The Original Post
A thread on the Something Awful forums, here, aimed at accumulating evidence of election fraud and error in the 2004 Presidential elections inspired me to do some detective work of my own. I made use of the Florida Department of State's county-by-county results and compared those results with voting method information gleaned from VerifiedVoting.org's "Verifier" service.
I think that I should state that while I voted for Kerry, this is in no way meant as propoganda or slander against President Bush or the Republican party. This is merely a probe of the results that I believe ought to be done by the major news outlets, but has not yet been.
There has been some brouhaha about the reliability of exit polling, so I'm going to eschew exit poll numbers in my examination and rely only on the official election counts provided by the state of Florida and machine information from VerifiedVoting.org.
Here are the results of the election, statewide:
We can see that Bush garnered a 52/47 victory in Florida.
I assembled the data into something that I could make further analysis of, and found some surprising results. The first thing I was able to realize was that there was a definite split between Bush and Kerry when examined in conjunction with the voting technology on which the vote was cast.
While the touchscreens returned a result that is nearly the reverse of the state's 52/47 Bush victory, precincts using optical scan technology were much more inclined to return results favorable to Bush.
When votes are compared by the manufacturer of the machine on which they were cast, the results are not as dramatic:
Diebold's machines are more favorable to Bush than on average, ES&S's machines nearly mirror the statewide results, and Sequoia's machines returned nearly the reverse of the statewide results, giving a 52/47 victory to Kerry.
The biggest surprise came when I looked at the vote when compared by specific model of machine on which the votes were cast:
While the first three machines deliver a relatively even split, consistent with what has been observed above, the last two machines, the Model 100 and Optech 3P Eagle, delivered a 61/38 and 60/38 Bush victory, respectively. The greatest Bush victory, 77.73/21.9, was achieved on Optech 3P Eagles in Baker county. Oddly, these were the only Optech 3P Eagles manufactured by Sequoia Voting Systems, as opposed to ES&S. The counties utilizing Model 100s and Optech 3P Eagles represented 15.24% of the Floridan vote.
The Model 100 and Optech 3P Eagle are manufactured by ES&S, with the exception of the Optech 3P Eagles in Baker county. ES&S also manufactures the iVotronic, a touch-screen machine. The Model 100 and Optech 3P Eagle are both optical scan machines. ES&S's machines, without a doubt, returned the most favorable results for the President.
I thought it'd be interesting to see what percentage of the state's votes were cast on each machine, and also what percentage of a candidates votes were delivered by each machine:
These results show that while the Model 100 and Optech 3P Eagle showed a great preference to Bush, their impact was not as material as the other three machines. Indeed, if all votes cast on the Model 100 and Optech 3P Eagle were thrown out, Bush would still have won the state with a 50.55/48.62 result.
Many have tried to explain this trend by saying that the optical scan machines were utilized by poorer, more rural areas which tend to favor the President. I investigated this claim by plotting the percent of votes counties cast for Bush and Kerry on the y-axis of a scatter plot, and the number of total votes in that county on the x-axis:
We can see that while smaller counties on the whole vote for President Bush, there are a few exceptions. We can also see that the largest county's results are more favorable than the second or third largest counties. Among all those red lines on the left of the graph we can also see that Kerry's greatest victory was achieved in one of the states smallest counties, Gadsden. Gadsden county used Model 100s manufactured by ES&S.
ES&S has had problems with their machines in the past. I managed to find some thorough voting machine documentation here (PDF). Among ES&S's slipups, these stood out as most relevant to me:
- September 2002, Union County, Florida. A programming error caused ES&S Model 100 machines to read 2,642 Democratic and Republican votes as entirely Republican in the September 2002 election. The ballots program in the memory packs read the ballots incorrectly. The vendor, ES&S, accepted responsibility for the programming error and paid for a hand recount.
- November 2002, Scurry County, Texas. A landslide victory for two commissioner candidates caused poll workers to question the results. The chip in the ES&S 650 contained an incorrect ballot program. ES&S sent a new chip, and the county officials also counted the votes by hand. The opposing candidates actually won by large margins.
- November 2002, Taos, New Mexico. A software programming error caused the Sequoia Optech optical scanner to assign votes to the wrong candidates. Just 25 votes separated the candidates in one race; another race had a 79-vote margin. After noticing that the computer was counting votes under the wrong names, Taos County Clerk Jeannette Rael contacted the programmer of the optical machine and was told it was a programming error.
Union County, the county named in the September 2002 incident, returned a 73/27 Bush victory this year, on ES&S Model 100s. Only 9 of Florida's 67 counties were more decidely in favor of the President.
There is no smoking gun in the Florida elections, but the results still have their pecularities. It's true that if optical scan votes were not counted, Kerry would have won with about the same margin that Bush did. It's true that rural counties tended to use optical scan machines more, and the wideheld view is that rural voters prefer Bush. It's true that machines manufactured by ES&S, a company that has had problems on the past, gave a higher percentage of votes to President Bush than did other machines.
I'd like to compare Florida with Ohio for the last, small part of this examination. Florida and Ohio were both the most important states in this election, and they show some surprising parallels.
First, the statewide results:
We can see that Bush wins, 51/49. Now, an examination of the votes compared by the machine on which they were cast:
Although the chart shows dramatic differences in both the "E-Voting: Touchscreen" and "Optical Scan:Precinct-Based" categories, we can ignore them. Those categories were occupied by only one county each, and are not representative of a trend among many counties.
That said, there is a similar boost for Bush on optical scan machines. The difference is not as dramatic, but is still not representative of the 51/49 Bush victory in Ohio.
Further data is sparse on Ohio, and so examinations of the sort I was able to do with Florida are difficult. I wrote VerifiedVoting.org on the matter:
I'm compiling information on the elections in Ohio and Florida, and am making use of your verifier service, as it is the only source of its kind.
The data on Florida is a boon, but the data on Ohio is sparse on specific model information. Do you know where I might be able to get such information? After scouring ohio.gov I came up empty-handed.
You can view my spreadsheets and charts, categorized by vendor, model, and technology here: http://www.wwu.edu/~adamsb6/vote/ From that page you can choose to download .xls files, or you can click on the webpage link to view an xml web spreadsheet. The web feature does not work in Firefox, if you happen to get an error.
I received this reply:
Hi Brandon - Thanks so much - I know Ohio is thin. Some
states offer easy to find information and lots of it - others none;
in some cases we had to make phone calls to each county to
learn more. In Ohio, we relied for what we did find on a state
voting activist group called CASE-Ohio (www.caseohio.org)
to provide us with what you see there; it was obtained by them
through individual phone calls but in some cases the info was
difficult to obtain. They may have acquired additional details
by now and you might like to contact them through their
website. Much of the state still used punchcards for Nov 2,
having decided NOT to purchase paperless touchscreens in
31 counties after the state passed legislation calling for a
voter-verified paper ballot by 2006. Rather than spend the $$
now and have costly retrofits later, they decided to hold off until
they could get systems that come with a voter-verified paper
If I learn more I will get back in touch with you, and I hope
you will contact me if you obtain additional details also.
I'm glad that Ohio has passed legislation requiring voter-verified paper ballots, but confused by the secrecy. Why would the information be at all difficult to obtain?
Some news stories have emerged on this subject:
Glitch gave Bush extra votes in Ohio
November 5, 2004
COLUMBUS, Ohio (AP) -- An error with an electronic voting system gave President Bush 3,893 extra votes in suburban Columbus, elections officials said.
Franklin County's unofficial results had Bush receiving 4,258 votes to Democrat John Kerry's 260 votes in a precinct in Gahanna. Records show only 638 voters cast ballots in that precinct.
Bush actually received 365 votes in the precinct, Matthew Damschroder, director of the Franklin County Board of Elections, told The Columbus Dispatch.
State and county election officials did not immediately respond to requests by The Associated Press for more details about the voting system and its vendor, and whether the error, if repeated elsewhere in Ohio, could have affected the outcome.
Bush won the state by more than 136,000 votes, according to unofficial results, and Kerry conceded the election on Wednesday after acknowledging that 155,000 provisional ballots yet to be counted in Ohio would not change the result. (Full Ohio results)
The Secretary of State's Office said Friday it could not revise Bush's total until the county reported the error.
The Ohio glitch is among a handful of computer troubles that have emerged since Tuesday's elections. (Touchscreen voting troubles reported)
In one North Carolina county, more than 4,500 votes were lost because officials mistakenly believed a computer that stored ballots electronically could hold more data than it did. And in San Francisco, a malfunction with custom voting software could delay efforts to declare the winners of four races for county supervisor.
In the Ohio precinct in question, the votes are recorded onto a cartridge. On one of the three machines at that precinct, a malfunction occurred in the recording process, Damschroder said. He could not explain how the malfunction occurred.
Damschroder said people who had seen poll results on the election board's Web site called to point out the discrepancy. The error would have been discovered when the official count for the election is performed later this month, he said.
The reader also recorded zero votes in a county commissioner race on the machine.
Workers checked the cartridge against memory banks in the voting machine and each showed that 115 people voted for Bush on that machine. With the other machines, the total for Bush in the precinct added up to 365 votes.
Meanwhile, in San Francisco, a glitch occurred with software designed for the city's new "ranked-choice voting," in which voters list their top three choices for municipal offices. If no candidate gets a majority of first-place votes outright, voters' second and third-place preferences are then distributed among candidates who weren't eliminated in the first round. (E-vote goes smoothly, but experts skeptical)
When the San Francisco Department of Elections tried a test run on Wednesday of the program that does the redistribution, some of the votes didn't get counted and skewed the results, director John Arntz said.
"All the information is there," Arntz said. "It's just not arriving the way it was supposed to."
A technician from the Omaha, Neb. company that designed the software, Election Systems & Software Inc., was working to diagnose and fix the problem.
JACKSONVILLE, N.C. - Several N.C. counties reported problems with voting, including Carteret, where more than 4,000 early votes were lost because the electronic voting system could not store the volume of votes it received.
It was unknown what effect the problem would have on local or state voting results or what action the state Board of Elections might take. Two statewide races, for superintendent of public instruction and agriculture commissioner, remained unresolved and too close to call Thursday.
In Carteret, the county was told by the manufacturer of the voting system that its units could store up to 10,500 votes, but the limit was actually 3,005 votes.
While waiting for the final precinct Tuesday night, election workers began tallying the absentees, no-excuses ballots and Election Day tallies. They found their numbers didn't match the computerized printout.
"That's when we started looking," said Sue Verdon, secretary of the Carteret County Board of Elections.
Local officials said UniLect Corp., the maker of the system, acknowledged it had given the county wrong information about how many votes the system can store.
Verdon said that after a ballot is cast, the electronic vote is stored in one of the units, known as a black box. Because the county thought the capacity was 10,500 votes, only one box was used during early voting.
The loss of the 4,530 votes didn't appear to change the outcome of the county races, but that wasn't the issue for Alecia Williams of Beaufort, who voted on one of the final days of the early voting period.
"The point is not whether the votes would have changed things, it's that they didn't get counted at all," Williams said.
State Board of Elections Executive Director Gary Bartlett said he did not know if the lost votes could somehow be retrieved from the computer equipment. He said there was some discussion of whether the votes could be extracted from the unit's internal drive.
In Mecklenburg County, a discrepancy in unofficial totals posted on election night had officials recounting early voting results Thursday.
The county election office said before the election that 102,109 people voted early or returned valid absentee ballots. But unofficial results from election night showed 106,064 people casting early and absentee votes for president.
The difference was pointed out by county Republicans on Wednesday.
Elections director Michael Dickerson said late Wednesday that he was unsure if there was a problem but acknowledged the situation was unusual. He said election officials had not lost any votes.
"I'm confident in the process we did election night," he said. "I'm never confident [in numbers] until we post the official results."
A recount could change results in the race for at-large seats on the county commission, where unofficial results showed Democrats pulled off a surprising sweep, ousting two Republican commissioners.
According to the unofficial tally, Republican Ruth Samuelson trailed third-place Democrat Jennifer Roberts by 1,921 votes. Republican Dan Ramirez was a distant fifth.
There are three at-large seats on the county commission.
In Craven County, vote totals in nine of the county's 26 precincts were electronically doubled. The outcome of most races isn't expected to be affected.
But in close races, particularly the contest for the Craven County Board of Commissioners District 5 seat where candidates Leon Staton and Anthony Michalek are separated by a mere 35 votes, the outcome could change.
In Onslow County, a software error changed the order of finish in the county commissioners race. The error didn't change who won, just who finished first through fifth.
The error occurred when a floppy disk that compiles voting data from the counting machines was programmed incorrectly, said Kim Strach, a deputy director with the State Board of Elections, who was in Onslow County to assist with the election.
Bartlett said there are an estimated 72,469 provisional ballots still to be counted in the state, with four counties not yet reporting.
Until those ballots are added into Tuesday's unofficial results, it's hard to speculate what impact, if any, voting changes in individual counties might have, officials said.
Congressmen's Letter to Comptroller General
November 5, 2004
The Honorable David M. Walker
Comptroller General of the United States
U.S. General Accountability Office
441 G Street, NW
Washington, DC 20548
Dear Mr. Walker:
We write with an urgent request that the Government Accountability Office immediately undertake an investigation of the efficacy of voting machines and new technologies used in the 2004 election, how election officials responded to difficulties they encountered and what we can do in the future to improve our election systems and administration.
In particular, we are extremely troubled by the following reports, which we would also request that you review and evaluate for us:
In Columbus, Ohio, an electronic voting system gave President Bush nearly 4,000 extra votes. "Machine Error Gives Bush Extra Ohio Votes," Associated Press, November 5.
An electronic tally of a South Florida gambling ballot initiative failed to record thousands of votes. "South Florida OKs Slot Machines Proposal," Id.
In one North Carolina county, more than 4,500 votes were lost because officials mistakenly believed a computer that stored ballots could hold more data that it did. "Machine Error Gives Bush Extra Ohio Votes," Id.
In San Francisco, a glitch occurred with voting machines software that resulted in some votes being left uncounted. Id.
In Florida, there was a substantial drop off in Democratic votes in proportion to voter registration in counties utilizing optical scan machines that was apparently not present in counties using other mechanisms. http://ustogether.org/election04/florida_vote_patt.htm
The House Judiciary Committee Democratic staff has received numerous reports from Youngstown, Ohio that voters who attempted to cast a vote for John Kerry on electronic voting machines saw that their votes were instead recorded as votes for George W. Bush. In South Florida, Congressman Wexler's staff received numerous reports from voters in Palm Beach, Broward and Dade Counties that they attempted to select John Kerry but George Bush appeared on the screen. CNN has reported that a dozen voters in six states, particularly Democrats in Florida, reported similar problems. This was among over one thousand such problems reported. "Touchscreen Voting Problems Reported," Associated Press, November 5.
Excessively long lines were a frequent problem throughout the nation in Democratic precincts, particularly in Florida and Ohio. In one Ohio voting precinct serving students from Kenyon College, some voters were required to wait more than eight hours to vote. "All Eyes on Ohio," Dan Lothian, CNN, November 3, http://www.cnn.com/2004/ALLPOLITICS...log/index.htm..
We are literally receiving additional reports every minute and will transmit additional information as it comes available. The essence of democracy is the confidence of the electorate in the accuracy of voting methods and the fairness of voting procedures. In 2000, that confidence suffered terribly, and we fear that such a blow to our democracy may have occurred in 2004.
Thank you for your prompt attention to this inquiry.
John Conyers, Jr. Jerrold Nadler Robert Wexler
Ranking Member Ranking Member Member of Congress
House Judiciary Committee Subcommittee on the Constitution
View my Excel spreadsheets:
View my Excel spreadsheets on the web (IE or Safari required, d'oh!)