Audience-based indicators for news media performance: a conceptual framework and findings from Germany

Many attempts to conceptualize and to assess the performance of media systems or single news media outlets focus on the “supply side” of public communication, operationalized as characteristics of the news content and the form of presentation. These characteristics indicate the potential performance of news media; they are a necessary but not a sufficient condition for media performance. In order to assess the actual performance of news media we need to know what kind of audiences they reach, how they fulfil their users’ news-related interests and needs, and how they contribute to their users’ perceptions of the news environment. In this article, we propose a conceptual framework for the definition of audience-based indicators for news media performance. We apply this framework to data gathered as part of the 2019 Reuters Institute Digital News Survey for Germany. We compare 42 news media, both online and offline, in regard to their reach in different parts of the population, and to their audiences’ interest in news and politics, their trust in media, and their perceptions of the overall performance of German news media. The findings underline that news media performance is a multidimensional concept and that there are different ways in which news media can perform. Furthermore, the particular type of media, technically or organizationally, still matters when it comes to audiences’ expectations and perceived functions.


Introduction
Media performance is a particularly relevant issue today.In regard to media-related changes, the ongoing processes of convergence (Sparviero, Peil, & Balbi, 2017) and deep mediatization (Hepp, 2020) are challenging all media to (re-)define their contribution to public communication and their position within the media environment.In regard to societal changes, recent political and social developments have substantial implications for democratic processes and social cohesion, leading to serious expectations regarding the media-and the extent to which the media fulfil these expectations.The need to evaluate the performance of individual media is particu-larly urgent for the area of news as it tends to be the key source that people use to keep themselves up-to-date on what is going on in the world and to build their opinion about current political issues.
When it comes to investigating news media performance, research often focuses on the supply side of the communicative process, that is, the content that is offered.By means of content analysis, this study measures a range of indicators, for example, the public relevance of the sampled news, the pluralism of topics and opinions, and the deliberative characteristics of the text (Jandura & Friedrich, 2014;Weiß et al., 2016).However, these indicators only reflect the communicative potential offered by the media, a necessary but not a sufficient con-dition for successful communication.In order to evaluate the degree to which they fulfil their communicative functions and actually "perform," we need to know if there are users who use these media for certain reasons and which effect those media might have.
While there might be much consensus that the users' perspective should be included in the assessment of media performance, there is controversy over the way users can be involved (Hasebrink, 2011).In recent years, several researchers have developed a user-centered perspective on media performance.For instance, in order to explore the often conflicting relationship between normative objectives, journalistic standards and users interests, Urban and Schweiger (2014) investigated the extent to which media users are able to judge news texts in regard to a range of normative criteria.While this approach begins from a normative perspective, Costera Meijer and Bijleveld (2016) adopt a user-centered perspective by investigating what users experience as "valuable journalism.''Current research demonstrates that media performance from an audience perspective can be analyzed on a variety of levels.Some studies refer to users' views of journalism as a profession (e.g., van der Wurff & Schoenbach, 2014).Others are interested in the perceived performance of the news environment in a specific country (e.g., Arlt, 2019;Jackob et al., 2019).Some researchers set out to measure and compare the performance of different news brands (e.g., Kim, Baek, & Martin, 2010).Finally, there exists a breadth of research that examines the perceived performance of specific pieces of journalism (e.g., Urban & Schweiger, 2014).This article refers to the level of single news media: We aim to develop audience-based indicators for the performance of single news media and their specific contribution to public communication.First, we propose a conceptual framework for the definition of audience-based indicators.Second, based on data accumulated in the 2019 Reuters Institute Digital News Survey, we apply this framework to German news media.Finally, we will discuss the findings in regard to the implications involved in the continuous monitoring of news media performance.

Conceptual Framework
As argued above, the assessment of media performance has to go beyond the characteristics of media content that represent their potential performance.In order to grasp the actual performance it has to include the perspective of media audiences.We define media performance as the communicative impact of the respective media in regard to normative expectations concerning public communication; it refers to the communicative functions that news media should fulfil as prerequisites of informed opinion building, democratic participation, and social cohesion (McQuail, 1992).Our conceptual framework builds on research on media uses and media effects that distinguishes: a) observable contacts of users with specific media as analyzed by audience research (Webster, Phalen, & Lichty, 2005); b) users' motivations for selecting specific media as analyzed in line with a uses-and-gratifications approach (Ruggiero, 2000); and c) the impact interactions with these media have, analyzed within the framework of media effects (Potter, 2011).In what follows, for each of these perspectives on audiences-contacts, motivation, impact-we will discuss audience-based indicators that reflect the relevant normative expectations concerning the performance of news media.
In regard to media interactions, theories on media functions within democratic societies emphasize the need for general participation; in the ideal case, every citizen should be informed about issues of public relevance.Thus, from this perspective, the use of news media is a prerequisite for modern democracies (Ytre-Arne & Moe, 2018).If a news program reaches just a few users, it is unlikely that it has a major communicative impact, thus its performance can be considered as low-even if, according to content analyses, it offers high quality content.Against this background, an important criterion for the performance of news media is the size of its audience.However, reach might be misleading if it is treated as the only indicator.Its main advantages-it is easy to measure, to understand, and to communicate-are necessarily linked with important disadvantages: It is not sensitive to differences between users with different social backgrounds and news-related interests and needs.
In order to compensate these limitations, we refer to normative arguments stressing that we have to go beyond the size of the audience by looking at its "structure," that is, its composition by different social groups (e.g., Morley, 1980).In this respect, we observe a tension between different specifications of normative objectives.Some normative discussions, for example, regarding the remit of public broadcasters, define a balanced representation of all social groups as a benchmark (e.g., Thomass, Moe, & d'Haenens, 2015).In this respect, high media performance refers to the principles of social inclusion and of equal access to information for all parts of society.This criterion is in conflict with professional orientations that stress the fact that communicators should carefully consider who their target groups are in order to be able to tailor content to the interests and needs of these particular groups (e.g., Taneja & Mamoria, 2012).From this perspective, media outlets that successfully reach specific target groups would be expected to perform high.Thus, based on normative considerations in regard to the structure of the audience, we cannot think about media performance as a one-dimensional concept.Instead, there are different ways of gauging high performance: Some media perform highly by reaching a diverse audience, while other media perform highly by fulfilling the particular needs of specific social groups.For the explorative objectives of the study at hand, we decided to choose age as an indicator for the social structure of news audiences.As many studies on news consumption have shown, younger and older groups differ substantially in the sources they use for news (Newman, Fletcher, Kalogeropoulos, & Kleis Nielsen, 2019), and public service media are often criticized for not reaching a young audience (Schulz, Levy, & Nielsen, 2019).
Within the overall media and communications environment, news media set out to serve specific functions.Following the uses-and-gratifications approach (e.g., Ruggiero, 2000), we argue that individuals who are very interested in public issues are more likely to use specific news media because they appreciate the experience that these media provide and the gratifications they seek.A news medium that reaches an audience, which is particularly interested in news and uses them quite frequently, seems to offer something that exactly meets these interests-and, therefore, demonstrates a measure of high performance.One might use the term 'news-ness' (Edgerly & Vraga, 2020) to describe this phenomenon: News that succeed in reaching people who are highly interested in news are obviously perceived as news.However, similar to the above argument on audience structure, this indicator can be interpreted in a different way: News that successfully reach those who are interested in news seem to neglect those who are less interested in news and, therefore, contribute to a widening societal gap (e.g., Schulz et al., 2019).Other news, then, might be regarded as performing high, because they succeed in reaching less-interested segments of the population.Once again, we have to face the fact that "performance" is revealed in many different guises.
News are offered to keep people up-to-date on current affairs, to let them know about political issues, to help them understand societal conflicts, and to assist them in building an opinion on relevant issues.So, when it comes to assessing news media performance, we should include measures that indicate the extent to which news media have an impact on their audiences.This argument touches on a highly complex and controversial area of research that is shaped by epistemological and methodological doubts about the measurement of media effects (Potter, 2011).It would go beyond the scope of this study to offer a fully theorized set of indicators to measure the communicative impact of news media.For pragmatic reasons concerning data availability, we propose to make use of indicators included in the Reuters Institute Digital News Survey that measure how respondents perceive the performance of news media in their country.
One concept that stands out as an often researched indicator of the audience's view of news media performance is trust in news (Engelke, Hase, & Wintterlin, 2019;Jackob et al., 2019).One may argue if trust in news, in relation to news use, is rather an independent or a dependent variable-both directions make sense.Nevertheless, in the context at hand, we regard it as an indicator of media performance if the audience of a news medium has a particularly high level of trust in news.
Besides trust, we consider audiences' perceptions of the extent to which the news media in their country are performing well in regard to specific functions.The Reuters Institute Digital News Survey includes indicators for five functions that reflect scholarly discussions (Newman et al., 2019): a) "Watchdog," i.e., the degree to which the news media monitor and scrutinize powerful people and businesses; b) "relevance of topics," i.e., if the topics chosen by the news media feel relevant to users; c) "tone," i.e., if the news media find a balance between negativity and positivity in their description of events; d) "immediacy," i.e., if the news media keep the users up to date about what is going on; and e) "depth," i.e., if the news media help users understand the news of the day.
Figure 1 provides an overview of our conceptual framework with its three perspectives and the respective indicators.Our empirical approach as presented below starts from the assumption that these perspectives are interrelated and that single news media can be characterized by a specific pattern of indicators representing specific grades of media performance.

Method
In order to empirically apply the conceptual framework on news media's performance from an audience perspective we draw on data that were collected by the Reuters Institute Digital News Survey.Beginning in 2012, the Reuters Institute for the Study of Journalism has been running an annual survey on news consumption in many countries around the world.The survey in 2019 (Newman et al., 2019) was conducted in 38 countries

MoƟvaƟon
Interest in news and policƟcs

Impact
Trust and perceived funcƟons by YouGov through an online questionnaire over the months of January and February 2019.In each country, sampling was based on and weighted according to nationally representative quotas for age, gender, region, and education.Respondents who said that they had not consumed any news in the past month were filtered out (around 3% across countries).Due to the methodology and the use of online samples, we have to be aware that the results reflect online users who use news at least once a month and who are more oriented to online communication.Our analyses use the German part of the survey that is organized in cooperation with the Leibniz Institute for Media Research | Hans Bredow Institute as a national partner.The German sample includes n = 2,022 respondents.

Diversity versus specificity
One of the characteristics of this study that make it particularly fruitful for our research question is that, for each country, it includes information on the use of the most relevant national news outlets representing different types of media.This makes it possible to compare these news media in regard to their media performance.In the German survey, respondents were asked if they have used specific online and offline sources for news in the last week.Most of these sources were specific news brands.In addition, three generic categories were used that reflect the highly regionalized media landscape in Germany: "regional newspapers," "public service radio," and "private radio."Brought together, our analyses can build on data for 42 news brands (see Table A1 in the Supplementary File for a list of these brands).
For each of these brands we analyzed the characteristics of their respective audiences.The basic indicator is the overall reach of the brand; in this case, reach is defined as the percentage of people who said that they have used the news outlet within the last week.Following our conceptual model, the audiences of these brands are then described by indicators for structure, motivation, and impact.
As for structure, we analyzed the reach of each medium in the youngest (18-24 years) and the oldest group (65 years and older).In addition, we calculated the difference between these two groups, with positive values indicating a wider reach among older people, negative values indicating a wider reach among younger people, and values close to zero indicating a similar reach in both groups.Finally, we calculated the mean age of the audience.
As for motivation, we defined the following indicators.First, the mean frequency of news use was calculated based on respondents' claims about how often they access news on a nine-point-scale, from 0 = never to 9 = more than 10 times a day.Second, interest in news was measured by asking "How interested, if at all, would you say you are in news?" with a five-point-scale from 4 = extremely interested to 0 = not at all interested.Third, interest in politics was measured by asking "How interested, if at all, would you say you are in politics?"using the same scale.
As for impact, in this case the perceived performance of news media in Germany, we defined indicators referring to trust in news and to perceived functions of news media.In regard to trust, we used mean values of respondents' level of agreement with two statements: "I think you can trust most news most of the time" and "I think I can trust most of the news I consume most of the time."The five-point-scale had a range from 1 = strongly disagree to 5 = strongly agree.In addition, we calculated the difference between these two items: Higher positive values indicate a larger gap between trust in news media the respondents use themselves, and trust in media in general.In regard to perceived functions of news media, we used means of the level of agreement (from 1 = strongly disagree to 5 = strongly agree) with the following items: "The news media monitor and scrutinize powerful people and businesses" ("watchdog"); "The topics chosen by the news media do not feel relevant to me" (recoded, higher values indicate higher relevance of topics, "relevance"); "The news media often take a too negative view of events" (recoded, higher values indicate higher satisfaction with the tone; "tone"); "The news media keep me up to date about what's going on" ("immediacy"); "The news media help me understand the news of the day" ("depth").
Table A1 lists all these indicators for the 42 news brands.In the first step of our analysis they serve as descriptive indicators for individual media brands' performance.In the second step, in order to identify distinct patterns of news media performance, we ran a hierarchical cluster analysis based on standardized values of these indicators.It is important to note that this methodological approach takes a media-centered perspective: We describe news media by characteristics of their aggregate audiences.From a user-centered perspective, we know that most users use more than one news medium and compose more or less complex news repertoires (Hasebrink, 2017); thus the aggregate audiences as described in our study partly overlap because some users belong to several audiences.However, this does not confine our argument to the idea that single news outlets' performance can be assessed and compared based on characteristics of their respective audiences.

News Brands' Reach
The key indicator for any kind of research on media audiences is the number of individuals who actually use them.A wide reach indicates that an outlet has a high potential to have a communicative impact.For each of the selected news outlets, Table A1 shows the percentage of people who said they have used it in the last week (column 1).Tagesschau, the daily news broadcast provided by the public service broadcaster ARD, has the highest reach (48.2%), followed by the aggregate category of regional newspapers (33.8%) and other tele-vision news formats.The aggregate categories of regional public and private radio channels reach about one-fifth of the population, respectively.Among the printed news media, the tabloid BILD has the widest reach (11.0%), other national newspapers, often considered as the "quality press," reach less than half of this figure.The four major weekly print magazines reach between five to ten percent of the population.In regard to online news sources, the online services provided by a number of magazines (Spiegel Online, Focus Online), newspapers (bild.de),and television broadcasters (tagesschau.de,n-tv.de)reach more than one-tenth of the population.In addition, the general platforms of internet service providers (t-online.de,web.de, gmx.de) play a significant role in Germans' news consumption.

Structure of Audiences
As shown above, some discussions on news media's performance refer to the question whether these brands reach all parts of society or if they are instead targeted to specific groups.In normative terms, this distinction is quite ambivalent since there are good reasons to value both of these objectives.On the one hand, reaching all parts of society corresponds with democratic and pluralist values; it reflects an inclusive communicative function.On the other hand, targeting news to the specific needs and preferences of particular groups can help to better reach these groups and to support them in building their own opinion.
In order to illustrate this argument we selected the age distribution of news brands' audiences as an example.Table A1 presents the reach of the news brands among respondents between 18 and 24 years-old (column 2) and older than 65 years-old (column 3).Column 4 shows the difference between these percentages in terms of how much higher the reach in the group 65 years-old and older is than in the group between 18 and 24 years-old.Column 5 shows the average age of the audiences.
Television news audiences are characterized by a substantially wider reach in the older group than it is in the younger group; thus the average age of their audience is higher than for other news media.A significant exception is ProSieben Newstime.In line with this private channel's strategic orientation towards young audiences, its news broadcast has a wider reach in the youngest group.As for radio and print media, the differences between the two groups are rather small.In many cases, online news' reach is higher in the younger group; this is particularly true for online news offered by print media brands.The opposite is true for the news offered by internet service providers that have a wider reach among the older age group.
As previously discussed, it remains debatable whether a balanced reach in different parts of the population is an objective in itself or if it is preferable that news media try to reach specialized audiences in order to fulfil specific needs and preferences.So, one might say that ARD Tagesschau, despite its high reach in the total population, fails to reach young audiences-the reach in the oldest group is more than 40 percentage points higher than it is for the youngest group.In this case, this discrepancy between the general reach and the reach among young people can be alleviated by the fact that, within the youngest group, ARD Tagesschau still has a wider reach than all other news sources.However, it is worthwhile to consider news brands' ability to reach specific audiences that tend to use less news as a relevant indicator for media performance-beyond the general reach.

Interest in News and Politics
In this section, we take as our starting point that news are made for people who are interested in current affairs.We can, therefore, regard it as an indicator for news media performance when the people they reach frequently use news and are highly interested in it.However, similar to the discussion in the previous section, one might argue that an even more important indicator of performance is the extent to which these news media are able to reach those audiences who are less interested.In regard to this issue, Table A1 shows three indicators: the average frequency of news use (column 6); the average interest in news (column 7); and the average interest in politics (column 8).
As far as frequency of news is concerned, the first observation is that almost all news brands' audiences use news more often than the overall population (mean = 6.73).This is a logical result of the fact that these audiences are defined based on the statement that they have used a particular news medium.With that in mind, it is rather surprising that the audiences of the three private television stations' news output barely reach the average of the whole population, which means that the news audiences they reach are less likely to use news than the audiences of all other news brands.On the opposite end of this scale, there is a clear finding that users of online news, particularly those originating from print media brands, are the most frequent users.
The second indicator, interest in news (Table A1, column 7), correlates highly with the previous one (r = .75,n = 42, p < .01).Again, audiences of private television news are the least interested, together with regional newspaper Rheinische Post, tabloid BILD, and one of the internet service providers.What these media seem to have in common is that they are used for reasons other than news.On the opposite side of this scale, we find online news that originate from television broadcasters and print media, and the national newspapers that are regarded as the "quality press." The third indicator, interest in politics (Table A1, column 8), is also highly correlated with frequency of news use (r = .77,p < .01),and correlates even more strongly with interest in news (p = .89,n = 42, p < .01).The main observations mentioned above, therefore, hold for this indicator, as well: The audiences of online news that originate from print media and television broadcasters as well as national newspapers are most interested in politics, while the audiences of private television news, tabloids, regional newspapers, and internet service providers are least interested.There is a moderately negative correlation between interest in politics and reach (r = −.35,n = 42, p < .05)which indicates that news brands with larger audiences reach a broad range of people that tend to be less interested in politics, while some more specialized brands with smaller audiences reach those people who are particularly interested.
In view of the low and, with the exception of interest in politics, insignificant correlations between reach and the three indicators discussed in this section, we can conclude that frequency of news use and interest in news and politics represent an independent dimension of news media performance.Some news brands-for example, public television news-perform highly by reaching a large audience, while others-for example, online news provided by print or television organizationsperform highly by serving the expectations of people who are particularly interested in news and politics.In addition, one might argue that some media brandsfor example, private television broadcasters or tabloidsdemonstrate high performance by reaching those people who are less interested news and politics.Again, we are confronted with the ambivalence of indicators for media performance: Depending on the exact objective, it may be worthwhile to serve those who have a particular interest in news, or to reach those who, due to a lack of interest in news, would otherwise not consume any news content at all.

Trust in News
As discussed in the theoretical section of this article, trust in news can be regarded as one potential outcome of using a particular news brand and, as such, indicate a certain level of performance for that brand.With respect to this concept, Table A1 shows findings for trust in news media in general (column 9) and for trust in the media the person actually uses (column 10).In addition, we calculated the difference between the answers to these questions, which indicates the perceived gap between news that are actually used and news in general (column 11).This gap might be regarded as a particularly interesting indicator for news media performance.
Starting with general trust in the news media, the average value of this indicator for the whole sample tends slightly to the positive end of the scale (mean = 3.22).The fact that the general trust of the audiences of most news brands is higher than this average indicates that news users tend to trust more in news.However, there are some audiences below this average: Interestingly, at the lower end of this scale, we find national newspapers (Handelsblatt, Welt) and magazines (ZEIT); on the upper end of the scale, we observe a mixture of online news from television broadcasters, public television news and regional newspapers.Obviously, trust in the news media that are actually used (Table A1, column 10) is moderately higher (mean = 3.53).This indicator is strongly correlated with general trust (r = .81,n = 42, p < .01),there are, therefore, no substantial differences between the news brands with the highest and the lowest values.
The difference between trust in the media that are actually used and media in general (Table A1, column 11) can be interpreted as the perceived gap between the performance of one's preferred news media and the media that are used by the average population.In the top ranks, we find national newspapers and magazines, online and offline, and online news provided by public television outlets.On the other hand, indicating a low gap between one's preferred media and media in general, there are private television news, online and offline, and, more difficult to interpret, the national newspaper Tageszeitung.This indicator is strongly correlated with all three indicators in the previous section, i.e., news frequency (r = .55,n = 42, p < .01),interest in news (r = .45,n = 42, p < .01)and in politics (r = .51,n = 42, p < .01).We can conclude from these data that news media audiences that are more interested in news and politics also tend to perceive a wider gap between media in general and their preferred media.

Perceived Functions of News Media
For the final step of this analysis, we examine the degree to which audiences of different news brands believe that the media in general fulfil certain journalistic functions: "watchdog" (see Table A1, column 12); "relevance of topics" (column 13); "tone" (column 14); "immediacy" (column 15); and "depth" (column 16).If the audience of a news brand strongly believes that the German news media fulfil these functions, we conclude that this news brand is likely to contribute to this function-or at least to suggest that the media demonstrate a high level of performance on these terms.
Across the news brands, "watchdog," "immediacy," and "depth" are highly correlated (r > .65,n = 42, p < .01).These items also have a strong correlation with trust in news in general and trust in used news.Consequently, these functions, together with trust in news, represent the core of the perceived performance of the news environment.With regard to the audiences of specific brands, the perception of the "watchdog" function is strongest for audiences of online news from television broadcasters and of television news, while newspaper and magazine readers are most critical in this respect.High "immediacy" is perceived by audiences of online and television news while low "immediacy" is perceived by the readers of some newspapers.
The perception that the topics chosen by the news media feel relevant (Table A1, column 13) is moderately correlated with "immediacy" and "depth" (r > .39,n = 42, p < .05).In addition, it is correlated with the overall reach (r = .45,n = 42, p < .01), the frequency of news use (r = .41,n = 42, p < .01),and trust in media used (r = .49,n = 42, p < .01).This pattern indicates that the personal relevance of topics is the function that is most closely linked to the likelihood to use news and to trust these news media.This perception is, therefore, highest among the audiences of news media with a high reach, for instance public service television and radio news and their online affiliates, and some online news provided by magazines (ZEIT Online and Spiegel Online).
The fifth function "tone" is not correlated with any of the other functions.Overall, the differences between the news brands are rather minor, indicating that this item cannot contribute to a reliable assessment of news media performance.

Patterns of News Media Performance
The findings presented so far support the assumption discussed in Section 2 that media performance cannot be operationalized as a one-dimensional concept.While some of the proposed indicators are correlated, others are not.Together with the argument that some of these indicators are inherently ambivalent, this observation leads to the question of whether or not we can identify specific patterns of media performance.In order to answer this question we conducted a hierarchical cluster analysis (Ward algorithm based on z-standardized values) on the basis of the indicators presented above.After an inspection of solutions between two and ten clusters with regard to the Euclidian Distance between clusters, and the interpretability in terms of specific media types, we decided to work with the seven clusters solution (Table 1).
In Cluster 1, we find the four news broadcasts that are offered by German public broadcasters and the generic category "regional newspapers."Key characteristics are the very high reach, particularly among older groups resulting in a high difference between the age groups; but even among the youngest group, they reach the second highest value.While news frequency and interest in news and politics are slightly below average, these audiences demonstrate a degree of trust toward news media in general and particularly toward the media they use on a regular basis.In regard to the perceived functions of news media, they have high values in the relevance of topics, immediacy, and depth.
Cluster 2, the largest one, includes private television news (3), the generic category "private radio," the tabloid BILD, television news channels (2), and online platforms (3); and the two selected examples for regional public radio and regional newspapers.Their reach is rather high, older groups are better reached than younger groups.As the most evident characteristic, these audiences are least interested in news and politics.Indicators for trust and perceived functions are slightly below the average.Public radio, online news from magazine news brands (3, except stern.de),and the online affiliates of public television news program Tagesschau and the national newspaper Süddeutsche Zeitung make up Cluster 3. Together, they have the highest reach among the youngest group, while the reach among the oldest group is lower.Frequency of news use is very high, the same is true for trust in the news they use-the high difference between trust in used news compared to trust in media in general indicates a strong belief among these audiences that the media they use are more trustworthy than other media.In addition, similar to Cluster 1, these audiences are more likely to perceive a high relevance of topics and immediacy of news media.
Cluster 4 includes national newspapers (4, except Tageszeitung) and the weekly newspaper ZEIT together with, perhaps unexpectedly, online news from N24 and BILD.The average reach is rather low, with almost no difference between the youngest and oldest group.The audiences of these news media are very skeptical about news media in general and also about the media they use: they trust them less than any other cluster.This goes along with rather low values regarding the perceived watchdog function, the relevance of topics, and immediacy.
In Cluster 5 we find printed magazines (3), the daily newspaper Tageszeitung, and online news from different kinds of media, n-tv, Stern, and ZDF Heute.Their reach is rather low, with no difference between younger and older groups.Their audiences are slightly above average with regard to their interest in news and politics, and they are quite trustful towards media in general and used media.In addition, they are more likely to perceive a watchdog function and immediacy of the news media.
Cluster 6 includes three of the four online newspapers.As for Clusters 4 and 5, the overall reach is rather low; however, they have a clear bias towards younger audiences.Frequency of news use and interest in news and politics are higher than in any other cluster, while trust in news media in general is rather low.For this cluster, we observe the highest difference between trust in used media and in media in general.With regard to the functions of news media, these audiences seem to lack the watchdog function and in depth news coverage that would help audiences to better understand the news of the day.
Cluster 7, the smallest cluster, includes the online news offered by private broadcasters (2).Their reach is very low, particularly among older groups.These audiences are low in terms of frequency of news use and interest in politics, but highest in trust.The difference between trust in used media and media in general is lower than in any other cluster.As for perceived functions, they have the highest values regarding watchdog and tone, and the lowest value regarding relevance of topics.
The seven clusters underline that news media performance has to be conceptualized in a multi-dimensional way: There are different ways for news media to perform.While some news media are good at reaching a large audience (e.g., Cluster 1), others manage to inform younger audiences (e.g., Clusters 4 and 6).Some media are good at serving those who are particularly interested in news and politics (e.g., Cluster 6), others successfully reach people who are less interested (e.g., Cluster 2).Some media contribute to the impression that media fulfil a watchdog function (e.g., Clusters 5 and 7), while others contribute to the impression that news media select relevant topics (Clusters 1 and 3).We will discuss the implications of these patterns below.

Conclusion
In this article, we have proposed a conceptual framework for the definition of audience-based indicators for news media performance that allow for the assessment of individual news providers' contribution to the news environment.This framework includes the overall reach of a news medium, the structure of its audience-in this case illustrated by different age groups-the audiences' motivation to use news, their trust in news media, and its perceived functions.Based on these indicators, we identified different patterns of media performance that reflect the multi-dimensional character of normative expectations.
Before we discuss the findings we have to emphasize a couple of limitations of our approach to conceptualize news media's performance from an audience perspective.First, there is no coherent theoretical approach that can guide the definition of indicators.While we have built the general framework on the basis of conceptual distinctions from audience and reception studies, the selection of specific indicators had to be done in an exploratory way-and within the range of items that have been used by the Reuters Institute Digital News Survey.Second, in this survey, trust and perceived media functions are measured by single items and not by established scales, which might go along with a lack of reliability and validity.Third, we have no data on audiences' direct evaluation of the news media they use; for these reasons we had to use indirect indicators for the perceived performance of the news media in general.And fourth, applying hierarchical cluster analysis is an exploratory process that depends on the news media included in the sample, the set of indicators, and the exact algorithm.It is necessary, therefore, to carry out corresponding analyses for other data sets.
Despite these limitations, we believe that our study can contribute to a more comprehensive and meaningful discourse on news media performance which includes the audience perspective.An initial, and somewhat striking, finding is that the clusters of media performance are largely characterized by specific media types.In most cases, news brands that belong to the same type of medium have been attributed to the same cluster.Furthermore, there does not seem to be very high commonalities between offline and online news offered by the same company.These observations stress that the distinction between different types of media-with respect to technical and organizational issues-remains relevant in the digital news environment.In this respect, the process of convergence does not level out the established images of news brands and media technologies.
The observed patterns of media performance reflect the fact that there are different ways to contribute to public communication: While traditional television news and regional newspapers perform high by informing large parts of the population, online newspapers' and magazines' contribution relates to the fact that they reach specific groups such as, in this case, young people.Online newspapers best serve the interests of those who are particularly interested in news and politics, while private television and private radio as well as tabloid newspapers and online platforms succeed in disseminating news to those who are less interested.Television audiences perceive the news environment in general as rather trustworthy, readers of national newspapers are most skeptical in this respect.Readers of online newspapers and magazines have much more trust in the media they use than in media in general.Readers of magazines and some online news appreciate the watchdog function, the immediacy, and the depth of news media coverage, while audiences of public radio and television and of online magazines appreciate the relevance of the topics as offered by news media.
In regard to our objective of developing audiencebased indicators for media performance, these findings could be considered frustrating, since the performance of media is often intuitively imagined as a onedimensional scale between "good" and "bad."The reality of public communication is actually much more complex.The fact that we do not find this one-dimensional scale and that there are different ways to perform should not lead to the conclusion that everything that news media do is good.As we have argued above, we define news media performance as communicative impact in regard to normative expectations concerning public communication.As we have argued and as the findings reflect, normative objectives might be ambivalent or at times conflict with each other.It is necessary, therefore, to understand that there are different ways to "perform" and that different kinds of news media contribute to the overall news environment in specific ways.
The proposed conceptual framework for the assessment of audience-based indicators for media performance can serve as a multi-dimensional benchmark for assessments of and public debates on individual news media's performance.For instance, public television and regional newspapers (Cluster 1) might be satisfied with their wide reach and the finding that their audiences think that news media in general succeed in covering relevant topics.At the same time they should consider how to better reach young audiences and to contribute to their audiences' perception that news media fulfil their watchdog function.As another example, we can look at the largest group of news media including private television and radio, tabloid newspapers, and online platforms (Cluster 2), that reach audiences that are least interested in news and politics.From a normative point of view, the potential argument that they succeed in serving people who are less interested in news only holds if it can be shown that they actually offer reliable news that help their audiences to better understand the worldthis means we need to embellish the analysis with information on the actual content.
This leads to future perspectives for our research.We are currently preparing further analyses in the following directions.First, the operationalization of our conceptual framework has been based on the German data collected as part of the Reuters Institute Digital News Survey.This gives us an opportunity to extend the analysis to other national news environments and to compare these environments in regard to the respective patterns of media performance.Second, considering a comprehensive model on media performance, in this article we have excluded the role of content as provided by different news media.The next logical step is to fuse audience data with data from content analyses of the same news media.As part of the collaborative project "Media Performance and Democracy" (https://en.mediaperformance.uni-mainz.de)we are currently working on merging the data presented here with content analyses that have been conducted on the most relevant news media in Austria, Germany, and Switzerland (Weiß et al., 2016).This will allow for an empirical link between normative indicators for the potential performance of single news media and indicators on performance as perceived by their respective audiences.This approach will assist in further developing a continuous monitoring of news media performance in changing news environments.Given the serious challenges that news media currently face, economic pressures especially, political interventions in media freedom, a high degree of media skepticism, and the emergence of socalled "fake news," these kinds of continuous efforts to observe news media's performance will be an important task for future research.

Figure 1 .
Figure 1.Conceptual framework for audience-based indicators of news media performance.

Table 1 .
Clusters of news media based on performance indicators.: Base = 2,022 respondents; * = highest and # = lowest values of the respective indicators.For details regarding the underlying scales, see TableA1(in the Supplementary File). Notes