Behind the Controversy By Mark Lindeman and Rick Brady PartI page3

VNS closed its doors in January 2003 after a brutal 2002 general election, in which its newly overhauled data system barely functioned and yielded no usable results on election night. ABC, CBS, CNN, Fox News, NBC, and the Associated Press formed the National Election Pool (NEP), which contracted with Edison Media Research and Mitofsky International to conduct the 2004 exit polls.

The 2004 general election exit poll can be likened to a good news/bad news story. The good news was that NEP was able to provide useful analytical results in real time (unlike VNS in 2002) and made no incorrect projections (unlike VNS in 2000). Although the best estimators gave Kerry the lead in four states that Bush ended up winning—Iowa, Nevada, New Mexico, and Ohio—the lead in each case was well within the margin of error.

The bad news was that not only did exit poll error favoring the Democratic candidate reach a new high—and not only did the error alter the apparent balance of the race—but the early, erroneous results were publicized sooner and more widely than ever before. Leaked early exit poll data began surfacing on various blogs before noon EST and are credited with having caused afternoon fluctuations in the stock market. Exit poll data have always leaked, and early results were posted on the internet in both 1996 and 2000. But the internet was a different place in 2004, and the increased popularity of blogs and political websites facilitated the wide distribution of leaked data.

The error in the estimates was widespread and, in some cases, startling. The best estimator at poll closing overstated Kerry’s performance in twenty-four states by more than one standard error; Bush’s performance was overstated by a similar margin in only four states. The disparity in these results, clearly not the product of random chance, provoked some accusations of intentional poll bias and of fraud in the official count. For their part, the NEP members and subscribers wanted to know why they paid for early exit poll data that proved to be of little worth in planning their coverage on election evening. In response to these concerns, Edison/Mitofsky prepared and publicly released the most comprehensive analysis of an exit poll available to date. The Edison/Mitofsky report concludes that, in contrast to Florida in 2000, survey error within the precincts accounted for most of the net error in the estimates.

The 2004 exit polls have evoked a clash of competing narratives. From Edison/Mitofsky’s perspective, the election night projections exemplify the proper use of exit poll data in concert with vote count data—while the midday leaks and premature predictions of a Kerry victory show how not to use exit polls. For some observers, the exit polls provide crucial evidence that Kerry actually won the popular vote in 2004—and that Edison/Mitofsky and the NEP have been disgracefully attempting to conceal the evidence ever since. But beyond this polarized debate remain more nuanced questions about the lessons of 2004, which bear both on how exit polls are conducted and how their results are presented.

Mark Lindeman is an assistant professor in the political studies program at Bard College, and Rick Brady is an M.C.P. candidate in the School of Public Administration and Urban Studies at San Diego State University.