The Media in Elections: Methods of Study
Landmark Study of the Media in Elections: A Review of the Patterson Approach
If the mass media are the principal means of communicating the elements of an election to the public, then the logical first question is: What exactly are the media communicating? The news is, in form at least, the principal source of the public's information and impressions on election subjects (we will consider other prime sources, such as candidates' ads and televised debates, a bit later). As Thomas Patterson pointed out in his study, The Mass Media Election: "Election news results from a series of decisions made by news organizations about what to observe, what to report, and what emphasis to place on various parts of the coverage."14 Patterson's comprehensive study, which analyzed the media in the 1976 presidential election, is the landmark exemplification of how to analyze news coverage of elections. It has been both a guide and an inspiration for the major studies conducted since it was first published in 1980, and its findings remain a foundation of our understanding of this field.
The principal method for studying what has been communicated as election news is content analysis, which is the analysis of the number, placement, size, and content of election-related stories in the news. In somewhat more formal terms, content analysis is "the systematic counting, assessing and interpreting of the form and substance of communication."15 As that latter definition suggests, content analysis can be applied to other forms of election communications such as political ads, speeches, and debates. In doing a content analysis of the news media's communications about an election, especially a presidential election, a practical problem presents itself at the start: It is not feasible to study all news stories in all mass media that reach the entire national public during the entire period of the campaign. Some choices must be made about samples that will stand for all relevant stories in all media. Patterson analyzed a randomly chosen large sample of stories (N = 6,567) appearing in the three major network news programs and Time and Newsweek. Two newspapers each in two locations were also monitored, one of which was one of the nation's leading papers, the Los Angeles Tbnes, while the others were medium-sized, metro area papers. The sto¬ries were chosen to cover the entire formal election period: January 1 through the November election day.
Regarding content, each of the stories was analyzed for the following: the date of the story, whether it was accompanied by a photo or film, its length, its position in the medium (on the front page/"leading" the broadcast or on an inside page/later in the broadcast), the principal subject of and actor in the story, whether the story "reflected favorably or unfavorably on this actor," and "references to the candidate's personal and leadership qualities, styles, back¬grounds, issue stands, values, campaign progress, and similar things."16 Assis¬tants then carefully coded each story and its elements, that is, they assigned numbers or letters to identify the political actor and so on and to signify the
273
Chapter 10 The Media in Elections I
value of each element found in each story. This also shows that the unit of analysis in this analytic effort was the stoiy. We will note a different unit of analysis shortly.) With the entry of this data into a computer, the magnitude and nature of the elements of news coverage can be discovered over all the media, or for print versus TV news, or certain aspects can be compared candidate by can¬didate, and so on.
But what of the receipt of those news stories by the public and their response? The ultimate concern in a democracy is surely the impact of cam¬paign communications on "we the people." Correspondingly, Patterson's study went beyond analyzing what was communicated and added a second basic-dimension to the analysis: who was communicated to by what media and what their responses were. Many studies have focused only on the nature of media communications, but what gave the Patterson study such power was that he looked both at the communications through the news media and at the receiving public, and connected the two. Patterson's principal method of studying the publics attention and response sought to overcome the limitations of the simple cross-sectional survey. Such a survey involves a single interview with each respondent at a single point in time, usually just after election day. But such an approach requires people to accurately recall just what the)- saw and read and how those communications affected their response to candidates and issues from as much as ten or more months in the past—and to separate media effects from nonmedia effects.
Patterson used the panel survey method to overcome those limitations. In his study, a total ("panel") of "1,236 eligible voters were questioned as many as seven times in the course of the . . . campaign about their media use, their impressions of the candidates and the campaign, their awareness of the election's issues, their interest in the campaign, and similar topics."17 But given the post-1968 nature of the presidential nomination and election process, the timing of these successive "waves" of interviews was important. If we are to systematically establish how media communications have affected people's thinking, it is also important to establish peoples' states of mind before they are exposed to such communications. Correspondingly, the first wave of interviews is needed to con¬struct a "baseline" of peoples' orientations, level of interest, awareness of issues and candidates, and so on. Interestingly, although Patterson recognized and incorporated this logic, the new nomination politics and the media age got a lit¬tle ahead of him. His baseline interviews were scheduled before the first primary in New Hampshire, which had been the first delegate selection event to receive media significant attention; but in 1976, for the first time, the Iowa caucuses— then occurring in January—received a great deal of attention (as he documented in the section on news coverage). For optimum purposes, in short, his baseline was a little late. (It was still basically adequate, however, as we will show later on.) After the baseline interviews, Patterson timed the waves of interviews to
274
The Media in Elections: Methods ol Study
capture peoples' attention and responses at the major stages of the process, that is, after the early primaries (April), the late primaries (June), the party conven¬tions (August), and late in the general election (October). As it happened, there were also three debates between the two major party candidates; telephone interviews were done after the first and second debates. Phone interviews were also conducted after the election to establish whether the respondents had voted and for whom if they had; that is, what, in the end, was their actual voting behavior and how did it relate to the nature of the news coverage?
But a question could be raised about interviewing the same panel of people (even such a large panel) seven times: Would the repeated interviewing of those people tend to sensitize them to the election process in general and the particu¬lar subjects asked about and hence, alter the very state of mind that the study was trying to ascertain? Thus, if repeated questioning makes people pay more attention to election news than they normally would have, thereby developing more awareness and knowledge of candidates and so on, then it is not media impact that is being discovered but something artificially stimulated by the study itself. This reinterview-sensitizarion problem did not turn out to be the significant factor one might expect, however. The principal reason is that most people are worried more about their families, jobs, finances, neighbor's barking dog, and so on; they do not tend to place a heavy emphasis on politics. Corre¬spondingly, as soon as the panel interview is over, they tend to quickly forget their talk with interviewers for some abstract academic project and return to thinking about those everyday matters that clearly and tangibly impinge upon them. Patterson included a feature in the study to check for the sensitization possibility: he added new respondents (to serve, in effect, as control groups) to the second and third waves of interviews. He found no significant differences in responses between those people and the people in the original group of the same demographic and political sorts.
A final aspect of Patterson's methods should be mentioned. In studying the news coverage, the prime national media could be fairly fully monitored in 1976 by looking at the three network news shows and two (or all three) of the major newsmagazines. Local news media are also important in the mix of sources of campaign communications, though, and it was not feasible to study all media reaching all areas, so choices had to be made of what locations and particular "media markets" to study. Choosing one or a few locations allowed the measure¬ment of virtually the entire significant media environment of those interviewed. Patterson chose two areas ("media markets"): the major metropolitan area of Los Angeles, which had a very diverse population, and Erie, Pennsylvania, with a metro population of 270,000 and a more homogenous, blue-collar, heavily Catholic character. But here we also see a limitation of selecting only two areas and these two areas in particular; both areas are urban, smaller cities and rural areas are not represented. (It is worth noting that Patterson did include other
275
Chapter 10 The Media in Elections I
local areas in the original design of the study, but it did not prove feasible for him to include that material in the analysis in his book.18)
The Patterson study made a significant contribution to the systematic study of the med