"
The fact that, in the 2004 election, all voting equipment technologies except paper ballots were associated with large unexplained exit poll discrepancies all favoring the same party certainly warrants further inquiry."
(from summary of
US Count Votes' exit poll analysis - March 31, 2005)
On March 31, 2005 US Count Votes' National Election Data Archive Project released a report entitled "Analysis of 2004 Presidential Election Exit Poll Discrepancies," on the discrepancies between 2004 National Exit Polls conducted by the Edison/Mitofsky Consortium and the official tabulated election results. The report found "the hypothesis that the voters' intent was not accurately recorded cannot be ruled out and needs further investigation." The 27-page report bears the imprimatur of an impressive array of academics including nine Ph.D's associated with major U.S. universities including The University of Pennsylvania, Temple, Wisconsin, Utah, Notre Dame, Cornell, Southern Methodist, Illinois at Chicago, and Case Western Reserve.*
CONTEXT:
As election 2004 approached most of the national polls suggested a razor-close Presidential election. On the eve of the election, polls by Marist, Democracy Corps and FOX all had John Kerry with a narrow 1 to 2% lead in the popular vote, while polls by NBC/Wall Street Journal; Zogby; ABC News; The Washington Post, Harris, CBS News and CNN/USA Today (Gallup) showed George W. Bush with a similar 1 to 2% lead. Gallop's final projection including assignment of undecided voters actually projected a tie at 49% for both candidates.
Most of the polls were showing a general erosion in support for George W. Bush as the election approached. Kerry also began to surge in many of the "battleground" states and, according to several polls, appeared headed for a victory in Ohio and Florida. The poll trends also suggested that the majority of "undecided" voters were breaking for Kerry.
On the eve of the election, the Zogby Poll called the election for John Kerry, giving him 311 electoral votes to 213 for George W. Bush. Zogby showed Bush with a slight popular vote edge of 1%, but called several "battleground" states for Kerry including Florida, Iowa, New Mexico and Ohio (all went to Bush in the "official" count). Zogby found Nevada and Colorado "too close to call" (both "officially" went to Bush).
Considerable controversy had arisen over electronic voting technology which would tally approximately 30% of the votes. The technology, including codes and software was owned by companies with strong Republican ties. Several studies revealed problems with the technology -- including susceptibility to hacking. Many of these electronic voting machines left no paper record, rendering a recount of their tallies impossible. Bills in both houses of Congress designed to mandate a voter-verified paper record were denied a floor vote by the Republican Leadership.
Beginning about midday and continuing into the early evening on Election Day Edison/Mitofsky Exit Polls, intended for release after the polls closed, began to leak. They showed John Kerry ahead in Iowa, Nevada, New Mexico and Ohio with Florida a virtual tie. Zogby had also called all of these states except Nevada ("too close to call") for Kerry. Early exit polls also showed Kerry with a 9-point lead in Pennsylvania which he eventually won by a much slimmer 2-point margin.
"No matter how one calculates it, the discrepancy cannot be
attributed to chance."
US Count Votes notes in its abstract that the "exit pollster of record for 2004 election was the Edison/Mitofsky Consortium. Their national exit poll results projected a Kerry victory by 3.0 % whereas Bush won the official count by 2.5 %." (a 5.5 percentage point variance representing more than four times the 1.3% margin of error). The report notes that several previous analyses had determined that the odds of such a large variance occurring by random chance ranged from 1 in 959,000 to 1 in 1,240. "No matter how one calculates it, the discrepancy cannot be attributed to chance."
"Edison/Mitofsky disavowed the results of their own poll, saying they cannot be construed as evidence that the official vote count was corrupted." They hypothesized that "Kerry voters were more amenable to completing the poll questionnaire than Bush voters." US Count Votes notes, however, that "Edison/Mitofsky's own exit poll data does not support their theory."
Shortly after midnight on election night, Edison/Mitofsky "adjusted" its exit polls to mirror the official tabulated election results. "These adjusted results remain posted, as of this writing," notes US Count Votes. Their report rejects a sinister motive, but notes nevertheless that "the effect of the process was at the very least confusing and served to blunt public awareness of the dramatic exit poll-vote count discrepancies during the critical period immediately following the election. The US Count Votes report notes that "slight" adjusting of poll results is legitimate, but that "the justification for doing so rests entirely on the assumption that the reported election results are in fact accurate, as reflected by a small and undramatic discrepancy between exit poll results and vote counts." Instead Edison/Mitofsky's exit poll results "had to be substantially, in fact dramatically, adjusted. Such substantial discrepancies and need for such dramatic adjustment raised a bright red flag. Edison/Mitofsky ignored this red flag and simply substituted the adjusted data set, which has been generally employed without acknowledgement."
Edison/Mitofsky has to this day refused all requests for the release of its raw data, including a request from Congressman John Conyers, senior-ranking Democrat on the House Judiciary Committee, who requested the data in conjunction with his investigation of 2004 election irregularities.
US Count Votes notes that there are "three primary explanations for the discrepancies" between the exit polls and the official vote tally:
1. Chance. Both US Count Votes and Edison/Mitofsky agree that "random statistical sampling error can be ruled out." The discrepancy is simply too large.
2. Innacurate Exit Polls. This is the theory advanced by Edison/Mitofsky, based on their hypothesis that Bush voters were more reluctant than Kerry voters to respond to exit polls. According to US Count Votes the data that Edison/Mitofsky offered in their report show how implausible this theory is."
3. Inaccurate Election Results. "Edison/Mitofsky did not even consider this hypothesis, and thus made no effort to contradict it" despite the fact that some of their exit poll data "may be construed as affirmative evidence for inaccurate election results. We conclude that the hypothesis that the voters' intent was not accurately recorded or counted cannot be ruled out and needs further investigation."
US Count Votes notes that "after last November's elections there were thousands of reports of irregularities" including: voting machine shortages; ballots counted and recounted in secret; lost, discarded and improperly rejected registration forms and absentee ballots; touch screen machines that registered "Bush" when voters pressed "Kerry;" precincts in which there were more votes recorded than registered voters; precincts in which reported participation rate was less than 10%; high rates of "spoiled" ballots and under-votes in which no vote for president was recorded; etc.
"These problems arose in the context of vote recording and counting systems developed, provided, and maintained primarily by a handful of private vendors with partisan ties, and where nonauditable voting equipment which cannot provide assurance that the votes are counted as cast, tallied about 30% of the national vote."
The US Count Votes report asserts that, in the absence of auditablilty or transparency in our election system, exit polls are of paramount importance as a check of the overall integrity of official election results.
The report explains that Edison/Mitofsky conducted the 2004 exit polls on behalf of the National Election Pool representing major national press and TV news services. The polls were conducted in every state plus a nationwide exit poll (with a substantial sample size of 13,047). Confidential exit poll data showing John Kerry ahead of George Bush in several key "battleground states" were leaked on the afternoon of November 2nd and, as previously noted, projected a 3.0 % Kerry victory which was to materialize into a 2.5 % Bush victory in the official vote count.
While Edison/Mitofsky has refused to make its raw data available, analyses are still possible because of "screen captures" of the raw data published to the internet on election night. The report cites earlier studies of these data revealing statistically significant discrepancies of exit poll results concentrated in five states, including the critical battleground states of Florida, Ohio and Pennsylvania. Two independent studies by Ph.D's in statistics had concluded that the probability of these discrepancies being concentrated in these particular three critical states (rather than in any randomly selected group of three states) is less than 1 in 330,000.
On January 19, 2005 Edison/Mitofsky released a 77-page report that acknowledged widespread discrepancies between their exit polls and official counts, admitted that the differences were far greater than can be explained by sampling error, and asserted that the disparity was "most likely due to Kerry voters participating in the exit polls at a higher rate than Bush voters." US Count Votes' analysis disagreed:
I. Random Chance: Without getting into complicated statistical analyses here, both US Count Votes and Edison/Mitofsky concluded that "random chance as a possible explanation for discrepancies between exit polls and official election results can be dismissed."
II. Exit Poll Error: US Counts Votes notes that "Exit Polling is a well-developed science, informed by half a century of experience and continually improving methodology." US Count Votes notes further that the sample precincts selected by Edison/Mitofsky "accurately predicted the results in their respective states with only a small observed bias (0.3%)." This data reflects positively on the Edison/Mitofsky sample. Edison/Mitofsky hypothesized a significant bias in the sample based on the "reluctant Bush responder" theory, however they provided no evidence to support this hypothesis, and it is contradicted by historical trends and by exit polls from 2004 state primaries. Edison/Mitofsky claimed it could not "measure the response rate by Kerry and Bush voters," but hypothesized a 56% response rate for Kerry voters and a 50% response rate for Bush voters, apparently because these percentages "would account for the entire... error that we observed..." US Count Votes observes that "no data in the E/M report supports the hypothesis that Kerry voters were more likely than Bush voters to cooperate with pollsters and, in fact, the data provided by E/M suggests that the opposite may have been true."
US Count Votes evaluated Edison/Mitofsky data in precincts where 80 % or more of the vote went to either Bush or Kerry. "The reluctant Bush responder hypothesis would lead one to expect a higher non-response rate where there are many Bush voters," yet the E/M data reveals that the response rate was actually higher in the stronger Bush precincts than it was in the stronger Kerry precincts. "This data bears directly on the plausibility of the report's central hypothesis, and it goes in the wrong direction," notes US Count Votes. "This fact undermines the report's central premise that Kerry supporters were more likely than Bush supporters to participate in the exit polls. US Count Votes also notes the fascinating detail that the discrepancy between the exit polls and the recorded vote was dramatically higher in precincts where Bush received 80-100% of the votes ... An alternate theory that is more consistent with the data is that "corruption of the official vote count occurred most freely in districts that were overwhelmingly Bush strongholds." {districts where audits would also be least likely to occur}.
US Count Votes reports that for the Edison/Mitofsky "reluctant Bush responder" theory to be true, Kerry voters would have had to be significantly more cooperative in Bush strongholds than in Kerry strongholds, contradicting previous experience and observations in the 2004 election that voters in a small minority tend to be less willing to respond to exit poll interviewers, not more as the Edison/Mitofsky theory requires. "The required pattern of exit poll participation by Kerry and Bush voters to satisfy the E/M exit poll data defies empirical experience and common sense under any assumed scenario," concludes US Count Votes.
32 states had both Senate and presidential races on a single exit poll survey. If Bush supporters were refusing to fill out the survey the exit poll should have been just as poor in the senate race as in the presidential race. Yet "the senate polls were significantly more accurate."
The evidence, quite simply, does not support the Edison/Mitofsky hypothesis that Kerry voters were more cooperative with exit pollsters than were Bush voters, an hypothesis that is crucial to Edison/Mitofsky's disavowal of the accuracy of its own 2004 exit polls.
III. Inaccurate Election Results. If the discrepancies between exit poll and election results cannot be explained by random sampling error and the "Reluctant Bush Responder" hypothesis is inconsistent with the data, "then the only remaining explanation - that the official vote was corrupted - must be seriously considered" concluded US Count Votes. Edison/Mitofsky never considered this possibility, "always assuming the correctness of the election results without providing supporting evidence for that assumption."
Data provided by Edison/Mitofsky in its report reveal significant differences between exit polls and the reported vote inconsistent with chance where Optical Scanners, Punch cards, Touch screens, and Mechanical voting machines were in use. Only precincts using paper ballots showed results consistent with chance. Edison/Mitofsky concluded that vote machine type was unrelated to the margin of error, but failed to explain how they arrived at that conclusion or whether they drew a distinction between votes tallied in the precinct or at a central location. In the absence of this data US Count Votes points to a limited study in New Mexico that showed that touch screen voting machines "produced significantly higher rates of under-votes in the presidential race than did New Mexico's optical scan machines." "Similar audits of other states' election results are needed" suggests US Count Votes.
US Count Votes concludes that "the many anecdotal reports of voting irregularities create a context in which the possibility that the overall vote was substantially corrupted must be taken seriously. The hypothesis that the discrepancy between exit polls and election results is due to errors in the official election tally remains a coherent theory."
IV. Misleading Use of Adjusted Exit Poll Data. US Count Votes notes that there is a considerable difference between "weighted" data, designed to adjust the data to conform to the demographic composition of the electorate, and for which there is general agreement, and "adjusted" data where the already demographically weighted results are further "weighted" and forced to conform with the election results. As noted earlier, Edison/Mitofsky posted election "adjusted" exit polls to its web site on election night 2004 where they remain to this day. Again as noted earlier, this data is highly questionable because of the extensive weighting required to align the two very different results. Edison/Mitofsky substituted the adjusted data for the weighted data early on "without acknowledgment," thus "perpetuating the confusion and misleading impressions" this substitution created. "We see no constructive reason for E/M's practice in this regard" concludes US Count Votes.
In summarizing its findings, US Count Votes contests the notion advanced by critics that firmer proof is required before a thorough investigation is launched into irregularities in the 2004 election. "We feel strongly that this is the wrong standard. One cannot have proof before an investigation." "In fact, the burden of proof should be to show that the election process is accurate and fair. The integrity of the American electoral system can and should be beyond reproach."
"The fact that, in the 2004 election, all voting equipment technologies except paper ballots were associated with large unexplained exit poll discrepancies all favoring the same party certainly warrants further inquiry."
The report concludes: "The absence of any statistically-plausible explanation for the discrepancy between Edison/Mitofsky exit poll data and the official presidential vote tally is an unanswered question of vital national importance that demands a thorough and unblinking investigation."
One wonders how any one could argue with that conclusion!
___________________________
* Authors and Endorsers of the US Count Votes Analysis:
Josh Mittendorf, Ph.D. Temple University Statistics Department
Steven F. Freeman, Ph.D. Visiting Scholar & Affiliated Faculty, Center for Operational Dynamics, University of Pennsylvania
Brian Joiner, Ph.D. Professor of Statistics and Director of Statistical Consulting (ret), University of Wisconsin
Frank Stenger, Ph.D. Professor of Numerical Analysis, School of Computing, University of Utah
Richard G. Sheehan, Ph.D. Professor, Department of Finance, University of Notre Dame
Paul F. Velleman, Ph.D. Associate Professor, Department of Statistical Sciences, Cornell University
Victoria Lovegren, Ph.D. Lecturer, Department of Mathematics, Case Western Reserve University
Campbell B. Read, Ph.D. Professor Emeritus, Dept. Statistical Science, Southern Methodist University
Jonathon Simon, J.D. Alliance for Democracy
Ron Baiman, Ph.D. Institute of Government and Public Affairs, University of Illinois at Chicago