How Fair Was the Election of 2004

While the 2004 vote was still being cast, a possibly tipsy New York Republican Representative Peter King appeared on the White House lawn with a blue plastic drinking cup in his hand. He told a documentary film-maker, "Its over! The election’s over-we won." When asked what that meant, he added, " It’s all about but the counting, and we’ll take care of the counting." King is one the better members of Congress and he seemed to be disquieted this knowledge.The election victory of 2004 violated two long-standing rules of thumb.( 1) Last minute undecided votes usually do not break for the incumbent. (2) An incumbent whose approval rating falls below 50% usually loses. These rules are not written in stone. Questions should be raised about the 2004 election because the results are not statistically probable and there was much evidence of gaming.

Recently, Christopher Hitchens published a piece in the March 2003 Vanity Fairentitled "Ohio's Od Numbers" and argues that the Federal Election Commission should take a careful look at the strange outcomes in Ohio. A sharp critic of John Kerry, Hitchens only looked at a few obvious problems and did not view the outcome from the standpoint of a statistician. He also gave full weight to the Republican argument that the Ohio election could not have been fixed because that would have required the cooperation of a significant number of people.There was less physical intimidation of minority voters in 2004 than in 2000, but there were probably even more efforts to suppress the minority vote. There were was less irrefutable evidence of efforts to almost openly fix machines, but those that did turn up almost always favored the GOP.

Almost gone was the shifting large numbers of votes to obscure minor party candidates for president. We can no longer expect to see the obvious and huge shifting of votes that occurred in Valusha County and elsewhere in Florida in 2000 or the unauthorized reprogramming of a large number of electronic machines, as occurred in southern Georgia in 2002. However, the election was replete with the kinds of anomalies and strange, unpredictable outcomes that alarmed statisticians and seasoned election observers.The election was unusual because there was such a great disparity between the exit polls and the actual results. The exit polls called for a Kerry victory , but the actual results gave Bush another term in the White House. In 42 of 51 states ( and District of Columbia), the states moved more toward Bush than the exit polls indicated. When it was clear that the exit polls appeared to be very wrong, the Edison and Mitofsky exit poll results were "rebalanced" early in the morning of November 3 so they came closer to actual results. An important assumption in the "reweighing" was that Bush got every vote he garnered in 2000; no one died and no one changed his mind.

This data was published by CNN and a CALTECH/MIT study, based on the modified figures, pronounced that the disparity between the exit polls and actual results was not great enough to warrant further inquiry. The actual exit poll data became available two weeks later and great disparities or "red shifts" in favor of Bush. In Delaware it was 10%,in New Hampshire 9.8%, in North Carolina 8%; in Ohio 6.2%, and Florida 6%. (2)In ten of the eleven battleground states, the differences between tallied and exit poll margin favored Bush. For two generations, the media’s exit polls have never been more than a tenth of a percent off. Freeman and Mitteldorf noted that there was a huge 10.0 average within precinct error in precincts that went at least 80% for Bush. On the other hand, the WPE for similar Kerry precincts was 0.3. Even Republican consultant Dick Morris told FOX News that "Exit polls are almost never wrong." The discrepancies were far outside the allowances for random error and chance. Statisticians have concluded that the chances of the wide variances of 2004 occurring were less that one in a thousand.

The chances that Kerry only received 47.1% of the vote in Florida are only three in a thousand. The best explanation Gallup pollsters could come up with was that Kerry people were much more willing to participate in exit polls than Bush backers. This became the official explanation of the pollsters. Yet 56% of those approached completed the surveys in Bush precincts, compared to 53% in Kerry bastions. (3) In states where a variety of means were used to count votes, there was no meaningful disparity between exit poll results and actual tabulations. However, there was a great difference between the exit polls and reported results in states were electronic programs were used to register and count votes. (4) The three corporations that control the manufacture and maintenance of electronic voting machines and the programs that tabulate votes are controlled by people with strong GOP ties. Moreover, their vote tabulating programs are easily hackable.

The many statistical anomalies of 2004 almost all benefitted the GOP, and they were a sort that could not be readily investigated. Indeed, Republican computer security expert Chuck Herrin noted that they "benefitted us 100% of the time." In the election of 2004, the electoral count was settled favor of George W. Bush in Ohio and Florida, both states where electronic voting and tabulation was predominant. A Quantitative Methods Research Team at the University of California at Berkeley examined the Florida election and found that "a county’s use of electronic voting resulted in a disproportionate increase in votes for President Bush " and that the random chances of this happening were "less than once in a thousand...." They conservatively estimated that Bush received between 130,000 and 260,000 more votes than could be statistically expected." Some voting machines there actually counted votes backward against Kerry. Bush carried the state by 5% , but lost it by 2% according to the exit polls. From a common sense but not rigorous statistician’s perspective, the results appear even more difficult to believe.. Among the 57 counties that used vote-scanning technology to count votes, 29 had been overwhelmingly Democratic. This time then became Bush bastions. The GOP gained 128.45% in those counties, and the Democrats lost 21%. In Liberty County, which is 88% Democratic, Bush gained 700%. Where optional scan technology was not used, voting followed predictable patterns. In Broward County, voting workers said there were boxes of uncounted absentee votes that were later removed from the court house. Before the election there, 50,000 applications for absentee ballots somehow got lost in the postal system. 6Analyzing the races in Ohio, Florida, and Pennsylvania ( where a Senate seat was involved), Professor Ron Baiman thought the discrepancy between exit polls and tabulated votes could only occur once in 155,000,000 times.

The race was especially close in Ohio, where Bush eventually was declared the winner by less than 130,000 votes. There was not one elected statewide official who was a Democrat, and the secretary of state, J. Kenneth Blackwell, who counted the votes, was the Bush campaign chairman. Every possible legal and marginally legal device was used against Kerry. . In Franklin County there was a pattern of taking voting machines out of strongly Democratic precincts so that the lines would be longer. There were 580 absentee votes in Trumbull county which could not be matched with voters in the registration list. If this were extrapolated statewide, it would come to 62, 513 fraudulent absentee votes. A former ES&S employee was allowed to tamper with the vote counting computer in Auglaize County on October 16. In Warren County( about 20 miles northeast of Cincinnati–Kings Island), the court house was locked down so that reporters could not observe counting. In the past, there was room for the media in the court house during elections. It is important to know this is one of the last counties in Ohio to close its polling places. That county had a 33% increase in votes cast over 2000. In Perry County, more votes were cast than people who signed the register as on site voters or absentees/ In Butler County, the Kerry-Edward ticket ran far behind other Democrats. In the Columbus area, there were 4,258 votes in a Gahanna precinct and 260 for Kerry. Fortunately this was corrected. However, there were only 800 voters there. Miami County was very slow counting its votes and only released its complete total when almost the entire state had been heard from. Bush picked up an amazing 19,000 more votes than in 2000 in this rural county. , Kerry carried 31.38% of the voter, compared to 36.38 for Gore.. The county boasted an almost impossible increase in turnout, with the new votes shared between the candidates in these proportions. Some precincts had nearly 98% turnouts. In Mahoning County, many voters reported that they attempted to vote for Kerry on the video machines, but the vote kept turning up for Bush. Some might recall that this is Youngstown and naturally wonder is some kind of corruption were involved. In Lucas County where Toledo is located, , numerous machines consistently malfunctioned. Outside of Xenia, polling officials applied special standards to prevent black students from Wilberforce University

The Ohio recount was also marred by allegations of fraud. Triads Corporation, that services the punch card voting system in 41 counties was found to be sending people to court houses to prepare computers for recounts. The clearest case of abuse was in Hocking County, where Michael from Triad appeared to brief workers on "tricky" questions attorneys might ask and to work with tabulator and computer/ He asked which precincts had been designated for recount and then worked on the computer, telling the employees not to turn it off as the correct results were then on display. He also helped employees post a "cheat sheet" to help them with the recount. The deputy election commissioner who blew the whistloe on this conduct was subsequently fired by Hocking County, which was presured to do so by Secretary J. Kenneth Blackwell.In Lucas County, a Diebold technician reprogrammed machines prior to the recount, and in suburban Sylvania Precinct 3 the programming card was reprogrammed..Voting officials in Shelby County admitted they had destroyed critical materials, making a recount there impossible to the recount, Secretary of State Blackwell ordered voting officials in the counties to prohibit others from examining poll registers, a violation of Ohio law which made it impossible for Democratic observers to analyze the election or make sense of the recount.

Strange occurances seemed to mark the recount in many of the counties of northwestern Ohio. Moreover, Blackwell did not impound a single machine or computer, so a trustworthy recount was impossible. Moreover, he refused to appear before a House committee investigating the Ohio outcome even though its chairman was a Republican who would have treated him gently. In the same spirit, the Ohio Atgtorney General has demanded that the State Supreme Court discipline four attorneys who unsuccessfully litigated in a quest for a more thorough recount. Much more could be said about the abnormalities in the Ohio recount. One wonders why any rigging was even necessary in view of the fact that Ohio was certain to certify the original results.Before the election, many efforts were made to hold down the minority vote.The Republican Secretary of State in Ohio attempted to invalidate voter registration forms in Cleveland because they were not printed on pound stock paper.9 A quarter of a million African Americans in Ohio received letters warning them that they could not vote; among the reasons given were being registered by the NAACP.

Misleading pamphlets were also sent to Milwaukee blacks in an effort to discourage their voting. John Perzel, speaker of the Pennsylvania House of representatives, openly stated it was necessary to suppress the black vote in Philadelphia. 11In Las Vegas, Nevada , the Republicans hired Voters Outreach of America to register voters. The firms managers were caught destroy Democratic registration forms. Employees said they were only being paid for Republican signature. In Oregon, the same firm was discovered to be repeating this performance. However, there it sometimes used public libraries, claiming to be official registrars. Its employees were reported to have told University of Oregon students that by registering as Republicans they were somehow fighting the sexual abuse of children. Similar problems appeared in West Virginia.. 12 In Florida, election officials tried to remove another 22,000 Blacks from the voter roles, even though they were not felons. Republican Secretary of State Glenda Hood disqualified an additional 10,000 registration forms of likely Democrats because they had checked in only one place rather than two that they were citizens of the United States.

Former President Jimmy Carter, whose center has monitored elections around the world, noted that the process in Florida did not meet international standards in part because it is operated by people holding "strong political biases," and he predicted that the 204 Florida election would again be characterized by irregularities. 13The usual efforts of those in power to prevent some groups from voting were much in evidence. By placing outmoded equipment in black precincts, the chance of having many "spoiled votes" was greatly increased. In the 47 Dayton, Ohio precincts that went heavily for Kerry, the spoilage rate was 5.16%, compared to 1.31% for the rest of Montgomery County. This is partly why blacks have a 800 % greater chance of not having their votes counted than whites. Similarly, placing too few machines in black precincts created great lines and discouraged people from voting. The situation was so blatant in Ohio, that the very Republican Columbus Dispatch felt obliged to report on it. It found that the number of machines in Democratic parts of Franklin county were reduced by 17 while the Republican precincts received eight more machines.

In Democratic Cuyahoga and Summit Counties, old machines were used and the average number of votes per machine was less than half those per machine in the Columbus area. The provisional ballot, was a new technique made possible by HAVA, allowed people to provisionally vote if they showed up at the wrong place or broke some other regulation. These votes were usually not counted . Even Hitchens, a sharp critic of Kerry, noted that the massive bottlenecks "had a tendency to occur in working-class and, shall we just say, nonwhite precincts." Many people simply gave up and did not bother to complete the provisional ballots that officials eventually decided to offer them.

In Gambier, Ohio, there was only working machine where Kenyon College students voted. Many of them gave up and refused to bother with the provisional ballots because they believed the paper ballots would not be counted. These techniques were widely used, especially in Ohio, and boosted the number of uncounted votes to the 3,000,000 mark In addition there were 95,000 lost votes in Franklin County, 12,000 in Mahoning, and 6,400 in Warren County. In some southwest white Republican counties,, C. Ellen Connally, an unfunded African American running for Ohio Supreme court, somehow ran far ahead of John Kerry. 14 In two predominantly black Cleveland precincts, this also occurred. The normal patteren in presidential election years is that the highest number of votes would be at the top of the ticket, and that there would be a sharp drop-off thereafter.Social scientists are supposed to be pragmatists, so it is probably unfair to complain about efforts by a political campaign to suppress the opposition vote. Similarly, extremely stringent regulations for the identification of some voters and increased efforts to eliminate legitimate black voters as part of programs to purge felons might be considered benefits of incumbency.

Perhaps some would draw the line at capriciously changing voting place locations or how many new machines are placed in black precincts might cross the line for many realists. If all the above were considered part of rigging election, many would be justified in concluding that John Kerry should be sworn in as President in 2005. But leaving these things aside, we still have results that defy any reasonable statistical or common sense analysis. Whether the election of 2004 should have gone to John Kerry cannot be established, even though a shift of 65,000 votes in Ohio would have made the difference. Electronic machines and computerized tabulation systems are extremely difficult to audit and there is very little possibility that this situation will change. We can expect more elections like this in the future. One piece there provides extensive background information on the potential for fraud offered by electronic voting machines and computerized counting systems

.1. To view the film clip---http://homepage.mac.com/duffyb/Movies/peterking.mov (Some who viewed this film clip concluded he may have been a bit tipsy.)2. Alastair Thompson, "Complete US Exit Poll Date Confirms Net Suspicions," Scoop.com ( November 17, 2004)3. Steven F. Freeman, "The Unexplained Exit Poll Discrepancy," (November from voting. 10, 2004) stfreema@sas.upenn.edu; Jonathan D. Simon and Ron P. Baiman, "Who Won the Popular Vote? An Examination of the Comparative Validity of Exit Poll and Vote Count Data,"The Free Press.com ( December 29, 2004)4. Michael Meacher, "Did Dubbya rig the election?" New Statesman (November 29, 2004), 22-235. Herrin, "Empathy Training for Compassionate Conservatives,"6. "UC Berkeley Research Team Sounds ‘Smoke Alarm’ for Florida E-Vote Count," (November 18, 2004) ucdata.berkeley.edu; Sean Sabatini, "Exit Polls and Vote Fraud: A user-Friendly Explanation," OPEDNews.com( undated); "Countdown," MSNBC.com ( November 3, 2004)7. John Conyers and others to. Kenneth Blackwell ( December 2, 2004), Truthout.com; Richard Hayes Phillips, "Hacking the Vote in Miami County," The Free Press.com (December 25, 2004).8. William Rivers Pitt, "Proof of Ohio election Fraud Exposed," Truthout.com ( December 15, 2004); Bob Fitrakis, Steve Rosenfeld, and Harvey Wasserman, "American Democracy Hangs by a Thread in Ohio," Yurca Report.com ( December 15, 2004); Toledo Blade, January 5, 20059. "The Poll Tax Updated," The New York Times, October 7, 2004)10. ABCNews.com (October 31, 2004)11. Bob Herbert, "Days of Shame," New York Times (November 1, 2004)12. PBS.com (October 13, 2004); Laura Kurtzman, "GOP Paid Firm Faces Voter Fraud Charge," San Jose Mercury (October 14, 2004); Adrienne Packer, "Voter Fraud Alleged," The Las Vegas Review-Journal (October 14, 2004); Associated Press and KGW News, "KGW Report Prompts Oregon Voter Fraud Investigation." (October 13, 2004)13. Jimmy Carter, "Still Seeking a Fair Florida Vote," Washington Post (September 27, 2004); Kurtzman, "GOP Paid Firm Faces Voter Fraud Charge."