Media-centered reporting styles An international comparison of election campaign coverage in TV news using qualitative comparative analysis

Transcription

1 National Centre of Competence in Research (NCCR) Challenges to Democracy in the 21 st Century Working Paper No. 72 Media-centered reporting styles An international comparison of election campaign coverage in TV news using qualitative comparative analysis Florin Büchel University of Zürich Institute of Mass Communication and Media Research (IPMZ) Division of International and Comparative Media Research Andreasstrasse 15 CH 8050 Zürich January 2014

2 Abstract Reporting style is an important concept for comparative communication research. This study aims to demonstrate how a cross-national analysis of election news coverage can serve as a meaningful example for characterizing different styles in political news reporting. Based on Qualitative Comparative Analysis (QCA), contours of three different types of political reporting styles emerge: (1) a strongly interventionist, strategy-focused approach; (2) a moderately interventionist, policy-centered approach and (3) a non-interventionist approach. The relationship between assertive news management style (political public relations) and assertive journalistic response (journalistic intervention) was found to constitute an important dimension of political reporting styles. Findings show that while national contexts still matter in explaining different reporting patterns, there is also evidence for the emergence of a transnational news culture. The results further suggest that national news cultures (at least as manifested in election news coverage) have aligned despite the fact that the political and media structures in the countries analyzed remain very different. For this study, a content analysis of election news coverage in eight countries was conducted. Two television newscasts per country, usually one from a public broad-caster and one from a commercial station, were selected. In each case the 30-minute flagship evening news program of the respective channel was chosen. All stories that were clearly electionrelated and aired in the last four weeks before polling day went into the analysis. In total, 1827 stories were coded (comprising soundbites), constituting 59.8 net program hours. In addition to descriptive analyses, a comprehensive multi-step QCA has been conducted. It yielded a solution indicating that political reporting styles can be differentiated along two perspectives: (1) whether election news coverage is centered around the journalistic voice or the political voice, and (2) whether campaigns are more issue-focused or rather focused on strategic electioneering topics. It will come as no surprise that the final mapping shows important differences (but also remarkable similarities) to the infamous triangle of Hallin and Mancini (2004). Key words: Political communication, mediatization, election campaign coverage, TV news, soundbite research, reporting styles, international comparison, qualitative comparative analysis 1

3 Reporting style as concept for comparative communication research differs from research into media systems or journalistic cultures in that it does not focus the attention on institutional arrangements or attitudinal profiles but on content features. This study aims to demonstrate how a crossnational analysis of sound- and imagebites can serve as a meaningful example for characterizing different political news cultures under the influence of the metaprocess mediatization. The starting point is the United States where the average sound bite meaning a block of uninterrupted speech by a candidate on television news shrunk from 42 seconds in 1968 to about 8 seconds in 2004 (Farnsworth & Lichter, 2007; Hallin, 1992). Patterson concluded that the US have reached the point of voiceless candidates because they have been given fewer and fewer chances to speak for themselves; in fact, candidates are said to have been squeezed out of election coverage (Patterson, 1993, p ). This study argues that this development, although significant, may not be the whole story. First, it seems useful to broaden the distinctly American perspective in this line of research and inquire to what extent soundbite news is an element of Western news culture in general. Secondly, it might be too early to dismiss short soundbites as a normatively shoddy reporting style in a priori set terms, since it is also possible that media-centered reporting styles contribute to watchdog journalism, potentially enlightening the citizens thus an explicit normative assessment of possible consequences for democracy has to be theoretically drawn out. Thirdly, it seems necessary to clearly identify the theoretical concepts and explanatory factors behind soundbite news that are at play in different media systems. Finally, a comprehensive Qualitative Comparative Analysis (QCA) which on the basis of all relevant variables discerns several distinct reporting styles was conducted. The paper at hand is a follow-up study to the analysis of Esser (2008). Soundbites and the Mediatization of Politics This study follows Hallin s (1992) definition of soundbites as film segments within a news story that show someone speak without interruption. In the literature, shrinking candidate soundbites during 2

4 election campaigns are usually seen as an indicator of an increasingly media-centered reporting style where the initiator of election narratives is the journalist, not the candidate. From this perspective, shrinking candidate soundbites refer to a reporting style that emphasizes the role of the journalist who summarizes, contextualizes and evaluates politicians messages and who basically does the speaking for the candidates (Farnsworth & Lichter, 2007). As a result, much of what voters hear about the candidates is rarely from the candidates themselves. This, it is assumed, undermines the authenticity of electoral coverage (Patterson, 1993) and makes it more difficult for viewers to fully make sense of a candidate s policies (Donsbach & Jandura, 2003). The concept of soundbites thus seeks to analyze how pronounced media interventionism is in specific national environments (Esser, 2008). It is thus an indicator for the amount of mediatization found in a country at a given time. The degree of media interventionism or mediatization of politics depends on how prevalent media logic vis-à-vis political logic is in political news coverage. Political logic is (with regard to political media contents) a journalistic reporting style correspondent to the wishes of politicians. This includes e.g. descriptive reporting, which integrates long and many quotes by politicians; hard news; consensus-orientation; etc. Furthermore, media logic is a hybrid that includes both a commercial and also a professional media logic. On the one hand, the commercial media logic refers to the commercialization of the media and therefore to the trend that makes the media dependent on advertising, with viewer levels and press circulation becoming the currency according to which media define their advertising prices (e.g. Golding & Murdock, 1996). This leads to Mazzoleni s understanding of media logic that refers to the specific narrative techniques, presentational styles and production formats that news organizations use in order to take full advantage of their own medium in a society-wide struggle for people s attention (Mazzoleni, 2008). The preferred news values and storytelling techniques include concepts like simplification, personalization, strategic framing, conflict-orientation, etc. (e.g. Strömbäck, 2008, p. 233). On the other hand, media logic can be considered as a professional media logic, emphasizing that 3

5 journalism has become an independent and unique institution itself, uncoupled from its former host institutions like politics or religion. What is important to emphasize is the antagonism between these two logics: for example, while political deliberation needs a serious and integer reporting style, it cannot exclude e.g. emotions and scandals. Mediatization also implies that actors within the public sphere will try to anticipate the rules of the media and act accordingly in order to get attention in the media, convey a positive image of themselves, etc. a process sometimes called self-mediatization. Regarding politicians, this means for example hiring spin doctors and communication experts using marketing techniques such as opinion analysis in order to tell the politicians how to talk about what, staging events such as campaign rallies just for the media, using provocative and conflictive linguistic forms and so on in order to get the attention of the media (e.g. McNair, 2000). Journalists in a professionalized media environment, however, will recognize these attempts to influence them and might react with their own counter-strategies in order to prove their independence (Kerbel, 1998, 1999): for example, they might cut out as much as possible from the original statements of politicians in order to tell the story with their own words and frames. So when journalists feel that their critical information-providing role is threatened by politicians attempts to determine what they can report and say, they will substitute products of their own instead of presenting what the politicians are trying to project (Zaller, 1998). In other words, media interventionism has become more pronounced through mediatization (Strömbäck & Esser, 2009). Whether and how journalists within different national contexts intervene is an indicator for different reporting styles among them. Mediatization can be an opportunity as well as a problem for societies. Thus, we can exemplarily identify four different reporting styles. Regarding political logic, the opportunity (1) for democracy is that we are dealing with a serious, high quality journalism that regards politics and politicians as important and that accordingly grants the politicians a lot of space in the media in order to present them- 4

6 selves and their vision. This is especially important during election campaigns, since media coverage is almost the only way for citizens to get to know their candidates and make a reasonable decision about who to vote for. On the other hand (2), this bears the great danger that journalism becomes a toothless lapdog, a pure political mouthpiece for the politicians, basically allowing them to channel any propaganda through the media. Turning further to media logic, the opportunity (3) here is that we have a critical type of journalism that understands himself as a watchdog that has to check the actions of the powerful in order to function as a fourth power in the political system. This is undoubtedly an important function of the media. However, the problem on the other hand (4) might be that the reporting style is not really critical, but rather some kind of apolitical and cynical journalism that mainly wants to ridicule the politicians just to create pseudo-conflicts that might attract viewers. To conclude, it should be avoided to a priori treat the media logic as the bad thing and the political logic as the normatively desirable way of reporting (or the other way around). Explaining Sound- and Imagebites in a Comparative Perspective From a comparative perspective, how can we explain the use of sound- and imagebites theoretically? The explanandum in this model is the average length of a soundbite by politicians. Since research on media and content production has often been conducted at the individual level without taking into account the larger structure in which journalists operate, Shoemaker and Reese (1996) suggested that there is a hierarchy of influences at different levels (e.g. individual level, organizational level, national level, etc.) shaping news production. Thus, different contextual factors on different levels (namely national macro- and organizational meso-level) are taken into account when explaining media interventionism. Basically, the macro-level factors are divided into variables regarding the political structure and the media structure of a country, while the meso-level conditions are divided into factors concerned with the political culture of politicians and the media culture of media organizations. Figure 1 shows the basic model of analysis, including the different conditions on the two levels as well as the outcome (re- 5

7 porting style, as manifested by the use of soundbites). ==[Figure 1 about here]== National Macro-Level Factors On the macro-level, there are firstly indicators regarding the political structure of a country. Several indicators are located on this level of explanatory factors. It can be assumed that the amount of confidence that people have in politics and the national government might influence how journalists report election campaigns: while the amount of confidence within the general public might be an indicator for the level of confidence in politics among the journalists themselves, it can also be assumed that the journalists would emotionalize, scandalize and personalize the coverage of campaigns more if they feel that their audience is cynical about politics anyways and that people might just turn off the TV set if politics is covered. Furthermore, the level of transparency within a political system is important. This factor could have several different implications for journalistic reporting styles: on the one hand, it is possible that an intransparent political system (e.g. if donations to political parties are not obliged to be transparent) arouses the watchdog in journalists, so that they feel motivated to thoroughly investigate these hidden backstage processes; on the other hand, it is also possible that in face of this intransparency they resign to a purely cynical reporting style. Furthermore, regarding transparent political systems a similar (although probably reversed) argument can be made: this could lead to serious quality journalism, reporting about financial flows and probable special interest policies that stand behind respective money lenders and thus enlightening the public about the mechanisms of power; however, it could also provoke an exaggerated faith in politicians (since everything is transparent anyway) and thus contribute to a sort of lapdog journalism. Finally, a further condition regarding the political structure can be found with regard to the competitiveness of elections. Here, I assume that journalists will respond with horse-race journalism focusing on strategic aspects if elections generally are a close race within a country; furthermore, this could also lead to a more aggressive political culture among the politicians, influencing reporting styles via the meso-level of political cultures. 6

8 Secondly, we have to consider conditions on the macro-level that can be attributed to the media structure of a respective media system. For this set of factors we can resort to the media system characteristics as identified by Hallin and Mancini (2004), namely the journalistic professionalism (also see Hallin, 1996), inclusiveness of the press market, political parallelism and the role of the state. Journalistic professionalism means that journalism is differentiated as an occupation and institution from other social institutions, particularly politics, and follows a distinct common culture and distinct modus of operation. For Hallin and Mancini (2004) professionalism mainly means autonomy from outside influences and outside control over one s work. A highly professionalized form of journalism would thus follow a news logic that is succinctly different from the logic of the political system, thus contributing to a more interventionist reporting style. Furthermore, the inclusiveness of the press market asks whether newspaper products generally have a high reach, and also whether they have a high reach among minorities and marginalized majorities (e.g. working class, women, etc.). This is not only a quantitative distinction, but also one showing the nature of newspapers and its relation to audiences in a country (Hallin & Mancini, 2004, p. 22). The absence of this sort of inclusive press market thus also means in some kind of converse argument that the public heavily relies on electronic media such as television for political information (also see Norris, 2000; Hallin & Mancini, 2004, p. 24). Thus, it can be assumed that television journalism might be more interventionist in countries where the press market is less important. The third media system characteristic that Hallin and Mancini describe is the amount of political parallelism found in a media system. This relates to the question whether certain media organizations have traditionally close ties to certain political parties (also see the discussions about external and internal pluralism in a media system). External pluralism and political parallelism are thought to induce a rather commentary oriented journalism, which might be an indicator for high media interventionism. It s also perfectly justifiable though to argue that high political parallelism will lead to less media interventionism, since journalists in such systems are relatively close to the political power holders (also see 7

9 Pfetsch, 2004) and thus might grant politicians more space to present themselves in the media (even if this probably only applies for politicians of certain parties). Finally, Hallin and Mancini argue that the role of the state in taking action regarding media policies is also important. This includes for example the importance of the public broadcasting channels in a country. Since these are usually legally obliged to produce diverse media content that is devoted to a public service orientation, we can assume that public broadcasters tend to produce a more serious journalism than their commercial counterparts, integrating many views and actors that might not be newsworthy enough for commercial enterprises. As the case of the United Kingdom and the BBC might show, the presence of a dominant and high quality public broadcaster in a country can also lead the commercial channels within the same media system to follow the high standard of the public market leader. Organizational Meso-Level Factors Furthermore, conditions on the organizational meso-level have to be considered. Again, let s firstly look at factors regarding the political culture of a party. The style of election campaigns is directly related to how media might cover the campaign. The general hypothesis here would be that the more professionalized (and thus self-mediatizing) a campaign is, the more media interventionism is provoked. So, for example, the more a politician tries to influence the media to convey a positive image of him using marketing techniques, the more journalists will intervene to prove their independence. An indicator for this type of political professionalism is for example whether candidates mainly show themselves in staged events such as campaign rallies where they have a lot of control over the whole communication situation and don t need to fear nasty questions (since only supporters are present anyways), or whether they are more open to present themselves in less controlled circumstances. Furthermore, the specific media culture found in a media organization should also directly influence reporting styles. Again, the amount of professionalism (and thus the likelihood of resorting to interventionist strategies) is the important factor on this level. For example, do journalists break in on 8

10 political candidates or do they let them talk? Do they grant them the last word and image in an election campaign story aired on television, or do they insist to present a concluding statement (which is thought to have the most lasting impression) themselves? These factors are directly related to media interventionism. Finally, it is also important what it is that the politicians talk about. So, the soundbite topics have to be taken into consideration. However, it s not always clear whether this is a feature of the political or the media culture: since it s possible both that the journalists will use only soundbites that are e.g. aggressive as well as that the politician mainly uses aggressive statements in his campaign, the topics that are discussed can belong both to the media and the political culture. For example, if journalists cut out only a short attack soundbite from a speech of one hour by a politician, it s rather an indicator for media culture; but if those are almost the only soundbites that a candidate voices, it s an indicator for political culture. Sample To analyze the amount of media interventionism, a content analysis of all election related stories in the last four weeks before polling day on the flagship evening news bulletins of British, French, Swiss, Danish, Italian, Spanish, US-American and German TV channels from 2007 to 2011 has been conducted. In the United States, we content-analyzed ABC World News Tonight and NBC Nightly News, in Great Britain the 10 o clock news on (the public broadcaster) BBC1 (called news at ten ) and ITV1 (simply called news ), in Germany Tagesschau on (the public broadcaster) ARD and RTL Aktuell, in France the 8 o clock news on TF1 and (the public broadcaster) France 2 (called 20heures on both channels), in Spain Telediario on (the public broadcaster) TVE1 and Noticias on Antena3, in Italy TG1 on (the public broadcaster) Rai Uno and TG5 on Canal5, in Denmark Avisen on (the public broadcaster) Denmark Radio and Nyhederne on TV2 and finally in Switzerland Tagesschau and 10 vor 10 on (the public broadcaster) SF1. All broadcasts follow similar formats, are anchored by prominent journalists, and pro- 9

11 vide a net news time between 22 minutes (mostly on the privately-owned, advertising-financed stations) and 30 minutes (mostly on the license fee-financed public service broadcasting stations). However, the Spanish channels show rather long newscasts of ca. 45 minutes. The study at hand is a follow up to a study conducted by Esser (2008), who examined four of these countries across time, while this study focuses on a cross-country comparison only, but with eight countries. First, all news shows were recorded in their entirety. Then all those items, however brief, that were labeled or framed as general election stories, were identified and marked in a logbook. In operational terms, the universe of this study consists of all stories that contained a clear verbal or visual reference to the upcoming general election usually, but not always, in the form of a background logo or an anchor remark. Some of these stories did not mention the candidates at all, other offered lengthy sit-down interviews with them. It is note-worthy that most American soundbite studies exclude interview stories a priori from their samples (e.g. Farnsworth & Lichter, 2007) which leads them to report shorter soundbite lengths for U.S. candidates than our analysis does. Yet it would be unfair (and in fact a severe case of selection bias) to ban all those stories from the analysis in which journalists try to respond to scholarly criticism by actively inviting candidates to participate in the news discourse in a more lengthy manner. Specially trained teams of international coders were assigned to stories depending on their language skills and country familiarity. They then were trained and supervised by the principle investigators until reliability tests showed coder-researcher agreement above 90 percent for all relevant categories. Most categories required only simple stop-watch measurements. The official campaign periods in the examined countries differ substantially in length. The only way to construct strictly equivalent samples for all campaigns in all countries was to analyze all electionrelated stories that aired in the last 20 workdays before polling day. This yielded a total number of 1827 stories (in which soundbites were found) on the above mentioned stations with an overall length 10

12 of ca program hours that were eventually coded and form the empirical basis of this study (see table 1). ==[Table 1 about here]== Measures Outcome: Length of Candidate Soundbites The first measure to explain is the outcome. What we want to explain is the average length of a candidate soundbite in election campaign news. In the content analysis, this measure has been directly clocked by coders (rounded to seconds). Furthermore, for soundbites by the candidates it has been coded which candidate it is, although the candidates themselves are of no direct concern for this study. National Macro-Level Factors The data for the national macro-level factors have been collected from various secondary sources. Firstly, for the political structure indicators had to be found that measure the amount of confidence people have in their governments, the transparency of a political system and the competitiveness of elections in a country. For the confidence in a government, two measures from the Gallup polls have been used (see Gallup, Inc., 2013): the direct question asking whether one has confidence in the national government and a question asking whether one has contacted an official recently (both measured as percentage of people that answered with yes on the level of countries). Furthermore, for the transparency and competitiveness, indicators could be found within the data from the Democracy Barometer (see Democracy Barometer, 2013), which is a project at the University of Zurich and University of Berlin devoted to analyzing different types of democracies as well as the democratic quality of these types. The indices derived from this data source are direct measures for the mentioned concepts. Secondly, for the media structure, indicators had to be found measuring the media systems characteristics as described by Hallin and Mancini (2004). These data were gathered in a joint project at the IPMZ (see Brüggemann, Engesser, Büchel, Humprecht, & Castro, 2013) and describing the whole opera- 11

13 tionalization in detail would greatly exceed the scope of this paper. The data stem from various sources, such as the World Press Trends (see WAN-IFRA, 2010), the World Value Survey (see World Values Survey Association, 2007), the European Election Studies (see European Election Studies, 2009), the expert survey European Media Systems Survey (see Popescu, 2010) and the journalist survey Worlds of Journalism (see Hanitzsch, 2010). We constructed indices for the four media system characteristics out of these data. In this project, we split the characteristic role of the state into several distinct subindicators in the current analysis, only the dimension regarding the importance of the public broadcaster has been used. The aim of the original project was to see whether a quantitative, exploratory mapping of these data (using cluster analysis and QCA) would lead to similar media system models as those conceptualized by Hallin and Mancini (2004) however, the measures for the different media system characteristics are perfectly suited to function as explanatory factors as well. Organizational Meso-Level Factors All data on the organizational meso-level are from the content analysis. Firstly, regarding the political culture, I argued in chapter 2 that the amount of professionalism that political parties have reached in their communication attempts is important. In the context of elections, one indicator for the level of professionalism is whether candidates mainly show themselves in staged events where they have a lot of control over the whole communication situation and don t need to fear nasty questions, or whether they are more open to present themselves in less controlled circumstances. This has been measured with the content analysis: for example, a campaign rally is a highly controlled situation, since only supporters are present there anyways; a speech in the parliament is already less controlled, and in a spontaneous interview on the street there is no control at all. In the analysis, the percentage of fully controlled situations has been used as indicator. Furthermore, regarding media culture, the fact whether the candidate or the journalist is granted the last word and the last image of an aired election story has been used as an indicator (the only indicator on story- instead of soundbite-level: percentage of stories with candidates or journalists having 12

14 the last word and / or the last image). Finally, it has been argued that what the politicians talk about (topics of soundbites) is utterly important as well it is, however, not clear whether this is a feature of media culture or political culture (respectively it can be both, depending on the situation). This variable that describes the topics that politicians talk about have been coded in the content analysis as well. The possible values for this variable are attack (attacking an opponent), defense (defending one s own position), issues (policy) and campaigning (electioneering statements, such as the chances to win or commenting recent polls etc.). Taken together, these indicators are the explanatory measures in the analysis. Table 2 outlines them quickly to give an overview. ==[Table 2 about here]== The main data-analysis strategy used in this study is QCA, which is a well-established technique in political science and slowly being applied in media and communication science (e.g. Nguyen Vu, 2010; Brüggemann & Kleinen von Königslöw, 2013, forthcoming; Downey & Stanyer, 2010). The main reasons for using QCA in this case are the medium-sized sample of news outlets and the ability to easily identify very complex interactions of conditions, which is highly complicated, using other methods such as multilevel-regression when one wants to model three-way and higher interactions. QCA was developed by Charles Ragin (1987, 2008) by applying (fuzzy) set theory to questions of social science. It combines quantitative, variable-based logic and qualitative, case-based interpretation. Set theory assumes that (a) causality is a combination of various conditions (explanatory variables) that eventually produce an outcome; (b) several different combinations of conditions may produce the same outcome; and (c) depending on the context a given condition may have a different impact on the outcome. Thus different causal paths may lead to the same outcome. Compared to other methods like cluster analysis, QCA allows the researcher to study differences and commonalities between cases in greater detail (since it employs a case-oriented interpretation). QCA is based on set theory, and thus codes all cases as being either absent or present according to their 13

15 variable values. Normally a high value indicates the presence of a certain variable, while a low value means its absence. The cut-off point on the scale which divides the absent and present cases has to be set by the researcher, based on theoretical case knowledge. This procedure is called calibration and can be seen as one of the major strengths of QCA as qualitative knowledge and quantitative data are combined within this step. Set theory and logic know two different mechanisms of causality (e.g. Ragin, 2000, p. 90ff): necessary conditions and sufficient conditions (see figure 2 for an illustration). For necessary conditions, each case that shows an outcome should also show the respective condition, whereas for sufficient conditions, each case that shows the condition also shows the outcome. In terms of Boolean algebra or set theory, a good way to visualize these mechanisms of causality is to imagine Venn diagrams (see figure 2 again): in a sufficient condition, the outcome is a superset of the condition ( if a case shows the condition, it also shows the outcome, but not all cases that show the outcome show the condition ); in a necessary condition, the outcome is a subset of the condition ( if a case shows the outcome, it also shows the condition, but not all cases showing the condition must show the outcome ). The goal of QCA is usually to identify combinations of conditions that are sufficient for a specific outcome. These are often called INUS-conditions, which stands for insufficient but necessary condition in a combination of conditions that itself is unnecessary but sufficient (see Mackie, 1965). ==[Figure 2 about here]== While building upon Boolean algebra and binary logic, QCA is not restricted to dichotomously coded sets. By integrating the so-called fuzzy-logic (see Ragin, 2000, 2008), the amount of membership a case has in a set can be graded. Fuzzy values range from 0 to 1, with 0.5 representing the qualitative anchor above which a case is a member of a specific set. The values between 0.5 and 1 (respectively between 0.5 and 0) represent the strength of the membership of a case (respectively of the nonmembership). The best way to visualize necessary and sufficient conditions in fuzzy-logic are XY-plots (see figure 2 again): if all cases are above the main diagonal, we are dealing with a sufficient condition, 14

16 if all cases are below the main diagonal it is a necessary condition. The space of the plot can be further divided to see which cases are irrelevant, which ones can or cannot be explained and which ones are ( true or untrue ) contradictions to the relation. For fuzzy algebra and Boolean algebra, the same procedures are applied to calculate logical conjunctions ( logical and-combination, intersection of sets), logical disjunctions ( logical or-combination, union of sets) and negations ( logical not ):2 a conjunction returns the minimum of all values on combined conditions; a disjunction returns the maximum and a negation is 1 minus the value on a condition (see Ragin, 2006, p. 11). In Boolean algebra, the result will obviously always be 1 or 0, but not in fuzzy algebra. Parameters of fit (consistency and coverage) can be calculated both for necessary and for sufficient conditions (see figure 2; also see Ragin, 2008, p. 44ff). Consistency is the parameter that tells us how consistent or contradictory the cases are, while coverage shows how (ir)relevant a condition is (in other words, how strong the explanatory power of a condition is). The rough proceeding of a QCA involves several important steps. These are shortly outlined here to give an overview. First of all, the variables that are measured for a certain number of cases should be calibrated into fuzzy sets (see tables 4, 5, 6, 7 and 3 in the appendix). The researcher has to decide which cases are (non-)members of a particular set, and if he wants to use fuzzy sets how strong their membership is. Once this is done, a qualitative label has to be added to the name of the particular set (compared to the variable ): for instance, the outcome in this study will no longer be the variable length of a candidate s soundbite, but the set of cases with long candidate sound bites. When the variables are calibrated into sets, the researcher can examine the set relations between conditions and an outcome. Furthermore, for the identification of sufficient conditions, the so-called truth table analysis is usually implemented. The first step in a truth table analysis is to create the truth table itself. It is a table that shows all possible combinations of conditions. Then, each case is assigned to the row in the truth table to which it corresponds. The next step in the truth table analysis is to examine for each row 15

17 whether the respective cases show the outcome or not consistency values are calculated to help the researcher to identify rows that clearly show the outcome consistently and rows that are contradictory. Usually, a consistency value of higher than 0.9 is considered satisfactorily consistent. Once this is done, the analysis itself can be applied: the truth table rows that show the outcome are combined with logical conjunctions between the conditions within one truth table row and with logical disjunctions between the truth table rows. This expression is then minimized using the rules of Boolean algebra. Consistency and coverage values are finally calculated both for single solution paths and for the overall solution. The last step of the QCA eventually is the interpretation. It is very important that the researcher draws the results back to the single cases by interpreting the solution with regard to the cases rather than the variables / sets (see Wagemann & Schneider, 2010, p. 410). A very useful visualization to help the interpretation is an XY-plot with the overall solution term in the X-axis and the outcome in the Y-axis. ==[Figure 3 about here]== In a multi-step approach (see figure 3), I first identified conditions at the national level that were sufficient for the outcome and used this solution as a condition in the second step of the analysis to identify interactions between macro- and meso-conditions (see Schneider & Wagemann, 2013). However, unlike the two-step approach as described by Schneider and Wagemann, I opted for a multi-step approach, breaking up the different conditions on the different levels into succinct analyses (see figure 3). In the first step, the indicators from the meso-level should be included in the analysis (three different analyses), theoretically resulting in four distinct solution paths for the meso-level; in the next step, the measures from the macro-level are introduced (two different analyses), resulting again in four solution paths; in a third and fourth step, the respective four solution paths should be brought down to two for each level, which would result in four distinct solution paths that can be analyzed in a final fifth step. However, in the effective empirical praxis this theoretical analysis model had to be slightly modified to result in reasonable solutions. 16

18 Results: Explaining Journalistic Reporting Styles Outcome: Length of Candidate Soundbites First of all, let s look at the average length of candidate soundbites (which is the outcome in the second, explanatory step). It was mentioned earlier that most soundbite studies exclude interview stories for the analysis, which potentially leads to findings of shorter soundbites. Thus, we report the average soundbite length both for non-interview soundbites, for interviews and for both together (total). This is on the one hand an indicator whether the exclusion of interviews might contribute to a certain sampling bias as mentioned earlier in the theory section, and on the other hand the differentiation works as a guidance to decide what to do in the following analyses. Tables 3 and 8 show the average length of these candidate soundbites across the examined TV channels and countries (values printed red are the actual values that will be explained in the second step of the analysis). Looking only at soundbites that are not extracts from interviews, the briefest candidate soundbite is indeed found in the US-American election campaign (7.4 sec), which can be explained by the high degree of media autonomy and commercial media structure, as well as the skeptical and less party-oriented communication culture in the United States. Furthermore, the longest (noninterview) candidate soundbites are identified in the Italian (17.5 sec) and Spanish elections (19.6 sec), which can be explained by the historically strong ties between state and media in the Mediterranean media systems and their shorter tradition of independent journalistic culture. Finally, Switzerland shows an average candidate soundbite almost as long as in Italy (16.3 sec), while France, Germany and Denmark show rather short candidate soundbites (France: 12.3 sec; Germany: 11.0 sec; Denmark: 9.0 sec). An interesting fact comparing Switzerland to Italy is that while they show almost the same average length of packaged candidate soundbites, Switzerland has the lowest (N = 124) and Italy the highest (N = 1147) density of soundbites. Therefore, we can conclude that Italian politicians had a lot of opportunities to speak through the media in the 2008 snap election (high density) and were also given a lot of time for their statements (long packaged candidate soundbites), whereas in Switzerland the number of 17

19 election stories was already lower to begin with (also see table 1). So for the non-interview soundbites, there are three groups of countries, those with above-average (non-interview) soundbite length (Spain, Italy, Switzerland), below average (United States, Germany, Denmark) and average soundbite length (France, Great Britain). Spain and Italy thus allow for the largest integration of politicians in their election campaign coverage, but the same is true for Switzerland. Further analyses will show whether the candidates in those three countries were treated with the same friendliness, or whether they were also confronted with challenging questions and probing interviews. ==[Table 3 about here]== Furthermore, we also have to look at interview soundbites (respectively the total length of candidate soundbites, which are not differentiated by interviews) to fully understand the phenomenon: generally, soundbites are longer when interviews are included in the analysis (the UK is an exception here, it seems that they compress interview soundbites even more than non-interview ones). The shortest average soundbite (interviews included) is identified in French channels (12.3 sec), not in the US- American anymore. French channels do not really show much interviews though to begin with (N = 2). The longest soundbite is still found in the Spanish channels, where the difference between noninterview soundbites and interview soundbites is strikingly high: candidates in Spain get on average and in total the possibility to speak 34.7 sec without interruption, which is almost twice as long as the country with the second-longest soundbites (Italy, 18.9 sec). The very long soundbites in Spain when interviews are included can be explained by the fact that Spanish channels devoted entire newscasts to lengthy sit-down interviews with the main candidates. Finally, some differences between channels in the respective countries can be found. In Spain and Italy, the commercial channels show strikingly longer soundbites than the respective public-service channels (in Italy especially obvious regarding interview soundbites: 26.3 sec in Rai Uno vs sec in Canal5; in Spain especially eye-catching regarding the total average soundbite length: 29.7 sec in TVE1 vs sec in Antena3). An explanation for this 18

20 pattern in the southern European countries might be the historically close proximity of journalists to power-holders, so the commercial channels interestingly show a rather sacerdotal reporting style. In Italy, we also have to take notice of the fact that Silvio Berlusconi, who ran for office in the 2008 snap election, has considerable influence over the commercial channel Canal5. The long soundbites in interviews especially in the Spanish and Italian channels might also be an indicator that journalists do not really ask the hard questions in these two countries. Thus, regarding the second research question, we can identify some country-specific patterns (short soundbites especially in France, Germany, Great Britain, Denmark and the United States, long soundbites especially in Switzerland, Italy and Spain), but also some channel-specific patterns (commercial broadcasters in southern European countries show a more sacerdotal reporting style than their respective public-service counterparts). Another finding is that a priori excluding interview soundbites might indeed contribute to some sort of selection bias, thus it would be preferable in future soundbite research to explicitly draw the difference between interview and non-interview soundbites. ==[Figure 4 about here]== Explaining Interventionist Reporting Styles After having looked descriptively at the average soundbite length of candidates, the second step in the analysis is to explain the soundbite length. As laid out in section 3, this will be done in several steps, combining the factors on the different levels. A first QCA on the meso-level is applied to analyze how the different topics combine into INUS-conditions regarding the outcome (short candidate soundbites). The solution is shown in figure 4 (XY-plots and solution paths). Two distinct solutions leading to short candidate soundbites have been found: (1) the absence of defense topics combined with the presence of campaigning topics (French and US-American channels) represent a mediatized campaign in which candidates do not need to defend their own positions, but mainly present statements regarding the electioneering aspect of the campaign ( horserace ); (2) the presence of defense topics either combined with an absence of attack topics (German 19

21 public service channel ARD and Danish channels) or with the presence of substantial topics (German and Danish public service channels ARD and Denmark Radio as well as the British channels) represents a substantial campaign in which candidates need to defend themselves and either do not attack the political opponents regularly or do talk about substantial issue topics often (focus on policy issues). These two solution paths show that both a substantial issue-centered political and media culture as well as a strategy-centered culture can lead to short candidate soundbites; in terms of our normative assessment of election campaign coverage, the former would be judged positively with the buzz phrase fourth power, while the latter would be classified as problematic and cynical journalism. ==[Figure 5 about here]== Furthermore, the next step in the theoretical model would have been to calculate QCAs for the factors regarding the political culture ( control ) respectively the media culture ( last word and image ). However, this had to be modified slightly to produce reasonable results: these two levels were combined in one single QCA using high control over the communication situation, last word by the candidate, last word by the journalist and last image by the candidate as conditions. Again, two distinct solution paths leading to short candidate soundbites have been found (see figure 5): (1) the absence of many stories in which the candidate is granted the last word combined with a highly controlled political campaigning culture (US-American channel NBC and German public service channel ARD) represent a highly mediatized campaign in which political candidates do show themselves mainly in fully controlled situations where they don t need to worry about getting daring questions, to which journalists react with media interventionism by not granting the politicians the last word in news stories (and also by using short soundbites); (2) stories in which the journalists insist on having the last word in stories, but still grant the candidates at least the last image in a story regularly represent a media-centered coverage that is slightly less interventionist than the first reporting style, but still focused heavily on the journalists themselves (French channels as well as the German and British private channels RTL and ITV). 20

22 Let s turn to the macro-level conditions now and see firstly how the indicators regarding the political structure influence the length of candidate soundbites. Again, two solution paths leading to short candidate soundbites have been found (see figure 6): (1) the strange situation in the US that people do not have confidence in their national government, but still regularly voice their opinions to officials represents an unresponsive political system in which people do get in contact with their representatives, but aren t really confident about their actions;6 (2) a transparent political system (Denmark, Great Britain and France) in which, for example, financial flows to the campaigns have to be made public, also leads to short candidate soundbites in certain countries, which is an indicator for potential investigative journalism. ==[Figure 6 about here]== Furthermore, another QCA has been applied to see how the media system characteristics on the level of media structure are connected to short candidate soundbites. One of the four media system characteristics stands out in this solution as important all on its own: the amount of journalistic professionalism (see figure 7). Out of all the countries present in this study, only the Mediterranean countries Italy and Spain do not show professionalized journalism, and those are also the countries where candidates are granted a lot of space in TV news to present themselves. British, French, German, Danish and US-American channels are all members in the set of countries with professional journalism, and they also all do show rather short candidate soundbites. However, one true contradiction is present here for the first time: Switzerland does show a high amount of journalistic professionalism, but nevertheless candidates are granted a lot of space in election campaign news. This could have to do with the fact that in a small country such as Switzerland, the national elites (such as journalists on the public service broadcaster and politicians) are closely related and often enough even know each other, thus leading to a rather sacerdotal reporting style despite professional journalism. Another ad-hoc explanation is that in Swiss elections, it is not as obvious as in the other countries who the actual political candidates are, 21

23 which is due to a complicated election system regarding the allocation of executive functions; thus, since the executive ( Bundesrat ) comprises seven equally important politicians from different parties (determined in a proportional election), usually only few of those executive politicians are exchanged in national elections, which leads to an advantage of incumbency for the governing candidates and also to a reduced necessity to actually campaign in the forefront of elections. ==[Figure 7 about here]== Finally, all the solutions derived in individual QCAs can be combined in a last step of analysis. In the theoretical model of analysis, the idea has been to use the four solutions found on the meso-level (respectively the three solutions on the macro-level) each in another analysis to first bring down the number of solutions in order to be able to eventually combine them in a last step; however, the individual solutions found on the two levels already had strong explanatory power, leaving little to reduce. Thus, the strategy of analysis had to be modified slightly again to find a reasonable final solution: the two solutions regarding the topics of soundbites as well as the transparency of a political system and the journalistic professionalism of a media system have proven to be the most meaningful. Using these four conditions in the final QCA produced three distinct solutions paths leading to short candidate soundbites, all of them related to professionalized journalism (see figure 8): (1) professionalized journalism in an intransparent political system, explaining Germany and the US (with Switzerland present as a true contradiction again); (2) professionalized journalism combined with an issue-centered campaigning culture ( substantial topics ), explaining Danish and British channels as well as the German public broadcaster ARD ( Northern European reporting style ); and (3) professionalized journalism that meets a lot of mediatized soundbite topics, which is the case for US-American and French channels (representing a highly mediatized campaign). ==[Figure 8 about here]== How would we need to normatively assess these three different reporting styles? The solution 22

24 path showing professional journalism in an intransparent political system (comprising the USA and Germany) which leads to short candidate soundbites (Switzerland is a contradiction to this solution) is the one showing a lot of overlap to other solutions: the US-American channels are also present in the third solution, while the German channel ARD is also present in the second solution. Only the German private broadcaster RTL is explained solely by this solution. RTL has not been explained in the first step of the analysis that evaluated the topic conditions (see figure 4), thus it cannot belong to one of the other two final solutions. Looking at the meso-level values for RTL (see table 7), we see that RTL does show a combination of topics that is unique for the channels with media-centered reporting styles (understood as showing short soundbites; only the Spanish channels show the same combination of topics): absence of substantial topics, but presence of attack, defense and campaigning soundbites. This is an indicator for a rather mediatized campaign, but the fact that defense topics are still present also shows that politicians regularly need to defend their positions despite the lack of space for substantial topics in RTL soundbites. Its public service counterpart, ARD, shows both defense and also substantial topics. So while in all other countries, differences between the two respective channels seem negligible, the two German channels actually show some differences. ARD also belongs to the second solution of a Northern European reporting style dedicated to substantial topics, which would qualify as a reporting style dedicated rather to a political logic of serious journalism, while the presence of RTL only in the first solution path (intransparent political systems) is an indicator for a reporting style dictated by the cynical part of media-logic. Regarding the second solution path (Northern European reporting style) that combines the presence of lots of substantial topics with a high degree of journalistic professionalism on the macro-level, it also needs to be argued that this reporting style is more devoted to a serious political logic regarding the topics, but also more to a watchdog journalism out of the possible reporting styles devoted to media logic (since defense as well as substance topics are present, but still the politicians are only granted limited space to mediatize themselves in order to influence the campaign coverage). 23

25 The BBC that is present in this solution path might be the exemplary ideal-type of this reporting style. Finally, the US-American and French channels that show highly mediatized topics in an environment of professionalized journalism are exemplary for a reporting style that might be rather cynical, with little substance and defense topics and a focus on strategic campaigning topics. The process of the campaign itself and the personalities of the candidates seem to be of higher interest for both the media and the politicians than actual policy issues, discussing for example social problems and probable political solutions. Finally, we can also quickly check whether those cases that do show long candidate soundbites (Switzerland, Italy and Spain) share a particular combination of conditions (out of the conditions used for the final analysis of short candidate soundbites). They do indeed share three conditions, or rather the lack thereof: they all show an intransparent political system, no substantial topics (as defined with the first analysis step) and no mediatized topics. RTL, which was kind of a strange case (together with Switzerland) in the analysis all along, also shows this combination of condition, which would qualify RTL as a contradiction here. So despite the intransparency of the political system, Spanish and Italian channels (as well as the Swiss channel SF1) show a rather sacerdotal reporting style with long candidate soundbites, providing space for the self-mediatization efforts by politicians. Additionally (see table 7), there is a difference between the Spanish and Italian channels on the one hand and the Swiss channel on the other: the channels in the Mediterranean countries also often grant the candidates the last word and the last image in a TV news story, another indicator for a toothless lapdog journalism in these countries. The Swiss public broadcaster does not do this (also see table 5), in fact it has the lowest score for granting the candidates even only the last image (9.8% of all stories) and the second lowest score for leaving them the final word in a story (2.0% of all stories, under-matched only by RTL with no stories with the last word by the candidate at all). These findings are again evidence of a special situation in Switzerland, being a small country with many direct-democratic elements and with a complex election 24

26 system regarding the political executive. Discussion The findings suggest that journalistic intervention is not a phenomenon which is specific to the United States only. Several of the results presented indicate transnational trends, with three discernible news cultures: two interventionist and one non-interventionist styles. Firstly, regarding the two different interventionist cultures, we find that channels that do not show long candidate sound bites share different conditions: firstly, they all do show high journalistic professionalism; additionally, the US- American and French channels show high amounts of mediatized soundbite topics, while the Northern European channels show a high density of substantial soundbite topics instead. The conclusion here is that the candidates in the United States and France are punished by journalists for mediatizing the topics of their campaigns, whereas the journalists in Great Britain, Denmark and in the German public broadcaster ARD regularly grant the candidates the opportunity to say something about the normatively important substance topics in news stories, but still grant them only short soundbites to do so. Secondly, regarding the non-interventionist reporting style, we find that channels that do show long candidate sound bites (Italy, Spain and Switzerland) share three conditions as well: they show an intransparent political system combined with a lack of both substantial as well as mediatized soundbite topics. Especially the channels in the Mediterranean countries are affine to a rather sacerdotal reporting style of lapdog journalism: the candidates are also regularly granted the opportunity to say the last words in a story; in Switzerland though, which also shows long candidate soundbites, we do not find many stories with the last word granted to the candidates, indicating a rather serious and issue-centered reporting style, but with some moderate interventionism. A few conclusions can be drawn by taking the analyses together. Firstly, it is striking how similar the channels within one country are: apart from the German channels, no further solutions were found with different solution paths for the different channels in one country. A conclusion here is that the 25

27 differences between public and private broadcasters are impressively small when it comes to election campaign coverage. Possibly, miscellaneous elections (also within one country) might differ substantially and be very heterogeneous: they might be autonomous events with different contextual factors making them unique, so that journalists eventually report the same event, no matter whether they are in a public or a private organization. Thus, news about this event are very similar between different news outlets, but differ between various elections. This might explain the striking differences between (clusters of) countries, but similarities within countries. Ideally, further analyses across time should study the differences between miscellaneous elections within one country preferably for a few countries in one analysis. Secondly, the different reporting styles that were found represent different transnational trends. The cultures identified bear similarity to Patterson s (2008) mapping of Western journalists role positions and Hallin and Mancini s (2004) typology of Western news systems. Patterson s study is based on survey responses, Hallin and Mancini s on institutional features, and my own much more modest undertaking on actual news content, explained with meso and macro-level factors. All three data sources have to be seen together to investigate the fundamental assumption behind comparative political communication research. It states that specific constellations of media and political structure (as manifested on the organizational, national and transnational level) characteristically shape the facets of cultures (as manifested, among others, in the ideas, practices and products of news journalists). With regard to soundbite news, our study was able to give a fairly specific assessment of the explanatory power of factors located on the different levels. While national contexts still matter the most, our evidence for the emergence of a transnational news culture is equally intriguing. The findings suggest that national news cultures (at least as manifested in sound- and imagebites) have aligned along several discernible transnational news cultures despite the fact that the political and media system structures in the countries analyzed remain very different. News cultures may be a more autonomous unit of anal- 26

28 ysis than previously thought; future research ought to address this question more systematically. In the words of Blumler and Gurevitch (1995, p. 74), the assumption states that different features of the structures, norms, and values of political [and media] systems will differentially promote or constrain political communication roles and behaviors within those systems. 27

29 References Blumler, J. G., & Gurevitch, M. (1995). The Crisis of Public Communication. London: Routledge. Brüggemann, M., Engesser, S., Büchel, F., Humprecht, E., & Castro, L. (2013). Towards an Empirical Typology of European Media Systems: Revisiting Hallin and Mancini (2004) [Conference Paper]. Mainz: Paper Presented at the DGPuK Annual Conference in Mainz, Germany, May Brüggemann, M., & Kleinen von Königslöw, K. (2013, forthcoming). Cosmopolitan Coverage. Causal Recipes for Patterns of Foreign News Coverage in European Newspapers. European Journal of Communication, 28(4). Democracy Barometer. (2013). Democracy Barometer [Secondary Data]. Zurich, Berlin: University of Zurich and Social Science Research Center Berlin. Retrieved May 26, 2013, from Donsbach, W., & Jandura, O. (2003). Chances and Effects of Authenticity. Candidates of the German Federal Election in TV News. Harvard International Journal of Press / Politics, 8(1), Downey, J., & Stanyer, J. (2010). Comparative Media Analysis: Why Some Fuzzy Thinking Might Help. Applying Fuzzy Set Qualitative Comparative Analysis to the Personalization of Mediated Political Communication. European Journal of Communication, 25(4), Esser, F. (2008). Dimensions of Political News Cultures: Sound Bite and Image Bite News in France, Germany, Great Britain, and the United States. Harvard International Journal of Press / Politics, 13(4), European Election Studies. (2009). World Value Survey [Survey Data]. Mannheim: European Election Studies. Retrieved May 26, 2013, from Farnsworth, S. J., & Lichter, R. S. (2007). The Nightly News Nightmare: Network Television s Coverage of U.S. Presidential Elections, Lanham, MD: Rowman & Littlefield. Gallup, Inc. (2013). Gallup World View [Polling Data]. Washington, D.C.: Gallup, Inc. Retrieved May 26, 28

30 2013, from Golding, P., & Murdock, G. (1996). Culture, Communications, and Political Economy. In J. Curran & M. Gurevitch (Eds.), Mass Media and Society (Second ed., p ). New York: St. Martin s Press Inc. Hallin, D. C. (1992). Sound-Bite News: Television Coverage of Elections, Journal of Communication, 42(2), Hallin, D. C. (1996). Commercialism and Professionalism in the American News Media. In J. Curran & M. Gurevitch (Eds.), Mass Media and Society (Second ed., p ). New York: St. Martin s Press Inc. Hallin, D. C., & Mancini, P. (2004). Comparing Media Systems. Three Models of Media and Politics. New York: Cambridge University Press. Hanitzsch, T. (2010). Worlds of Journalism [Journalist Survey Data]. Munich: Worlds of Journalism. Retrieved May 26, 2013, from Kerbel, M. R. (1998). Edited for Television: CNN, ABC, and American Presidential Politics (Second ed.). Boulder, CO: Westview. Kerbel, M. R. (1999). Remote and Controlled: Media Politics in a Cynical Age (Second ed.). Boulder, CO: Westview. Mackie, J. L. (1965). Causes and Conditionals. American Philosophical Quarterly, 3(2), Mazzoleni, G. (2008). Mediatization. In L. L. Kaid & C. Holtz-Bacha (Eds.), Encyclopedia of Political Communication (Second ed.). McNair, B. (2000). Journalism and Democracy: An Evaluation of the Political Public Sphere. London: Routledge. Nguyen Vu, H. N. (2010). Money Matters. A Cross-National Study of Economic Influences on TV News [Doctoral Thesis at the University of Zurich]. Zurich: University of Zurich. Retrieved May 26, 2013, from 29

31 Norris, P. (2000). A Virtuous Circle: Political Communications in Postindustrial Societies. Cambridge: Cambridge University Press. Patterson, T. E. (1993). Out of Order. New York: Knopf. Patterson, T. E. (2008). Political Roles of the Journalist. In D. A. Graber, D. McQuail, & P. Norris (Eds.), The Politics of News, the News of Politics (p ). Washington, D.C.: CQ Press. Pfetsch, B. (2004). From Political Culture to Political Communications Culture: A Theoretical Approach to Comparative Analysis. In F. Esser & B. Pfetsch (Eds.), Comparing Political Communication: Theories, Cases, and Challenges (p ). New York: Cambridge University Press. Popescu, M. (2010). European Media Systems Survey [Expert Survey Data]. Essex: European Media Systems Survey. Retrieved May 26, 2013, from Ragin, C. C. (1987). The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. Berkeley, Los Angeles, London: University of California Press. Ragin, C. C. (2000). Fuzzy-Set Social Science. Chicago, London: University of Chicago Press. Ragin, C. C. (2006). User s Guide to Fuzzy-Set / Qualitative Comparative Analysis 2.0 [Computer software manual]. Tucson, Arizona: University of Arizona, Department of Sociology. Retrieved August 23, 2011, from Ragin, C. C. (2008). Redesigning Social Inquiry: Fuzzy Sets And Beyond. Chicago, London: University of Chicago Press. Schneider, C. Q., & Wagemann, C. (2013). Set-Theoretic Methods for the Social Sciences. Cambridge: Cambridge University Press. Shoemaker, P. J., & Reese, S. D. (1996). Mediating the message: Theories of Influences on Mass Media Content. New York: Longman. Strömbäck, J. (2008). Four Phases of Mediatization: An Analysis of the Mediatization of Politics. Harvard International Journal of Press / Politics, 13(3),

32 Strömbäck, J., & Esser, F. (2009). Shaping Politics: Mediatization and Media Interventionism. In K. Lundby (Ed.), Mediatization: Concept, Changes, Consequences (p ). New York: Peter Lang. Wagemann, C., & Schneider, C. Q. (2010). Standards of Good Practice in Qualitative Comparative Analysis (QCA) and Fuzzy-Sets. Comparative Sociology, 9(3), WAN-IFRA. (2010). World Press Trends [Secondary Data]. Darmstadt, Paris: WAN-IFRA. Retrieved May 26, 2013, from microsites/world-press-trends World Values Survey Association. (2007). World Value Survey [Survey Data]. Michigan: World Values Survey Association. Retrieved May 26, 2013, from Zaller, J. (1998). The Rule of Product Substitution in Presidential Campaign News. Annals of the American Academy of Political and Social Science, 560(1),

33 Tables Table 1: Sample Description Content Analysis Country TV Channel N Election Stories Mean Story Length Cumulative Story Length Total Length of Election News Germany 2009 RTL sec 83 min ARD sec 149 min 232 min USA 2008 NBC sec 158 min ABC sec 176 min 334 min Spain 2008 Antena sec 249 min TVE sec 378 min 627 min Italy 2008 Canal sec 197 min Rai sec 273 min 470 min Denmark 2007 TV sec 241 min DR sec 250 min 491 min Switzerland 2011 SF sec 143 min 143 min France 2007 F sec 297 min TF sec 230 min 527 min Great Britain 2010 ITV sec 366 min BBC sec 397 min 763 min 32

34 Table 2: Measures Explanatory Conditions Political Media Structure Culture Transparency, Confidence, Voiced Opinion, Competitioni Control Topics Professionalism, Political Parallelism, Press Market, Public Broadcasting Last Word / Image 33

35 Table 3: Outcome Length of Candidate Soundbites Country TV Channel Excluding Interviews: Length of Candidate Soundbite (N) Only Interviews: Length of Candidate Soundbite (N) Total: Length of Candidate Soundbite (N) ABC 7.8 sec (257) 29.5 sec (55) 13.5 sec (312) USA 2008 NBC 6.9 sec (200) 43.3 sec (32) 12.7 sec (232) Total 7.4 sec (457) 33.8 sec (87) 13.2 sec (544) BBC 14.2 sec (557) 10.5 sec (46) 18.8 sec (603) GB 2010 ITV 13.3 sec (728) 8.6 sec (74) 12.5 sec (802) Total 13.7 sec (1285) 9.4 sec (120) 13.1 sec (1405) RTL 9.1 sec (173) 13.2 sec (12) 9.6 sec (185) D 2009 ARD 12.3 sec (205) 47.5 sec (6) 15.3 sec (211) Total 11.0 sec (378) 30.3 sec (18) 12.9 sec (396) TF sec (517) - (0) 11.6 sec (517) F 2007 F sec (598) 10.0 sec (2) 12.9 sec (600) Total 12.3 sec (1115) 10.0 sec (2) 12.3 sec (1117) CH 2011 SF sec (99) 19.2 sec (25) 17.5 sec (124) Canal sec (241) 31.5 sec (47) 20.8 sec (519) I 2008 Rai sec (565) 26.3 sec (63) 17.4 sec (628) Total 17.5 sec (1037) 28.6 sec (110) 18.9 sec (1147) TV2 8.7 sec (324) 25.8 sec (129) 14.1 sec (453) DK 2007 DR 9.3 sec (241) 23.9 sec (95) 13.6 sec (336) Total 9.0 sec (565) 24.9 sec (224) 13.9 sec (789) TVE sec (606) 47.4 sec (93) 29.7 sec (699) E 2008 Antena sec (276) 57.3 sec (82) 42.5 sec (358) Total 19.6 sec (882) 52.2 sec (175) 34.7 sec (1057) 34

36 Figures Figure 1: Basic Model of Analysis Macro- and Meso-Level Conditions 35

37 Figure 2: Visualization of Sufficiency and Necessity in Set Theory 36

38 Figure 3: Multi-Step QCA Model of Analysis 37