Re-emerging Discourse:
Assessing the German Press Coverage of the Census Debates 1987 and 2011
Master Thesis
Research Master Communication Science Graduate School of Communication
University of Amsterdam
Marco Lünich 10393684
University of Amsterdam Supervisor: dr. Richard van der Wurff
Abstract
This study assessed the working of the public sphere in the recurring census debate in Germany. With regard to developments in the media, society, and politics potential differences were investigated between the debates in 1987 and 2011. A content analysis of press coverage was conducted for different types of press (i.e., quality newspapers, the alternative press, and the tabloids). Indicators were based on the normative criteria of democratic public sphere models. Results show that the overall coverage of the census decreased significantly in 2011, even
though privacy concerns are of ongoing importance for the citizenry. The findings did not show a comprehensive trend for the mediated public sphere. The analysis of the process of discourse, the styles of speech, and the discussion of solutions suggest that the media coverage performed worse in 2011. However, the latter debate showed an improvement towards more balance of speakers, as journalists and political periphery speakers had a higher share of arguments. Overall, the diversity of speakers is similar compared with previous discourses and there was greater empowerment and more responsiveness in the arguments than in prior studies. Such findings might be attributed to the less intimate, but highly controversial topic at hand. Furthermore, the results suggest ambiguous developments between the tabloid press and outlets from the quality and alternative press in terms of incivility and the justification of arguments. Contrary to the discourse in 1987, tabloids reported more neutral but also more superficial in the latter debate. In general, the public sphere fell short of its normative ideals. Still, the diversity of speakers and the discussion of solutions show a considerable participation and solution orientation of the census debate. Consequently, future research of other issues may put an emphasis on the evaluation of speakers and the content of their arguments in different arenas of the public sphere.
Keywords: Public Sphere, Deliberation, Content Analysis, Census Debate, Discussion of Problems and Solutions
Re-emerging Discourse: Assessing the German Press Coverage of the Census Debates 1987 and 2011
A national population census is a recurring event in developed countries. In a census the state aims at gathering certain data about its people such as the general population count, age distribution, and living circumstances. In the European Union a census is required to take place every ten years in each member state. However, different countries use different methods to compile the national statistics. For instance, in Germany the last two major censuses took place in 1987 and 2011. They were based on responses from a majority of the population, as the government decided not to use existing databases to assemble the data, but to directly survey the people (or a randomly chosen one-third of the population and all house owners in 2011,
respectively). The censuses thus personally concerned all 80 million inhabitants. Considerable controversy was raised because a census is accompanied with privacy concerns and issues such as the costs, the objectives, and usefulness as well as the representativeness of the inquiry and its method.
As a consequence, there are nearly identical arguments for and against a census in 1987 and 2011 that resulted in a societal debate in each year. In this study, the focus lies on a complete assessment of the quality of a discourse in the media that plays out over a limited period of time of media coverage around the date of the census. This discourse is picked up again after several years have passed with no explicit discussion about a census in between. Thus, the approach at hand falls in line with Wessler (2008), who calls for more research on discourse in the media and the public sphere in a comparative perspective.
Consequently, this study aims at contrasting substantial changes and their magnitude with respect to a well-defined discussion topic in the mediated public sphere. In this regard, the debate
differs from studies that arbitrarily investigated issues that had no specific time frame. For instance, Ferree, Gamson, Gerhards, and Rucht (2002a, 2002b) analyzed the abortion discourse in the USA and Germany, while Wessler (1999) addressed the German press coverage on drugs over the years, adding a long-term process perspective on the workings of the discourse.
To a certain extent, those previous studies addressed over-arching changes in the landscape of journalism, the media, society, and politics. Their research objects, however, are broad and involve a lot of different multi-dimensional topics (e.g., illegal vs. legal drugs, medical vs. recreational drugs). Hence, they differ from the census debate with respect to the topical and temporal specifics of the discourse in the public sphere.
The Censuses and the Public Sphere
According to Marcinkowski there are certain conceptions defining the general public sphere. To begin with, “the term refers to the institutionalization of a realm of social life for the exchange of information and opinions” (2008, p. 4041) that is described above with the societal discussion of the pro and contra concerning the census. To be more precise, it “is the act of free citizens gathering together for debate in order to achieve a rational regulation of public affairs” (p. 4041) such as the execution of the census. The public sphere is seen as “a political process in which common cause is built through the search for solutions to problems initially encountered as private concerns” (Johnson, 2009, p. 1, emphasis added). Eventually, the public sphere plays out in different “spaces in which public communication regularly takes place: streets and squares, formal and informal gathering places, publicly meeting institutions of the political system, and the arena of the mass media” (Marcinkowski, 2008, p. 4041f.).
This study is particularly interested in how the public debate on the censuses in 1987 and 2011 played out in the mass media. For most people, mass media outlets are primary sources to
learn about information and opinions concerning critical societal issues (McQuail, 2010). Newspapers especially are seen as significant providers for daily information concerning public affairs and recent developments (Ghanem, 2008). Despite steadily decreasing circulation numbers, they are still seen as relevant for the formation of public opinion and are a crucial source of information for decision makers in politics and the economy (Schenk & Mangold, 2011). Thus, this study investigates the quality of debate with special regard to the media coverage of the major press outlets from different types of press that exist in Western media systems such as Germany (McQuail, 2010; Hallin & Mancini, 2000).
To assess the quality of the debate, we will first give an overview of the normative conceptions of different democratic models of the public sphere. Their most important criteria for evaluation refer to the participants, the process, and the styles of speech present in a public discourse and are extended by the public discussion of solutions. Based on such normative criteria, we will introduce indicators from the literature that help to address and measure them. We will then turn to an analysis of the press coverage in 1987 and 2011 that tries to answer the following research question:
How did the public sphere in Germany operate in the public debate on the national censuses in 1987 and 2011?
In answering this question this study aims at using a comparative perspective on the public discourse in the media about a recurring issue. Addressing the different qualities of this mediated public discourse in a democratic society in depth might lead to an understanding of how societies and their media manage controversial and complex issues over time. Insights will be gained of what a debate looks like if it reappears after a certain time interval without
serve as an indication for other substantial issues that nations occasionally face, such as social reforms or dangerous technical developments, but also war and natural disasters. In such instances it is important to understand how discourse in the public sphere plays out, so all stakeholders in a democratic society can make informed decisions and take well-conceived actions.
Theoretical Models of the Public Sphere and Their Normative Criteria
Jarren and Donges (2002) suggest that any empirical-analytic conceptions of the public sphere and respective indicators are necessarily based on a variety of normative assumptions constructed on distinct concepts of democracy. Thus, differences in the normative assumptions of the theories of the public sphere concerning the media and journalism have to be thoroughly discussed before any empirical investigation can begin.
Strömbäck (2005) reviews the normative implications for journalism based on different models of democracy. As his procedural democracy model has no normative implications for journalism, only the competitive model, the participatory model, and the deliberative model are discussed. Those models have specific demands for journalism and the public sphere with regard to how the discourse should look. In particular, Strömbäck places significant emphasis on how the stakeholders in the mediated debate should behave and be involved with the discourse. He also addresses the mechanisms of the process and style of debate in the media. This is very similar to the guiding questions of Ferree et al. (2002a) in their respective discussion of different models of the public sphere (see also Wessler, 2008):
• “Who participates? • In what sort of process?
The distinct models of the public sphere have different normative perspectives and consequently different demands and answers to the questions above. Based on the similar demands of Strömbäck and Ferree and colleagues the normative criteria for the different models will now be introduced and possible indicators from previous research are connected to their concepts. Eventually, the performance of the different concepts and indicators might suggest which model of the public sphere can describe the public debate of the censuses in 1987 and 2011.
The Participants in the Public Debate
With regard to all of the different models of the public sphere an important question is usually who participates and contributes to a discussion. Models such as the representative liberal model of Ferree et al. (2002a) and Strömbäck´s concept of competitive democracy (2005) see it as sufficient if the diversity of speakers appropriately reflects the composition of political and ideological perspectives as found in parliament to guarantee a free marketplace of different ideas. This way, all relevant societal interests are represented via party speakers that voice their interest on behalf of the voters. The voter consequently does not participate in the discussion but can make an enlightened decision at election time based on the media´s ability to transparently document and report on issues. In this conception, the focus lies predominantly on the expertise of the speaker that qualifies him or her to participate and contribute to the discussion.
In contrast, other models require popular inclusion and thus place more emphasis on the participatory element of democracy and its citizens. “The common thread in participatory liberal theories is the desirability of maximizing the participation of citizens in the public decisions that affect their lives” (Ferree et al., 2002a, p. 295). More specifically, Strömbäck states for the participatory model of democracy that “the role of citizens is active. In order for democracy to
thrive, people must be actively engaged in public life and part of political processes and decision making” (2005, p. 340).
With regard to who gets to speak in the deliberative model of the public sphere, Wessler calls the assessment of the diversity of speakers “the central yardstick for assessing public deliberation” (2008, p.1). Thus, deliberative democracy also calls for popular inclusion. The measurement of the diversity of speakers must consequently take into account a broad range of possible participants. For instance, it should include speakers from the political center (e.g., from government, administration, and judiciary) and the periphery of the political system (e.g., civic associations, ordinary citizens, artists, etc.) to distinguish actors with institutional power from actors with less of such power (Wessler, 2008; Ferree et al., 2002b).
The Process of Public Discourse
While the assessment of the diversity of speakers is an important aspect of the public sphere and its normative conceptions, a second relevant perspective for the evaluation of the public sphere is in “what sort of process” (Ferree et al., 2002a, p. 316) the discourse is held. To begin with, the process of discussion can foster empowerment. Ferree et al. define empowerment as “the extent to which the mass media provide support and encouragement of a sense that, by acting together, grass-roots constituencies can influence the policies and conditions that affect their daily life" (2002b, p. 236). This resembles Strömbäck´s call for journalism to “mobilize the citizens’ interest, engagement and participation in public life” (2005, p. 341).
Empowerment provides people with the opportunity and encouragement to speak up in the media and is an important normative criterion in participatory models of the public sphere. With regard to the opportunity for citizens to speak up on issues and different from the
representative model, their appearance can give a first hint at the empowerment of ordinary citizens in the media.
While the aforementioned indicators refer to their respective concept´s presence in the article, the following indicators are present in the arguments made by individual speakers to give more detailed insights into the discussion (Wessler, 2008; Ferree et al., 2002b). For instance, presented arguments should justify how society and its citizens profit in order to help the citizens understand. Steenbergen, Bächtiger, Spörndli, and Steiner (2003) address this as the content of a justification. They propose to base the reasoning for or against an argument on either specific group interest or the common interest of citizens. Furthermore, empowerment is seen as the encouragement to show citizens that they have the capability to bring about social change and to show ordinary citizens how to act to overcome a problem. For instance, arguments can explain that the solution to a problem can be brought about on the individual level of the citizen (Kim & Willis, 2007). Beyond that, the speaker can encourage or even demand that the citizens perform a specific action in an argument (Kensicki, 2004; Semetko & Valkenburg, 2000).
While empowerment is important in participatory conceptions of the public sphere, discursive or deliberative conceptions place a high value on the deliberative quality of the discussion. Not only is it important that citizens have their say in the debate, but that the sharing of their ideas plays out in a considerate, well-reasoned, and mutually recognized manner. Even though Ferree et al. (2002a) refer to the process of discourse as deliberative most aspects of the deliberative quality of discourse touch upon the style of utterances and consequently how ideas should be presented.
The Styles of Public Discourse
With regard to the question of how ideas in public democratic discourse should be presented there are different styles of speech that the different models of the public sphere refer to. While the norm in the representative liberal model by Ferree et al. (2002a) is a detached and civil style their participatory liberal model recognizes different kinds of styles that might also include incivility – that is the absence of certain standards of politeness and courtesy
(Papacharissi, 2004). The reason is that “polemical speech acts or symbols that capture the emotional loading of public issues as well as their cognitive content can play a very important mobilizing role” (Ferree et al., 2002a, p. 298).
According to the deliberative quality of a discourse, which is seen as a crucial normative criterion in the discursive model, there are additional indicators of different deliberative
dimensions that refer to the style of speech (Ferree et al., 2002a). Rinke, Wessler, Löb, and Weinmann (2013) single out three major aspects that are based on theoretical considerations in the literature of mediated deliberation (Gastil, 2008; Wessler, 2008).
Deliberation starts with the need for mutual respect that includes especially the presence of civility mentioned above. Furthermore, based on Bennett et al. (2004), deliberation requires responsiveness, which “involves recognizing, incorporating, and rebutting the arguments of others” (Ferree et al., 2002a, p. 306). On the one hand, this means that the discussion should include speakers with opposing views and of different tone – positive, neutral or negative – towards the issue. On the other hand, it is important that their arguments refer to each other in a dialogue (Ferree et al., 2002a; Rinke et al., 2013; Wessler, 2008). This dialogue involves the type of arguments of the speaker, which can either be about (1) the census in general, (2) another
speaker without addressing his or her argument, or (3) about another speaker´s argument. It also involves rebuttals, i.e., a speaker refers to another speaker´s argument and tries to refute it.
Eventually, a necessary requirement of deliberation is reason-giving (Rinke et al., 2013; Wessler, 2008). This is addressed by the extent to which one justifies one´s arguments.
Steenbergen et al. (2003) introduce the level of justification of an argument as a necessary part of reason-giving that assesses the extent to which an argument is justified by the speakers. This ranges from giving no reference why a speaker regards his argument as justified, over arguments that give a broad general justification, up to specified justifications, where an argument contains a detailed reasoning for its justification.
The presented normative criteria and their indicators for the public discourse help to assess the state of the mediated public sphere. Expecting potential differences in terms of the indicators in between both census debates, we turn to closer considerations of the potential developments in society and the media.
The Census Debate in the Public Sphere
While there are no specific considerations that the societal situation in Germany has dramatically changed between the census debate in 1987 and 2011 (i.e., a system change such as in the Soviet countries), there are some likely differences because of general developments in the media that need to be addressed. Concerning the question who is participating in the public sphere, mass communication research already has a long tradition of communicator research analyzing the different stakeholders in public communication. A recurring theme here is that of elite dominance, in which a small elite of political and societal actors receives the biggest share of attention and opportunities to speak (Norris, 2000; Schudson, 2003; Tuchman, 1978). While the participatory and discursive conceptions of the public sphere call for popular inclusion,
Wessler (2008) mentions two reasons why this is most likely not a realistic case and how it could even be detrimental to public discussion: On the one hand there is the limited carrying capacity of the media and on the other hand “it would also undermine the capacity of public deliberation to produce new, interesting, relevant, and competent ideas rather than just any kinds of ideas” (2008, p. 3).
The development of increasing personalization of politics (McAllister, 2007) and the mediatization of politics (Mazzoleni & Schulz, 1999) might add to the effect of elite dominance, but also concern the presentation of ideas. Wessler (2008) argues that the media´s focus on personalization, i.e., the focus on (powerful or famous) persons might restrict the likelihood for deliberation, which rather concerns the discourse of issues. Accordingly, the mediatization of politics requests speakers to adhere to a media logic. Attention is given to the speaker who is able to cater to the requirements of the media and their preferred news values such as negativity and unexpectedness (Blumler & Kavanagh, 1999; Harcup & O’Neill, 2001). It might thus be assumed that professional speakers from the political center are better prepared than ordinary citizens to fulfill such requirements of the media – be it because they are better trained or more experienced for such situations (Negrine & Lilleker, 2002). Consequently, it could be expected that the diversity of speakers is imbalanced, tending to favor speakers from the political center and a certain kind of presentation of ideas that devalues reasoned and in-depth discussion of arguments. It might also be expected that the higher the focus on a limited set of speakers (especially from the political center) the lower the deliberative quality of the discourse as there would initially be fewer chances for different speakers to engage in mutual discourse (Bennett et al., 2004).
Prior empirical research suggests that the state of the diversity of speakers, of
empowerment, and of deliberation falls short of its theoretically imagined ideal (Ferree et al. 2002b, Bennett et al., 2004). There is no argument to assume that this has changed for the better, but rather to have worsened based on the aforementioned considerations.
RQ1: To what extent is there a difference between the diversity of speakers in the census debate in Germany in 1987 and 2011?
RQ2: To what extent is there a difference between the sort of process of the census debate in Germany in 1987 and 2011?
While the two former questions address the participants and process of discourse in the public sphere, there are also considerations concerning the style of debate. With regard to the media system and its latest development introduced above there are also potential influences on said styles. For instance, research that investigates TV news and sound bites of speakers finds less and less time dedicated to the arguments of politicians and other speakers (Esser, 2008; Hallin, 1992). Accordingly, it might be suspected that there is also less room for dialogue in newspapers. This might also be associated with the general trend towards horse race journalism (Iyengar, Norpoth, & Hahn, 2004). This trend where the media focuses on a candidate´s
performance and not on the public issue might not be as severe as sometimes imagined (Sigelman & Bullock, 1991). Still, it might negatively influence the amount of space given to dialogue.
As far as other important aspects of the style of debate, there might be considerable differences between the different types of press (McQuail, 2010). For instance, one could also assume that especially tabloid newspapers paid higher attention to utterances that were uncivil. After all, tabloids are known for favoring sensationalism (Örnebring & Jönsson, 2004). Then
again, the market for newspaper journalism significantly changed in the last years because of dwindling circulation numbers, which resulted in a newspaper crisis (Siles & Boczkowski, 2012). Consequently, there could be increased competitiveness between newspapers altogether that diminished differences in reporting news between the distinct press types (Blumler & Kavanagh, 1999). This might eventually also lead to less variation regarding the discussion in 2011 in terms of style of speech of the reported arguments. Based on the developments in the media system described above, there might be differences concerning the styles of speech in the debate.
RQ3: To what extent is there a difference in the styles of speech in the census debate in Germany in 1987 and 2011 and in the different newspapers?
The Discussion of Solutions
When we look at the conceptual components of the models of the public sphere and the discussed indicators, it becomes obvious that they are mostly referring to the procedure and style of the discourse. None of the indicators directly discuss the content of the arguments and the necessity to address concrete solutions or at least the process of deriving to them. While
questions on speakers and the process and style of their discourse are important, the discussion of solutions is also a crucial part of the function of a democratic media system from a normative perspective. All normative models aim for a working democracy with outcomes based on different forms of closure for which solutions need to be found (Ferree et al., 2002a). For instance, Gastil states one of the core requirements for the media with regard to mediated deliberation: “Present the broadest possible range of solutions to problems, including nongovernmental and unpopular ones” (2008, p. 52). Even the least demanding form of
procedural democracy supports the marketplace of ideas and thus the discussion of various solutions (Strömbäck, 2005).
Consequently, solutions to overcome issues fit well into the normative criteria of the mediated public discourse, since media theory sees the mass media for better or worse as important agents of social change (McQuail, 2010). More specifically, when discussing the normative implications for journalism Strömbäck sees the following requirement for journalists: “They should try to find the information necessary to understand a particular issue, and be able to link factual conditions, underlying moral values, and proposed solutions as to what
consequences are likely” (2005, p. 337, emphasis added). He demands that politics is not framed as a strategic game such as in horse race journalism (Broh, 1980; Rinke et al., 2013), but that “journalism should frame politics as a continuous process of finding solutions to common problems” (Strömbäck, 2005, p. 341).
This study consequently also wants to emphasize the content of the discourse and the empirical assessment of the discussion of solutions in particular. Similar to the complex latent constructs introduced above (e.g., empowerment, deliberative quality), the discussion of solutions possesses different dimensions and consequently different indicators. Thus, previous empirical approaches on the discussion of solutions are introduced to give an idea of what is known about the mediated discussion of solutions and what aspects it should involve. Empirical Approaches to the Discussion of Solutions
Concerning the discussion of solutions, research results so far are rather daunting. In general, research finds the discussion of solutions in the media to be somewhat rare (Trumbo, 1996; Eickelkamp, 2011). Kensicki (2004) finds that US media coverage often does not mention and discuss solutions when it comes to topics such as pollution, poverty, and incarceration. Kim
and Willis (2007) find mixed results in their investigation on news framing of obesity in the US. While solutions on the societal level received greater coverage over time, solutions on the individual level decreased. Based on such previous findings there are potential differences between the two debates in 1987 and 2011 in terms of the discussion of solutions.
RQ4: To what extent are solutions discussed in the census debate in Germany in 1987 and 2011? Empirical communication research follows different approaches in content analyses to investigate the aforementioned aspects of discussing problems and suggesting solutions and remedies. Kensicki (2004) suggests that readers receive a general impression concerning critical societal problems and their solutions in the media coverage. Thus, a multi-dimensional approach to the discussion of a solution can give a better idea of how readers learn about potential
improvements and how and by whom those improvements might be achieved. A frequently found basic approach initially asks if a story suggests whether there is a solution to a problem (Kensicki, 2004; Semetko & Valkenburg, 2000). Some approaches go beyond this evaluation and measure the solution discussion in more detail (Zhou & Moy, 2007; Kim & Willis, 2007).
To begin with an argument can refer to a problem or a problematic issue. This is not necessarily the case but most often serves as a starting point to introduce solutions for said problem. The leading question assessed in prior studies (Trumbo, 1996; Saguy & Almeling, 2008) is: “Does the argument address something as problematic?”. In a next step it is important whether the argument sees room for improvement or if there is already a particular solution to the problem. This is addressed by the indicator question “Does the argument suggest that there is an improvement or a solution to the problem?” that is introduced in various studies (Semetko & Valkenburg, 2000; Trumbo, 1996; Eickelkamp, 2011; Kensicki, 2004). Research by Saguy and Almeling (2008) and Kim and Willis (2007) furthermore distinguishes between the levels on
which a solution can be achieved. In his or her argument the speaker can see the solution either on the individual level, the level of the group/an institution, or on the societal level. The indicator question consequently asks, “On which level can the improvement be brought about?”.
Subsequently, it is of interest if the speaker mentions specific measures or proposed actions that might be able to alleviate and improve problematic situations (Zhou & Moy, 2007, Kim & Willis, 2007). It is thus of interest, if the argument addresses a specific treatment, action or measure that aims at improving the problematic solution. While mentioning a specific action is one possibility other studies specifically evaluate whether there is a call to action. The speaker can suggest or even demand a specific course of action from someone. Based on prior studies, this can be assessed by asking “Does the argument make an explicit call or suggestion for action to improve the problematic situation?” (Zhou & Moy, 2007; Kensicki, 2004; Semetko &
Valkenburg, 2000).
All in all, those studies give a first idea of how to approach different aspects of the discussion of solutions. However, each of them addressed distinct dimensions individually. Most of them did so on the article level only, while the present study aims at investigating the five aspects of the discussion of solutions also on the argument level.
Method and Data
The research questions were investigated by conducting a standardized quantitative media content analysis of German press coverage in 1987 and 2011.
Content Analysis Time Frame and Sample
The time frame for each census debate has been selected based on a similar study conducted by Hagen (1993) that based its analysis on a pre-study. The latter found that the relevant media coverage took place roughly five months prior to and one month after the
respective census date (25th of May, 1987 and 9th of May, 2011). Consequently, this time frame (1st of January until 30th of June) was adopted for this study´s media coverage in 1987 and 2011, respectively, to achieve comparability.
Four newspapers were selected that reflect a broad variety of influential German press outlets, ranging from established quality newspapers over yellow press tabloids and alternative media: The Frankfurter Allgemeine Zeitung (FAZ) is a center-right newspaper (Hagen, 1993) with a mostly subscription-based circulation of 355.000 units in 1987 and 360.000 in 2011. The Süddeutsche Zeitung (SZ) is a center-left newspaper (Hagen, 1993) with a mostly subscription-based circulation of 370.000 in 1987 and 430.000 units in 2011. The Bild Zeitung is a daily tabloid newspaper with a circulation of roughly 4.800.000 units in 1987 and 2.900.000 units in 2011. Taz – die tageszeitung is the dominant alternative left-wing newspaper with a mostly subscription-based circulation of 84.000 units in 1987 and 18.500 units in 20111.
All articles by the four newspapers that addressed the census and were issued within the time frame were collected. Due to the unavailability of the newspapers (except the taz) in digital format for 1987, archived newspapers were consulted and screenshots of relevant pages and articles were taken. It is possible that a few articles that mentioned the census might have been accidently but unsystematically missed because all material needed to be manually scanned on microfilm. For the press coverage in 2011, all articles were downloaded from digital databases or bought directly from the publishing company. Due to the abundance of articles in 1987 for three of the four newspapers (taz, FAZ, and SZ) a random sample of exactly one-third of the articles was drawn and coded.
1
Despite its decline in hard copy circulation the taz still plays a considerable role in the German media landscape, which can also be seen in the 5 million unique visits to their online outlet in May 2011 where the taz publishes all of its print articles free of charge.
Coding Procedure and Pre-Test
An extensive pre-testing phase ensured sufficient reliability of the codebook. In a first step, after a detailed discussion of the codebook and coder training, two German native speakers coded a sample of articles that addressed the census but did not belong to the sampling
population for the eventual study (such as local reports from other newspapers). Following this first coding procedure the problematic categories were adjusted, coding categories were condensed, and coding instructions were clarified and/or extended.
In a second step, coding 32 arguments from randomly selected articles, the same two coders revealed satisfactory inter-coder reliability based on Krippendorf´s Alpha coefficients. All Alpha values of the measures are reported below. The reliability values were obtained by opting for a “hard” coding approach whereby coders only coded a category if indicators were manifest in the argument (Früh, 2011; Berelson, 1952). Thus, it could be that sometimes a meaning in-between the lines was present, but was not coded because of potential ambiguous interpretation by the coders. In the experience of the coders there was hardly any ambiguity present in the press coverage, as the focus of the present study was on political news and information. Still, the data at hand might underestimate certain categories for the sake of good inter-coder reliability. Measures
The codebook of the content analysis addressed the concepts and constructs that were introduced in the theoretical section. Most measures were consequently based on previous studies as indicated above. The recording units were on the one hand the whole article and on the other hand each unique argument made in the article (Riffe, Lacy, & Fico, 2006). Initially, the argument and its components need to be defined: Adapted from Ferree et al. (2002b), an argument is a directly repeated or indirectly referenced utterance (i.e., speaking act) by an
identifiable actor over a distinct subject within an article. There are consequently three criteria for demarcation that distinguish an argument (1) from other arguments, (2) from arguments from other actors, and (3) from general statements that are not counted as arguments at all. Arguments are distinguished from each other by their thematic reference to a specific topic that they
evaluate. If the topic changes there is a new argument. There also needs to be a speaker that gives the argument. This speaker is either directly giving the argument (Helmut Kohl says: “XYZ”) or the author of the article is citing the speaking act (Helmut Kohl said XYZ). It is also possible that an argument is made, but no speaker is identified. In such a case the argument is attributed to the journalist (e.g., “XYZ is not good”). A new argument is coded when the speaker changes. Mere statements that refer to general actions and are not directly or indirectly repeated (evaluative) speaking acts by any speaker (e.g., “Kohl arrived in Bonn.”) are no arguments. Formal Categories on the Article Level
The first categories from the codebook addressed the formal features of the respective article and were coded on the article level. Those variables were the article ID, the date of publication, and the outlet.
Categories on the Argument Level
The most important unit of analysis was the argument. Categories coded on the level of the argument were the type of speaker2, who made the argument (Krippendorf´s α = .959). A list of potential speakers that reflects a broad variety of different societal stakeholders was adopted from studies conducted by Wessler (2008, 1999) and Rinke et al. (2013). Their measure does not just cover the requested broad range of possible speakers but also differentiates between the political center and the periphery to distinguish between actors with and without political power.
The tone of the argument (α = .946) concerning the census was adapted from Schuck and de Vreese´s study on public support for the EU enlargement (2006). With this measure the speaker´s argument can be evaluated as either positive, negative, or balanced/neutral (“indicating no dominant evaluative direction”, p. 14) towards the census.
The type of argument (α = .938) was based on the distinction by Ferree et al. (2002b) of what the argument referred to. It could address the census itself, another speaker or another speaker´s argument. This distinction was made to distinguish statements that addressed another speaker or his or her argument and could consequently contain incivility and responsiveness. For instance, an indicator for rebuttals (Ferree et al., 2002b) is whether the argument rejects or approves another speaker’s argument (α = .937). Incivility (α = .933) can be measured by several indicator questions adopted from Papacharissi (2004) that consider different dimensions of incivility such as impoliteness and mutual respect (see also Rinke et al., 2013; Zhang, Cao, & Tran, 2012).
Based on Steenbergen et al. (2003) the measures level of justification (α = .904) and the content of the justification (α = .932) were used. The former evaluated the argument as having either no justification at all, or having a general or a specified justification. The latter measures whether the argument is based on a specific group interest or the common interest of the people.
Eventually the five questions related to the discussion of solutions were introduced that were discussed above. First, it was coded if the argument discussed something as problematic (α = .848). In a next step it was assessed if the argument mentioned at least the possibility of an improvement or a specific solution (α = .875) of the situation. Supposing this was the case, it had to be assessed on which level (α = .896) the argument locates the possible improvement
next category assessed if the argument contained specific measure or action (α = .938) that might bring about the improvement. Another category assessed whether a call to action (α = .798) was made or if a specific measurement or course of action was explicitly suggested or demanded.
Method of Analysis
To answer the four research questions, the findings of the respective indicators on the argument level will be compared for 1987 and 2011 on the basis of Chi-Square difference tests. Subsequently, for each research question we will compare the findings over all newspapers between the two debates. As research question three concerns differences in the style of speech in the media coverage in the different types of press there will be an additional investigation of the arguments in the newspapers from the quality press (FAZ & SZ), the alternative press (taz), and the tabloid press (Bild) between the two debates.
Results
As discussed above a random sample of one-third of all articles was drawn from three of the four newspapers for the 1987 period. This resulted in 84 articles in the taz that referenced the keyword “census.” The samples of the FAZ and SZ consisted of 65 and 49 coded articles, respectively. Only for the Bild were all 67 articles from 1987 coded. On the contrary, in 2011 all articles that addressed the census were coded. There were 35 articles coded for the taz, 10 for the FAZ, 14 for the SZ, and 9 articles coded for the Bild. Looking at the number of articles in both census debates there were altogether fewer articles in 2011 (N= 68) than in the 1987 sample (n = 263). On average there were 3.7 arguments per article in 1987 and 5.2 arguments per article in
20113. The data analysis was eventually performed for n = 969 arguments in 1987 and N = 356 arguments in 2011.
Diversity of Speakers
To answer the first research question that asks whether there is a difference in the diversity of speakers the results show there are some significant changes between the two debates. Overall in 1987 almost half of the speakers were from the political center (see Table 1). This share went down to less than one-third in 2011. The share of speakers from the periphery rose at the same time from one-quarter to one-third. Journalists represented the biggest change, increasing their share from 22.4% to 35.7%. This difference in the distribution of the speaker groups between the two debates is highly significant (Χ2(3, N=1325) = 46.06, p < .001).
Table 1
Amount of arguments of different speakers
1987 2011
Type of speaker No. of
arguments Relative share No. of arguments Relative share
Political Center total 474 48.9% 107 30.1%
Political Periphery total 243 25.1% 115 32.3%
Journalists (Internal) 217 22.4% 127 35.7%
Other 35 3.6% 7 2.0%
N 969 356
Note. χ2(3, N=1325) = 46.06, p < .001; Cramér’s V = .186
Eventually, in 2011 the share of arguments was more equally distributed between the speakers than in 1987, with each group making roughly one-third of the arguments.
The Process of Discourse
The second research question asks whether there are differences in the process of discourse between the years. Here the focus is on the indicators that evaluate the concept of
3 In 33 articles there was no argument found. Those were short reports that just addressed facts and events without any quoted statements and evaluation.
empowerment. Those indicators are the appearance of citizens and social movements as speakers, as well as a justification based on the interest of the people and the individual level where a solution can be brought about. Also of interest is whether a speaker who sees the solution on the individual level of citizens calls them to action in his or her argument. Assessing the prevalence of speakers that are citizens or social movements the results show that taken together their share of arguments did significantly differ between 2011 and 1987 (χ2(1, N=1325) = 6.33, p = .012). In 2011 the share of citizens and social movements was 13.2 percent of all arguments compared to 8.6 percent in 1987.
Furthermore, the justification of arguments can contain a reference to the interest of the people, i.e., the common interest. The results show that 2.7 percent of all arguments in 1987 and 4.5 percent of arguments in 2011 contained a reference to the interest of the people. In 5 percent of all arguments in 1987 and 7 percent of the cases in 2011 the justification was based on a specific group interest. Arguments that were not justified by any interest amounted to 92.4 percent in 1987 and 88.5 percent of all arguments in 2011. There was no significant difference of referenced interests as the content of justification between the two points in time (χ2(2, N=1325) = 5.15, p = .076).
Moreover, an argument can suggest that a solution can be brought about on the individual level of the citizen. In 1987 10.4 percent of all arguments referred to this level of the solution. In 2011 this share amounted to 11.2 percent of the arguments. This difference was nonsignificant and thus the content of justification stayed constant between both points of time (χ2(1, N=1325) = 0.18, p = .671). In a next step it is interesting how often a speaker, who sees the solution on an individual level, is explicitly encouraging or appealing to citizens to take action. In 1987, 43.6 percent of arguments that see the solution on the individual level make a call to action (0.5 % of
all arguments in 1987). In four out of ten of such arguments in 2011 (0.5 % of all arguments in 2011), the speakers are calling the citizens to action. There is no difference in the share of such arguments between both points in time (χ2(1, N=141) = 0.15, p = .700).
Different Types of Arguments in the Census Debates
Before discussing the styles of speech in the census debates, the results for the different types of arguments in the debate are addressed, as some of the indicators only apply to certain types of those arguments. The distribution shows a significant difference between the types of given arguments in 1987 and 2011 (see Table 2). The majority of arguments at both points in time did directly address the census and did not refer to any other speakers. In roughly 15 percent of the cases in 1987 and 2011 the argument did only address another speaker. Reference to another speaker´s argument was prevalent in one out of four arguments in 1987. This number declined to 13 percent of all arguments in 2011.
Table 2
Distribution of the types of arguments in the newspapers in 1987 and 2011
1987 2011
Type of Argument concerns … No. of arguments Relative share No. of arguments Relative share … the census. 566 58.4% 261 73.3% … another speaker. 143 14.8% 48 13.5% … another speaker´s argument. 260 26.8% 47 13.2% Total 969 356 Note. χ2(2, N=1325) = 30.43, p < .001; Cramér’s V = .152
The Style of Speech in the Census Debate
On the one hand, the third research question asked whether the styles of speech differed between the two census debates. On the other hand, it addresses the difference in styles of speech between the distinct newspapers, especially as it concerns their type of press. Consequently, the results of different indicators such as the incivility of the debate and the deliberative quality are reported. First, the difference between the two points in time is addressed, and then the results for the different newspapers are evaluated.
Differences Between the Two Census Debates
With regard to uncivil arguments, the results show that of the arguments referring to another speaker or to his or her argument 13.2 percent were uncivil in 1987 and 8.4 percent were uncivil in 2011. This difference between the two years was non-significant χ2(1, N=498) = 1.60, p = .206). On the contrary, the tonality of the arguments between the two years differed
significantly (see Table 3). Table 3
Tonality of the arguments in 1987 and 2011
1987 2011
Tone of argument No. of arguments Relative share No. of arguments Relative share
Negative 158 16.3% 80 22.5%
Neutral/Balanced 679 70.1% 233 65.4%
Positive 132 13.6% 43 12.1 %
Total 969 356
The majority of arguments in 1987 and 2011 were neutral or balanced. There were more negative than positive arguments in both debates. However, in 2011 even more of the arguments were negative while relatively fewer arguments were positive.
In 1987, of all arguments that concerned another speaker´s argument (n = 260) 63.8 percent were rebuttals and 11.2 percent did approve of what another speaker said. One-quarter of such arguments did neither approve nor rebut the other speaker. In 2011 (n = 47), the results show that 76.6 percent of all arguments that addressed another speaker´s argument were rebuttals and 10.6 percent approved it while 12.8 percent of the arguments contained neither rebuttal nor approval. There is no significant difference between the two years (χ2(2, N=307) = 3.57, p = .168).
Concerning the justification of arguments, the results show that in 1987, 1.5 percent of all arguments were not justified at all while that share amounted to 21.9 percent of all arguments in 2011. A general justification was given to 60.1 percent of the arguments in 1987 and to 50 percent of the arguments in 2011. In 1987, 38.4 percent and in 2011, 28.1 percent of all articles received a specified justification. The findings show that in 2011 significantly less general justification and less specified justification were given to the arguments than in 1987 (χ2(2, N=1325) = 166.14, p < .001).
Differences Between the Distinct Types of Press
As far as the differences between the types of press, the results for 1987 show that 12.7 percent of all arguments in the quality press and 9.7 percent of arguments in the alternative press were uncivil. This share amounted to 26.2 percent in the tabloid press. The differences between the press types were significant (χ2(2, N=403) = 7.16, p = .022). In 2011, there were no
Only in the alternative press were 10.5 percent of the arguments uncivil. Thus, the amount of uncivil arguments in the alternative press did not significantly differ between the two debates (χ2(2, N=200) = 0.38, p = .846).
Concerning the tonality between the different types of press, the results show that almost three-quarters of all arguments in the quality press and the alternative press in 1987 were
balanced or neutral and that this share significantly decreased in 2011 (see Table 4). In 2011, more arguments evaluated the census as positive or negative, while in the quality press the share of negative arguments and in the alternative press the share of positive arguments grew the most.
Table 4
Tonality of arguments in the press types in 1987 and 2011
1987 2011
Press type Tone of argument No. of arguments Relative share No. of
arguments Relative share Quality pressa Negative 58 11.1% 29 24.4% Neutral/Balanced 379 72.5% 73 61.3% Positive 86 16.4% 17 14.3% Subtotal 523 119 Alternative pressb Negative 77 22.5% 49 25.5% Neutral/Balanced 249 72.8% 118 61.5% Positive 16 4.7% 25 13.0% Subtotal 342 192 Tabloid pressc Negative 23 22.1% 2 4.4% Neutral/Balanced 51 49.0% 42 93.3% Positive 30 28.8% 1 2.2% Subtotal 104 45 Total 969 356 Note. a . χ2(2, N=642) = 14.60, p = .001; Cramér’s V = .151 b . χ2(2, N=534) = 13.92, p = .001; Cramér’s V = .161 c . χ2(2, N=149) = 26.42, p = .001; Cramér’s V = .421
In contrast, for the tabloid press the results show that more than half of the arguments in 1987 had a positive or negative tone. However, the share of arguments that are neutral or balanced significantly increased to 93 percent in 2011.
Concerning the rebuttal or approval of other speakers’ arguments, there were no
differences between the distinct types of press in 1987 (χ2(4, N=403) = 3.08, p = .433) and 2011 (χ2(4, N=95) = 5.16, p = .271). To the contrary, there were significant differences in the level of justification between the distinct types of press in 1987 (χ2(4, N=969) = 46.86, p < .001) and 2011 (χ2(4, N=95) = 89.61, p < .001). Of all arguments in 1987, there were 0.4 percent in the quality press, 2 percent in the alternative press, and 5.8 percent in the tabloid press that were not justified at all. Those shares increased in 2011 to 29.4 percent of all arguments in the quality press, 6.2 percent in the alternative press, and 68.9 percent of all arguments in the tabloid press. The Discussion of Solutions
The last research question concerned potential differences in the discussion of solutions between the two debates. Indicators for the discussion of solutions evaluated if the argument suggested a problem, indicated the possibility of a solution, and potentially even introduced a specific solution or measure. The argument could discuss on which level the solution could be brought about. Furthermore the speaker could demand or appeal to take a certain action. While the argument did not necessarily have to refer to a problem to suggest a solution, the indicators that accrue in the presence of a potential improvement or solution in the argument are connected with each other. To begin with, there had to be a potential improvement before the speaker could refer to the level on which it could be brought about and what the solution and its measures could specifically look like. Thus, the indicators and the findings of the discussion of solutions are related with each other.
Out of all arguments in 1987, 88.6 percent addressed a problem or a problematic issue. In 2011, this was the case in 70.8 percent of all arguments. This difference was highly significant (χ2(1, N=1325) = 61.33, p < .001). Furthermore, 58.5 percent of the arguments saw opportunity for a solution or an improvement in 1987, while in 1987 this was the case in 37.9 percent of the arguments. Again this difference between the years was highly significant (χ2(1, N=1325) = 44.32, p < .001). In 1987, 46.3 percent of the arguments mentioned a specific measure or a solution to improve a problem. This share amounted to one-third of all arguments in 2011. The difference between the years was also highly significant (χ2(1, N=1325) =17.72, p < .001). Taking all arguments into consideration, in 1987 there was a call to action in 18 percent of them while this was the case in 10.1 percent of the arguments in 2011. The difference is highly significant as well (χ2(1, N=1325) =12.01, p = .001).
Discussion
Before discussing the results, we will briefly refer to some important limitations of the scope of this study. To begin with, the two census debates lie 24 years apart. Recent surveys show that data security and privacy issues still play an important role in Germany (DIVIS, 2013). However, there might be potential events or developments in society and politics besides
changes in the media system that had an effect on the issue and its discourse (e.g., a difference in sense of privacy). Especially, as the overall lower amount of articles and arguments suggest less debate in the press coverage, one might argue that the debate was of less interest in 2011. Additional research should therefore also introduce the perspectives of the communicators and the audience in the census debate, as this study does not consider what the stakeholders in the debate did with the information they received in the media.
Concerning the development of the digital media and the proliferation of the Internet and online news in particular, this study takes into account only the traditional press media.
Therefore, parts of the debate that took place in the public sphere of online media in 2011 were not included and addressed. However, due to reasons of comparability between the specific outlets in 1987 and 2011 there was no inclusion of online news in this study. Future research should take online media into account and could presumably analyze differences in the census debate in 2011 and prospective population censuses. Likewise, the focus on the press coverage as one important part of the public sphere does not consider the discourse that can be found in other media such as television or in interpersonal encounters of citizens. Lastly, the findings at hand do only apply to the German case of the census debate, especially as other countries use different methodologies for their census and have different cultures of discourse.
With regard to the results, the different amount of arguments first needs to be addressed. It was shown that the absolute number of articles and arguments in 2011 was lower than for the sample in 1987. In general, there was less space and attention given to the census debate in 2011 by the print media. However, if an article in 2011 addressed the census it contained on average two more arguments than an article in 1987. This difference might be based on more short articles concerning a lot of smaller events during the census in 1987.
Starting with the first research question and whether there is a difference in the diversity of the speakers, the results showed some relevant changes between the years. The census debate in Germany showed a higher elite dominance of speakers in 1987, but a more evenly balanced diversity of speakers with a greater appearance of arguments from political periphery speakers and journalists in 2011. In line with Hallin´s (1992) research on the development of news sound bites, the results suggest that media coverage was more journalist-centered, as the journalists in
the German press coverage devoted less time and space on presenting arguments by other speakers in 2011. They preferably delivered the arguments themselves.
Furthermore, the higher share of arguments from speakers of other parts of society than the political center suggests a greater popular inclusion than in public discourse on other issues. This share of arguments from the periphery is even greater in 2011 than in 1987. The numbers at both points in time resemble more those that Ferree et al. (2002b) found in the US than in
Germany in their study on the abortion discourse in both countries. They showed that 29 percent of speakers in the US were only from the periphery, and 15 percent in Germany. Wessler (1999) found similar numbers for the media coverage on drugs in Germany with citizens making up 16 percent of the speakers. A reason for the different numbers might be that it was easier for journalists to find engaged speakers from the periphery. The topic of the census affected every citizen and was presumably less intimate than drug consumption and abortion.
Concerning the second research question and whether there was a difference in the process of the discourse the results above addressed changes in the indicators of empowerment. It was shown that the debates only differed in the amount of space that was given to the civil society speakers. The results indicated that the share of arguments made by citizens and social movements was lower in 1987 than in 2011. In the former debate, less than one in ten arguments was made by a speaker belonging to the two groups and rose to more than one out of eight arguments in 2011. This is consistent with the share of such speakers that Ferree et al. (2002b) found in the abortion discourse in Germany. Otherwise, the process of discourse appeared to be identical in both debates. Regarding the indicators such as the content of justification and the level of the solution and its encouragement, the results suggest a certain amount of
in the media system and in political communication described above did not have any effect on the extent of empowerment and the process of communication. Concerning the content of
justification, the arguments quite scarcely referenced anyone´s interest at all. Thus, in most cases it is not made clear to citizens how they can profit or be put at a disadvantage directly by the census. Based on the results, neither are many arguments present in the media coverage that refer to specific group interests (e.g., the interest of corporations, etc.). Furthermore, in a majority of the arguments the speakers do not suggest that individuals are able to do something to improve a situation. In the arguments, in which they do see that a solution can be brought about on the individual level, the speakers actively and explicitly encourage the individual citizen to take action four out of ten times in both debates.
As far as the prevalence of the different types of arguments, it was shown that in both debates the majority of arguments addressed only the census without any reference to other speakers or their arguments. To begin with, the share of arguments addressing other speakers and their arguments that covered certain aspects of the style of speech in the third research question were consequently less common in 2011.
On the one hand the third research question addressed differences in the style of speech of such arguments between both debates. One the other hand, it was asked whether there were any differences between the two debates regarding the type of press of the newspapers and the presented arguments. For both parts of the third research question, the results show a mixed picture. For the general debate, there was no difference in civility and the responsiveness, which addresses rebuttals and approvals of others’ arguments. A greater share of arguments had a positive or negative tonality in 2011, but at the same time the extent of the level of justification performed worse in 2011 than in 1987. More than one out of five arguments did not give any
justification in 2011 compared with 1.5 percent of the arguments in 1987. This development shows that the level of justification decreased the most in the tabloid press. One could argue that the debate became more superficial as arguments became less justified. With regard to the press types, the tabloid press differed from the quality and the alternative press as it showed stronger and reverse changes between the two debates. Contrary to expectations of higher negativity and sensationalism in the tabloid press, their media coverage appears to be more detached and neutral but also more superficial in the latter debate.
The results show that overall the share of uncivil arguments appeared no different in 1987 than in 2011 and that about one out of ten arguments addressing another speaker was uncivil. This share is almost identical with the assessment by Ferree et al. (2002b) of the abortion issue. More specifically, in 1987 the tabloid press contained a significantly higher share of uncivil arguments than the quality and alternative press. In 2011, there were no uncivil arguments at all in the tabloid press and in quality newspapers. The share of uncivil arguments in the alternative press, however, stayed constant between the debates, attributing to the constant incivility in both debates. The tendency of increasing negativity and sensationalization in the media coverage did not lead to more incivility in the press coverage of the census debate. There was actually no room for incivility at all in the revision of the debate in 2011, except for the alternative press that was also the most negative towards the census.
Furthermore, the tonality of arguments differed between the two debates of the census. While still a majority of arguments neither approved nor disapproved of the census, this share was lower in 2011. As one might falsely conclude that the debate was more heated in 2011, it has to be kept in mind that the overall amount of arguments in the media coverage was substantially lower in this debate. Still, the results suggest that a higher share of arguments stated a clear
evaluation of the census. As far as the tonality, the results indicated a reverse development between the tabloid press, and the quality and alternative press. As the share of arguments with a specific tone rose for the latter two press types to almost 40 percent, the share of such arguments dropped from more than half of all arguments to less than ten percent in the tabloid press in 2011. This also suggests a greater neutrality of the tabloid press in the latter debate.
Concerning the dialogic structure, the results show that there was no difference between the two debates and the press types in terms of the share of arguments that rebutted or approved of another speaker´s argument. In general, the majority of such arguments in both debates were rebuttals. Only about one in ten arguments approved of what another speaker said. Interestingly, those results suggest that there was quite a substantial amount of responsiveness as speakers in both debates referred to another´s arguments and actively rebutted or approved them at least to a certain extent. In a study by Bennett et al. (2004), no responsiveness at all was found in the debate about the World Economic Forum and the Word Social Forum. As the authors suggested, findings will vary depending on the context of the debate and issue. Speakers in the present study were not organizationally and ideologically separated, which could have led to greater
responsiveness in the debate. There were speakers pro as well as contra the census from the political center and from the political periphery.
Regarding the last research question and asking for differences in the discussion of solutions, the results show differences for all indicators between the two debates. More precisely, each of the indicators performed worse in 2011 than in 1987. This shows that solutions appear not to be clearer and perspectives not more consolidated in the media when an issue reappears in the discourse. Still, there is a substantial discussion of solutions for both points in time.
level of justification) there was a more prevalent discussion of solutions and less decrease between the two debates and the different press types.
Conclusion
The answers to the four research questions give an impression concerning the initially introduced guiding questions and their normative criteria of the different models of the public sphere. Other studies that investigated the abortion issue and the drug discourse in Germany came to the conclusion that the German public sphere mostly resembled the representative liberal model of the public sphere. The findings of the present study suggest a similar result.
However, showing a more balanced diversity between the speakers of the political center and periphery in 2011, the findings indicate better popular inclusion of all parts and groups of society and consequently a higher participatory element to the discourse. This is supported by the presence of empowerment indicators in the media coverage at both points in time. Such
differences to other discourses might be based on the topic of the debate. The census affected all citizens and was a less intimate issue than abortion or drugs. Moreover, the specifics of the census debate result in a certain extent of responsiveness and dialogue that is apparently lacking for other issues such as the globalization debate (Bennett et al., 2004). This is arguably based on the palpable individual impact of the census on all citizens that required their cooperation in the survey and that politics and the media could consequently not ignore. They had to personally address the arguments of the adversaries of the census as information campaigns alone did not appease the protesters and boycotters, whose participation was needed.
While some of the indicators for the process of discourse and the styles of speech were scarcely found in the arguments, the focus on the discussion of solutions and therefore on the content level yielded interesting additional results. On the one hand the analysis of the discussion
of solutions showed that the census debate in 2011 was less solution-oriented compared with that of 1987. This suggests that discourses can vary in the extent on which they focus on solutions and that they do not necessarily improve over a longer period of time. On the other hand there was still quite a substantial reference to problems and solutions in both census debates. While the speakers did not talk to each other and engage in an ideal deliberation, the discourse was after all addressing the need to find solutions.
In accordance with other studies, it was shown that the deliberative qualities of the discourse did not meet the ideal standards of the normative criteria of participatory or discursive models of the public sphere. Following Graber (2003), the media and the quality of their debate fall short of their expected ideal roles in a democratic discourse. The political culture of a country might be a more decisive element in the proceedings of a democracy. However, the content of the arguments in this study was still oriented towards finding solutions for the issue of the census. Accordingly, a suggestion for future research involving the public sphere might be to pay more attention to the speakers and the actual content of their argumentation and less on how and in which process ideas are shared in the media.
Further investigations that involve different issues and their debates can give additional insights into the interrelationships of the speakers in a debate and their discussion of solutions in changing contexts. Particularly, comparing differences between countries and between different media systems and journalistic cultures appears to be worthwhile (Hallin & Mancini, 2000). As there are already considerable differences between different types of outlets in the press within one country, it is interesting how the census debate and the discussion of solutions play out in different parts of the public sphere over time. Social movements especially might have carried their engagement in the debate to online media in an attempt to better organize and coordinate
support for their concerns (van de Donk, Loader, Nixon, & Rucht, 2004). It is up to further evaluation, how such developments affect the workings of the public sphere and the necessary search for solutions to societal issues.
References
Bennett, W. L., Pickard, V. W., Iozzi, D. P., Schroeder, C. L., Lagos, T., & Caswell, C. E. (2004). Managing the Public Sphere: Journalistic Construction of the Great Globalization Debate. Journal of Communication, 54(3), 437–455. doi:10.1111/j.1460-2466.2004.tb02638.x Berelson, B. (1952). Content analysis in communication research. Glencoe, Ill.: Free Press.
Blumler, J. G., & Kavanagh, D. (1999). The Third Age of Political Communication: Influences and Features. Political Communication, 16(3), 209–230. doi:10.1080/105846099198596
Broh, C. A. (1980). Horse-Race Journalism: Reporting the Polls in the 1976 Presidential Election. Public Opinion Quarterly, 44(4), 514–529. doi:10.1086/268620
DIVSI. (2013). Umfrage: Politische-Einstellungen – Datenschutz so wichtig wie die Bewältigung der Finanzkrise. DIVSI. Retrieved from https://www.divsi.de/umfrage-politische-einstellungen-datenschutz-so-wichtig-wie-die-bewaeltigung-der-finanzkrise/
Eickelkamp, A. (2011). Der Nutzwertjournalismus. Herkunft, Funktionalität und Praxis eines Journalismustyps. Köln: Herbert von Halem Verlag.
Esser, F. (2008). Dimensions of Political News Cultures: Sound Bite and Image Bite News in France, Germany, Great Britain, and the United States. The International Journal of Press/Politics, 13(4), 401–428. doi:10.1177/1940161208323691
Ferree, M. M., Gamson, W. A., Gerhards, J., & Rucht, D. (2002a). Four models of the public sphere in modern democracies. Theory and Society, 31(3), 289–324.
Ferree, M. M., Gamson, W. A., Gerhards, J., & Rucht, D. (2002b). Shaping Abortion Discourse: Democracy and the Public Sphere in Germany and the United States. Cambridge, UK ; New York: Cambridge University Press.
