Mitofsky Claim to "Rule Out Vote Fraud" is Bunk
Math Logic Proof Shows that ESI Analysis of
Ohio and National Exit Poll Data is Bunk
-- Mitofsky and
ESI's Claim to "Rule Out Vote Fraud" is Proven
The National Election Data Archive
George W. Bush could have won the 2004 presidential election due to large-scale vote fraud without election data showing the patterns that the Election Science Institute (ESI) and pollster Warren Mitofsky claim must exist if vote fraud had occurred. The analysis that exit pollster Warren Mitofsky presented at the October 14, 2005 American Statistical Association fall conference has been proven mathematically useless for testing exit poll data for vote fraud.
In the 2004 presidential election, John F. Kerry won according to exit polls. Yet George W. Bush won according to the official election results. Exit pollster Warren Mitofsky in a January 2005 paper, stated that the discrepancy between election and exit poll results was caused by Kerry voters responding to exit polls more than Bush voters. From January to June, the National Election Data Archive (NEDA) used algebraic methods to show that this "reluctant Bush responder" explanation was refuted by the available exit poll data. However, in a June 2005 paper, and at the October 14 American Statistical Association fall conference, Mitofsky presented another hypothesis that he claimed "kills the vote fraud argument" in the 2004 presidential election.
Math Logic Proves that ESI's Latest Analysis Purporting to Rule out Vote Fraud Is Invalid
Mitofsky and Election Science Institute (ESI) argue that
"If systematic fraud or error in vote counting [favoring Bush] occurred [in precincts] in 2004 but not in 2000, then Bush would have done significantly better in those precincts in 2004 [than in 2000], and we would see larger exit poll discrepancies in those precincts."
That is, ESI claims that if precincts had vote fraud, then we would expect better Bush performance in those precincts in 2004 than in 2000. Mitofsky showed that precinct-level Bush vote increases from the 2000 election were not correlated with larger exit poll discrepancies, and concluded that vote fraud could thus be ruled out in the 2004 presidential election.
However, to cite just one possibility, what if the Democrats won the 2004 turnout battle big-time so that the effect of vote fraud in those precincts was to rescue Bush from a worse performance than in 2000 and bring him up to even?
The National Election Data Archive (NEDA) uses mathematical logic to prove that ESI's logic is incorrect and that any analysis of vote fraud based on it is meaningless. NEDA, in its proof, shows how Bush could win in 2004 due to large-scale vote fraud and yet have higher exit poll discrepancies where the Bush vote share is less in 2004 than in 2000 (not more as ESI claims).
In other words, no conclusion on the occurrence of vote fraud can be reached via the analysis used by Mitofsky with ESI.
Any mathematician utilizing the discipline of Math Logic can easily check the validity of NEDA's proof by reading “Mathematical Proof that Election Sciences Institute's Test to Rule Out Vote Fraud is Logically Incorrect” which can be found at
NEDA requests that:
1. Mitofsky and ESI please shoot only straight arrows into the fray from now on -- by logically and mathematically checking their hypotheses of the 2004 exit poll discrepancies before publicly releasing them. Time could better be spent on implementing a national election data archive system to analyze election data as soon as polls close; and
2. the polling firm Edison/Mitofsky release their arsenal of 2004 raw precinct-level unadjusted data for the entire United States, as was done for Ohio, so that independent researchers could statistically ascertain whether vote fraud probably occurred or not. Further, precinct identifiers are needed to allow investigation into the causes of some impossible election results in some precincts where the sum of all non-responders to the exit poll plus the number of all responders who said they voted for Bush is less than Bush’s official vote share.
Any valid comparison of the 2000 and 2004 elections to test for vote fraud would require 1) the unadjusted exit poll discrepancy data for the 2000 election, and 2) consideration of other issues such as the influence of third party candidates, voter turnout increases, and changing precinct demographic and geographic characteristics .
NEDA will release its own analysis of the precinct level Ohio exit poll data on November 2, 2005 in a report, “The Gun is Smoking: Ohio Precinct-level Exit Poll Data Show Virtually Irrefutable Evidence of Vote Miscount”.
The National Election Data Archive (NEDA) a nonprofit organization of statisticians and mathematicians devoted to the accuracy of U.S. vote counts.