Notes from the Underground
Why the 2004
Election Matters More than Ever
Part 1: The Meaning of the Legend
“Scoop” Independent News
Part 1 of series
Election 2004: The Urban Legend
Imagine a night at the theater.* A magician comes on stage with a corpse in tow. A doctor from the audience confirms that it is in fact the very real human corpse of a middle aged white male. The magician passes his hand over the corpse just once. It gets up, dances a gig, and leaves the stage. The reanimated middle age man who was once dead returns for an encore.
You’re aghast! You go back stage and confront the magician, “How did you do that?” The magician responds sincerely, “I have no idea.”
Does that make him a magician?
Now imagine that after you question the entertainer, he rolls out another corpse, which is undoubtedly a quite dead middle aged white male. The magician says, “Just pass your hand over the corpse once.” You do, and the corpse arises, dances a gig and leaves the dressing room asking the first person he sees where to get a cab.
Does that make you a magician?
On election night 2004, the networks came on the air and announced that George W. Bush had won the presidential contest to become 43rd president of the United States.
Earlier in the day, there were leaked reports revealing the results of the networks’ own exit polls conducted by a distinguished polling firm. The reports had the White House in a panic. Bush was sure to lose given the trends. According to the exits, he was losing his base, the rural segment of the population that had carried him to victory in 2000. Turnout in the Republican suburbs was not much greater than in the country as a whole, and new voters were going for Kerry 60% to 40%.
The final leaked poll was enough to bring broad smiles to the faces of Democratic leaders and committed campaign workers who had gathered in union halls and hotel ballrooms across the nation.
Then, as if by magic, the 11 pm Election Day vote tallies told a different story. These were accepted by the network reporters as an ex cathedra dictate from the American electorate. We were told that the pious Red prevailed once again over the decadent Blue, a replay of 2000 we were told. Bush was reelected.
The optimistic mood of the Kerry campaign and Democratic faithful was crushed in the twinkling of an eye. What happened? What about the exit polls?
The still corpse of the Bush campaign had been reanimated. It arose from the death of certain defeat, danced a gig, and trotted off center stage to do its considerable damage for the next three years: death and destruction in Iraq; dismantling of the United States Constitution; the abandonment of Katrina’s survivors (for all the world to see); augmented by an impressive and elaborate parade of other calamities that are all attributed to this feat of magic on election night.
Steve Rhodes Creative Commons
How Did They Do It?
Quite simply … by magic. According to the final exit poll, the only heat Bush won, he had two million less votes in the rural segment of the population. That segment went from 23% of the electorate in 2000 to 16% in 2004. Bush made marginal gains in the suburbs. He was headed for disaster rolling into the cities. He picked up steam in cities with populations 50,000 to 500,000, by breaking even in 2004 after a 17% loss to Gore in 2000.
But it was big city dwellers that passed their collective hand over the Bush corpse and brought it to life. He was on life supports before the big city totals were factored in. All that Kerry had to do was match the Gore big city percentage and he would be the next president. According to the day after election final exit poll, big city turnout was up 66%, Bush votes increased 153% (Fig. 5) over 2000 there, and white voters (Figs. 6 & 7) in big cities went from five million in 2000 to nine million in 2004.
Had some would-be campaign operative passed his hand over our largest cities and reanimated white males in sufficient quantity to save the seemingly doomed Bush and doom the rest of us?
Click for big version
Election 2004: The Urban Legend
On June 13th of this year, “Scoop” Independent News published Election 2004: The Urban Legend. I wrote the article based in large part on the research of Internet poster Anaxarchos. The figures cited above from the final national exit pool exit poll and the absurdist conclusions forced from those figures demonstrate that there is no reason to have faith in the final poll result and, as a result, no reason to believe that there is a coherent narrative to justify the election results and the Bush victory.
We’re like the incredulous audience member who went back stage to confront the magician. Even though we can do the trick ourselves by passing our hands over the questionable reported results and the final exit poll to justify continued political life to someone who looked like a sure loser, there’s a foul magic to the process.
Where are the Critics of the Urban Legend?
When the article was published, it received wide spread attention across America’s only uncensored news source, the Internet. Multiple sites posted the article in full, not a common event for a 7,500 word analysis. Major figures in the free and fair elections movement provided their endorsement including Mark Crispin Miller and Ernest Partridge.
We anticipated a full scale assault from friends of the network’s long time polling company, Edison Mitofsky (EM). Nothing much materialized. This was surprising since our reporting and interpretation of the network – EM presentation of the urban results dooms that poll to the status of a failed effort, at the very least, and, more likely, one of the biggest ever failures in public opinion polling.
Anaxarchos Responds to the Missing Critics
Recently, I received a letter from Anaxarchos containing his remarkable comments on the few criticisms offered and, more importantly, an elaboration on the initial article. I’d encourage you to read the full letter (see Appendix) as well as this article.
Anaxarchos: “Having looked carefully at the critical
reviews, it appears to me that your critics have entirely
missed the import of your piece and its underlying analysis.
I could review many of the subsidiary points they raise, but
that seems unimportant compared to the two larger points
that they don’t mention.”
He’s correct. Those who ridicule critics who question the results of the 2004 election were restrained to say the least. This was surprising. The Bush defenders have left no criticism of the election results unturned, particularly those related to the exit polls. Why the restraint?
There were no substantive responses to Urban Legend because there could be none. The claim that turnout in the big cities (500,000 or greater) went up 66% was demolished entirely through simple political commentary. Why would urban residents’ turnout in waves propelling Bush to victory when the rest of the country was only at a 16% increase in turnout? What had Bush done for them to justify this first ever rousing level of support? More importantly, when in our history did an incumbent president lose share and actual votes in his strongest area (in this case, the rural segment) and gain steam and secure an election victory in hostile territory (the big cities)?
The claim of the 66% increase in turnout was also put to a final rest by the incorporation of actual city turnout data made available on election night and finalized shortly there after. Specifically, actual city voting results showed that city turnout increases were only about 16%, (Chart 1) the reported average for the country. These big city results were, in some cases, reported on election eve by the very networks that paid for the exit polls and by the exit pollsters who claim to reconcile their final results to the election results. One must wonder if the right hand was giving to the left the full story.
Could the polling
company and their sponsors, the major networks (plus CNN and
the Associated Press) have been this ignorant of what was
happening in New York City? The results reported on local
news outlets owned by the networks showed a 12% increase in
turnout? That’s 54 points below the claimed urban increase
of 66%. New York is, after all, the headquarters of the
television network poll sponsors and near the headquarters
of the polling company. Did they simply ignore these
results in their haste to produce their version of the final
exit poll the day after the election? And why wasn’t
there any comment on the more than obvious disparity between
the actual results for big cities, particularly on turnout,
and the polling results they continued to show long after
the certified vote count for big cities became available to
everyone. This is a critical question addressing the
integrity of the entire exit polling and reporting process
The Entire Narrative of the Election
Anaxarchos elaborates the first big error of the exit pollsters and network consortium
Anaxarchos: “It seems to me that the
most important implications of “Urban Legend” are these:
1) The entire narrative of the 2004 election is built on the foundation of the exit polls. There is virtually no other real-time source of data on who voted how, why, and where. Indeed as the critics of the use of exit polls for fraud detection have pointed out on many occasions, this voter survey is precisely what the exit polls are “intended” to provide, and why they are funded by the consortium of media outlets, the NEP. The Charlie Cook reference in your piece was typical. The Exits provided the sum total of the data behind his analysis of the election.”
Based on the final exit poll two distinguished analysts, Charles Cook and Ruy Teixeira stuck their necks out in different directions. Cook called the Bush victory a display of political genius and immediately made a fundamental mistake. He claimed that defections from the Kerry camp by black, Latinos, and Jewish voters had done the trick for Bush. Had he examined the data available at the time, he would have known that there were only marginal changes in these groups. Teixeira was more precise as Anaxarchos points out:
“Unfortunately, so committed was Teixeira to the
impossibility of widespread election fraud, that he assumed
that there was disconnect between urban data as the NEP
defined “urban” and county data, with the observation
that, “urban doesn’t mean urban and rural doesn’t mean
rural”. Teixeira promised a detailed county analysis to
reconcile the differences. Of course, no such
“reconciliation” was forthcoming. My guess is that
Teixeira, like Cook, underestimated the magnitude of the
“reconciliation” that would be required and also
underestimated the final turnout of the 2004 election which
only further widened that gap.”
One of the most astute analysts, Cook, jumped to the self-informed conclusion that the Bush urban victory had to be due to a shift in ethnic voting. It’s easy to see why. He was unaware that the white big city vote increased from five million in 2000 to nine million in 2004. We can suppose that it never occurred to him that such a thing could or would happen. Why would we expect him to check the exit turnout rate against actual city voting totals?
Teixeira’s response and follow up are even more perplexing. He’s the author of The Emerging Democratic Majority and a recognized polling expert. After dropping his confusion of terms argument, he promised a county analysis to show how Bush won, a common response of establishment Democrats. But he never produced the study? Why? Maybe he stared into the abyss and the abyss stared right back.
He dismissed claims of fraud based on exit poll analysis by writing “… it is possible that the magnitude of these corrections has been greater than normal.” That depends on what your definition of normal is. What’s normal about increasing turnout by a factor of four (16% actual to 66% claimed) to achieve an absurd result? The basis for the urban data correction (actual city results) was available when he made this statement. Had he bothered to look? We’d like to hear from him on this and the questions we outlined clearly in the original article (presuming he’s given up his role as a Democratic apologist for questions about Bush election integrity).
So what does this mean?
Anaxarchos: “ It means at a
minimum that either one must try to support the indications
of the Exit Polls that the Bush winning margin in 2004 came
in the Urban centers, implausible as that seems, or one must
craft a new narrative of the 2004 presidential election.
Believe it or not, the former option is not nearly as
difficult as the latter. Your critics have missed what it
means to simply declare that “the Exit Polls must have
been wrong”. With that dismissal, much of the supporting
evidence for how Bush “won” in 2004 disappears as
For over 30 years, the way we’ve made sense out of “who voted where and why” is through exit polls which are designed to and accepted as answering those very questions. There have been few complaints, other than Florida 2000 when the exit poll showed a narrow Gore victory. Given the trashing of 100,000 mostly minority spoiled ballots, who could criticize the pollsters if they initially showed a Gore victory as a result of interviewing voters in minority precincts whose ballots had been “spoiled.”.
If we don’t know how Bush won, ratifying the election results is mindless magic. If we don’t demand an understanding of how he won, then can we dismiss the notion of election fraud made over and over with to an ever widening and receptive audience? Are elections the one area of administration activity that escapes critical analysis? Perhaps the election fraud doubters have been listening to Alberto Gonzales and his crew on these questions.
Anaxarchos offers a compelling case for the election polls failure across the board, not just in the big cities.
“Consider the following:
If the Bush winning margin did not come in the cities, where did it come from? If the urban vote as reported by the Exits is incorrect, then the remainder of the Exit Poll narrative must also be incorrect. It is true that the big city vote underlines the anomaly but take a look at the three-category demographic (Urban, Suburban, and Rural) and you get a slightly more muted version of the same story. If the cities don’t hold Bush’s winning margin, then that clearly means that it must have come from somewhere else. While the erosion of the Bush rural margin is significant, reversing it is not enough. We must also “offset” the loss of Bush’s urban margin in the suburbs and we must do this while constantly living under the overhang of an 18% increase in turnout (which clearly favored Kerry). The result is that the Exit Polls must not only be “wrong” in the cities, they must also be “wrong” across the board and this to a significant degree. In truth, the degree of this “wrongness” must increase as we go from city to countryside because, as we have seen, the Exit Polls weight the Bush urban margin into existence.”
Painful choices regarding the outcome of the 2004 presidential election.
We can accept the official election results simply as reported by discarding or denying any and all questions and anomalies. Doing so makes us no better than the uncritical magician in the opening passage. It just happened. We don’t know why. We agree that it doesn’t make much sense but that’s just the way it is (in this best of all possible worlds). Move along.
We can accept the election results and totally dismiss the exit poll adjustments as indicative of a flawed poll that should be dismissed. Our argument here is no better than in the first option. Its faith based. That’s just the way it is but we’ll discuss it a bit, feign erudition, and impress you with our obscure knowledge of polling methods and math.
Or we can face the reality and the dreadful conclusion. There’s no way to tell if Bush truly won the vote total in 2004 while there are many reasons to doubt that he did. The parallel measurement of the actual vote, the exit poll, can only concoct a Bush victory through egregious adjustments to its own raw data for the big cities. Why would such adjustments be required? Was the measurement off for the smaller cities where Bush gained 17 points over 2000? Was it off for the suburbs and rural segment? What about the voluminous reports of voter suppression and voting irregularities across the nation; reports including consistent vote flipping from Kerry to Bush?
If there were no problems with the actual vote count, problems that the exit poll analysis clearly indicates, why on earth would two thirds of Ohio counties destroy the ballots and election records from 2004 well before the required retention period?
And what about this question, perhaps the simplest of all with the greatest potential for understanding just what happened in 2004? Why does the network consortium refuse to release the raw data for 2004? The raw data has been closely guarded by the pollsters and the networks despite at least two requests for examination of this data by now Committee on the Judiciary Chairman, John Conyers, Democrat, Michigan.
Has that data suffered the same fate as the destroyed Ohio ballots?
Would the handling of the raw data that produced this unbelievable narrative embarrass the networks and indicate that they should have known shortly after the election; that they certainly know by now, without any doubt, that there are huge problems with the final exit poll, the poll the national election pool and its polling company have defended to consistently and vigorously?
Or would the freeing of this privately held data concerning our public election show what many suspect: the real winner of the 2004 election is not sitting in the White House.
You can be sure that the four major networks, CNN, and the Associated Press would be in court right now demanding the release of the exit poll data were it any concern other than them holding back the data from the rightful public review demanded.
Permission to reprint in part or whole with a link to this article in “Scoop” and attribution of authorship.
Appendix: Full Letter from Anaxarchos to Michael Collins
Thank you for sending me the reviews for your article, “Urban Legend”, and congratulations on the overwhelmingly positive response you have received. I was a little disturbed at the few negative criticisms that you sent along. Having looked carefully at the critical reviews, it appears to me that your critics have entirely missed the import of your piece and its underlying analysis. I could review many of the subsidiary points they raise, but that seems unimportant compared to the two larger points that they don’t mention. It seems to me that the most important implications of “Urban Legend” are these:
1) The entire narrative of the 2004 election is built on the foundation of the exit polls. There is virtually no other real-time source of data on who voted how, why, and where. Indeed, as the critics of the use of exit polls for fraud detection have pointed out on many occasions, this voter survey is precisely what the exit polls are “intended” to provide, and why they are funded by the consortium of media outlets, the NEP. The Charlie Cook reference in your piece was typical. The Exits provided the sum total of the data behind his analysis of the election. Unfortunately, the data he relied on was “implausible” and thus his “analysis” was equally so. Neither was Cook the only one to trip over that anomaly. Ruy Teixeira also noticed the same “implausibility” within days of the election. Unfortunately, so committed was Teixeira to the impossibility of widespread election fraud, that he assumed that there was disconnect between urban data as the NEP defined “urban” and county data, with the observation that, “urban doesn’t mean urban and rural doesn’t mean rural”. Teixeira promised a detailed county analysis to reconcile the differences. Of course, no such “reconciliation” was forthcoming. My guess is that Teixeira, like Cook, underestimated the magnitude of the “reconciliation” which would be required and also underestimated the final turnout of the 2004 election which only further widened that gap. AlterNet.Com
So what does this mean? It means at a minimum that either one must try to support the indications of the Exit Polls that the Bush winning margin in 2004 came in the Urban centers, implausible as that seems, or one must craft a new narrative of the 2004 presidential election. Believe it or not, the former option is not nearly as difficult as the latter. Your critics have missed what it means to simply declare that “the Exit Polls must have been wrong”. With that dismissal, much of the supporting evidence for how Bush “won” in 2004 disappears as well. Consider the following:
If the Bush winning margin did not come in the cities, where did it come from? If the urban vote as reported by the Exits is incorrect, then the remainder of the Exit Poll narrative must also be incorrect. It is true that the big city vote underlines the anomaly but take a look at the three-category demographic (Urban, Suburban, and Rural) and you get a slightly more muted version of the same story. If the cities don’t hold Bush’s winning margin, then that clearly means that it must have come from somewhere else. While the erosion of the Bush rural margin is significant, reversing it is not enough. We must also “offset” the loss of Bush’s urban margin in the suburbs and we must do this while constantly living under the overhang of an 18% increase in turnout (which clearly favored Kerry). The result is that the Exit Polls must not only be “wrong” in the cities, they must also be “wrong” across the board and this to a significant degree. In truth, the degree of this “wrongness” must increase as we go from city to countryside because, as we have seen, the Exit Polls weight the Bush urban margin into existence. In fact, the weightings decrease significantly as we move from more to less urban territory, and this has been previously presented as an indication of the rural accuracy of the Exit Polls in comparison to the tallied vote count. How do we now reverse that? At the very least, this begs for a serious investigation as you called for.
Yet, the story gets worse. The election narrative starts with the vote count but it doesn’t end there. Certainly the accepted narrative of the election, universally reported by the major news outlets, of “values voters”, “security moms”, and the like, all derived from the Exit Polls and all the products of “weighting”, become much less compelling if the overall narrative is undermined. But, this part of the accepted narrative is also the more trivial. There are some much more important implications here. It is not simply that the election narrative based on your unlikely “Urban Legend” is wrong by itself. It also undermines the use of that story to refute competing narratives which were unceremoniously rejected at the time of the election itself. Consider this:”In a stunning admission, an elections manager for NBC News said national news organizations overestimated President George W. Bush's support among Latino voters, downwardly revising its estimated support for President Bush to 40 percent from 44 percent among Hispanics, and increasing challenger John Kerry's support among Hispanics to 58 percent from 53 percent. The revision doubles Kerry's margin of victory among Hispanic voters from 9 to 18 percent. Ana Maria Arumi, the NBC elections manager also revised NBC's estimate for Hispanic support for Bush in Texas, revising a reported 18-point lead for Bush to a 2-point win for Kerry among Hispanics, a remarkable 20-point turnaround from figures reported on election night.
"Latino presidential partisan preferences did not change significantly from four years ago," said WCVI's president, Antonio Gonzalez, in his presentation before the National Association of Hispanic Journalists…
"”But I repeat, NBC has set an example for network poll integrity by taking a giant step away from the Edison International/Mitofsky election results, and toward WCVI's findings. For example, today NBC stated that 70% of its respondents came from non-urban areas and 30% from urban areas, while acknowledging that 50% of Latino voters come from urban areas. This admission could explain the difference in their results and WCVI's. They under-represented Latino urban voters (who are more likely to vote democratic) and over-represented Latino non-urban votes (who are more likely to vote republican). We hope the other networks follow suit with more adjustments in their findings," Gonzalez concluded. HispanicBusiness.Com
According to its exit poll survey, the Institute found that Latino voters supported democratic presidential candidate John Kerry over President George W. Bush by a margin of 65.4% to 33%.”
The problem with the story above is that the real implication of such a sampling error among Hispanics was not considered. According to the “official” narrative, the shift of Hispanic voters toward parity was one of the most important pillars of the Bush “victory” in 2004. From “Urban Legend”, we know that this problem was most likely a weighting problem and not a sampling problem per se. But, the issue is not confined simply to Hispanic voters. Another pillar of the victory was a small but significant shift among black voters away from Kerry, universally reported as an artifact of the Republican use of political or religious “wedge issues” in the election. Perhaps one of the most important facts revealed in the “Urban Legend”, however, was that the Exit Polls reported a 40% increase in the black vote overall in comparison to 2000, but, simultaneously, virtually no increase in the black big city vote. We thus have widespread examples of exit poll responders appearing where they are not: black and Hispanic voters, with more conservative and Republican tendencies popping up in the suburbs, and a mass of urban GOP whites materializing in the cities. But… if these things didn’t happen, how could Bush possibly “win”?
The inverse of this is equally striking. The suggestion from the Exit Poll anomalies, above, is that 2004 was actually a rerun of the 2000 election with 16 to 18% greater turnout. In fact, you also saw that the pattern of both the weighted and unweighted Exit Polls for 2000 and the unweighted Polls for 2004 are remarkably similar. But if this was true, how is it possible for Bush not to lose?
Did the Exit Polls really pick up “ghost voters” in the cities and thus expose widespread election fraud? Who knows? There are actually some states in which the Bush urban margin improves through a process similar to the one implied, by “Urban Legend”. Yet, it is more likely that the Exit Polls picked up an anomaly in the larger election and the attempt to reconcile this anomaly creates the “ghosts”. It is also possible that something completely different occurred which actually gave the election to Bush, but nothing in such an outcome is possible without overturning the Exit Polls in their entirety and creating not just a new narrative for the election itself but also explaining the massive variation of the Polls themselves. To attempt to take any other position is fundamentally dishonest and genuinely “faith-based”.
2) I have already run on too long but, while point #1 above explains the extrinsic implications of “Urban Legend”, there are some intrinsic implications as well. In the spring of 2005, Edison/Mitofsky, the polling organization responsible for the 2004 exit polls, released their analysis of the exit poll discrepancy. Instead of blaming precinct selection or methodology, the polling organization made a spirited defense of both. The alternative explanation was that a breakdown had occurred in the sampling of voters in what Mitofsky claimed were accurately chosen precincts. Since that time, numerous panels representing the statistical establishment have convened and, each time, have supported Mitofsky’s original conclusions. Explanations of various presumed sampling problems, “within precinct errors”, “shy voters” and the like, have been numerous and tiresome. Because of “Urban Legend”, it also seems that these were entirely irrelevant. How is it that the august scholars and expert panels missed the most fundamental anomaly of the urban vote? This isn’t just missing the forest for the trees. This is more like missing the forest fire for the toadstools. I will go into this in some depth in the future if you have an interest.
Note to Anaxarchos: I have an interest. Mike