A close look at the 2002 Alabama Governor’s race suggests that the fraud perpetrated in that election was more obvious than even the 2000 or 2004 U.S. Presidential elections. The final official results were Riley 672,225, Siegelman 669,105 – a difference of 3,120 votes, representing a margin of 0.2% of the total.

It would behoove us to have a good understanding of how this happened, because I have a slight suspicion that we’ll see more of this sort of thing this November, and it should be much easier to prevent if we know what to look for.


Election night, November 5, 2002, Bay Minette, Alabama

Republican controlled Bay Minette is the county seat for Baldwin County, Alabama. In 2002, Baldwin County used optical scan machines to tabulate vote counts from paper ballots filled out by voters and fed into the machines. The paper ballots themselves are saved, which means that they are available for recounting in case of close or contested elections.

The machine tabulated results from each precinct in the country are recorded on individual “data packs”, which are picked up by sheriff deputies after the polls close and delivered to the Bay Minette Board of Elections, which then use a central tabulator to tabulate the county-wide vote count.

The initial vote count for Governor for Baldwin County, reported from the Bay Minette tabulator at 10:45 p.m., was quite surprising to say the least. It reported: Riley (R) 30,142, Siegelman (D) 11,820, and the Libertarian candidate, John Sophocleus, 13,190. Although it was expected that Siegelman would lose Baldwin County, the margin of the loss not believable, as he had lost Baldwin County in the Governor’s race in 1998 by only a little over four thousand votes. Furthermore, the idea of his losing to the Libertarian candidate was not plausible.

So, “someone” from the sheriff’s office went into the tabulation room to look into the matter and returned a few minutes later, announcing that the problem had been fixed. The new totals, which were reported at 11:04 p.m. and picked up and distributed by the AP, were: Riley 31,052, Siegelman 19,070, and Sophocleus a much more reasonable 937. The pickup of 7,250 votes by Siegelman was enough to give him a slim state-wide victory.

But two minutes later, at 11:06 p.m., the results were changed again, reducing Siegelman’s total back down to 12,736, a decrease of 6,334 votes, which gave the election back to Riley. William Pfeifer, the Baldwin County Chairman of the Democratic Party, was just outside the tabulating room at 11:04 when the second report, giving Siegelman the victory, was announced. But he didn’t find out about the reversal until he returned home and turned on the news.


Next morning, November 6, 2002, Bay Minette

The next morning, Pfeifer arrived at the probate court building in an attempt to speak with probate officials to find out what had happened. Pfeifer relates his experience:


The board certified the election results a little after 10:30 a.m., and Riley gave his victory speech around 11:00 a.m.


Failed request for recount

Two days later, Pfeifer petitioned for a hand recount of the Baldwin County ballots. But Alabama Attorney General Bill Pryor ruled later that day that the seals on the boxes containing the ballots could not be broken without a court order to do so. He claimed that his ruling was based on the Alabama Constitution. Don Siegelman contested the ruling and continued to seek a recount, which may have been the reason that he was framed for bribery and sent to prison, as testified to by Dana Jill Simpson:


In an apparently unrelated incident, Bill Pryor was appointed to the 11th Circuit Court of Appeals by George W. Bush during a Congressional recess in 2004.


Statistical anomalies

It is worth noting that when the original computer error was identified, which resulted in adjusting Siegelman’s vote upwards from 11,820 to 19,070 and reducing the Libertarian candidate’s vote downward from an implausible 13,190 to 937, there were also five other races that had to be re-adjusted at the same time. But when the third and final report was issued, the totals in those five other races remained as what they had been adjusted to, while Siegelman’s vote total was the only one that was re-adjusted.

I’ve already noted that Siegelman’s vote total in 2002 in Baldwin County was totally out of line with what would be expected from his performance in 2004.

In addition, James H. Gundlach, a professor of sociology at Auburn University, performed an analysis of the data and concluded that someone with a wireless connection must have changed the tallies. He presented his analysis at the 2003 annual meeting of the Alabama Political Science Association in a paper titled " A Statistical Analysis of Possible Electronic Ballot Box Stuffing”. In that paper Gundlach emphasized the reduction in Siegelman’s vote count from 19,070 to 12,736, saying that such a dramatic decrease is



A discussion of central tabulator mediated election fraud

County central tabulators receive vote counts from all precincts throughout the county. They generally receive the counts electronically by modem, and they receive a whole bunch of physical evidence (tapes from individual voting machines, memory cards, provisional ballots, etc.) as well. The central tabulators tabulate and report the vote counts for the whole county and by precinct, using processes that vary from state to state. These processes can be quite complicated, as indicated by this article from Verified Voting, which explains how people can monitor the tabulation process.

The “pre-tabulator” vote counts for individual precincts are the vote counts that are posted by the individual precincts shortly after poll closing on Election Day. The “post-tabulator” vote counts are the vote counts that are reported out by the county central tabulator, and those are the official counts. For obvious reasons, the pre-tabulator and post-tabulator vote counts should match in a fair election.

There are reasons, I believe, to think that central tabulator mediated fraud is a more practical way to influence a national or state-wide election than is programming vote switching for individual voting machines. Individual voting machines register perhaps one hundred votes per machine. So consider how many individual voting machines would have to be rigged to change the results of a presidential election.

County central tabulators, on the other hand, tabulate the results for a whole county, which in large counties may account for a million or more votes. So you’d have to rig the results of ten thousand individual voting machines to achieve the impact of rigging the results of a single large county central tabulator.

Let’s now consider some examples of likely central tabulator election fraud:

November 2000, Election Day, Volusia County, Florida
The TV networks initially called Florida for George W. Bush based on a “computer glitch” in the central county tabulation of votes in Volusia County, which mysteriously subtracted 16,022 votes from Al Gore’s total. Gore subsequently conceded the election to Bush, but then retracted his concession when the problem was discovered and the votes were given back to Gore, at which time the election was then declared a virtual tie – pending a vote recount that was never completed. So probably the glitch in Volusia County had nothing to do with the final election results. But still, one has to wonder about the reasons for such an error.

November 2004, Election Day, Cleveland, Ohio
The combination of exceptionally long voting lines throughout Cleveland on Election Day 2004 on the one hand, and yet surprisingly low official voter turnout in Cleveland, is very perplexing, especially since Cleveland used punch card voting, which is not subject to the delays that electronic voting tends to cause. That finding alone suggests foul play, since long voting lines should be associated with high voter turnout, not low voter turnout. And since Cleveland is a very heavily democratic city with over three hundred thousand registered voters, the potential for fraud is obvious.

Because I was very suspicious of this I tried to ascertain whether or not the pre-tabulator and post-tabulator vote counts for Cuyahoga County matched. The post-tabulator vote counts were published on the Cuyahoga county web-site, so that part was easy. I then requested the pre-tabulator vote counts from the Director of the Cuyahoga County Board of Elections, Michael Vu. Though Vu repeatedly promised to obtain those for me, he never followed through. So I collaborated with Ray Beckerman’s Ohio Project to conduct an audit to obtain the pre-calculator vote counts. The initial audit of 15 precincts (out of 1458 in Cuyahoga County) showed a net loss to the Kerry/Edwards ticket of 140 votes. However, the audit was never able to be completed.

The other way that the vote count in Cuyahoga County could have been confirmed would have been to conduct a county-wide hand recount of the votes. The rules of the Ohio recount specified that a 3% recount of each county would be conducted, and if any discrepancies were found in the recount (between the pre-tabulator and the official post-tabulator precinct counts), then a county-wide hand recount would be conducted. No vote discrepancies were found in the Cuyahoga County 3% recount. However, many anomalies were observed at the recount, and two election workers were convicted of rigging the vote count.

November 2004, Election Night, Warren County, Ohio
When election officials in Warren County, Ohio, performed the final tabulation of votes for their county, they decided to do so in private, locking out all reporters from observation of the process. Their initial excuse for this was that they didn’t want reporters to interfere with the counting process. Later, they changed that excuse to say that the FBI warned them of a terrorism alert of grade 10 on a 1 to 10 scale. That claim was later denied by the FBI, and county officials refused to name the FBI agent whom they claimed gave them the warning. Several months later I called Erica Solvig, the reporter who broke the story, in an attempt to find out more about what happened. She told me that she wasn’t at liberty to discuss it.


Lessons that Democrats should learn from all this

1) In any close election against a Republican, consider very carefully the possibility of election fraud before conceding.

2) Democrats should push hard to make sure that paper trails are required for all elections, so that the citizens of our country don’t have to be at the mercy of privately owned electronic voting machines that essentially count our votes in secret. And when paper trails are available, there is no reason in the world why, in a democracy, hand recounts shouldn’t be done whenever the results of an election appear suspicious or questionable. Election officials should never be allowed to pick what precincts are counted.

3) Prevention of central tabulator-mediated election fraud :
If county central tabulator fraud is perpetrated, the official post-tabulator precinct counts will not match the pre-tabulator counts, which are calculated at each precinct in the county shortly after poll closing. The post-tabulator counts are easy to identify, since they are the official counts and will be posted on the county Board of Elections web site as soon as the results become official.

The pre-tabulator counts are more difficult to obtain. In my attempt to help confirm fraud in the 2004 presidential election, I tried to obtain pre-tabulator counts in order to compare them with post-tabulator counts. Because of the difficulties I had obtaining those counts I talked with voting rights organizations to ascertain how I could obtain them. I was astounded to hear from them that they had also tried but had rarely been able to obtain the pre-tabulator counts.

Thus, it appears that within weeks or days following the 2004 election, the pre-tabulator vote counts either tended to disappear, or else county boards of elections were generally unhelpful in making them available to enquiring citizens.

But they must be available at the time of poll closing, since each precinct must report them to the county central tabulator. In many jurisdictions, they are required to be publicly posted at each precinct at the time of poll closing. But even if they aren’t posted, there should be no reason why poll watchers couldn’t obtain them. In fact, that is one of the most important tasks of today’s poll watchers.

Identifying substantial mismatches between pre-tabulator and post-tabulator vote counts should signal a high likelihood of election fraud. In any county where that occurs in a close race, automatic hand recounts should be required.