Two Journalisms? Linear and Curvilinear Relationships Between Journalists’ Role Ideals and Degree of Democracy

ABSTRACT If there are authoritarian, transitional, and democratic political and media systems, are there associated authoritarian, transitional, and democratic role profiles for journalists? The relation between countries’ degree of democracy and the importance of journalistic role conceptions among journalists in a country is analyzed based on the World Journalism Study (Wave 2). I find that journalists in democratic political systems generally subscribe to fewer role conceptions than journalists in authoritarian political systems; Journalism-for-Democracy is more defined by what it is not (i.e., journalists reject the typical role conceptions of Journalism-for-Authoritarianism and Journalism-in-Transition) than by what it is. Journalism in transitional political systems (i.e., hybrid regimes and deficient democracies) is characterized by a specific set of role conceptions that are linked to political activism, which journalists in full democracies tend to reject. The study reassesses which role conceptions might be typical for which kind of political systems leading to revised hypotheses regarding the link between democracy and journalistic role conceptions.

empirically how journalists' role conceptions correspond to a country's degree of democracy. Whatever relationship this study finds, it is most likely the outcome of a complex nexus of interdependent processes and should not be interpreted in causal terms, one way or the other (Zelizer 2013).
Journalists, journalism, and news media are regarded an integral part of modern liberal-representative democracies. They-or the "public sphere" that is unthinkable without news media-take center stage in normative theories of democracies as well (e.g., Chambers 2003). However, this does not mean that journalism only exists in democracies. Quite to the contrary. Nazi Germany had its journalism (Fröhlich and Holtz-Bacha 2003), the Soviet Union had its journalism (Siebert, Peterson, and Schramm 1963), and contemporary non-democratic regimes have "their" journalism, too (e.g., Tong 2019). This insight was recognized early-on (Siebert, Peterson, and Schramm 1963), but the relations between role conceptions and regime type remain to be explored and mapped.
In a nutshell (and with a bit of exaggeration of the normative undertones), Siebert, Peterson, and Schramm (1963) describe two species of journalism: A «good» type of journalism that is typical for Western democracies (distinguishing the libertarian and the social responsibility theories of the press), and a «bad» type of journalism that is typical for authoritarian regimes (distinguishing a generic authoritarian and a Soviet Communist theory of the press). While many have criticized the "Four Theories", it is apparent it is one of the most influential books on media systems ever written (Rantanen 2017). And its basic bipolar understanding of media systems (and associated "journalisms") in democracies versus authoritarian regimes (see also Ogan 1982) has rarely been challenged. Therefore, it is worthwhile to use the data collected in the second wave of the WJS (WJS2) to test whether the assumption of bipolarity of media systems finds expression in differences in how journalists view their role.

Roles and Role Perceptions
Roles are a set of expectations towards how an individual or group who takes the role (or is attributed the role) should behave (Hanitzsch and Vos 2018). It is external in the sense that it is about what "others" or "society" or "those-in-charge" expect an individual to do, but they are internal in the sense that it is one's own perception of the expectations of others that shape behavior (Ajzen 1991). Role expectations and role perceptions are not the only influences on human behavior-they are one important factor that is taken into consideration. For instance, the Theory of Planned Behavior (Ajzen 1991) views role expectations (called "subjective norms" there) as a main factor influencing behavior.
I use role expectations as the most important individual-level indication of what journalistic norms are in different countries. Often, expectations and norms are discussed at systemic or institutional levels ("the press", "the media") but in the end will have to be enacted at the individual level according to the hierarchies of influences model (Shoemaker and Reese 2014). This justifies studying expectations at the individual level and aggregating them to the country level, despite substantial within-country variation of role perceptions. I presume that, ceteris paribus, role perceptions of journalists will powerfully shape newswork and in aggregate contribute to a highly different news environment across countries. If regime type and journalistic role perceptions are closely correlated, this would manifest in a sharp contrast between news in countries with democratic versus authoritarian regimes.

Existing Typologies of Role Perceptions
Systematic investigations into journalistic roles perceptions started in the United States (Johnstone, Slawski, and Bowman 1976;Weaver, Willnat, and Wilhoit 2019) and slowly widened to other Western countries (e.g., to Germany: Donsbach and Klett 1993;Köcher 1986). These studies, unsurprisingly, focused on roles that are salient in these democratic countries, either because they play a role in the justification of journalism, or because they are contested. Weaver, Willnat, and Wilhoit (2019) classify the roles included in the US surveys 1971-2013 into "watchdog", "disseminator", "mobilizer", and "adversarial" functions. A major question was to what degree journalists would focus on "neutrality" and "objectivity" in their coverage versus monitoring or opposing the powerful. Comparative surveys that include the United States and several European countries put more emphasis on the debate between subjective/interpretative versus objective reporting, the degree of partisanship, and how it is enacted (Donsbach and Klett 1993). This already suggested that adding new countries with different political systems widens the spectrum of roles that researchers consider. WJS1 and WJS2 vastly expanded the number and diversity of countries it included and hence added a whole new set of role conceptions, culminating in a more encompassing conceptualization of 18 roles categorized into 6 wider classes: (1) informational-instructive roles, (2) analytical-deliberative roles, (3) critical-monitorial roles, (4) advocative-radical roles, (5) developmental-educative roles, and (6) collaborative-facilitative roles (Hanitzsch and Vos 2018). Classes 1-3 were part of the set of relevant role conceptions included in surveys from the outset. Classes 4-6 only slowly emerged in theoretical discourse (and empirical measurement) as the inclusion of developing countries and non-democratic countries broadened the view. Classes 1-4 dominate in the open-ended responses by European journalists (Standaert 2022). Classes 5-6 are somewhat more visible if the entire set of WJS2 countries is used (Standaert, Hanitzsch, and Dedonder 2021).

Normative, Cognitive, Practiced and Narrated Roles
Hanitzsch and Vos (2017) have conceptualized the relationship between normative roles (expectations and ideals), cognitive roles (internalized norms), practiced roles (behavior in a role), narrated roles (discourse about roles), which can be imagined as a feedback loop: (1) normative roles are internalized into (2) cognitive roles, cognitive roles are enacted into (3) practiced roles, reflection on practiced roles leads to (4) narrated roles, and narrated roles affect the normative roles (normalization). All these stages are bundled in the measures used in WJS2: journalists in the survey will express norm importance in their work as they perceive it. This will reflect their role ideals (normative) but put even more emphasis on roles they have internalized (cognitive) and that they practice (practiced) or performed (Mellado 2015), and they tell us about the roles in the "language" the questionnaire permits them to use (narrated). And respondents may differ in how they mix normative, cognitive, practiced, and narrated role components in their response.

Democracy Indices
Recently, the overemphasis of the conceptual linkage between democracy and journalism has been criticized (Zelizer 2013), and for good reasons. Rather than presuming or predefining journalism and democracy as inherently linked, this study explores the empirical relationship between the two, breaking up the presumed relationship into a more nuanced view: Some aspects of journalistic self-conception are more or less closely linked to degree of democracy in a country (positively or negatively, linearly or non-linearly), while other aspects are unrelated to the degree of democracy in a country (Hanitzsch and Vos 2018), and individual-level variation of role conceptions is considered, allowing a more realistic and nuanced assessment of how journalistic role conceptions are linked with democracy.
Defining democracy and finding valid indicators is inherently difficult. All measurements of a "degree of democracy" depend on the specific understanding of democracy that measure is built around (Giannone 2010). Common criticisms of measures of democracy include the changes in definitions over time, lack of transparency, or even ideological biases (Giannone 2010). For this study, I decided to use the democracy index yearly published by The Economist (a newspaper) [the "Democracy" index] with 2018 data (Figure 1). This is also to ensure continuity with previous scholarship on the relationship between democracy and professional autonomy that uses the Economist index (Hamada 2021;Reich and Hanitzsch 2013). The Economist's index (ranging from 0 = fully authoritarian to 10 = full democracy) is based on answers from country experts to a standardized 60-question questionnaire that capture the aspects of electoral process and pluralism, civil liberties, government effectiveness, political participation, and political culture. The greatest weakness in the index is that the number and choice of experts is not transparent and may differ between countries. The index has substantial face validity and convergent validity with other indexes (Mainwaring, Brinks, and Pérez-Liñán 2001), however. I compared the Economist's index with Freedom House's "Freedom in the World index" and the various "Varieties of Democracy" indices, and found a correlation of r > .889 between the Economist's Democracy Index and any of the alternative indices (Table A1).

The Two Journalisms
When talking about normatively loaded concepts such as journalism and democracy, it is helpful to explicate any prescriptive claims (Geise et al. 2021). In step one, I treat "journalism" as a mostly descriptive category that is defined by its function in society. In step two, journalism can be subdivided normatively into "good" and "bad" forms of journalism from the normative perspective of democratic theories (Nerone 2013;Strömbäck 2005).

Journalism
First, I seek to establish a mostly descriptive-functional understanding of journalism that minimizes normative elements from theories of democracy. Scholars tend to use the term journalism in different ways (Deuze 2005;Nerone 2013). The broadest and most widespread understanding defines journalism as a profession or a social system centered around the function to choose what information enters the public sphere with a high potential for receiving widespread attention throughout a society. Thereby, it contributes significantly to generating and maintaining the public sphere, feeding it with content (Shapiro 2014). It defines journalism by its social function, not its content, its values, or its formal status (Shapiro 2014), minimizing normative undertones. Therefore, it is important to dismiss the thought that journalism is always normatively "good" or "desirable": In this definition, e.g., Nazi Germany had its journalism that fed its public sphere (Young 1987), but one that is viewed as (highly) undesirable from the normative standpoint of theories of democracy.
Western scholars tend to view the Western ideal of what journalism is and should be as "journalism" in general, but it is just one specific shape that it can take, and one ideal it can follow (Nerone 2013;Voltmer 2013). The conceptual link between democracy and journalism is socially constructed (Zelizer 2013) or applies only to a rather narrow definition of journalism.
"Good" Versus "Bad" Journalisms As a second step, I will add a normative viewpoint based on theories of democracy that distinguished "bad" and "good" forms of journalism. Roughly viewed, there are two larger classes of "ideals".
One, the ideal according to procedural and competitive democratic theories (Strömbäck 2005) where journalism reports independent of whether it supports or harms the current government (through the lens of democratic theory: "good" journalism). Citizens can adequately observe the competence and responsivity of the current government. Thereby, journalism facilitates peaceful democratic government change if the performance of the current government does not generate a majority in a well-informed citizenry (Nielsen 2017). I call the first ideal Journalism-for-Democracy. This ideal of what journalism should be like to fulfill its basic functions in a democracy is associated with specific role expectations. This results in an ideal role profile for a Journalist-in-Democracy.
Two, the ideal of authoritarian systems where journalism is supposed to unconditionally support the current regime and its policy (through the lens of democratic theory: "bad" journalism) (Ogan 1982). It hinders peaceful government change (for which institutions may be lacking anyway) and at any cost attempts to stabilize the current government. In many ways, this (historically older) "counter draft" is the ideal how journalism should be in an authoritarian regime: Journalism-for-Authoritarianism (McNair 2009;Siebert, Peterson, and Schramm 1963). It is also associated with an idealized role profile for a Journalist-in-Authoritarianism.
I do not assume that all journalists in a democracy [authoritian regime] subscribe to the role profile of "Journalism-for-Democracy" ["Journalism-for-Authoritarianism"]. For instance, there will be some journalists who work towards authoritarianism (or according to an authoritarian logic) in a democracy. The connection between democracy (authoritarianism) and Journalism-for-Democracy (Journalism-for-Authoritarianism) is a normative-ideal connection because they are assumed to "match" and complement each other.

Journalism-for-Democracy (D-Roles)
The Journalism-for-Democracy role profile is characterized by the implementation of freedom of the press, free flow of information, free competition of ideas in the "public sphere", journalism as a "fourth estate" as a guardian of democracy and sentinel against abuse of power (McNair 2009;Siebert, Peterson, and Schramm 1963;Voltmer 1999). It is conducive to a well-informed and politically competent electorate fit for electing their representatives based on representatives' performance. If the current leadership loses legitimacy or proves ineffective, it will be removed after the next election at the latest (Dahl 2000), or resign in anticipation of electoral disaster.
It would correspond to specific individual-level role conceptions for what journalists should do and which functions they should fulfill: (a) Provide timely and relevant information dissemination (Voltmer 1999). (b) Provide coverage in terms of "objective" selection and presentation of issues, information, and protagonists (Schudson 2001). (c) View themselves as agents of a "fourth estate" that holds accountable powerholders for any misconduct and abuse of power (Voltmer 1999). Journalists would not view their coverage as actively influencing public opinion, but rather as helping the audience to form their own opinions based on the best available information and arguments (Voltmer 1999). This does not preclude that journalists practice partisan journalism and take an active role in opinion formation; for instance, the emphasis of issues, facts, and arguments can vary, and different conclusions can be drawn from the same facts. Partisan and active roles are compatible with democracy and a specific understanding of "objectivity" that also allows for interpretative approaches to journalism (Donsbach and Klett 1993) as long as it is a part of a larger system of news organizations that together represent all major ideologies (external pluralism). Table 1 summarizes which indicators in WJS2 are expected to represent the Journalism-for-Democracy role profile (+). I call these roles "D-roles". They mostly match the "informational-instructive", "analytical-deliberative", and the "critical-monitorial" classes of role conceptions (Hanitzsch and Vos 2018). High autonomy [D1, D2] reflects freedom of the press. There is empirical evidence of a positive relationship with degree of democracy from the WJS1 (Reich and Hanitzsch 2013) and WJS2 (Hamada 2021). The marketbased financing of journalism (viewed as a cornerstone of its autonomy from politics) requires maximizing the audience [D7]; securing a wide audience is also key in most public-service news organizations. Motivating people to participate and offering a forum for opinion expression [DT2, DT3] reflects that the citizens should participate in the democratic political process; open deliberation requires tolerance (e.g., to cope Table 1. Overview of hypothesized relationships: role ideals that match "Journalism-for-Democracy" (D-roles), "Journalism-for-Transition" (T-Roles), and "Journalism-for-Authoritarianism" (A-roles).

Journalism-for-Democracy
Journalism-for-Transition

Journalism-for-Authoritarianism
Hypothesized effect of degree of democracy . According to Journalism-for-Democracy, news reporting would be "objective", a mirror of reality; the media should provide information and analysis that helps citizens form their opinions autonomously [D3, D4, D5, D6]. Finally, the ideal of journalism as a "Fourth Estate" requires active monitoring for potential abuse of power by the government [DT1].
H1: The more democratic a country, the more do journalists in that country endorse the roles typical for the ideal role profile of Journalism-for-Democracy (D-roles).

Journalism-for-Authoritarianism (A-Roles)
Journalism-for-Authoritarianism is characterized by top-down steered communication campaigns that restrict the flow of information and the competition of ideas to those ideas that fit the regime's ideology and help implement regime-advocated policies, stabilize, or solidify the regime. It is designed to instruct the population into believing the authoritarian leadership is legitimate and effective even if that would not be the case. This type of journalism would in many ways be oppositional to the role conception profile in Journalism-for-Democracy. Table 1 summarizes which indicators in WJS2 are expected to represent the Journalism-for-Authoritarianism role profile (+). I call these roles "A-roles". Journalismfor-authoritarianism actively utilizes the power of journalism and its effects in the interest of the current political regime. The associated role conceptions mostly conform to the "collaborative-facilitative" and to a lesser degree to the "developmental-educative" and the "advocative-radical" classes of role conceptions (Hanitzsch and Vos 2018).
Conveying a positive image of political leadership [A1], supporting government policy [A3], and supporting national development [AT4] are the most obvious instances of authoritarian roles-they combine the bringing about of media effects and a regime-supporting goal.
Influencing public opinion [AT2] and setting the political agenda [AT1] are less obvious instances of authoritarian roles as they endorse bringing about media effects but do not explicitly mention a regime-supporting goal. I still classify these roles as authoritarian for three reasons: (1) The roles' focus on goal-directed media effects holds the potential of being interpreted and exercised in an authoritarian fashion. (2) Exerting goal-directed influence has an inherent authoritarian component, namely that one knows what is best for the audience and has a mission to influence the audience into that direction; somebody else (the one who sets the goals) determines what is best for the citizenry, and journalists are expected to exert their power to promote this way of thinking. (3) Supporting the current government is one of the most likely goals (yet not the only one) that journalists may want or be pressured to pursue when exerting this influence. H2: The more democratic a country, the less do journalists in that country endorse the roles typical for the ideal role profile of Journalism-for-Authoritarianism (A-roles).

Journalism-for-Transitional-Societies (T-Roles)
Between the two "ideal" types of journalism-Journalism-for-Democracy and Journalismfor-Authoritarianism-there will be mixtures and in-betweens, just like there are deficient democracies and hybrid regimes (Voltmer 2013).
Mixture of D-roles and A-roles. This can manifest as a simple mixture or middle-ground of A-roles and D-roles. Political, judicial, or military attempts to "purge" journalism through repressions against specific outlets or individuals will affect the mixture by removing some persons from the profession, attracting others, and intimidating yet others. Changes in journalism education will change role conceptions gradually. This suggests that countries with a medium democracy score will display a moderate level of both authoritarian and democratic role ideals; this may reflect either many journalists who moderately endorse these roles, or a polarization where some endorse and some others refute a role conception.
Unique T-, DT-, and AT-roles. But in addition to this "mixture hypothesis", there is good reason to believe that transitional political systems would give rise to an idiosyncratic set of typical "transitional" role conceptions.
The main reason is that "transitional" countries are likely to swing either into the more democratic or the more authoritarian direction in the proximate future (Mufti 2018), and journalists have incentives to try to influence where the country is heading. Neither side would be particularly happy with the hybrid regime. The trajectory of democratization is usually not linear: trends towards democracy or authoritarianism are not one-way roads but may stall or reverse, and journalists and play a role in the process and the outcome where a country is heading. This tipping-point nature and uncertainty where hybrid political regimes are heading would plausibly leave traces in journalists' role self-conception.
I expect a fierce ideological activist journalism that participates in a "battle of ideas". It would lead to a journalism that is more militant, combative, polarized, and effectoriented. Transitional role (T-role) conceptions seek to exert the media's power in a way that could plausibly be used to facilitate regime change or transformation of the political system in either direction. Since these activist journalists would push into different political directions, only role conceptions with no or an open-ended ("change", "national development") goal formulation are likely to emerge as T-roles.
Another motivation behind pursuing T-roles is self-preservation: should the political system tilt into the opposite direction, journalists have to fear persecution, e.g., in terms of murder, imprisonment, or occupational bans; and with a combative role ideal, they may improve the chances that "their" side is winning.
Some T-roles will overlap with D-roles (DT-roles): Being a watchdog for political leaders [DT1] can reveal misconduct by leaders. This can serve as a powerful tool to push for political change. Also, mobilizing people (e.g., to participate in protests) [DT2] and creating conditions for free expression of opinion (for one's own side) [DT3, DT4] can strengthen political movements that push for political transition.
There are also pure T-roles that I expect are typical of journalism in transitional societies and untypical in established democratic as well as authoritarian regimes: Being an adversary of the government [T1] is clearly a revolutionary role conception, and being an advocate for social change [T2] expresses an inherent motivation towards stimulating regime change.
Finally, there are roles that overlap with authoritarian roles that do not specifically mention the goal of supporting the current regime: setting the political agenda (putting one's side's issues on the agenda) [AT1], influencing public opinion (to support one's side) [AT2], and supporting national development (by stimulating regime change) [AT3]. Utilizing the media's power to stimulate political change is the common denominator behind these role orientations. I expect a generally higher importance of Troles in countries with a moderate degree of democracy (Table 1). H3: In countries with a moderate degree of democracy, journalists endorse the roles typical for Journalism-for-Transition (T-roles) more than in countries with a high or a low degree of democracy.
This expected curvilinear (inverse U-shaped) relationship between degree of democracy and importance of these roles can be asymmetric, either skewed towards democracy (DT-roles: popular in transitional systems and in democracies) or skewed towards authoritarianism (AT-roles: popular in transitional systems and authoritarian regimes).

Universal Journalistic Roles (U-Roles)
There are other roles that are less political in the sense that they are less vulnerable to being politically instrumental, neither from a democratic nor from an authoritarian perspective. I therefore expect that endorsement of these role conceptions is more individualized and does not systematically co-vary with the degree of democracy in the country. This expectation relates to the roles conceptions to "educate the audience" (U1), "provide advice, orientation and direction for daily life" (U2), "tell stories about the world" (U3), "provide entertainment and relaxation" (U4), "monitor and scrutinize business" (U5).

Method
Design I conducted a secondary analysis of the second wave of the WJS2 which is a standardized survey of 27,567 journalists from 67 countries. The field guide for the study specifies that journalists should be professionals who earn half or more of their income from their journalistic work, and that only media that regularly publish news should be included. Outlets should be chosen to represent the structure of the media system in terms of geography and types of outlets. Four different interviewing modes (online, mail/e-mail, telephone, personal) were used. The field guide recommends personal or telephone interviews, but country partners could deviate from the recommendation for practical reasons and even mix modes. Social desirability bias typically affects results more in personal or telephone interviews compared to online or mail responses (Kreuter, Presser, and Tourangeau 2008). Also, the political regime type will most likely have affected the propensity to participate in the study, as well as the answers, e.g., through fear of retribution. Table A2 documents all countries, the sample size in the country, and the dominant mode of interviewing used. I added country-level data for degree of democracy in a country as rated in the Democracy index (The Economist Intelligence Unit 2019). The WJS2 coverage of countries (n = 67, see Figure 2) ensures substantial variation in both democracy indices and journalists' role conceptions.

Measurement
This study draws on three questions in WJS2: Perceived autonomy for selecting news stories (C9), perceived autonomy in selecting aspects within news stories (C10), and perceived importance of various roles (C12). WJS2 covers a broad set of role ideals that reflects the six-class conception of role ideals in journalism (Hanitzsch and Vos 2018) that includes a vast number of D-role, A-Role, T-Role, and U-Role items. I decided to mix the role-importance questions with these two autonomy questions as experience of autonomy on the job is predicted to be an inherent part of Journalism-for-Democracy. It does not indicate the presence a specific role expectation. Rather, it indicates the absence of role expectations that are experienced as constraining one's leeway in central parts of journalistic work. The questions and items are documented in the online appendix. The full questionnaire is available from the WJS website.

Model-Based Classification
There is an elegant way for testing the hypothesized relationships between roles and degree of democracy: I estimate mixed models with individual journalists nested in countries. This was done for each role ideal importance rating (23 item-level models) and for each class of role ideals (6 class-level models: D, DT, T, AT, A, and U roles). The model features random intercepts for 67 countries (all item-and class-level models) and for the role importance items assigned to the respective class (only class-level models), but no other random components. For each item or class, three such models are compared that include different fixed effects: (1) only an intercept (null model M1: f f (x) = β 0 + a country + b Item ), (2) only the intercept and degree of democracy "x" (linear model M2: f(x) = f (x) = β 0 + β 1 x + a country + b Item ), and (3) the intercept, degree of democracy "x", and degree of democracy "x" squared (curvilinear model M3: f (x) = β 0 + β 1 x + β 2 x 2 + a country + b Item ) with a country ∼N(0,σ 2 a ) and b item ∼ N(0,σ 2 b ) (b item is included only in the class-level models).
Based on these models, I compare models using χ 2 -tests of nested models: If model (2) outperforms model (1) and the sign of β 1 is positive, the role will be classified as D-role. If β 1 is negative, the role will be classified as an A-role. These two roles are mutually exclusive. Second, I compare models (3) and (2) with χ 2 -tests of nested models. If model (3) outperforms model (2) and the sign of β 2 is negative (inverted U), I classify the role as T-role, possibly in addition to an A-role or a D-role component. Third, if neither model (2) adds explanatory power to model (1), nor model (3) adds explanatory power to model (2), I classify the role ideal as U-role: its importance does not seem to respond systematically to the degree of democracy.

Visual Inspection
The model-based tests will be corroborated by visual inspection of plots of country-level averages for each role conception (WJS2 data, y-axis) and the degree of democracy in the respective country ("Economist" data, x-axis), together with curve fitted according to model M3 specified above. For A-roles, curves should continuously point downwards. For D-roles, curves should continuously point upwards. For T-roles, curves should exhibit an inverted U-shaped form: these roles are most important at medium levels of democracy in a country. Other than the more narrowly designed statistical model used (that can only detect linear or curvilinear relationships), a visual inspection can reveal any sort of non-linear relationship in the data. It can thereby rule out the possibility that a linear or curvilinear relationship is not the optimal way to describe the data.

Factoring or not Factoring the Dependent Variable?
Factorizing the dependent variables into latent dimensions was considered to increase the reliability of measures and facilitate display of results. I decided not to factorize the single items for the testing of hypotheses since the items are conceptualized as meaningful single-item measures of theoretical constructs; their theoretical grouping reflects conceptual commonalities (Hanitzsch and Vos 2018) rather than expected empirical correlations. The "Further exploration" subsection in the results section documents results for factorization based on country-level average covariance, however. Thanks to the project's open data policy, it is easy to replicate all analyses made here with the R code I document in the online appendix.

D-Roles (H1)
Contrary to H1, agreement with D-role ideals did not generally increase with increasing democracy score of the country. Entering "degree of democracy" as predictor into the model (M2) does not improve the model (χ 2 (1) = 0.552; p = .457) and the regression weight of degree of democracy is not significant: β 1 (SE) = −0.011 (.014); p = .430 (Table A3).

Visual Inspection
The visual inspection corroborates the conclusions that H1 is rejected while H2 and H3 receive empirical support (Figure 3 center panel, Figure 4 top panel).

Reclassifying Role Ideals for Next Working Hypotheses
In the next step, I visually inspected and ran models for each role ideal separately to make suggestions as to how to re-classify which role ideal responds how to democracy Figure 3. Relationships between degree of democracy (x-axis, 0-10) and importance of role ideals (yaxis, z-standardized), grouped based on hypothesized (left), original (center) and revised (right) role classifications.
( Table 2). The turnover table (Table 3) shows that A-roles and U-roles were more frequent than hypothesized while D-and DT-roles were much rarer than hypothesized. Figures A1  and A2 show the item-by-item fitted curve of the original and the revised role classification. Table A3, Figures 3 and 4 compare the patterns obtained based on the original versus the revised classification.

Dimensions of Journalistic Role Ideals
For further explorations of the patterns between role ideals and democracy, I use the updated role classification I arrived at after testing the hypotheses. The acceptance of D-roles, A-roles, and T-roles in a country serve as indicators of three dimensions of country-level roles ideals in journalism: (1) A democracy compatibility dimension (Droles, with all ideals aiming at autonomy and reality-adequacy) ("Authentic autonomy";   ; T-Role: M3 > M2 and B 2 < 0. *p < .05, **p < .01, ***p < .001. α = .84; ω = .79), (2) an authoritarianism compatibility dimension (A-roles, with most ideals aiming at goal-directed shaping of public opinion outcomes as a kind of "Social Engineering"; α = .93; ω = .69), and (3) a transitional-system compatibility dimension (T-roles, with most ideals aiming at activation and mobilization of citizens, or enabling of their participation) ("Activation"; α = .85; ω = .71). These three dimensions satisfy measurement requirements very well for the country-level averages. However, they do not form empirical dimensions at the individual level (Table A4). Individual-level role perceptions prove to be highly idiosyncratic, but the country-level averages can be mapped well onto dimensions. Scores were extracted as mean indices of the measures included in the respective dimension, which were then z-standardized.

Clusters of Journalistic Role Ideals
Clustering along these three dimensions ([1] Autonomy; [2] Social engineering; [3] Activation) in a model-based profile analysis (using the Mclust package: Fraley and Raftery 2007) yields several useful solutions with good model parsimony (BIC), ranging between 2 and 5 clusters of different shapes. I chose the five-cluster EEE solution (EEE means that in the vector space, clusters are constrained to be the simplest ellipsoids Table 3. Reclassification turnover table between hypothesized and detected relationships between role idea and democracy score. with equal shape, volume, and orientation) because it is most informative (not statistically, but substantially viewed) at a reasonable level of complexity that does not overstrain our analytical capabilities ( Figure A3). Diagnostics of four best-BIC cluster solutions (EEE-5, EEV-4, VEE-4, and VVE-2) are plotted in Figures A4 and A5. Figure A6 shows how the five clusters merge into two clusters step by step-for those interested in the more parsimonious clustering. I label the five clusters (in decreasing order of degree of democracy): (A) Type I: Western Democratic role ideals (USA cluster; n = 21), (B) Type II: Eastern Democratic role ideals (Japan cluster; n = 5), (C) Type III: Unconsolidated Democratic role ideals (Chile cluster; n = 18), (D) Type IV: Unconsolidated Authoritarian role ideals (Egypt cluster; n = 18), and (E) Type V: Consolidated Authoritarian regime role ideals (China cluster; n = 5) ( Figure 5; Table A5). The clusters can be ranked according to the average Economist democracy index score of the countries assigned to the cluster, in decreasing order: (A) Type I Western Democratic . Not all the clusters are clearly distinct from one another in their degree of democracy, however. The Eastern Democratic cluster, for instance, is not significantly different from the Unconsolidated Democratic and the Unconsolidated Authoritarian clusters in its degree of democracy (Table A5). This partly results from the low number of cases in that cluster that leads to wide confidence intervals, but the estimates for degree of democracy are not very different (e.g., between Type II and Type III). The pattern suggests three tiers regarding degree of democracy: Type I countries with "Western Democratic Journalism" have the highest democracy score. Type V countries with "Consolidated Authoritarian Journalism" have the lowest democracy scores. Type II, III, and IV countries (with "Eastern Democratic Journalim", "Unconsolidated Democratic Journalism", or "Unconsolidated Authoritarian Journalism") have medium democracy scores.
The degree to which journalists endorse "social engineering" roles decreases with degree of democracy. The degree to which journalists endorse "activation" roles is affected by two factors: (1) it tends to be greater in authoritarian versus democratic regimes; (2) it is lower in consolidated types (I, II, and V) and higher in unconsolidated types (III and IV). Unconsolidated systems face an ongoing process of political transformation or have recently experienced such a process. "Autonomy", in contrast, is only high in Western and Unconsolidated Democracies (I, III). Unconsolidated Authoritarian (IV) countries have a moderate level of autonomy. Both Eastern Democracies (II) and Consolidated Authoritarian systems (V) have a very low degree of autonomy ( Figure 5; Table A5). The geographical distribution of the types is displayed in Figure 6.

Outlier Analysis
Comparing the clusters, the scores on the dimensions, and the democracy classification of the country yields an interesting map (Figure 7): Endorsement of journalism role ideals and democracy status correspond well, but there are outliers where there is a mismatch between the journalistic role ideals and the political system. That is where data points/ country deviate substantially from the regression lines in Figure 7.
Autonomy outliers. Journalists in Sudan, Turkey, USA, Portugal, and several Eastern European countries express a greater degree of autonomy than the degree of democracy in the country would suggest. This pattern seems to be typical for Unconsolidated Democracies and for Western Democracies with relatively low democracy scores. Norway, Japan, South Korea, Malaysia, Hong Kong, Singapore, Tanzania, and Qatar (this includes all Type II countries) exhibit lower degrees of autonomy than one would expect based on the countries' degree of democracy.
Social Engineering outliers. Regarding the Social Engineering dimension, Russia, Turkey, Hong Kong, France, Italy, Belgium, Czech Republic, and the USA endorse Social Engineering roles much less than expected based on their degree of democracy. This pattern occurs in the democracy-compatible types I-III, but not the two Authoritarian role ideal types IV-V. Journalists from Malawi, Mexico, India, Colombia, and Botswana endorse Social Engineering role ideals more than would be expected based on their degree of democracy. This pattern is mostly found among in countries with Unconsolidated Authoritarian role ideals.
Activation outliers. Journalists in Sudan, Sierra Leone, Tanzania, El Salvador, Mexico, Colombia, Botswana, Israel, Spain, and Chile endorse Activist role ideals more than would be expected based on their degree of democracy. This pattern is mostly found among countries with an unconsolidated political system (III, IV), probably due to the stronger ideological struggle there. Journalists from Russia, China, Qatar, Bhutan, Hong Kong, Hungary, Singapore, Czech Republic, France, USA, Italy, Belgium, UK, Germany, Austria, Switzerland, Netherlands, and Iceland exhibit a more passive set of role ideals than would be predicted based on their degree of democracy. This pattern is typical for countries with consolidated political systems (I, II, V). Mixed types. There are also some countries that are ambiguous regarding their classification into clusters and are best described as a mixture of types: Chile (45% III, 36% IV, 20% II), Cyprus (88% III, 12% I), Greece (70% I, 30% III), Malaysia (87% IV, 13% V), Oman (51% V, 49% IV), Russia (57% II, 42% III), Serbia (80% III, 20% IV), and South Korea (71% II, 29% III).

Discussion
Implications Two Journalisms? This study has shown that there are both roles that are associated with the democracy-authoritarianism dimension of political systems and with a stable-transitional (or: consolidated-unconsolidated) dimension of political systems. The idea of a bipolarity of "two journalisms" is not entirely wrong, but clearly a simplification.
On the democracy-authoritarianism dimension, some (few) role ideals are clearly more important in democracies whereas some (many) role ideals are rated as more important among journalists who work in authoritarian systems. The study also showed that this association between roles and regime types had not been entirely clear before this study, leading some hypotheses to collide with empirical reality.
Classification and re-classification of roles. Figures 2 and 3 show how much more consistent the visual patterns become when we regroup the items based on the data. For now, the updated classification serves to demonstrate how our expectations have changed based on the WJS2 data. A new test with an independently collected set of data is needed to test the revised hypotheses. WJS3 is currently in the field, featuring more than 100 countries around the world, offering an opportunity for such an independent test of the updated hypotheses.
Journalism-for-Democracy/D-roles. In a nutshell, we can say that Journalism-for-Democracy is more characterized by what it is not than by what it is-the absence of clear role ideals and a more passive definition of one's role. Relative to authoritarian systems, journalists in democracies will usually feel a greater degree of autonomy and have a stronger belief in an objective reality that can be reported "as it is" ("Authenticity-driven autonomy"). While this notion is naïve epistemologically, it expresses an important normative role to try to disentangle observable facts from their interpretation. This disentanglement assists citizens in their judgment of the government's performance. Journalism-for-Democracy is also characterized by refuting the importance of exerting goal-directed influence on the audience ("Social engineering"). Journalists in democracies define their role as not being a player in political-ideological struggles and not playing anyone else's game. This conflicts with empirical reality, as researchers have shown that journalists take an active part in political opinion formation (Kepplinger, Brosius, and Staab 1991;Patterson and Donsbach 1996), and often end up supporting government frames and policies (Bennett, Lawrence, and Livingston 2007). However, journalists' self-image states otherwise-and induces some restraint and subtleness in how journalists affect opinion formation in countries where D-roles dominate.
Journalism-for-Authoritarianism/A-roles. Conversely, we found very clear-cut role markers for Journalism-for-Authoritarianism. In countries where A-roles dominate, journalists embracing the effects they can have on the audience and on public opinion to change the world for the better according to their goals ("social engineering"). Journalists in authoritarian regimes (1) are well-aware of their potential impact and (2) view it as an inherent part of their job to utilize these effects for political ends, whatever that end would be. This stands in contrast to some initial expectations I had: mobilizing political activity, providing a forum for expression, providing orientation/direction, promoting tolerance, maximizing the audience, and providing entertainment sound very much like being typical in a democratic political system. However, journalists rate them as more important if they work in authoritarian political systems. Providing entertainment also stands out as an A-role, which may reflect the function of "entertainment-for-narcotization" (Lazarsfeld and Merton 1957) to either distract the population from grievances, or increase their satisfaction, and stabilize the political system.
Journalism for Transitional Societies/T-roles. Finally, some role ideals are particularly pronounced in "hybrid regimes" and "deficient democracies", as the democracy index classifies them. Such transitional role ideals often overlap with those that are typical in authoritarian systems (AT-roles), but are even more important for journalists in systems with intermediate political systems. Those are tied to political activism and political influence as long as they are not explicitly serving the stability of the current regime but rather may challenge the current regime: monitoring political leaders, providing a forum for opinion expression, promoting tolerance, set the agenda, influence public opinion, stimulate social change, contribute to national development.
Clusters of role ideals. I looked for clusters of journalistic role ideals at the country level based on three dimensions: an autonomy dimension (D-roles load positively), social engineering dimension (A-roles load positively), and an activation dimension (T-roles load positively). Three out of five clusters of role ideals seem to be compatible with democratic political systems: the classical Western Democracies (type I, USA) form one cluster or role ideals that contrasts with established Eastern Democracies (type II, Japan), and finally a cluster of Unconsolidated Democracies (type III, Chile). The analysis shows the diversity of journalistic role ideals even among (mostly) democratic countries.
In the authoritarian spectrum, we find two types: one set of role ideals seems to be typical for Consolidated Authoritarian countries (type V, China) and one seems to be typical for Unconsolidated Authoritarian countries (type IV, Egypt). In both type IV and V countries, social engineering is viewed as important. Type V is characterized by extremely low autonomy and moderate activation. In type IV, autonomy is moderate and activation is high. There are clear-cut differences in the constellation of role ideals typical for the clusters.
However, in terms of degree of democracy, I have demonstrated that several clusters display a similar degree of democracy and there are no statistically significant differences between some clusters. This illustrates that totally different constellations of role ideals can coincide with a similar degree of democracy. The journalisms in such two countries are similarly compatible (or incompatible) with democracy, but on different dimensions of role ideals. It shows that there are different pathways to and different shapes of "journalism-for-democracy", but also several pathways for how "journalism-for-democracy" can transform into a form of "journalism-for-authoritarianism".
Only the countries with a "Western Democratic Journalism" consistently score very high on the democracy index. This may partly reflect a Western bias in both concepts or both measurements as both the modern notion of democracy and of "good" journalism originate from Western countries. However, there are clear substantial differences, as the combination of high perceived autonomy and clear refutation of social engineering roles is a unique feature of Western countries and the typical understanding of journalism encountered there. This combination could be a factor in developing and stabilizing a fully democratic political system in the long run-a hypothesis that warrants further exploration. Against that hypothesis, one might argue that transplanting a "Western Democratic Journalism" into a country will cause estrangement rather than democratization if it conflicts with the cultural or the political context.
When journalism and political system do not match. The data show a striking alignment between country-level averages of journalistic role ideals and political systems' degree of democracy. However, this general pattern conceals two other truths that should be acknowledged: (1) There are some cases where journalists' role ideals do not seem to match the political system where we observed them. Prominently, journalists in Russia [DI: 2.94, "authoritarian regime"] have a slightly higher autonomy score, a substantially lower social engineering score and a substantially lower activation score than would be expected for an "authoritarian regime"; this means that it is closer to Type I (USA) than would be expected. The data were collected in 2014/15 (Anikina, Frost, and Hanitzsch 2017), however, such that current patterns of media control during the Ukraine War (Reporters without Borders 2022) cannot yet find expression in the data. At that time, Russia had a relatively recent history of a more democratic political system in the 1990s (see Figure A8) that may still have influenced role conceptions. Also, a unit-or item-response bias is possible if democratically minded Russian journalists had a greater inclination to participate or journalists felt inclined to give answers that shed a "favorable" light on journalism in Russia.
Another consistent outlier is the U.S. (DI: 7.96) where role ideals are characterized by higher autonomy, lower social engineering, and lower activation than would be predicted from its democracy score-probably owing to the long-standing democratic tradition in the U.S. The shadow of the past seems to explain at least some of the outliers.

Limitations and Desiderata
Some factors could reduce validity and reliability of role conception measurements: First, not all role conceptions are covered by WJS2. Second, the questions/items can be interpreted differently in different contexts. Classical notions of journalistic ideals from Western democracies are subject to reframing and reinterpretation in other parts of the world and in other political systems. For instance, "objectivity" according to Liberal ideology would mean "what can be perceived and can be verified as truth" whereas Marxist-Leninist ideology interprets "objectivity" in terms of what matches the truth as expressed in Marxist insights (Siebert, Peterson, and Schramm 1963, 119). Third, we cannot disentangle normative, cognitive, practiced, and narrated roles. These could probably be disentangled by asking additional questions, but this would mean that even more valuable interview time would be used on exploring role self-perceptions of journalists. Fourth, social and political pressure on journalists, particularly in non-free countries, may have led to both sampling bias and response bias, e.g., when journalists "loyal" to the regime are more likely to participate than oppositional journalists, and/or oppositional journalists respond untruthfully in fear of repression. The data may therefore misrepresent the differences in journalistic role ideals between free and non-free countries. Furthermore, there is substantial within-country variation between individual journalists, ranging from 57 to 91%, depending on the role ideal considered (Table A2). For instance, 22% of Chinese and 42% of Russian journalists were above the center-of-the-scale of the Autonomy scale. 23% of Norwegian and 22% of U.S. journalists are below the center-ofthe-scale of the Autonomy scale, the measurement most immediately indicative of a Journalism-for-Democracy. Interestingly, many D-and U-roles are widely endorsed while many A-role are only popular in some countries.
Like the measurement of role ideals, measurement of democracy can be criticized; the findings prove robust when replicating the analyses with alternative measurements of democracy (e.g., VDEM indices, Table 2). But democracy is itself multidimensional such that a bipolar spectrum between democracy and authoritarianism may oversimplify matters (Lijphart 1999). Finally, countries for which no data could be obtained will most likely systematically deviate from the countries included in the study. The WJS2 has, for example, a better coverage of Western Europe than, e.g., Western Africa or Central Asia. WJS3 will further extend the selection of countries.
Despite these concerns, the data reveal substantial differences between states with different political systems and help deepen our understanding of the correlations between political systems and journalistic role ideals-showing that (1) there is more than just two journalisms around the globe, (2) there is more than one Journalism-for-Democracy, (3) more than one pathway how a more democratic journalism can take shape, and (4) Journalism-for-Democracy looks different from what we imagined it would.

Disclosure Statement
No potential conflict of interest was reported by the author(s).