Wednesday, November 10, 2004

David Cobb pushing for recount in Ohio

I've looked through the official results for the state of Florida and have found some a strange fact that no one else has either noticed or attempted to offer an explanation for. That fact is that voters who turned out but did not cast a vote for President are clearly more numerous on some machines, groups, and counties than others.

This anomaly becomes apparent even when sorting the vote by voting technology used, which only has two categories. I'll be referring to ballots cast without Presidential votes as "nonvotes." Let me know if there is a better word to describe them.

In the graph above, we can see that while touchscreen voting machines contributed 54% of the state's vote, they managed to contribute a much greater percentage of the state's nonvotes, 61%. There is more than a seven point spread between votes and nonvotes contributed. We can also see that optical scan voting machines performed in the reverse, contributing a lower percentage of the state's nonvotes than it contributed to the state's turnout.

In this graph we witness a distinct difference between the amount of voter turnout and nonvotes that Diebold's machines contributed to the statewide totals. While Diebold contributed more than 30% of the state's total turnout, it only contributed about 19% of the state's total nonvotes. Diebold's percentage of nonvotes contributed was only about 60% of what its total turnout contributed was. Diebold is the only manufacturer whose machines contributed less to the amount of nonvotes than to turnout.

The other two manufacturers, ES&S and Sequoia, exhibited a difference between themselves, as well. ES&S's percent nonvotes contributed to the state total was approximately 120% of its percent turnout contributed. Sequoia's nonvote contribution was about 112% of its percent turnout contribution.

Another calculation shows that 0.24% of the ballots Diebold counted included no vote for President. ES&S's and Sequoia's machines respectively returned 0.48% and 0.45% of ballots with no Presidential vote. This data is shown below, and is in my Excel spreadsheet, as well.

Vendor Vendor Turnout Sub-Total Vendor Nonvote Sub-Total Vendor's % Statewide Turnout Vendor's % Statewide Nonvotes Vendor's % Statewide Turnout - % Statewide Nonvotes
Diebold Election Systems 2,354,808 5,666 30.82% 18.57% 12.25%

Percent Nonvotes 0.24%

Election Systems and Software, Inc. (ES&S) 3,743,986 17,888 49.00% 58.63% -9.63%

Percent Nonvotes 0.48%

Sequoia Voting Systems, Inc. 1,541,525 6,955 20.18% 22.80% -2.62%

Percent Nonvotes 0.45%

The last comparison I was able to make of this kind was nonvotes and turnout categorized by machine model.

This graph shows us what we already knew in the second graph. The Accuvote-ES2001B is the only machine manufactured by Diebold in the state, and so its percent nonvotes and percent turnout contributed to the state totals will be the same as Diebold's numbers in the vendor comparison.

The new information contained in this graph is that behavior is mostly uniform among machines of the other two manufacturers. The AVC Edge, iVotronic, Model 100, and Optech 3P Eagle all contributed a higher percentage of the state's total nonvotes than they did to the state's total turnout.

More calculations reveal that ES&S's optical scan machines, the Model 100 and Optech 3P Eagle, were the only machines that counted more than 0.5% of their ballots as not having a vote for President. This data is below.

Model Model Turnout Sub-Total Model Nonvote Sub-Total Model's % Statewide Turnout Model's % Statewide Nonvotes Model's % Statewide Turnout - % Statewide Nonvotes Voting Tech's % Statewide Nonvotes as a % of Statewide Turnout
AccuVote-ES2001B 2,354,808 5,666 30.82% 18.57% 12.25% 60.26%

Percent Nonvotes 0.24%

AVC Edge 1,531,477 6,862 20.04% 22.49% -2.45% 112.21%

Percent Nonvotes 0.45%

iVotronic 2,590,145 11,855 33.90% 38.86% -4.96% 114.62%

Percent Nonvotes 0.46%

Model 100 364,622 1,894 4.77% 6.21% -1.44% 130.08%

Percent Nonvotes 0.52%

Optech 3P Eagle 799,267 4,232 10.46% 13.87% -3.41% 132.60%

Percent Nonvotes 0.53%

This is a list of some facts that I did not think needed to be supported by charts and graphs:

  • The ten counties with the lowest percentage of nonvotes made use of Diebold Accuvote machines. All but one, Monroe County, voted for Bush. Monroe was Kerry's second narrowest win in Florida, as he edged out the President by only 0.47%.
  • The ten counties with the highest percentage of nonvotes used optical scan machines. Eight were manufactured by ES&S, one by Diebold, and one by Sequoia. The machine manufactured by Sequioa, the Optech 3P Eagle, is the only such machine manufactured by Sequoia. All other Optech 3P Eagles in Florida were manufactured by ES&S. In these counties are both Bush's second greatest win and Kerry's narrowest victory. Bush won 78/22 in Baker county, which utilized the sole Optech 3P Eagle made by Sequoia and reported 0.93% of ballots as not having a Presidential vote. Kerry won by 0.21% in Orange county, which used Optech 3P Eagles manufactured by ES&S and 0.68% of ballots cast as not having a vote for President.

There are two more peculiarities that don't require much complex calculation at all to recognize.

The first is that Osceola county's turnout and Presidential votes are exactly the same. Osceola had 0 nonvotes. I thought perhaps that Osceola was a smaller county, as that could be a possible explanation for not even one person not casting a vote for President on their ballot, but it's not. Osceola county, in which 82,204 ballots were cast, accounted for 1.08% of Florida's total turnout. This is not a small number - the average (median) county in Florida accounted for only 0.55% of Florida's total turnout.

Osceola was the sole county whose turnout figures were updated on November 8 that still had a a turnout that was lower than the number of votes it reported cast for President. On that update, 82,178 votes were cast and 81,917 voters were reported to have turned out. 261 more votes were cast than voters were reported to have turned out. Osceola used Diebold Accuvote machines for its elections.

The second oddity is Volusia county. Volusia has the second smallest difference between its turnout and vote numbers, exceeded only by Osceola's difference of 0. 254 of the 229,193 ballots cast in Volusia county, 0.11%, were reported not to have included a vote for President. On its own, this would not be too strange, but Bev Harris of just recently updated her site with an account of her experience in Volusia. Here is that account:

TUESDAY NOV 16 2004: Volusia County on lockdown

County election records just got put on lockdown

Dueling lawyers, election officials gnashing teeth, film crew catching it all.

Here's what happened so far:

Friday Black Box Voting investigators Andy Stephenson and Kathleen Wynne popped in to ask for some records. They were rebuffed by an elections official named Denise. Bev Harris called on the cell phone from investigations in downstate Florida, and told Volusia County Elections Supervisor Deanie Lowe that Black Box Voting would be in to pick up the Nov. 2 Freedom of Information request, or would file for a hand recount. "No, Bev, please don't do that!" Lowe exclaimed. But this is the way it has to be, folks. Black Box Voting didn't back down.

Monday Bev, Andy and Kathleen came in with a film crew and asked for the FOIA request. Deanie Lowe gave it over with a smile, but Harris noticed that one item, the polling place tapes, were not copies of the real ones, but instead were new printouts, done on Nov. 15, and not signed by anyone.

Harris asked to see the real ones, and they said for "privacy" reasons they can't make copies of the signed ones. She insisted on at least viewing them (although refusing to give copies of the signatures is not legally defensible, according to Berkeley elections attorney, Lowell Finley). They said the real ones were in the County Elections warehouse. It was quittin' time and an arrangment was made to come back this morning to review them.

Lana Hires, a Volusia County employee who gained some notoriety in an election 2000 Diebold memo, where she asked for an explanation of minus 16,022 votes for Gore, so she wouldn't have to stand there "looking dumb" when the auditor came in, was particularly unhappy about seeing the Black Box Voting investigators in the office. She vigorously shook her head when Deanie Lowe suggested going to the warehouse.

Kathleen Wynne and Bev Harris showed up at the warehouse at 8:15 Tuesday morning, Nov. 16. There was Lana Hires looking especially gruff, yet surprised. She ordered them out. Well, they couldn't see why because there she was, with a couple other people, handling the original poll tapes. You know, the ones with the signatures on them. Harris and Wynne stepped out and Volusia County officials promptly shut the door.

There was a trash bag on the porch outside the door. Harris looked into it and what do you know, but there were poll tapes in there. They came out and glared at Harris and Wynne, who drove away a small bit, and then videotaped the license plates of the two vehicles marked 'City Council' member. Others came out to glare and soon all doors were slammed.

So, Harris and Wynne went and parked behind a bus to see what they would do next. They pulled out some large pylons, which blocked the door. Harris decided to go look at the garbage some more while Wynne videotaped. A man who identified himself as "Pete" came out and Harris immediately wrote a public records request for the contents of the garbage bag, which also contained ballots -- real ones, but not filled out.

A brief tug of war occurred, tearing the garbage bag open. Harris and Wynne then looked through it, as Pete looked on. He was quite friendly.

Black Box Voting collected various poll tapes and other information and asked if they could copy it, for the public records request. "You won't be going anywhere," said Pete. "The deputy is on his way."

Yes, not one but two police cars came up and then two county elections officials, and everyone stood around discussing the merits of the "black bag" public records request.

The police finally let Harris and Wynne go, about the time the film crew arrived, and everyone trooped off to the elections office. There, the plot thickened.

Black Box Voting began to compare the special printouts given in the FOIA request with the signed polling tapes from election night. Lo and behold, some were missing. By this time, Black Box Voting investigator Andy Stephenson had joined the group at Volusia County. Some polling place tapes didn't match. In fact, in one location, precinct 215, an African-American precinct, the votes were off by hundreds, in favor of George W. Bush and other Republicans.

Hmm. Which was right? The polling tape Volusia gave to Black Box Voting, specially printed on Nov. 15, without signatures, or the ones with signatures, printed on Nov. 2, with up to 8 signatures per tape?

Well, then it became even more interesting. A Volusia employee boxed up some items from an office containing Lana Hires' desk, which appeared to contain -- you guessed it -- polling place tapes. The employee took them to the back of the building and disappeared.

Then, Ellen B., a voting integrity advocate from Broward County, Florida, and Susan, from Volusia, decided now would be a good time to go through the trash at the elections office. Lo and behold, they found all kinds of memos and some polling place tapes, fresh from Volusia elections office.

So, Black Box Voting compared these with the Nov. 2 signed ones and the "special' ones from Nov. 15 given, unsigned, finding several of the MISSING poll tapes. There they were: In the garbage.

So, Wynne went to the car and got the polling place tapes she had pulled from the warehouse garbage. My my my. There were not only discrepancies, but a polling place tape that was signed by six officials.

This was a bit disturbing, since the employees there had said that bag was destined for the shredder.

By now, a county lawyer had appeared on the scene, suddenly threatening to charge Black Box Voting extra for the time spent looking at the real stuff Volusia had withheld earlier. Other lawyers appeared, phoned, people had meetings, Lana glowered at everyone, and someone shut the door in the office holding the GEMS server.

Black Box Voting investigator Andy Stephenson then went to get the Diebold "GEMS" central server locked down. He also got the memory cards locked down and secured, much to the dismay of Lana. They were scattered around unsecured in any way before that.

Everyone agreed to convene tomorrow morning, to further audit, discuss the hand count that Black Box Voting will require of Volusia County, and of course, it is time to talk about contesting the election in Volusia.

# # # # #

I don't endorse Bev Harris's account as the truth, because it is rather sensationalist and accusatory. It did cause my interest to pique when I saw that Volusia county was so far from the norm.

I believe that what I've shown here furthers the argument that there are unexplainable differences between optical scan and touch screen voting machines. Prior evidence invoked Bush and Kerry, and could be easily explained away by stating that poorer, more rural counties tend to utilize optical scan machines, and those rural communities favor conservatives like Bush. I don't know of any evidence that rural voters are less likely to fill out a ballot and not vote for President than any other voters, though.

You may download my Excel spreadsheets by clicking here.

You may view my Excel spreadsheets online (IE or Safari required) here.


Post a Comment

<< Home