Article — From the August 2005 issue
- Current Issue
SIGN IN to access Harper’s Magazine
1. Sign in to Customer Care using your account number or postal address.
2. Select Email/Password Information.
3. Enter your new information and click on Save My Changes.
Subscribers can find additional help here. Not a subscriber? Subscribe today!
Article — From the August 2005 issue
In the summer of 2003, Representative Peter King (R., N.Y.) was interviewed by Alexandra Pelosi at a barbecue on the White House lawn for her HBO documentary Diary of a Political Tourist. “It’s already over. The election’s over. We won,” King exulted more than a year before the election. When asked by Pelosi—the daughter of House Minority Leader Nancy Pelosi—how he knew that Bush would win, he answered, “It’s all over but the counting. And we’ll take care of the counting.”
King, who is well known in Washington for his eccentric utterances, says he was kidding, that he has known Pelosi for years, that she is “a clown,” and that her project was a “spoof.” Still, he said it. And laughter, despite the counsel of Kenneth Blackwell’s press flack, seems an inappropriate response to the prospect of a stolen election—as does the advice that we “get over it.” The point of the Conyers report, and of this report as well, is not to send Bush packing and put Kerry in his place. The Framers could no more conceive of electoral fraud on such a scale than they could picture Fox News Channel or the Pentagon; and so we have no constitutional recourse, should it be proven, finally, that the wrong guy “won.” The point of our revisiting the last election, rather, is to see exactly what the damage was so that the people can demand appropriate reforms. Those who say we should “move on” from that suspicious race and work instead on “bigger issues”—like electoral reform—are urging the impossible; for there has never been a great reform that was not driven by some major scandal.
“If a nation expects to be ignorant and free, in a state of civilization,” Thomas Jefferson said, “it expects what never was and never will be.” That much-quoted line foretells precisely what has happened to us since “the news” has turned into a daily paraphrase of Karl Rove’s fevered dreams. Just as 2+2=5 in Orwell’s Oceania, so here today the United States just won two brilliant military victories, 9/11 could not have been prevented, we live in a democracy (like the Iraqis), and last year’s presidential race “was, at the end of the day, an honest election.” Such claims, presented as the truth, are nothing but faith-based reiteration, as valid as the notions that one chooses to be homosexual, that condoms don’t prevent the spread of HIV, and that the universe was made 6,000 years ago.
In this nation’s epic struggle on behalf of freedom, reason, and democracy, the press has unilaterally disarmed—and therefore many good Americans, both liberal and conservative, have lost faith in the promise of self-government. That vast surrender is demoralizing, certainly, but if we face it, and endeavor to reverse it, it will not prove fatal. This democracy can survive a plot to hijack an election. What it cannot survive is our indifference to, or unawareness of, the evidence that such a plot has succeeded.
CORRECTION: The print version of “None Dare Call It Stolen” contained the following line, which was incorrect: “on Election Day, twenty-six state exit polls incorrectly predicted wins for Kerry.” The correct number was five states. Although we regret the error, the context surrounding it bears further explanation.
The mistake was brought to our attention by a letter from Warren Mitofsky, founder of Mitofsky International, which, along with partner Edison Media Research, has conducted exit polls of every presidential contest since 1996. In the letter, Mr. Mitofsky stated that not only was the figure for twenty-six states incorrect, so, too, was the assertion that Edison/Mitofsky’s exit polling contained any mistakes whatsoever. “One hundred-twenty-three races for President, Senator, Governor, and propositions,” Mr. Mitofsky wrote, “were called without error.” He further attributed our misstep to “confusing the reports by bloggers with the exit poll my partner and I did.”
Perhaps. But a closer inspection of what Mr. Mitofsky actually means by “called without error” could indicate otherwise. On January 19, 2005, Edison/Mitofsky released a report that, while continuing to maintain that no election projection mistakes were made, did acknowledge the existence of serious “differences between the exit poll estimates and the actual vote count.” In thirty states, the voter estimates produced by Edison/Mitofsky data was wrong to a statistically significant degree (twenty-six states for Kerry, four for Bush). Our mistake came in failing to recognize that in twenty-one of the twenty-six instances in which the estimates incorrectly named Kerry as the front-runner, he ultimately carried the state, only by a smaller margin than indicated by the exit polls. Still, an apparent logical disconnect would seem to exist. How could the estimates be wrong but not the final projection? To answer this question, a clear picture of the difference between estimates and final projections is needed.
On Election Day, exit poll interviewers submit their results to Edison/Mitofsky three times, during regularly scheduled “calls,” the last of which comes shortly before the close of the polls. These results do not contain official vote numbers, which is important. Many people would assume that Edison/Mitofsky’s final projections exclusively utilize the information collected at the polls and sent in during the calls; however, this is not the process. Edison/Mitofsky’s report makes clear that it does not “rely solely on exit polls in its computations and estimates.” When the voting is complete, actual vote numbers are combined with the exit poll responses and “as in past elections, the final exit poll data used for analysis . . . [is] adjusted to match the actual vote returns.” So, even if the exit poll estimates are erroneous, Edison/Mitofsky still isn’t wrong-because they just add in the actual vote numbers to ensure everything checks.
This practice is by no means secret, although perhaps the average voter or election-night network-television watcher might not have been aware of it. I certainly wasn’t. Maybe knowing this should serve to highlight the risks of viewing exit polls as a hedge against improprieties in the vote count. Or perhaps that is precisely the best use for them. The chances that the state exit poll estimates erred by such a wide margin was one 1 in 16.5 million, according to a study by the National Election Data Archive Project. One final key point remains: of the five states Edison/Mitofsky had Kerry leading that he eventually lost, Ohio was one.
More from Mark Crispin Miller: