In November 2005, Ohio voters soundly defeated a package of election reform measures placed on the ballot as voter initiatives. In the days leading up to the election, a Columbus Dispatch poll predicted that Issue 2 (early voting) and Issue 3 (campaign finance reform) would pass, Issue 4 (redistricting commission) was too close to call, and Issue 5 (election administration) would be defeated. In actuality, the results differed from the polls by wide margins: Issue 2 received 36.7 percent of the vote, Issue 3 received 33.1 percent, Issue 4 received 30.3, and Issue 5 received 29.9 percent of the vote.
Not long after the election I had occasion to meet with a group of people who expressed concern about the discrepancy between the pre-election polling results and the actual vote tallies. Many in this group of active participants in the political process intimated, without actually declaring so, that such a discrepancy could result only from some kind of fraud or vote tampering.
While the individuals mentioned above merely expressed concern over the gap between pre-election polls and the actual vote, exit polling data from the day of the 2004 presidential election spawned an election contest challenging the certification of Ohio's electors for George W. Bush. In Moss v. Bush, 1 plaintiffs described the gap between exit poll results and the actual vote as "dramatic," 2 and contended that the "inescapable conclusion is that there was election fraud in connection with the vote counting in Ohio." 3 Plaintiffs argued that the irregularities were sufficient at least to set aside the results of the election, if the court was unwilling to declare the Kerry-Edwards ticket the winner in Ohio. 4
Both of these examples point to the impact of exit polling on popular confidence in the outcome of elections. Both the group I spoke with and the plaintiffs in the Moss v. Bush challenge trusted the results of polls more than they trusted the vote counting outcome as announced by election officials, despite Ohio's two-party checks on the operations of elections.
Over-reliance on polling to predict the outcome of an election should have ended with the presidential election of 2000. In 2000, the television media announced the outcome of the presidential election, relying in part on exit polls conducted by the now-defunct Voter News Service (VNS). Subsequently, the networks were forced to retract those declarations as the actual vote counting indicated a vote too close to call. According to VNS, the reasons for the inaccuracies of the exit polls included mistakes in their estimations of both the size and outcome of the absentee vote, sampling errors at the precincts, how VNS used past election results to inform current exit polling, and failure to use actual vote tallies to update exit polling data. 5
During the 2004 presidential election, the exit polling data once again proved problematic. Early reports based on exit polls put John Kerry ahead of George W. Bush, but by the time the actual votes were tallied, Bush claimed the Electoral College victory. The Edison/Mitofsky consortium that conducted the polls attributed the inconsistencies mainly to Kerry voters participating in exit polls at higher rates. 6 This higher participation rate was in turn blamed on a cacophony of reasons, including distance restriction on interviewers, weather, multiple precincts voting at single polling places, and interviewer characteristics. 7 An additional explanation is that fewer and fewer voters are agreeing to participate in exit polling. 8
Despite the explanations proffered by the organizations that conducted the polls, skeptics continue to rely on the discrepancies between exit polling and actual vote counts to claim that the vote counting, rather than the exit polling, was tainted. The Moss v. Bush case relied exclusively on the exit poll reports to make the case that the voters actually elected Kerry but "election fraud (or other irregularity) occurred in the counting of the vote in Ohio and a number of other states." 9 Similarly, a group of scholars and statisticians, rejected Edison/Mitofsky's own explanation for inaccurate exit polls in 2004. 10 Following detailed statistical analysis of the exit polling data, the group declared that "If the discrepancies between exit poll and election results cannot be explained by random sampling error; the 'Reluctant Bush Responder' hypothesis is inconsistent with the data; and other exit polling errors are insufficient to explain the large exit polling discrepancies, then the only remaining explanation - that the official count was corrupted - must be seriously considered." 11
These responses reveal that exit polls are being used as proxies for actual votes. John Zogby, the experienced and well respected CEO of polling firm Zogby International, argues that exit polling is useful for some purposes, but not for predicting the outcome of an election. 12 He posits that exit polls provide information about the demographics of voters and turnout, which can in turn be used for modeling future elections. However, neither the media who report exit polls as predictions of the outcome of elections, nor the vote watchers who rely on those predictions, limit their uses of exit polling, instead assuming that exit polling data will be nearly synonymous with actual vote counts.
However, some trends in how elections are conducted may reduce the ability to gather exit poll data, if the trends do not cause the demise of exit polling altogether. Among those trends are:
The combined impact of these changes could mean that exit polling is no longer available as a tool.
If exit polls were used solely for the informational purposes suggested by Zogby, exit polls should have no detrimental impact on voter confidence in the outcome of elections. However, in the two most recent presidential elections, exit polls have instead contributed to a continuing chorus of claims of "irregularities" at the least, and outright election fraud at the worst. Turnout in elections in the United States is already lower than many of us would wish, and creating distrust in the validity of the outcome of elections can only serve to further lower peoples' desires to participate. While the information gathered through exit polls is useful for politicians and social scientists, the poor record of their use by the media in the last two presidential elections and the negative impact on voter confidence in the outcomes of those elections, combined with trends in election administration likely making exit polling even more difficult to conduct with accuracy, leads to the conclusion that exit polls should no longer play their historical role in election night reporting.
1. Moss v. Bush, Verified Election Contest Petition, December 17, 2004, Supreme Court of Ohio, 2004 WL 3153018.
2. Id. at ¶ 70.
3. Id. at ¶ 73.
4. Id., Prayer for Relief at ¶ 1.
5. Howard Kurtz, Errors Plagued Election Night Polling Service; VNS Report Also Faults Networks in Fla. Blunder, Washington Post, December 22, 2000 at A01.
6. Evaluation of Edison/Mitofsky Election System 2004" prepared by Edison Media Research and Mitofsky International for the National Election Pool (NEP) Jan. 19, 2005, p. 3-4, available at http://exit-poll.net/election-night/EvaluationJan192005.pdf.
7. Id. at 4.
8. Report of the Commission on Federal Election Reform, "Building Confidence in U.S. Elections," organized by the Center for Democracy and Election Management, American University, September 2005, § 7.2, available at http://www.american.edu/ia/cfer/report/report.html.
9. Moss v. Bush, Plaintiff's Brief, Claim for Relief ¶72.
10. "Analysis of the 2004 Presidential Election Exit Poll Discrepancies," U.S. Count Votes' National Election Data Archive Project. March 31, 2005, available at http://electionarchive.org/ucvAnalysis/US/Exit_Polls_2004_Mitofsky-Edison.pdf.
11. "Analysis of the 2004 Presidential Election Exit Poll Discrepancies," p. 18.
12. John Zogby, "The Polls, the Pundits, and the Election of 2004," in Elections 2004, a publication of the International Information Programs of the Department of State, September, 2003, available at http://usinfo.state.gov/products/pubs/election04/polls.htm.