database, that's where they keep the votes." Harris then had Dean double-click on a file in that folder titled "Central Tabulator Votes," which caused the PC to open the vote count in a database program like Excel.

In the "Sum of the Candidates" row of numbers, she found that in one precinct Dean had received 800 votes and Lex Luthor had gotten 400.

"Let's just flip those," Harris said, as Dean cut and pasted the numbers from one cell into the other. "And," she added magnanimously, "let's give 100 votes to Tiger."

They closed the database, went back into the official GEMS software "the legitimate way, you're the county supervisor and you're checking on the progress of your election."

As the screen displayed the official voter tabulation, Harris said, "And you can see now that Howard Dean has only 500 votes, Lex Luthor has 900, and Tiger Woods has 100." Dean, the winner, was now the loser.
Watch streaming video of Bev Harris and Howard Dean here.

UPDATE II (via Corrente): From the archives, an article by Mark Crispin Miller (author of Cruel and Unusual and The Bush Dyslexicon) examining certain statistical irregularities from California's recall election last spring.

| | Technorati Links | to Del.icio.us