The following exchange of letters appeared in TomPaine.com on 5/20/05, ten days after Russ Baker's original article, and buried in a section that few readers see. It is reproduced with permission.

What I hope people will take away from this the exchange: it is not unusual that two or more competing models offer an explanation. With the Ohio exit polls, scientists and academics are trying to unravel what happened. This takes time. On the other hand, there are people who lack the qualifications to understand the issues without translation. Those people tend to inflame, rather than inform the debate.

The basic issues, as I see them, are these:

• One could explain the Ohio exit polls either as evidence of fraud or as incompetent exit polling.
  

• The person who could do the most to resolve the issue, Warren Mitofsky of Edison/Mitofsky exit polling has not made the data freely available. Failure to share data raises alarms among those who have dealt with intellectual fraud before. That concern may not be justified, but failure to share data inevitably raises suspicions

• The main model explaining the exit polls as pollster error (called the "Shy Republican" or "Reluctant Responder" model) has problems, points that are not easily explained.

• There are reasons external to the exit polls, such as the Conyers Report, that make fraud a more plausible explanation.

Fraud is not likely to be regarded as proven unless someone comes forward to confess. That's unlikely. We only know about Iran-Contra because a plane crashed. But circumstances do not allow us to remain indifferent or passive. Each of us has to reach a determination as to what happened and act accordingly. Even absent complete evidence, we have to decide which side we are on.

I believe that our freedoms, especially the freedom to choose our leaders based on facts and not smears, have been stolen from us. This exchange of letters is one small part of making the case, hoping to persuade you, the reader, that justice is on this side, and not on the side of ignoring questions of electoral fraud. But, please, choose a side.

I think Russ Baker's reply illustrates my point about inflaming, rather informing. Unlike him, I want my opponent to have his say, because he makes a better case against himself than anything I could say.

After reviewing his reply, readers may enjoy reading our FAQ, point #13 on why Using a Pen Name is a Sign of Moral Turpitude. Baker evidently doesn't know the joke's on him.

Exiting The Polls

Re: What Didn't Happen In Ohio by Russ Baker

Let me say in advance that I do not have a dog in this fight. One of the assumptions implicit in Professor Steven Freeman's analysis of the exit polls is that the polls were competently designed and honestly reported. When I read the Edison/Mitofsky report and tried to coax his organization into facilitating access to data, it became evident that there was no basis to assume that. This is not to detract from Freeman's analysis, which is probably right, but merely to say that I would want to scrutinize Mitofsky's data and methodology much more closely before drawing conclusions from it.

The journalistic deficiencies in Baker's article are capsulized in one simple observation regarding his methods of reasoning. Mr. Baker assures us that the academics who believe the exit polls indicate a Kerry win are insufficiently trained in the science of poll interpretation.

So, who are the paladins he trots out to defeat their arguments? Warren Mitofsky, whose company and future stand or fall on the work under discussion. A psychology graduate student who has written a children's book. A bachelor's level graduate with some graduate coursework in polling (and many years working as an employee of polling firms). And someone with no relevant credentials (but plenty of chutzpah), Rick Brady (resume ). Brady's work  actually contradicts Baker's implied characterization of it, saying "Dr. Freeman is right in concluding that statistical explanations of the discrepancy of the discrepancy to date are inadequate." 

Junior people can, of course, be right when senior people are wrong. Mr. Baker's trump cards, however, deliver an underruff.

As for Liddle's work , let us stipulate that it is an original, interesting, and positive contribution that deserves to be recognized in any debate of exit polling. However, as she concedes in a footnote, her work does not refute the possibility that the vote was miscounted. It only provides an alternate explanation.

It is in the body of the work that her lack of experience shows.  She makes a great deal out of the fact that "Whole Precinct Error" (WPE) is not perfectly independent of precinct partisanship. A secret that one learns only outside of the classroom is that real-life data are not usually as obliging as those in textbooks. Confounding of the variables is not unusual, a fact that Liddle might have learned from her recent experience on the Florida vote , but evidently has not. Rather, contrary to the available evidence (combined response rates in strong Kerry precincts were actually lower than in strong Bush precincts), she accepts as plausible the notion that Bush voters were less likely to respond to pollsters. Since the quality of the data is uncertain, her model is not automatically disqualified, but she needed to do a good job of explaining the point, and didn't.

She also failed to do checks of her model that an experienced researcher would have done. One (extremely obvious) check is looking at more than one state. For the limited data she analyzed, as Josh Mitteldorf of Temple University pointed out to me, her model predicts that WPE will go to zero at both extremes of partisanship. In the real data, this does not appear to be the case in strong Bush precincts. In those strong Bush precincts, the median is much higher than the mean. In layman's terms, this means that there are a few precincts with very high WPE, and some with very low WPE. Low WPE precincts would not contradict Liddle's hypothesis, but the high WPE precincts suggest a question.

I think that the vice president of USCountVotes, Bruce Odell may have answered the question Liddle should have asked on WPE by noting that any model should explain not only the WPE, but also the response rate and any differences between mean vs. median. Out of courtesy to Odell, I will leave it to him to present his results at the opportune time. But neither Liddle nor Odell is the point of this letter. Apparently, Russ Baker lacked the courtesy to talk to USCV and he can't plead a lack of journalistic experience. In his case, it comes across as a lapse of journalistic integrity.

The sense that Baker is going outside the ethic of normal journalism is amplified by a number of other troublesome aspects of his article. He said he did "spot checks on a few popular fraud anecdotes" to conclude that there was no serious problem with the election. Evidently, he missed the Conyers report, which questioned whether Ohio's electors were lawfully chosen.  Baker's main thesis, the notion that those of us who question the election should shut up, is based on an appeal to anonymous authority ("many Democratic officials, election reform advocates, and analysts"). Appeals to authority are always questionable, but appeals to anonymous authorities are laughable. Using writers of limited credentials to disparage or belittle writers of greater credentials is an ugly tactic, one employed by proponents of so-called "intelligent design" and opponents of the theory of global warming in their war on science.  Misrepresentation of Rick Brady's actual conclusions was egregious, and should have been caught by an editor.

The interpretation of the exit polls will be argued out over time. As is common in these situations, more than one model may fit the data. Reaching a fair evaluation may require looking outside the limited set of data Warren Mitofsky has made feely available. Russ Baker and TomPaine, however, have taken a pre-emptive strike against the truth by rushing into print with a weakly founded and poorly researched article. In this, they have politicized what should be a scientific issue and done much harm . . . especially to their own reputations.

Oliver Dawshed

Oliver Dawshed acts as an informal advisor to USCountVotes.org, the organization whose work on the 2004 exit polls was attacked in Baker's piece. This letter is strictly his own.

Russ Baker replies:

Mr. Dawshed criticizes me for citing "many Democratic officials, election reform advocates and analysts" who aren’t convinced from the evidence that a large-scale “theft” occurred in Ohio - as opposed to individual acts of incompetency, obstructionism, etc. “Appeals to authority are always questionable, but anonymous ones are laughable,” he declares.

Well, that’s a good one. Because “Oliver Dawshed” is not this person’s real name. For whatever reason, the person behind the fictional “Dawshed” employs a pseudonym. So much for laughable anonymous appeals.

On to other matters. In a widely circulated e-mail, Mr. “Dawshed” accused TomPaine of not publishing his original letter. Yet TomPaine never received that letter. The very fact that Mr. Dawshed would rush to an incorrect conclusion about TomPaine’s desire to censor debate - and distribute that allegation without even checking first - tells you something about him and his methodology.

What about bias? Dawshed says he does “not have a dog in this fight.” Read on to his bio following the letter, where casually mentions that “acts as an informal advisor to USCountVotes.org.” No dog in the fight? It’s a Doberman.

Dawshed criticizes me for not calling the USCV folks for comment. This perplexes me. USCV has already published their material and circulated it widely. In my writing in TomPaine, which is principally a journal of opinion and not a daily newspaper, I remark upon USCV’s work. There’s no obligation in that format to call for comment, and, given the extensive USCV argumentation on display, no apparent point. Finally, I’d just note this hypocrisy - neither “Dawshed” nor his cohorts who write about me have ever called me to ask any questions either.

Dawshed says that the work by Brady, whom I cite, is actually at odds with my own point. Hardly. Brady says only that USCV’s Freeman “is right in concluding that statistical explanations” offered thus far about the discrepancy are inadequate. By this, he doesn’t mean to agree with Freeman on anything of substance, just that he feels Mitofsky’s people need to more extensively document their claims.

Regarding Warren Mitofsky: They suggest that he is part of a deliberate cover-up. Here’s their proof: He blames the gap between exit polls and the final tabulated result on screw-ups with his own product rather than “fraud.” Rationally speaking, if Mitofsky’s future really depends on the outcome here, as Dawshed says, then he should be arguing for the accuracy and quality of his work, which he is not doing. Perhaps Mitofsky is actually an honest man. Now that’s a shocking notion.

Since the editors at TomPaine asked that I be brief, I won’t go point by point over all of the aspersions Mr. “Dawshed” casts on me and on those whom I believe offer credible critiques of the conspiracy theory. As for his contention that I misrepresent aspects of their analysis, I’ll just note that none of them has contacted me to say so.

Finally, let me urge open-minded readers to focus on the bottom line: Those who attack me because I say I don’t see evidence of a conspiracy to steal the election through either computers or some other means are not interested in a fair and open inquiry. They are angry, and no one - even a friendly skeptic - can be spared.

http://www.tompaine.com/articles/20050520/playing_fair_your_letters.php
© 2005 TomPaine.com
( A Project of The Institute for America's Future)

or

Copyright, © 2005 M. E. Cowan and Oliver T. Dawshed. All Rights Reserved.
Permissions for reproduction or extensive quotation may be obtained from
     or