The Evidence for Ballot Tampering in Escambia County, Florida

PDF

Home

An examination of data collected by the Miami Herald(1) during its examination of overvoted ballots provides convincing evidence of fraud in the counting of absentee ballots in Escambia County. More than 21,500 absentee ballots were cast in Escambia, of which the Herald identified 296 ballots as overvoted. All these overvoted ballots contained three or more marks(2) in the presidential race. Not one ballot contained marks for exactly two presidential candidates. The odds against this result — no overvoted ballots with exactly two marks — are so high that “astronomical” would be an understatement.

Optically scanned ballots with distinct marks for two different candidates are a common occurrence. Such overvotes occur less frequently on absentee ballots than on ballots cast at the polls — but only when there is no “voter protection” at the polls. Various factors can influence the rate at which such ballots are cast, including ballot design (the rates at which overvotes are cast, and “double-marked” ballots occur, are significantly higher in those counties that split their presidential race between two columns than in the counties where all the candidates were listed in one column) and the make-up of the absentee pool of voters.

Even allowing for such factors, the complete lack of double-marked overvotes in the Herald data from Escambia County’s absentee votes should raise questions that warrant an investigation into possible ballot tampering. In most counties, there are more double-marked overvotes than overvotes with three or more marks. No other county even comes close to the results found in Escambia County, where overvoted absentee ballots were exclusively marked three or more times. The odds against this result occurring naturally are so prohibitive that, unless the Herald data is completely erroneous, fraudulent tampering with the ballots is the only explanation.

According to the Orlando Sentinel(3), more than 10,000 ballots were duplicated in at least 26 Florida counties because they were “damaged or defective” (as permitted by law).(4) These duplicates included ballots that contained overvotes and undervotes where the voter’s intent was determined. All but a handful were absentee ballots duplicated in Republican-dominated counties where votes cast at the polls were scanned and tabulated in individual precincts.

Nearly one quarter (2,400) of these duplicated ballots came from Escambia County alone, where duplicated ballots represented more than 11% of the absentee ballots cast.(5) According to the Sentinel, “Escambia’s ‘duplicating team’ of more than a dozen poll workers went to great lengths — working until 2 a.m. — to make sure their absentee voters got a second or third look to have a mistaken ballot corrected and duplicated.”

The Sentinel also states that the duplication of ballots was done “with no outside scrutiny,” even though Florida law requires that duplicating ballots must be done “in the presence of witnesses.”

Other Florida counties show results that warrant investigation. In both Bay and Santa Rosa Counties, for example, the percentage of absentee ballots that were overvoted is so much smaller than average that ballot tampering must be investigated both counties. At least two other counties (Flagler and Seminole) have very low rates of double-marked overvotes that also should be investigated.

Escambia’s absentee overvote rate, in itself, is not low enough to raise suspicion; the absolute lack of any double-marked ballots is what is so noteworthy.

In the discussion of what occurred in Escambia County, these terms are defined as follows:

• Mark — Marks, mostly made with a pen or pencil, that a scanner would detect. They may be deliberate marks for a listed candidate; they may denote the presence of a possibly valid write-in vote; or they may be erasures and stray marks that have been erroneously identified by the machine as “votes.”
• Double-marked — A ballot that is overvoted with exactly two marks.
• Multiple-marked — A ballot that is overvoted with three or more marks.
• Total overvotes — The total number of ballots that the machine read as overvotes (“double” plus “multiple”).
• Recovered votes — Ballots on which the Miami Herald was able to determine “voter intent”; not ballots duplicated and hand-counted by the counties.
• Unrecovered double — The total double-marked ballots minus the recovered votes; the minimum number of double-marked ballots for which the Herald was unable to determine voter intent. This is a minimum, not an absolute number, because some of the recovered ballots may have had three or more marks. This number is included to contrast the numbers for Escambia and the other counties — to show that Escambia County is the only county that had no double-voted ballots.
• OC — Optically scanned ballots that were tabulated by machine in a central location.
• OCH — Optically scanned ballots that were tabulated by machine in a central location, and subsequently hand-counted.
• OP — Optically scanned ballots tabulated at the precinct where they were cast.

• Table 2, which provides totals for votes cast and breaks them down by absentee and poll cast votes, includes only votes cast for Bush or Gore, not for other presidential candidates. (6)

• In Table 3, for those counties/categories where the Herald found ballots that indicated voter intent, the ratio number for unrecovered double votes represents the highest possible number of counted votes for each unrecovered double vote. Those numbers would be lower (indicating that unrecovered double voted ballots occurred more frequently) if some of the Herald recovered ballots had three or more marks. In those counties/categories that the Herald did not “recover” any votes, the ratio numbers would not change.

What the Data Show

Excerpt from Table 1 — Overvoted Ballots in Optically Scanned Counties

The table illustrates the following significant details:

• There were no double-marked ballots among the Escambia County overvoted absentee ballots examined by the Herald. Escambia is the only county that has no double-marked ballots.

• All counties — except Escambia — have unrecovered double-marked ballots.

• In the rest of Florida, there is no massive disparity between the number of absentee ballots with multiple marks, and the number with just two marks. Escambia County is unique in this regard.

• Almost half the Escambia County ballots that were overvoted at the polls were double-marked, and fewer than 30% of those double-marked ballots showed voter intent. In other words, the county-wide pool of voters — from which the absentee voter pool is drawn — shows no tendency to overvote their ballots in any unusual fashion. The overall pool of voters from Escambia county is not unique, and there is no reason to believe that the absentee voters in Escambia would be unique.

Excerpt from Table 2 — Counted Votes and Undervotes

Excerpt from Table 3 — Overvote Ratio

The disparity in the absentee ballot pool cannot be so easily explained. Factors such as ballot design and the overall makeup of the pool of absentee voters could account for the disparities in most counties, but not in Escambia.

The ratio table shows that a “normal” occurrence is one unrecovered double-marked ballots for each 250–900 ballots cast.(8) Five counties have frequencies that are so anomalous that investigation is warranted. In Flagler and Seminole counties, the ratio is not so high that it creates a presumption of fraud. Bay and Santa Rosa counties have frequencies anomalous enough to indicate fraud. Escambia County, which has a minimum ratio of 1 unrecovered double-voted ballot for every 21,222 valid absentee votes cast (assuming that the next ballot cast would have an unrecoverable double-voted ballot) is completely “off the map.”

Even assuming that the rates from Bay and Santa Rosa counties were somehow normal, thus making Escambia’s rate slightly less unimaginable without assuming fraud, the differences in the distribution of overvoted ballots make it impossible to consider Escambia’s numbers as indicative of anything but ballot tampering.

The relationship between the occurrence of unrecovered double-voted ballots and the occurrence of machine counted overvotes is illustrated in the ratio tables.

In all counties except Escambia, there is never more than a 1:6 ratio between the number of unrecovered double-marked absentee ballots and the number of all machine counted overvoted absentee ballots. (Lake County, which had the 1:6 ratio is somewhat of an anomaly in that regard. Only one other county approaches that rate (1:4.5 in Hernando)(9) and the overwhelming majority of counties have a ratio less than 1:3.) If there had been even one double-marked absentee ballot in Escambia, its ratio would have been close to 1:300.

Conclusions

Circumstantial evidence provides additional support for the conclusion of fraud. Although Escambia County had spent more than $500,000 on precinct-based scanning equipment that had the potential to provide voters with overvote and undervote protection, Escambia County did not turn on this feature at the polls. The professed reason was to save money on duplicate ballots — which would have cost 23 cents each. It would have cost less than $750 to provide duplicate ballots at the polls to the 3,201 people who overvoted in the Presidential race. This explanation also contrasts with the decision to duplicate more than 2,400 absentee ballots, at a cost of more than $550 not including labor.

Ballot tampering in a federal election is both a federal and state offense. The decision to turn off voter protection, then “correct” only overvotes and undervotes only on absentee ballots also represents illegal discriminatory conduct.

According to the latest census, 24.1% of Escambia citizens are African American, and according to the state of Florida, 16.6% of Escambia’s registered voters were black. Blacks make up 30% of the registered Democrats in Escambia (and only 2% of the Republicans) and 86% of black voters are registered as Democrats (6% are registered as Republicans). In this election, there was more than a 10-point difference in the Bush/Gore margin between poll-cast and absentee votes, which suggests that there was a significantly lower percentage of blacks among absentee voters than among voters at the polls.

There can be little doubt that the Escambia County Canvassing Board would be well aware of these demographic realities.

The Escambia County Board offered various explanations why they handled absentee ballots differently from poll cast ballots, none of which hold up to scrutiny. Turning off voter protection at the polls in order to “keep things moving at the polls” or to “save money” might be considered by some to be valid reasons; but they do not explain why poll-cast votes were not treated the same as absentee votes when it came to counting them under a “voter intent” standard.

The decision to turn off voter protection at the polls but to correct overvoted and undervoted absentee ballots must therefore be seen as a deliberate decision to provide different levels of “protection” to two different sets of voters, with the set of voters containing a disproportionate number of blacks receiving the lower level of voter protection. Even assuming that the County officials were unaware of any racial disparity on election night, those officials had the data that showed the racial disparity in sufficient time after the election to go back and provide the poll-cast ballots with the same treatment received by the absentee ballots.

Aside from the evidence of ballot tampering, the higher rates of rejection for ballots cast in minority neighborhoods requires investigation and prosecution of Escambia County’s elections officials to assure Americans, and especially African Americans, that federal elections will be conducted in a nondiscriminatory manner. It is a violation of the Florida and U. S. Constitutions, as well as of the Federal Civil Rights Act and Voting Rights Act, to administer elections in a way that discriminates against African Americans; under the Voting Rights Act, it is not necessary to show intent to discriminate; a discriminatory effect constitutes violation of the law. If the Department of Justice does not investigate Escambia County and prosecute those responsible for the disparate treatment of black voters, it will justify the suspicion that the Bush/Ashcroft Justice Department has no interest in preventing or prosecuting criminal racial discrimination.

Tables

The county-wide “recovered” numbers are from a table published online by USA Today on 11 May 2001 in conjunction with the Herald ballot examination. The table is no longer accessible online.

The absentee precinct-counted “recovered” numbers are from the Herald recount web site search engine (also no longer available online). The counties are divided into three distinct groups for the following reasons:

• There is a huge difference in the occurrence of overvotes between counties that count ballots at the precincts and counties that count them in the central elections office; centrally tabulated ballots (OC) have many more overvotes.

• According to the Orlando Sentinel, 26 counties, almost all of them precinct tabulated counties (OP), duplicated and hand-counted the overvoted and undervoted absentee ballots.

• Some of the centrally tabulated counties hand-counted all their overvoted and undervoted ballots — in six centrally tabulated counties (OCH), the Herald found no unrecovered overvoted ballots. All of these factors influence the likelihood of overvotes included in the data, as well as the kinds of overvotes included.

When looking at the precinct-tabulated data, keep in mind that some counties (Escambia, Manatee, and possibly others) did not turn on the overvote protection. The data also suggest that some counties did not use voter protection consistently throughout the entire county, whereas others (Leon, Seminole, and Flagler, for example) provide voter protection in all precincts.

The data for Calhoun County are not included in the table because that information is not available in http://bushboyzstolethevote.com/votes.htm. Washington County is also not included because the overvote data did not isolate its absentee ballots. There may have been no overvoted absentee ballots in that county at all. This would be unusual, but given that the total number of absentee ballots in Washington County was less than 5% of the ballots cast in Escambia County, the lack of any overvoted ballots is not so anomalous as to suggest fraud.

Table 1. Overvoted Ballots in Optically Scanned Counties

• Mark — Marks, mostly made with a pen or pencil, that a scanner would detect. They may be deliberate marks for a listed candidate; they may denote the presence of a possibly valid write-in vote; or they may be erasures and stray marks that have been erroneously identified by the machine as “votes.”
• Double-marked — A ballot that is overvoted with exactly two marks.
• Multiple-marked — A ballot that is overvoted with three or more marks.
• Total overvotes — The total number of ballots that the machine read as overvotes (“double” plus “multiple”).
• Recovered votes — Ballots on which the Miami Herald was able to determine “voter intent”; not ballots duplicated and hand-counted by the counties.
• Unrecovered double — The total double-marked ballots minus the recovered votes; the minimum number of double-marked ballots for which the Herald was unable to determine voter intent. This is a minimum, not an absolute number, because some of the recovered ballots may have had three or more marks. This number is included to contrast the numbers for Escambia and the other counties — to show that Escambia County is the only county that had no double-voted ballots.
• OC — Optically scanned ballots that were tabulated by machine in a central location.
• OCH — Optically scanned ballots that were tabulated by machine in a central location, and subsequently hand-counted.
• OP — Optically scanned ballots tabulated at the precinct where they were cast.

Table 2. Counted Votes and Undervotes