Election polls have a long history of a symbiotic relationship with the media, dating back to the nineteenth century (Converse 1987; Frankovic 2008). However, it was not until the 1920s that polls insinuated themselves into the news operations of election coverage on a regular basis. Before the advent of the modern polling period, the major political poll operating in the United States was conducted by a leading circulation magazine, the Literary Digest. Based upon its subscriber lists, eventually supplemented by information from telephone directories and automobile registrations, the Literary Digest conducted large mail surveys to produce estimates of the outcome of presidential elections. While the magazine made good estimates from its data in 1928 and 1932, its 1936 estimate that predicted a Landon win over Roosevelt was a disaster (Squire 1988). Of greater significance, George Gallup had started a new public polling operation and promised his first major news client, the Washington Post, that his methods could outperform the Literary Digest. When he did, a new era of public polling for news organizations was born.
The nature of the symbiotic relationship was, and remains, that public polling organizations needed the news media to publicize the work they did during elections in order to promote their availability for private work for commercial clients, where they made their money. And the news organizations were attracted to the ability of polls to generate news content based upon their election forecasts which coincided with their natural inclination to cover elections with “horse race” journalism which focuses on who is ahead and by how much. At the same time as Gallup began his relationship with a series of newspapers that carried his syndicated column from the American Institute for Public Opinion, Elmo Roper established a similar relationship with Fortune magazine (Zetterberg 2008). And as American campaign techniques have been transported to overseas political systems, so has the use of polls in the news coverage of them (see, e.g., Worcester 1980; Weimann 1990; Brettschneider 1997; 2008).
These partnerships still exist today, although changes in technology and the structure of the news business have altered them. In the 1970s, a number of forces converged to move news organizations into independent data collection rather than continuing their reliance on syndicated data gatherers like Gallup, Roper, or Harris. This was the period when network television news was competing with newspapers for pre-eminent coverage of elections, including live on-the-air coverage on election night. In the polling industry, telephone surveys were replacing face-to-face interviewing techniques, and the advent of low-cost personal computers and supporting software coincided with the development of computer-assisted telephone interviewing (CATI) operations and the design of random digit dialed (RDD) sampling.
Major partnerships developed in the US between CBS News and the New York Times, ABC News and the Washington Post, and NBC News and the Wall Street Journal. Each of these partnerships involved a major morning newspaper and a significant television news operation that coincidentally shared the same effective deadline early in the evening of the same day. Similar joint polling operations have sprung up in a number of other countries, including in Germany between the Allensbach Institute and the Frankfurter Allgemeine Zeitung, for example, and in the United Kingdom between the British Broadcasting Corporation (BBC) and Globescan, as well as a variety of research institutes. The television news operations get to produce a brief version of a polling story first for broadcast, while the newspaper operations produce longer stories, typically with more analysis, for the next morning’s paper. By the late 1990s, these news organizations were also independently producing web-based content for simultaneous release as well.
In the beginning, the partnerships were justified on the basis of newsworthiness criteria. Why should a pre-eminent collection of journalists have to rely upon editorial decisions made by someone at the Gallup Organization (or Harris or Roper for that matter) about which questions should be included in a poll and when it should be fielded? When the paired news organizations set up their operations, there were shared resources each committed to the operation (like the phone banks in the advertising operations of the newspapers that were not being used in the evenings), and they contributed jointly to the development of the questionnaires. Each poll generated a dataset that could be analyzed separately by each organization, and it often produced different lead stories based upon different assessments of the most newsworthy content from the poll.
In 1992, the Gallup Organization entered into a joint agreement with CNN and USA Today to produce poll content. In addition to work on analysis, Gallup editors eventually began to appear in brief video segments on the cable network. In 2006, in what might be described as business coming full circle across 70 years of work, Gallup announced it was severing its relationship with CNN because of declining cable ratings for their segments and an interest in establishing its own e-broadcasting presence, including with AOL. It maintains its partnership with USA Today because of its large readership in the United States and its worldwide presence. We are waiting to see whether similar “divorces” occur with other pairings of media and polling organizations in other parts of the world.
Political Polls In Election Campaigns
Campaigns and elections are of special interest to news organizations because they share many features that make them ideally suited to coverage. They are high-impact events that affect many readers and viewers; in the case of presidential elections or even statewide elections, they affect everyone in the circulation or broadcast area. They usually involve known figures who are eager to be news sources and willing to be quoted. They involve conflict and drama, and they occur on a schedule. And election day provides a clear end to the event with a declared winner and loser.
All of these features provide a clear and well-understood storyline for news organizations, and they enable them to carefully plan their coverage in terms of the allocation of Polls and the Media resources (including reporters), the partitioning of the news hole, and the organization around key events in the political calendar, including the announcement of candidacies, primary elections or nominating conventions, debates between the candidates, and major policy addresses (Weaver 2008). Every one of these events is also amenable to different forms of polling that assesses the relative standing of the candidates at each stage of the campaign, especially in terms of the “horse race” measure of who is ahead and by how much; who won a debate; and the impact of significant campaign events like planned speeches or inadvertent slip-ups.
The polls also played a central role in the capture of election coverage by electronic broadcast media from newspapers. While news organizations sometimes shared the content of pre-election polls and analyzed them carefully for trends, the early television coverage of election night results was based upon the tabulation of raw votes from across the country. These votes came in very slowly and were relatively meaningless unless the news organization had detailed information about the geographical locations where they came from and the historical voting patterns there. It was quite common in this period for newspaper coverage of elections not to contain any significant results and analysis of them until the one or two days following the election. In order to overcome this problem and gain a reporting advantage, the US television networks developed new polling techniques to assist with election night coverage. At first they simply applied sampling theory to select “key precincts” where the raw votes could be tabulated on a representative basis. But they quickly saw the limits of such techniques for extended analysis of such factors as gender or attitudinal patterns underlying voter preferences. So exit polls were developed as a method to add a significant explanatory note to the understanding of social and political forces at work in the electorate. This technique added depth and immediacy to television coverage of elections, as well as an overall immediacy to counting votes more quickly.
The exit poll phenomenon has taken off in other countries as well, assisting electronic news organizations, especially television networks, in capturing election night coverage. Depending upon the nature of the electoral system, these exit poll operations focus less on the division of the popular vote and more on the control of the national legislature or parliament. In the United Kingdom, one set of exit polls is designed to capture the winners of the marginal or competitive seats and hence the control of the House of Commons. In other countries, different interviewing procedures are employed, including the use of telephone surveys of those who have already voted, as in the Philippines. But in each case, the goal is the rapid reporting of the outcome of elections in as close to real time as possible and before newspapers can report the same results.
Problems With Media Use Of Polls
The problems with the media’s use of polls derive from their reliance on “trial heat” or “horse race” polls that measure who is ahead and behind in a political contest (Patterson 2005). Polls could contribute in a number of significant ways to election coverage if they were used in a more varied set of ways that gave the voters a stronger voice in the coverage. They can be used at the start of the campaign to provide information about which issues the public thinks are most important and their sense of how much attention Polls and the Media the candidates are devoting to them. They can be used during the campaign to assess how much the voters are learning about the candidates and the issues they are discussing. And they can be used at the end of the campaign to measure voters’ satisfaction with the content and tone of the candidates’ speeches, ads, and debate performances, as well as with the media coverage of the campaign.
In the standard news story about the campaign that does not include such polling information, journalists search out party leaders, candidates, and campaign staff members for assessments of the campaign. These are all strategists of one kind or another who have a vested interest in the outcome of the campaign and will “spin” a story to their side’s advantage. This often involves statements about what the voters are learning or how they are reacting to the candidates and events of the campaign. Why not use polls to give the voters their own voice in this coverage rather than providing partisan or self-interested interpretations of what is going on in the public’s mind?
One of the realities that contemporary news organizations face is that their audience sizes and shares have been declining. This has produced a number of associated trends in news coverage. One is a relative decline in hard news coverage, supplanted by an increasing amount of soft coverage. Furthermore, news organizations have turned to a number of techniques to try to engage their remaining audience members and provide some interaction with the production of news. Both of these trends are complemented with the increased use of polling data, although not always of the highest quality. Polls enable news organizations to report on what their readers and viewers are thinking, their patterns of behavior, and their attitudes and preferences. The hope is that this will forge a closer tie between the news organizations and the audience.
At the same time, news organizations are some of the most profitable corporations in the business world, and they are under continuous pressure to maintain these high profit margins for their shareholders and investors. One consequence is that often they resort to inexpensive data collection that includes dial-in or web polls that involve self-selected samples of viewers and readers who are sometimes asked to respond to simple single question surveys. The emphasis is on the sheer number of responses, for the purpose of engaging as many audience members as possible, rather than the representative nature of them. These features make Internet polls especially attractive to news organizations, although significant questions about their methods remain.
- Brettschneider, F. (1997). The press and the polls in Germany, 1980–1994: Poll coverage as an essential part of election campaign reporting. International Journal of Public Opinion Research, 9, 248–265.
- Brettschneider, F. (2008). The news media’s use of opinion polls. In W. Donsbach & M. W. Traugott (eds.), Handbook of public opinion research. London: Sage.
- Converse, J. (1987). Survey research in the United States: Roots and emergence, 1890–1960. Berkeley: University of California Press.
- Frankovic, K. (2008). Pre-election polls and exit polls. In W. Donsbach & M. W. Traugott (eds.), Handbook of public opinion research. London: Sage.
- Patterson, T. E. (2005). Of polls, mountains: U.S. journalists and their use of election surveys. Public Opinion Quarterly, 69, 716–724.
- Squire, P. (1988). Why the 1936 Literary Digest poll failed. Public Opinion Quarterly, 52, 125– 133.
- Weaver, D. H. (2008). Attitudes of journalists toward public opinion research. In W. Donsbach & M. W. Traugott (eds.), Handbook of public opinion research. London: Sage.
- Weimann, G. (1990). The obsession to forecast: Pre-election polls in the Israeli press. Public Opinion Quarterly, 54, 396–408.
- Worcester, R. M. (1980). Pollsters, the press, and political polling in Britain. Public Opinion Quarterly, 44, 548–566.
- Zetterberg, H. (2008). The start of modern public opinion research. In W. Donsbach & M. W. Traugott (eds.), Handbook of public opinion research. London: Sage.