The political agenda-setting dynamics of television news : a cross-national comparison of the political agenda-setting power of Commercial and Public Service broadcasting news

Master Thesis – MSc Political Science (Political Economy track)

Bram Meesters Student nr. 12299243

The Political Agenda-Setting Dynamics of Television News:

A cross-national comparison of the political agenda-setting power of Commercial and Public Service broadcasting news.

June 2019 19.305 words

Supervisor: dr. J. Berkhout Second Reader: dr. D. van der Pas

1

1. Table of contents

2

2. Introduction

‘You talk about undermining the public discourse. It seems to me that simply going through and finding lone things that sound bad out of context, and then (…) hitting people with them, is a way for you to make a quick buck on BBC off the fact that I’m popular and no-one has ever heard of you.’

Ben Shapiro (Conservative commentator form the United States)

‘Uh, there’re not many bucks to be made on the BBC, unlike American broadcasting, Mr. Shapiro.’

Andrew Neil (Journalist for the BBC)

This is an excerpt of a heated conversation on the BBC’s Politics Live between U.S. conservative commentator Ben Shapiro and the BBC’s Andrew Neil (Politics Live, 2019). This interview got was eventually terminated by Shapiro, after which it became viral. The fragment above introduces two central facets of this research.

Firstly, Neil claims that there is not much money to be made on the BBC. This relates to the fact that the BBC is a public service broadcaster, whose income largely consists of a licensee fee (Benson & Powers, 2011, p. 55). The height of this fee is determined by the British Department of Culture, Media and Sport over a period of multiple years. The fact that the BBCs audience share does not influence its funding is a deliberate decision to ensure that journalism is devoid of commercial influence (Avery, 2004, p. 1851). This stands in sharp contrast with commercial broadcasters, whose income mainly consists of advertisement incomes.

Second, the accusation that American broadcasting is more commercialised than British broadcasting introduces the fact that that differences in journalistic culture (Van Dalen, de Vreese & Albæk, 2012; Albæk et al., 2014) and funding arrangements (Benson & Powers, 2011; Aalberg et al., 2010; Curran et al., 2009) persist across countries. In fact, the statement by Neil that American broadcasting is more market-driven is correct (Aalberg & Cushion, 2016, p. 9). For example, U.S. per capita spending on public journalism is less than four dollars, compared to ninety dollars in the United Kingdom (Benson & Powers, 2011, p. 61).

3 Neither the United Stated not the United Kingdom are included in this analysis. Yet, this exchange highlights the diverging effects that broadcasting arrangements and country-level factors have on journalistic conduct.

The subject of journalistic conduct is relevant because the media has a normative role in society. The press has two functions for democratic theory (Blumler & Kavanagh, 1999, p. 226). On the one hand, representative democracy requires that citizens are well-informed, in order for them to make informed voting decisions (Brants & van Praag, 2017, p. 402). Hence, the press should inform its audience, and provide a platform for debate, while reporting from a position of autonomy, neutrality, and objectivity (Brants & van Praag, 2006, p. 29). The press should also highlight the views of different political parties and inform about the policy issues which reflect these parties (Brants & van Praag, 2017, p. 402).

On the other hand, the media is often perceived in democratic theory as a “fourth branch” of government (Cater, 1959). In that sense, the press needs to be a platform for the public good and should report beyond the concerns of national political actors to reflect the needs and opinions of the public (Sparrow, 1999, p. xv; Kleinnijenhuis, 2003, pp. 163-164). Then, the mass media provides a check on political institutions and serves as a political institution in its own regard (Cook, 2006). In other words, the press is sometimes expected to mirror what is going on in politics (Bennett, 1990), and sometimes to lead politics (Bartels, 1996; Baumgartner, Leech & Jones, 1997, p. 350).

The press is not always successful in fulfilling the last function. That is not to say that the press exactly mirrors political actors (Cook, 2006, p. 162), as journalists retain the final control over what they consider newsworthy. However, original news reports often fail to get political attention, because political actors cannot allocate an infinite amount of attention to everything at once, and have to make decisions which issues they deem the most relevant (Baumgartner et al., 1997, p. 350). But sometimes news reports can have important political consequences. The Watergate scandal in the 1970s has shown that original journalistic reports can dominate the political agenda for months, and can have important political ramifications (Lang & Lang, 1981). But journalistic output can also have a more subtle influence on the political agenda. The conditions under which media items become subject of political attention has been studied by a growing body of political agenda-setting research (Walgrave & van Aelst, 2006; Walgrave & van Aelst, 2016). The question many of these studies are concerned with, and the research question of this study, is as follows:

4 This study employs a cross-national causal analysis of the media’s influence on

parliamentary questions in the Netherlands, France, and Canada. The first part of this thesis summarizes the literature on political agenda-setting. It is shown that the fields of political science and communication science both studied the influence of agendas on one another, and that a novel stream of political communication research combines these two approaches into one. The main hypothesis of this thesis is that news items by public service broadcasters have a greater chance of inducing parliamentary questions than news items by commercial

broadcasters. Moreover, the country’s media system and the obtrusiveness of issues are hypothesized to affect the agenda-setting power of news items.

This theory is tested through a causal analysis of individual media items across the three countries. These media items are coded based on aggregate level characteristics. After this, it is tested whether these media items induced parliamentary questions. In conclusion, this study finds no statistically significant support for the hypothesized relationships, although it still offers valuable contributions and insights for further research.

5

3. Literature review:

To find out why some media items get political attention and others do not, the literature review will firstly explain how political science research has dealt with the more general question of why issues make it on the political agenda. This question has been a central subject of empirical and theoretical scrutiny since it came on the agenda of political science in the late 1950s.

Firstly, this study will provide some conceptual clearance of the terminology used, particularly with respect to the usage of the terms ‘political agenda’ and ‘policy agenda’. After this, the origins and development of political agenda-setting research will be outlined. Seeing as this study concerns the influence of media on the political agenda, the sections that follow will outline the development of media agenda-setting research. It will be shown that political scientists borrowed aspects of communication scientist’s research regarding media effects on the public agenda, and applied it to their study of the political agenda.

In the final part of this section, the differences between the two concepts of public service broadcasting and commercial broadcasting will be discussed. Communication science studies have repeatedly shown that journalists working within these two broadcasting produce a distinct output of news, and differ with regard to their role conception. Hence, it is argued that distinguishing between these two broadcasting structures is relevant in media agenda-setting research. This is a novel explanation which no previous research has touched upon, which is a research gap that this study aims to fill. Finally, the need for further research on the agenda-setting function of these two media types will be noted, before moving to the

theoretical framework of this study.

Conceptualising agendas and agenda-setting

The related concepts of ‘agenda(s)’ and ‘agenda-setting’ mean different things to different scholars. For many political scientists, the term ‘political agenda’ refers to the list of items which policymakers have formally accepted for active and serious consideration (Cobb & Elder, 1971, p. 906; Cobb, Ross & Ross, 1976, p. 126). In this context, agenda-setting – sometimes termed as agenda-building (Rogers & Dearing, 1988, p. 556; i.e. Cobb & Elder, 1971) – refers to the process ‘through which problems or issues come to command the active and serious attention of government as prospective matters of public policy’. According to these definitions, the political agenda is an explicit part of the policymaking process (Van

6 Dalen & van Aelst, 2014, p. 44), and concerns attention to subjects by government officials and policymakers.

However, more recent scholarship takes on a broader definition of the political agenda. For these scholars, the political agenda consists of the list of issues to which political actors devote their attention (Green-Pedersen & Walgrave, 2014, p. 1). Researchers who adopt the latter definition of the political agenda include political actors beyond government officials (i.e. Kingdon, 1984, p. 3), and recognise that politics is not always explicitly connected to policymaking. Political agenda-setting is, then, the process through which issues are selected by political actors that deserve political attention (Green-Pedersen & Walgrave, 2014, p. 6).

Although the two streams of research above are both concerned with the concepts of the political agenda, their definition of this term is different. Hence, this study distinguishes between policy agenda-setting research and political agenda-setting research. Researchers within the former category deal with the policy agenda, and are concerned with the prioritization of issues for the sake of policymaking by government officials. Researchers within the latter category deal with the political agenda, which explains why political actors allocate their attention to certain issues, but not to others (Walgrave & van Aelst, 2016, pp. 158-159). Having clarified this distinction, the next sub-sections will explain how research on the policy agenda and the political agenda developed.

Origins of agenda-setting research

Agendas can only include a limited list of items at any given time. This feature of agendas finds its origin in the study of human nature. These studies, and particularly the work of Simon (1967; 1983), emphasize two constraints on organisational efficiency; a limited amount of time, and the inability of humans to process large amounts of information.

Humans are naturally constrained by the scarcity of time, which is why they are only able to allocate attention to a limited number of topics (Simon, 1967, p. 100). As noted by Lippmann (1922, pp. 31-35; Simon, 1967, p. 101), individuals are only able or willing to spend a limited amount of time to allocate attention to certain subjects. Hence, one of the most important decision that politicians need to make is the decision to allocate time and attention to one topic over others (Simon, 1967, p. 100). Political organisations, being comprised of humans, similarly face the constraint of time (Simon, 1967, pp. 100-101). Although political bodies are comprised of a large number of people, the allocation of attention to different subjects is limited in many instances. Delegating work to

7 (sub)committees or bureaucracies partly offsets the restricted attention of legislators, but legislatures can still only attend to a restricted number of topics in each session.

Moreover, the inability of human organisations to allocate an infinite amount of time to every subject limits their ability to process information (Cobb and Elder, 1971, p. 901). Organisations dealing with policy questions are particularly exposed to a large amount of information from a variety of sources (Simon, 1983; Kingdon, 1984, p. 37; Baumgartner & Jones, 2005, p. 10). By delegating the task of processing information to a large number of people through bureaucratic systems, political bodies are able to serve their daily routine functions (Simon, 1967, p. 100). However, issues with a high level of political saliency often require focused attention by a great number of persons, if not the whole polity. Hence, polities are only able to allocate attention to a restricted number of political issues, because they are restricted by the amount of time they can allocate and by their ability to process information.

In light of the finding that attention is scarce in human organisations, it became

conventional by the 1960s to apply this conception to legislators, executives, and government bureaucrats (i.e. Simon, 1967, p. 102; Walker, 1977, pp. 424-425). The studies of

Schattschneider (1957) and Bachrach and Baratz (1962) were particularly influential in bringing the issue of attention to the forefront of the political science debate, by relating this finding to the concept of power.

Schattschneider, in his inaugural address to the American Political Science Association, defines politics as ‘the management of conflict’ (Schattschneider, 1957, p. 936). Politicians, he claims, are unlikely to seek out conflict when they are unsure about the prospects of winning (Schattschneider, 1957, pp. 936-937). Rather, politicians will attempt to strategically influence which issues are up for debate, in such a way that the issues that are salient reflect conflicts which they can win. Then, power is expressed by actors who determine what politics is about, as they are able to define the choice of conflicts, which in turn allocates more power to these actors. As such, Schattschneider claims, ‘the definition of alternatives is the supreme instrument of power’ (Schattschneider, 1957, p. 937).

This conception of power stands in sharp contrast with pluralist thought, which defines power as the ability of political actors to influence the outcome of key political decisions through participation in the decision-making process (Bachrach & Baratz, 1962, p. 948). Bachrach and Baratz (1962), who built on the works of Schattschneider, criticise this approach as it ignores the ability of actors to limit the scope of political decision-making to issues that are relatively safe for them. While acknowledging the merits of a pluralist view on power, they argue that power has a ‘second face’ (Bachrach & Baratz, 1962, p. 947), more in

8 line with Schattschneider’s definition of power. As they note: ‘to the extent that a person or group (…) creates or reinforces barriers to the public airing of policy conflicts, that person or group has power’ (Bachrach & Baratz, 1962, p. 949). As a consequence, billions of potential conflicts are reduced to a mere few by powerful actors (Schattschneider, 1957, p. 935). The most (dangerously) controversial issues are avoided, which causes the democratic system to be highly stable over time, with a bias towards issues which serve actors who are powerful enough to determine which issues are important (Walker, 1966, p. 291; Schattschneider, 1960, pp. 64-69; Schattschneider, 1957, pp. 935-936; Gamson, 1968, p. 19-21).

By linking the scarcity of politicians’ attention to questions of power, the studies of Schattschneider and Bachrach and Baratz have inspired a great many political scientists to study the political agenda (Green-Pedersen & Walgrave, 2014, p. 1; Walker, 1977, p. 423). In particular, these scholars aim to find out why previously dormant issues are sometimes transformed into salient political controversies (Cobb & Elder, 1971, p. 900). In other words, the general question asked is, in the words of Cobb and Elder (1971, p. 905): ‘[W]hat

determines the agenda for political controversy within a community?’. This question is relevant for political science, because political attention to an issue is often perceived as a precondition for political action – or policy – concerning that issue (Walgrave, Boydstun, Vliegenthart & Hardy, 2017, p. 548).

Policy agenda studies

Although Schattschneider and Bachrach and Baratz emphasized the political nature of agenda-setting, scholarship on political agenda-setting mainly focused on the role that agendas play in the policymaking process, rather than on the broader subject of political attention (Green-Pedersen & Walgrave, 2014, p. 2). Specifically, the study of policy agendas attempts to explain how new policy measures are selected for adoption (Baumgartner & Jones, 1993, p. 238). For instance, Cobb & Elder (1971, p. 900) use an agenda-setting approach to examine how masses shape and determine major policy issues. Kingdon (1984) too focuses on the role of agendas in policymaking (Green-Pedersen & Walgrave, 2014, p. 2). Although he does distinguish between issues which are on the policymaking agenda and on the political agenda (Kingdon, 1984, p. 4), his research predominantly covers the policy agenda (Kingdon, 1984, p. 5). However, as the next subsections show, this approach has theoretical and methodological flaws.

9

The problem of case selection

What characterised many of these policy agenda studies is that they often conduct case studies dealing with political attention to specific policy issues (Rogers & Dearing, 1988, p. 576), characterised by landmark pieces of legislation. For instance, Walker (1977) examines the factors which put safety legislation for automobiles, mining, and workers on the policy agenda. His study revolves around three cases of ‘legislative breakthroughs’ (Walker, 1977, p. 431) in the U.S. Senate, namely the Highway Safety Act of 1966, the Coal Mine Health and Safety Act of 1969, and the Occupational Health and Safety Act of 1970 (Walker, 1977, pp. 432-443).Similarly, Kingdom (1984, pp. 240-242) focuses on the issues of health and transportation, and uses 23 cases of prominent legislative proposals to understand how issues get on the legislative agenda. The study of Elder and Cobb (1984, pp. 115-116)

correspondingly centres around the circumstances which gave rise to U.S. legislature addressing elderly care, such as the enactment of Medicare in 1965 and Supplemental Security Income in 1972.

Policy agenda studies such as the ones mentioned above have two major flaws by limiting their scope to specific policy breakthroughs. Firstly, by focusing on policy these studies ignore a sizeable aspect of politics which is symbolic, rather than substantive, in nature. Attention is symbolic if it is ‘brief, infrequent, devoid of policy content and not strongly supportive […] of actions by others’ (Shull & Ringelstein, 1989, p. 46). Although symbols can be used to influence policymaking, politicians may also use symbolic language to persuade actors outside politics, such as members of the public, to accept their views (Shull & Ringelstein, 1989, p. 45). For example, speeches or parliamentary questions can be used to attract public attention to issues and influence the public agenda through media coverage (Van der Pas, van der Brug & Vliegenthart, 2017, p. 495). A narrow focus on the enactment of certain policies ignores this symbolic function of politics. Hence, the explanations that policy agenda studies offer on the question why certain topics enter the policy agenda cannot be applied to the broader political agenda, which includes all agendas which are political in nature, including symbolic political actions (Walgrave & van Aelst, 2016, pp. 158-165).

Second, policy agenda studies suffer from selection bias. The unit of analysis for these studies are often successful instances of legislation being passed regarding specific topics. The outcome variable of these analyses are instances of exceptionally successful

implementation of certain policies. This is a theoretical limitation, because such an approach does not include instances in which parties failed to put an issue on the political agenda. Thus, this approach ignores the finding by Bachrach and Baratz (1961, p. 949) that power has a

10 second face, because it does not include instances when policy conflicts are barred from being raised.

Moreover, this selection bias creates methodological problems. By basing their case selection only on instances of policy success, policy agenda studies have no variance in the dependent variable (Collier & Mahoney, 1996, p. 72). This is problematic because the factors that give rise to policy success might also be present in cases of policy failure. As such, these studies are potentially subject to systematic error (Collier & Mahoney, 1996, pp. 56-57). For researchers within the quantitative tradition, this fact is often enough to discard the outcome of such studies altogether (i.e. King, Keohane & Verba, 1993, p. 130).

Many qualitative researchers have a more nuanced view (i.e. Rohlfing, 2012, p. 6; Collier & Mahoney, 1996, pp. 73-74) and hence argue that such research can have merits, for example by discovering novel explanations. In this way, policy agenda studies are valuable because they highlight the complex nature of the agenda-setting process (Walker, 1977, p. 430; Elder & Cobb, 1984, pp. 117-118; Kingdon, 1984, p. 2). However, these studies cannot be generalized to the full spectrum of cases (Collier & Mahoney, 1996, p. 71; Baumgartner & Jones, 1993, pp. 47-48). This defect is explicitly recognised by Kingdon (1984, p. 242). Some other studies are less nuanced in this regard. For example, Walker (1977, p. 430) aims to offer ‘a general explanation of the conditions that usually allow issues to claim a place on the [political] agenda’, and Elder and Cobb (1984, p. 118) test contextual factors which give rise to policy success. Such research aims cannot be met when the dependent variable is restricted to policy successes. All in all, by limiting their case selection to instances of policy success, early policy agenda scholars could not provide a concluding answer to the question why certain issues get on the political agenda and others do not.

The problem of endogeneity

Early studies of policy agendas suffer from another deficit, besides their focus on a limited number of cases. These studies also have a tendency to underemphasize the impact of public or media agendas on the policy agenda (Green-Pedersen & Walgrave, 2014, pp. 3-7). Political participants without formal government positions, such as interest groups,

academics, journalists, and the general public (Kingdon, 1984, pp. 45-70) conceivably have an influence on politicians. While the impact of interest groups and experts on political agendas has been studied by a score of political scientists by the 1980s (Baumgartner, 2010, pp. 521-522), they found that the media and the general public had a limited amount of direct influence on the political agenda.

11 In fact, some scholars play down the importance of outside influences altogether. The study by Walker (1977) again, is exemplary. Walker (1977, p. 431) argues that U.S. Senators introduce new issues to the political agenda, in the hope that this will lead to an increase in political support through a powerful winning coalition. If successful, such coalitions can commit themselves to other logically related issues, and can thereby emit political dominance for a long period of time (Walker, 1977, pp. 444-445). According to this view, agenda-setting is the result of an endogenous process of coalition-building. Outside actors, such as journalists and lobbyists, are only important in the agenda-setting process insofar as they react to the issues raised by Senators (Walker, 1977, p. 431).

Kingdon (1984, pp. 61-62) similarly dismisses the impact of media on government agendas. By and large, media merely report on what is going on in government. This reflects the fact that, up until the 1990s, agenda-setting scholars tended to stress endogenous factors in accounting for changes in the policy agenda (Walgrave & van Aelst, 2006, p. 89; Green-Pedersen & Walgrave, 2014, pp. 3-4). Agenda change was predominantly explained as being a result of actions by individual political entrepreneurs (Baumgartner & Jones, 1993, p. 48). In fact, some authors (i.e. Grossmann, 2014) still downplay the influence of exogenous factors on political agenda-setting. This is unfortunate because, as the next subsections show, external factors such as the media do have an important agenda-setting function. This finding was recognised by scholars who focus on the broader political agenda rather than the narrower policy agenda.

Political agenda studies: from policy to politics

In the 1990s, political scientists gradually moved away from the narrow focus on policy agendas towards the broader study of political agendas. Since then, the focus of agenda scholars has shifted back from questions regarding policymaking towards questions related to politics as a whole, such as on how political institutions and political actors generate attention of politicians to some issues but not others (Green-Pedersen & Walgrave, 2014, p. 3).

Thereupon, and in the footsteps of Schattschneider and Bachrach and Baratz, political attention became a legitimate subject in its own right for agenda scholars. The seminal study of Baumgartner and Jones (1993) has been pioneering in this shift (Green-Pedersen, 2014, p. 3).

Whereas policy agenda studies focus on policy change, many other political science fields emphasize the incremental nature of politics (Baumgartner & Jones, 1993, p. 238; Grossmann, 2014, pp. 15-16). The punctuated equilibrium theory of Baumgartner and Jones

12 (1993, p. 238) combines both perspectives, by emphasizing stability and change. As they argue, political institutions and stable issue definitions cause stability, whereas a sudden change in the definition of issues can cause equilibria to be broken. Baumgartner and Jones (1993, p. 6) note that groups are sometimes able to establish monopolies of political

understanding, which are supported by political institutions, in order to avoid constant conflict. As Schattschneider notes, this gives these groups considerable power. Therefore, groups different from those in power will actively strive for issue definitions which support their agenda, in which case equilibria of political understanding can be broken.

Baumgartner and Jones (1993, p. 57) predominantly base their findings on empirical evidence from attention to specific policy issues, such as nuclear power, smoking and tobacco regulation, pesticides, automobile safety, urban affairs, and the valence issues of drug-,

alcohol-, and child abuse. However, these issues all have different characteristics, and the focus is not merely on the mechanisms that gave rise to these issues, but also on how some issues receded from the agenda. Moreover, Baumgartner and Jones examine how the external forces of media attention, public opinion, and interest groups influence the policy agenda. In that sense, this study is an improvement from policy agenda studies.

In addition, by emphasizing the impact of institutions and political actors in the agenda-setting process, Baumgartner and Jones have inspired political scientists to move beyond ‘attention to policy [towards] questions related to the politics of attention’ (Green-Pedersen & Walgrave, 2014, p. 3; Baumgartner & Jones, 2005). Thus, by the early 2000s, agenda scholars increasingly asked the questions proposed by Schattschneider and Bachrach and Baratz. The composition of the political agenda increasingly became explained by looking at the

interaction and power relations of actors, by institutions, and by outside forces. The narrow focus on individual policy agendas was replaced with a more comprehensive focus on the political agenda as a whole. Hence, besides budgetary and legislative agendas, these studies included the interaction of the political agendas of parties and parliament, with outside forces such as the mass media’s agenda (Green-Pedersen & Walgrave, 2014, pp. 3-4). Agenda work has moved toward the impact that these agendas have on one another. Increasingly, scholars realized that there is no such thing as ‘the political agenda’ but only a restricted group of loosely associated political agendas (Walgrave & van Aelst, 2006, p. 94). As Green-Pedersen and Walgrave (2014, p. 4) note, the question of agenda-studies increasingly became: ‘how does issue attention “jump” from one agenda to the other, and how do agendas interact?’.

Therefore, one increasingly vibrant type of political agenda studies become the study of the media agenda’s influence on the political agenda. In fact, the media were found to be an

13 important force which could punctuate policy equilibria by Baumgartner and Jones (1993, p. 20). They found that changes of media attention to certain topics often induce changes in the degree of attention by politicians. However, the study of media influences on the political agenda does not just have its origins in the field of political science. As the next sub-section shows, communication scientists laid the groundwork for modern students of the media’s influence on the political agenda.

Study of media influence on political agendas

Communication science

Before media effects were extensively incorporated in studies of political agenda-setting, they were frequently analysed by communication scientists. Rather than focusing on media effects on the political agenda, scholars within the communication science tradition predominantly analysed the effect of the media on the public agenda. Agenda-setting is then concerned with ‘the process through which the mass media communicate the relative

importance of various events or issues to the public’ (Rogers & Dearing, 1988, p. 556). Thus, the concept of ‘agenda-setting’ refers to a similar process whether in the context of political science or communication science (Walgrave & van Aelst, 2016, pp. 158-159). The difference between both approaches is the subject of the media’s influence; while political science research focused on political responses to the media, communication science research focused on the public’s susceptibility to media influence.

The study by McCombs and Shaw (1972) is widely credited as being the first empirical study of the mass media’s influence on the public agenda (Rogers & Dearing, 1988, p. 563; van Aelst et al., 2008, p. 497; Wolfe, Jones & Baumgartner, 2013, p. 178). McCombs and Shaw (1972, p. 184) find that voters often share the media’s definition of what issues are important, which suggests that the media have a strong agenda-setting influence on the public. This is due to the media being the primary source of political information for the national public, and hence their perception of which issues are important mirror the most prominent issues in the mediums they follow (McCombs & Shaw, 1972, pp. 176-177). The article by McCombs and Shaw induced a large numbers of communication scientists to study the relationship between the media and the public agenda, as is evident by the fact that the study is one of the most cited media effects theories (Walgrave & van Aelst, 2016, p. 158; Rogers & Dearing, 1988, p. 586). In fact, the power of the media to set the public agenda has been verified by numerous studies (Wolfe et al., 2013, p. 178).

14

Political Communication Research

Given the magnitude of media agenda studies in communication science, and the empirical evidence for the agenda-setting function of media, it is questionable why little empirical policy agenda studies incorporated media effects. In fact, some scholars (i.e. Wolfe et al., 2013) argue that both fields still operate more-or-less independently. Despite this disciplinary independence, there have been instances where political scientists found that the media had affected the political agenda.

For example, Cohen (1963, p. 37) argues that, in the case of foreign policy, the press commonly promotes certain policies. Sometimes this happens in coordination with

policymakers, whereas other times media are able to push an issue on the agenda independent of policymakers. Similarly, Cobb and Elder (1971, p. 909) note that the media can play an important role in putting issues on the political agenda. Both argue that independent reporters can raise issues by acting as opinion leaders (Cohen, 1963, p. 37; Cobb & Elder, 1971, p. 909). However, these statements are merely supported by anecdotal cases of individual reporters. Empirical studies on the relationship between the media and politics emerged in the late 1980s (Walgrave & van Aelst, 2006, p. 89).

Bennett (1990, pp. 120-122) found a strong correlation between op-ed page opposition in the New York Times to the Reagan administration’s Nicaragua policies and congressional opposition to the subject. For Bennett (1990, pp. 106), these results show that journalists tend to index the voices and viewpoints displayed in their medium to the viewpoints expressed in government debates. However, his empirical data can just as easily be interpreted as evidence for media influence on foreign policy, because causal inferences cannot with certainty be drawn from correlation measures (King, Keohane, Verba, 1993, pp. 75-76).

To overcome this problem of interpretation, many agenda-setting scholars have since employed time-series analyses to deduce the causal relationship between media attention and political actions (van Aelst & Walgrave, 2011, p. 297). For instance, Baumgartner, Jones, and Leech (1997) employ time-series data to estimate whether media attention preceded

congressional hearings on the topics of smoking and tobacco, nuclear power, and drug abuse. Although they find support for an agenda-setting function of the media, their data is crude, and it is difficult to estimate the effect size and precise causal relationship of media attention on congressional attention (Baumgartner et al., 1997, p. 363).

Through more sophisticated statistical tools, such as autoregression, the quality of time-series analyses can be enhanced (Freeman, Williams, Lin, 1989, p. 874). This method ensures a causal relationship because media coverage precedes political action while preceding

15 political action is also controlled for (van Aelst & Walgrave, 2011, p. 297). Bartels (1996, p. 9) recognised the value of this method. He found that media and politics influence each other and that the direction of influence depends on the issue and the medium (Bartels, 1996, pp. 22-23). Moreover, Bartels (1996, p. 25) advocated for further application of advanced statistical methods in the study of the political impact of the mass media.

Modern political communication studies on agenda-setting

A combination of three research traditions

In the early 2000s, an approach emerged which combined elements of the three approaches mentioned above (see Figure 1). The article by Walgrave & van Aelst (2006, p. 89) has been particularly influential in this approach, as these authors explicitly call for an integration of political agenda studies and communication studies on media effects. In fact, subsequent studies of the agenda-setting relationship between media and politics have done just that (Sevenans, 2018, p. 245). Modern agenda-setting scholars combined the advantages of each of the three streams of research into one research agenda.

As a point of departure, scholars within this research agenda recognise that information is scarce, and that the allocation of attention to certain subjects and not to others is an

important political resource (i.e. Vliegenthart et al., 2016a, pp. 283-284). This notion is borrowed from political agenda scholars in the tradition of Schattschneider (1957) and Bachrach and Baratz (1962). Yet, modern media agenda-setting studies have accepted the criticism that many studies within the policy agendas tradition were too narrowly focused on cases of specific policy successes (Walgrave & van Aelst, 2016, p. 158). In fact, before 2005 Walgrave and van Aelst (2006, p. 90) noted that almost all studies which they considered focused on a specific policy issue. Between 2005 and 2015, this was the case for just over half of the studies (Walgrave & van Aelst, 2016, pp. 161-163).

Similarly, previous research by research on the media’s political agenda-setting power laid the groundwork for more sophisticated methodological designs. Simple time-series analyses were supplemented with cross-sectional analyses, causal micro-analyses, in-depth interviews, and surveys (Walgrave & van Aelst, 2016, pp. 160-173; Sevenans, 2018, pp. 245-246). In addition, these studies frequently include cases beyond the United States, or multiple cases. This is an improvement because agenda-setting was previously understudied outside the United States (Walgrave et al., 2008, p. 815). As a matter of fact, all agenda-setting studies before the 2000s in Table 1 only study the United States case, which is problematic because agenda-setting is likely different across countries.

16 Moreover, modern media agenda studies borrowed from public agenda scholars a focus on the conditional character of the agenda-setting power of the mass media. Communication scholars stress that a medium does not automatically influence an individual, but that

cognitive processes affect to what extent an individual is affected by media messages

(Walgrave & van Aelst, 2006, p. 99). What is more, media effects, such as framing, are shown to have an effect on how the public evaluate political issues or politicians (Wolfe et al., 2013, p. 178).

To some extent, politicians may be affected by cognitive processes and media effects in the same way as citizens (Walgrave & van Aelst, 2006, p. 99). However, the behaviour of political actors in the political arena is different from the behaviour of ordinary citizens, as politics is a matter of strategy (Schattschneider, 1957, p. 937). The decision to allocate

attention to one issue over the other is not made through a cognitive process, as is the case for public agenda-setting (Wolfe et al., 2013, p. 184; Walgrave & van Aelst, 2006, p. 99; van Dalen & van Aelst, 2014, p. 44). Rather, political agenda-setting is a behavioural process in which political actors strategically choose which issues should make it on the agenda, and which issues should be excluded.

Figure 3.1: Development of research on the media’s political agenda-setting function (see also Rogers & Dearing, 1988, pp. 561-563).

1920s Lippmann 1922

Early Political Communication scholars

Policy/political agenda scholars

Public agenda scholars

1950s Schattschneider 1957

1960s Cohen 1963 Bachrach & Baratz 1962 Simon 1967 1970s Cobb & Elder 1971 Cobb & Elder 1971

Cobb, Ross, Ross 1976

Walker 1977 McCombs & Shaw 1972

1980s Elder & Cobb 1984

Kingdon 1984 1990s Bennett 1990 Baumgartner & Jones 1993

Bartels 1996 Baumgartner et al. 1997

2000s Modern Political Communication Agenda Scholars

Soroka, 2002b; Walgrave & van Aelst 2006; Walgrave et al. 2008; Thesen, 2013; Vliegenthart et al 2016a; Van der Pas et al. 2017; Sevenans, 2018

17 Despite this discrepancy, scholars of political agenda-setting were inspired by

communication scientists’ emphasis on the conditional nature of agenda-setting (Walgrave & van Aelst, 2006, p. 99). Where public agenda scholars emphasized the effects of framing, cognitive effects (Wolfe et al., 2013, p. 184), and individual characteristics such as political interest (i.e. McCombs & Shaw, 1972, p. 186), political scientists began to take media input and political context into account (Walgrave & van Aelst, 2006, p. 104).

Three contingency factors are often noted. Firstly, the media’s impact depends on the type of political agenda. For example, media effects are often shown to be larger on symbolic or legislative political agendas than on substantive or executive agendas (Walgrave & van Aelst, 2006, p. 95; i.e. Walgrave et al., 2008, p. 831). Secondly, it matters which issues are examined. Like the political agenda, the media’s agenda is limited to a number of issues due to the constraint of time (Baumgartner, Jones & Leech, 1997, p. 349; Albæk et al., 2014, p. 42). Yet, not all issues have an equal effect on the political agenda. Walgrave et al. (2008, pp. 831-832) found that it is less likely for issues such as defence, foreign affairs, and economic policy to impact the expected media influence, which is more likely for other – more

sensational – issues. Thirdly, it is often hypothesized that media effects depend on the type of medium that is studied (Walgrave & van Aelst, 2006, p. 92), particularly the distinction between television and newspapers. Findings on this subject are inconclusive, as some studies show that newspapers have a stronger agenda-setting influence on politics (i.e. Walgrave et al., 2008, p. 832), whereas others find that television has a stronger impact (Van Aelst et al., 2008, p. 499).

Medium types

The contingency effect of the political agenda and issue types have been studied quite extensively. The study of legislative and executive agendas is quite balanced, and many studies take into account a variety of factors to measure the political agenda, and an increasingly wide variety of issue typologies (Walgrave & van Aelst, 2016, pp. 161-163). However, these studies show little variation with regard to which type of medium is employed (Walgrave & van Aelst, 2016, p. 167). Most studies deal with traditional media outlets such as newspapers, television, and radio, with newspaper studies being the most popular. This leaves the role that internet-based media and social media as political agenda-setters understudied, despite the increasing reliance on such media as a source of political information (Aalberg & Cushion, 2016, p. 1).

What is more, very little studies incorporate data from multiple media outlets. This is unfortunate because public agenda studies have shown that different media outlets have

18 divergent effects on the public agenda or public attitudes, particular through comparisons between television news and newspapers (i.e. Carter & Greenberg, 1965; McCombs & Shaw, 1972). Inspired by studies in the latter category, some media agenda studies, such as

Walgrave et al. (2008), form an exception, as they distinguish between data from newspapers and television sources and their effects on political agenda types. Van Aelst and colleagues (2008) also distinguish between multiple media outlets, although this article relies on elite surveys rather than empirical data which tracks issue attention, which makes it difficult to gauge the extent to which the agenda-setting power of newspapers and television differs. Vliegenthart and Walgrave (2011) similarly use data from both newspapers and television media, however, they do not differentiate between the two and rather lump both types together under the same variable. Despite these exceptions, then, there is still a lack of literature within political science which compares the effect of different media types on the political agenda (Walgrave & van Aelst, 2016, p. 167).

Specifically, one distinction that is rarely made is between public service broadcasting (PSB) news and commercial broadcasting news media. This is unfortunate because the next sub-section will show that communication science studies have found a variety of differences between these two outlet types.

Public service broadcasting versus commercial broadcasting

Public service broadcasting (PSB) can be defined as a form of broadcasting which is ‘based on the principles of universality of service, diversity of programming, provision for minority audiences (…), sustaining an informed electorate, and cultural and educational enrichment’ (Avery, 2004, p. 1851). Put simply, PSB aims to provide the public with

programmes they want to watch, and to satisfy broader social purposes such as education and the promotion of citizenship (Armstrong & Weeds, 2007, p. 81). Unlike commercial

broadcasters which depend on advertising, PSB is primarily funded through licence fees, taxes, or other non-commercial funding options (Avery, 2004, p. 1851). This is done, on the one hand, to ensure universal access. On the other hand, this pressures public journalists to actively connect to all sectors of the public, in order to justify public funding (Curran et al., 2009, p. 19; Massey, 2004, p. 1359). Nevertheless, in some countries – such as the

Netherlands, France, Germany, and Canada – licence fees or public funds are supplemented by advertisement revenue (Benson & Powers, 2011, pp. 9-10).

Regardless of the precise funding arrangements, public service journalists have a different outlook on their profession than their commercial rivals. Public service journalists’

19 role conception differs considerably from commercial journalists (Albæk et al, 2014, p. 178). Commercial broadcasters, aiming to make money and attract high-spending audiences (Curran et al., 2009, p. 19), employ a more pragmatic approach, which focuses on political conflict over political substance (Albæk, 2014, pp. 36-49). What is more, journalists within the commercial sphere more often share partisan viewpoints and focus on items with a higher entertainment value. Public service journalists, on the other hand, have a sacerdotal role conception. This means that they regard certain institutions or events as intrinsically

important, and entitled to coverage regardless of the newsworthiness of the item (Blumler & Kavanagh, 1999, p. 226). Moreover, public service journalists place a higher value on their function as providers of political information, and on their impartial role (Albæk et al., 2014, pp. 48-49).

Because public service broadcasters have a different role conception and a different financial arrangement than commercial broadcasters, their output of news differs (Aalberg & Cushion, 2016, pp. 9-12). Hard news, defined as politically relevant stories, based on

impersonal and unemotional reporting (Reinemann et al., 2012, p. 234; Aalberg & Cushion, 2016, p. 10), is more prevalent in public service broadcasting (Aalberg et al., 2010, p. 262). On the other hand, soft news, which has individual relevance rather than political relevance, and is based on personal and emotional reporting, is relatively more prevalent in commercial broadcasting television. This difference is confirmed by multiple empirical studies. Aalberg et al. (2010) show that commercial news has a limited focus on news and current affairs

compared to public service broadcasting in a cross-national comparison of 6 countries. Moreover, the study by Esser et al. (2012) finds that public service broadcasters produce longer and more in-depth political coverage in 13 European countries. Likewise, Brants and van Praag (2006, pp. 35-36) found that the Dutch public broadcasting news focused more on substantive issues during campaign times, whereas its commercial counterpart generally focused more on “horse-race coverage”, such as opinion polls and reflection.

This distinctive news agenda has an effect on the country in which these broadcasters operate. In general, the existence of a strong public service broadcasting system in a country increases the frequency and length of news and current affairs output of the country in

question (Esser et al., 2012, p. 251). This is partially a result of the substantial market share of many public service broadcasters, which makes them the dominant source of news for many citizens (Aalberg & Cushion, 2016, p. 1). This effect is strengthened by the fact that

commercial channels have a tendency to adopt similar informative programmes as established by public channels (Esser et al., 2012, p. 267; Aalberg & Cushion, 2016, p. 11). As a result,

20 citizens’ knowledge about political events in countries with a strong PSB-system is higher than in countries in which commercial broadcasters dominate (Curran et al., 2009, p. 12, Aalberg & Cushion, 2016, p. 17). Thus, role conceptions and news output differ considerably between PSB and commercial broadcasters. The precise effect of these differences depends on the country (van Dalen et al., 2012, p. 912) and on the media system (Albæk et al., 2014, p. 173). Moreover, public service broadcasting has a distinct and positive effect on the political information environment and on citizens’ knowledge.

Yet, these differences are mediated by some countertrends. Firstly, the quantity of commercial broadcasting channels has significantly increased since the 1980s (Curran et al., 2009, p. 6; Aalberg & Cushion, 2016, p. 19). Second, the broadcasting sector was increasingly deregulated (Curran et al., 2009, p. 9), particularly in many Mediterranean countries (Esser et al., 2012, p. 251). Finally, the audience size and influence of public broadcasters is

contracting (Curran et al., 2009, p. 6). Unlike a few decades ago, television news can now be escaped and substituted by a number of other mediums (Albæk et al., 2014, p. 5). Social media and the internet have challenged the dominance of old broadcasting media formats (Aalberg & Cushion, 2016, p. 1). To stay relevant within an increasingly commercialized broadcasting supply, and a decreasing television audience, some scholars report an increasing trend towards commercialisation (Hallin & Mancini, 2004, p. 284; Blumler & Kavanagh, 1995, pp. 217-219). Nevertheless, scholarship continues to show that important differences persist between public service broadcasters and commercial broadcasters (Aalberg & Cushion, 2016, p. 19; Esser et al., 2012, p. 267; Brants & van Praag, 2006, p. 37), and between countries with a strong PSB-system and countries without such a system (Hallin & Mancini, 2017, pp. 162-164; Albæk et al., 2014, p. 173; Esser et al., 2012, p. 267; Aalberg et al., 2010; Brants & van Praag, 2006, p. 39).

Broadcasting mediums and agenda-setting

Where does this leave the role of public service broadcasting media in agenda-setting? The previous sub-section shows that there is a scholarly consensus that commercial and public service media have persistently different role conceptions, news content, and country-wide effects. Moreover, previous research has shown that different media outlets have a different effect on the political agenda. Thus, it is conceivable that differences between the

broadcasting arrangements of mediums have an influence on their agenda-setting power. Seeing as television is the most powerful agenda-setting medium (van Aelst, 2008, p. 502), such a difference has an important effect on the composition of the political agenda.

21 An empirical analysis of the different agenda-setting effects of PSB and commercial broadcasters can provide more clarity on the question of why certain media items get introduced to the political agenda, whereas others do not. However, to the knowledge of the author, there have been no empirical studies which compare the agenda-setting power of

PSB-news items to commercial news items. Typical agenda-setting studies which analyses the

impact of television news media on the political agenda lump television media together (i.e. van Dalen & van Aelst, 2014) or focus on the largest television channels, irrespective of whether these are publicly or commercially funded, and do not conceptually distinguish between the two broadcasting types (i.e. Vliegenthart & Walgrave, 2011; Valenzuela & Arriagada, 2010). Given the methodological maturation of agenda studies and the increasing intercommunication between the fields of communication science and political science, it is puzzling why such research has not been conducted before.

This study aims to fill this gap. As a starting point, this study shares the concern of political agenda scholars, who aim to find out why politicians address certain issues over others. This question is answered by focusing on the role that media play in setting the political agenda. Specifically, the theoretical framework of this study shares a focus with previous communication science work on the distinction between public service media and commercial media. The full theoretically framework of this study also incorporates

hypotheses which are derived from public agenda scholars who focus on the influence of journalistic autonomy and of issue obtrusiveness on the political agenda-setting power of media items. The result is a methodologically advanced study which incorporates elements of public- and political agenda research. As such, this research falls within the scope of modern political communication studies which emphasize the contingency of media effects on the political agenda.

22

4. Theory

In answering the research question, the position is taken that distinguishing between commercial broadcasting news and public service broadcasting news provides a valuable insight into the relationship between media and politics. It is hypothesized that public service news has a greater agenda-setting effect than commercial news. Two mechanisms motivate this hypothesis.

The first mechanism is concerned with the prioritization of issues by journalists. As noted, one of the defining features of public service broadcasting is that it primarily receives its funding from public funds, such as license fees and taxing (Avery, 2004, p. 1851; Aalberg & Cushion, 2016, pp. 4-5). As a result, PSB news and CB news have a different output. On the one hand, PSB can afford to have a more sacerdotal role conception and prioritize hard news with a low entertainment value, because such channels are not guided by commercial pressures. On the other hand, public service journalists prioritize these issues to justify public funding. Hence, as is dealt with in more detail in the literature review, public service news has a greater output of politically relevant news, which intuitively has a higher chance of

influencing the political agenda compared to news items with little political relevance. The second mechanism is concerned with the prioritization of issues by politicians. Even when controlling for the output of media, it is conceivable that public service media have a greater effect on the political agenda due to their principles of universal access and appeal to a broad audience, and their substantial market shares. These factors are relevant because the most important reason for political actors to adopt media issues is the fact that media coverage is associated with public opinion (Walgrave & van Aelst, 2006, p. 100).

As noted, politicians are exposed to a large variety of information, but are only able to process a part of it. Within this information-rich environment, politicians have to prioritize t-to which problems they give attention and which problems can be (temporarily) dropped from consideration (Baumgartner & Jones, 2005, pp. 10-11). To do so, they have to filter their sources of politically relevant information and will favour some sources over others (Baumgartner & Jones, 2005, pp. 57-60). Politicians can be expected – and are indeed empirically shown – to focus on issues which resonate with citizens, as this increases their chance to get re-elected (Baumgartner & Jones, 2005, pp. 249-269). Logically, this means that politicians should favour sources that give an insight into what the public finds important. Media sources are often used for this purpose (Kingdon, 1984, p. 60). It is conceivable that

23 the substantial and broad audience of public service media would increase the chance that politicians adopt issues from public service media over commercial media, as this is potentially more reflective of public opinion.

All in all, there are valid reasons to believe that the political agenda-setting effect of PSB news items is greater than items on commercial news. Therefore, the first hypothesis is as follows:

Hypothesis 1: News by Public Service Media has greater agenda-setting power than news by commercial media.

There are, however, other factors which might affect the media’s agenda-setting power. For one thing, there might be a difference in agenda-setting between countries with distinct media systems. Van Dalen and van Aelst (2014, pp. 58-59) give preliminary evidence for the hypothesis that journalists in more autonomous media systems have greater agenda-setting power.

In their seminal analyses of the relationship between media and politics in Western countries, Hallin and Mancini (2004) made a classification between three types of media systems. According to Hallin and Mancini (2004, pp. 73-75), journalistic autonomy is greatest in democratic corporatist countries. In these countries, journalistic professionalism is high, and media autonomy is guaranteed by the state. Journalistic professionalism is similarly strong in liberal countries, albeit without the kind of formal organisation that exists in

democratic corporatists systems. Therefore, commercial pressures limit journalistic autonomy. The lowest level of political autonomy is found in polarized pluralist countries, where

political pressures restrict the autonomy of journalists. Political parallelism is relatively high in polarized pluralist countries (Hallin & Mancini, 2004). This means that mediums are closely linked to parties through organizational linkages or ideological linkages of their journalists or audience (Seymour-Ure, 1974, p. 173; Van der Pas et al., 2017, p. 493).

This rank order of media systems by their degree of journalistic autonomy is supported by empirical evidence. Albæk et al. (2014, pp. 41-46) show that political pressures greatly constrain the journalistic autonomy of journalists in Spain, a polarized pluralist country. Yet, political pressures are less apparent in the United Kingdom – a liberal country – and

especially less apparent in Denmark and Germany, which are both democratic corporatist countries.

24 As noted, Van Dalen and van Aelst (2014) expected that a greater degree of political autonomy increases the agenda-setting effect of media in a country. In countries characterized by an autonomous media, the press is able to operate independently from outside interference. Because autonomous journalists are better able to take on its function as a watchdog of

politics (Cohen, 1963, p. 35), they should more frequently influence political actors than less autonomous journalists (Van Dalen & van Aelst, 2014, p. 48). In countries with greater political control over the media, the media is expected to follow the political agenda rather than the other way around. The results of van Dalen and van Aelst (2014, pp. 53-54) indicate that this is true for television media, which is able to exert a greater agenda-setting influence in countries with more autonomous media.

In light of these findings, it is expected that media have a greater influence on the political agenda in media systems in which journalists have more autonomy. Hence, political agenda-setting power of the media is greatest in democratic corporatist systems, and least pronounced in polarized pluralist systems. Liberal media systems are expected to fall in between the other systems in terms of the media’s agenda-setting power. This leads to the second hypothesis of this study:

Hypothesis 2: Media agenda-setting power is greater in media systems with a higher level of journalistic autonomy.

Moreover, the media’s agenda-setting power is contingent on the types of issues they bring up (Walgrave & van Aelst, 2006, p. 93-94). Public agenda studies have shown that the media has a lower effect on the public agenda for issues which are more obtrusive (Demers et al., 1989; i.e. Zucker, 1978). In this context, issues are defined as obtrusive if people have a personal experience with it independent from media exposure (Demers et al., 1989, p. 797). Unobtrusive issues, on the other hand, are devoid of personal experience, and it would be hard for individuals to observe these issues without media attention (Walgrave & van Aelst, 2006, p. 93).

This theory, which is concerned with public agenda-setting, can also be applied to political agenda-setting. Walgrave et al. (2008, p. 832) and Soroka (2002b; Walgrave & van Aelst, 2006, p. 93) have shown that media effects focusing on unobtrusive issues have a greater effect on the political agenda compared to obtrusive issues, provided that they have a concrete effect on the population. Like the public, politicians are dependent on the media to bring attention to issues which they cannot directly observe through personal experience

25 (Walgrave et al., 2008, p. 820). Thus, it is anticipated that issues which are unobtrusive – in other words, issues which are not directly observed by political actors – have a greater agenda-setting potential for the media. This leads to the third hypothesis:

Hypothesis 3: News items focused on unobtrusive issues have a greater chance of a political response than news items focused on obtrusive issues.

Finally, it should be noted that media agenda-setting effects may be largely endogenous (Sevenans, 2018, p. 250). As Cook (2006, p. 159) notes, in addition to influencing politics, the media are also a product of politics. Rather than having an independent agenda-setting function, the media often serves as a channel of communication between political actors (Kingdon, 1984, p. 59). Political actors may publicize their plans through the media, before political action on this subject is taken (Sevenans, 2018, p. 250). In these instances, it may seem like the media has influenced politics, while the agenda-setting effect is in fact

endogenous. The media is then situated in the middle of the policymaking process, rather than being an outside force of influence (Cohen, 1963, p. 40; Kingdon, 1984, p. 59).

As a matter of fact, politicians are becoming increasingly preoccupied with seeking publicity (Cook, 2006, p. 162). Through carefully considered PR-campaigns, political actors try to represent issues and events ‘in ways that gain support, shape action, and influence outcomes’ (Bennett, 2016, p. 101). Politicians are often successful in getting their messages across, as media depend on political sources for reliable information on certain issues (Cook, 2006, p. 162; Bennett, 2016, p. 100; Sevenans, 2018, p. 250).

Hence, rather than being seen as an independent watchdog, the media is often perceived as being merely semi-independent; it still has the final say over its journalistic outcome, but is highly dependent on political officials (Cook, 2006, p. 163). The relationship between the media and politics is then reciprocal rather than unidirectional (Van Dalen & van Aelst, 2014, p. 44; Kleinnijenhuis, 2003, p. 182; Walgrave et al., 2008, p. 830). In many instances, the media respond to the political agenda. This effect is rarely controlled for in media agenda setting studies (Sevenans, 2018, p. 251; i.e. Thesen 2013). This study forms an exception to this by taking this finding into account and controlling for the reciprocal and endogenous nature of political agenda-setting by the media.

Figure 4.1 summarizes the theoretical framework of this study. The next section will outline this study’s research design.

26 Figure 4.1: Theoretical Framework

H1: TV broadcasting

type

H3_{: Issue Obtrusiveness}

Political agenda H2: Journalistic

autonomy, Media system

27

5. Research Design

To test the theoretical framework outlined in the previous section, this study employed a cross-national quantitative analysis of the media’s influence on the political agenda.

Quantitative analyses are frequently employed in relation to questions about the media’s political agenda-setting power (Walgrave & van Aelst, 2016, pp. 160-164; van Aelst & Walgrave, 2011, p. 296). Most of these studies use time-series analyses to investigate the causal relationship between the media agenda and the political agenda (i.e. Bartels, 1996; Baumgartner & Jones, 1997; Walgrave et al., 2008; Van der Pas, 2014; Vliegenthart et al., 2016a; Vliegenthart & Damstra, 2019).

Thesen (2013) deviated from this approach by employing a causal micro-analysis, which is a more in-depth research design to address causality (Walgrave & van Aelst, 2016, p. 164; Sevenans, 2018, pp. 245-246). Where time series analyses test causal mechanisms on the level of issue categories, a micro-level analysis addresses causality on the level of individual news stories (Thesen, 2013, p. 370). Consequently, the study of this thesis can also be

classified as a causal micro-analysis. An original data set had to be created specifically for the research aim of this thesis because this kind of analysis is seldomly applied to questions of political agenda-setting by the media. As a matter of fact, Thesen’s (2013) study is frequently cited as the only example of the application of such an approach in this context (Walgrave & van Aelst, 2016, p. 164; Sevenans, 2018, p. 246).

The rest of this section will further specify the research design of this study, by operationalising the most important independent and dependent variables employed in this study. Furthermore, this section ends with a motivation for the case selection and a

description of the dataset of this study.

Operationalisation of variables

Main independent variable: media agenda

As noted, this study concerns instances in which the media agenda influences the political agenda. Since this study employed a micro-level causal analysis, the media agenda was classified in terms of individual news items. In order to operationalise this concept, it was important to specify the programmes from which these items are sampled.

28 To provide a satisfactory test for the hypothesis that public service media has a greater agenda-setting power than commercial media, news items were sampled from both of these medium types. Specifically, this study consulted data from each country of the best-watched national newscasts, of commercial and public service broadcasters. The choice for the largest newscast was made because these attract large audiences, and are thus more likely to

influence or reflect public opinion. This increases the legitimacy of these programmes in public debate (Esser et al., 2012, p. 256). In addition, the most watched newscasts provide more political information, as the provision of political information plays a greater role in programmes which cater to a larger audience (Esser et al., 2012, p. 255).

The larger share of political information in these programmes and the greater legitimacy they receive in public debate cause these programmes to be most-likely cases for political agenda-setting to occur. If no effect is found for the best-watched news items, which have a greater chance to reflect the public opinion and have a greater chance of being politically relevant, then it is unlikely to find an effect for lesser-watched news items. The sampling of news programmes on the basis of audience share has been employed by a multitude of political agenda studies (i.e. Aalberg et al., 2010; Esser et al, 2012; Curran et al., 2009). Individual news items were selected from these news programmes. Specifically, the first four news items of each newscast were analysed. This is in line with the most-likely sampling procedure, as political news is highly visible at the beginning of newscasts (Albæk et al., 2014, p. 34; McNair, 2000, p. 17). Hence, the supply of political information is especially high at the beginning of television newscasts.

In sum, the first four news items were sampled from the best-watched newscasts of commercial- and public service broadcasting channels in each country included in the analysis. Causal relationships require temporal succession, which means that a cause must temporally precede the hypothesized effect (Sevenans, 2018, p. 249). Since this study concerns itself with the manner in which media attention causes an effect on the political agenda, this study’s data only includes newscasts which were aired on Mondays. Political attention, which is operationalised in the next sub-section, was measured on subsequent days of the week, to ensure temporal secession.

Main dependent variable: political agenda

In order to measure whether political agenda setting took place, data was collected which signals if individual news items were subject to parliamentary questioning. There are two reasons why parliamentary questions are used over other frequently used measures of the

29 political agenda, such as legislation, bills, executive orders or budgets (Walgrave & van Aelst, 2016, p. 159).

First of all, as noted in the literature review, the political agenda is broader than the policy agenda. A large part of politics is symbolic or not directly related to policymaking. Parliamentary questioning is a good way to gauge the political agenda because such questions are representative for the symbolic parliamentary agenda (Walgrave, Soroka & Nuytemans, 2008, p. 823). MPs often ask questions as a means of communication to the public (Van der Pas et al., 2017, p. 495). Nonetheless, parliamentary questions can also be seen as part of the policymaking process. Specifically, they are part of the problem identification phase, which is the first phase of the policymaking process that occurs before policies get formulated and adopted (Cobb & Elder, 1981, p. 394). Moreover, parliamentary questions are important means for Members of Parliament (MPs) to put issues on the legislative agenda (Van der Pas, 2014, p. 49) and force government officials to address certain issues (Van der Pas, van der Brug & Vliegenthart, 2017, p. 495). Thus, parliamentary questions conceptually cover both the symbolic and substantial nature of the political agenda.

Second, media influence on parliamentary questioning is relatively strong, due to the public and symbolic nature of this institutional venue (Walgrave & van Aelst., 2006, p. 96; Pritchard & Berkowitz, 1993). What is more, parliamentary questions are positioned at the beginning of the policymaking process, at which time the media has the highest agenda-setting influence (Walgrave & van Aelst, 2016, pp. 159-160; Vliegenthart et al., 2016, p. 841). Hence, in line with the operationalisation of the media agenda, the political agenda was operationalised in a way that makes it more likely to find cases of political agenda-setting by the media.

As noted, the media items included in this study are taken from newscasts on Monday evenings. To find out whether these media items caused a political reaction, the author coded whether these news items were mentioned in the form of parliamentary questions in the week following the news item. The question asked in this coding process was similar as the one asked in the study of Thesen (2013, p. 375): Was there a match between the individual news item and parliamentary questions asked in the following week? The parliamentary question had to deal with exactly the same event, action or statement which defined the news story, although the news source did not have to be explicitly mentioned by name. Moreover, the number of parliamentary questions was also coded. No distinction was made between oral or written questions.