TIA’s 2004 Election Model


Exit Poll Analysis:  The Essential DU Threads

Exit Poll Analysis:  7/12/05 Update

2004 Election Collection




Other Links

2004 Election Results & Discussion

Election Reform, Fraud, and Related News

Tim Lohrentz on Exit Polls

Diane Perlman on Facing Election Fraud



Try the Newest Version of the Interactive Election Model Here.

(You’ll Need Excel.)

The Unanswered Question: Who Really Won In 2004?

With An Introduction by DemocraticUnderground Poster Autorank


According to the vote tabulators, in the 2004 presidential election George W. Bush won a stunning victory that defied all odds, particularly those applied by unbiased statisticians. He won despite trailing in most state and national polls. He won despite an approval rating of less than 50%, usually the death knell for an incumbent presidential candidate. He won despite trailing in the three National Exit Polls three timelines from 4pm to 12:22 am (13047 respondents) by a steady 48%-51%, miraculously winning the final exit poll (with only 613 additional respondents, totaling 13,660). This poll was weighted (altered) to meet the reported election result on the assumption that the reported result was accurate -- quite an assumption. The final poll showed a stunning reversal of the Kerry 51%-48% poll margin, which had been measured consistently all day by the same polling group: major news/networks and polling firm Edison-Mitofsky.

The analysis of exit polls and documented fraud in this election began on the Internet. A number of academics posted detailed work showing the near-impossible odds of Bush overcoming deficits in the state exit polls and the National Exit Polls. Much of this analysis comes from TruthIsAll (TIA), a poster on DemocraticUnderground.Com. TIA has a background and several degrees in applied mathematics. Using various elements of the national and state exit polls and other data sources, he produces results that are thorough, detailed, sober and compelling. He shows ALL data and calculations, while encouraging others to check his math. Only once did he make a minor math error, after asking DUers to check his calculation of probability that at least 16 states would deviate beyond their exit poll margin of error and go for Bush. The answer turned out to be one in 19 trillion! The debates on DemocraticUnderground's "2004: Election Results and Discussion" forum are legendary and have attracted observers from all over the Net.

Before the election, TIA produced a daily update of his Election Model site. On 11/1/04, based on extensive statistical analysis, he projected a Kerry win of 51.63% to 48.38%, using a combined average of national polls, and of 51.80% to 48.2% using a Monte Carlo simulation of individual state polls. After the polls closed, data from the Edison Mitofsky NEP survey (sponsored by the major television networks and CNN) was unintentionally released over the Internet. This was internal network data, embargoed from public use, data with statements like “Estimates not for on-air use” and “This page cannot be displayed.” The networks had locked down this data for their own use in an “electronic cover-up” that was offensive to those who knew the story. Luckily for all of us, Jonathan Simon downloaded the exit poll data and saved the CNN screen shots! The Edison-Mitofsky (EM)-Corporate Media (CM) “embargoed data” was available for anyone with eyes to see it and a mind to review it.

TruthIsAll immediately began analyzing and publishing analyses on the forbidden data. Looking at the demographics on the second to last E-M major network poll, he laid out the set of improbable circumstances needed for Bush to win: “To believe Bush won the election, you must also believe….” This post was cited by Will Pitt in a major blog, which gave it wide visibility on the Net. “KERRY WON THE FEMALE VOTE BY A HIGHER PERCENTAGE THAN BUSH WON THE MALE VOTE…AND MORE WOMEN (54%) VOTED THAN MEN (46%).” It was all right there, polling results that we were never intended to see. But this was only the beginning. There are over 100 individual analytical postings that demonstrate the tremendous odds against a Bush win. This high-level analysis dovetailed with and was confirmed by on-the-ground stories of voting rights violations all over the country, particularly in Ohio.

The key data sources for TIA’s analysis are the four EM National Exit Polls and the 50 state exit polls. For those who doubt the reliability of exit polling, there has been a trend toward accuracy within 0.4% since 1998. These Exit polls are endorsed heartily by international voting rights activists -- the CarterCenter, for example -- and even the Bush administration, which used them, ironically, in the Ukraine elections to demonstrate fraud and call for a new election.

At 12:22 am on November 3, the national exit poll of 13,047 respondents showed Kerry to be the winner by 51% to 48%, matching TIA’s pre-election projection. The poll was “un-weighted,” meaning the EM and CM had yet to apply weighting “adjustments”: percentages and weights applied to all the demographic categories to match the poll results to the reported vote count! Imagine if this technique had been applied by exit pollsters in the first Ukrainian election to show victory for the incumbent, who had committed gross election fraud. Yet this odd technique of turning a poll into a ratification of the actual voting results was applied in the American election. The final exit poll, with 13,660 respondents, showed a stunning reversal of fortune for Bush. The poll results were “re-weighted” to create a Bush “victory margin.”

The odds against the deviations from the state and national exit polls to the final vote count are astronomical. In addition, there is the consistency of the “pristine” exit poll timeline from 4 pm (8349 respondents) to 7:30 pm (11,027) to 12:22 am (13,047).

In addition to the gender-based evidence cited above, TIA has shown that some weightings for the question “How did you vote in 2000” are mathematically impossible. For example, the final poll claims that 43% of all 2004 voters were former Bush 2000 voters. But 43% of 122.3 million, the number of votes in the 2004 presidential election, is 52.59 million, and Bush only got 50.46 million votes in 2000, approximately 1.75 million of them from voters who have since died. Therefore, Bush’s final poll exit poll numbers, WHICH WERE MATCHED TO THE VOTE, had to be off by 4 million votes.

The analysis also demonstrated that other voter statistics make it impossible for Bush to have won. Even if all Bush voters from 2000 showed up and voted for him, he still needed an additional 13 million votes. He didn’t get them from new voters and those who did not vote in 2000; those voters preferred Kerry by an almost 3-to-2 margin. Because of this, a Bush victory required that he must win a whopping 14% of Gore 2000 voters, all of whom had to return to vote in 2004. But Gore voters were angry; they came back to defeat Bush once again after having the election stolen from them.

Logical absurdities and inconsistencies in Election 2004 abound. The data, analysis, and narrative are available here for open-minded individuals who want to form their own conclusions about “Stolen Election 2004.”

This work is just part of a comprehensive set of election fraud work and analysis provided by the dedicated voting rights activists in DemocraticUnderground.Com’s “2004: Election Results and Discussion” forum, a unique Net resource.

Corporate America controls the media and we get manufactured news.  Corporate America now controls the voting machines and we get manufactured elections.