The Functions of Exit Polls

Michael X. Delli Carpini

The purpose of the data produced by VNS is threefold. First, and from the point of view of the media foremost, is the ability to use these voter polls, in conjunction with other data, to project the winners and losers of races before the actual vote tallies are complete. This aspect of exit polls has been the most controversial over the years (predating the formation of VNS), and the decision to shut down VNS after its 2000 and 2002 debacles is most closely tied to this issue.

While the media’s interest in the predictive power of exit polls is understandable, and arguably adds to the interest and drama of election night coverage, from a broader perspective this is the least important reason why such polls have been and will continue to be valuable. A second purpose has been to provide a snapshot of the American electorate, taken moments after they leave their respective polling places. Information about not only who people voted for, but more importantly, why they did so, can be illuminating to the public and policymakers alike. Further analyses of particular subgroups, trends in voting over time, the public’s issue concerns, and so forth, all add to the value of these periodic assessments of the American electorate. In the best of worlds these glimpses into the electorate’s mood can serve to spark public and elite conversations that are as crucial to the democratic process as the electoral outcomes themselves.

The media certainly are aware of the value of this kind of analysis, filling their election night coverage with graphs, charts and commentary designed to educate the viewing public about emerging trends and patterns. There are limits to these instant analyses, however - limits driven in part by the constraints of television journalism (“keep it short,” “keep it simple”) and in part by the fact that serious analysis, whether journalistic or academic, can seldom be successfully accomplished or effectively communicated in the few minutes allotted to researchers to develop, graphic experts to design or reporters to explain. Some of these shortcomings are better addressed by the print media, who inevitably use VNS data in their subsequent coverage, but even here the use of this data lasts for no more than a day or two following an election, quickly becoming “old news.”

It is at this point that the third purpose of exit polls becomes evident - to provide scholars with data that can be used for more systematic, theory-driven research. This purpose clearly has been the least important to VNS’s media sponsors, perhaps best exemplified by the fact that as of this writing, they have no clear plans for assessing how much of the 2002 data is salvageable or for making the data publicly available (a group of scholars and practitioners are currently in conversation with the networks in an effort to develop such a plan). Even if the data become available, however, academic research based on exit polls suffers from the extreme opposite problem of journalistic research, taking months or years to complete and seldom finding its way back into the public debate.

Looking at this state of affairs, my own view is that the most important issue concerning the future of exit polls, of VNS or of its replacement is not should they continue (I believe they should) or even do they affect election outcomes or voter turnout (the evidence here is mixed and indeterminate, though my own view is that they can and that this is something that needs to be addressed). More important is the fact that exit polls provide a potentially invaluable and currently underutilized tool for increasing public, policymaker, media and academic understanding of what voters (collectively and broken into important subgroups) are thinking. In reexamining how exit polls will be conducted in future elections, my hope is that some serious attention will be paid - by journalists and scholars alike - to how these data are used - on election night, in subsequent (and preferably ongoing) media coverage and analysis, and (preferably in more timely and relevant ways than in the past) by the academic community.

Finally, assuring that future exit poll data are readily available for secondary analysis by scholars, practitioners and other media outlets is crucial. At a minimum this would mean taking seriously the need to release these data in a timely and accessible way, ideally by including this goal in the design and financing of whatever new exit polling entity emerges. Even more valuable - but admittedly less likely - would be providing greater opportunities for more scholars (and others) to advise on the questions asked, and perhaps even opportunities to add specific questions to the surveys - questions that may be of less immediate relevance to journalists but that could greatly enhance the longer-term value of exit polls to deepen our understanding of the American electorate. However, while this latter goal is certainly consistent with the media’s public interest obligations, given their economic and organizational priorities, it seems likely that meeting this goal would require a very different structure - and perhaps a completely separate institutional and financing arrangement - than has existed or is currently being contemplated.

Michael X. Delli Carpini is the director of the Public Policy Program for the Pew Charitable Trusts.

Return to the Roundtable