It was well known in the days prior to the 2004 Presidential election that a Bush victory was highly unlikely without Bush carrying both Ohio and Florida. As Election Day unfolded, spirits in the Kerry camp were running high, as it became evident that Ohio’s 20 electoral votes would determine the victor, and Kerry had a comfortable lead in the Ohio exit poll. Even CNN’s right wing hack, Robert Novak, acknowledged that it would be an uphill climb for Bush.
But as the results came in from Ohio, optimism in the Kerry camp began to fade, and by late evening their remaining hope was narrowed down to strongly Democratic Cuyahoga County, and especially Cleveland, where reports of large pre-election increases in new voter registration and exceptionally high voter turnout had circulated. But this remaining hope soon faded, as it became clear that the voter turnout from Cleveland was in fact miserably low, and by noon the next day John Kerry conceded the election.
It now appears plausible or likely, from a variety of accumulated evidence, that Kerry may have been cheated out of close to 100,000 votes in Cuyahoga County, or more. The evidence for this falls into three categories: 1) Deletion of votes from highly Democratic Cleveland precincts, perhaps through electronic manipulation of central tabulators; 2) Vote switching from Kerry to Bush, perhaps through ballot order rotation tricks; and, 3) Massive purges of voter registration. Let’s consider the evidence for each of these:
Evidence for electronic deletion of votes via central tabulatorsStrange relationship between voter turnout and machines per voterIn Ohio as a whole, as discussed in
Section IV, page 3 of the Democratic National Committee (DNC) report on the Ohio election, there is a discussion about how, in general, voter turnout is strongly related to the ratio of machines per voter. The reason for this correlation is that insufficient numbers of machines per voter can result in reduced voter turnout because of voters leaving the voting lines when they are unable to wait several hours to vote.
However, in Cuyahoga County this normal relationship is inexplicably reversed, so that voting machines per voter is negatively associated with voter turnout. What could this reversal of the normal relationship between turnout and machines per voter mean? One possibility would be that some strongly Democratic precincts that had plenty of machines available for their voters had large numbers of votes electronically, and fraudulently, deleted from the official count by central tabulators. This would have caused the apparent voter turnout from these counties to be low, thereby upsetting the normal relationship between number of machines and voter turnout.
Implausibly low voter turnout in ClevelandRichard Hayes Phillips, a statistical expert in identifying statistical anomalies, whose findings have been widely publicized, has stated that there are at least 30 precincts in Cleveland with inexplicably low voter turnout, ranging as low as 7.1%.
These findings translate into a 51% voter turnout in Cleveland, which is very low compared to the remainder of the county, which experienced a 73% voter turnout.
Evidence for vote switching from Kerry to BushStrange relationship between the Kerry vote and other variables in Cuyahoga County In
Section VI of the DNC report, Professor Walter Mebane concludes that there was no widespread fraud in Ohio of a sufficient degree to overturn the results of the Presidential election. This conclusion is based on strong within precinct correlations between Kerry’s percentage of the vote and four other variables, including the Hagan (Dem. candidate for Governor) percent of the vote in 2002, the Fingerhut (Dem. candidate for Senate) percent of the vote in 2004, voting No on Issue 1 (2004 referendum to ban gay marriage), and percent African-Americans.
I see two problems with Mebane’s conclusion of no widespread fraud sufficient to overturn the election. First, the correlations that he describes do not in any way rule out the kind of scenario described above, under the “Evidence for electronic deletion of votes via central tabulators” section, as long as the deleted votes are deleted randomly from those precincts where they’re deleted.
Secondly (and more relevant to this section of my post), the correlations that were so strong in other areas of the state were noticeably weaker in Cuyahoga County. With regard to the Kerry/Hagan correlations, 6 of the 17 outliers (meaning that the correlation was much weaker) in the state were Cuyahoga County precincts. And with regard to the Kerry correlation with the other variables, Mebane states: “Only seven of Ohio's 88 counties deviate significantly from that pattern." One of those seven counties is Cuyahoga, and Cuyahoga is the only county in the state where there is actually a negative correlation between the Kerry vote and “No” on Issue 1.
So the question we should ask is, if the strong correlation between the Kerry vote and the other variables rules out wide-spread “vote switching” type fraud in the rest of the state, then do the deviations from that pattern in Cuyahoga County mean that a certain amount of that type of fraud likely occurred there? And if so, then how many votes might that have accounted for? Answers to those questions are not at all evident in the DNC report.
Here’s what Professor Mebane wrote to me in response to my query on this subject: “I don't know what went on in Cuyahoga County. As I wrote in several places in the DNC report, there were many anomalies in the data from Cuyahoga County that warrant further investigation.”
Switching of votes through ballot order rotation confusionDr. Phillips also noted at least 16 precincts where votes intended to be cast for Kerry were apparently shifted to other candidates, likely a result of a phenomenon known as
“ballot order rotation”. This is where the candidates are listed in different orders on the ballot in different precincts, so that if a voter votes in the wrong precinct (with a different ballot order) the machine will record the vote for a candidate who was unintended by the voter. Given the large number of voters who voted in the wrong precinct and the fact that this phenomenon occurred largely in highly Democratic precincts, Kerry would lose a lot of net votes because of this. It is difficult to tell how many votes this phenomenon cost Kerry. I can’t make out that Kerry lost more than 1,000 votes through this specific mechanism, but Bill Bored and rosebud57 feel that this may have been a much bigger problem than that. I hope that they contribute to the discussion on this thread, at least with regard to this issue.
Evidence for voter registration fraud According to a report in the New York Times by
Kate Zernike and Ford Fessenden, there were 230,000 new voter registrations in Cuyahoga County in 2004. Yet, according to official Ohio SOS figures, there were only 119,273 new voter registrations in Cuyahoga County between March and November of 2004. If we assume the Zernike/Fessenden article to be accurate, there would have been 110,727 additional registered voters in Cuyahoga County. But let’s suppose that the missing registered voters came from Cleveland rather than from the suburbs of Cuyahoga County – not an unlikely scenario, given the fact that Cleveland is far more Democratic than its suburbs, so that is where fraud would likely have been targeted.
The fraud that is suggested by the above findings could have occurred in one or two ways, or a combination of the two. This could have simply involved purging of legal registered voters from Cuyahoga County. Or, it could have involved electronic deletion of votes via the county’s central tabulator, with fraudulent manipulation of the voter registration figures so that the voter turnout wouldn’t appear implausibly low.
Other evidence – recount fraud A fair recount of the Cuyahoga County vote would have lain to rest the above claims if the recount had shown the official vote count to be accurate. In order to ensure a fair recount, Ohio’s procedures require when a recount is duly authorized (which it was) that a random 3% sample of the county’s precincts be counted first, and if that count shows any discrepancies with the official vote total, then a full recount of the county must proceed. The reason for the requirement that the initial selected precincts be random is that otherwise state officials – if they were overly partisan and if they knew that fraud had been perpetrated in certain precincts – could abuse the purpose of the recount by selecting precincts for the recount that they knew to be “clean”.
Unfortunately, a random recount never occurred in Cuyahoga County (as was the case in numerous other Ohio counties), as precincts were selected non-randomly by state officials, thus throwing into question the validity of the recount. Consequently, two Cuyahoga County Board of Elections workers have recently been
indicted, on six felony counts each, for this act. And don’t be fooled by the statement in the article that says that this crime for which the two workers were indicted would not have changed the results of the election. Nobody can know whether that is a true statement or not, since a valid recount of the vote has not yet been performed. In fact, the above noted considerations would suggest that a recount may very well have changed the results of the election. And what reason would the Cuyahoga County Board of Elections workers have had to risk being indicted for election fraud if there was no good reason for them to believe that the actions they took might be necessary in order to prevent an overturning of the election results?
Let’s make some reasonable assumptions and add up the net votes that were plausibly or likely stolen from the Kerry/Edwards ticketAssumption 1 – voter turnout in Cleveland was equal to voter turnout in the remainder of Cuyahoga County According to current official figures, voter turnout in Cleveland was 51.23%, and voter turnout in the remainder of Cuyahoga County was 73.21%. So, why is it plausible to believe that voter turnout in Cleveland was equal to that in the remainder of the county? First, there was the massive new voter registration drives, coupled by the fact that according to the DNC report, precincts with greater new voter registration were generally characterized by higher voter turnout. We know from the Fessenden/Zernike reports that the Democrats accomplished massive new voter registration in Cuyahoga County. Though the breakdown wasn’t mentioned in the article, where do you think that these voter registration efforts by the Democrats were targeted: Cleveland, where Kerry won 83% of the vote, or the remainder of the county, where he won 53% of the vote? I think it’s reasonable to assume that most of it was done in Cleveland.
Secondly, there is the evidence noted above of fraud designed to decrease voter turnout in Cleveland. And thirdly, several anecdotal reports that I’ve heard testify to the fact that voter turnout in Cleveland was off the charts.
So, let’s suppose that voter turnout in Cleveland was 73.21 %, rather than 51.23%. That would mean a net of 21.98%. Multiply that by 165,578, and you get
36,394 additional votes.
Assumption 2 – The NY Times reporters were correct about new voter registration in Cuyahoga County, and Blackwell’s official figures are wrong It seems to me that this is a no-brainer. Did the NY Times reporters have any reason to lie about voter registration efforts in Ohio? Did Blackwell?
So, we have, as noted above, 110,727 additional registered voters in Cuyahoga County. It seems to me that the vast majority of these additional votes would have come from Cleveland. Again, why would Democrats target a part of the county where Democrats had gathered 53% of the votes, rather than 83%, as they did in Cleveland, unless they targeted those precincts in the remainder of Cuyahoga County that were heavily Democratic? So, with 110,727 additional voters and a 73.21% turnout, we get 81,063 additional voters for Cleveland. Given that Kerry won 83.27% of the vote in Cleveland, compared to Bush’s 15.88%, that gives Kerry a net margin of 67.39%. Multiply that by 81,063 votes, and you get
54,628 addition net Kerry votes.
Assumption 3 – No within precinct vote switching in Cuyahoga County, other than the approximately 800 votes that we know aboutThis assumption hurts Kerry’s chances, but I don’t know how to calculate how many votes Kerry lost due to vote switching due to “ballot order rotation”, so I’ll just leave that blank for now.
Additional net votes for Kerry as a results of the three above assumptionsWe have 36,394 additional votes for Kerry as a result of assumption 1 and 54,628 as a result of assumption 2. Adding those together, along with a few hundred votes from assumption 3
gives Kerry a net of about 92,000 votes in Cleveland alone.
Where do the remainder of the votes needed for a Kerry victory come from? The above assumptions leave Kerry about 26,000 votes shy of victory. Where would the additional needed votes have come from:
Uncounted ballots in OhioThere are at this time more than 106,000
ballots in Ohio that remain uncounted.
Addition of votes to Republican CountiesEvidence that votes were electronically added to Republican counties after the polls closed include the additional of
19,000 votes to Miami County’s total after 100% of precincts had reported (resulting in a net gain for Bush of 6,000 votes) and the illegal
“lockdown” in Warren County, rationalized on the basis of a transparently false “national security” basis.
Vote switching in Mahoning CountyThis involved touch screen machines that registered votes for Bush when voters attempted to vote for Kerry, and indicated in this
EIRS analysis and this
Washington post report of 25 machines that demonstrated this behavior.
Voter suppression by insufficient allocation of voting machines in Franklin CountyAs demonstrated by
Elizabeth Liddle (Febble) and others, this probably resulted in a net loss for Kerry of 17,000 votes. This phenomenon also occurred in many other areas of the state, but was only thoroughly quantified in Franklin County.
More voter registration fraudVoter registration fraud was not limited to Cuyahoga County. According to
Ford Fessenden’s report, new Democratic registration exceeded Republican registration by large amounts throughout the state. Yet, this is
contrary to official figures released by the Ohio Secretary of State’s Office.
Fraudulent recount throughout Ohio
Samples for the recount were chosen in a non-random manner, contrary to established state procedure, and every effort appears to have been made to ensure that results of the 3% sample recount would match election day results, so as to prevent the occurrence of county-wide hand recounts. Perhaps the most flagrant example of this was
Sherole Eaton’s testimony that a Triad technician in Hocking County modified a vote tabulator prior to the recount and advised election officials on how to manipulate voting machinery to ensure that a hand recount would match the machine recount. Ms. Eaton was fired from her job as a result of this transgression. How many others witnessed similar events but did not possess enough courage to risk their livelihood in order to make their observations public, as Ms. Eaton did?
Furthermore, Mr. Blackwell has steadfastly refused to testify under oath with regard to the numerous “irregularities” associated with the Ohio election, and has made every effort to bar the public from access to essential documents that might shed some light on what happened on election day.
What can be done to better document all this?First and foremost a full and fair recount needs to be accomplishedLegal suits are still pending on this.
Matching of Central tabulator count with precinct counts in Cuyahoga CountyThis action could definitely tell us if central tabulator fraud occurred in Cuyahoga County, and to what extent.
Documentation of voter registration fraudThis may require getting documentation from the NY Times reporters who documented and wrote the story on the massive voter registration efforts in Ohio, especially in Cuyahoga County. I have tried to contact them, unsuccessfully.
Documentation of voter turnout in ClevelandThe major part of this analysis depends on this. If anyone is aware of information that would verify that voter turnout in Cleveland was indeed much higher than what the official figures show, please let us have that information.
Vote switching in Cleveland due to ballot order rotationsIn my analysis I assumed no additional votes for Kerry due to this issue because I didn’t know how to quantify it. If anyone can help do that, please let us know.