Media, Public Opinion, and Political Action

Holli Semetko. Handbook of Media Studies. Editor: John D H Downing, Denis McQuail, Philip Schlesinger, Ellen Wartella. Sage Publication. 2004.

Research on the subject of public opinion and political action from a media studies perspective has expanded considerably in recent years. Whereas in the mid-1980s, concern about media influence characterized a comparatively small portion of what Kinder and Sears (1986) described as research on public opinion and political action, nowadays it is hardly possible to think of the media as anything but central to research in this area. Many of the studies in this area can also be described as within the interdisciplinary subfield of political communication research, which has also expanded considerably over the past two decades.

Some recent international examples illustrate the variety of reasons why the media are at the center of public debate in many societies around the world. Italy, for example, is classified as an established democracy in comparative research, but many have come to question this status given Prime Minister Berlusconi’s controlling interest in commercial broadcasting in the country and his government’s political influence over public broadcasting (see also Hallin & Papathanassopoulos, 2002). Central and Latin American countries, many of which are societies in transition from military regimes, provide numerous examples over the past two decades of how the news media contribute to the rise and fall of political power (see, e.g., Waisbord, 1995, 1997, 2000). In Russia, Ukraine, and other post-Soviet societies in transition, the media are often at the center of debate not only because of their general pro-government bias (see, e.g., Mickiewicz, 1988, 1999; Semetko & Krasnoboka, 2003) but also because of the many human rights violations against journalists. And in China, where the news media are heavily constrained and access is government controlled, concerns about press freedom to diverge from the party line take on special urgency in light of the sudden acute respiratory syndrome (SARS) epidemic in 2003.

Studies of public opinion and political action are at an interdisciplinary crossroads, and an array of quantitative and qualitative methods characterizes research in this area. The growing body of research on public opinion and political action can still be described as going the most deeply into the U.S. case. Although it may be appropriate to describe journalists as political actors and the news media as a political institution in any society (Cook, 1998; Patterson, 2002; Schudson, 2002), less is known about other parts of the world where different institutional conditions prevail. Growing scholarly interest in media, public opinion, and political action on a global scale is not only because of opportunities presented by the larger number of countries involved in democratization processes over the past two decades but also because of the global expansion of commercial media (McChesney, 1997).

Even in an established democracy such as the Netherlands, which is representative of European multiparty parliamentary systems, in which news organizations are largely independent from party influence and levels of party identification and electoral turnout are comparatively higher than the United States, there is evidence that one’s viewing behavior has consequences that cannot always be thought of as “virtuous”: Watching television news regularly on the public service channels, for example, has positive effects on political learning, feelings of political efficacy, and turning out to vote, whereas regularly opting for commercial television news has consistent and significant negative effects on these variables. This research thus supports a “dual-effects” hypothesis (Aarts & Semetko, 2003). Especially given the wide range of commercial broadcasting options that citizens in most countries now have, the role of television in politics today is under conditions quite apart from, for example, Britain’s first television elections in the late 1950s and early 1960s, when everyone had only a couple of channels to which they could turn for news and reporters were quite reverent and descriptive (as opposed to evaluative) in their coverage of candidates and leaders (Blumler & McQuail, 1969).

There is also special concern about the potential consequences of contemporary media environments for democracy building in societies in transition, in which party systems are still in their infancy, news organizations and journalists are often in precarious positions, and there are low levels of party identification and electoral experience among citizens. In these societies in transition as well as in established democracies, analysis of survey data can often show a positive relationship between media use and satisfaction with democracy, trust in institutions, and other measures of political attitudes (Dalton, 1996; Norris, 2000), but given the superficial and limited range of questions about media use in those surveys, such a general conclusion may mask the more complex set of relationships that may be operating.

In the limitations of one chapter, it is impossible to cover the historically important studies from communication, political science, psychology, and sociology that contribute to our present understanding of the media’s role in public opinion formation, public political action, and the link between these two processes. I have therefore chosen to focus primarily on fruitful opportunities for interdisciplinary and multi-methodological research in this area by reviewing some of the main findings from research that has been conducted from a media studies perspective over the past two decades. Such a perspective takes media as a key part of the focus of public opinion and political action research, with an emphasis on one or more of the following: use, contents, institutions, professionals, and media systems. I also do not include all of the research published in the past 10 to 20 years in this area but instead concentrate on some of the concepts, hypotheses, and perspectives that I believe are important for the future. In particular, I focus on the research about the formation of and changes in attitudes, the research on social movements, and studies of frames and framing effects.

Studying Attitude Stability and Change

The question of consistency in political and policy attitudes was the focus of Philip Converse’s seminal research on attitude stability. Converse (1962, 1964) argued that most Americans have no consistent pattern to their opinions or beliefs and that their opinions do not depend on ideological principles. An analysis of American National Election Study (ANES) panel data from 1956, 1958, and 1960 found that so many changed their opinions on policy issues that it appeared as if answers had been given at random. The general conclusion was that for most individuals, policy attitudes were for the most part nonattitudes. From this perspective, shifts in public opinion are largely explained by a lack of knowledge, interest, and ideology that results in randomness of opinion.

There are many contrasting views on public opinion and what moves it. The work by Page and Shapiro (1992) stands out because it describes a “rational public” whose opinions are moved by information in a way that reveals consistency between policy preferences and basic values. They focus on aggregate public opinion in the United States and show that it is largely stable over time on many aspects of public policy. This does not necessarily contradict Converse (1962, 1964), who described individual-level opinion change as largely random because these random changes at the individual level could appear stable at the aggregate level. Page and Shapiro (1992) argue, however, that public opinion moves in response to events or new information about an issue. They identified parallel publics (cohorts or demographic groups) who used similar standards to assess issues and whose opinions largely moved in the same direction, and they conclude that opinion is rational and that attitudes exhibit consistency.5

These two examples illustrate one of the central problems in public opinion research from a media effects perspective. It is the tension between the individual level and the aggregate level. Most studies tend to focus on one level or the other and often employ only a single methodological approach.


Each method has its own advantages and disadvantages. Experiments allow for the maximum control on the part of the researcher of the environment within which the study takes place and therefore result in high internal validity, allowing us to specify cause-and-effect relationships. But media effects experiments also have special disadvantages, not the least of which is the laboratory (not a natural setting) and the stimulus material (which may distort what is actually in the news). Surveys, by contrast, may be nationally representative and are conducted in naturalistic settings. Cross-sectional surveys nevertheless rely on reported measures of communication exposure and correlations between these and reported attitudes. A panel survey, in which the same group of respondents is interviewed on separate occasions with the advantage of the naturalistic setting, permits causal analysis of effects. Panel studies are not without problems, however, the most common ones being attrition rates and contamination effects. In addition to these approaches, depth interviews, focus groups, and deliberative events are also common to research public opinion change and have been the basis for some pioneering studies on how people talk about and process political news (Gamson, 1992; Graber, 1988, 2001), frame it (Neuman, Just, & Crigler, 1992), and interpret, argue, and “deliberate” political issue information (Price, Cappella, & Nir, 2002). Content analysis has been often ignored by public opinion researchers, even though reliable and valid content indicators can be important for an understanding of the sources of media effects. Because each of the above methods of research has its own inherent strengths and weaknesses, a genuine understanding of the effects of communications on attitudes ideally requires a design in which a number of methods are used in a complementary fashion.

Multilevel Analysis

Public opinion research that is survey based has also developed sophisticated tools for taking into account the contexts in which public opinion is measured. And data can be at different levels—for example, the individual level, the organizational level, and the system level. Multilevel analysis permits researchers to model the different layers in the structure of multilevel data and determine how layers interact and influence the dependent variable (Steenbergen & Jones, 2002). Multilevel analysis is an especially relevant form of modeling data for the study of media influence in a comparative research design in which contextual variables are also important (Shah, McLeod, & Yoon, 2001).

In the process of European political and economic integration, in which many countries are giving up some aspects of authority to another level of government and publics are forming opinions about national-level as well as European-level issues and institutions, multilevel analysis is especially useful for the study of opinion formation (Rohrschneider, 2002). It can also be useful to assess the impact of the news on political behavior—specifically, the decision to vote or, more appropriately, not to vote. In the 1999 European Parliament elections, for example, turnout was down from the previous election and reached the exceptionally low levels of 23% in the United Kingdom and 30% in the Netherlands and Finland, in comparison with highs elsewhere of 64% in Spain, 71% in Italy, and 70% in Greece.

Research on electoral turnout in European parliamentary elections has, for the most part, included some of these contextual variables alongside traditional demographic characteristics (education, age, gender) but, with few exceptions (Blumler, 1984) and until recently, has largely ignored the contribution of the media or the campaign information environment to mobilizing or demobilizing electors (Niedermyer & Sinnott, 1998; Sinnott, 1995, 1998; van der Eijk & Franklin, 1996). Theories about party competition and factors influencing electoral turnout, such as the day an election is held or the presence of more than one election in one country on the same day, can be tested using a multilevel analysis that permits the integration of quantity and tone of media reporting (about the 1999 European parliamentary election campaigns in each country) into the model predicting turnout or abstention. This model also includes potentially important national contextual influences such as, for example, concurrently held national or local elections, Sunday voting, and the presence of a viable anti-European Union (EU) party (Banducci & Semetko, 2002).

The Media as a Facilitator of Opinion and Action: The Link between the Individual Level and Societal Level

Media are often conceptualized as the primary connection between an individual and the society. One important line of research deals with the citizen as a consumer of news and media information, as well as a member of a community, and how mediated information and personal experience influence opinions and action. A central focus has been to identify the ways in which media uses and contents influence individual opinion, within the context of one’s personal network of political discussants. This has emphasized the consequences of exposure to information in the media, in comparison with exposure to information via personal networks, for the development and perception of political attitudes.7 Does the information in the media enhance or diminish the influence of personal experience and networks on the formation of political attitudes?

One study focused on individuals’ personal experience with a problem or issue and how this can become politicized in the context of news and media information sources. By reporting people’s experiences and linking them to the experiences of others, the media help people to interpret their own personal experience as part of a larger societal trend. According to Mutz (1994), for example, “Mothers who have lost their children to automobile fatalities have long been against drunk driving, but until the issue became highly publicized through the efforts of Mothers Against Drunk Driving, their personal experiences were not highly politicized” (p. 691). In this way, the media may help the victim of a crime or the unemployed worker to come to realize that she or he is one of many across the country or in the community. The media therefore may contribute to the politicization of one’s personal experience because this can, in turn, affect political opinions and political preferences and can potentially induce people to take political action. The problem of unemployment is one that is often covered in the news and is one with which many individuals also have personal experience. According to Mutz (1994),

Exposure to unemployment news appears to strengthen the impact of personal experiences [with unemployment] on presidential performance ratings. Heavy unemployment coverage also increases the extent to which perceptions of national unemployment conditions are generalized from personal experience. Overall, … mass media may … help legitimize the translation of private interests into political attitudes. (p. 689, emphasis added)

Another study addressed the extent to which individuals learn more about political views dissimilar to their own via the media in comparison with one’s interpersonal network of political discussants. This is an important question not least in the U.S. context, in which the media have been criticized for not offering a wide range of political viewpoints. Yet Mutz and Martin (2001) find that

individuals are exposed to far more dissimilar political views via news media than through interpersonal political discussants. The media advantage is rooted in the relative difficulty of selectively exposing oneself to those sources of information, as well as the lesser desire to do so, given the impersonal nature of mass media. (p. 97)

The media are therefore described as “facilitating communication across lines of political difference.” Impersonal influence refers to the important role played by media in shaping perceptions of societal-level trends and developments. According to Mutz (1998),

Media play a particularly important role in shaping impersonal perceptions, and impersonal perceptions, in turn, play a particularly important role in shaping political judgments. Nonetheless, citizens are not necessarily doomed to excessive conformity or to puppet-like manipulation at the hands of those who control media content… the normative social influence conveyed by interpersonal political discussion is obviously important, but the American public also demonstrate a great deal of independence, even in face-to-face settings. (p. 270)

The American penchant for demonstrating independence in political discussions is by no means a universal trait. The “spiral-of-silence theory,” one of the most widely debated theories in public opinion research, is based on the idea that one’s willingness to express one’s own political opinions is constrained by how she or he perceives the climate of public opinion (Noelle-Neumann, 1974). Dozens of studies have been conducted to address key hypotheses in the theory over the past two to three decades. A review of this body of research shows that inconsistencies in conceptualization, operationalization, and a lack of attention to macroscopic variables explain much of the variance in the spiral-of-silence effects. Scheufele and Moy (2000) conclude,

Spiral of silence studies in different cultures have failed to take into account culture-specific variables that may mitigate the importance of opinion perceptions as predictors of individual behavior or attitudes. In other words, cross-cultural differences are key factors in predicting speaking out, the key dependent variable in spiral of silence research. As a result, we call for the return to a more macroscopic focus in spiral of silence research. (p. 3)

The perceptions of the climate of opinion, or the distribution of public opinion on an issue, can also influence an individual’s willingness to participate in political activities such as expressing one’s opinion and other forms of participation (Scheufele & Eveland, 2001). Another study has found an important interaction effect between reliance on hard news media and the frequency of political discussion. The effects of hard media use are different for people who often talk about politics with others in comparison with those who do not (Scheufele, 2002).

New Media and Civic Engagement

How citizens connect and engage with civic life has been potentially enhanced with the arrival of the Internet. Amidst public concern about the lack of interest that young people have in reading newspapers or watching news, the Internet provides new opportunities for entertainment and information. A multilevel research design that takes into account individual-level media use and community-level variables explored the influence of both on civic engagement, measured as interpersonal trust and participation. Although, not surprisingly, the study found that reliance on the media for information purposes is positively related to the production of social capital, whereas entertainment uses of the media are negatively related to civic participation, there was also an important and surprising interaction effect (Shah, McLeod, et al., 2001):

Informational uses of mass media were also found to interact with community context to influence civic engagement. Analyses within sub-samples find that among the youngest adult Americans, use of the Internet for information exchange more strongly influences trust in people and civic participation than do uses of traditional print and broadcast news. (p. 464)

A related study based on the same data set further explores the associations between Internet use as compared to other media among different age cohorts and reinforces the point that Internet influence on the production of social capital is primarily confined to the younger (“Generation X”) cohort. The authors conclude (Shah, Kwak, & Holbert, 2001),

Although the size of associations is generally small, the data suggest that informational uses of the Internet are positively related to individual differences in the production of social capital, whereas social-recreational uses are negatively related to these civic indicators. Analyses within sub-samples defined by generational age-breaks further suggest that social capital production is related to Internet use among Generation X, while it is tied to television use among Baby Boomers and newspaper use among members of the Civic Generation. (p. 141)

Political Learning

Doris Graber’s (2001) research demonstrates how vital visuals are for the retention and understanding of information, particularly political information. Television and, increasingly, the Internet, as it becomes more graphics than text driven, are therefore important sources of information for political learning. According to Graber (2001),

Audiovisuals ease two major information-processing problems: failure to embed information in long-term memory and inability to retrieve it when needed…. All else being equal, when messages include visuals rich in relevant information, memory is enhanced … its accuracy improves as well. Good visuals make a situation more graphic and vivid… they are etched more deeply into memory initially than non-visual messages. In turn, because they are more easily recalled, they are frequently refreshed, which then prevents fading. (pp. 33–34)

Research shows that there is a difference between unsophisticated and sophisticated television viewers in coping “with messages that require complex processing at both the verbal and the audiovisual levels” (Graber, 2001, p. 35; see also Rahn & Cramer, 1996). One reason why audiovisuals do not consistently enhance learning or recall is because many TV news visuals are “totally uninformative” (Graber, 2001, p. 35). In comparison with television news, news on the Internet is received by a more interested, active (and interactive) audience.

Media, Public Opinion, and Protest as Political Action

Much of the research on protest and collective action has taken a comparative historical approach with case studies from the 19th century or earlier. Political opportunity structure is an important theoretical concept in the arsenal of social movement theories that has provided a framework for much of the collective action research in which the 20th century is the focus (McAdam, Tarrow, & Tilly, 2001; Tarrow, 1998). Political opportunity structures refer to “consistent—but not necessarily formal or permanent—dimensions of the political environment that provide incentives for people to undertake collective action by affecting their expectations for success or failure” (Tarrow, 1998, p. 85). Public opinion is one of those consistent dimensions of the environment. But the role of public opinion, particularly how the media reporting of events and of public opinion contributes to perceptions of “incentives” for undertaking collective action, has not been central to the research conducted from within this perspective. There has been attention to rhetorical framing of arguments from those working within a political opportunity structure approach (Koopmans & Duyvendak, 1995), but as a whole, the discussion of the media takes the form of references to other studies involving framing (Gamson & Modigliani, 1989) and some attention to the way various actors, including the media, frame political opportunity (Gamson & Meyer, 1996). There has been a call for an effort to “insert a dimension of public discourse within a political opportunity perspective” (Koopmans & Statham, 2000, pp. 36-37), but although content analysis has been used, it has not been for the purpose of assessing the contribution of the media to collective action from a political opportunity structure perspective. A cross-national comparative study uses a content analysis of quality newspapers in Germany and Britain to provide a record of the “claims-making” activities of various groups, but the way this has been done fails to consider from a political opportunity perspective any questions about media power in shaping news agendas or media effects on collective action processes. There remains the problem of using the media as an indicator of claims-making activities in the public sphere, without acknowledging that the news organization or outlet can be an independent actor in the process.

Social protest, social movements, and collective action provide a fascinating setting for the study of the influence of media and communication process on public opinion, as studies of the student, antiwar, and environmentalist movements have shown (see, e.g., Gamson, 1988; Gitlin, 1980; Hallin, 1986). Public opinion and media coverage of events and opinions may help to facilitate or diminish opportunities for collective action. Research has shown that conceptions of public opinion are embedded in the news coverage of social protest (McLeod & Hertog, 1992). There are common approaches to framing protest in the news (Hertog & McLeod, 1995), and the term protest paradigm has been used to describe the formulaic approach that journalists take in reporting social protest (McLeod & Detenber, 1999). These common characteristics of protest reporting can have important consequences for the opinions of audiences. For example, when the news is framed in a way that supports the status quo in the coverage of one social protest, an experiment found that “viewers were more critical of and less likely to identify with the protesters, less critical of the police, and less likely to support the protesters’ expressive rights” (McLeod & Detenber, 1999). Support for the status quo in the news coverage also led to lower estimations of the effectiveness of the protest, perceptions of its newsworthiness, and public support for the protest.

The framing of issues in the media also has effects on the mobilization of social movements. Media framing of issues has been described as an alternative or complementary explanation to the influence of public opinion, elite cues, and political opportunity structure on movement mobilization (Cooper, 2002). Cross-national comparative research on protest shows that “congruence” between the media’s and the movement’s framing of the issues, for example, can facilitate mobilization but that mobilization is hindered if there is “divergence” (Cooper, 2002).

The student peace movements during the intensification of the Vietnam War and the media coverage of the opposition to the war, along with the shifts in public support for the war over time, provide a basis for addressing questions about information, opinion change, and political action. The Vietnam War was the case used by John Zaller (1991, 1992) to explain both individual-level and aggregate-level changes in one model of U.S. public opinion. Whereas past research at the individual-level “modeled mass opinion change as a two-step process involving reception of political communication and acceptance or rejection of that communication,” Zaller (1991, p. 1215) proposed “a two-message version of the reception-acceptance model, in which citizens are exposed to two opposing communication flows, either or both of which may affect their opinions.” Different levels of attention to politics and different political values among citizens, along with variation over time in the intensity of oppositional messages in the media, interact in this model to explain both aggregate-level shifts and individual-level changes over time.

The Conditioning Effects of Political Involvement

The more politically involved are generally characterized by greater political knowledge, attentiveness, exposure to political information, and attachment to certain political values or beliefs. The literature does not provide a clear answer on what to expect concerning the relationship between attitude change and political involvement. Converse (1964) would predict that citizens with the lowest levels of exposure, interest, knowledge, and beliefs would change only randomly or not at all; those who were at the highest levels would remain stable or change systematically; and those in the middle levels would be most open to influence from the information environment. Following Converse, Zaller (1992) also argues that a more detailed distinction in levels of awareness or attentiveness is necessary because the relationship between attitude change and political awareness may be nonlinear. Awareness or attentiveness is operationalized with the use of a scale or index of political knowledge by Zaller (1991, 1992), and the expectation is that citizens with intermediate knowledge levels would be most likely to be affected by information. His model of public opinion change has inspired a host of studies over the past decade.

There continues to be debate, however, about the validity of the measures used to establish political knowledge levels. The importance of political knowledge as a predictor of political involvement is acknowledged in the literature, but there remains disagreement over how it should best be measured (see, e.g., Delli Carpini & Keeter, 1993; Graber, 2001; Luskin, 1990; Mondak, 2001). Recent research likens the factual political knowledge questions used in most survey-based studies (including the ANES) to a high school civics exam and questions the ability of these questions to capture actual political understanding and awareness (Graber, 2001; Mondak, 2001).

Agenda Setting, Priming, and Framing

Although some researchers have attached the word theory to each of the above terms, others refer to agenda setting, priming, and framing as key concepts or processes in public opinion research. There is also debate about whether and how these concepts, processes, or theories, in terms of effects on audiences, are actually related (McCombs, Lopez-Escobar, & Llamas, 2000; Scheufele, 2000). I do not wish to elaborate the debate over the status of these as concepts or theories or take sides in the discussion over their independence or interdependence. Below I discuss briefly the developments in agenda-setting research before turning to key questions in priming and framing research.

Agenda Setting

Much of the research testing the agenda-setting hypothesis has drawn on two sources of data. One is content analysis of the news. This is used to establish the most important issues in the news. The other measures public opinion. The latter has included cross-sectional designs, time series of comparable cross-sectional surveys, more elaborate panel studies, and experiments. Support for the hypothesis of media agenda-setting effects has been found in hundreds of studies over the past few decades (for reviews, see McCombs, 1981, 2004; McCombs, Einsiedel, & Weaver, 1991; McCombs & Shaw, 1993; Rogers & Dearing, 1988; Rogers, Dearing, & Bregman, 1993; Stempel, Weaver, & Wilhoit, 2003; Swanson, 1988; Weaver, McCombs, & Spellman, 1975; Weaver, Graber, McCombs, & Eyal, 1981). These studies provide substantial support for the hypothesis that the most prominent issues in the news are also the issues that become the most important in public opinion. The fact that support for the agenda-setting hypothesis has been found using all of these of methodological approaches further strengthens the argument for a powerful news media.

The term itself was first coined by McCombs and Shaw (1972), based on their community study of media agenda setting in a U.S. presidential campaign, which involved content analysis and cross-sectional survey data. The study was “pioneering” for providing evidence of a strong and significant correlation between the campaign agenda in the media and in public opinion, as well as setting forth the hypothesis that agenda setting is a process led by the news media (Rogers & Dearing, 1988).

A yearlong panel study of the 1976 U.S. presidential election, designed specifically to test the agenda-setting hypothesis, provided further evidence of the direction of the causal link from media agendas to public agendas (Weaver et al., 1981). The authors identified the relative strengths of television and the newspapers as influences on public opinion and established that agenda-setting effects varied over time, with the strongest during the spring and summer and the weakest during the final 12 weeks of the campaign, particularly with respect to unobtrusive issues such as foreign affairs, the environment, and government credibility. Patterson’s (1980, p. 159) multiwave panel study of the same 1976 campaign also found that newspapers were more important vehicles than television news for voter learning about the issues and established that the more heavily the issue was reported in the press, the more readers with low or moderate interest in the election learned about the issue.

Iyengar and Kinder (1987, p. 12) tried to diminish the artificiality of the experimental setting and drew their sample from heterogeneous groups of people (not only college sophomores) and followed a strategy of conceptual replication in a series of experiments testing the agenda-setting power of television news. Findings combined from three sources—a series of sequential experiments each lasting 1 week (with pre- and postexperimental questionnaires), additional experiments that required only postquestionnaires, and an analysis of time-series data from national public opinion polls—provided further substantial support for the agenda-setting hypothesis. They thus concluded, “By attending to some problems and ignoring others, television news shapes the American public’s political priorities. These effects appear to be neither momentary, as our experimental results indicate, nor permanent, as our time-series results reveal” (Iyengar & Kinder, 1987, p. 33). The study was innovative in its focus on television news as an agenda setter in normal (nonelection) periods.

Media power to influence public agendas may be conditioned by a number of factors. One is the extent to which real-world indicators reinforce or diminish the media message. Another is the type of issue—“unobtrusive” issues may be more susceptible to agenda setting. News about foreign affairs and foreign peoples, with whom viewers have little or no direct experience, is an example of an “unobtrusive” issue (see also Baum, 2002). Media influence may also vary with the type of public agenda. An individual’s civic agenda (his or her perception of the most important issues or problems facing the community) may be quite different from an individual’s personal agenda (his or her opinion about the most important problems he or she is facing). When an individual’s personal agenda is the focus of research, then there is considerably less support for the agenda-setting hypothesis (McLeod, Becker, & Byrnes, 1974).

Election campaigns outside the United States have been the focus for a number of studies that have not found evidence in support of the agenda-setting hypothesis. British general elections in the 1980s and 1990s have provided more than one example in which media agendas and audience agendas failed to coincide (see, e.g., Miller, Clarke, Harrop, LeDuc, & Whiteley, 1990; Norris, Curtice, Sanders, Scammell, & Semetko, 1999), and in Germany too, agenda-setting effects failed to materialize in the historic 1990 election (Semetko & Schoenbach, 1994).


In a study of citizens’ responses to Watergate, Weaver et al. (1975) argued

that for persons with a high need for orientation about politics, mass communication does more than merely reinforce preexisting beliefs. In fact, the media may teach these members of the audience the issues and topics to use in evaluating certain candidates and parties, not just during political campaigns, but also in the longer periods between campaigns. (p. 471)

They refer to the process that came to be later described by the term priming in political communication research. Other early studies also provided evidence of the media’s role in shaping the standards by which citizens evaluate political leaders and candidates (see, e.g., Patterson, 1980; Patterson & McClure, 1976; Protess & McCombs, 1991; Weaver et al., 1981). The term priming has been defined broadly by Fiske and Taylor (1984) as the effects of prior context on the interpretation and retrieval of information, as well as more specifically by Iyengar and Kinder (1987) and Krosnick and Kinder (1990) as changes in the standards used by the public to evaluate political leaders. Their experiments also provide support for the priming hypothesis.

Krosnick and Kinder (1990) used a simple knowledge test (a count of the number of political elites a respondent was able to identify correctly) to partition their respondents into two groups—political “novices” (with low knowledge) and “experts” (with high knowledge)—and argued that the novices are most susceptible to media influence. Krosnick and Brannon (1993) investigated the (separate and combined) roles of knowledge, interest, and exposure in the priming of evaluations of U.S. President George Bush over the period surrounding the 1990–1991 Gulf War when his approval ratings jumped more than 20 percentage points. They concluded that the relationship between knowledge and priming effects is more complex than previous research suggests and actually revised the conclusion of Krosnick and Kinder (1990) with respect to the effects of knowledge. Krosnick and Brannon point toward a social psychological model of information processing:

Greater knowledge constitutes a greater ability to interpret, encode, store, and retrieve new information. And higher levels of exposure and interest are associated with a greater likelihood of forming online political evaluations and a dilution of priming effects due to a wider range of knowledge being activated by media coverage. (p. 972)

Miller and Krosnick (2000) went further to elaborate the importance of trust when the moderating role of political expertise in information processing is considered. They argue,

Media coverage of an issue does indeed increase the cognitive accessibility of related beliefs, but this does not produce priming. Instead, politically knowledgeable citizens who trust the media to be accurate and informative infer that news coverage of an issue means it is an important matter for the nation, leading these people to place greater emphasis on that issue when evaluating the President. Thus, news media priming does not occur because politically naive citizens are “victims” of the architecture of their minds, but instead appears to reflect inferences made from a credible institutional source of information by sophisticated citizens. (p. 301)

There are at least two recent overviews of the literature on priming. One brings together the 30-some studies from the fields of psychology, communication, and political science that deal explicitly with media priming, and its well-developed discussion of what constitutes media priming makes this a theoretically important study as well as a valuable critical assessment of the literature (Peter, 2002). Another is a meta-analysis of the priming literature that incorporates the research in the areas of violence as well as politics, based on 26 published articles representing 42 studies, and in conclusion raises questions about whether media priming actually shares common characteristics with the type of priming studied by cognitive and social psychologists (Roskos-Ewoldsen, Klinger, & Roskos-Ewoldsen, 2002). The study suggests that one of the important questions for future media priming research, which is a characteristic of psychological priming research, is whether more intense media primes result in stronger priming effects.

These two overviews of the priming literature also emphasize the need to further distinguish, both theoretically and operationally, priming effects from what has been described as that which is “chronically accessible” (see, e.g., Lau, 1989). Another relevant study of priming effects put it this way: “Future research should conceptualize priming more broadly to include considerations of both the accessibility of cognitions in short-term memory and the pathways among information in long-term memory” (Domke, Shah, & Wackman, 1998).

The priming literature as a whole suggests that effects are mediated by levels of political involvement, as measured by knowledge, exposure, and interest. But the groups most likely to be primed and the sign or direction of the priming effect remain unclear. In fact, the complex interactions between knowledge, exposure, and interest led Krosnick and Brannon (1993) to revise the conclusion of Krosnick and Kinder (1990) on the knowledge and attentiveness (exposure and interest) groups most likely to be primed and the sign or direction of the priming effects of these variables. There is also some evidence to suggest that priming effects may occur across the board and may not always be mediated by levels of political involvement (see also Peter, 2002).


Framing research, like agenda-setting research, focuses on the relationship between issues in the news and the public perceptions of these issues. But it also goes further and “expands beyond agenda-setting research into what people talk or think about by examining how they think and talk about issues in the news” (Pan & Kosicki, 1993, p. 70; see also Pan & Kosicki, 2001). Another study actually concludes “that a model combining the theories of agenda setting and framing provides a better explanation for the shifts in aggregate opinion than either theory on its own” (Jasperson, Shah, Watts, Faber, & Fan, 1998, p. 205).

Our understanding of frames and framing effects has been the subject of much research over the past few decades (e.g., de Vreese, 2002; de Vreese, Peter, & Semetko, 2001; de Vreese & Semetko, 2004, in press; Edelman, 1993; Entman, 1991, 1993; Fiske & Taylor, 1984; Gamson, 1992; Goffman, 1974; Graber, 1988, 1993; Iyengar, 1991; Jasperson et al., 1998; Neuman et al., 1992; Price, Tewksbury, & Powers, 1997; Semetko & Valkenburg, 2000; Tuchman, 1978; Zaller, 1992). Frames set the parameters “in which citizens discuss public events” (Tuchman, 1978, p. iv). They are “persistent selection, emphasis, and exclusion” (Gitlin, 1980, p. 7). The process of framing refers to selecting “some aspects of a perceived reality” to enhance their salience “in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation” (Entman, 1993, p. 53). Framing effects have been defined as “changes in judgment engendered by subtle alterations in the definition of judgment or choice of problems” (Iyengar, 1987, p. 816) or “one in which salient attributes of a message (its organization, selection of content, or thematic structure) render particular thoughts applicable, resulting in their activation and use in evaluations” (Price et al., 1997, p. 486). A working definition of framing that brings together the study of frames and framing effects is proposed by Stephen Reese in the prologue to Framing Public Life: “Frames are organizing principles that are socially shared and persistent over time, that work symbolically to meaningfully structure the social world” (Reese, Gandy, & Grant, 2001, p. 11).

Two possible approaches to identifying frames in the news emerge from a review of the literature:

The inductive approach involves analyzing a news story with an open view to attempt to reveal the array of possible frames, beginning with very loosely defined preconceptions of these frames (see, for example, Gamson, 1992). This approach can detect the many possible ways in which an issue can be framed, but this method is labor intensive, often based on small samples, and can be difficult to replicate.

A deductive approach involves predefining certain frames as content analytic variables to verify the extent to which these frames occur in the news. This approach makes it necessary to have a clear idea of the kinds of frames that are likely to be in the news, because the frames that are not defined a priori may be overlooked. This approach has the advantage that it can easily be replicated, it can cope with large samples, and it can easily detect differences in framing between media (e.g., television vs. press) and within media (e.g., highbrow news programs or newspapers vs. tabloid style media). (Semetko & Valkenburg, 2000, pp. 94-95)

A number of news value-driven frames emerge from a review of the literature. The conflict frame, for instance, has been the subject of much discussion in the United States, where reporters have been criticized for being too negative in the coverage of politics (Capella & Jamieson, 1997; Patterson, 1993). Neuman et al. (1992) identified several different frames that were common in the U.S. news coverage of a range of issues, and these frames included conflict, economic consequences, human impact, and morality frames. Iyengar (1991) explicitly studied how audience members framed who was responsible for various social problems, after he exposed them to issues imbedded within what he described as frames: episodic, which refers to stories that report specific events, and thematic, which refers to stories that involve analytical, contextual, or historical coverage. One of the problems with much of the research on frames in the news is the lack of specification of measures and the lack of attention to the validity and reliability of measures.

Taking a deductive approach to analyzing frames in the news based on a review of the most common frames in the news—and in an effort to bring greater validity and reliability to the study of news frames—one study of national news coverage of politics in one European country developed reliable scales to measure “conflict,” “human interest,” “economic consequences,” “morality” and “attribution of responsibility” frames in the news (Semetko & Valkenburg, 2000, p. 100). The authors found that although the use of frames varied significantly by news outlet (the more sensationalist outlets vs. the more sober ones), the most common frame in political news was attribution of responsibility, followed by conflict. They also measured the extent to which news was “episodic” or “thematic” in Iyengar’s (1991) terms.

The attribution-of-responsibility frame was defined as that which presents an issue or problem in such a way as to lay blame or credit for its cause or solution either to the government or to an individual or group (Semetko & Valkenburg, 2000). This was an effort to further develop the work of Iyengar (1991), who studied how audiences framed responsibility but who never actually measured how the news framed responsibility for causing and solving social problems. Iyengar argued that the medium of television—specifically, television news, by covering an issue or problem in terms of an event, instance, or individual (“episodically”) rather than in terms of the larger historical social context (“thematically”)—encourages people to offer individual-level explanations for social problems. That explained why his framing experiments found that the poor woman on welfare is held responsible for her fate rather than the system or the U.S. government.

Television news in Europe, as in the United States, is also predominantly “episodic,” but the content analysis of European political news revealed that at the same time, responsibility was often attributed to the government and not the individual. These findings suggest that although television news in many countries may be predominantly episodic because of news values and preferences for news formats, the way in which responsibility is framed in the news is influenced by the political cultures and social contexts in which the news is produced. In Western European democracies, unlike the United States, where the welfare state is comparatively strong, the government is expected to provide answers to social problems. Television news therefore can be episodic and at the same time frame the government (rather than the individual) as responsible for social problems such as poverty. Although much of the work on framing is conducted in one country and primarily the United States, it is nevertheless important to consider how other contexts may circumscribe the generalizations one may be tempted to draw from the data.

Media and Elections

Studies of elections and campaign-induced opinion change have incorporated some aspects of a media studies perspective. The sheer amount of U.S. literature now available on questions about opinion change and campaign influence, as well as media uses and impacts in elections at various levels, is so large that a thorough discussion of the findings cannot fit within the parameters of this chapter. A number of studies, for example, based on presidential campaigns in the past two decades emphasize the importance of presidential campaigns (Holbrook, 1996; Popkin, 1994), call into question the so-called “minimal effects” model of elections (Finkel, 1993), and identify the impacts of campaign events, media appearances, and advertising on voting (Shaw, 1999; Shaw & Roberts, 2000). The U.S. 2000 presidential campaign also saw major studies conducted by Thomas Patterson and others at Harvard University (see, and there were a number of major research projects on the media and the campaign launched by researchers at the Annenberg School for Communication at the University of Pennsylvania. A special issue of Political Communication (Bimber, 2002) was devoted to the topic of political communication in this election, which ironically had more people paying attention to it after Election Day than before.

Media effects may be cognitive (effects on political knowledge), attitudinal (effects on political opinions), or behavioral (effects on turnout and vote choice). A number of election studies in the 1980s and 1990s identified significant media effects in national election campaigns in a number of countries outside the United States. These include Britain (Curtice & Semetko, 1994; Miller et al., 1990; Norris et al., 1999), France (Semetko & Bórquez, 1991), Germany (see, e.g., Finkel & Schrott, 1995; Pfetsch & Voltmer, 1994; Schmitt-Beck & Schrott, 1994; Semetko & Schoenbach, 1994), the Netherlands (Aarts & Semetko, 2003), Canada (Carty & Eagles, 2000; Jenkins, 1999; Mendelsohn, 1996), and New Zealand (Vowles et al., 2002).

The influence of media in referendum campaigns has only more recently received scholarly attention. Because a referendum is fought on one issue, and because that issue can often cut across party lines, the news media may be a more important source of information to voters who cannot fall back on their traditional party sympathies for guidance on how to vote or motivation on whether to vote. As the European Union member states increase from 15 to 25 in 2004, national referendums on issues about European integration will become increasingly common. Research on the influence of the media in European referendums to date has revealed that the campaign news matters not only for vote choice but also for citizens’ perceptions about the campaign and the various aspects of the referendum issue, as well as for evaluations of political leaders during the campaign and in the period after the referendum result (de Vreese & Semetko, 2002a, 2002b, 2004, in press).

Negative Campaigning

The effects of negative political advertising in elections has been the topic of much U.S. research and was also the focus of a symposium in the American Political Science Review (Volume 93, December 1999). A meta-analysis of the large body of research into the effects of negative advertising, however, shows that there is no clear answer to the question of whether it is actually effective. Research by Kim Kahn and Patrick Kenney (1997, 1999) on the effects of information on intense and hard-fought campaigns in U.S. Senate races found that the uses and impact of information by those with high social capital were different from those with low social capital. The latter rely more on party identification, issues, presidential approval, and assessments of personal and economic circumstances, whereas those with higher social capital are less affected by changes in the campaign environment. And with respect to the likelihood of turning out to vote on Election Day, Kahn and Kenney (1999) found that negative campaigning also had an influence. But the tone of the campaign has a greater impact on those with low social capital, such as no party attachment, little or no interest in politics, and less political knowledge, particularly when the nature of the negative campaigns shifts from legitimate criticisms to unsubstantiated attacks.


This discussion of public opinion and political action has summarized some of the developments in research from a media studies perspective, as well as some research that has not emphasized media and media effects as central in the processes of social change. Research from a media studies perspective considers the characteristics of media content and the factors that produce media content, the actions of journalists and news organizations, and how these together may influence public perceptions.

A media studies perspective provides a potentially fruitful avenue for theory building. In the study of collective action, for example, it would be worthwhile to devote more attention to the impressive sociological literature on political opportunity structures and how this might be elaborated from a media studies perspective. Such an approach would aim to integrate the media as a political actor into the larger arena of opportunity structure.

Another avenue for theory building that is becoming increasingly important to research conducted from a media studies perspective is a comparative angle. All research is comparative, in the sense that it is impossible to discuss change or absolutes without reference to something else. Comparisons over time are important, and assumptions about change over time are often looming in the background, if not already at the forefront, of statements about the media and society. More important for theory building are comparisons across groups. In highlighting some of the more recent research in public opinion and political action, this chapter has also mentioned a variety of different groups that have a special meaning for research from a media studies perspective. Age cohorts or generations are one example of such groups. As discussed above, Generation Xers, baby boomers, and the civic generation vary in their forms of civic engagement, and use of different media accentuated these differences. Apart from the issue of the digital divide, which exists between groups within countries as well as among countries, the Internet itself, even in a highly wired community, is found to be more meaningful for the civic engagement of young people than it is for older citizens. In societies in transition, cohort analysis might be important for understanding how journalists perceive and perform their jobs as well as for how citizens use the media and how they relate to different political actors and media institutions.

Another area that deserves closer attention concerns how to measure or describe those who are politically involved in comparison with those who are not, as this may have important consequences for understanding media effects. Political knowledge indicators have been commonly used to group individuals into “high” and “low” groups for purposes of distinguishing differences in media effects, but this needs to be seen within a larger perspective on what constitutes political knowledge and how knowledge may assist in the processing of political information. This is also part of a larger discussion on information processing, a topic that has generated considerable interest (Graber, 2001).

And given that party systems, media systems, and electoral systems are organized primarily at the national level, nation-states are also an important focus for cross-national comparison. These sorts of comparisons may be common to comparative political scientists (see, e.g., Gunther & Mughan, 2000), as well as those involved in comparative historical sociology and the study of cultural change, but the comparative perspective on media studies is (relatively speaking) rather underdeveloped (Blumler, McLeod, & Rosengren, 1992). A notable early study in the area of media theory shed light on the importance of considering post-Soviet and Eastern European examples for internationalizing media theory (Downing, 1996). Another study in political communications focused on innovations in party political campaigning in democracies and societies in transition (Swanson & Mancini, 1996). And another emphasized the importance of going beyond the typical “Western” case studies for building media theory (Curran & Park, 2000).