Quality of Understanding in Communication among and between Political Parties, Mass Media, and Citizens: An Empirical Study of the 2013 Austrian National Election

The study investigates the extent to which political parties, the mass media, and citizens follow qualitative principles demanded by the public sphere concept in political campaign communications. Using the index of a quality of understanding (IQU), it analyses the press releases and Facebook posts of political parties, newspaper articles, and responses by citizens in the form of comments in newspaper forums and on parties’ Facebook pages (N=7,525) during the 2013 Austrian national election. Considering that the quality of understanding of public discourse is measured on a 100-point scale, which serves as a benchmark representing perfect understanding, observed real-world values are often rather low. Austrian political parties scored the highest IQU of 28.35 points, and hence can be described as most closely following the principles of an ideal communication orientation. The quality of understanding is the lowest in everyday political discussion on Facebook, where political parties’ posts have an IQU of 17.97 points. The difference of 10.38 points to the highest achieved value of 28.35 reveals different deliberative communication practices between well-considered and strategically formulated communication in press releases as well as newspaper articles and everyday communication including citizens’ comments on Facebook and newspaper articles, which take different configurations.

political parties strive to legitimize their actions and interests in order to gain citizens' votes. Hence, election campaigns represent a situation in which competing parties present diverse and often conflicting topics, positions, preferences, and interpretations. Thompson (2008, p. 502) has noted, "[s]ome basic disagreement is necessary to create the problem that deliberative democracy is intended to solve. (…) If the participants are mostly like-minded or hold the same views before they enter into the discussion, they are not situated in the circumstances of deliberation." In the deliberative process, political decisions are legitimized when they are based on a debate that permits the circulation of information, positions, arguments, and ideas because then actors are oriented toward mutual understanding (Habermas, 1992). Citizens, whose active participation in the decision-making process is very limited, usually make their voting decisions based on these public discourses and negotiations of political actors (cf. Curran, 1991). Thus, political actors are accountable for clear and transparent communication because their legitimacy depends on public support. Primarily it is the mass media that distribute "the information necessary for citizens to make an informed choice at election time" (Curran, 1991, p. 29). The role of the journalist is to objectively distribute the information provided by the political parties. However, in this process, the journalist also positions her-/himself as ' advocate of the public discourse' (Burkart, 1998), who reflects, critically examines, and comments on the information. "[I]nterpretation and analysis can provide an important background for audiences and facilitate a deeper understanding of issues" (de Vreese. Esser and Hopmann, 2017, p. 4) and hence support the goal of creating mutual understanding. In this process, the media can actively shape the public discourse. Citizens engage in public discourse to express their opinion on political matters, interact with others, and even to persuade others (Canter, 2013;Penney, 2016). Certainly, citizens do not always follow the high ideals of deliberative quality (cf. Dahlgren, 2005), but to successfully interact with others they at least have to make an effort, because only then will the communication process continue (Habermas, 1992).
For the analysis of the communicative throughput and its quality, this study builds on earlier work by Burkart and Russmann (2016) using the index of quality of understanding (IQU) (Index für Verständigungsorientierung, VOI), which is based on Habermas's conception of deliberative democracy and builds on the idea of core deliberative values. Habermas's (1987) perspective of a deliberative democracy requires public understanding. From this perspective, the question of how to communicate asks whether actors are oriented towards mutual understanding and to the extent to which they try to achieve it. Therefore, this article begins with a brief discussion of the work of Habermas, who has strongly influenced deliberative theory, and previous studies measuring the deliberative quality of communication. It then moves on to introduce and explain the IQU. To contribute to the discussion on the deliberative quality of communication, this study focuses on politics through the analysis of political parties' press releases and Facebook posts, on the media through the analysis of newspaper articles, and on citizens through the analysis of their comments on newspaper articles and their comments on parties' Facebook posts during the 2013 Austrian national election. These aspects are further elaborated in the section outlining the setting of this study. This is followed by the analysis of the quality of understanding of political parties', the media, and citizens' communication in a realworld setting. The article ends with a discussion of the significance of the findings.
theoretical Background and literature Review First, it should be noted that the findings of empirical studies on the deliberative quality of communication defy direct comparison, because despite "shar[ing] a common core of values, the empirical studies actually adopt diverse concepts of deliberation and examine different consequences under a range of conditions" (Thompson, 2008, p. 501; see also Kuyper, 2018). However, despite the use of different measurement concepts, previous empirical studies share similar normative perspectives of ideals of deliberative democracy when defining criteria measuring the quality of deliberation ascribing to the work of John Rawls (1971), Joshua Cohen (1989) and, as this study does, especially Jürgen Habermas and his Theory of Communicative Action (1984,1987). Habermas (1984Habermas ( , 1987 defines rational conditions for mutual understanding in communicative action. To agree on something, actors have to accept the validity of each other's statements. In seeking this end, they implicitly claim that what they say is true, correct, and authentic, or in any case to be accepted as reasonable and, if problematized, can be rationally justified in these respects (Habermas, 1984(Habermas, , 2008. Habermas identifies four validity claims (Geltungsansprüche): intelligibility (i.e., participants use the proper grammatical rules), truth (i.e., participants are certain that they are talking about something the partner accepts as real), truthfulness (i.e., participants agree on being honest with and not misleading each other), and legitimacy (i.e., participants believe that they are acting in accordance with mutually accepted values and norms). Communication will only continue as long as actors do not doubt the fulfillment of the four validity claims. However, like democracy, such an ideal speech situation (Habermas, 1984) "is something that we can use as a critical standard for judging the quality of actual talk, but it is not something humans can live up to" (Gastil, 2008, p. 22). In his later work Habermas (1992, p. 396; translation by the author) himself emphasizes the ideal speech situation as a counterfactual pragmatic presupposition; having only "the sense of a methodical fiction." Reality often falls short of these ideals. In our everyday communication, claims are open to criticism and justification.
To reach mutual understanding, participants must make a sincere effort to engage in discourse about the particular problem under consideration. In this regard, a discourse about a topic is carried by doubts. Following ideal democratic standards, participants have "to make of understanding of political discourse. For instance, it helps to find answers to the question of the extent to which the political discourse among and between parties, the media, and citizens comes close to the benchmark of genuine deliberation.
The quality of understanding depends on the strength of the five indicators (i.e., subindices). For instance, the quality of understanding is high when participants (a) state not only their positions on a particular problem but also the reasons behind them, (b) propose specific solutions for debatable issues, (c) deal with other participants in a more or less respectful way, (d) refer to each other, and (e) express relevant doubts concerning positions and other participants. In the following, I provide a detailed description of the five indicators and their subindices.

Method and Measurement: index of a Quality of Understanding
The IQU provides an overall evaluation of the quality of understanding of public discourse. 1

statement of Reasons
First, participants of public discourse want to convince others of the importance of their issues (Gerhards, 2003). They want to persuade them to accept and to support their stances. Thus, participants "should give reasons for [their positions and] decisions, and respond to the reasons that citizens give in return" (Gutmann & Thompson, 2004, p. 3; see also Cohen, 1989). Generally, well-grounded positions tend to gain in importance and less well-grounded positions lose in importance (Gerhards, 1997). A few empirical studies that examine the effects of deliberation in political debates (e.g., Kuhlmann, 1999;Steenbergen et al., 2003) highlight the importance of this principle to reaching mutual understanding as for consensus to be reached, arguments must be well explained and logically presented (Spörndli, 2004).
To analyze the quality of understanding, it seems to be important to differentiate degrees to which reasons (justifications, arguments) are stated. The IQU distinguishes four degrees, assigning a value to each: 0 = no statement of reasons is given for a position, a political act, or political behavior; 1 = generalized statement of reasons: indefinite, short justifications based on common sense; 2 = simple statement of reasons: the argument is justified by a single fact; 3 = specific statement of reasons: supported by facts and figures, expert opinion, or science.
In each unit of analysis (i.e., a single press release, newspaper article, Facebook/newspaper forum post or comment), up to three variables in order of their appearance are coded. For instance, once three statements of reasons in a press release, article, post, or comment have been coded, no additional statements of reasons are counted. The subindex statements of reasons is measured by (a) the number of justifications and (b) their degrees. Thus, higher values indicate more justifications at higher degrees. Each score aggregates (up to) three coded variables. Thus, the maximum score for degrees of statements of reasons is 9, which is given if three specific statements of reasons (concerning the main statement) have been coded. The subindex (standardized by the maximum value) is calculated using the following formula: INDBEG (Index of Statement of Reasons) = (CumulativeValue3VariableStatementsofReasons/9) × 100.

Proposals of solutions
Participants' ideas and opinions about public issues often differ. Thus, ideally, participants will introduce varying solutions for a particular problem addressed in the discussion (cf. Cohen, 1989). Based on the analysis of abortion discourse in Germany, Gerhards, Neidhardt & Rucht (1998) note that, to convince others of their views and opinions, the political actors offered differing solutions for a particular problem.
Again, the IQU distinguishes between the quality of proposals of solutions: 0 = no proposals of solutions are provided for a problem; 1 = partial proposals of solutions: the participants introduce an idea; 2 = precise proposals of solutions: a detailed concept is introduced and its de facto implementation is outlined. The subindex for proposals of solutions is aggregated the same way as the INDBEG, but its maximum value is smaller. Hence, the divisor in the formula is 6: INDLOES (Index of Proposals of Solutions) = (CumulativeValue3VariableProposalsofSolutions/6) × 100.

Respect
A political discussion requires "maintaining a degree of respect for yourself and your fellow participants" (Gastil, 2008, p. 10; see also Fishkin, 1991 on mutual respect). Respect means that participants actively listen to one another to understand different viewpoints (Gutmann & Thompson, 2004;Spörndli, 2004). Participants can question one another's positions and actions as long as criticism remains respectful (Spörndli, 2004). The analysis of debates in the German Conference Committee shows that mutual respect is a factor that positively influences the process of reaching consensus (Spörndli, 2004).
The IQU defines respect for positions and other political actors as another condition for mutual understanding in communicative action and differentiates between three respect levels: 0 = disrespectful expression: expressions with an explicit, negatively valenced attribute; 3 = respectful expressions: expressions that are neither negatively nor positively valenced; this applies to all units of analysis that are neither disrespectful nor explicitly respectful; 4 = explicitly respectful expressions: expressions with an explicit positively valenced attribute.
This subindex differs from the subindices of the first two indicators insofar as the respect level indicates a negative parameter. To prevent a negative index value and thus allow for aggregation of the respect level with the two other only positively evaluated indicators (statements of reasons and proposals of solutions), the value 0 is assigned to the negative attribute of respect (respectless expressions), thereby assigning higher values to respectful and explicitly respectful expressions to emphasize their (exclusively) positive notion. The subindex for respect is aggregated the same way as the INDBEG and the INDLOES, but its maximum value is higher. Hence, the divisor in the formula is 12: INDREGEY (Index of Respect) = (CumulativeValue3VariableRespect/12) × 100.

Reciprocity
To ask that participants listen to each other implies that people will respond to what others have said. Political discourse is characterized by reciprocal actions between many people. "Reciprocity can therefore be defined as a basic condition for deliberation. If citizens do not listen to each other and interact (…), there can be no deliberation, only monologue." (Kies, 2010) Specifically for the discourse of politics and the mass media with citizens, this means that citizens have to be treated as more than mere objects of legislation or passive subjects to be dominated; citizens are agents with their own interests, who directly or through their elected representatives participate in politics (Gutmann & Thompson, 2004;Kies, 2010).
Again, the IQU distinguishes between three different kinds (resp. values) of reciprocity: 1 = monologue: a message does not refer to a previous message; participants are only stating their opinions, ideas, etc.; 2 = initiation: a message that may initiate a new discussion (e.g. by asking questions and calling for responses from fellow participants; 2 = response: a message that explicitly responds to a previous message. The subindex for reciprocity is measured by a single value per unit of analysis. The subindex (standardized by the maximum value) is calculated using the following formula: INDREZ (Index of Reciprocity) = (CumulativeValue1Variable Reciprocity/2) x 100.

Doubts (Validity claims)
Following the Habermasian perspective of understanding, when interacting participants disbelieve other participants' validity claims, they sometimes cast doubts on other participants and their messages. Thus, by justifying their positions, behaviors, or actions, participants will try to dispel doubts. The IQU distinguishes between four types of doubts: Doubts are expressed (1) about intelligibility if participants question whether a statement is formulated in such a way that the members of the addressed public can understand it; (2) about truth if participants question whether a statement pertaining to a specific circumstance (e.g., a situation, fact, or occurrence) is a proven fact; (3) about truthfulness if participants claim that another participant is not trustworthy (e.g., lying, lack of honesty or integrity); and (4) about legitimacy if participants question the appropriateness of others' actions, interests, and behaviors.
First, counter variables are used to draw conclusions about the different types of doubts: zw1verst = number of doubts of intelligibility (max 3), zw2wahei = number of doubts of truth (max 3), zw3wahaf = number of doubts of truthfulness (max 3), zw4legi = number of doubts of legitimacy (max 3). Then, the cumulative value for the subindex for doubts is aggregated in the same way as for the other indicators: zweifsum = zw1verst+zw2wahei+zw 3wahaf+zw4legi (max 3); INDZWEIF (Index of Doubts) = (zweifsum/3) × 100.

index of a Quality of Understanding (iQU)
The indexing process is based on a scoring system that considers the maximum possible value of each of the five subindices. The total score is transformed into a scale from 0 to 100. Owing to standardization (of the maximum value), each subindex is integrated with the same weight into the IQU, although the indicators have different manifestations at different levels. Thus, each of the five subindices contributes to the IQU with the same weight: A total score of 100 is unlikely in any analyzed communication, and especially in everyday political discourse such as that surrounding an election campaign. However, the IQU provides an analytical template to evaluate the democratic quality of discourse and, thus, can be applied to the analysis of all kinds of discourse in different settings-online as well as offline. In the current study, the IQU may provide answers to the question of how deliberative democracy is displayed in different settings that are interlinked at a specific point.

the setting
Austria is a representative democracy. The country's political system has a multi-party structure and a consensus-based culture (Fallend, 2010). However, in the 1990s, "the erosion of consociationalism has led to a more dynamic and open system of political contestation" (Melchior, 2005, p. 13). At the turn of the century, "a new differentiation emerged wherein an agreement-based model of administration shifted towards a conflict-and competition-oriented based model" (Fallend, 2010, p. 173). Today, more political actors, media outlets, and citizens are demanding political participation and comprehensive democratization.
Political parties hold press conferences, give interviews, and fireside chats, and they send out press releases to distribute their messages. To examine the quality of understanding of parties' communication the presented empirical study focuses on press releases because press releases "closely reflect the daily campaign. In these documents parties try to launch their main theme de jour" (Norris et al., 1999, p. 44). Press releases are usually wellconsidered, deliberately, and strategically formulated. In their press releases, parties seek approval for themselves and their positions, they justify their views and actions so recipients can sift the better from the worse (Levine, 2003). They are thus an important source for determining the political discourse of the parties in the election campaign because, more than any other means of communication, press releases reflect the party agenda consistently and comprehensively (Walters, Walters & Gray, 1996).
Most likely, this communication will reach citizens via the mass media. Melischek, Rußmann & Seethaler (2010) emphasize that, in Austria, press releases influence media coverage, at least to an extent. The mass media raise "people's awareness of issues that are of public concern, provide a platform for debate, serve as a watchdog" (Maia, 2018, p. 350) and, hence distribute "the information necessary for citizens to make an informed choice at election time" (Curran, 1991, p. 29). Austria (with its consensus-based political culture) has a democratic corporatist media system (Hallin & Mancini, 2004), which is committed to a commentary-oriented journalism. Newspapers provide an adequate opportunity for the broader electorate to inform itself about public issues and to form opinions, as well as promoting a rational political discourse (Dahlgren, 2005;Levine, 2003). Newspapers have a strong position in the Austrian media market due to the historically early establishment of the mass press and its coexistence with a parallel, though low circulating, diverse party press (Seethaler & Melischek, 2006). A current study comparing political journalism by de Vreese and colleagues (2017) describes news media coverage in Austria as "issue-focused interpretive" (p. 171). It contains "a good deal of political substance (i.e., hard news indicators)" (de Vreese et al., 2017, p. 171), but the news media are also an active shaper of public opinion. Through their interpretations and analysis, journalists provide additional information to give their audience a deeper understanding of issues , and hence they support the goal of creating mutual understanding.
Citizens' communication about politics increasingly happens online. Social media and online forums are the new neighborhood pubs. Here Habermas' (1990) vision of small coffee house discussions becomes reality. The online public sphere contributes to political transparency and promotes deliberative processes (Dahlberg, 2001a). Social media and online forums provide some of the same functions as face-to-face communication for the discussion of current topics (Grieve et al., 2013). People engage in commenting activities "to express a personal opinion on the subject matter of the story (…) and to interact with other readers" (Canter, 2013, p. 607). Sometimes even to persuade others and to initiate open and equal deliberative dialogue (Penney, 2016). Facebook was and still is the most important social media platform for Austrian parties because it is the most popular platform in Austria (Statista, 2018). Certainly, most people are on Facebook for other reasons than political discourse and only a few are linked to a politician's or a party's social media page. However, Facebook fulfills functions such as "community-building, social capital, and civic engagement away from the home and the workplace. Politics here aligns itself with broader repertoires of self-expression and lifestyle values. Politics in Facebook goes to where people are, not where we would like them to be" (Chadwick, 2009, p. 30). Certainly, compared to parties' press releases and newspaper articles, which theoretically more closely follow deliberative criteria, citizen's communication follows more everyday reasoning. However, social media as well as newspaper forums allow grasping citizen's political communication and they connect and empower even more people than the corner pub.

Research questions
Using the index of a quality of understanding, the current research seeks answers to the following questions: Only press releases, newspaper articles, and Facebook posts with a reference to Austrian federal politics were selected and thus its central topic can refer to the polity, politics, or policy dimension of politics. For newspaper articles, only those were included which were either published on the homepage of the website of the daily newspaper (comparable with page 1 of the print edition) or on the pages with political reporting with reference to Austria (Die Presse and Der Standard: Domestic politics, Kurier: Domestic politics, Krone: Politics section). A purely geographical reference to Austria such as when Austria or an Austrian location is only mentioned as a venue (e.g. for a meeting of international organizations), is not regarded as a reference to Austrian politics. Furthermore, press comments, interviews, letters to the editor as well as articles in all other parts of the newspaper are not included.
The study focuses on the last six weeks of the 2013 Austrian election campaign. Because of the elaborate analysis of the IQU and limited resources, I examined only a subsample of the data: First, I reduced the material to three of the six weeks, the first (pre-election period), third (halftime), and sixth (final stage) week prior to Election Day (see Table 2). This selection allows for a reduction in the amount of material, while still considering the full period of six weeks spanning the various stages in the hot phase of the election campaign. Comments on newspaper articles and Facebook contributions posted within these three weeks may not have been made during the same three weeks. Therefore, comments made within the first and second week are subsumed in the pre-election period, comments made within the third and fourth week are subsumed in the halftime period, and comments from the fifth and sixth week are subsumed in the final stage. Second, owing to the great number of posts and comments on the Facebook pages of the FPÖ, The Greens, and Team Stronach, a random sample was drawn comprising about 500 posts and the associated comments from each party's page. The approximately 500 posts and comments on the Facebook pages of each NEOS and the SPÖ served as a benchmark for the sample size, as these parties represented the middle of the range for posting frequency. If a post received fewer than 50 comments, all of them were coded. If a post received more than 50 comments, regardless of the number of comments (e.g., 250 comments), (approximately) 30 succeeding comments were randomly selected and coded. I used the same approach for newspaper articles with more than 50 comments. For the ÖVP, all 32 published posts and comments in the relevant period were coded entirely.
The single press release, the single newspaper article, the single Facebook post, as well as the single newspaper and Facebook comment, is treated as the unit of analysis. Especially, user comments in newspaper forums and on Facebook are often without substantive value to the political discussion. Participants are only referring to trivia, nonsense, or giving plain encouragement for the political actor such as you are the best. In the case of a party's posts, this might include only information about an upcoming event such as a TV debate. These posts and comments were not coded. Participants need to give some kind of relevant and substantive information about political issues to contribute to the online discourse. Comments with substantive value in general indicate a statement, an opinion, and/or an idea. Only those were analyzed according to the principles of a quality of understanding and are included in the analysis.
Holsti's coefficient was used to test intercoder reliability. Overall, the intercoder percentage agreement for each variable falls within the (acceptable) range from 0.66 to 1.0. 3

communication Orientation among and between Political Parties, Media, and citizens
The data in Figure 2 reveal an IQU between 17.97 and 28.35 points (on a 100-point scale) across political parties, the media, and citizens. The mean provides an overall quality evaluation. Both of these extreme scores were measured for party communications: The lowest IQU was measured for Facebook posts and the highest IQU was measured for press releases. Based on a Kruskal-Wallis rank sum test it can be concluded that means significantly differ between the analyzed forums (p < 0.001) with the exception that the difference between newspaper forum comments and press releases was narrowly non-significant (p = 0.065).
It is important to notice that the indicator reciprocity was not coded for press releases. Parties use press releases to distribute their messages. Initiating a new discussion or calling for a response from another party are only given implicitly. When measuring the IQU for press releases without the indicator reciprocity a higher IQU of 35.44 points is reached-compared to an IQU of 28.35 points with the indicator reciprocity. This score is comparable to previous findings (Burkart & Russmann 2016) measuring the IQU without the indicator reciprocity for parties' press releases over four decades in the Austrian national election campaigns in 1970 (38.30 points), 1983 (37.26 points), 1999 (36.88 points), and 2008 (35.69 points). Hence, the presented results for press releases for the 2013 national election campaign are in line with previous findings.
In the 2013 Austrian national election, the IQU for newspaper articles is 25.64 points and hence lower than the IQU for parties' press releases. This is also in line with previous findings measuring the IQU over four decades (1970: 19.62 points, 1983: 17.64 points, 1999: 18.53 points, and 2008: 18.81 points; Burkart & Russmann, 2016)-even though, in 2013 the IQU for newspapers is a little higher. Analyzing the role of the media in Austrian election campaigns from 1970 to 2008, Seethaler and Melischek (2014) note that the media have overcome "its former role as mere transmitters" (p. 258) and is "predominantly choosing a proactive" (p. 272) role in the campaigning process. This manifests itself in greater journalistic autonomy. The media are shaping public discourse (cf. de . Hence, in the 2013 election campaign, the media not only covered the parties' agendas, but also provided more background and promoted a deeper understanding of issues (de Vreese et al., 2017) leading to a higher quality of understanding of news media coverage. The IQU for newspaper forum comments of 27.74 points is very close to the score for newspaper articles. Facebook communication shows the lowest quality of understanding.
To get a more precise picture of the quality of understanding of public discourse between political parties, the media, and citizens, each indicator is analyzed separately as shown in Table 3. Overall, the results show only a few patterns for the quality of understanding within and between the three actors of public discourse, but these shall be discussed in more detail.
On the party-level, a Kruskal-Wallis rank sum test reveals no statistically significant difference in IQU scores for press releases between the different parties (p = 0.064), but it does for parties' Facebook posts (p = 0.001) ( Table 3). Thus, for press releases, we can only speak of a tendency  (Sommer, 2008). But the party shows the lowest scores for statements of reasons and proposal of solutions for press releases. Conversely, on Facebook the FPÖ introduced a lot more proposals of solutions than the other parties did and scored above average on the subindex statements of reasons. Even though, in the party comparison, the Greens cast more doubts on other political actors and their messages and scored above average on the subindex statements of reasons on Facebook, they-contrary to the FPÖ-score highest on proposals of solutions in press releases, but did not introduce a single proposal of solutions in their Facebook posts. The Greens were also more respectful than any other party. In contrast to these two oldest opposition parties, the two incumbents and coalition partners, SPÖ and ÖVP, have the lowest IQU in party communication.
Both parties score (rather) low on the subindex doubts for press releases and lowest for Facebook (SPÖ: 5.56 points, ÖVP: 3.70 points). It seems that raising doubts is reserved for opposition parties, rather than incumbents. A Mann-Whitney-U-Test was calculated to determine if there were differences in the IQU between government parties (SPÖ and ÖVP) and opposition parties (FPÖ, Greens, NEOS, Team Stronach). For Facebook posts, opposition parties (Mdn = 16.11) had a significantly higher IQU than the governing parties (Mdn = 11.67), z = -3.205, p = 0.001, r = -0.32. For press releases, the IQU for governing parties (Mdn = 27.50) did not differ from the IQU of the opposition (Mdn = 27.78), U = 23877.500, z = -1.134, ns, r = -0.05.
With an IQU of 17.97 points for parties' posts and an IQU of 22.71 points for citizens' comments, Facebook communication showed the lowest quality of understanding (Figure 2). For parties' Facebook pages, a Kruskal-Wallis rank sum test reveals that the IQU differs significantly between parties' posts (p = 0.001) as well as between citizens' comments (p < 0.001) ( Table 3) With the exception of the Facebook account of the Team Stronach (IQU for posts = 26.17 points; IQU for citizens' comments = 18.73 points), the data show that citizens follow the principles of quality of understanding in their comments to a greater extent than the parties do in their posts. Following the Team Stronach community, the FPÖ community has the second lowest IQU for citizens' comments.
Comparing the IQU for Facebook comments along ideological lines with Greens, SPÖ, and NEOS as parties (primarily) on the left side of the ideological spectrum and FPÖ, Team Stronach, and ÖVP as parties on the right side of the ideological spectrum, the results show that Facebook communities of (rather) left parties (Mdn = 22.78) have a significantly higher IQU than communities of right parties (Mdn = 15.56), U = 433267.000000, z = -10.530, p < 0.001, r = -0.22.
For newspaper coverage, a Kruskal-Wallis rank sum test reveals that the IQU differs significantly between newspaper articles from different outlets (p = 0.001) ( Table 3). Pairwise comparisons indicate considerable differences between the groups Kurier-Standard (p = 0.003), Kurier-Presse (p < 0.001), and Kurier-Krone (p =        1 FPÖ-ÖVP (p < 0.00l), FPÖ-Greens (p < 0.00l), Team Stronach-Greens (p = 0.016) and SPÖ-Greens (p = 0.006).   -0.20. This is also in line with previous findings measuring the IQU for these newspapers over four decades, which showed a higher IQU for Der Standard (Burkart & Russmann, 2016). Surprisingly, Der Standard scored lowest on the subindex statements of reasons and the Krone scored highest. Nevertheless, in Der Standard the tone is more respectful, the newspaper introduces many proposals of solutions and casts more doubts than the other three newspapers, which increases the quality of understanding of their articles. The Kruskal-Wallis rank sum test reveals that the IQU also differs significantly between newspaper forums (p < 0.001).
Pairwise comparisons indicate significant differences between five of the six groups: Kurier-Presse (p = 0.003), Kurier-Krone (p = 0.003), Kurier-Standard (p = 0.003), Presse-Standard (p < 0.001), and Krone-Standard (p < 0.001). Table 3 illustrates that the IQU for forum comments of Der Standard (32.65 points) is the highest, followed by Die Presse (26.74 points), Krone (26.62 points), and Kurier (25.42 points). To determine if there were differences in the IQU between quality newspapers (Der Standard and Die Presse) and tabloid newspapers (Krone and Kurier) a Mann-Whitney-U-Test was calculated. The data reveal a similar picture for the communities as for the articles: Forums of quality newspapers (Mdn = 31.67) had a significantly higher IQU than forums of tabloid papers (Mdn = 26.67), U = 2204320,000, z = -10.373, p < 0.001, r = -0.15.
Moreover, just as on Facebook, citizens' communication in the media shows a higher quality of understanding than the communication (input) it refers to-albeit differences in newspapers are much smaller than on Facebook. But no similar communication patterns can be revealed for citizen's communication in newspaper forums and on Facebook. In newspaper forums, citizens cast more doubts and comments are more reciprocal, but they are less respectful than on Facebook.
The comparison between party, media, and citizen communication also reveals similarities between parties and the media as well as differences between their communication and citizens' communication. Communication in parties' press releases and newspaper articles that is usually well-considered and/or strategically formulated not only uses generalized statements of reasons but an equal or greater amount of simple statements of reasons, as well as using specific justifications more frequently than other groups (see Table 4).
For instance, as illustrated in Table 4, of the 746 statements of reasons found in the 454 press releases 44%  Table 4 further illustrates that 81-86% of instances of doubts in the analyzed outlets are doubts on legitimacy and almost no doubts were cast on intelligibility. Differences are revealed for doubts on truth, more prominent in newspaper articles and forum comments, and truthfulness, more prominent in press releases and on Facebook. Of course, accusing a political actor of lacking honesty or integrity cannot really be addressed in discourse and can only be examined in subsequent actions.

Discussion and conclusion
This paper responds to the call to consider the deliberative qualities of communication among and between politics, the media, and citizens, particularly by examining the quality of understanding of their discourses. Using the index of quality of understanding (IQU), the study analyses the extent to which the communication of these three actors of the political realm, who are highly connected, follows the principles of deliberation in the 2013 Austrian national elections.
Considering that the quality of understanding of public discourse is measured on a 100-point scale, which serves as a benchmark, the results are rather low for the 2013 Austrian national election. Austrian political parties scored the highest IQU of 28.35 points, and hence can be described as most closely following the principles of an ideal communication orientation, followed by communication in newspapers. The quality of understanding is the lowest in everyday political discussion on Facebook, where political parties' posts have an IQU of 17.97 points. However, as this value is 10.38 points below the highest achieved value of 28.35, these communications also reveal deliberative communication practices of different configurations and magnitudes. With deliberation being an inherently normative concept, it is also debatable how much deliberation can be realistically expected. Bächtiger and Parkinson (2019, p. 43) note that "[w]e will not (or only very rarely) find perfectly deliberative (…) actors in the real world, but actors who use narratives and rhetoric to bolster formal arguments, who are respectful but do so with some strategic impetus as well, who use reasons to persuade but also to mobilize." The findings reveal differences between well-considered and strategically formulated communication and everyday communication: Press releases are more sophisticated, acknowledging the standards of deliberation to a greater extent, by frequently using simple and specific statements of reasons as well as precise proposals of solutions. In their press releases parties seek approval for themselves and their positions, they justify their views and actions so the better can be sifted from the worse and they present their ideas for the next legislative period. Their aim is to persuade citizens (and political opponents) of the importance of their issues and stances. Newspaper articles seek to inform their readers and give them input for when participating in public discourse. In this sense, political parties and the media act as is expected of them as main institutional actors in a democratic system. In contrast, citizens' communication is dominated by generalized statements of reasons, while precise proposals of solutions are rarely present. This might raise the question of how successful newspapers have been in informing citizens, but that is a topic for another study. Moreover, citizens' communication online is often more spontaneous. Parties' communication on Facebook resembles citizens' communication. As it can be assumed that most campaign communication on social media is also planned and thought through, it looks like political parties adopt an everyday communication style on Facebook. In contrast to parties and newspapers, rather informing communication, citizens' comments are reciprocal. Citizens engage in commenting practices to express their opinions-just like they do in face-to-face communication in their neighborhood pub. This is more often the case for Facebook communities on the left sight of the ideological spectrum. There, IQU is above the average IQU score, whereas the IQU of Facebook communities on the right side of the ideological spectrum is below the average IQU score. Future research should further examine the communities and particularly the actively participating citizens. For the analyzed sample, the data show that only a very few citizens can be described as active participants in public discourse. For instance, the 2,166 analyzed Facebook comments were posted by 1,497 different users (names), whereby only 137 users commented three or more times. The highest number of comments by a single user (within the analyzed sample) was 28, followed by 21 comments.
Findings for parties' communication on Facebook reveal that, in the present study, the two incumbents have the lowest IQU, whereas the two oldest opposition parties have the highest IQU and score high on raising doubts-primarily by questioning the legitimacy of the governing parties' actions, interests, and behaviors. Election times are times when the previous legislative period is questioned and this needs to happen in a way that everyone can understand it. In sum, two-thirds of all doubts cast in campaign communication are doubts on legitimacy. This is in line with previous findings (Burkart & Russmann, 2016).
Unlike on Facebook, the quality of understanding in newspaper articles seems to influence the quality of understanding in forum comments. The media need to be aware of its positive impact on citizens' communication practices and, especially with growing mis-and disinformation as well as negative communication practices such as hate speech in newspaper forums the media need to foster deliberative practices. Parties, however, should consider that Facebook is the platform that enables them to reach more citizens directly than any other platform. Deliberation, and thus a high quality of understanding, is important because it leads to better informed citizens, strengthens a more tolerant attitude, promotes a sense of community, increases the frequency of political information searches and political activities of citizens, as well as increases the frequency of making more informed decisions about public affairs (Gastil, 2008;Ryfe, 2005).
A limitation of this study is that it did not control for the length of utterance. For instance, specific statements of reasons or precise proposals of solutions have a greater probability of occurring in longer press releases and articles than in shorter Facebook comments. Given the large sample of this study (see Table 2) this criterion was neglected. Another opportunity for future research is the extension beyond one election campaign. Although findings for press releases and newspaper articles in the present study resemble those of a previous study on four election campaigns (Burkart & Russmann, 2016), extending the period of analysis might improve the robustness of the conclusions. In this context, the question arises as to whether the quality of understanding of political parties, the media, and citizens' communication might be different in non-election periods. Studies on non-election periods are generally rare, but as all analyzed outlets are used year-round, it is important to consider this in future study designs.

Notes
1 The description of the IQU method follows Burkart & Russmann (2016

Funding information
For the analysis of Facebook postings funding was received from the City of Vienna (grant number MA62-II/281277/17).