In Print: Polling, the News Media, and Politics January 2006

Elections have a special place in American journalism, for several reasons. One reason is that we live in a democracy, and public opinion has such a central role in the functioning and legitimacy of our government. Elections, and the campaigns leading up to them, are the defining political act in the United States. They represent a point at which most Americans devote more time to thinking about politics and public affairs than normal. The election of public officials with broad public supports—and the “mandates” that might be involved—provides a critical underpinning of our system of representation.

At the same time, elections make great news. They involve important issues and, eventually, well-known figures. They operate under a system of rules that most citizens are familiar with. They occur on a fixed schedule, involve substantial conflict, and ultimately come to a neat resolution on Election Day with the declaration of winners and losers. Moreover, campaigns are populated by willing sources interested in talking with journalists and having their side of the story presented in the best possible light. For all these reasons, there is a strong symbiotic relationship between journalists and candidates. They rely on each other for success, even though they often seem at odds with each other.

One of their common interests is how the public feels about the campaigns, the issues, and the candidates—what the “public mood” is. This is complicated by the fact that there is no common or standard definition of what “public opinion” is. Some believe it is an attribute of an individual that can be expressed out loud or in public. Others believe that it is the aggregated opinions of people, measured and summarized. By this definition, it is not public unless it is disseminated widely and made known to others. Under these circumstances, a poll or survey is the ideal mechanism for measuring public opinion, and the news media are the ideal vehicles for making these opinions known to others.

In the old days, both candidates and journalists relied on various “experts” for these assessments. They included party leaders, elected officials, and such unobtrusive indicators as the size of crowds that turned out for scheduled events. But the size of a crowd, for example, is an imperfect measure of public opinion because it is often difficult to associate a good measure of valence or affect with sheer numbers of participants or to gauge the intensity of feelings associated with the views that its members hold. For a very long time, what was missing in American politics was a way to produce systematic and reliable measures of public opinion—information that could be used to plan or revise strategy or to contextualize reporting of what the candidates were saying and doing.

Whereas politicians and journalists have always been interested in knowing about public opinion, the extensive application of survey research techniques did not begin until the 1930s, when the “founding fathers,” such George Gallup, Elmo Roper, and Archibald Crossley, began to collect and publish opinion data. From the start, their efforts were possible because of relationships they established with newspapers and magazines. They needed the mass media to serve as outlets for the wide dissemination of their results because their public opinion business was a way of promoting their firms’ proprietary work for commercial clients. And these news outlets were always looking for new and timely content.

After World War II, improvements in sampling methods and increasing commercial demand for survey research led to an increase in polling. The candidates themselves turned to public opinion polls as an integral part of their own research efforts, using the information to supplement analyses of historical voting patterns. At the same time, the public dissemination of data rapidly accelerated after the important news organizations in the United States—the networks and major metropolitan daily newspapers—began to collaborate on their own independent data collection. Now the news coverage of election campaigns is filled with poll results, both from polls leading up to a primary or general election day and then from exit polls of voters leaving their balloting places. The former data are used to predict winners and to explain and dissect the campaign, and the latter are used to provide poll-based explanations of the “meaning” of the outcome, as well as to project the winners.

News organizations and journalists justified their entry into the polling business because they believed that the use of poll data contributed to their objectivity in producing news about politics. When they purchased results from the Gallup poll or one of its competitors, they acquired useful nonpartisan content at a reasonable price. Technological shifts that reduced the cost of polling—most notably the penetration of telephones into virtually every American household and the availability of microcomputers that serve as low-cost interviewing devices and data analysis machines—raised the prospect of independent data-collection activities. And news organizations further justified this on the basis of increased editorial control: they could ask whatever questions they wanted and put studies into the field whenever they wanted if they ran their own polling operations.

The technology of polling is always changing. In the past decade, a number of telecommunications technologies and government regulations have hampered the conduct of telephone surveys. This includes the proliferation of cell phones, call screening and call blocking, and the plague of telemarketing that led to the recent advent of Do Not Call List regulations. Separately, starting in the late 1990s, some organizations began to use the Internet to collect preelection poll data. This is still problematic for a number of reasons in the United States, including the limited number and range of people who have regular access to computers and the World Wide Web, the difficulty of drawing good probability samples and achieving high response rates, the ability of individuals to express their opinions more than once, and new federal- and state-level regulations that restrict some of the ways that pollsters can use the Internet to recruit sample respondents. Eventually some of these problems are likely to be overcome, but the current regulatory and social environments are unpredictable and it is uncertain what the long-term effects will be on pollsters.

Since we edited the last version of this book, the exit poll operation embodied in the Voter News Service (VNS) went through two jarring elections. In 2000, which was the closest presidential election in American history and was beset by a protracted legal battle that started in Florida, the networks and the exit poll operation had a bad outing that prompted congressional hearings. The sponsoring parties tried to correct their systems before the 2002 election, but there was not enough time. On Election Night 2002, they were unable to produce data for projections and analysis, and VNS went out of business, to be replaced by the National Election Pool (NEP). They… built an entirely new software system for the 2004 election coverage….

These are the technological and business trends that have accelerated the production of polling data and their increased use in news making, and they highlight the problems that pollsters and news organization have overcome. They explain why we have more polls, but they do not tell us what difference polls make or what impact they have on American political life. These are more subjective issues, but we do have a view on them.

There is growing body of literature, increasingly compelling in terms of the evidence mustered, that indicates that polls have a substantial impact on the American political process. Poll results have an impact on the vitality and viability of candidacies, affecting who can raise money, organize a field staff, and secure volunteers. News coverage containing poll results has an impact on assessments that citizens make of candidates and how, and even if, they decide to vote. And polls clearly have an effect on how campaigns are covered, as reporters, editors, and producers use this information to make decisions about who to cover and how to frame the coverage.

We are not opposed to polls and polling; on the contrary, we see election polls in terms of their still largely unfulfilled potential. There is plenty of room for them to make a substantial contribution to levels of citizen knowledge and understanding of the political process, including the provision of information about how fellow citizens see the political world in terms of issues and how they respond to candidates and their campaigns.

Unfortunately, these possibilities go largely unrealized because too much campaign reporting is devoted solely to who is ahead and who is behind—a form of “horserace” coverage to which polls easily lend themselves. Polls are also used to support explanations of campaign strategy and dynamics, rather than focus on the issues that concern voters and their appreciation and understanding of what the candidates have to say about them.

Our hope is that if citizens understand more about how polls are conducted, analyzed, and reported in the media, they will be able to think about other ways in which such information would be useful to them. And on an informed basis, they will be able and want to exert pressure on news organizations to alter some elements of their coverage so they will be more responsive to the informational interest and needs of their readers and viewers.