Running head: DELIBERATIVE QUALITIES OF NEWS FRAMES Deliberative qualities of generic news frames: Assessing the democratic value of strategic game and contestation framing in election campaign coverage

News frames are patterns of news construction journalists rely on to present information to their audiences. While much of the research on news frames has focused on their identification and effects, less work has investigated the specific contributions these different frames make to democratic life. Value judgments about distinct news frames are often not generated in a systematic fashion, not grounded in democratic theory, and/or not supported by empirical evidence. In this article, we address these problems by arguing for and extending normative assessment as a standard operating procedure to determine the democratic value of political communication phenomena. We demonstrate the usefulness of normative assessment by showing how two important generic news frames (politics as a strategic game and as a substantive contestation) contribute to a deliberative public discourse prior to a general election. Using data on television news coverage of the German federal election campaign in 2009, we investigate how these frames are related to the inclusiveness and civility of public discourse and the extent to which it features exchanges of substantive reasons for political positions. Results show that mediated democratic deliberation suffers consistently from strategic game framing while contestation frames make ambivalent contributions. Implications for political communication scholarship as well as journalistic practice are discussed. DELIBERATIVE QUALITIES OF NEWS FRAMES 3 Introduction News framing analysis is a rich subfield of political communication with a rapidly growing number of studies (see Borah, 2011; D’Angelo & Kuypers, 2010). Research in this field has shown that—apart from more focused issue-specific news frames—news coverage in general is characterized by a number of generic news frames (Iyengar, 1991; Neuman, Just, & Crigler, 1992; Schuck & de Vreese, 2006; Semetko & Valkenburg, 2000). Generic news frames are patterns of news content that “transcend thematic limitations and can be identified in relation to different topics, some even over time and in different cultural contexts” (de Vreese, 2005, p. 54). These patterns have their roots in journalists’ more or less conscious professional norms and routines (Gans, 1979; Tuchman, 1980). It is striking that the occurrence or even dominance of certain generic news frames is usually not systematically linked to the democratic performance of the news media at large. Thus we have little solid knowledge about what a certain amount of, for example, conflict or human interest or consequences framing in the news media means for the functioning of democracy. More specifically, we do not know how these frames are to be judged from the perspective of different theoretical views of democracy. If relations between the framing concept and normative demands formulated in democratic theory are drawn in the literature, this is most often done in studies of framing effects, which focus on the malleability of citizens’ opinions and preferences (e.g., Druckman, 2001; Druckman & Nelson, 2003). It is this well-documented reactivity of citizens to frames in the news and other types of communication that makes them consequential and important to study. But although some studies of news framing relate news reporting characteristics to democratic theory (e.g., Entman, 2004), most studies lack coherent and systematic value judgments about their findings (see Lawrence, 2010). Thus the growing body of knowledge about news framing does not tell us as much as it could about the democratic performance of the news media and the democratic value of key DELIBERATIVE QUALITIES OF NEWS FRAMES 4 journalistic routines. News framing analysis carries an unfulfilled potential both of understanding the functioning of modern democracies and of guiding democratic practice: If news framing research was able to specify what exactly is right or wrong with the prevalence of certain generic news frames, and which normative perspective it is that guides such judgment, it could also make more precise contributions to adjustments in the world of communication practice. What we propose in this paper, therefore, is to undertake a systematic normative assessment of common generic news frames by studying their empirical association with normatively desirable features of news content. Such an undertaking needs three ingredients: (A) data on commonly used generic news frames, (B) a relevant theoretical source of explicit normative standards by which to assess the democratic value of these generic news frames, and (C) a well-developed, systematic assessment procedure. (A) In this paper, we focus on the strategic game frame and the contestation frame. These two generic news frames are of interest not only because they are particularly common in election campaign coverage. They are important because they capture the two main dimensions of political competition: politics as an instrumental competition of interested actors (i.e., as strategic game) and as a substantive competition of policy ideas (i.e., as contestation). These two types of competition lie at the heart of most established theories of democracy (see Held, 2006), even if they are emphasized to varying degrees by different theories. Strategic game framing in election coverage is usually contrasted with issue coverage (Aalberg, Strömbäck, & de Vreese, 2012) while contestation framing is usually contrasted with a framing of substantive political positions as uncontested (i.e., only one position is presented or multiple positions are presented but remain unrelated). Thus a single news story can exhibit both a strategic game and a contestation frame—an empirical association that we look into below. DELIBERATIVE QUALITIES OF NEWS FRAMES 5 (B) In assessing the normative value of these often-used news frames we work with the framework provided by theories of deliberative democracy (e.g., Gutmann & Thompson, 2004; Habermas, 2006). We ask: What are the deliberative qualities of news stories characterized by strategic game and contestation frames? News framing processes are an integral element of a larger process of mediated deliberation (Maia, 2009, 2012). Some scholars have pointed to the relevance of media frames for the deliberative quality of public communication and have studied frames pertaining to individual issues (e.g., Maia, 2009; Pan & Kosicki, 2001; Simon & Xenos, 2000). But a normatively inspired news framing analysis has not been conducted across a range of issues or for generic news frames. This is what we set out to do, using data on the television news coverage of the 2009 German federal election campaign as our case in point. (C) Finally, we concur with Althaus (2012) that systematic normative assessment should become a standard procedure in political communication research and present an extension of his conception below. In doing so, we develop empirical hypotheses which we test in the second part of this paper. Empirical scholarship can benefit from explicit connections to normative thinking in several ways. For one, normative theories can focus attention on a range of phenomena that might otherwise be overlooked (Althaus, 2012; Lazarsfeld, 1957, p. 41; Peters, 2008, p. 67). Second, systematic normative assessment can broaden the range of normative standards by which to evaluate empirical findings (Althaus, 2012, p. 98). And last but not least it highlights the relevance of specific empirical findings in terms of broader philosophical concerns and thus can make political communication scholarship relevant beyond our own ranks (Althaus, 2012, p. 99). These general benefits of normative assessment are just as relevant to news framing research as they are to other areas of political communication. Connecting News Framing and Mediated Deliberation Research DELIBERATIVE QUALITIES OF NEWS FRAMES 6 Why should news framing and deliberation research be connected in the first place? Have they not lived separate lives for good reasons in the past? It is important to note that we are not talking about small-group deliberation here, but about mediated deliberation or, more precisely, about deliberative features of news coverage. We are not looking for specific communication settings or arrangements that are particularly conducive to or have been explicitly designed to foster deliberative exchanges. Instead we look at deliberativeness as a variable quality of any kind of communication, including the mediated political communication that television news provides. Deliberative news content—characterized by civil and respectful exchanges of ideas and justifications—is a vital component of a larger conception of a deliberative democratic system (Wessler, 2008). Vis-à-vis citizens, deliberative news media content can serve as a reservoir of conflicting positions and justifications that reduce information costs drastically. It can also serve as a model for deliberative behavior in everyday political talk. Vis-à-vis political decision-makers, deliberative news media content may enhance the quality of decisions by prompting political elites to avoid decisions and positions that are difficult to justify and therefore may entice widespread disapproval in media discourse (Wessler, 2008, pp. 5–7). The degree to which these salutary effects of deliberative news media content are realized is of course subject to empirical investigation and will vary according to circumstance. The important point is that there are good normative reasons why news media content should be deliberative. If this is the case for news media content in general, does it also hold for news framing choices in particular? After all, news framing may be seen as a primarily strategic, not deliberative, form of action: Through generic news framing journalists try to package the news in a way that is attractive to their audiences rather than trying to fulfill the prescriptions of theorists of deliberative democracy. Why should it then make sense to subject generic news frames to seemingly incommensurable normative standards? First, we do not have to assume that journalists follow a deliberative logic of action. In fact, we do not need to assume any DELIBERATIVE QUALITIES OF NEWS FRAMES 7 kind of logic at all to be able to apply a normative perspective inspired by deliberation theory. Political news culture (i.e., the content produced by journalists; see Esser, 2008) can be studied without a need to also study journalism culture (i.e., the professional selfunderstanding and institutional roles of journalists; see Hanitzsch, 2007). Citizens may learn from different positions and justifications presented in the media even if these positions and justifications are presented to them merely for the instrumental purposes of maximizing political support (on the part of politicians) and audience shares (on the part of journalists) (Schultz, 2006). Thus the journalistic choice of a generic news frame may have salutary or detrimental effects regardless of whether it was made in order to facilitate deliberation or for other reasons. In addition, principles, including those of public deliberation, can become consequential when embedded in professional standards of journalistic action (Bro, 2010). Empirically, studies of both journalists’ professional attitudes (e.g., Voakes, 1999) and the content they produce (e.g., Gerhards, Neidhardt, & Rucht, 1998) indicate that they frequently harbor and make explicit ideals about facilitating substantial societal debate or a fair and open exchange of ideas. The inclination towards those ideals is likely to vary with the organizational form of the specific media outlet. For example, public service broadcasting stations in Europe are often governed by laws or codes of good practice that explicitly call for inclusive, substantive and fair treatment of disagreement that go far beyond general legal provisions against libel. Individual journalists will also vary in their propensity toward facilitating meaningful deliberation. But it is not plausible to categorically juxtapose framing decisions and normative orientations towards deliberation or to assume that framing choices should be unaffected by such orientations. As we will show below, the use of strategic game and contestation frames is connected to deliberative features of news coverage on both a conceptual and an empirical level. DELIBERATIVE QUALITIES OF NEWS FRAMES 8 In recent years these deliberative features of news discourse have been operationalized for empirical study in varying fashion (e.g., Bennett et al., 2004; Ettema, 2007; Ferree, Gamson, Gerhards, & Rucht, 2002; Graham & Witschge, 2003; Maia, 2009; Rohlinger, 2007; Wessler, 2008). Selecting criteria against which to judge the quality of media content is an often neglected exercise that implicates theoretical decisions that need to be made carefully. In light of normative arguments and extant empirical evidence, which democratic standards are most central and applicable to the specific phenomenon under study? Deliberative democratic theory has sometimes been interpreted by empirical researchers to demand that a substantive consensus over contested issues must be established among participants for public discourse to be successful (e.g., Ferree et al., 2002; Gerhards, 1997). A second, similarly demanding yardstick that has been employed in mediated deliberation research is the idea that discourse participants should publicly admit to reversing their opinions in light of arguments opposing their original position (Maia, 2009). While substantive consensus and reversibility of opinions certainly constitute deliberative ideals, a specific theory of deliberation under conditions of competition and mass mediation would not expect that they be actually fulfilled. The triadic structure of public controversies in a mass media forum, where participants address specific segments of the public to garner their support rather than talk to each other, makes such outcomes highly unlikely (Peters, 2008, p. 239). Not surprisingly, setting such unrealistically high standards for mediated deliberation has led to dim conclusions about its democratic potentials. From a more refined normative standpoint, however, mass-mediated deliberation may still contribute positively to democratic life even in the absence of substantive consensus and public reversals of opinion by presenting reasoned dissent among public speakers (see Wessler, 2008). We therefore concentrate on those four criteria for deliberative news content that are essential for such reasoned dissent to come about: DELIBERATIVE QUALITIES OF NEWS FRAMES 9 1. Inclusiveness of speakers: This criterion focuses on the social diversity of individuals given a voice in news discourse. It is particularly concerned with the degree to which groups from the political periphery (citizens, intellectuals, etc.) occur in the news (Maia, 2009). Various theorists emphasize the importance of an inclusive public sphere for the realization of both moral values (like political autonomy and equality of citizens) and cognitive gains in the deliberative process (e.g., Bohman, 1996; Cooke, 2000; Gutmann & Thompson, 2004). 2. Civility: This criterion aims at the degree to which public discourse is conducted in a polite and respectful manner. It also pertains to politeness and respect in the presentation of political discourse as it is produced by journalists and news editors (e.g., Rohlinger, 2007). As part of the “throughput” of mediated deliberation, deliberative democratic theories highlight the importance of civility as a precondition for the reasoned exchange of ideas between speakers in the news and the desired ensuing weighing of conflicting arguments by citizens. Such civil exchanges may lead to a change in preferences based on “good reasons” or at least the mutual acknowledgment of legitimate opposition (e.g., Dryzek, 2005; Gutmann & Thompson, 1996; also Mutz, 2007). 3. Responsiveness: This criterion is concerned with the degree to which ideas and opinions are related to one another and put into a “dialogue” either by letting opposed speakers refer to one another directly or by letting a journalist present and connect their views (e.g., Bennett et al., 2004). Deliberative democratic theories value responsiveness as a prerequisite for the reflexive refinement of arguments or even the reversal of positions. As a consequence, responsive news content is valuable as it increases the quantity and quality of arguments available for media audiences’ own deliberations (e.g., Bennett et al., 2004; Wessler, 2008). DELIBERATIVE QUALITIES OF NEWS FRAMES 10 4. Reason-giving: This criterion concerns the degree to which actors provide substantial reasons for their positions in a given political controversy. This demand extends to journalists who are supposed to elicit reasons from political actors and render them to their audiences (e.g., Ettema, 2007). Giving reasons as public political justification is central to most normative accounts of deliberative democracy. It is either argued to be a necessary component of any demanding notion of public political rationality (Kuhlmann, 1999) or even considered a necessary prerequisite for the realization of further democratic values like equality or autonomy (Forst, 2007; for an overview of this literature see Chambers, 2010). In this view, high-quality democratic communication involves speakers from various social backgrounds who engage in public debate characterized by civil exchanges of explicitly justified positions and ideas. Normative Assessment of Generic News Frames Normative claims in news framing research, when present, tend to take the form of ad hoc statements and represent assertions more than well-founded evaluations. Few studies pursue a program of gauging systematically and explicitly those aspects of news frames that are relevant to democratic theory. Systematic normative assessment offers a solution here. It is an analytical procedure that aims at identifying and explicating the criteria underlying value judgments about empirical research findings (Althaus, 2012). But it is not normative in itself. Rather, it is a clarifying practice that exposes the multiple connections between the empirical world as we observe it and the normative world as we would like it to be—acknowledging that there are multiple ideal worlds as developed in different democratic theories. Althaus (2012) distinguishes four “levels” of normative assessment. In his typology, the more explicit the use of normative criteria becomes, the higher the level of assessment reached. On the first level, the criteria used for evaluating an empirical finding are not spelled out at all resulting in DELIBERATIVE QUALITIES OF NEWS FRAMES 11 mere assertions about the normative value of the finding. On the second level, an explicit normative criterion is given (“Finding A is troubling because it lacks quality X”). On the third level, a standard is provided for judging the normative status achieved (“Finding A is troubling because it has less than Y amount of quality X”) (Althaus, 2012, p. 100). On the fourth level, normative assessment relates several criteria for judging findings to one another. This practice thus excavates possible conflicts between normative theories in the assessment of specific empirical observations. Examples of this kind of sophisticated normative assessment are rare but do include seminal works like those of Baker (2002) and Ferree et al. (2002). However, recent theoretical (e.g., Anderson, 2007; Christians, Glasser, McQuail, Nordenstreng, & White, 2009; Strömbäck, 2005) and empirical (e.g., Freelon, 2010; Trappel & Maniglio, 2009) political communication research registers increasing engagement with the kind of complex normative questions posed in fourth-level normative assessment. We like to add to this typology that fourth-level normative assessment need not necessarily be inter-theory assessment but can just as well be intra-theory assessment whenever an empirical finding speaks to multiple normative values invoked by a single normative theory. An obvious benefit of normative assessment is that it enables researchers to come to more transparent normative conclusions about political communication phenomena, conclusions that can stake a more legitimate claim to validity as they are explicitly grounded in developed normative theory. We aim at reaping this benefit by assessing different generic news frames. Extending Althaus’ proposal, we follow a two-step procedure in conducting our normative assessment. Any theoretical construct carries both conceptual and empirical implications. When assessing a construct normatively, it is thus helpful to distinguish between conceptual normative assessment and empirical normative assessment. Conceptual normative assessment is an examination of a concept for its logical implications with respect to a set of normative values. It involves investigating the intension of a concept (i.e., the qualities or attributes it connotes, see Hurley, 2008). Conceptual normative assessment establishes DELIBERATIVE QUALITIES OF NEWS FRAMES 12 whether the construct’s definition speaks to a certain theory such that the mere presence of the phenomenon made observable by the construct must be judged positively or negatively (or ambivalently) according to that theory. Empirical normative assessment, on the other hand, is an examination of the empirical relations of the phenomenon denoted by a concept with other normatively relevant phenomena. Assessing strategic game frames The academic literature on election coverage shows a considerable lack of agreement about the differences between “horse-race frames” (e.g., Russonello & Wolf, 1979), “strategic frames” (e.g., Schulz & Zeh, 2007), and “game schema” (e.g., Patterson, 1980, 1993) or “game frames” (e.g., Lawrence, 2000). Consequently, these terms have often been used interchangeably and varying proposals for their integration have been made (e.g., Binderkrantz & Green-Pedersen, 2009; Faßbinder, 2009; Lengauer, 2007). In a recent review article, Aalberg, Strömbäck and de Vreese (2012) propose to think of game and strategy frames as distinct components of a larger generic strategic game frame. In their conceptualization, a game-framed news story centers around which candidate or party is winning or losing, who supports whom, and what the outcome of a political conflict might be. A strategy-framed news story, in contrast, centers around the instrumental actions of political actors, particularly their public relations efforts and political marketing decisions. We follow this typology and include both strategy and game elements of political reporting in our analysis of strategic game frames (see below). Strategic game framing as a generic framing device requires normative assessment for two reasons. First, it has been a staple of election news coverage across different national contexts. Despite some variation across political and media structures and cultures, election coverage is characterized by a focus on strategic aspects of campaigns in most democracies (Kaid & Strömbäck, 2008). This includes Germany where this has been true for television DELIBERATIVE QUALITIES OF NEWS FRAMES 13 news since the mid-1990s (Genz, Schönbach, & Semetko, 2001; Schulz & Zeh, 2007). Second, strategic game frames are frequently claimed to harm democracy by trivializing politics and crowding out substantive issue coverage (e.g., Farnsworth & Lichter, 2007; Patterson, 1980, 1993). However, these claims usually fall short of a rigorous assessment of the phenomenon and resemble what Althaus (2012) calls “normative assertions”. And indeed, the traditional negative connotation of strategic game framing has been called into question by research showing that it need not be associated with a decrease in substantive coverage in certain contexts (e.g., Bartels, 1988; Gan, Teo, & Detenber, 2005). More to the point, the value of framing politics as a strategic game from the perspective of different normative models of democracy remains largely unknown (Wessler, 2008, p. 9). As a concept, strategic game framing directly speaks to the inclusiveness criterion only. This is particularly true in an election context, when a strategyor game-framed political message focuses on the politics of a handful of political protagonists that are up for election. This narrow focus privileges a thin slice of those affected by policies (i.e., political decisionmakers) when it comes to the allocation of public attention and speaking opportunities. And while it does not entirely preclude experts and “common people” from getting airtime a focus on the powerful contenders in gameand strategy-framed coverage is inevitable. It follows that a dominant presence of strategic game frames in election coverage is per se problematic in this regard. By contrast, the degree to which strategic game framing is associated with levels of civility, responsiveness, and reason-giving in the news is less clear and can only be determined through empirical normative assessment. When we consider civility, we first need to recognize that strategic behavior is inherently actor-bound. Therefore, if journalists choose to present politics as a horse-race of strategically acting candidates and parties, they should be likely to emphasize the lines of conflict between them. We thus expect that strategic game DELIBERATIVE QUALITIES OF NEWS FRAMES 14 framing often involves the reporting of clashes between political opponents and that this increases the likelihood of uncivil behavior finding its way into the news: H1: Election coverage containing a strategic game frame will exhibit a lower level of civility than coverage without a strategic game frame. The relation of strategic game frames to the responsiveness of news stories (i.e., the degree to which they connect opposing political ideas and positions) is not immediately clear. On the one hand, the focus on strategic competition in strategic game coverage might seem to give journalists opportunities to compare and contrast the programs and positions of political competitors. On the other hand, it seems more plausible that a strategy focus leads them to emphasize individual political actors and their actions and prevents them from drawing connections between the substantive positions of candidates and parties. We thus hypothesize: H2: Election coverage containing a strategic game frame will exhibit a lower level of responsiveness than coverage without a strategic game frame. Strategic game coverage has traditionally been asserted to lead to a lack of “substance” in news reporting (e.g., Lichter, 2001). While the conceptualization of substance is often not clear-cut, we argue that any demanding version of the concept must reflect a need for speakers in the news to present reasons for their points of view, especially in a deliberative model of democracy (Ettema, 2007). It indeed seems plausible that the coverage of an election campaign as a strategic game among candidates or parties diverts attention away from their policy positions (Kerbel, Apee, & Ross, 2000), let alone their substantive justifications for them. We thus expect: H3: Election coverage containing a strategic game frame will exhibit a lower level of reasongiving than coverage without a strategic game frame. Assessing contestation frames DELIBERATIVE QUALITIES OF NEWS FRAMES 15 Contestation framing is a way of presenting the news that explicitly contrasts two or more points of view on a particular topic, the key point here being that such framing focuses on the substantive aspects of political competition. Empirical investigations of contestation frames are rare. In their study of major television news in six countries, Cottle & Rai (2006) found a contestation frame in 8.2 to 15.4 percent of news stories. Contestation frames have not been discussed as distinct generic frames in the literature before but bear clear conceptual semblance to the more commonly investigated generic conflict news frame. The conflict frame was first introduced by Neuman et al. (1992). Like the contestation frame, it emphasizes the journalistic construction of news in terms of polarized, contentious forces. However, Neuman et al. (1992, p. 64) explicitly connected the conflict frame to game and horse-race news frames, thus conceptualizing it in strategic rather than substantive terms. Later adaptations of the conflict frame incorporated more substantive elements like “disagreement” and “points of divergence” (de Vreese, Peter, & Semetko, 2001, p. 109), but still combined it with strategy and game (see also Semetko & Valkenburg, 2000). The recent proposal of a related “contest” frame (Hänggli & Kriesi, 2012) restricts it to strategic competition (“politics”) but does not conceptualize a corresponding frame that would capture the communication of substantive political competition. Therefore, despite its conceptual lineage with the conflict frame, the value added of introducing the contestation frame as a distinct generic news frame is that it allows for a clear-cut distinction between strategic game elements on the one hand, and substantive (or ideational) elements of political competition on the other, instead of conflating them in one unitary concept. 4 This generic frame can thus regularly be observed in TV news reporting. Findings from news studies investigating cognate concepts like “multiperspectivalness” (Benson, 2009; Gans, 1979, 2003, 2011) and “dialogic structure” (Ferree et al., 2002; Rohlinger, 2007) further attest to its empirical relevance. DELIBERATIVE QUALITIES OF NEWS FRAMES 16 Much like strategic game framing, contestation frames have normative overtones in the literature and typically are valued positively. However, statements that contestation frames generally contribute to deliberation (Cottle & Rai, 2006, p. 171) or constitute “evidence of ... positive and democracy-nurturing behavior” (Robertson, 2010, p. 519) remain vague in that they do not link the frame’s characteristics and consequences with specific values espoused in normative democratic theory. Considering the four criteria of deliberativeness cited above, we find that attaching a positive value to the contestation frame concept apriori is justified to the extent that it connotes responsiveness. Responsiveness is the normative value most central to theories of deliberative democracy. The “epistemic dimension” of democracy (Habermas, 2006), that is, the desired cognitive gains of the public use of reason, cannot be realized without the substantive exchange of different views on common issues. In our analysis below, we therefore treat contestation as the positive indicator of a responsive discourse. Conceptually, the responsive contestation frames behave neutrally towards the degree of inclusiveness, civility and reason-giving in the news. Therefore, their relations to these criteria of news quality are subject to empirical assessment. When looking at speaker inclusiveness no self-evident relationship to contestation frames reveals itself. In principle, contestation framing may involve actors from the center as well as the periphery of the political system. However, the competition of ideas denoted by contestation does not necessitate a focus on elite actors, but may rather invite inclusion of elite-challenging peripheral actors more than when issues are framed in a non-responsive fashion. We thus hypothesize: H4: Election coverage containing a contestation frame will exhibit a higher level of inclusiveness than coverage without a contestation frame. What is the relationship of contestation frames in television news with the level of civility? The opposing viewpoints featured in contestation frames are likely to be presented not by DELIBERATIVE QUALITIES OF NEWS FRAMES 17 journalists themselves but by personal representatives, if only to guard journalists against accusations of political bias. We expect that, particularly in the run-up to a major election, debate between opposing speakers, even if substantive, bears a potential for personal confrontation and uncivil rhetoric aiming at delegitimizing the opposing speaker. Hence we hypothesize: H5: Election coverage containing a contestation frame will exhibit a lower level of civility than coverage without a contestation frame. Our last empirical question concerns the relation of contestation frames to the extent of reason-giving in news coverage. Contestation frames imply explicit links between two or more opposing viewpoints, by either a journalist or other speakers. We expect that this confrontation of views and issue positions builds up communicative pressure on political protagonists to justify their positions and thus increases the number of reasons presented to the audience: H6: Election coverage containing a contestation frame will exhibit a higher level of reasongiving than coverage without a contestation frame. Design and Methods To test these hypotheses, we analyzed news coverage from the four television channels with the highest market shares in Germany: ARD and ZDF (both public service) as well as RTL and Sat.1 (both commercial). The analysis included all news items mentioning the 2009 federal election in the 112 nightly newscasts aired on the four channels in the four weeks preceding the election (August 30, 2009 to September 26, 2009). This left us with 175 news stories from ARD’s Tagesschau (n = 43), ZDF’s heute (n = 47), RTL Aktuell (n = 50), and Sat.1 News (n = 35). The coding scheme involved two units of analysis: the news item (for the strategic game and contestation variables) and the individual utterance (for the types of speakers, DELIBERATIVE QUALITIES OF NEWS FRAMES 18 civility, and reason-giving variables). All utterances, primary or quoted, were coded. The 175 news items comprised a total of 1424 utterances. We coded five key variables. (The complete coding protocol is available in the online appendix.) First, we determined the extent of strategic game framing as compared to issue coverage using a fine-grained topic variable. Based on past research and the election manifestos of the five parties represented in the German national parliament we listed political topics relevant to the election and categorized them as either “policy” or “strategic game”. Drawing on the conceptualization by Aalberg et al. (2012, p. 167), we employed a total of four mutually exclusive thematic indicators of strategy framing (election campaign/party PR; election metacoverage) and game framing (coalition speculations/election outcomes; opinion polls) (see Strömbäck & Dimitrova, 2011 for a similar operationalization). The second key variable on the item level was contestation framing. This variable captured the presence of opposing views in the same news item but only if a link was established between a claim and a counterclaim by a journalist or another speaker (Wessler, 2008). On the utterance level, we coded three variables. To gauge the inclusiveness of actors in the newscasts, we coded the types of speakers using an extensive list of actor types grouped into three broader categories: the center of the political system (legislative/party, administrative/government as well as judiciary actors), the periphery of the political system (special interest organizations, social movements, experts, intellectuals, citizens affected by a problem/policy, celebrities, and other media personnel), and journalists presenting the particular news show analyzed. Civility was operationalized as its inverse, the absence of incivility. Incivility is most often characterized in the deliberation literature as unusually impolite forms of verbal or physical social behavior. Drawing on several other studies as well as on pretests of our instrument, we developed an operationalization of incivility that distinguishes between verbal DELIBERATIVE QUALITIES OF NEWS FRAMES 19 and visual forms of incivility as well as between incivility committed by speakers and uncivil styles of journalistic presentation. On the verbal side, incivility means the use of “hot button language, that is, words that are likely to outrage opponents” (Ferree et al., 2002, p. 239). Somewhat weaker violations of the civility norm are tapped with the indicator “derogatory style”. “Interruption by a speaker” and “personal attack” (Mutz, 2007, p. 624) account for uncivil behavior between speakers. Verbal incivility also occurs when journalists interrupt speakers, which can also happen by truncating a speaker’s statement if that makes the utterance semantically incomplete or pointless. On the visual side, incivility was coded when speakers showed each other “nonverbal disrespect”, for example by rolling their eyes (Mutz, 2007, pp. 624–625). An indicator of uncivil journalistic practices was the use of certain camera operations, particularly extreme close-ups of speakers involved in a dispute (Mutz, 2007, p. 623). The last variable we coded was reason-giving. Reasons are linguistic expressions that support an argumentative structure and do not consist of factual information, polemic, threats, and narration only (Schultz, 2006). They offer either explanation or proof (Kuhlmann, 1999), both of which can be seen as attempts at fulfilling two of Habermas’ validity claims (1987): An explanation supports the validity claim of comprehensibility, a proof fosters the validity claim of truthfulness by making a traceable connection to reality. Following Gerhards et al. (1998), we coded whether a speaker gave a reason for their position or not. All coding was done by two of the authors. Final intercoder reliabilities were established based on a double-coding of newscasts selected at random from the full sample until the suggested threshold of 50 coding acts (Neuendorf, 2002) was reached for every variable. Krippendorff’s alpha (α) was ≥ .83 for each variable, surpassing the .80 threshold level proposed by Krippendorff (2004). We prepared our data for analysis in two steps. First, we constructed binary indicators from each original variable to limit the number of cells in the contingency tables for log-linear DELIBERATIVE QUALITIES OF NEWS FRAMES 20 analysis, thus avoiding excess numbers of zero-count cells that can bias parameter estimates (Knoke & Burke, 1980). This also served to make the interpretation of results more straightforward. Second, we transformed the three utterance-level variables of interest into dichotomous item-level indicators. We thus finally worked with five dichotomous variables on the news item level. They indicate the presence of a strategic game frame and/or a contestation frame in the item as well as whether it contained at least one instance of incivility and/or reason-giving by a speaker and/or at least one speaker belonging to the periphery of the political system. The frequency of news stories corresponding to each combination of our variables of interest can be considered in the context of a 2 × 2 × 2 × 2 × 2 design (strategic game frame vs. no strategic game frame; contestation frame vs. no contestation frame; includes uncivil utterance vs. includes no uncivil utterance; includes utterance giving a reason vs. includes no utterance giving a reason; includes a speaker of the periphery vs. does not include a speaker of the periphery). We used log-linear modeling to analyze the multiway contingency table corresponding to this data structure. Log-linear modeling is typically used for modeling the log of cell counts in a contingency table as functions of the constituting categorical variables. It is especially useful for the analysis of association and interaction patterns involving multiple response variables (Agresti, 2002). As a first step here, it helps us to see whether the relationships we posit in our hypotheses hold beyond controls for all other possible associations between the five variables. We present results in the order of hypotheses and begin with the omnibus multivariate results before we present follow-up bivariate analyses. Empirical Findings For the 175 news stories included in our analysis, we selected the most parsimonious well-fitting model in a hierarchical log-linear analysis: Starting from the saturated model, we gradually reduced the number of included effects until any further exclusion of an effect DELIBERATIVE QUALITIES OF NEWS FRAMES 21 would significantly decrease the fit of the model with our data. After the model reduction process had thus come to a halt, we removed the only remaining three-way interaction effect in the model, which only marginally decreased model fit (χ = 4.09, df = 1, p = .043). Since the resulting model still fits the data well (LR χ = 17.14, df = 19, p = .58), we selected it as the final model. Table 1 reports tests of all effects retained in the selected model, adjusted for all other effects in the saturated model. [Table 1 about here] Three out of the six effects predicted in our hypotheses are statistically significant in the omnibus test with a fourth (Contestation × Reason-giving) bordering significance (χ = 3.68, df = 1, p = .06). This multivariate finding provides reassurance that these effects are robust, even in the presence of controls for different types of news stories. However, it tells us nothing about the direction and magnitude of the relationships signified by the significant effects. We therefore went on to look at them individually. Looking first at the univariate distributions of the two generic frames, we find that, consistent with prior news research, the majority of election stories on television news (68.0%) featured a strategic game frame of the election while only a minority of all election news stories (42.3%) was framed in terms of a contestation between opposing substantive positions. The 2009 federal election campaign thus was covered more as a strategic competition between candidates and parties than as a substantive competition between the ideas they represented. We next looked at the associations of strategic game frames with different indicators of news deliberativeness. The bivariate associations with all indicators included in the loglinear model are displayed in Figure 1. First of all, we found clear support for the assumption that strategic game framing, at least as operationalized here, privileges political elites (i.e., the “political center”): Only one-third (36.1%) of stories containing a strategic game frame DELIBERATIVE QUALITIES OF NEWS FRAMES 22 featured at least one representative of the political periphery while this was true for more than half (58.9%) of the election news stories that carried no strategic game frame. [Figure 1 about here] The bivariate relationship between strategic game frames and the occurrence of uncivil discourse remained significant beyond controls in the initial log-linear analysis. A closer look at the upper right chart in Figure 1 reveals that the direction of the relationship conforms to the prediction formulated in our first hypothesis: Election stories providing a strategic game frame of the election were more likely to also contain at least one uncivil feature of televised discourse. As explained above, contestation frames present at least one response of at least two opposed speakers to one another by definition. We therefore used it as our indicator of responsiveness to test our second hypothesis. The bivariate effect was not necessary for the most parsimonious log-linear model to fit the data which indicates a weak association at best. A look at the lower left corner of Figure 1 gives nuance to this finding. It shows that strategyand game-framed election stories were somewhat less likely to also provide substantive contestation than stories without a strategic game component (39.5% vs. 48.2%). The second hypothesis thus receives some, but not overwhelming, support from our data. Finally, we were interested in how strategic game coverage was related to the extent of giving reasons for political action presented to the audience by either journalists or other speakers. The initial log-linear analysis indicated the presence of a relevant bivariate relation beyond controls. While almost seven out of ten election stories without strategic game reporting presented at least one substantive reason to their audiences this was true for less than half of the strategic-game-framed stories (69.6% vs. 47.1%, see Figure 1). Levels of reason-giving were thus lowered when reporting was occupied with the campaign as a strategic game which supports our third hypothesis. [Figure 2 about here] DELIBERATIVE QUALITIES OF NEWS FRAMES 23 We now turn to the hypotheses concerning the content correlates of contestation frames. Results of the bivariate analyses are given in Figure 2. Hypothesis four predicted that the use of contestation frames would come along with a greater inclusion of actors from the political periphery. However, the effect is not retained in the final log-linear model indicating the absence of an association. And indeed the bivariate analysis shows that the share of contestation stories featuring periphery speakers was only marginally higher than that of other election stories (45.9% vs. 41.6%). The inclusivity advantage of contestation news framing was thus negligible in this campaign and hypothesis four not supported. We went on to look at the association of contestation frames with the degree of incivility in televised political discourse. The results are staggering. In the log-linear analysis, contestation frames emerge as strongly related to the likelihood of uncivil transgressions in news reporting. Looking at the graph in the upper right corner of Figure 3, we find that while only fewer than half (45.5%) of all election news stories without a contestation frame present incivility of some sort to viewers, almost three fourths (73.0%) of contestation stories did this—an incivility gap of 27.5 percentage points that supports our fifth hypothesis. The last correlate of contestation frames we investigated was their impact on the levels of reason-giving in televised election discourse. The bivariate effect was retained in the final log-linear model, but it was only bordering significance beyond controls (see Table 1). The unclear status of the effect becomes clear upon inspection of the unpartialled bivariate association plotted in the lower left corner of Figure 2. The expected deliberative advantage of contestation frames with regard to reason-giving is readily visible: Two out of three news items that presented two or more opposing speakers responding to one another also presented at least one substantive reason for a political claim (67.6%). News items that did not employ a contestation frame provided a reason only 44.6% of the time. Our sixth and final hypothesis thus was supported by the data. DELIBERATIVE QUALITIES OF NEWS FRAMES 24 On the whole, the data support most of our hypotheses. The assumption that strategic game frames systematically restrict access to public discourse for peripheral actors was confirmed as were our expectations that they go along with lower levels of civility, responsiveness and reason-giving in public discussion during the run-up to an election. Our analyses of contestation frames in campaign coverage provide support to our hypotheses that such journalistic framing of elections entails relatively low levels of civility as well as a relatively high likelihood for substantive political reasons to make their way into the news. Contrary to expectation, however, journalists did not employ contestation frames to allow for a greater representation of peripheral voices in public pre-election discourse. Discussion The results of our study point to several missed opportunities for a more deliberative journalistic construction of the 2009 federal election campaign in Germany. In sum, the television news discourse of the election was dominated by a depiction of the campaign as a strategic game—a result that is consistent with prior findings (Genz et al., 2001; Schulz & Zeh, 2005, 2007). This campaign transpired against the backdrop of a stable lead in the polls by the ultimately successful conservative-liberal camp. Yet journalists relied on familiar techniques of dramatization and a focus on politics as a strategic, not substantive, contest to tell their story, with the negative implications for the degree of inclusiveness discussed above. Further, this study shows that, at least in this campaign, the negative assessment of strategic game coverage is justified if it is held against a deliberative model of democracy. Due to its systematic association with decreased levels of civility, responsiveness, and reason-giving the deliberative value of this generic framing device is low. Strategic game frames effectively curtailed opportunities for speakers appearing in the news to engage with policy issues and opposing positions, arguably doing little to inform spectators of the campaign process about substantive disagreements between the competing parties and candidates. We do not argue DELIBERATIVE QUALITIES OF NEWS FRAMES 25 that there is a deterministic relationship between strategic game frames and their negative byproducts we observed in this campaign. But we do suggest that television journalists be cognizant and try to minimize them in the future. The picture emerging for contestation frames is more complex. First off, we established that the inclusiveness and responsiveness coming along with substantive televised contestation benefit large-scale democratic deliberation. However, less than half of all news stories we analyzed contained an explicit contrast or confrontation of available political options which leaves some room for improvement in the coverage of future campaigns. The results of our empirical normative assessment of contestation frames are ambiguous with regard to their relationship to other deliberative qualities. We found that contestation frames were associated with more uncivil televised discussion, which is certainly disquieting. At least in election campaign coverage, the presentation of substantive political disagreement in television news seems to spawn some unproductive clamor. However, we were also able to show that journalistic decisions to frame the election campaign as substantive contestation meant that the news audience was more likely to learn about substantive reasons for the political positions of candidates and parties. In this regard, contestation frames put audience members in a better position to deliberate the vote choices put in front of them in the election. Contestation framing is obviously related to television’s particular potential to expose its viewers to political disagreement (see Goldman & Mutz, 2011; Mutz & Martin, 2001). One reading of our findings, then, has us cautioned not to overstate or generalize the salutary effects of this potential, unlike it is often done in the literature (e.g., McLeod et al., 1999; Price, Cappella, & Nir, 2002; Scheufele, Hardy, Brossard, Waismel-Manor, & Nisbet, 2006): In the specific case we studied here at least, the presentation of substantive disagreement was not positively related to other deliberative qualities of television news across the board. Therefore, it would certainly be worthwhile for journalists to try to look beyond the mere presentation of disagreement over issues among political elites in an effort to provide a deeper DELIBERATIVE QUALITIES OF NEWS FRAMES 26 and more dispassionate understanding of the programmatic and ideological fissures that lie