In “Public Opinion Quarterly”, 1972, XXXVI, 2
In choosing and displaying news, editors, newsroom staff, and broadcasters play an important part in shaping political reality. Readers learn not only about a given issue, but also how much importance to attach to that issue from the amount of information in a news story and its position. In reflecting what candidates are saying during a campaign, the mass media may well determine the important issues—that is, the media may set the "agenda" of the campaign.
IN OUR DAY, more than ever before, candidates go before the people through the mass media rather than in person. The information in the mass media becomes the only contact many have with politics. The pledges, promises, and rhetoric encapsulated in news stories, columns, and editorials constitute much of the information upon which a voting decision has to be made. Most of what people know comes to them "second" or "third" hand from the mass media or from other people.
Although the evidence that mass media deeply change attitudes in a campaign is far from conclusive, the evidence is much stronger that voters learn from the immense quantity of information available during each campaign. People, of course, vary greatly in their attention to mass media political information. Some, normally the better educated and most politically interested (and those least likely to change political beliefs), actively seek information; but most seem to acquire it, if at all, without much effort. It just comes in. As Berelson succinctly puts it: "On any single subject many 'hear' but few 'listen'." But Berelson also found that those with the greatest mass media exposure are most likely to know where the candidates stand on different issues. Trenaman and McQuail found the same thing in a study of the 1959 General Election in England. Voters do learn.
They apparently learn, furthermore, in direct proportion to the emphasis placed on the campaign issues by the mass media. Specifically focusing on the agenda-setting function of the media, Lang and Lang observe:
The mass media force attention to certain issues. They build up public images of political figures. They are constantly presenting objects suggesting what individuals in the mass should think about, know about, have feelings about.
Perhaps this hypothesized agenda-setting function of the mass media is most succinctly stated by Cohen, who noted that the press "may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about."* While the mass media may have little influence on the direction or intensity of attitudes, it is hypothesized that the mass media set the agenda for each political campaign, influencing the salience of attitudes toward the political issues.
To investigate the agenda-setting capacity of the mass media in the 1968 presidential campaign, this study attempted to match what Chapel Hill voters said were key issues of the campaign with the actual content of the mass media used by them during the campaign. Respondents were selected randomly from lists of registered voters in five Chapel Hill precincts economically, socially, and racially representative of the community. By restricting this study to one community, numerous other sources of variation—for example, regional differences or variations in media performance—were controlled.
Between September 18 and October 6, 100 interviews were completed. To select these 100 respondents a filter question was used to identify those who had not yet definitely decided how to vote—presumably those most open or susceptible to campaign information. Only those not yet fully committed to a particular candidate were interviewed. Borrowing from the Trenaman and McQuail strategy, this study asked each respondent to outline the key issues as he saw them, regardless of what the candidates might be saying at the moment. Interviewers recorded the answers as exactly as possible.
Concurrently with the voter interviews, the mass media serving these voters were collected and content analyzed. A pretest in spring 1968 found that for the Chapel Hill community almost all the mass media political information was provided by the following sources: Durham Morning Herald, Durham Sun, Raleigh News and Observer, Raleigh Times, New York Times, Time, Newsweek, and NBC and CBS evening news broadcasts.
The answers of respondents, .regarding major problems as they saw them and the news and editorial comment appearing between September 12 and October 6 in the sampled newspapers, magazines, and news broadcasts were coded into 15 categories representing the key issues and other kinds of campaign news. Media news content also was divided into "major" and "minor" levels to see whether there was any substantial difference in mass media emphasis across topics. For the print media, this major/minor division was in terms of space and position; for television, it was made in terms of position and time allowed. More specifically, major items were defined as follows:
1. Television: Any story 45 seconds or more in length and/or one of the three lead stories.
2. Newspapers: Any story which appeared as the lead on the front page or on any page under a three-column headline in which at least one-third of the story (a minimum of five paragraphs) was devoted to political news coverage.
5. News Magazines: Any story more than one column or any item which appeared in the lead at the beginning of the news section of the magazine.
4. Editorial Page Coverage of Newspapers and Magazines: Any item in the lead editorial position (the top left corner of the editorial page) plus all items in which one-third (at least five paragraphs) of an editorial or columnist comment was devoted to political campaign coverage.
Minor items are those stories which are political in nature and included in the study but which are smaller in terms of space, time, or display than major items.
The over-all major item emphasis of the selected mass media on different topics and candidates during the campaign is displayed in Table i. It indicates that a considerable amount of campaign news was not devoted to discussion of the major political issues but rather to analysis of the campaign itself. This may give pause to those who think of campaign news as being primarily about the issues. Thirty-five percent of the major news coverage of Wallace was com-
MAJOR MASS MEDIA REPORTS ON CANDIDATES AND ISSUES, BY CANDIDATES
(le colonne riportano le issues di Nixon, Agnew, Humphrey, Muskie, Fallace, Lemay; l’ultima riporta i totali)
posed of this analysis ("Has he a chance to win or not?"). For Humphrey and Nixon the figures were, respectively, go percent and 25 percent. At the same time, the table also shows the relative emphasis of candidates speaking about each other. For example, Agnew apparently spent more time attacking Humphrey (22 percent of the major news items about Agnew) than did Nixon (11 percent of the major news about Nixon). The over-all, minor item emphasis of the mass media on these political issues and topics closely paralleled that of major item emphasis.
Table 2 focuses on the relative emphasis of each party on the issues, as reflected in the mass media. The table shows that Humphrey/ Muskie emphasized foreign policy far more than did Nixon/Agnew or Wallace/Lemay. In the case of the "law and order" issue, however over half the Wallace/Lemay news was about this, while less than one-fourth of the Humphrey/Muskie news concentrated upon this topic. With Nixon/Agnew it was almost a third-just behind tne Republican emphasis on foreign policy. Humphrey of course spent considerable time justifying (or commenting upon) the Vietnam War- Nixon did not choose (or have) to do this.
The media appear to have exerted a considerable impact on voters' judgments of what they considered the major issues of the campaign (even though the questionnaire specifically asked them to make judgments without regard to what politicians might be saying at the moment) The correlation between the major item emphasis on the main campaign issues carried by the media and voters' ^dependent judgments of what were the important issues was +.967. Between minor item emphasis on the main campaign issues and voters' judgments, the correlation was +-979- In short, the data suggest a very strong relationship between the emphasis placed on different campaign issues by the media (reflecting to a considerable degree the emphasis by candidates) and the judgments of voters as to the salience and importance of various campaign topics.
But while the three presidential candidates placed widely different emphasis upon different issues, the judgments of the voters seem to reflect the composite of the mass media coverage. This suggests that voters pay some attention to all the political news regardless of whether it is from, or about, any particular favored candidate. Because the tables we have seen reflect the composite of all the respondents, it is possible that individual differences, reflected in party preferences and in a predisposition to look mainly at material favorable to one's own party, are lost by lumping all the voters together in the analysis. Therefore, answers of respondents who indicated a preference (but not commitment) for one of the candidates during the September-October period studied (45 of the respondents; the others were undecided) were analyzed separately. Table 3 shows the results of this analysis for four selected media.
The table shows the frequency of important issues cited by respondents who favored Humphrey; Nixon, or Wallace correlated
MASS MEDIA REPORT ON ISSUES, BY PARTIES
INTERCORRELATIONS OF MAJOR AND MINOR ISSUE EMPHASIS BY SELECTED MEDIA WITH VOTER ISSUE EMPHASIS
(a) with the frequency of all the major and minor issues carried by the media and (b) with the frequency of the major and minor issues oriented to each party (stories with a particular party or candidate as a primary referent) carried by each of the four media. For example, the correlation is .89 between what Democrats see as the important issues and the New York Times's emphasis on the issues in all its major news items. The correlation is .79 between the Democrats' emphasis on the issues and the emphasis of the New York Times as reflected only in items about the Democratic candidates.
If one expected voters to pay more attention to the major and minor issues oriented to their own party—that is, to read or view selectively—the correlations between the voters and news/opinion about their own party should be strongest. This would be evidence of selective perception. If, on the other hand, the voters attend reasonably well to all the news, regardless of which candidate or party issue is stressed, the correlations between the voter and total media content would be strongest. This would be evidence of the agenda-setting function. The crucial question is which set of correlations is stronger.
In general, Table 3 shows that voters who were not firmly committed early in the campaign attended well to all the news. For major news items, correlations were more often higher between voter judgments of important issues and the issues reflected in all the news (including of course news about their favored candidate/party) than were voter judgments of issues reflected in news only about their candidate/party. For minor news items, again voters more often correlated highest with the emphasis reflected in all the news than with the emphasis reflected in news about a favored candidate. Considering both major and minor item coverage, 18 of 24 possible comparisons show voters more in agreement with all the news rather than with news only about their own party/candidate preference. This finding is better explained by the agenda-setting function of the mass media than by selective perception.
Although the data reported in Table 3 generally show high agreement between voter and media evaluations of what the important issues were in 1968, the correlations are not uniform across the various media and all groups of voters.
The variations across media are more clearly reflected in Table 4, which includes all survey respondents, not just those predisposed toward a-candidate at the time of the survey. There also is a high degree of consensus among the news media about the significant issues of the campaign, but again there is not perfect agreement Considering the news media as mediators between voters and the actual political arena, we might interpret the correlations in Table 5 as reliability coefficients, indicating the extent of agreement among the news media about what the important political events are. To the extent that the coefficients are less than perfect, the pseudo-environment reflected in the mass media is less than a perfect representation of the actual 1968 campaign. Two sets of factors, at least, reduce consensus among the news
INTERCORRELATION OF MASS MEDIA PRESIDENTIAL NEWS COVERAGE FOR MAJOR AND
media. First, the basic characteristics of newspapers, television, and newsmagazines differ. Newspapers appear daily and have lots of space. Television is daily but has a severe time constraint. Newsmagazines appear weekly; news therefore cannot be as "timely". Table 5 shows that the highest correlations tend to be among like media; the lowest correlations, between different media.
Second, news media do have a point of view, sometimes extreme biases. However, the high correlations in Table 5 (especially among like media) suggest consensus on news values, especially on major news items. Although there is no explicit, commonly agreed-upon definition of news, there is a professional norm regarding major news stories from day to day. These major-story norms doubtless are greatly influenced today by widespread use of the major wire services —especially by newspapers and television—for much political information. But as we move from major events of the campaign, upon which nearly everyone agrees, there is more room for individual interpretation, reflected in the lower correlations for minor item agreement among media shown in Table 5. Since a newspaper, for example, uses only about 15 percent of the material available on any given day, there is considerable latitude for selection among minor
In short, the political world is reproduced imperfectly by individual news media. Yet the evidence in this study that voters tend to share the media's composite definition of what is important strongly suggests an agenda-setting function of the mass media.
The existence of an agenda-setting function of the mass media is not proved by the correlations reported here, of course, but the evidence is in line with the conditions that must exist if agenda-setting by the mass media does occur. This study has compared aggregate units - Chapel Hill voters as a group compared to the aggregate performance of several mass media. This is satisfactory as a first test of the agenda-setting hypothesis, but subsequent research must move from a broad societal level to the social psychological level, matching individual attitudes with individual use of the mass media. Yet even the present study refines the evidence in several respects. Efforts were made to match respondent attitudes only with media actually used by Chapel Hill voters. Further, the analysis includes a juxtaposition of the agenda-setting and selective perception hypotheses. Comparison of these correlations too supports the agenda-setting hypothesis.
Interpreting the evidence from this study as indicating mass media influence seems more plausible than alternative explanations. Any argument that the correlations between media and voter emphasis are spurious—that they are simply responding to the same events and not influencing each other one way or the other—assumes that voters have alternative means of observing the day-to-day changes in the political arena. This assumption is not plausible; since few directly participate in presidential election campaigns, and fewer still see presidential candidates in person, the information flowing in interpersonal communication channels is primarily relayed from, and based upon, mass media news coverage. The media are the major primary sources of national political information; for most, mass media provide the best—and only—easily available approximation of ever-changing political realities.
It might also be argued that the high correlations indicate that the media simply were successful hi matching their messages to audience interests. Yet since numerous studies indicate a sharp divergence between the news values of professional journalists and their audiences, it would be remarkable to find a near perfect fit in this one case. It seems more likely that the media have prevailed in this area of major coverage.
While this study is primarily a sociology of politics and mass communication, some psychological data were collected on each voter's personal cognitive representation of the issues. Shrauger has suggested that the salience of the evaluative dimension—not the sheer number of attributes—is the essential feature of cognitive differentiation. So a content analysis classified respondents according to the salience of affect in their responses to open-ended questions about the candidates and issues. Some voters described the issues and candidates in highly affective terms. Others were much more matter-of-fact. Each respondent's answers were classified by the coders as "all affect," "affect dominant," "some affect but not dominant," or "no affect at all" Regarding each voter's salience of affect as his cognitive style of storing political information, the study hypothesized that cognitive style also influences patterns of information-seeking.
Eschewing causal language to discuss this relationship, the hypothesis states that salience of affect will index or locate differences in the communication behavior of voters. But a number of highly efficient locator variables for voter communication behavior already are well documented in the research literature. Among these are level of formal education and interest in politics generally. However, in terms of The American Voter's model of a "funnel" stretching across time, education and political interest are located some distance back from the particular campaign being considered. Cognitive style is located closer to the end of the funnel, closer to the time of actual participation in a campaign. It also would seem to have the advantage of a more functional relationship to voter behavior.
Examination of the relationship, between salience of affect and this pair of traditional locators, education and political interest, showed no significant correlations. The independent effects of political interest and salience of affect on media use are demonstrated in Table 6. Also demonstrated is the efficacy of salience of affect as a locator or predictor of media use, especially among persons with high
Both salience of affect and media use in Table 6 are based on the issue that respondents designated as the most important to them personally. Salience of affect was coded from their discussion of why the issue was important. Use of each communication medium is based on whether or not the respondent had seen or heard anything via that medium about that particular issue in the past twenty-four hours.
High salience of affect tends to block use of communication media to acquire further information about issues with high personal importance. At least, survey respondents with high salience of affect do not recall acquiring recent information. This is true both for persons with low and high political interest, but especially among those with high political interest. For example, among respondents with high political interest and high salience of affect only 36 percent reported reading anything in the newspaper recently about the issue they believed to be most important. But among high political interest respondents with low salience of affect nearly six of ten (58.3 percent) said they acquired information from the newspaper. Similar patterns hold for all the communication media.
Future studies of communication behavior and political agenda-setting must consider both psychological and sociological variables; knowledge of both is crucial to establishment of sound theoretical constructs. Considered at both levels as a communication concept, agenda-setting seems useful for study of the process of political consensus.
PROPORTION OF MEDIA USERS BY POLITICAL INTEREST AND SALIENCE OF AFFECT
da:The Anatomy of Agenda-Setting Research
by Everett M. Rogers, University of New Mexico, James W. Dearing, Michigan State University, and Dorine Bregman, Institut d'Etudes Politiques de Paris
“Journal of Communication”, 1993, vol. 43, 2
We include studies of media agenda setting, public agenda setting, and policy agenda setting, and refer to the theoretical interrelationships among these three types of research as the agenda-setting process. Media agenda setting includes those studies that conceptualize the mass media news agenda as the main dependent variable of study. Public agenda setting includes those studies that conceptualize the relative importance of issues to members of the public as the main dependent variable of study. Policy agenda setting includes those studies that conceptualize the issue agenda of governmental bodies or elected officials as the main dependent variable of study. Increasingly, some agenda-setting studies include two or three of these dependent variables in their design. Our conceptualization of the agenda-setting process is sufficiently broad to draw intellectual relationships among investigations that may be quite diverse in their conceptualization. However, all agenda-setting studies share an obvious concern with the relative importance of public issues and a less obvious concern with the general functioning of public opinion in a democracy. Ultimately, research on the agenda-setting process seeks to offer one explanation of how social change occurs in modern society.
Why Communication Researchers Research Agenda Setting.
What attracts communication scholars to investigate agenda setting? One main reason for their interest is that agenda-setting research appears to offer an alternative to the scholarly search for direct media effects on attitude change and overt behavior change. Earlier mass communication research had found limited effects, which seemed counterintuitive to many researchers, especially to those (such as McCombs and Shaw) who had previous mass media experience. Further, early mass communication PhDs felt that the media's main purpose was to inform, rather than to persuade or change overt behavior. So they looked for cognitive effects, like agenda setting, in which people are told what to think about. Many of the agenda-setting researchers stated that the main justification for their work was an attempt to overcome the limited-effects findings of past research. For example, McCombs stated in a 1981 overview:
Its [agenda setting 's] initial empirical exploration was fortuitously timed. It came at that time in the history of mass communication research when disenchantment both with attitudes and opinions as dependent variables, and with the limited-effects model as an adequate intellectual summary, was leading scholars to look elsewhere.
Recently, Carragee et ai. (1987) assessed the contribution of agenda-setting research to understanding effects in this way:
Despite important shortcomings, the agenda-setting approach has con tributed to a more advanced understanding of the media's role in society'. It has helped to change the emphasis of mass communication research away from the study of short-term attitudinal effects to a more longitudinal analysis of social impact. This is no small contribution (p 42)
After the mid-1980s, public agenda-setting scholars began to break out of the McCombs-and-Shaw-style methodological mode, while still pursuing the basic question of how the public agenda is set. For example, Shanto lyengar and his collaborators (see, for instance, lyengar & Kinder, 1987) began to investigate the microlevel agenda setting of individual respondents in a series of laboratory experiments with doctored television newscasts (constructed so as to overstress some news issue). Researchers conducted over-time, single-issue studies in which time becomes a variable of study, with the media agenda, the public agenda, and the policy agenda indexed as variables, and subjected to either pretest or posttest designs in order to understand the causal relationships involved in the agenda-setting process (Cook et al., 1983; Leff, Protess, & Brooks, 1986; Protess et al., 1991). Others used rnultirnethod approaches (Rogers et al., 1991).
This proliferation of research approaches used to study the agenda-setting process has given this research specialty a renewed intellectual energy. The original and basic question McCombs and Shaw (1972) asked— whether the media agenda affects the public agenda—was answered fairly thoroughly by the 1980s. Today, a variety of different research methodologies are being used to probe other, related questions about the agenda-setting process. These include:
1. How is the media agenda set?7 How is the policy agenda set? How do the media agenda, the policy agenda, and the public agenda collaboratively influence each other, if they do?
2. Why do "real-world indicators" of an issue not play an important role in the agenda-setting process? For instance, the issue of drug abuse rose to a high position on the U.S. agenda irrt986, and stayed in this high priority for several years. It finally declined after 1989, while the long-term trend in the number of drug-related deaths in the U.S. has slowly and almost linearly declined (Danielian & Reese, 1989).
3. What are the cognitive processes involved in the agenda-setting process at the individual level? Here experiments by lyengar and others, centering on such concepts as priming and framing, are important.
4. How can we measure the public agenda more accurately? Often researchers use a single survey question, such as "What is the most important problem facing the U.S. today?" to represent the public agenda. The exact wording of this question is crucial. For example, while only 6% of the respondents in a 1988 national sample of U.S. adults responded to this question with the word "AIDS," more than 70% of the same rQspon-dents said that AIDS was the most important health problem facing the U.S. at that time (Rogers et al., 1991).