What We Know About the Media’s Agenda Setting Power

Why Political Actors Adopt Media Issues

While recent work has found that media coverage does affect the political agenda, it has been much less vocal as to why political elites respond to media cues. In other words, a firm theoretical foundation for the found effect of setting the political agenda is still lacking (but see an attempt by Jones & Wolfe, 2010; Walgrave & Van Aelst, 2006). The situation reminds us of the early days of the public agenda setting work. When McCombs and Shaw found in their 1972 study that citizens’ issue priorities were strongly associated with media coverage of issues, a theory of why citizens adopt media cues was lacking as well; early public agenda setting work was undertheorized. Only much later, the underlying mechanism of agenda setting—the storage of information into memory and the accessibility of that information (Scheufele & Tewksbury, 2007)—was specified.

To further develop the field of political agenda setting, more efforts need to go to specifying the underlying theoretical mechanisms. In contrast to public agenda setting, where the mechanism is an accessibility effect—when people are asked what they care about they look into their memory for the things they have experienced (in the media) recently and/or frequently—political agenda setting might be more a matter of applicability. Depending on their position in a political system, political actors have a specific task. Government MPs, for example, have different tasks and pursue different goals than opposition MPs. Political actors respond to media coverage when the information encapsulated in the media signal matches their task, when the media cue is applicable to their goals. The essential difference with public agenda setting lays in the fact that the political agenda setting effect is a behavioral effect, it deals with what political actors are doing, while the public agenda setting effect is a cognitive effect, it deals with what ordinary citizens are thinking. That is why political agenda setting is more a matter of applicability (to the behavioral task) than of accessibility (from memory). In a new study, Sevenans, Walgrave, and Epping (2016) disentangle the different types of attention individual political elites pay to media stories. They find politicians’ cognitive recall, just like that of citizens, to be driven by the accessibility of the media stories; their informal behavior and their intention to act on these stories, in contrast, is affected by applicability. This finding directly supports the applicability notion.

Media coverage is used in politics because it is suitable for political elites. It is relevant to furthering their goals. The information encapsulated in media coverage forms a useful political resource. First, the media form a formidable source of factual information about societal problems and their potential solutions. Politicians simply learn about society via the news. They not only learn from the media about problems and solutions but also about public opinion (Herbst, 1998). They even learn from the media about what other political actors are doing or up to (Davis, 2007).

Additionally, how the media bring this information often fits political actors’ goals. Media coverage is fast and succinct, it is negative (Lengauer, Esser, & Berganza, 2012), often framed in a conflictual way (Neuman, Just, & Crigler, 1992), and frequently attributes responsibility to specific actors for things that go bad (Semetko & Valkenburg, 2000). All these features make news an attractive resource for politicians to nurture their political activities and to react on media coverage openly or more covertly. Especially for the opposition, for example, the media produce a constant stream of bad news stories that can be employed in their fight with the government. No wonder some studies found the media to be a stronger agenda setter of the opposition’s than of the government’s agenda (for example, Walgrave et al., 2008).

Third, the media not only deliver information that may be useful, they also form a platform that political actors need to connect with their voters. Politicians vie for media coverage. One way to get into the news is by reacting on it. Since media attention is cyclic (Downs, 1972), adopting media cues increases the chances that one will get media coverage (see also van Santen, Helfer, & Van Aelst, 2015; Wolfsfeld & Sheafer, 2006). That is why most studies found the political reaction to media coverage to be immediate—if politicians wait to react to media coverage, they risk that the momentum is already gone (for example, Walgrave et al., 2008).

One way to further specify why and how the media affects the political agenda and to deepen the proposed applicability logic, is to look more meticulously into the exact content of the media coverage. While political agenda setting essentially is a transfer of mere issue salience from the media to politics, examining how the issue is covered in the news, is a promising way forward. Technically speaking, the question is whether how the news is framed moderates the political agenda setting effect. In other words, is some coverage more applicable to political actors’ tasks than other coverage? The more general policy literature, which does not deal so much with media coverage, but with how issues are framed and portrayed, suggests that framing should make a difference. Both Schattschneider (1960) and Kingdon (1984) emphasized the importance of the definition of an issue to let it move up the agenda, and they considered politics to be about a struggle to let one’s specification of the issue prevail. Recent policy work showed, for example, that framing the issue of capital punishment in the United States in terms of eradicating the risk that innocent people would be found guilty and sentenced to death, has had major consequences for its gradual abolishment (Baumgartner, De Boef, & Boydstun, 2008; for another example, see Rose & Baumgartner, 2013). Earlier work in the United States found that negative news, compared to positive news, leads to different, and more, policy activity (Baumgartner & Jones, 1993).

A small but growing portion of recent studies has begun to include media framing and tone in political agenda setting studies. These studies offer strong support for the applicability mechanism as a driver of the media effect. The work by Thesen (2011, 2013) on Denmark is exemplary in this respect. Based on a detailed study of parliamentary questions, he shows that news with a negative tone and news that attributes responsibility for the bad situation to the government, leads to more questioning behavior of opposition than of government MPs. Such news is, of course, much more useful for the opposition than for the government; in fact, it is downright damaging for the government. Interestingly, Thesen finds that government MPs use the news as well: when it is positive, they draw on it to defend their policy record. In a recent study covering parliamentary questions in Sweden and Spain, Van der Pas (2014) finds that parties’ reaction to media coverage about integration and immigration in the European Union is conditioned by how the news is framed. Parties react only when the frame the media uses matches their own understanding of the issue. She concludes:

When media reporting provides a context in which their frame prevails, their policy solutions appear more plausible, so it makes sense to strike iron when it’s hot and discuss the issue in parliament at that moment. In contrast, if parties broach a topic while the framing of it in the media is in discord with their platform and framing, they will have a hard time finding support for their policies (Van der Pas, 2014, p. 46).

A third study, investigating the moderating role of tone and framing is a recent work by Sevenans and Vliegenthart (2015). Drawing on data from Belgium and the Netherlands, they find that MPs in these countries ask more questions in response to the news when it has been framed in a conflictual way. Finally, another new study by Sevenans, Walgrave, and Epping (2016) showed that political news, news that is told from a political perspective and featuring political actors—one may call this a “political” frame—gains more attention from political actors than non-political news. In sum, a handful of recent studies indicate that framing and tone of the media may matter as they increase the political usefulness of the news. This is a promising avenue, but more work is needed. The found patterns require replication in more countries and, especially, a broadening of the frames beyond attribution of responsibility and conflictual framing seems to be in order.

When thinking about the theoretical underpinnings of the effect of news coverage on the political agenda, another question is to what extent political adoption is proportional to the media’s attention. In fact, the policy agenda literature heavily supports the idea of the disproportionate reactions of institutions to incoming, societal signals (Jones & Baumgartner, 2005). Due to cognitive and institutional friction, these scholars show, policy outputs are not in proportion to the societal inputs. Time and again, students of policy agendas, have found that all policy outputs—including questions in parliament, speeches, or bills—are non‑Normally distributed but are spiked, abrupt, and stochastic, while their societal input is, by definition, Normally distributed (Baumgartner et al., 2009; Jones, Sulkin, & Larsen, 2003). These findings strongly suggest that the political adoption of media cues—the media are just delivering one type of societal signals—may be disproportional as well. Yet, there is no work that examines this potentially non-linear reaction of politics on the media. All recent studies in Table 1 and in the earlier studies, simply modeled the media effect as being a linear effect, supposing that a one-unit increase in media attention has a similar effect on the political agenda on a high level as on a low level of media attention. But this is very unlikely. So, a second theoretical advance would be not only to dig deeper into the applicability of the media but also to conceive of the media effect as a potentially non-linear effect (see also Wolfe, Jones, & Baumgartner, 2013).

Disaggregating Political Agenda Setting Effects

Students of the media and political agendas are beginning to get a better grip on why the political agenda follows media cues. The phenomenon of the media agenda affecting the political agenda in itself is by now well documented and appears to be widespread. But our knowledge of the exact mechanisms of media power is shallow. Instead of looking further for the mere existence of media effects, the field should aim for making more theoretical advances. The concept of applicability appears to be a necessary element in any theory having the ambition to explain the media’s political agenda effects. The present studies dealing with the precise content of the news, its framing, and its tone, as well as studies finding differences between government and opposition parties or dealing with owned vs. non-owned issues, point the way.

We argue that the field can probably advance most by disaggregating its research designs. Currently, agenda setting is empirically investigated as a macro-phenomenon, general media coverage affects the output of political institutions. But underlying the macro behavior of institutions (e.g., parliaments), or the meso-behavior of organizations (e.g., parties), is the micro-behavior of individuals. The macro-level policy agenda work, invariably finding irregular and spiked attention to issues, builds on the premise that the limited cognitive architecture of individual human beings occupying positions in political institutions is driving the well-established macro-patterns. In fact, the whole bounded rationality approach to politics is based on the assumption that, ultimately, individuals attend to and process information (Simon, 1985). If individuals deal with information and let their attention be determined by the media agenda, then we should start looking at individual elites to further our knowledge about political agenda setting. In fact, it is the “half-way” disaggregation to parties that has taught us the most so far and pointed the way, for example, to the role of opposition or of issue ownership. Further disaggregating the evidence and designs and addressing individual elites is the natural next step.

The concept of applicability—information is used when it matches certain tasks or goals—relates foremost to individuals who have varying tasks and who set themselves different goals. Of course, institutions or organizations within those institutions have tasks and goals as well, but there is much more variation at the individual level. Note that our plea for disaggregated research designs targeting individual elites maps nicely onto the past evolution of the public agenda setting field. There as well, the research domain started with a number of compelling aggregate studies, but very soon, due a desire to develop better theory and to better understand why citizens’ issue priorities are affected by media coverage, the individual level studies pervaded the field and came to be the dominant methodological approach (the first individual-level public agenda study was probably Weaver, 1977).

Moving to the individual level not only allows us to formulate more precise theory but also to remedy some of the methodological problems aggregate-level studies are wrestling with. The bulk of time-series studies listed in Table 1, as well as the older studies, are mostly suggestive of media effects, but they cannot establish causality with absolute certainty. First, we saw an increase of studies that deal with multiple or even all issues. These broader studies give a more complete picture of the interaction between media and politics, but they do not include a measurement of the real world in their designs. This is in contrast with studies that focus on only one issue, such as the economy (and control for economic conditions or unemployment rates; see for example, Eshbaugh-Soha & Peake, 2005; Olds, 2013; Vliegenthart & Montes, 2014) or the environment (and control for emission statistics or food prices; see Delshad, 2012; Jenner, 2012). However, when these all-issue studies find that the media set the political agenda, this may be due to the fact that the media simply react faster on real world developments than political actors, making it appear that those who react to the media are just reacting to the real world. Added to that, media coverage itself is driven by leading political actors (see for example indexing theory by Bennett, 1990), which raises endogeneity issues, especially if political actors steer media coverage by concealed leaking of information or by appearing as anonymous sources in the news. A third problem the prevailing time-series method is wrestling with is that the issues are mostly aggregated at a very high level; for example, all economic issues are subsumed under one economic super issue (but see Eshbaugh-Soha & Peake, 2005; Olds, 2013). This makes it hard to examine whether, for example, news coverage regarding the mortgage crisis actually led to more parliamentary questioning on the mortgage crisis or rather to questions regarding another, maybe unconnected, economic issue. Of course, the fact that many time-series studies covering different issues in different countries uncovered remarkably similar patterns strongly increases confidence in the fact that we are dealing with a truly causal relationship, in which preceding media coverage leads to subsequent political attention. Still, causal inference and precise causal identification, especially for large-scale all-issue analyses, remains a challenging problem for time-series studies.

Studies at the individual level are less inflicted by these possible methodological hitches. Cause and consequence are more closely connected and more easily observable. Some studies already exist. For example, a recent study by Sevenans, Walgrave, and Vos (2015) questioned Belgian MPs about their reactivity to media stories. It found that individual MPs with particularly partisan goals—MPs who engage in the partisan battle and try to damage other parties—are more responsive to media cues than MPs who rather aim to make policy. The study underscored, again, the importance of the applicability of the media information. For the so-called “party warriors” among the MPs, media cues are more fit for their war-mongering task than for the policy-makers.

Ultimately, the field is, and should be, moving towards implementing full experimental designs that directly assess agenda reactivity by political elites to media coverage. Quickly gaining popularity in communications and political science alike (Iyengar, 2001), the experimental method is the best way to disentangle the causality and intricate mechanisms of agenda setting. The main problem of this approach is gaining access to political elites, which challenges its feasibility. Although much recent work has managed to survey elites in many countries (e.g., Deschouwer & Depauw, 2014), it appears to be the case that the access problem is even more outspoken in larger countries in which elites are likely to experience more time constraints and to be more overburdened. Further development of the field, drawing on the individualistic and experimental methods, may therefore remain restrained to smaller country studies. Still, we plead for survey-embedded experiments, or vignette experiments, in which elites are exposed to differently framed media stories, for example, or to information that comes from media and information that does not come from the media The main problem is that experiments with elites cannot study elites’ actual behavior, but can only examine what they tell us about their potential or intentional behavior following up on a media story they were asked to read or watch. In a sense, we remain stuck with elites cognitive reaction. The question remains to what extent elites’ reactions to media stimuli recorded in a survey are predicting their real world reactions, which is the classic external validity problem. Therefore, an alternative way forward is to focus more in-depth on the interaction between media coverage and policy makers in specific case studies.

In recent years, a handful of scholars analyzed the role of the media in concrete cases of policymaking, but such studies have remained rare so far. For instance, some studies focused on healthcare related topics, such as the work of Tieberghien (2014) on drug policy in Belgium, while Katikireddi and Hilton (2015) addressed the role of the media in the debate on alcohol minimum prizing in Scotland. In the Netherlands, Melenhorst (2015) investigated the legislative process that resulted in the legal regulation of remuneration for (semi)public officials, and Ardic, Annema, and van Wee (2015) studied two cases of road pricing policy. The case-study approach allows investigators to detangle when and why politicians react to certain types of news coverage and take into account both the amount of attention and the framing of the topic. The findings of these case studies on the policymaking role of the media show striking similarities. First, they all conclude that media coverage had an influence on the political debate and the actions of policy makers. Media attention is seen mostly as a way to keep a topic high on the public and political agenda, not just in the beginning but throughout the policy process. Second, the cited studies stress the strategic nature of these reactions: politicians never react automatically to the media but rather are selective in support of their interpretation of the issue at stake. Of course, case studies score lower in term of generalizability, but their supplementary insights on when, how, and why individual political actors use the media is certainly useful.

Wrapping up, a promising way forward is to study individual elites and the adoption of media issues on their personal agendas in greater detail, ideally in an experimental or in- depth way. This will lead to examination of the mechanisms underlying the macro-findings, to discover new mechanisms that cannot be studied at the aggregate level, and to directly address some of the methodological issues the aggregate level studies are struggling with. Of course, in the end, the goal must be to integrate the micro- and macro-level studies into one overarching theory of agenda setting by the media. As we argued here, the notion of the applicability of the information encapsulated in media messages is the most promising way to integrate existing and future work on the aggregate and the disaggregate level.

Systemic Variation in the Media’s Agenda Power

Above we mentioned the quasi absence of comparative work examining whether the media matter more, or differently, for setting the political agenda in some countries compared to others (see also Van Dalen & Van Aelst, 2014). We believe it to be highly likely that the media do not play identical roles across countries. The variations in media systems and political systems should affect how elites process and react to the news. Since the availability of comparative data is increasing, due to the existence of comparative datasets such as those of the Comparative Agendas Project and due to the spread of automated coding, we expect comparative studies to gain momentum. We can formulate several expectations that future comparative work could test.

Regarding media system differences, in systems with strong locally segmented media markets, for example, the media cater to the interests of their local audience. They mostly cover local stories, and these are less relevant for national MPs. In national media markets, in contrast, we expect more media responsiveness as the national media outlets bring news that may be relevant to nurture elites’ parliamentary and other activities. Another media factor may be the political parallelism of the media. In some countries, the media are more openly partisan than in others. They are associated with parties and/or they support specific parties (Groeling, 2013; Hallin & Mancini, 2004). This makes them cover different issues. In those countries, parties may be more reactive to befriended media than to media outlets affiliated with competing parties. This basically is what Vliegenthart and Montes found in their study of the political reactions to economic coverage for Spain (2014). In the Netherlands, a country with a much less partisan press, no such effects where found.

A third media system feature that may be of importance is the fragmentation of the media market. If many different outlets co-exist in a country, and if these outlets systematically cover different issues (for example, the latter is the case in Spain, according to Bonafont & Baumgartner, 2013), this should attenuate the impact of media coverage on politics. Eilders (1997) suggested, some time ago, that the consonance of the media matters for their political agenda effect, but no study has directly tested this—although a recent study by Sevenans, Walgrave, and Epping (2016) shows that stories that are covered in many news outlets at the same time have a higher chance of being the object of intended action by MPs than stories that receive much less consonant coverage.

In terms of political system features, one could think of many characteristics could make a difference. The most important of all possible political system differences is the electoral system, proportional versus majoritarian. It affects the amount of elected parties in a country and whether the government typically is a coalition government or not (Lijphart, 1999). In a proportional and fragmented party system, party competition and individual competition are tougher. More politicians have unsafe, or insecure, seats, and this may increase their responsiveness to the media. Inversely, earlier work has found politics to be more reactive, not less, to public opinion in two-party systems compared to multi-party systems (Soroka & Wlezien, 2010), suggesting just the opposite: more media reactivity in two-party systems. Either way, examination of these differences should be high on the research agendas of scholars. The power ratio between government and parliament may be another variable worth pursuing. When parliament is strongly dominated by the government and, in particular, when the information asymmetry between both central democratic institutions is large, MPs may rely more on mass media to get their information and to challenge or monitor the government.

It may be the case that media and political system features must be taken into account together to make sense of the media’s agenda influence. This is what Van Dalen and Van Aelst (2014) argue when they talk about the “power balance” between politics and mass media. Media systems with autonomous outlets that have their own independent resources exert more influence, they say. Political systems, with low cohesiveness and fragmented power, undergo more media influence. Based on political journalists’ perceptions of the media’s power to set political agendas, they test their approach and find that the media’s power is small in countries with weak media (small reach, low autonomy) and with political power concentrated in the hands of a limited number of political actors (few parties, one-party government). In countries with strong media, and power shared among multiple players, the opposite perception exists among journalists. Spain is a typical example of the former case, whereas Norway and Sweden are examples of the latter case. Van Dalen and Van Aelst’s work is the most elaborate effort so far to examine media power comparatively. Its main weakness is that it draws on the perceptions of journalists and not on a direct empirical analysis of the behavior of politicians. Wrapping up, we can conclude that more comparative work is needed.

Conclusion

The media are one source of influence on the political agenda. During the last decade, gradually more studies in more countries have empirically tackled the media’s power to set political agendas. Within the broader research on media and politics, media and political agenda setting has become one of the most active and vibrant subdomains. Most of this work confirms the existence of media effects on the political agenda, in particular on the parliamentary agenda. The media’s agenda power is contingent, though, on the issue at stake, on parties being in government or in the opposition, and on the ownership of parties over issues.

To make further progress in a subdomain that has gained a certain maturity lately, we put forward three avenues for further research. First, the media’s agenda impact must be better theorized. The question of why political actors adopt media issues is not well understood. We argue that the applicability of the media signal to the work and goals of the political actors should take center stage in such a theory. Political actors do not react mechanically to media coverage. They use it selectively to further their goals and strengthen their position. Second, moving forward can be accomplished by disaggregating the present research designs. Aggregate time series designs have been, and continue to be, valuable instruments to establish the presence of agenda effects. Yet, they are not well suited to refine the theory or to understand the precise mechanisms linking media coverage with political adoption. We argue that future research should look at individual elites, ideally using experimental designs or in-depth case studies. Third, as the field is almost completely void of comparative studies, we do not have a lot to say about whether the media matter more for the political agenda in some systems than in others. We call for more studies that theorize country differences, focusing both on media system and political system characteristics.

As data about media and political agendas become more readily available, we expect that the work on the effect of media coverage on the political agenda will expand further. We believe future work should invest in deepening and broadening our knowledge about the media’s agenda power. We hope this article lays out a useful research program for the years to come.