*Why it's not a free country anyway
*Examples where the united $tates has not respected majority rule before
*See our article refuting a certain line of approach to seeing Florida as a fraud
*See our article on Zogby as a case of how capitalism influences polling
*About Republican conspiracy theories of exit poll rigging
*Knowing what to ask for in the debate about exit poll stats
with contributions from A Contributor and HC123, November 16 2004
In a previous article, MIM reported that discrepant Florida voting results in 2004 reported in a paper by Kathy Dopp could be explained by the white political fraud known as the Panhandle "Dixiecrat." Six days later the Washington Post and similar papers wrote a story along the same lines, but with less detail, which is why one should always start reading at the MIM website. MIM may have been the first to shoot down the Florida optical scan theory in print, but now we have to report that a different "Kerry won" theory does have convincing evidence that cannot be shot down at the moment.
Since the time of the election, no one is denying numerous reports of glitches, including one in Ohio that garnered Bu$h over 3000 votes. Greg Palast has also discussed the Florida-2000-redux with the uncounted paper ballots that are "spoiled" but come disproportionately from oppressed nationality voters. The question remains whether these glitches add up to a stolen election where some glitches have been found and some yet to be uncovered. Statisticians have become involved because of 1) the exit polls with samples of the voting data 2) a desire to predict voting patterns based on voter registration, race etc. as a backup check on the system's integrity. The statisticians are well-placed to be able to answer whether or not glitches added up to a bigger story.
MIM can confirm all of the following aspects of "Kerry won" theories.
1) Exit polls have been used before by governments including the u.$. government watching other countries to check on whether someone rigged voting. Major imperialist news outlet US News & World Report already speculated prior to the u.$. elections how electronic voting can be rigged--in Venezuela (which of course means it can also be done in the united $tates). We would only that not only Bu$h himself, but some governments other than Uncle $am would have the ability and motivation to rig the results.
2) Conventional bourgeois wisdom and MIM wisdom says no one would bother trying to fix Massachusetts against Kerry or Texas against Bush, because it would be too risky. Neither side claimed before the election that Bush was going to win Massachusetts or that Kerry was going to win Texas. There were only a minority of "battleground" states, and those who do not know that just were not paying attention before the election.
3) Pundits and paid professionals in both the Democratic Party and Republican Party said before the election that it would come down to "FLOHPA," meaning Florida, Ohio and Pennsylvania.
None of this so far can be contested by anyone credibly.
Here is where the conspiracy theory comes in and it is misleading many people on the pro-Bu$h side. The question is how an election would be rigged. It's important to understand that if we do not answer the question, we cannot run a statistical test on it. Without a statistical test, there is no way to compare the various forms of data and argument descends to anecdotes on both sides.
At the moment to really oppose a "Kerry won" theory with evidence, we see no recourse but statistical tests the likes of which have circulated on the Internet. There simply is not an airtight and centralized voting system in Amerika, and we at MIM even doubt that Amerika has the political umph to ever have one with all the civic and computer training it would require to have one. Furthermore, we at MIM have seen Uncle $am interfere in too many elections globally to believe that some of that might rub off at home. Hence, statistics are the only recourse to answer the question of whether or not discovered election glitches are the tip of the iceberg of systematic fraud.
If we agree that trying to fix Massachusetts, Texas, etc would be too bold and risky, then statistical testing should focus on the battleground states. Given that the campaigns talked about the battleground states the whole campaign, and that most states did not enter any discussion, the proper test of a rigging theory is in those battleground states. Whether pro-Bush Republicans would admit this or not, MIM must report to its readers that in fact, constructing a battleground state rigging is the most reasonable a for a conspiracy theory. The US News & World Report article on Venezuela already raised that very assumption in that country, prior to November 2nd. (This is another way of saying that if US News & World Report were consistent, it would be on this story and testing the evidence for a battleground state vote rigging.)
Factually, MIM can confirm the following about "Kerry won" theories:
1) Exit polls did favor Kerry all afternoon, especially in the battleground states-- and people such as Bush Sr. were aware of them. The reactionary "Drudge Report" put up a web page to deny the importance of the exit polls. Of course, it is important to deny their significance, because people who want to vote should vote anyway, and not give up just because Kerry won all the eastern time zone battleground states (as it appeared at that time.) (We oppose voting but we also oppose pragmatism, an unprincipled approach. Those who wanted to vote in principle should have voted and had no impediments. MIM's interest in this question stems from a desire to know the exact nature of the political system and to speak against injustice wherever it may be.)
An accounting of the elite's view of the exit polls election day is here:
2) Opponents of the exit polls put forward their own conspiracy theory which says that Democrats told people to go answer interviewer questions to stack the exit polls. Needless to say, a conspiracy theory to rebut a conspiracy theory leaves the country no where and no one has offered any evidence for the Republican version of conspiracy theory that adds up to an overall story.
3) Scattered reports about the gender of respondents at polling places do not add up to an overall statistical test or reason to dismiss exit poll data especially when the exit polls of several battleground states work together in the same statistical test.
The New York Times itself wrote repeatedly in its story November 5 titled "Report Says Problems Led to Skewed Surveying Data" by Jim Rutenberg that the report of the polling consortium "didn't have a full explanation." There were only ad hoc and post hoc stories of why the National Election Pool survey supposedly failed. MIM strongly suspects that the journalistic organizations lack the statistical competence, organizational attention to statistics and statisticians with sufficient courage to consume their data correctly. If a consortium failed, it would require outside expertise to be able to see the motivations or incompetence involved if any. See our story on what to look for as a statistics consumer in the current controversy.
An example of the assertion of gender skew along with many other brief and unsystematic assertions is here:
At what time in the exit polls did they have a sample of 59% wimmin is the appropriate question. We at MIM rather doubt that Zogby would have posted his poll results after 5pm on his website and left them there all night when he had a sample with 59% wimmin. For such to leak out at 2pm November 2 would not surprise us, but to see such a result on a web page at 5pm and after would surprise us. We would have to buy a conspiracy theory of partisan Democrats' trying to cause some voters not to vote by posting such results that showed a Kerry blow-out. We're certainly prepared to adopt that conspiracy theory about Zogby, but we need the evidence first.
4) We have heard the assertion that late day voters are more Republican than early day voters, and we have some evidence for that which would be especially relevant by 2pm but less relevant as time goes on. For discussion of the "time of day" effect on polls and the logic or lack thereof in a conspiracy to rig exit polls, see another article on that subject.
5) We have heard the assertion that Republicans do not like to answer polls, but we have not seen systematic evidence for that being relevant in this instance of and we have also not seen how it is possible that pollsters would allow that to happen by the time they posted their late returns. People like Zogby would notice, so we are left with a conspiracy theory about Democrats again.
Zogby's motivations are not transparent. If he has substantial clients who would want him to perpetrate a conspiracy to depress Republican voting in the western time zone, then that would make sense of the suggested conspiracy--whether those clients knew what they were doing or not. If he is trying to gain new clients, then posting that Kerry was going to win with over 310 electoral votes on election night was not going to help him in his bid for credibility. Hence, if there was an underweighting of Republicans in what Zogby was looking at, he would want to know.
The most substantial "Kerry won" argument still standing is that the combined polling results of the exit polls in the battleground states compared with the results are implausibly far apart.
An example of this sort of argument is here:
The unexplained exit poll discrepancy
We found two other similar discussions and none of the arguments in this vein
faced a decisive rebuttal in MIM's analysis. It means that the struggle in this arena should continue.
The samples involved were simply too large and point rather decisively to either a conspiracy by Democrats in polling or a conspiracy by Republicans to steal the election or both conspiracies simultaneously. None of the rebuttals of the "Kerry Won" side available as of November 16 made any argument on this except to question the reliability of the exit poll data. In other words, none of the critics perceived how strong the data was and they all assumed that which was to be proved--that the election was fair. It's not enough to say one persynally did not see the exit poll data. (MIM can vouch that it did.) Many did and that is not disputed credibly. A full criticism of the exit poll data would be proof of a conspiracy or an unparalleled breakdown in the exit polls with a connection to how that breakdown influenced the poll. We should warn in advance that the breakdown theory of the exit polls is highly unlikely, because it involves too many states, not to mention precincts. Nonetheless, the best attack on the exit polls should come from the exit poll data itself, but it has not been released to the public.
If Zogby-International does not answer the clamor regarding its results, we will be more inclined to accept the conspiracy theory about Zogby. MIM is quite sure that it knows better than to release results based on a sample of 59% wimmin. Likewise, a proper rebuttal with systematic data seems to be within the reach if not the competence of CNN that bought exit poll data. MIM is not saying the number crunching is finished by any means, and even Ohio's provisional and absentee votes are only just now be counted, but
Here we will also mention the Cal-Tech/MIT study for the vote project called "Voting Machines and the Underestimate of the Bush Vote."
1) The Cal-Tech/MIT study for the vote project (VTP) asserts but does not show anyone how it is that the exit polls are not "designed to predict the outcome of the election." We have not seen any papers in connection to the original setup of the polling consortium that did the exit polls and VTP did not quote any.
Even if the consortium did not tell its journalistic clients of any intention to serve as a piece of a historical conflict over the nature of the voting system when it set up the system, the point would remain that to settle an argument over voter fraud given the existence of proven individual discrepancies, the exit polls are the best or second-best available option depending on whether it is better than prediction by demographic and party registration characteristics. (Ordinarily we would expect the exit polls to be better.)
It is also a historical fact that exit polls have been used before in just the manner they are being discussed now. Those of us with long TV memories remember their being used quite well till the Florida 2000 debacle. US News & World Report and critic of the "Kerry won" theorists Dick Morris have both admitted that the exit polls have been used in the immediate past by Republicans in charge of the administration for verification against fraud. The fact that Amerikans are not used to thinking of themselves as living in a Banana Republic requiring such verification does not change the history and it does not mean that the correct answer is to use the vote results to refute the poll results or to adjust the poll results till Bu$h is shown a victor.
2) While the Cal-Tech/MIT "Voting Technology Project" admits that the discrepancy between exit polls and the final vote tally was far outside the margin of error (or confidence interval that one would normally set up), "Voting Technology Project" still takes a myopic empiricist approach in which it tests for the idea that someone fixed every single state, instead of just the battleground states. Hence, the VTP did not rebut the most substantial "Kerry won" argument.
While admitting that the Bu$h vote exceeded pollsters' findings (CNN, Zogby etc.), the "Voting Technology Project" finds that the overall vote seems reasonable. In other words, VTP understands that the overall discrepancy between exit polls and voting results is absurd, so it takes recourse to the individual results. The real question is whether the battleground states dragged up the whole average for the discrepancy, and based on the VTP's own data, the conclusion would have to be "yes," another disturbing fact that comes to light from examining the VTP paper. MIM had not seen any overall data indication of a battleground versus non-battleground discrepancy prior to the VTP report. VTP's finding that several exit polls underestimated the Kerry vote actually reinforces the "Kerry won" conspiracy that the fix was on in the battleground states, because it tends to show that overall the exit polls were not obviously systematically skewed. (MIM has not seen a statistical test on whether there is a difference in the skew between battleground and non-battleground exit polls. We will report it when we see it.)
The VTP results unintentionally suggest that when we exclude the battleground states, we will see approximately equal numbers of state exit polls underestimating the Kerry vote and underestimating the Bush vote. This fact reinforces the "Kerry won" vote-rigging argument. Despite that fact, it may still be true that a weighted average of those results would point to a Democratic conspiracy to rig the exit polls even in non-battleground states. It is possible there are two simultaneous conspiracies and even that can be measured and tested.
3) The "Voting Technology Project" found no evidence for a fraud particular to voting types (paper/scan/touchscreen), as MIM argued previously in its article on Florida and optical scanning. From previous arguments about Venezuela, it had already come to light that the mechanism to cause voter fraud through electronic balloting already exists. The only question is whether someone exploited that mechanism and how many times in how many places.
As of this date, like VTP, MIM has also not seen any evidence that successfully distinguishes along the lines of paper/optic scanning/touch screen or voter prediction based on demographics. While the argument for a fix in Florida certainly has a surface plausibility to it, we agree with the "Voting Technology Project" thus far on this question. As far as we can see in looking at some data in Florida and Pennsylvania, this question stands confounded with geography and the urban/rural distinction.
4) The VTP called for the polling consortium to release the data. MIM agrees.
Based on the above, MIM reports the following:
1) There is no convincing evidence or argument yet available to rebut the "Kerry won" theory based on the combined results of battleground state exit polls. The VTP paper on the subject is in fact self-contradictory.
2) Some of the arguments used against the "Kerry won" theory are also "conspiracy theories," but with less evidence to back them.
3) Kathy Dopp succeeded in bringing the energy of many to bear on the topic, but her original paper focused on the difficult job of proving the particular mechanism of conspiracy. There were not enough touch screen precincts in the rural areas to prove her point either way, and so the rural Dixiecrat explanation fit the data more generally. After Kathy Dopp, others have adopted a less ambitious approach asking whether in fact there is a discrepancy in the overall data that a statistician can discern. The answer is "yes."
4) We cannot rule out yet that there were two conspiracies--one by Democrats to rig the exit polls and one by Republicans to rig the voting in the battleground states. There are appropriate statistical tests for that as well. At first glance, it appears to MIM that exit polls in fact support this dual-conspiracy theory. If not, a likely outcome would be that there was no conspiracy by either side in the non-battleground states.
Given the exit poll data that ultra-Democrats saw on CNN till 1:30 a.m. and Zogby's web page as of 5pm election day and long after, ultra-Democrats have a basis for reasonably concluding that someone fixed the battleground states. MIM joins the call for a release of the exit polling data as it occurred in stages and a public discussion of the methods involved to replace the anecdotes and speculations surrounding the exit polls now. If the government has to pay Zogby and the exit polling consortium to release the data publicly, it should.
[If you have seen any convincing evidence regarding these issues and would like it added here, you can write to [email protected]]