The Unacknowledged Scandal of Electoral Fraud

First published, in truncated form and without notes, in Humanist Perspectives 165 (Summer 2008): 11-16, The present text restores my original notes, adding as well some references to analyses published since 2008, and includes the final pages of my text that were omitted in the original publication.


1. The New York Times shows its teeth

“Vote Fraud Theories, Spread by Blogs, Are Quickly Buried.”1 With this incisive headline Tom Zeller of the New York Times declared, ten days after the event, the fatuity of sceptical analyses of George W. Bush’s victory in the November 2, 2004 presidential election, and promised a rapid interment for the products of those zombie statisticians and conspiracists whose groans and howlings were echoing through the blogosphere.

I must confess to being of that company, having first declared the 2004 U.S. presidential election fraudulent in an article written on the following day and published online on November 5, 2004.2 A hasty judgment? We’ll see when we get to the evidence.

But how good, on the other hand, was Zeller’s research? Investigative journalist Greg Palast was one of the people he contacted in preparing his article. In November 2000, Palast had broken the story of the massive and illegal disenfranchisement of African-Americans in Florida (a decisive factor in that election),3 and in October and early November 2004 he had denounced parallel illegalities in a series of articles (two of which I cited in my piece).4 Zeller’s inquiries consisted, Palast has written, of two sneering questions: was he a “sore loser,” or was he a “conspiracy nut”?5

Would it be unfair to suggest that “research” of this kind seems of a piece with other recent behaviour on the part of the New York Times? After all, this is a newspaper that put its weight behind the Bush regime’s terror-mongering about Iraq’s nonexistent weapons of mass destruction in 2002-03—and that has since, despite a brief fit of penitence, gone on to endorse a parallel set of fictions about Iran’s equally nonexistent nuclear weapons program.6 Why should one imagine it any more likely to deal honestly with evidence that the Bush Republicans had succeeded—not for the first, but the second time—in stealing the presidency?

Despite the New York Times’ premature obituary, electoral fraud analysis is alive and well. If the evidence available when I wrote my first article was perhaps not conclusive, it has since become overwhelming. The Bush Republicans’ theft of the 2004 election from John Kerry is no longer a ‘theory,’ I would propose, but an established historical fact.7

Let’s consider some of that evidence, beginning with an episode which shows how perilously close the American corporate media have come to losing whatever vestigial sense of irony they may once have possessed.


2. Exit polls in the US and Ukraine: the central bank vs. the local savings-and-loan

Less than three weeks after the November 2nd election, in which Bush defied the expectations of the best-informed analysts8 by securing a second term in office, Republican Senator Richard Lugar issued a ringing declaration: “It is now apparent that a concerted and forceful program of election-day fraud and abuse was enacted with either the leadership or co-operation of governmental authorities.”9

Had Lugar been talking of his own country, instead of Ukraine, where a second-round presidential election had been held the previous day, on November 21st, his statement would have made banner headlines. It would also have been no less accurate. But how did Lugar, who was Chair of the Senate Foreign Relations Committee, and the other Republican heavyweights who joined him in denouncing the Ukrainian election, among them John McCain, Condoleezza Rice, and George W. Bush himself, know that it was fraudulent? Because Ukraine’s election was marked by a scandalous divergence between the official count and the exit polls.

The official vote tally credited Viktor Yanukovych, the candidate favoured by Russia, with 49.4 percent of the vote, and the western-leaning Viktor Yushchenko with 46.7 percent. However, the exit polls showed Yushchenko winning by a decisive margin. One national exit poll—financed by the right-wing U.S. think-tank Freedom House and the U.S. Democratic Party’s National Democratic Institute (NDI), and therefore itself not above suspicion—gave Yushchenko the presidency by 54 to 43 percent. Another probably more reliable exit poll showed Yushchenko winning with 49.4 percent of the vote to Yanukovych’s 45.9 percent.10 The election appears to have been stolen, then, with a divergence of at least 6.2 percent between the official count and what exit polls show to have been the electorate’s actual decision. Massive demonstrations in Kyiv, backed up by an international outcry and threats from the United States of dire economic consequences if the fraud was not remedied, forced a re-run, which was duly won by Yushchenko.

The elephantine irony here—which went unrecognized in the mainstream U.S. media—is that the same criterion which reveals Ukraine’s election to have been fraudulent also shows that the Bush Republicans had for a second time stolen the U.S. presidency. According to the official vote tally, Bush won the popular vote, with 51 percent to John Kerry’s 48 percent. His victory in Ohio, which in this election was the crucial swing state, gave him a majority in the Electoral College as well. But according to the national exit poll figures released by the corporate media on the evening of November 2, it was Kerry, not Bush, who had won by 51 to 48 percent. And according to the state-by-state exit polls, Kerry had also won Ohio, as well as other states—Nevada, Iowa, New Mexico, and probably Florida—awarded to Bush by the official vote tallies.

U.S. presidential elections, it must be remembered, are decided not by the popular vote, but by the winning of a majority in the Electoral College: each state selects a number of electors corresponding to the number of its Representatives and Senators in the U.S. Congress; they are supposed to vote, on a winner-take-all basis, for the candidate who won the popular vote in that state. Under this system, it is easy to predict in a close election which states will decide the final results. Both political parties knew in 2000 that Florida would be one of the crucial states, and in 2004, it appeared that the key states would be Florida, Pennsylvania, and Ohio: whoever won two of the three would become president. When Kerry won Pennsylvania, and Bush was awarded Florida, it all came down, on the evening of November 2nd, to the results from Ohio.

It was therefore potentially embarrassing that the figures from Ohio available in early November 2004 showed a divergence of 6.7 percent in Bush’s favour between the exit poll and the official vote count. (More recently released data showed that the Ohio exit-poll sample had been much larger than was initially indicated—and that the actual divergence in Bush’s favour was fully 10.9 percent.)11

Pundits were quick to insist—as in the title of a November 20th article by Richard Morin of the Washington Post—that “Exit Polls Can’t Always Predict Winners, So Don’t Expect Them To.”12 Two days later, when the Post carried breaking news of the controversy over Ukraine’s dubious election, it quoted a purported election-stealer from that country who held precisely the same opinion: “These [exit] polls don’t work,” declared Gennady Korzh, a Yanukovych spokesman. “And remember, if Americans believed exit polls, and not the actual count, John Kerry would be president.”13

There’s copious evidence that the exit polls which showed Kerry to have won by a wide margin in the Electoral College, and by more than seven million in the popular vote, are in fact highly accurate. Professionally conducted exit polls differ significantly from other forms of public opinion sampling. Pre-election polls, for instance, sample the responses of people whom pollsters estimate may be “likely voters”14 to hypothetical questions (such as “If next month’s election were today, which candidate would you favour?”). Exit polls, in contrast, randomly sample voters in precincts chosen, on the basis of demographical analysis and past voting patterns, as being typical of (for example) rural voters in Ohio’s Scioto County or inner-city voters in Cleveland. People who have voted moments before are asked to indicate, in confidential questionnaires, not what they might choose to do in the future, but what they have just done. Exit polls thus sample—confidentially—actual rather than hypothetical behaviour. Moreover, their sample sizes far exceed those of opinion polls. Opinion polls in the US commonly have a national sample size of around 2,000; the 2004 exit polls in each state had samples of about that size, and the national exit poll was based on responses from over 13,000 voters—giving it a margin of error of about 1.3 percent.15 More than 110,000 voters were sampled, in all, in the national and state exit polls. The fact that comparisons with the official vote counts showed a staggering “red shift” in Bush’s favour can therefore not be simply wished away.

Data available within days of the election revealed that in forty-one states there were deviations in Bush’s favour between the exit polls and the vote count, and that in seventeen states—including all of the so-called “battleground” or “swing” states—the “red shift” deviations exceeded the state polls’ margins of error. The release, long after the election, of more detailed exit poll data brought out the full enormity of the situation. As Dr. Steven Freeman of the University of Pennsylvania has observed, the mean deviation in Bush’s favour in the non-battleground or non-swing states was 5.4 percent, while in the eleven battleground or swing states—those in which, by common consent, the election was decided—the mean deviation in Bush’s favour rose to 8 percent. In the three key states—Florida, Pennsylvania, and Ohio—the average deviation in Bush’s favour was 9.1 percent. The statistical likelihood of such results occurring by chance is, effectively, zero.16

George Bush, it would appear, was playing the same game as Viktor Yanukovych, if for higher stakes. In a second article published in late November 2004, I proposed that “if stealing elections is like knocking off banks, the fact that one practitioner can dynamite the vault of the central bank and get away with it, while his less fortunate compeer draws unwanted attention by blowing out all the windows of the neighbourhood Savings-and-Loan, doesn’t make the former any less a bank robber than the latter.”17

But could the exit polls somehow have been systematically in error? Every attempt to show that this must have been the case has failed. Hypotheses that women (who show a greater tendency than men to support the Democratic Party) were over-sampled, or that Republicans held off voting until late in the day, or that Republicans were for one or another reason under-sampled (the opposite is demonstrably the case), or that the national exit poll’s margin of error was radically underestimated, have in turn been advanced, analyzed, refuted and dismissed.18

We’re left with the alternative hypothesis of a corrupted vote count. But for many millions of Americans, there was nothing hypothetical, or for that matter very subtle, about what happened to them in the 2004 presidential election.


3. What voters experienced: the varieties of vote suppression and fraud

Prior to election day, in state after state, Republican officials obstructed inner-city registration drives and purged voters’ lists (giving closest attention to areas with large African-American, Latino, and Native American populations). Greg Palast estimated that tactics of this kind reduced John Kerry’s support by something like a million would-be voters before a single election-day ballot had been cast.19

But making it onto the voters’ list didn’t guarantee success in actually voting. Campaigns of disinformation organized by the Republican Party sought to confuse inner-city voters as to when and where they could cast their ballots.20 In Ohio especially, a reduction in the number of voting sites and a deliberate undersupply of voting machines in Democratic-leaning urban precincts forced would-be voters to stand in line, in the November rain or in crowded corridors, for six or eight hours, or even longer.21 The physically frail, the elderly, and people with small children were effectively disenfranchised, while others with unforgiving employers faced a choice between voting and keeping their jobs. While vote-suppression tactics of this kind were not reflected in the exit polls, which sample actual rather than would-be voters, they do help explain a third or more of the 10 percent difference in turnout between Republican- and Democratic-leaning precincts in Ohio’s Cuyahoga County (which includes the Democratic stronghold of Cleveland).22

Merely casting a ballot did not ensure that it would be counted. Many minority voters arrived at the head of the line-ups only to discover that they had been deleted from the voters’ lists, or that their qualifications to vote were being challenged by Republican poll-watchers (who were equipped, in direct violation of the 1965 Voting Rights Act, with racial-profiling “caging lists” of black, Hispanic, and in some districts Jewish voters). In more than three million such cases, people were given “provisional ballots”—over a million of which went uncounted.23

Greg Palast’s analysis of U.S. Election Assistance Commission statistics revealed that the EAC’s official figure of 1,855,827 ballots cast but not counted in the 2004 presidential election is “missing data from several cities and entire states too embarrassed to report the votes they failed to count.” Correcting for this under-reporting brings the number of uncounted votes up to 3,006,380. But that is not the end of it: as Palast noted, a U.S. Census Bureau announcement published seven months after the election revealed—though only in a footnote on voter turnout—that, in Palast’s words, “The Census tabulation of voters voting ‘differs’ […] from ballots tallied by the Clerk of the House of Representatives for the 2004 presidential race by 3.4 million votes.”24 The distribution of these uncounted votes was far from random. On a map of Cleveland, Ohio, there is an uncanny correlation between those parts of the city which are 75 percent or more African-American, and those parts in which the proportion of uncounted ballots rose above 17 percent of the total ballots cast.25

Even when votes were counted, they were not necessarily counted as intended by the voter. Complaints to election-monitoring organizations poured in from across the country about the behaviour of the touch-screen electronic voting machines on which 30 percent of the votes nationwide were cast. The problems were large-scale: in Youngstown, Ohio, for example, up to thirty machines had to be “recalibrated” during the day, while many others were left uncorrected.26

The ‘errors’ were irreparable, because the touch-screen machines marketed by ES&S, Diebold, Sequoia and other manufacturers (most of which have track records marked both by fraud and by close links to the Republican Party) preserve no independent record of their transactions. They were also systematic: in nearly every instance, including 86 of the 88 cases documented by one monitoring organization, the machines were flipping Kerry votes to Bush.27 True machine errors would trend toward a neutral effect: a 97.7 percent ‘error’ rate in favour of Bush tells us we are looking at something quite different. And yet the US corporate media persisted in describing effects that were obviously the result of malicious programming as mere “glitches.”

However, the really wholesale fraud occurred at the level of precinct and county vote-tabulation machines. Officials in Republican-controlled Warren County in southwest Ohio claimed on the evening of November 2nd that a terrorist threat passed on by the FBI obliged them to “lock down” the building in which the votes for Warren and two adjoining counties were being tallied. With election observers, journalists, and other potential terrorists excluded, all three counties reported wildly improbable increases in voter turnout from their 2000 levels, and a corresponding and wholly implausible increase of more than 34,000 votes in Bush’s margin of victory. The FBI, it turned out, knew nothing of this terrorist threat, and emails exchanged by Warren County officials prove that they planned the lock-down a week before the election.28

Tabulation-machine fraud can also be demonstrated elsewhere in Ohio. Republican Miami County, for example, after sending in returns with 100 percent of the precincts reporting, had second thoughts and supplied another set of returns with nearly 19,000 additional votes. These second returns were improbably tidy: Kerry’s share of the vote remained, to one-hundredth of one percent, exactly what it had been in the first set of returns (33.92 percent), and George W. Bush was shown to have won the county by exactly 16,000 votes.29 The same clumsy hacking produced grotesquely uneven turnout figures—in the precincts of Concord South and Concord South West, for example, 94.27 and 98.55 percent respectively—while the adjoining precinct of Concord South East reported a more credible turnout of 56.55 percent of registered voters.30 Over-enthusiastic hacking in pro-Bush Perry County produced still wilder results—precincts reporting turnouts of 124.0 and 124.4 percent of the registered voters.31 As with Miami County, Ohio’s Republican Secretary of State Kenneth Blackwell happily certified these results as official.

No less shameful, in an opposite sense, are the certified reports from Cleveland, Ohio, where there is clear evidence that electronic vote tabulators were hacked. Inner-city precincts 92 percent or more of whose votes went to Kerry recorded voter turnouts of as few as 22.31 percent (precinct 6B), 21.43 percent (precinct 13O), 20.07 percent (13F), 14.59 percent (13D), and 7.85 percent (6C) of the registered voters.32 Thousands of African-Americans in these precincts, we are asked to believe, engaged in unprecedented voter-registration drives and then lined up for many hours in the rain—in order not to bother casting a vote.

I’ve focused here on examples from a half-dozen of Ohio’s eighty-eight counties, because that was the state where the election was finally decided. These could be multiplied, thanks to the work of Richard Hayes Phillips and other activists who have gone through the Ohio evidence county by county, developing detailed calculations of the numbers of votes stolen, or conjured up out of thin air, in each jurisdiction.33 But across the country, from North Carolina to the state of Washington, similar work has been done—enabled by the simple fact that it is difficult to steal an election, especially in a country whose electoral system is as chaotically subdivided as that of the United States, without leaving statistically and textually detectable sticky fingerprints behind.

Whether the fraud was ‘retail’ in scale, consisting of forged entries in polling books to accommodate hundreds of absentee ghost-votes for Bush,34 or the apparent incompetence of a lack of signage in multiple-precinct inner-city voting places which led to a carefully planned transfer of hundreds, or even thousands of punch-card votes from Kerry to Bush and to fringe candidates35—or whether it was ‘wholesale’ fraud, a matter of large-scale hacking made visible by gross disparities between hand-counted advance-poll or absentee votes and electronically tabulated election-day votes, or by the strange anomaly of a “down-ballot” Democratic candidate for lesser office repeatedly out-polling the Kerry-Edwards ticket in Republican counties in Ohio (a sure sign that the presidential votes were hacked, while those in a less important contest were left undisturbed)36—the criminal activities of the election-thieves have in many instances been detected, analyzed, and quantified.


4. The scale of the theft

As Steven Freeman and other analysts have observed, the exit poll data makes it possible to assess the overall scale of the fraud. Let’s run through the numbers.37

According to the official vote tallies, 105.5 million votes for president were cast in 2000, and 122.5 million in 2004. Gore won 51 million votes in 2000, and Bush 50.5 million; the remaining 4 million went to Nader and other third-party candidates. Assuming that voters in both camps were equally energized to vote again in 2004, and making allowance for the passage of four years, we can estimate that Kerry and Bush had a hypothetical base from 2000 of 49 million and 48.5 million voters respectively. The exit polls tell us that 8 percent of Gore 2000 voters swung to Bush in 2004 (and 1 percent to third-party candidates), while 10 percent of Bush 2000 voters went to Kerry. That expands Kerry’s base from 2000 to 49.5 million, while Bush’s shrinks to 47.5 million. What about people who voted in 2000 for Nader or other third-party candidates? According to the exit polls, 64 percent of them (2.5 million) voted in 2004 for Kerry, and just 17 percent (0.5 million) for Bush. We can therefore calculate that Kerry won 52 million of the actual total votes cast in 2004 by people who also voted in 2000, and that Bush won just 48 million.

Bush sinks still further out of contention when we add in the 21 million people who voted for the first time in 2004. The exit polls inform us that 57 percent of them (12 million) supported Kerry, 41 percent (8.5 million) supported Bush, and 2 percent (0.5 million) supported a third-party candidate. That means Kerry received—or should have received—a total of 64 million votes, and that Bush should have been credited with 56.5 million.

Compare these figures with the official results, according to which Bush received 62 million votes, and Kerry 59 million. It would appear, in round numbers, that the official tally credits Bush with 5.5 million more votes than he would have received from an honest vote count, and Kerry with 5 million too few.


5. Silences of the corporate media

How has it been possible for a story of such scale and importance to be ignored, indeed suppressed, by the mainstream media of a nation that is still nominally a democracy?

Possible or not, it has happened. In the notes to this essay, I have cited a number of important books and articles in which the evidence alluded to here is explored in detail. The books were published between 2004 and 2010: although they provide responsible and scrupulous analyses of matters of the utmost urgency to a democracy, not one of them, to the best of my knowledge, has received a review in the mainstream North American print media.

But there may be reason to suspect that the corporate media’s complicity in the corruption of the 2004 election goes beyond mere after-the-fact ignoring of the research and analysis carried out by citizen activists like Bev Harris and the mathematician-blogger ‘TruthIsAll’, by investigative journalists like Greg Palast and Bob Fitrakis, and by academics like Steven Freeman and Mark Crispin Miller.

Let’s return to the exit polls. They were conducted by a partnership of Edison Media Research and Mitofsky International (Edison-Mitofsky), which was employed by a media consortium calling itself the National Election Pool, and consisting of ABC News, the Associated Press, CBS News, CNN, Fox News, and NBC News. The job of Edison-Mitofsky was to supply not just successive iterations of exit polling data to its media subscribers, but also information about the incoming vote tallies. Curiously, it may seem, Edison-Mitofsky received raw feed about the vote tallies from the Associated Press, which was thus in the double position of being at once a supplier of crucial input and a purchaser of analytical output.

The Associated Press has been very reticent about the manner in which it acquired vote-tally data—but as reporter Lynn Landes has noted, one independent journalist in Chicago was able briefly to interview an AP representative at the Cook County election headquarters whose computer, unlike those of all the other media representatives, had a direct link to the mainframe vote tabulator.38

The computer security implications are appalling: anyone linked in this way to a central tabulator could just as easily alter the vote tally as download it. (County officials could have preserved ballot security—their primary responsibility, one would think—by transferring information to the AP representative on a series of data-transfer keys or of compact disks.) But there is another still more disturbing implication. Questioned by Landes, the AP refused to confirm or deny that it collected its raw data nationwide by means of direct-link access to vote tabulators.39 If we can assume that the system in place in Chicago was most probably reproduced elsewhere, it would follow that the Associated Press, with real-time access to the movement of data in electronic tabulators across the country, was in a position of at least passive complicity in the fraudulent manipulations of data that we know occurred in many of those machines.

At issue here are the structural relations involved in the tallying of votes and the reporting of vote-tallies. Extending beyond the simple cooperation that would be required in such an operation, these include interlocking directorial control on both sides. In 2004, at least 50 percent of the votes nationwide were counted on electronic tabulators supplied by ES&S (Election Systems & Software), a corporation based in Omaha, Nebraska, which is also the largest manufacturer of touch-screen voting machines. ES&S, which has figured since 1996 in some of the most flagrant vote-tabulation scandals on record,40 is owned by the ultra-conservative Omaha World Herald and by the McCarthy Group, which are in turn subsidiaries of Peter Kiewit Sons, a corporation notorious in eleven states (as Bev Harris observes in Black Box Voting) for its involvement in bid-rigging fraud.41 Lynn Landes notes that the Associated Press, a consortium of 1,500 US daily newspapers, is controlled by a seven-member executive committee, the Associated Press Managing Editors (APME)—whose vice-president in 2004 was Deanna Sands, managing editor of the Omaha World Herald.42

I draw no conclusions, beyond the obvious one that linkages of this kind between private corporations tallying the vote and disseminating the results are a natural consequence of governmental abdication to private interests of the responsibility for conducting honest elections.

Together with the rest of the National Election Pool, and Edison-Mitofsky, the Associated Press was involved shortly after the election in denying a request by another kind of consortium—a group of university scholars, including mathematicians, statisticians, and computer security experts, who were conducting public-interest research into electoral anomalies—for access to the detailed exit-poll data. This information, the media corporations said, was proprietary, and their own analysts could assess it, thank-you all the same, without external help.

The university experts, obviously enough, were anxious to see raw exit-poll data in order to securely establish the implications of the divergence between exit polls and vote tally. But why was the media consortium so reluctant to give them access to it? There is a story here worth telling, one that helps to explain why, in contrast to the Ukrainian electorate, most of the American public—together with the Democratic Party—remained so strangely passive in the face of accumulating evidence that the election had been stolen.


6. The November 2 exit polls—and the corrupted data of November 3

My first article on the 2004 election was prompted by my discovery, early in the morning of November 3rd, that the exit poll figures which on the evening of the 2nd had pointed to a clear Kerry victory had all been changed overnight, and now chimed with the official vote tallies that gave the presidency once more to Bush. What caught my attention was the curious fact that the altered figures (though subsequently cited by the corporate media and by academics as authoritative) were mathematically impossible.

On November 2nd the national exit poll posted by CNN, which was based by 9:00 p.m. EST on 13,047 respondents, showed that women voters (54 percent of the total) had favoured Kerry over Bush by 54 percent to 45 percent, while male voters had preferred Bush by 51 percent to 47 percent, with 1 percent of the vote in each case going to Ralph Nader’s third-party candidacy.43

At 1:36 a.m. EST on November 3rd, however, CNN posted a new set of numbers, based on 13,531 respondents. According to these figures, women’s votes (still 54 percent of the total) had gone 52 percent to Kerry, 47 percent to Bush, and 1 percent to Nader, while men’s votes had gone 54 percent to Bush, 45 percent to Kerry, and 1 percent to Nader. Kerry’s previous lead of nearly 3 percent had evaporated, and now Bush led by almost 1.5 percent. Although the number of respondents had risen by only 3.6 percent,44 there was a mathematically impossible swing of 4.5 percent from Kerry to Bush in voters’ reports of their choices.

A similar pattern was evident in the Ohio exit poll. According to the figures I noted on CNN at 7:32 p.m. on November 2nd, women voters (53 percent of 1,963 respondents in Ohio) had favoured Kerry over Bush by 53 percent to 47 percent, while male voters preferred Kerry by 51 percent to 49 percent. Kerry thus had a decisive lead over Bush of just over 4 percent.

But in new figures posted at 1:41 a.m. EST on November 3rd, a dramatic shift occurred: with a total of 2,020 respondents, women voters had split 50-50 in their preferences for Kerry and Bush, while men had tilted to supporting Bush over Kerry by 52 percent to 47 percent. The ‘adjusted’ exit poll showed Bush beating Kerry in Ohio by 2.5 percent. These fifty-seven additional respondents, I wrote, “must all have voted very powerfully for Bush—for while representing only a 2.8 percent increase in the number of respondents, they managed to produce a swing from Kerry to Bush of fully 6.5 percent.”45

The swing to Bush in the Florida exit poll was if anything more startling. At 8:40 p.m. EST on November 2nd, I took note of results posted by CNN that showed a near dead heat. Women voters (54 percent of 2,846 respondents) had preferred Kerry over Bush by 52 percent to 48 percent, while men preferred Bush over Kerry by 52 percent to 47 percent, with Nader taking 1 percent of their votes.

But the ‘adjusted’ results, derived from 2,862 respondents and posted at 1:01 a.m. on November 3rd, showed that women had preferred Bush over Kerry by 50 percent to 49 percent, while men had voted for Bush over Kerry by 53 percent to 46 percent, with 1 percent of their votes going to Nader. A mere sixteen additional respondents—0.55 percent of the total number—had produced a 4 percent swing to Bush.

I suggested that the adjusted November 3rd exit poll numbers amounted to “a late-night contribution … to the rewriting of history,” and speculated that should questions about electoral fraud in the 2004 presidential election ever become “insistent enough to be embarrassing,” pundits would be able to point to the final exit poll figures “in the decisive swing states of Florida and Ohio—and to marvel at how closely they reflect the NEP’s vote tallies.”46

That prediction wasn’t far off the mark. On November 11th, 2004, the political scientists of the Caltech/MIT Voting Technology Project published a paper which concluded that “there is no evidence, based on exit polls, that electronic voting machines were used to steal the election for President Bush.”47 Of course these investigators found no such evidence: they were using the November 3rd exit poll figures, which had been forced into conformity with the official vote tallies. As one might expect, the Caltech/MIT report’s conclusion was widely reported in the mainstream media—which took no notice of the devastating criticisms their paper received from scholars in other universities,48 or the lead author’s grudging acknowledgment in early December of his team’s embarrassing error.49


7. Complicity in deception by the exit pollsters?

While the Caltech/MIT team can be faulted for not researching their subject more thoroughly,50 there is also evidence that Edison-Mitofsky made concerted attempts to deceive the public about the nature of their ‘final’ exit poll data. November 4th's Washington Post carried a story by Richard Morin which sourced to Joseph Lenski of Edison Media Research an explanation of what had happened with the exit polls. According to this narrative, an Edison-Mitofsky server had malfunctioned shortly before 11 p.m.—“barely minutes before the consortium was to update its exit polling with the results of later interviewing that found Bush with a one-point lead”—and this “glitch prevented access to any exit poll results until technicians got a backup system operational at 1:33 a.m.” on November 3rd.51 But as the evidence I have cited shows, the claim that a final wave of exit poll interviews erased the three-point lead Kerry had held in “preliminary exit poll results,” and gave Bush instead “a one-point lead,”52 is demonstrably false.

The computer-glitch story is likewise dubious: it is contradicted by the updating time given by CNN for the ‘final’ Florida figures (1:01 a.m.), and it all too conveniently masks the labours Lenski and his colleagues must have devoted to massaging the exit poll data—not just the overall percentages, but all of the demographic break-down data as well—into conformity with the official vote tallies. This ‘final’ data couldn’t have been fiddled, the story implies: first, because the glitch “prevented access” to the data between 11 p.m. and 1:33 a.m., and secondly, because the figures that were put up after 1:33 a.m. were already waiting to be posted before 11 p.m. How very convenient.

In an article by Jim Rutenberg in November 4th's New York Times, Lenski was quoted again as offering an explanation of the divergences between early exit poll figures and the ‘final’ figures: “Mr. Lenski said in an interview yesterday that it was possible that more Democrats and more women were voting earlier, perhaps skewing the data in the afternoon. But, he said, by the end of the night the system’s polling data basically tracked with the actual results. ‘Sophisticated users of this data know the limitations of partial survey results,’ he said.”53

Lenski’s one definite statement of fact here is misleading. The national exit poll data, both the figures which were leaked to internet sources early in the afternoon of November 2nd and those which Edison-Mitofsky supplied to CNN and the other members of the National Election Pool that afternoon and evening, consistently showed Kerry winning the election. The data sets released to the NEP at 3:59 p.m. and at 7:33 p.m. and based, respectively, on 8,349 and 11,027 interviews, are now in the public domain (available at These data sets were undoubtedly “partial,” in that exit polling continued throughout the day, but both sets were weighted (a procedure designed to correct for sampling errors or imbalances), and both concur with the data, based on 13,047 responses, which I collected from CNN later that evening.

The conclusion is unambiguous: Kerry consistently led Bush by 51 percent to 48 percent. It is indeed true that “by the end of the night” of November 2nd to November 3rd the figures being circulated by Edison-Mitofsky as exit poll data supported the outcome of the vote tally. But if by “polling data” we mean numbers actually derived from the responses of voters to the exit poll questionnaires, as opposed to altered figures substituted for the data to make it appear to conform with the official vote tally, there was no point on November 2nd or afterwards at which the “polling data … tracked with the actual results.”

Information contained in a November 5th New York Times article by Jim Rutenberg supports this conclusion. Rutenberg noted that a post-election report written by Edison-Mitofsky and distributed by the NEP to its media subscribers acknowledged what it called “the inaccuracies of the projections produced by the early waves of exit poll data.”54 (The wording chimes with Lenski’s pretence that the corrupted November 3rd figures were based upon actual data derived from a final wave of interviewing.) However, Rutenberg also quoted the Washington Post’s managing editor, Steve Coll, who on November 3rd wrote in an online chat with readers that “The last wave of national exit polls we received, along with many other subscribers, showed Kerry winning the popular vote by 51 percent to 48 percent, if true, surely enough to carry the Electoral College.” Interviewed by Rutenberg, “Mr. Coll said his newspaper had to scramble to make last-minute changes to an article analyzing why voters voted the way they did”; and Coll—who like nearly every other journalist in the mainstream media excluded a priori the possibility that the official tallies might be suspect—declared of the survey data, “We think it wasn’t worth what we paid for it, that’s for sure.”55

Warren Mitofsky, the principal of Mitofsky International and a major contributor since the late 1960s to the development of the research model of exit polling, seems also to have participated in the deception. Keith Olbermann reported on November 24th, 2004 that after he referred to “the variance among the early and late exit polls, and the voting” during the previous evening’s MSNBC Countdown program, Countdown received what he described as a “strident” email from Mitofsky protesting against the program’s “misinformation,” and insisting that “no early exit polls” had been released by his company or by Edison Media Research: “the early release came from unauthorized leaks to bloggers who posted misinformation.”56

Mitofsky may have thought he could wish away the figures that he and his colleagues had supplied to the NEP on the afternoon and evening of November 2nd. After all, those percentages had been erased after midnight when CNN and other subscribers replaced them with the corrupted ones. He was perhaps hoping that people would forget that the Washington Post had published the final November 2nd data in their morning edition the following day, and would ignore the fact that screen shots of the November 2nd data had been preserved and circulated by two researchers, Jonathan Simon and Steven Freeman. But neither Olbermann’s remark nor the leaked early data posted by bloggers were “misinformation.”

It should now be clear why Edison-Mitofsky and the National Election Pool consortium were reluctant to release the detailed exit poll data to university researchers. As Mitofsky and Lenski must have been aware, and as the analysts hired by the various media organizations that made up the National Election Pool must also have recognized, the exit poll data constituted a serious challenge to the legitimacy of the election. In forcing the exit poll figures into conformity with the official vote tallies, Edison-Mitofsky were effectively covering up crucial evidence that the election had been stolen—as were the media organizations of the NEP when they remained silent about the anomalies.

Two excuses have been advanced for the exit pollsters’ behaviour. The first is that the forcing of exit poll results to match official vote counts has been standard practice in US elections for at least the past four presidential election cycles. This, regrettably, is true—and it might be taken as one measure of the corporate news media’s retreat from any residual critical and investigative functions. One should ask defenders of this practice to imagine, as a thought-experiment, how they would have responded to news that the exit polls in Ukraine’s 2004 presidential election had been forced into agreement with the vote tallies showing Yanukovych’s electoral triumph.

It has also been argued, as Richard Morin wrote in the November 4th Washington Post article I have already referred to, that adjusting the exit polls “to reflect the actual vote … in theory improves the accuracy of all the exit poll results, including the breakdown of the vote by age, gender and other characteristics.”57 Though widely repeated by pollsters and pundits, this claim is false, both in theory and in practice. It is false in theory because vote tallies contain no information whatsoever about demography, and a re-weighting of demographic data in exit polls to fit vote tally numbers is therefore necessarily an exercise in fiction.58 And it is false in practice because the demographic jigging required to produce the desired percentages robs the exit polls of any scientific value.

Edison-Mitofsky’s November 2nd figures showed, for example, that Hispanic or Latino votes had gone 56 percent to Kerry, 41 percent to Bush, and 3 percent to Nader. What trust could be given to their November 3rd claim that Kerry had actually received only 53 percent, and Bush 44 percent? Other evidence, including an exit poll taken by the William C. Velasquez Institute, suggested that Bush had in fact not even obtained the 35 percent of the Latino vote that he received in 2000. In what could only be regarded as a stinging rebuke to Edison-Mitofsky, NBC News took the unprecedented step on December 3rd, 2004 of revising its exit-poll estimates, reducing Bush’s support among Hispanics to 40 percent, and raising Kerry’s to 58 percent—thus doubling his lead among this group (which makes up 9 percent of the population) from 9 to 18 percent. NBC also altered its estimate of Bush’s Hispanic support in Texas, “revising a reported 18-point lead for Bush to a 2-point win for Kerry among Hispanics, a remarkable 20-point turnaround….”59 It would appear that NBC’s analysts concluded that Edison-Mitofsky’s November 3rd alterations had exacerbated an existing pro-Bush sampling error in the actual exit poll data.

There is evidence elsewhere in the November 2nd exit poll figures of a sampling bias that favoured the Republican Party. Although by the official tally Al Gore won the popular vote in 2000 by 540,000 votes, or 0.5 percent, the successive waves of November 2nd exit poll figures show 3 percent more Bush than Gore voters among respondents who said they had voted in 2000—a difference that is inflated to fully 6 percent in the November 3rd figures, according to which 43 percent of 2004 voters had voted for Bush in 2000, and only 37 percent of them had voted for Gore in 2000. These percentages tell us, among other things, that the active 2004 electorate included 52.6 million people who had voted for Bush in 2000—an interesting result, given that Bush received only 50.5 million votes in the 2000 election. In this case, a pro-Bush sampling error, instead of being corrected by re-weighting, has been exaggerated to the point of absurdity.

A much larger absurdity in the altered November 3rd exit poll figures has been identified by Michael Collins. Those November 3rd figures show that in the 2004 election Bush lost ground significantly among the rural and small-town voters who had been his core constituency in 2000: he received 2.5 million fewer rural votes in 2004 than he had in 2000, and his 20 percent lead over Al Gore in the small-town vote shrank in 2004 to a mere 2 percent lead over John Kerry. But he won his election victory, according to this altered data, in the major urban centres, where a massive supposed 66 percent increase in voter turnout was led by an increase of more than four million in the number of white voters—even though, in big cities, there was a minimal Republican get-out-the-vote campaign “and, in general, a minimal presence in the form of advertising and special events.”60

As Collins observes, it stretches credulity to think that Bush's victory could have been won in the large cities of the U.S.:

Never mind the fact that exit polling reported that 95%, 66%, and 80% of black, Latino and Jewish voters supported Kerry. Never mind the fact that these voters represent over 50% of the United States urban population. Never mind the fact that whites in big cities are the most liberal group of whites in the nation. Finally, never mind that the new voter findings nationwide showed a 3 to 2 advantage for Kerry and that 40% of the voters in the cities were new voters.61

But the crowning absurdity arises from the fact that the 66 percent increase in voter turnout in large cities reported in the altered November 3rd exit poll figures demonstrably never occurred. Collins notes that official data is available for twelve of the country's twenty-four big cities (accounting for 61 percent of the total big city population); in these cities voter turnout increased on average by 13.1 percent. A more than 100 percent average increase in voter turnout would be required in the other twelve big cities to produce the average 66 percent increase reported in the November 3rd figures.62 Bush's victory, Collins concludes, is an urban legend.

As these examples indicate, a forcing of exit poll data to fit divergent vote tallies doesn’t “improve its accuracy”; rather, it amplifies existing errors, and makes the corrupted exit polls useless for any honest purpose.

Given Edison-Mitofsky’s alteration of the exit poll figures and vigorous pretence that the altered figures were genuinely based on actual data, together with the corporate media’s suppression of evidence—behind which we have observed traces of at least passive complicity on the part of the Associated Press and the other members of the National Election Pool in the stealing of the election—it is hardly surprising that the response of most of the American populace, and of all the most powerful figures in the Democratic Party, was at best one of confused passivity.

Who, after all, was one to believe, the reports in the online Columbus Free Press of Bob Fitrakis and his colleagues, and a scattering of analysts posting at Democratic, or the New York Times, the Washington Post, and all the mainstream electronic media? A few fringe academics, or the great mass of university political scientists, who with honourable exceptions maintained a virginal unacquaintance with the writings of computer security experts, and remaining ignorant of the actual mechanics of the elections they analyzed, were able to denounce charges of fraud as tinfoil-hat theories that would have to involve a cast of many thousands of lurking conspirators?63

The alternatives offered to potential skeptics by the structure of corporate power were succinctly outlined by Tom Zeller of the New York Times when he showed his teeth to Greg Palast in that email of early November 2004: would you prefer to be pilloried as a “sore loser,” or as a “conspiracy nut”?


8. Consequences

But what have the consequences been of George W. Bush’s second stolen election?

Noam Chomsky has suggested that such a question (or indeed, the whole issue of electoral fraud) does not merit serious attention, since in his view any electronically induced errors in the vote count could only have had a random effect, and the public was in any case so grossly manipulated that an election could barely be said to have taken place—the Democratic and Republican parties being merely the left and right wings of the same predatory bird. Such a characterization of the two parties is accurate enough, but Chomsky’s attitude seems strange for a thinker who elsewhere takes an incrementalist view of political action, encouraging his readers to believe that every small step towards human liberation counts for something. If that is the case, then very large steps in the opposite direction ought likewise to be meaningful.

The most immediate consequence of Bush’s theft of the 2004 election was an assault of genocidal intensity upon the Iraqi city of Falluja, which since the spring of that year had been a centre of resistance to the occupying armies of the U.S. and the U.K. Though the city was effectively besieged during the month of October 2004, the all-out attack—which began with strikes on medical facilities, and targeted civilians with cluster munitions, depleted uranium ordnance, white phosphorus and other previously unknown chemical weapons—was launched only when news arrived of Bush’s re-election. However weak and vacillating John Kerry’s position on the war may have been, the Pentagon clearly understood that the attack on Falluja would go forward only if Bush remained in office.

And the purely domestic consequences? The right to vote and have one’s vote counted, the right of a simple majority to remove from office a politician or party of whom they disapprove, are foundational political rights: when they are lost, all other political, civic and human rights are at risk.

These foundational rights appear to have been lost in the United States. Previous elections were crooked, and sometimes stolen, but since 2004 there has been no more reason to trust the official results of any election in the United States than in Guatemala or Kenya. And what of other rights? The Military Commissions Act, passed in September 2006 (and unchallenged since November 2006 by the new Democratic majority in Congress), permits the President to arbitrarily redefine American citizens as “enemy combatants,” and subject them to arbitrary arrest, unconstrained interrogation, and indefinite imprisonment. Together with other Bush regime legislation passed since 2001, this act normalizes and makes permanent a state of exception which supersedes the Constitution and the Bill of Rights.

The country has entered—perhaps not irrevocably—a condition of yet-to-be-fully-activated fascism.




1  Tom Zeller, “Vote Fraud Theories, Spread by Blogs, Are Quickly Buried,” The New York Times (12 November 2004),

2  Michael Keefer, “Footprints of Electoral Fraud: The November 2 Exit Poll Scam,” Centre for Research on Globalization (5 November 2004),

3  Significantly, Palast’s reports appeared not in the U.S., but in Britain’s Observer and BBC television; with the exception of, no U.S. news outlet would carry the story until months after Bush was in the White House. See Greg Palast, “Jim Crow in Cyberspace: the Unreported Story of How They Fixed the Vote in Florida,” in The Best Democracy Money Can Buy (2nd ed., New York: Plume, 2004), pp. 11-81. For other evidence relating to the theft of the 2000 election (and about the New York Times’ role role in legitimizing the theft), see Daniel Lazare, The Velvet Coup (London: Verso, 2001); Robert Parry, “So Bush Did Steal the White House,” (22 November 2001),; Alastair Thompson, “Diebold Memos Disclose Florida 2000 E-Voting Fraud,” Scoop (24 October 2003),; Mark Crispin Miller, The Bush Dyslexicon (2nd ed., New York: Norton, 2002), pp. xxvii-xxviii; and Andrew Gumbel, Steal This Vote: Dirty Elections and the Rotten History of Democracy in America (New York: Nation Books, 2005), pp. 201-24.

4  Palast, “An Election Spoiled Rotten,” (1 November 2004),; and “Kerry Won. Here are the Facts,” (4 November 2004),

5  Greg Palast, Armed Madhouse (2006; rpt. New York: Plume, 2007), p. 187.

6  Strong evidence about the deceptions practised in relation to Iran’s supposed nuclear-weapon program was provided by physicist Gordon Prather in a long series of articles published between April 2004 and October 2009 and available at, The New York Time’s Michael Gordon, who co-wrote with Judith Miller one of the most scandalously mendacious articles about Iraq’s supposed weapons of mass destruction, went on to write in a similar vein about Iran, and to echo unsubstantiated accusations that Iran was supplying weapons used by the Iraqi resistance. See Alexander Cockburn, “This man will buy anything Bush is selling,” The First Post (13 October 2007),; and also Howard Friel, “The New York Times, Iran, and International Law,” ( 25 October 2007),

7  Support for this judgment can be found in the following studies: Bev Harris, with David Allen, Black Box Voting: Ballot Tampering in the 21st Century (Renton, WA: Talion Publishing, 2004); Michael Keefer, “The Strange Death of American Democracy: Endgame in Ohio,” Centre for Research on Globalization (24 January 2005),; Mark Crispin Miller, Fooled Again: How the Right Stole the 2004 Election & Why They’ll Steal the Next One Too (Unless We Stop Them) (New York: Basic Books, 2005); Bob Fitrakis and Harvey Wasserman, How the GOP Stole America’s 2004 Election & Is Rigging 2008 (Columbus: CICJ Books, 2005); Bob Fitrakis, Steven Rosenfeld, and Harvey Wasserman, eds., Did George W. Bush Steal America’s 2004 Election? Essential Documents (Columbus: CICJ Books, 2005); Fitrakis, Rosenfeld, and Wasserman, eds., What Happened in Ohio? A Documentary Record of Theft and Fraud in the 2004 Election (New York: The New Press, 2006); Steven F. Freeman and Joel Bleifuss, Was the 2004 presidential election stolen? Exit polls, election fraud, and the official count (New York: Seven Stories Press, 2006); Robert F. Kennedy, Jr., “Was the 2004 Election Stolen?” Rolling Stone (1 June 2006), available at the Centre for Research on Globalization,; Greg Palast, Armed Madhouse (2006; rpt. New York: Plume, 2007), pp. 187-263; Dennis Loo, “Never Elected, Not Once: The Immaculate Deception and the Road Ahead,” in Impeach the President: The Case Against Bush and Cheney, eds. Dennis Loo and Peter Phillips (New York: Seven Stories Press, 2006), pp. 29-57; and Richard Charnin, Proving Election Fraud: Phantom Voters, Uncounted Votes, and the National Exit Poll (Bloomington, Indiana: AuthorHouse, 2010). Charnin has published tirelessly on the exit poll evidence since 2004 under the blogger name ‘TruthIsAll’; an important collection of his work is available at TruthIsAll: The Unanswered Question: Who Really Won in 2004?, with an introduction by ‘Autorank’ (Michael Collins),

8  The national five-poll moving averages calculated by fourteen different polling agencies between October 29 and November 2, 2004 showed Kerry with a lead of between 0.3 and 3.3 percent over Bush; and Richard Charnin’s Monte Carlo Electoral Vote Simulation (with the challenger allotted 60 percent of pre-election-day undecided voters) predicted a 98-percent probability of a Kerry Electoral College victory. See Charnin, Proving Electoral Fraud, pp. 75-76.

9  See “Ukraine cities defy poll result,” BBC News (22 November 2004),; and “In quotes: World concern at Ukraine election,” BBC News (23 November 2004),

10  Peter Finn, “Partial Vote Results Show a Tight Race in Ukraine Runoff,” Washington Post (22 November 2004), As Peter Traynor wrote in The Guardian, the exit polls funded by Freedom House and the NDI, which gave Yuschenko an eleven-percent lead, “set the agenda for much of what has followed” (Traynor, “US campaign behind the turmoil in Kiev,” The Guardian [26 November 2004], available at

11  See Freeman and Bleifuss, Was the 2004 presidential election stolen?, p. 134.

12  Richard Morin, “Surveying the Damage: Exit Polls Can’t Always Predict Winners, So Don’t Expect Them To,” Washington Post (20 November 2004),

13  Quoted by Finn, “Partial Vote Results Show a Tight Race.”

14  As Freeman and Bleifuss show (pp. 166-73), the methods used by pollsters like Gallup to select “likely voters” lead them to under-sample young voters, those without land-line phones, and black and other minority populations (all of whom tend to vote Democratic)—thus producing, together with other methodological errors, results with a Republican bias of about 7 percent.

15  The margin of error stated by the pollsters on election day for the national exit poll was 1 percent. This figure presumably contained an allowance not just for random sampling error, but also for “clustering error” (which could have been a factor because the precincts within which voters were randomly sampled were selected by the pollsters). However, since they did not explicitly say so, I have raised their figure to 1.3 percent (an additional 30 percent above random sampling error is the allowance most statisticians prescribe for clustering error). See See Daniel M. Merkle and Murray Edelman, “A Review of the 1996 Voter News Service Exit Polls from a Total Survey Error Perspective,” in P. J. Lavrakas and M. W. Traugott, eds., Election Polls, the News Media, and Democracy (New York: Chatham House, 2000), p. 72.

16  Freeman and Bleifuss, Was the 2004 presidential election stolen?, p. 138.

17  Michael Keefer, “Election Fraud in America” Centre for Research on Globalization (November 30, 2004),; available at this website under the title “The Stolen U.S. Presidential Election: A Comparative Analysis.”

18  For detailed discussion of the falsity of the various hypotheses offered in attempts to invalidate the exit polls, see Freeman and Bleifuss, pp. 85-145, and Charnin, pp. 212-47.

19  Palast, “An Election Spoiled Rotten.”

20  Some of this disinformation took the crude form of letters sent to inner-city Cleveland voters telling them that Democratic Party supporters were to vote on November 3 (the day after the election). A subtler and more effective kind of disinformation was produced by a lack of signage identifying different voting precincts for which the same polling place was used. Thousands of voters joined the wrong line-ups and voted in the wrong precincts—which meant, because the punch-card ballots given to voters registered in different precincts listed the presidential candidates in different sequences, that many votes intended for Kerry (the candidate favoured by an overwhelming majority of Cleveland voters) went instead to Bush and to third-party candidates. See James Q. Jacobs, “Precinct Cross-Voting and Ballot Order in the Ohio 2004 Presidential Race,” 2004 Ohio Election—Analysis, Summary, Charts, and Spreadsheets (14 January 2005),

21  See Free Press Staff, “Franklin County, Ohio voting machine assignments, and other information,” The Free Press (20 November 2004),; Bob Fitrakis, “How the Ohio election was rigged for Bush,” The Free Press (22 November 2004),; Richard Hayes Phillips, “Stealing votes in Columbus,” The Free Press (23 November 2004),; and Richard Hayes Phillips, “Another Stolen Election: Favoritism in the Suburbs,” Lyric Poetry Website (26 November 2004), Even Kenneth Blackwell, Ohio's Republican Secretary of State, who was responsible for the vote suppression, the miscounting, and the subsequent cover-up in that state, acknowledged that “a seven-hour wait” outside polling stations “is clearly unacceptable”; see Blackwell, “How Ohio pulled it off,” The Washington Times (17 November 2004),

22  Cornell University statistician Walter S. Mebane's analysis of the shorting of voting machines in predominantly African-American precincts in Franklin County, which was included in the otherwise scandalously inadequate analysis of the Ohio election published by the Democratic National Committee, revealed that it produced a more than four percent reduction in voter turnout.

23  Palast, Armed Madhouse, pp. 199-208.

24  Ibid., pp. 189-90.

25  Ibid., p. 192 (where Palast reproduces maps prepared by Dr. Mark J. Salling of Cleveland State University).

26  Richard Hayes Phillips, “Default Settings in Mahoning County,” The Free Press (23 December 2004),

27  According to staff writers of The Nashua Advocate, over 97 percent of the vote-flipping incidents reported to the non-partisan Election Incident Reporting System (EIRS) favoured Bush: see “News: Election 2004: Who's Reading the Words of 'Internet Muckrakers'? Diebold, For a Start...,” The Nashua Advocate (14 January 2005),

28  Erica Solvig, “Warren's vote tally walled off: Alone in Ohio, officials cited homeland security,” Cincinnati Enquirer (5 November 2004),; Solvig, “No changes in final Warren Co. Vote count, Emails released Monday show lockdown pre-planned,” Cincinnati Enquirer (16 November 2004), On the suspicious results of the vote-counting, see Richard Hayes Phillips, “Election results in southwestern Ohio,” The Free Press (21 December 2004),

29  Richard Hayes Phillips, “Hacking the vote in Miami County,” The Free Press (25 December 2004),

30  Ibid., see also Bob Fitrakis, Steve Rosenfeld, and Harvey Wasserman, “Ten preliminary reasons why the Bush vote does not compute...,” The Free Press (3 January 2005),

31  Fitrakis, Rosenfeld, and Wasserman, “Ten preliminary reasons.”

32  These are the certified figures, from Cuyahoga County General Election: Official Results Report, which is no longer available on the web; the data can now be obtained from James Q. Jacobs' website ( The figures first released after the election, before there had been a partial counting of provisional ballots, were still more shocking: 21.8 percent (Cleveland 6B), 21.01 percent (13O), 19.6 percent (13F), 13.05 percent (13D), and 7.1 percent (6C). Cleveland precinct 10L was initially reported as having a 24.72 percent turnout—a figure which rose in the certified results to a 56.21 percent turnout. Perhaps by some accident all of the provisional and absentee ballots cast in this precinct were counted.

33  Richard Hayes Phillips, Witness to a Crime: A Citizen's Audit of an American Election (Atlanta: Canterbury Press, 2008). Phillips acknowledges the work of many people who assisted him in his investigations—which were hampered by a systematic destruction of evidence in most of Ohio's 88 counties.

34  Dr. Werner Lange, “More Votes than Voters in Ohio: Absentee Vote Inflated, Certified Vote in Doubt,” (12 December 2004), Dr. Lange found 580 such ghost votes in the 106 precinct poll books he was able to inspect in Trumbull County. This level of faked absentee votes, reproduced across Ohio, would have resulted in over 62,000 faked votes.

35  James Q. Jacobs, “Precinct Cross-Voting and Ballot Order in the Ohio 2004 Presidential Race,” 2004 Ohio Election—Analysis, Summary, Charts, and Spreadsheets (14 January 2005),

36  See Status Report of the House Judiciary Committee Democratic Staff, Preserving Democracy: What Went Wrong in Ohio (Washington, D.C.: U.S. House of Representatives, 5 January 2005, available at, B.1, p. 54 note 238, and p. 54-55 note 240.

37  These numbers are derived from Freeman’s work; they are rounded off to the nearest half-million.

38  Lynn Landes, ″Did Networks Fake Exit Polls, While AP 'Accessed' 2,995 Mainframe Computers?” The Landes Report (5 January 2005),

39  Ibid.

40  In 1996 the company was named American Information Systems (AIS). The most stunning upset of that year’s election was Republican Chuck Hagel’s victory in the Nebraska senate race. Hagel was CEO of AIS, resigning a fortnight before announcing his run for the senate, but retaining a financial interest in the company; the votes in Nebraska were all tabulated by AIS. This and other AIS-ES&S scandals are documented by Bev Harris.

41  Harris, Black Box Voting: Ballot Tampering in the 21st Century (Renton, WA: Talion Publishing, 2004, chapters 2 and 8.

42  The reference is to a 2004 article by Lynn Landes, no longer available online, but excerpted following ″Did Networks Fake Exit Polls...?”

43  These CNN exit-poll percentage figures are rounded off to the nearest integer; calculations based on them must therefore be regarded as approximations.

44  I mistakenly wrote in “Footprints of Electoral Fraud” that the number of respondents had risen “by less than 3 percent.” The small corrections I have made to some of my original calculations in no case affect the substance of the argument.

45  Keefer, “Footprints of Electoral Fraud.”

46  Ibid.

47  Charles Stewart III et al., “Voting Machines and the Underestimate of the Bush Vote: Version 2,” CalTech/MIT Voting Technology Project (11 November 2004),

48  See Steven F. Freeman, “The Unexplained Exit Poll Discrepancy,” University of Pennsylvania Center for Organizational Dynamics (revised version, 29 December 2004), p. 4. Freeman notes that Stewart’s team “used data in which the [vote] count is assumed correct to prove that the count is correct!” Peter Caithamer, argued, in “A criticism of the Caltech/MIT Voting Technology Project’s ‘Voting machines and the underestimate of the Bush vote’,” Crisis Papers (18 November 2004),, that the paper’s reliance on “modified exit polls” destroyed its credibility, since “analysis of these polls suggests that the modification was not legitimate.” Leonard Wayne showed, in “Problems with ‘Voting Machines and the Underestimate of the Bush Vote’” (23 November 2004),, that the corrupted exit poll data used by the Voting Technology Project group matched the election returns too closely to be real: “The odds that normally distributed data could by chance be so tightly clustered as in the VTP figures is less than 1 in 1,000.”

49  Charles Stewart III, “Addendum to Voting Machines and the Underestimate of the Bush Vote,” Caltech/MIT Voting Technology Project (5 December 2004),

50  Other political scientists have since trotted happily across the same pons asinorum. See, for example, Michael Traugott, Benjamin Highton and Henry E. Brady, “A Review of Recent Controversies Concerning the 2004 Presidential Election Exit Polls,” The National Research Commission on Elections and Voting: A Project of the Social Science Research Council (10 March 2005),, p. 5. Although the authors believe that the “initial versions of the data” were those made available by CNN on November 3rd, this paper is of value for its reminder that the NEP continued to tinker with the data, releasing on November 7th ‘exit poll’ figures that gave Bush a lead of fully 4 percent. As late as January 2006, elections expert Paul Gronke was still recommending the Caltech/MIT paper as authoritative: see his contributions to a discussion thread at “New Year’s Resolutions?” Blue Oregon (30 December 2005-22 January 2006),—and my responses, which are also available at this website.

51  Richard Morin, “Exit Polls: New Woes Surface in Use of Estimates,” Washington Post (4 November 2004): A29.

52  The “one-point lead” Morin mentions is presumably his rounding off of the 1.4 percent lead credited to Bush in the ‘final’ or corrupted national exit poll figures of November 3rd.

53  Jim Rutenberg, “”The 2004 Elections: The Electorate—the Polling; Survey Experts Cite Problems with Data and Interpretation,” New York Times (4 November 2004),

54  Jim Rutenberg, “Report Says Problems Led to Skewed Surveying Data,” The New York Times (5 November 2004),

55  Ibid.

56  Keith Olbermann, “Bloggermann: Zogby Vs. Mitofsky,” MSNBC News (24 November 2004),

57  Morin, “Exit Polls: New Woes Surface in Use of Estimates.”

58  An anonymous poster at the Democratic Underground site, who claimed to have worked in polling at the state and national levels since 1975, explained the issue very clearly: “The only place we learn anything about the voting patterns of demographic or political sub-groups is through an exit poll (or some other pre- or post-election poll) where the characteristics of the voters can be matched with what they say are their actual or intended votes. The actual votes cast … contain no information that would allow any adjustment whatsoever in refining the demographic/political sub-group analyses.” See ‘Fly by night’, post 14 in a discussion thread initiated by ‘TruthIsAll’, “I learned something about exit polls today,” Democratic Underground (17 January 2005),

59  William C. Velasquez Institute (press release), “NBC Makes Unprecedented Downward Correction in Latino Support for Bush,” Yahoo Finance (3 December 2004),

60  Michael Collins, “Election 2004: The Urban Legend,” Scoop: Independent News (13 June 2007),

61  Ibid.

62  Ibid.

63  In the first published version of this essay in Humanist Perspectives, which appeared without notes, I offered some guidance to “essential resources” in a follow-up note. I wrote that “Greg Palast provides essential evidence on the Republican Party’s large-scale vote-suppression campaigns and on the non-counting of votes; Bev Harris’s work on electronic fraud is of equal importance. For analysis of the crucial exit poll evidence, see Steven Freeman and Joel Bleifuss, and the postings of ‘TruthIsAll’. Bob Fitrakis and his colleagues have gathered together important collections of documents and analyses, while Mark Crispin Miller has analyzed the corporate media’s suppression of electoral fraud issues. The articles of Dennis Loo and Robert F. Kennedy offer useful summaries of the evidence. See also “Professor Steve Freeman Presentation on Stolen US Elections,” Second National Latino Congress, Los Angeles (October 2007),”