Letters: Debating Exit Polls, Part 2

January 28, 2005

This week concludes our debate on exit polls in Election 2004. Tune in next week, when letters from our readers re-emerge.

Oh, dear. Rebutting a rebuttal . Seldom a constructive use of time.  But then, to ignore Steve Freeman’s published comments about my TomPaine.com Ohio election research would be to acquiesce in the ongoing degradation of serious debate over real problems that bedevil the nation’s elections apparatus—and to endorse this distraction from fashioning viable systemic solutions.

Freeman is a leading proponent of the theory that the election in Ohio was deliberately stolen, and that proof of this lies both in a gap between exit poll results and final tallies, and in anecdotes of election day irregularities. In my article, I expressed doubts about Freeman’s argument, and in his rebuttal, he dismissed what I had to say.

I won’t go into everything Freeman says, although everything he says is either partially or completely wrong. But I will focus on a few key areas. (If you didn’t see his piece, you can find it here. You can also find my original piece here  or near the top of my own website, http://www.russbaker.com/.)

First of all: credibility. Freeman, a university professor, makes much of his academic credentials. Freeman writes: “A study of election integrity…requires an understanding of election practices and voting systems, and, most importantly, an ability and willingness to investigate a complex subject in which the data and the accompanying official pronouncements are themselves suspect. I hold degrees in both political science and systems science, and have received four national awards for best research paper of the year—on four different topics in three different fields.”

I won’t explore here the limitations of some of my own cocksure college professors, other than to say that academia has as many bad eggs as good, and that mere titles, positions and accolades don’t necessarily qualify someone to declare an election stolen.

Freeman, who selectively cites favorable journalists as credible authorities, seems to summarily ignore my own long, unsullied record of journalistic accuracy. The fact is, absent a Ph.D., one can do pretty well in getting to the bottom of things with any or all of the following: a keen mind, a good nose, a sense of fairness and healthy skepticism, and the ability to apply logic and ask good questions.

Probably the most disturbing part of Freeman’s theory is his myopic understanding of the ways in which presidential exit polls function. He writes: “Baker dismisses the validity of exit polls, but prominent survey researchers…political scientists…and journalists concur that they are highly reliable. As far back as 1987, political columnist David Broder wrote that exit polls "are the most useful analytic tool developed in my working life."

Actually, I never said that exit polls weren’t valid. I said that they are imprecise. And I said that, in exploring the particulars of what happened on November 2 with the people who did the exit poll, I learned about all manner of technical complication that could have affected the numbers and the perceptions—a source of potential problems that Freeman fails to deal with.

The outfit that did the exit polling faced myriad headaches—including apparent quality-control issues with some of their canvassers whose job was to convince voters to voluntarily complete surveys outside polling places. Many of them are hired through subcontractors;  some are more scrupulous in following rules than others; some relate to voters better than others and therefore perhaps get more accurate information. Probably the key factor was the election-day environment in which those people operated—almost circuslike, often combative, with teams of election watchers camped outside polling places wearing NAACP Election Protection T-shirts. It’s foolish or disingenuous to assert the accuracy of exit polls while denying the probability that some Republicans may have felt disinclined to say that they voted for Bush.

The pollsters intend to do what they can to rectify these matters in the future, but they underline a fundamental reality about exit polls—they are surveys. They are not exact replications of actual voting.

Suffice to say that if the author of some study concedes that it was flawed—and if he is considered generally expert and credible—as is true of exit pollster pere Warren Mitofsky—then there is no reason to insist that he is covering up some hideous plot. As it happens, besides being universally trusted and respected, the pollster is personally a lifelong liberal who had no use for Bush.

Freeman et al miss another key reality about exit polls. You can’t say that the exit poll results were right, in that there are no actual “results.” The canvassers working for the exit pollsters go to selected precincts, not to all precincts, and so whatever they report must then be massaged through a variety of processes designed to correctly adjust and extrapolate in order to ensure that reported totals accurately reflect the will of the voters. Therefore, what the Mitofsky people were doing was sending across, privately, to their media clients, a flow of numbers reflecting different calculations and variables. The numbers that Freeman and his acolytes have spread, virus-like, through the Internet are but a small drink from a steady, evolving stream from Mitofsky.

Most importantly, exit polls aren’t intended to be a check against election fraud. They are attempts to predict the final outcome of the voting. In order to improve the accuracy of their predictions, the pollsters constantly revise them until they are in conformity with an emerging actual tally of the real vote.

Instead of embracing this essential truth about how exit polls work (or don’t work), Freeman cites other people who believe them to be largely accurate. And indeed they are. Throughout the day, the Ohio numbers being produced by Mitofsky were only off by a few points, which made a big difference only because of the closeness of the Bush-Kerry race in that state.

Freeman moves on to cite evidence that the election was stolen. Principally, he credits a “commission” headed by Rep. John Conyers. But that “commission” didn’t extensively investigate first hand. Mostly it just took statements from voters and activists, statements that were not fully vetted for their accuracy or for the likelihood that specific cases could be broadly extrapolated. Clearly, there were many odd and troubling incidents, but the reality is that when you carefully tally up all of the best-documented irregularities, it is still not enough to reverse Bush’s Ohio victory.  Even some leading figures in contesting the Ohio results concede that point.

Certainly, as I said in my original TomPaine piece, some Republican officials, including Secretary of State J. Kenneth Blackwell, behaved in an overtly partisan and inappropriate manner. An investigation of his conduct is warranted. But there’s no evidence that Blackwell’s rulings actually resulted in a reversal of the election. 

Critics of the electoral mechanism (and I am certainly one of them) need to be very careful not to overshoot. It doesn’t do anyone any good to throw bad arguments at a good cause. Or to dispirit and needlessly anger people by persuading them that their vote was rendered meaningless by broad-based fraud, when you really don’t know that to be the case. The full truth is not yet clear, but promoting sensational scenarios does little to serve the public interest.

As Norman Mailer put it so well in a recent essay for the London Sunday Times , “Good hypotheses depend on real questions, which is to say questions that do not always generate happy answers.”