The Appeal of Media Populism: The Media Preferences of Citizens with Populist Attitudes

Although a growing body of literature points to the particular media diet of populist voters, we know too little about what specific media preferences characterize citizens with populist attitudes. This article investigates to what extent citizens with antiestablishment and exclusionist populist attitudes are attracted to attitudinal-congruent media content. We collected survey data using a nationally representative sample (N = 809) and found that citizens’ preferences for media content are in sync with their populist attitudes. Beyond having a tabloidized and entertainment-based media diet, populist voters self-select media content that actively articulates the divide between the “innocent” people and “culprit” others. These findings provide new insights into the appeal of different types of media populism among citizens with populist attitudes on different dimensions.

In previous studies, the theoretical assumptions of the media's relationship to populism were predominately based on right-wing populism (e.g., Caiani & della Porta, 2011;Mazzoleni, Stewart, & Horsfield, 2003). The Austrian tabloid newspaper Kronen Zeitung, for example, has been criticized for being favorable of the right-wing populist party FPÖ, and its populist news coverage is assumed to shape xenophobic sentiments among citizens (e.g., Karner, 2013). Following this rationale, readers of tabloid newspapers may have strong exclusionist populist attitudes as these outlets frame issues in a monocultural way. The relationship between media preferences and the core definition of antiestablishment populism, the opposition of the good people to the culprit elites, has been largely overlooked in extant literature.
Drawing on survey data collected among a representative sample of Dutch citizens (N = 809), we found that citizens with populist attitudes on different dimensions are attracted to different types of media populism. Correlational evidence shows that citizens' preferences for media populism were congruent with their populist attitudes, even when non-media-related factors are taken into account. Although we are unable to make causal claims, these findings provide important foundational evidence indicating that media diets and preferences are in sync with citizens' populist attitudes.

TWO METADIMENSIONS STRUCTURING CITIZENS' POPULIST ATTITUDES
Extant research has predominately conceptualized and measured "the" populist attitude as a one-dimensional concept (e.g., Akkerman, Mudde, & Zaslove, 2014;Hawkins, Riding, & Mudde, 2012). To arrive at a more precise conceptualization of populist attitudes sensitive to the variety of populist ideas expressed throughout the globe, we have proposed and tested a two-dimensional structure underlying populist attitudes (see Appendix A for a graphical depiction of the model; Hameleers et al., 2015). We briefly outline these two metadimensions here. Many scholars have emphasized that references to the centrality of the ordinary people are necessary but incomplete indicators of populism (e.g., Mudde, 2004). Rather, the opposition between the ordinary, good citizens and evil others in society defines the essence of populism (Mudde, 2004). This causal and moral relationship is vertically defined in the first dimension of populist attitudes: antiestablishment. People who interpret reality from this dimension construct the other vertically as the corrupt political elites who have betrayed the people's will (Ruzza & Fella, 2011).
The second dimension that can be distinguished is exclusionism. This dimension entails the perception of a horizontal opposition between the pure people and evil others in the heartland (Jagers & Walgrave, 2007). These others are constructed as undesired segments of the people, such as immigrants, people THE APPEAL OF MEDIA POPULISM with different religions, traditions or cultural values, or people who unfairly profit from the welfare state (e.g., Derks, 2006;Oesch, 2008). The exclusionist dimension thus taps into the in-group threat that people experience from within their nation: Others who do not belong to the heartland pollute the in-group's imagined community (Taggart, 2000).
It may be argued that our conceptualization of antiestablishment populist attitudes is strongly related to political distrust or cynicism whereas the exclusionist dimension is similar to xenophobic, nativist, or ethnocentric sentiments (e.g., Duckitt & Sibley, 2010). However, the conceptualization of populist attitudes proposed here can be distinguished from such concepts by its emphasis on a moral and causal divide between the in-group of the ordinary people and the vertically or horizontally defined other: The people are good and innocent, whereas the evil other is attributed responsibility for causing the people's problems (Hameleers, Bos, & de Vreese, 2016).

POPULIST MEDIA USE
Extant literature on the relationship between media use and populism has foregrounded several assumptions about what media appeal most to citizens with populist attitudes. In this study, we take a closer look at these theoretical ideas to put them to an empirical test. We first follow scholars who have argued that voters with populist attitudes primarily use tabloid and entertainment media (e.g., Albertazzi & McDonnell, 2008;Mazzoleni, 2008). Subsequently, we use the concept of media populism to unravel why this tabloidized and entertainmentbased media diet would be so attractive for citizens with populist attitudes: What are the central content features of tabloid and entertainment media that citizens with populist attitudes would be appealed to?

The Central Role of Tabloidized Media Diets
The assumptions concerning the relationship between the media and populism are predominately based on the idea that tabloid newspapers are more receptive and favorable of populist viewpoints than quality newspapers (e.g., Mazzoleni, 2008). On the receiver side, people who use tabloid media are said to be more aligned with populist ideas than people who use quality media (Mazzoleni, Stewart, & Horsfield, 2003). In line with this, empirical evidence demonstrated that voters for the populist Freedom Party in the Netherlands are actually more likely to use tabloid media than people who vote for mainstream political parties (Bos, Kruikemeier, & de Vreese, 2014). It therefore seems plausible that people who use tabloid media have stronger populist attitudes.
The tabloidized media diet of voters with populist perceptions can be explained by taking a closer look into the parallelism between populism and tabloid media. Quality newspapers are assumed to maintain a stronger relationship with the establishment, whereas tabloid newspapers depend more heavily on the mass audience (e.g., Art, 2006;Klein, 1998;Mazzoleni et al., 2003). Therefore, tabloid media are assumed to devote more attention to the worldviews of ordinary citizens than quality newspapers.
In line with selective exposure theory, people are expected to self-select political content that reflects their own views on society (Stroud, 2008). For tabloid outlets, this bias should be strongest for populist citizens, who are said to be low in political trust (Bos et al., 2013;Fieschi & Heywood, 2004). More specifically, as populist attitudes tap into the perceived centrality of common citizens while articulating distrust in others, tabloid media that articulate similar viewpoints should consequentially be most appealing to voters with such attitudes (e.g., Mazzoleni et al., 2003).
The appeal of tabloid media among people with populist attitudes can further be explained by the convergence of the core values of populism and tabloid media. Tabloid media share their ideological bias with populism (Hallin & Mancini, 2004). Both share a similar "thin" ideology grounded in the centrality of the common people and the circumvention of elites. Because this ideological core is also expressed in citizens' populist attitudes, people with stronger populist attitudes should be attracted most to tabloid media. In line with this, we expect that citizens who read tabloid newspapers have stronger populist attitudes than citizens who do not read tabloid newspapers (H1a).
Journalists of quality media adhere less to entertainment values and more to objectivity than journalists of tabloid media (e.g., Skovsgaard, 2014). On the sender side, this journalist practice translates into a stronger reliability on elite expert sources and less attention to the opinions and experiences of the ordinary people (Esser & Umbricht, 2013). Moreover, the coverage of quality newspapers is found to demonstrate a negativity bias toward populist leaders (Bosman & d'Haenens, 2008). On top of this, on the receiver side, readers of quality newspapers are found to be more supportive of the establishment's representation than readers of tabloid newspapers (Aarts & Semetko, 2003).
Because quality or elite media are assumed to give a voice to experts and elite sources rather than the ordinary people, and because their readers should be less distrusting and negative toward the establishment, we expect that citizens who read quality newspapers have weaker populist attitudes than citizens who do not read quality newspapers (H1b).

Entertainment Preferences
Empirical research has demonstrated that people who vote for populist parties have a specific media diet. Besides reading tabloid newspapers, they watch more THE APPEAL OF MEDIA POPULISM entertainment and soft news programs than other voters do (e.g., Bos et al., 2014). In line with this, the second media type that should appeal most to citizens with populist attitudes concerns entertainment and soft-news content (e.g., Klein, 1998;Mazzoleni et al., 2003). The appeal of entertainment and soft-news media among voters with populist attitudes can be explained by the centrality of ordinary citizens and the disdain of elites and experts articulated in both entertainment media and populism (Klein, 1998). Based on this congruence, people who prefer entertainment are more likely to hold stronger populist attitudes than people who do not prefer such media content (Mazzoleni et al., 2003).
Specifically, it has been demonstrated that exposure to entertainment-based content results in more political cynicism than exposure to hard news (Boukes & Boomgaarden, 2015). Political cynicism, in turn, relates positively to people's susceptibility to populist viewpoints (e.g., Bos et al., 2013). People who become more cynical because of exposure to entertainment media may thus be most susceptible to populist viewpoints. People exposed to hard-news content, in contrast, are supposed to be more aligned with the viewpoints of the elites. As they self-select into media use that is less likely to challenge the elites, they are less likely to be opposed to the elites themselves.
Based on the foregrounded theoretical assumptions and empirical findings, we propose the following hypotheses on the appeal of entertainment versus hardnews preferences among citizens: The more people prefer entertainment, the stronger their populist attitudes (H2a), and the more people prefer hard news, the weaker their populist attitudes (H2b).

The Appeal of Populist Media Content
Until this point, we have hypothesized that citizens with stronger populist attitudes can be characterized by their specific media use. We predicted that citizens with populist attitudes have a particular media diet consisting of tabloid, soft news, and entertainment-based content. Looking beyond citizens' media use, we also want to understand why tabloid and entertainment media may be so appealing for citizens with populist attitudes on the different distinguished dimensions.
In line with the concept of media populism, tabloid and entertainment media outlets are, more than other media outlets, assumed to actively engage in a populist style of communication themselves (Krämer, 2014;Mazzoleni, 2008). In doing so, journalists of these outlets frame issues in terms of the populist opposition between the common people and the culprit others. This frame has previously been defined as the "populist master frame" (e.g., Caiani & della Porta, 2011). By using this frame, journalists can define others both vertically as the corrupt elites and horizontally as the culprit societal out-groups, such as immigrants (e.g., Klein, 1998). Such coverage is, for example, used in an article of the British newspaper the Mirror: "While Brits endure crippling austerity with no end in sight, the rich have got richer again" (Beattie & Bloom, 2016, para. 1). In this newspaper article, the ordinary British citizens are implicitly framed as the innocent in-group. The rich elites are depicted as the culprits, whose self-interests harm the silenced majority: The elites deprive the hardworking native citizens from what they deserve.
These populist frames may affect people's perceptions of reality by providing them with a simplified, polarized definition of political issues. In doing so, populist media relate to citizens' negative stereotypes of the out-group and positive stereotypes of the in-group (e.g., Dixon, 2008;Hewstone, 1989). By activating these stereotypes, journalists who engage in media populism may contribute to "media based othering" (Krämer, 2014, p. 55). Negative news coverage of societal out-groups in turn enhances the chronic accessibility of negative stereotypes among citizens (e.g., Brader, 2005). People who prefer populist media content may thus perceive a binary divide in society themselves. This divide entails the antagonism between the imagined community of the blameless hardworking citizens and evil others, such as politicians, that threaten the purity of this community.
By distinguishing between different types of media populism, we can make more specific predictions about the appeal of populist media content among citizens with populist attitudes. Informed by the typology of populist communication foregrounded by Jagers and Walgrave (2007), we propose three types of media populism. This allows us to empirically assess the extent to which voters with populist attitudes prefer media content that uses specific types of populism to frame societal issues.
First, we can distinguish people centrality or empty media populism. This most "minimal" type of media populism emphasizes the centrality of the common people's will. In this definition, the opinions and experiences of ordinary citizens are the focal point of media coverage. In line with this definition, Uitermark, Oudenampsen, van Heerikhuizen, and van Reekum (2012) argued that entertainment television shows and tabloid newspapers are actively engaging in populist coverage by positioning the viewer-the ordinary, hardworking citizen-as central to the program or news event.
The second type of media populism distinguished in this article is anti-elites media populism. This type of media populism connects to literature that stresses how journalists of tabloid newspapers engage in populism themselves by emphasizing the binary opposition of the blameless people to untrustworthy elites (e.g., Krämer, 2014). This specific type of populist media content thus entails the disdain of elites, such as politicians or experts, who are perceived as a less credible and reliable source than the common people. For anti-elites media populism, the top-down analyses of elite experts, such as scientists, policymakers, or politicians, are consequentially perceived as less meaningful and less reliable than the down-to-earth experiences of ordinary citizens.
The final type of media populism distinguished in this article is monocultural media populism. This subtype of media populism contends that immigrants, ethnic minorities, refugees, and all other societal out-groups that do not belong to the populist heartland should not be given a voice in media coverage. Rather, journalists using this form of media populism provide a central stage for the common "national" citizen. Because journalists using monocultural media populism devote less attention to the opinions of horizontally constructed out-groups in society, this type of media populism connects most saliently to exclusionist populist communication (Jagers & Walgrave, 2007).
Against this backdrop, we can formulate specific predictions on the appeal of populist ideas used by journalists among citizens with populist attitudes on the two dimensions. As anti-elites media populism revolves around the opposition of the people to a vertical out-group, it should connect most saliently to people's antiestablishment populist attitudes. Monocultural media populism constructs the boundary between the people and others in a horizontal way. Therefore, it should connect mostly to exclusionist populist attitudes. People centrality media populism only touches upon the centrality of the in-group. As the "good" in-group is highlighted in both dimensions of populist attitudes, this type of media populism should be positively related to people's populist perceptions on both dimensions.
Based on this reasoning, we hypothesize the following: People with antiestablishment populist attitudes are appealed to people centrality and anti-elites media populism (H3a); people with exclusionist populist attitudes are appealed to people centrality and monocultural media populism (H3b).

The Media in Context
The tabloidized and populist media diet is not the only distinguishing feature of citizens with populist attitudes. Indeed, extant literature uses a number of interindividual differences to describe the profile of populist voters: age, gender, education, political knowledge, and perceived relative deprivation. In this study, we included these factors to assess the relative strength of the relationship between media use and congruent populist attitudes. We briefly discuss the potential relationship of these non-media-related alternative explanations here.
First, populist voters have mainly been characterized as younger men (e.g., Bos et al., 2013;Lubbers & Scheepers, 2000). Second, it has been argued that especially lower educated citizens are attracted to populist parties that simplify complex political issues in terms of the binary opposition between "us" and "them" (e.g., Bos et al., 2013;de Koster, Achterberg, & van Der Waal, 2013). Another interindividual factor related to populist attitudes is political knowledge. Previous research demonstrated that people who have less knowledge on institutions may regard them as more threatening (e.g., Anderson, 1998). Therefore, the less knowledgeable people are about politics, the more likely they will resort to populist perceptions that simplify political issues in terms of the distinction between the blameless people and the corrupt, threatening establishment.
Previous research has argued that the appeal of populism is strongly rooted in perceptions of relative deprivation (e.g., de Koster et al., 2013;Elchardus & Spruyt, 2012). In populist rhetoric, the government, the rich elites, and societal out-groups are blamed for depriving the common, hardworking citizens from what they in fact deserve (Elchardus & Spruyt, 2012;Grant & Brown, 1995). In line with this reasoning, people who believe that others always get more from the government than they get themselves are most susceptible to populist viewpoints.
To more precisely map the importance of populist citizens' media diet and preferences, we thus need to assess the relative contribution of the media in sketching the profile of populist voters while taking the aforementioned factors into account. Therefore, the final research question of this study is as follows: RQ1: What is the relative strength of the relationship between media use and congruent populist attitudes?

METHOD Sample
Institutional Review Board approval for the data collection was given on June 23, 2015. Polling company TNS NIPO collected the data by means of an online survey. From a panel of 124,000 citizens representative of the Dutch voting population in all regions of the country, 1,425 citizens were randomly selected and invited to participate. They received an invitation via e-mail or telephone.
Participants of the gross sample are allowed to complete a maximum of three surveys per month, and the agency optimizes involvement by making sure that participants are not overloaded but still receive regular invitations to stay attached to the panel. Because this large representative panel is mostly used for market research or election studies, participants are not frequently invited for social science research. As incentive, participants receive credits, which they can exchange for vouchers. Of the selection of eligible participants, 809 participants completed the survey. This relates to a response rate (RR1) and cooperation rate (COOP1) of 57%. The sample was representative of the national voting population in terms of gender, age, family composition, region, education, social class, and previous voting behavior. The mean age of participants was 51.07 years (SD = 17.25), 48.5% were female, and 23.9% were lower educated. As can be seen in Table B1 in Appendix B, the sample is indeed by and large representative of the Dutch adult population.  (Akkerman et al., 2014;Hawkins et al., 2012;Rooduijn, 2014) and people's populist opposition to economic elites (Elchardus & Spruyt, 2012).
Exclusionist Populist Attitudes. Based on the following six statements measured on a scale ranging from 1 (completely disagree) to 7 (completely agree), a 7-point scale of exclusionist populist attitudes was constructed (Cronbach's α = .91, M = 3.70, SD = 1.47): (a) Immigrants are threatening the purity of our culture. (b) Immigrants cost our country a lot of money that should rather be invested in our own people. (c) Our borders should be closed for immigrants. (d) Immigrants are responsible for a lot of our nation's problems. (e) Social benefits such as unemployment benefits and health insurance benefits are given to people who don't really deserve it. (f) People who are not originally from our country have no rights to receive our social benefits. Of all the participants, 8.8% scored 6 or higher on the scale. The development of these items was grounded in theory on exclusionist populism, nativism, and antiimmigration attitudes (e.g., Duckitt & Sibley, 2010;Jagers & Walgrave, 2007). By explicitly referring to a moral and causal distinction between the ordinary, native people as in-group and culprit out-groups, these items aimed to explicitly tap into populist sentiments (e.g., Albertazzi & McDonnell, 2008).
Entertainment and Tabloid Preferences. We measured the frequency of exposure to tabloid and quality newspapers by asking participants to indicate how many days in a normal week they usually read the tabloid newspaper Telegraaf and the quality newspaper Volkskrant. 1 As most people did not read 1 We included only one tabloid newspaper and one quality newspaper because the data available for other newspapers were outdated. The measures available for the other quality newspaper in the Netherlands, the NRC, pointed in the expected direction: Reading the NRC was significantly and negatively correlated to both the antiestablishment (r = -.10, p < .01) and the exclusionist (r = -.16, p < .01) dimensions of populist attitudes. these newspapers at all, the frequency of reading was recoded into a binary variable (0 = not reading this newspaper, 1 = reading this newspaper).
To operationalize entertainment and hard-news exposure and preferences, we asked participants to indicate their exposure to soft-news/hard-news media, as well as their preferences for entertainment and hard-news media formats. For media exposure, we asked participants to explicitly indicate how many days in a normal week they watched the soft news entertainment program Hart van Nederland. Moreover, we asked them how many days they watched the hard-news television program Nieuwsuur. Because approximately half of the sample reported not to be exposed to these media outlets, the frequency of watching these television shows was recoded into binary variables (0 = not exposed to these media formats, 1 = exposed to these media formats). Nieuwsuur was categorized as hard news based on its focus on rationality, impersonality, thematic framing, being nonemotional, being expert centered, and focus on an in-depth coverage of politics (Prior, 2003). Hart van Nederland, in turn, was categorized as soft news because it is more sensational, incident based, person centered, and more episodically framed (Boukes & Boomgaarden, 2015).
Preferences for entertainment content were measured with a single item measured on a 7-point scale from 1 (completely disagree) to 7 (completely agree): I enjoy spending an entire evening watching television shows and movies (Prior, 2003). Informed by Prior (2003), hard news preferences were measured on a three-item scale (Cronbach's α = .71, M = 3.94, SD = 1.34). This scale consisted of the following items: (1) I hate to miss the news; (2) I like complex news stories, even if it requires my full attention to comprehend it; (3) How many days in a normal weekday do you watch news programs, such as NOS Journaal or RTL Nieuws?
Preferences for Media Populism. We measured participants' preferences for the three conceptualized types of media populism with three separate statements measured on 7-point scales ranging from 1 (completely disagree) to 7 (completely agree). We used a confirmatory factor analysis (CFA) to ensure that these statements did not form a unidimensional scale (see Results section). People centrality media populism was measured by asking citizens to what extent they agreed with the statement that media content should pay more attention to ordinary people like themselves. To measure anti-elites media populism, people were asked to what extent they agreed with the statement that media content should ask elites rather than ordinary people to voice their opinion. This item was reverse-coded to indicate participants' preferences for this type of media populism. Third, people were asked for their agreement on the statement whether the media should devote more attention to minorities living in society, such as immigrants. This item was also reverse-coded to be an indicator of monocultural media populism.

THE APPEAL OF MEDIA POPULISM
Non-Media-Related Factors. Participants' level of education was originally measured with seven categories, which were recoded into lower and higher education. 2 Based on three multiple-choice questions on national politics, we constructed a three-item index for political knowledge, asking participants to identify the two parties in government, the minister of foreign affairs, and the leader of political party CDA (0 = all answers wrong, 3 = all answers correct; Cronbach's α = .75, M = 2.12, SD = 1.09). Based on three items, we constructed a 7-point scale of perceived relative deprivation (Cronbach's α = .85, M = 3.68, SD = 1.48). These three items are as follows: (a) If we need anything from the government, ordinary people like us always have to wait longer than others; (b) I never received what I in fact deserved; and (c) It's always other people who profit from all kinds of benefits.
Participants with populist attitudes, just like libertarians, liberals, and conservatives, may be distinguished by their preferences regarding salient societal issues (see Carmines, Ensley, & Wagner, 2012). Against this backdrop, we further explored how participants with antiestablishment and exclusionist populist attitudes be distinguished by their issue positions toward immigration, economic inclusionism, and European integration. We measured these issue preferences using the following three statements measured on 7point semantic differentials: (a) Immigrants should be allowed to keep their own culture/should fully adjust to our culture, (b) European integration has not gone far enough yet/has already gone too far, and (c) Income differences in society should decrease/increase.

Analysis
We conducted a CFA to check whether populist attitudes could empirically be distinguished from populist media preferences. In the next step, we assessed the relationship of the theoretically proposed populist media preferences to both dimensions of citizens' populist attitudes with ordinary least squares regression models in which we included nonmedia factors and controls. Table 1 and Table 2 demonstrate the results for each dimension separately. 3 2 Lower education indicates not completing an education/primary school/lower vocational or high school lower variant. Higher education indicates high school higher variant, university bachelor, or higher.
3 As robustness check, we also divided the sample into groups to find out if the results were similar when comparing citizens with stronger populist attitudes (M + SD) to citizens with weaker populist attitudes (M-SD). This analysis yielded similar results as reported in the article.

RESULTS
The three items measuring preferences for the three types of media populism correlated rather weakly (r = .19, r = -.07, r = -.02) and did not form a reliable unidimensional scale (Cronbach's α = .10). This supports our conceptualization that the different types of media populism are not tapping into a one-dimensional populist media preference. Next, we estimated a CFA model to investigate whether preferences for media populism could be validly distinguished from populist attitudes. The model in which preferences for media populism were included as indicators of both populist attitudes dimensions fitted significantly and substantially worse than the model with only the items measuring populist attitudes, Δχ 2 (6) = 156.14, p < .001. Moreover, the standardized regression weights of the preferences for media populism items and the two dimensions of populist attitudes are all relatively weak (r < .28), with the exception of a moderately strong correlation between the exclusionist dimension and preferences for monocultural media populism (r = .41). Still, even when incorporating only this best-fitting item as additional indicator of the exclusionist dimension, model fit decreased substantially and significantly, Δχ 2 (1) = 142.53, p < .001.
These results indicate that populist attitudes and populist media preferences are not tapping into the same underlying construct: Participants clearly distinguished between their political populist interpretations and their preferences for populism by the media.

The Media Diets of Citizens With Populist Attitudes
Now that we confirmed that populist attitudes and preferences for media populism are not tautological, we proceeded with mapping the media diets of participants with populist attitudes. 4 We first predicted that people who read tabloid  Note. Two-tailed tests. Unstandardized regression weights. Standard errors reported between parentheses. Pop. = populism. *p < .05. **p < .01. ***p < .001. newspapers have stronger populist attitudes than people who do not read tabloid newspapers (H1a). We found that reading a tabloid newspaper was not significantly related to antiestablishment populist attitudes (see Table 1). In line with our predictions, however, reading a tabloid newspaper was significantly and positively related to the exclusionist dimension (see Table 2). This means that participants who read a tabloid newspaper had stronger exclusionist populist attitudes than participants who did not read a tabloid newspaper. H1a can thus only be confirmed for the exclusionist dimension of populist attitudes.
Participants who read a quality newspaper scored significantly higher on the antiestablishment dimension than participants who did not read a quality newspaper, which contradicts H1b. In line with H1b, however, participants who read a quality newspaper had significantly weaker populist attitudes on the exclusionist dimension than people who did not read a quality newspaper. Overall, H1a and H1b can thus be confirmed only for the exclusionist dimension of populist attitudes.
Regarding the appeal of entertainment media, Table 1 shows that preferences for entertainment content related significantly and positively to the antiestablishment dimension: The more people preferred entertainment content, the stronger their populist attitudes on the antiestablishment dimension. This supports H2a. Hard-news preferences, in contrast, were not significantly related to participants' populist attitudes on the antiestablishment dimension, which is not in line with H2b. Moreover, participants' self-reported exposure to soft-news (entertainment) content or hard-news content was not significantly related to their antiestablishment populist attitudes, which contradicts both H2a and H2b.
For the exclusionist dimension, entertainment preferences, hard-news preferences and, watching soft-news/hard-news content were all not significantly related to participants' populist attitudes. Based on these findings, we have only found limited support for H2a and H2b: Only preferences for entertainment content were significantly and positively related to the antiestablishment dimension.
The results show that entertainment and tabloid media appealed to citizens with populist attitudes on different dimensions in different ways. Reading a quality or tabloid newspaper only related to the exclusionist dimension of populist attitudes in the expected direction and entertainment preferences only related positively to the antiestablishment dimension.

The Appeal of Media Populism
In the next step of the regression analyses, we included participants' preferences for media content that uses the three types of media populism (see Table 1 and Table 2). As can be seen in Table 1, participants who preferred people centrality and anti-elites media populism had stronger populist attitudes on the THE APPEAL OF MEDIA POPULISM antiestablishment dimension, which supports H3a. More specifically, the stronger the participants' preferences for media content that provides a stage for the ordinary people, the higher their populist attitudes on the antiestablishment dimension. Similarly, the stronger the participants' preferences for media content that circumvents elites, the stronger their populist antiestablishment perceptions. In line with our expectations, participants' preference for monocultural media populism was not significantly related to the antiestablishment dimension.
As shown in Table 2, participants who preferred media content with people centrality media populism and monocultural media populism had stronger populist attitudes on the exclusionist dimension, which supports H3b. This means that the more that participants preferred media content that centralizes ordinary citizens, the more they interpreted reality from an exclusionist populist frame of reference. In a similar vein, the more that people preferred media content that provides a stage for the monocultural in-group of the native citizens while devoting less attention to societal out-groups, such as immigrants, the stronger their exclusionist populist attitudes.
To sum up, participants who preferred media content stressing the centrality of ordinary people and the disdain of elite sources scored higher on the antiestablishment dimension. Participants who preferred media content highlighting the centrality of the people and a monocultural interpretation of society were most likely to hold exclusionist populist perceptions. H3a and H3b are thus both supported: The populist attitudes of citizens are in sync with their populist media preferences.

The Populist Citizen Beyond Media Preferences
In the next step, we assessed the relative strength of the appeal of populist media among participants with populist attitudes (RQ1). Beyond the media, perceived relative deprivation related strongly to citizens' populist attitudes on both dimensions (see Table 1 and Table 2). The more participants felt deprived, the stronger their populist attitudes. Level of education was also related to both dimensions of the people's populism: Lower educated participants held stronger populist attitudes than higher educated participants. Political knowledge was only related to participants' exclusionist populist attitudes. The less knowledgeable people were about politics, the more likely they were to interpret reality from an exclusionist populist frame of reference.
Once preferences for media populism were included in the regression models, the proportion explained variance of populist attitudes increased significantly for both dimensions (see Table 1 and Table 2): 30.8% of the variance in the antiestablishment dimension was explained by nonmedia factors alone. This increased to 33.5% when preferences for populist media content were included (see Table 1). For the exclusionist dimension, the proportion explained variance increased even more substantially from 45.3% to 54.4% (see Table 2). To answer RQ1, we need to focus on populist media preferences to more precisely explain which factors relate to citizens' populist attitudes, which is especially the case for the exclusionist dimension. At the same time, the findings show that media use alone is-obviously-far from sufficient to distinguish citizens with populist attitudes from citizens without such attitudes.
To further contextualize the findings on populist media use, we explored the issue preferences of participants with populist attitudes while controlling for all other variables. 5 Participants with exclusionist populist attitudes believed that immigrants should not be allowed to keep their own culture (b = .20, SE = .03, p < .001). This issue position was not salient among participants with antiestablishment populist attitudes (b = .01, SE = .03, ns). Participants with exclusionist populist attitudes had weaker economic-inclusionist attitudes (b = -.08, SE = .02, p < .001), but this position was not salient among those with antiestablishment populist attitudes (b = -.01, SE = .03, ns). Participants with exclusionist (b = .07, SE = 0.02, p < .01) and antiestablishment populist attitudes (b = .06, SE = .02, p < .01) both opposed EU integration.

DISCUSSION
In the midst of the mediatization of politics, media populism is argued to be a highly salient phenomenon across the globe (Krämer, 2014;Mazzoleni, 2008). Populism is also salient as an individual-level attitude on the demand side of citizens (Hameleers et al., 2015). A large body of literature has therefore claimed that the media play an important role in the global rise of populism (e.g., Mazzoleni et al., 2003;Mudde, 2004). It has even been argued that journalists actively engage in populist framing themselves (e.g., Caiani & della Porta, 2011;Mazzoleni, 2008). Still, we know too little about how citizens with populist attitudes can be distinguished from nonpopulists by their preferences for specific media content. Against this background, this article aimed to put the assumed appeal of media populism among citizens with congruent attitudes to an empirical test. To do so, we proposed a typology of media populism and related preferences for three types of media populism to the two core dimensions structuring citizens' populist attitudes.
In general, the results of this study provided limited support for the often assumed tabloidized media diet of citizens with populist attitudes, as we only found a tabloidized media preference among citizens with exclusionist populist attitudes (e.g. Karner, 2013;Krämer, 2014;Mazzoleni et al., 2003). This may be explained in the light of the higher threshold for the more distant, right-wing exclusionist dimension, which makes citizens with such philosophies easier to distinguish by their media preferences than citizens with less extreme, societally acceptable antiestablishment populist attitudes. Looking beyond media exposure, we further assessed if populist content features argued to be centralized by journalists in tabloid and entertainment news coverage appealed most to citizens with populist attitudes.
Doing so, we found that citizens' preferences for different forms of media populism were stronger related to populist attitudes than their self-reported tabloidized and entertainment-based media diet. If we, in line with extant literature, assume that the media are increasingly using populist frames to cover important societal issues (e.g., Caiani & della Porta, 2011;Krämer, 2014;Mazzoleni, 2008), these frames may be highly persuasive for citizens who prefer media content that simplifies issues in binary "us" against "them" oppositions. This can tentatively be interpreted as support for the assumption that media populism relates to populist frames of interpretations among citizens (Krämer, 2014). If the media, for example, frame immigrants and refugees as societal outgroups responsible for their own fate, citizens who prefer such media content may accept this view perceiving that horizontally constructed others indeed pose a severe threat to the purity of their heartland.
However, because citizens' tabloidized and entertainment-based media diet was not strongly related to their populist attitudes, one could argue that journalists of these media types may not be as overly populist in their news coverage as assumed in extant literature (e.g., Mazzoleni et al., 2003). Indeed, the results of recent empirical studies are still inconclusive with regard to the degree of populism expressed in tabloid newspapers versus quality newspapers (e.g., Akkerman, Mudde, & Zaslove, 2014;Rooduijn, de Lange, & van der Brug, 2014). Our key finding that citizens with populist attitudes prefer media content framed in a populist way may not be related to the actual supply of media populism in tabloid and entertainment formats. In future research, content analyses need to point out whether the specific media diet of populist voters is actually more populist than the media diet of voters for mainstream parties.
Most literature has based the assumptions of the relationship between the media and populism exclusively on right-wing populism (e.g., Uitermark et al., 2012). An important contribution of this study is that it revealed the specific media preferences of citizens with antiestablishment and exclusionist populist attitudes. We found that citizens who were attracted to different types of media populism interpreted reality according to different populist frames of references. Preferences for media populism that highlighted the centrality of the ordinary people appealed to citizens with populist attitudes on both dimensions. As people centrality provides the most "empty" or "minimal" definition of media populism, the relationship of people centrality media populism to both dimensions makes perfect sense (e.g., Jagers & Walgrave, 2007).
In line with this reasoning, the "complete" types of media populism appealed differently to citizens with exclusionist and antiestablishment populist attitudes. People who preferred media content that circumvents elites scored higher on the antiestablishment dimension, and people who preferred media content that circumvented immigrants and minorities by depicting a monocultural society scored higher on the exclusionist dimension.
Besides their specific media preferences, citizens with populist attitudes can be distinguished by their specific issue preferences (Carmines et al., 2012), most saliently regarding opposition to immigration and EU integration. Next to this, the nucleus of populist attitudes can be identified as perceived relative deprivation (see also Elchardus & Spruyt, 2012). Beyond the media, populist citizens can thus also be categorized by their interpretation of the sociocultural environment, which expands the "losers of modernization" thesis (e.g., Kriesi et al., 2006).
The findings of this article have practical implications. If the media partially contribute to the polarization of public opinion by inciting citizens' populist "us" against "them" perceptions, one could argue that media populism is an undesired, harmful phenomenon. On the receiver side, citizens can be made more aware of the potential effects of content that is framed in a populist way. Although citizens may not always be aware of their own populist philosophies, by understanding how the populist content of their media diet may affect their own "us" versus "them" interpretations of issues, citizens can adequately use their media literacy to more critically judge and, if desired, resist the persuasive potential of media content that frames issues in a populist way.
Our study has some limitations that can be addressed in future research. First and foremost, the findings of this study are insufficiently able to point to a causal relationship between media populism and populist attitudes. It could well be the case that populist attitudes are both cause and consequence from exposure to tabloid/entertainment content and preferences for media populism. Specifically, journalists may engage in populist news coverage in an attempt to appeal to a large perceived audience with populist attitudes. Alternatively, citizens may have become more populist because of their preference for and self-selection of media types that frame issues in a populist way. As we set out to explore the relationship between the appeal of populist media and populist attitudes, the causal order may be less relevant for the purpose of this study, which we consider a first foundational study in disentangling this relationship. Nevertheless, future research should more precisely assess the causal order of the described relationships, for example, by conducting experiments in which populist attitudes are the dependent variables or by pairing panel survey data with a content analysis of populism by the media. Only then we can start to assess how different dimensions of the people's populism are actually caused by media populism.
A second limitation concerns the generalizability of our findings. The media variables incorporated as predictors of populist attitudes in this study may not work THE APPEAL OF MEDIA POPULISM in the same way in countries with different media systems. However, the typology of media populism used in this article was grounded in internationally applied conceptualizations. Future research may further assess the role of differing contexts by conducting comparative research in countries that differ substantially in media systems and sociopolitical contexts.
All in all, this study has demonstrated that a two-dimensional structuring of populist attitudes allows for a better understanding of the peculiar media diets of populist voters on both the left and right end of the political spectrum. As different types of media populism appeal to voters in attitudinal-congruent ways, we can no longer assume that all populist citizens have the same media diet. Rather, the variety of populist viewpoints propagated by populist media seems to be congruent with the different ways in which citizens are populist themselves.