Selective Exposure to Populist Communication: How Attitudinal Congruence Drives the Effects of Populist Attributions of Blame

Selective Exposure to Populist Communication: How Attitudinal Congruence Drives the Effects of... Abstract Although it has been argued that populist communication only appeals to a specific audience, extant research has exclusively studied the effects of populist communication in forced-exposure media environments. Responding to this discrepancy, we conducted two experiments (N = 562 and N = 558) in which we manipulated the core idea of populist messages—attributing blame—in forced and selective exposure media environments. Our results demonstrate that citizens higher in relative deprivation are most likely to select populist messages. Irrespective of selective exposure, citizens’ populist attitudes are only positively affected if the populist message is congruent with their prior feelings of relative deprivation. These results provide important insights for the polarizing potential of media populism in a fragmented media environment. Populism has become influential in the political landscape of many democracies around the world. In recent years, the populist ideas expressed by a variety of actors and movements all around the globe have affected politics and society in important ways, potentially leading to a polarized electorate (e.g., Pappas, 2014). In populist discourse, the ordinary people are addressed as the “good” in-group. This in-group should consequentially be the focal point of political decision-making (Canovan, 1999; Mudde, 2004). The elites are constructed as a corrupt enemy that deprives the people from their deserved earnings, and are accused of failing to represent the people’s will. The core idea of populism thus revolves around the moral and causal divide between blameless ordinary people and culprit others (e.g., Hameleers, Bos, & de Vreese, 2017; Mudde, 2004). The persuasiveness of such populist blame attributions forms the backdrop of this paper: what are the effects of populist messages, and which citizens are most likely to select and be persuaded by them? It has been demonstrated that exposure to populist communication affects citizens’ political perceptions (e.g., Bos, van der Brug, & de Vreese, 2013). Extant literature further pointed to the media as an important supply-side factor facilitating the dissemination of populist ideas in society (e.g., Bos et al., 2013; Mazzoleni, 2008; Waisbord & Amado, 2017). Two explanations of the media’s role in the spread of populism have been forwarded. First, driven by media logic, the media are said to offer favorable attention to the newsworthy ideas of populist politicians (e.g., Mazzoleni, 2008). Second, journalists are themselves assumed to actively frame issues along the lines of a populist distinction in society (e.g., Krämer, 2014; Mazzoleni, 2008). This journalistic practice has been described as populism by the media or as media populism (Krämer, 2014; Mazzoleni, 2008). This implies that the media, and not only politicians, engage in populist communication by actively stressing the moral and causal divide between the blameless people and the culprit others. Backed up by empirical evidence, we argue that such mediatized attributions of blame can be regarded as a persuasive frame that activates populist schemata among receivers (Hameleers et al., 2017; Krämer, 2014). Such populist blame attributions may not convince all citizens. In line with selective exposure theory, people are only expected to select populist content if they have similar worldviews. Outside of experimental settings, people are only expected to expose themselves to populist content if their prior beliefs are congruent with it (Festinger, 1957). Therefore, traditional experiments that force participants into viewing messages that they would normally not select themselves do not reflect the fragmented media environment of citizens (Bennett & Iyengar, 2008). But what prior beliefs drive people’s selection of populist media content? What are the consequences of being forced to read incongruent populist messages? To answer these questions, we conducted two online survey experiments (N = 562 and N = 558) in which selective exposure was linked to different types of media populism framing different societally relevant issues. This allowed us to investigate whether it is indeed the case that populist messages are most persuasive among the part of the electorate that self-selects these messages, motivated by attitudinal congruence. Populist attributions of blame in media populism Populism revolves around the construction of a societal divide between ordinary citizens as a blameless in-group versus the “corrupt” elites as a culpable out-group (e.g., Caiani & della Porta, 2011; Mudde, 2004; Taggart, 2000). Populist discourse addresses the in-group of the people as being victimized by the corrupt elites. These elites do not represent ordinary citizens’ will and deprive them from cultural and material resources. The elites are accused of favoring other groups in society instead of their “own” people (Elchardus & Spruyt, 2016). Populist discourse presents issues in “black and white” or “us versus them” interpretations, hereby simplifying complex political issues into an antagonistic societal divide (Taggart, 2000). By repeatedly exposing people to such simplified messages that reinforce a positive image of the deprived people and a negative image of the elite and other out-groups, populist discourse is argued to be highly persuasive (e.g., Rooduijn, 2014). Attributing blame to the elites or societal out-groups forms the ideational core of populist messages (e.g., Hameleers et al., 2017). Populist ideas articulate the divide between us and them by scapegoating various out-groups for causing the crisis while absolving ordinary people of blame (e.g., Taggart, 2000). Populist communication provides citizens with an influential heuristic cue that helps them to translate complex socio-political issues into black and white terms (Hewstone, 1989; Iyengar, 1989). Attributions of responsibility hereby fulfill an important role in citizens’ political decision making, as they can heuristically use responsibility cues to punish the culprits (Marsh & Tilley, 2010). Attributions of blame guide citizens’ political opinions in important ways (Hobolt & Tilley, 2014; Iyengar, 1989). Specifically, if certain politicians are attributed blame, citizens evaluate them in more negative terms and are less likely to vote for them (Marsh & Tilley, 2010). It has been demonstrated that the media play an important role in the effects of blame attributions, as the framing of societal issues in terms of “who has done it” influences citizens’ own perceptions of blame (e.g., Iyengar, 1989). It must be noted here that attributing blame to political opponents is a common political phenomenon. Populist blame attribution differs from general attributions of responsibility by emphasizing both an antagonistic causal and moral divide between the good people and the culpable other. Beyond highlighting which actors are responsible, populist blame attributions emphasize that the elites are failing to represent the people’s will, depriving them from what they deserve. Populist messages attribute blame to different out-groups deemed responsible for ordinary people’s problems (Jagers & Walgrave, 2007). First, blame is attributed to the elites (e.g., Mudde, 2004; Rooduijn, 2014). The most salient elitist enemies are the politicians in government, who are accused of not representing ordinary people and their will (e.g., Hawkins, 2009; Mudde, 2004). Right-wing populist messages can additionally emphasize a societal divide between the native people as a blameless in-group and horizontally opposed others as culpable out-groups (Jagers & Walgrave, 2007). This populist divide entails the interpretation that refugees, immigrants, ethnic minorities, welfare-state profiteers, and all other out-groups living amongst the people are responsible for causing the problems of the native ordinary citizens, such as an increasing crime rate or a declining economy. The populist core idea of in-group favoritism and out-group hostility lies at the heart of media populism (Krämer, 2014; Mazzoleni, 2008). Media populism can be defined as journalists’ active reconstruction of populist ideas, acting independently of the political actors associated with populism (Krämer, 2014). These populist ideas are most saliently reflected in the journalistic interpretation of a divide between the “silenced majority” of ordinary citizens and the corrupt elites or culpable out-groups (e.g., Caiani & della Porta, 2011). This implies that, beyond disseminating the hard facts of events, journalists who engage in media populism actively emphasize or frame an interpretation of the background, causes, moral evaluations, and treatment recommendation when reporting on the news (e.g., Entman, 1993). Specifically, media populism attributes causal responsibility for negative developments while emphasizing the moral divide between the “good” people and the “evil” others. The treatment recommendation is to remove the elites or out-groups. The effects of populist communication on populist interpretations Populism is not only communicated by politicians or journalists. Citizens can interpret issues along the lines of populist distinctions themselves as well (e.g., Akkerman, Mudde, & Zaslove, 2014; Rooduijn, 2014). Such individual-level populist attitudes tap into citizens’ sentiments of a causal and moral divide between ordinary people and culprit others. The process through which populist messages affect populist attitudes can be described as “media-based othering” (Krämer, 2014). As a result of exposure to messages that attribute blame to elites or societal out-groups, citizens may interpret issues in congruent us versus them frames as well. This effect can be interpreted in light of the psychological mechanisms of social identity and stereotyping (e.g., Tajfel, 1978). These mechanisms postulate that positive images of the in-group and negative stereotypes of the out-group are primed and chronically accessible among receivers (e.g., Dixon, 2008). Exposure to populist messages may thus stimulate and activate populist attitudes among the public, which is in line with the literatures on trait activation and cognitive priming (e.g., Richey, 2012). Specifically, exposure to populist communication primes congruent us versus them cognitive clusters or schemata among receivers. As a consequence of exposure to populist communication, people’s negative stereotypes of the out-group and positive stereotypes of the in-group may be activated. Such stereotypical representations are reflected in populist attitudes that tap into the perceived opposition between the good people and the evil others (Akkerman et al., 2014). This corroborates a recent line of empirical research that has causally related exposure to populist communication to populist attitudes (e.g., Matthes & Schmuck, 2017). Selective exposure to media populism In the era of increasingly more fluid and user-controllable media diets, exposure to specific media messages cannot be taken for granted (e.g., Bennett & Iyengar, 2008; Stroud, 2008). Responding to this changing media landscape, a growing body of literature points to selective exposure as an important factor that needs to be taken into account in media effect studies (e.g., Arceneaux, Johnson, & Murphy, 2012; Iyengar & Hahn, 2009). Selective exposure can be defined as the guiding influence of people’s prior beliefs on their selection of media content (Stroud, 2008). Recent literature contends that, when offered the choice, people are most likely to expose themselves to pro-attitudinal as opposed to counter-attitudinal content. Media effects should consequentially occur most among people that actively choose to expose themselves to specific media content. The psychological underpinnings of selective exposure are rooted in theories of cognitive consistency (Iyengar & Hahn, 2009). In line with this, people are expected to select and filter novel information based on their prior beliefs (Festinger, 1957). Doing so, people pay attention to information that confirms their prior attitudes and beliefs while circumventing information that counters their priors. In line with this, people’s desire to adhere to their prior beliefs weighs more than getting it right, or making the most accurate decision based on an elaboration on all the factual information available to them (Lodge & Taber, 2013). These selection biases are psychologically explained by people’s intrinsic motivation to reduce cognitive dissonance. Although not necessarily conscious, people engage in selective processing strategies that avoid or counterargue inconsistent information (Festinger, 1957). Selective exposure can result in a polarization of beliefs (e.g., Stroud, 2008). The argumentation behind this view is as follows: as people with certain viewpoints expose themselves to congruent information, their beliefs will be reinforced and driven by the search for cognitive consistency. In addition, people become immune to attitudinal change by incongruent information. Although people may already hold a certain belief, deliberate exposure to congruent views may thus bolster their priors (e.g., Iyengar & Hahn, 2009). Aligning ourselves with this view on the polarizing potential of selective exposure, we expect that people who self-select attitudinal congruent populist content are those most persuaded by populist attribution of blame. How perceptual biases drive the effects of populist communication While some have argued that populism appeals to people that can be categorized as the “losers of modernization” (e.g., Kriesi et al., 2006) based on their social characteristics, many scholars have more recently gravitated towards an understanding that populist messages appeal most to people who perceive themselves to have lost out more than others in society. These vulnerable people should be those most attracted to populist sentiments that voice their grievances of being worse off than others (Elchardus & Spruyt, 2016). In line with this reasoning, the attitudinal filter of selective exposure to media populism can be described as relative deprivation: the perception that the out-group opposed to the people’s in-group unfairly receives economic and cultural resources at the cost of the in-group of ordinary citizens (Elchardus & Spruyt, 2012, 2016). Put differently, relative deprivation entails the perception of an unjust distribution of the society’s wealth: ordinary people “like us” never get what they deserve from society, whereas others always seem to profit (e.g., Hogg, Meehan, & Farquharson, 2010). Relative deprivation has been causally related to populism (e.g., Elchardus & Spruyt, 2012, 2016; Spruyt, Keppens, & van Droogenbroeck, 2016). As demonstrated by Elchardus & Spruyt (2016), populist perceptions are the consequence of an inherently negative perspective on the evolution of society and of the feeling of belonging to a group of people that is unfairly treated by society. Populist discourse is known to respond to a variety of grievances among society (Laclau, 1977). Populist discourse addresses these “victimized” people by claiming that their grievances are not heard by the corrupt elites. Against this backdrop, people who feel vulnerable and deprived have a higher tendency to be attracted to populist discourse, which claims to voice the grievances of the deprived people (Elchardus & Spruyt, 2016). Relative deprivation can be regarded as a salient prior belief driving people’s selection of attitudinal-congruent populist media content. H1: The more people experience feelings of relative deprivation, the more likely they are to self-select into exposure to congruent pro-attitudinal populist media content. It has been argued that populist cues that highlight the opposition between us and them are highly persuasive because they simplify complex societal and political issues into matters of who has done it, while absolving ordinary people of any blame (e.g., Mudde, 2004;,Rooduijn, 2014). The in-group can thus maintain a positive self-concept by attributing blame to others. This persuasive appeal has also been identified in studies on responsibility attributions (e.g., Hobolt & Tilley, 2014). H2: Exposure to populist content leads to stronger populist attitudes than exposure to non-populist content. We postulated that prior beliefs of relative deprivation drive the selection of populist content. In line with the premises of selective exposure theory, this implies that the media effects of populist blame attributions should occur most for those people who feel most deprived. For these people, the populist message is congruent with their priors. Motivated by a desire to avoid cognitive dissonance, their populist attitudes should be bolstered most (Festinger, 1957). H3: People exposed to attitudinal congruent populist content are persuaded most by media populism. Study 1 Method Design To investigate who selects populist content and how selective exposure and attitudinal congruence drive the effects of populist communication, we employed two experiments. The first experiment focuses on anti-elites media populism. Before exploring the role of attitudinal congruence, this experiment investigates who is most likely to select populist content. The design concerned a 2 (forced vs. selective exposure to media populism) × 3 (pro vs. counter. vs. balanced attitudinal stance) + 3 (pro vs. counter. vs. balanced self-select non-populist attitudinal stance) + control between-subjects factorial design. Because of the asymmetry in the number of cells, we allocated relatively more participants to the choice than the forced exposure conditions (Feldman, Stroud, Bimber, & Wojcieszak, 2013). Within the forced exposure conditions, participants were randomly assigned to one of the four experimental groups. In the choice conditions, participants self-selected into one of the six experimental groups or the control group. Sample The survey experiments were carried out on a diverse sample of Dutch citizens recruited by an international polling agency. Eligible participants in the company’s database (over 18 years, living in the Netherlands) were randomly selected and approached via e-mail. The samples are by and large representative of the voting population in terms of age, gender, and level of education. 759 individuals accessed the survey link. The study was completed by 562 participants (completion rate 74.0%). Their mean age was 49.29 years old (SD = 14.84). 46.7% were male and 53.3% were female. 24.5% of the participants were lower educated, 31.2% were higher educated, and 44.3% had a moderate level of education. Procedure The experiment was conducted online. After the informed consent procedure, participants completed a pre-test that included measures for demographics, moderating variables, and control variables. Participants were randomly allocated to either the forced or selective exposure media environment. In the selective exposure environment, participants were allowed to select one of the seven alternative news items. Next, they were exposed to the stimulus they selected themselves. In the forced exposure conditions, participants were randomly assigned to watch one of the three populist stimuli or the control stimulus. The selected or forced news item was visible for at least 30 seconds. This minimum exposure time was informed by the pilot study and was thoroughly pre-tested by the researchers. After 30 seconds, a “next” button appeared on the lower-left corner of the screen. After reading the news item, participants were forwarded to the post-test survey. This survey contained measures for the dependent variables and manipulation checks. The average response time was 15.64 minutes (SD = 20.28), of which 54.26 seconds were devoted to reading the stimuli (SD = 31.29). Once participants completed the survey, they were directed to the panel company environment, where they received incentive vouchers. Independent variables and stimuli In the selective exposure conditions, choice was manipulated by presenting participants with a cover story describing that people may prefer some news items above others available to them. To simulate their everyday media environment, participants were told that they should select one preferred article out of a list of alternative online articles. By clicking on a link showing the article’s headline, they were forwarded to their preferred article. In the forced exposure conditions, the cover story explained that people’s everyday media environment exists of a plethora of alternative news stories, and that they were exposed to one of these many options. The topic of the stimuli was the decreasing health care budget. In line with extant research on selective exposure, the articles’ attitudinal stance was manipulated into pro-attitudinal, counter-attitudinal, and balanced content (see Feldman et al., 2013). Pro-attitudinal messages connected the negative development to the elites, whereas counter-attitudinal messages connected it to the in-group (see stimuli in Appendix A and Table 1 for an explication of conditions in the various congruency groups). Table 1 Experimental Conditions and Attitudinal Congruence Attitudinal Congruence Stimuli  Congruent Stimulus  Incongruent Stimulus  Populist  People higher in perceived deprivation are exposed to a populist stimulus blaming the out-group  People lower in perceived deprivation are exposed to a populist stimulus blaming the out-group  Counter-Blame  People lower in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  People higher in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  Non-Populist  People higher in perceived deprivation are exposed to a stimulus connecting the situation to the elites  People lower in perceived deprivation are exposed to a populist stimulus connecting the situation to the elites  Counter–No Blame  People lower in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  People higher in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  Attitudinal Congruence Stimuli  Congruent Stimulus  Incongruent Stimulus  Populist  People higher in perceived deprivation are exposed to a populist stimulus blaming the out-group  People lower in perceived deprivation are exposed to a populist stimulus blaming the out-group  Counter-Blame  People lower in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  People higher in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  Non-Populist  People higher in perceived deprivation are exposed to a stimulus connecting the situation to the elites  People lower in perceived deprivation are exposed to a populist stimulus connecting the situation to the elites  Counter–No Blame  People lower in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  People higher in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  Populist blame attribution was manipulated by emphasizing the ideational core of populism in the article’s framing of the news event: the causal and moral opposition between the good, blameless ordinary citizens and the culprit, evil elites (e.g., Mudde, 2004; Jagers & Walgrave, 2007). To do so, the populist stimuli, for example, referred to “our” own people in need and “our” tax money when referring to the blameless in-group. The elites in government were attributed blame for ordinary people’s problems by stressing how they are corrupt, self-interested, and unable and unwilling to care for ordinary people. The balanced stimuli articulated a mixed interpretation: both the Dutch people and the government were connected to the development (non-populism) or blamed for it (populism). All other factors were kept constant across conditions. The same topics and problem interpretations were displayed across conditions. We randomized the display order of all headlines in the selective exposure conditions. In that sense, the bias of incidental selection was constant across conditions. A growing body of literature has pointed to the fragmented spread of populism in the media (e.g., Engesser, Ernst, Esser, & Büchel, 2016). This implies that the different elements of populist content may not be present at the same time in one line of text. To make sure that people could intentionally self-select populist content, the Manichean construction of us against them was represented in both the texts and headlines of the populist stimuli. Manipulation checks The manipulation of populist blame attribution was successful (F[1, 226] = 71.45, p < .001). This means that participants in the elitist blame attribution conditions were significantly more likely to believe that the article attributed responsibility to the elites (M = 5.39, SD = 1.34) than participants in the other conditions (M = 3.74, SD = 1.60). Participants in the counter blame conditions were more likely to believe that ordinary people were blamed (M = 4.76, SD = 1.81) than in the other conditions (M = 3.23, SD = 1.84). The manipulation of selective exposure also succeeded (F[1, 553] = 206.21, p < .001), which indicates that participants in the selective exposure conditions were significantly more likely to perceive they had the choice to self-select an article (M = 5.49, SD = 1.45) than those in the forced exposure conditions (M = 3.53, SD = 1.72). Measures Unless explicitly described otherwise, all items were measured on 7-point scales (where 1 = completely disagree and 7 = completely agree). Populist attitudes We used confirmatory factor analyses (CFA) to estimate a two-dimensional structure of populist attitudes, consisting of anti-establishment and exclusionist populist attitudes. The two-dimensional model fitted the data well (χ2[28] = 35.80, χ2/df = 1.28, p = 0.148; RMSEA = .02, 90% CI [0.001, 0.040]; CFI = 0.99)). All standardized regression weights pointed to a satisfactory convergent validity. The model’s discriminant validity was also satisfactory, and the correlation between both factors was .58. Model fit declined substantially and significantly when the correlation between factors was constrained to one. In the first experiment, participants’ populist attitudes were measured on a four-item anti-establishment populism scale (Cronbach’s α = .82, M = 4.37, SD = 1.29; see Appendix B for items). Relative deprivation We expected that perceived relative deprivation would drive participants’ selection of populist media content (also see Elchardus & Spruyt, 2016; Spruyt et al., 2016). We measured perceived relative deprivation on a five-item seven-point scale (see Appendix B; Cronbach’s α = .92, M = 4.33, SD = 1.42). In addition, we included measures of alternative explanations, such as political cynicism (Cronbach’s α = .88, M = 4.82, SD = 1.42), governmental trust (Cronbach’s α = .92, M = 3.02, SD = 1.44), and issue agreement (Cronbach’s α = .64, M = 5.42, SD = 1.21). We assessed whether relative deprivation and populist attitudes tapped into different underlying constructs. The CFA indicated that model fit declined significantly when the correlation between the two constructs was constrained to the value 1 (∆χ2 [1] = 4.83, p = .020). The correlation between both factors was reasonably high (r = .74). Still, based on the assessment of discriminant validity, the correlation was too weak to merge both factors. Populist attitudes and perceived relative deprivation should therefore not be considered as the same construct. We constructed a variable that indicates whether the stimulus chosen by or shown to the participant was congruent with his or her prior perceptions of relative deprivation. For someone lower in relative deprivation, for example, counter attitudinal messages were coded as congruent with this participant’s priors (Table 1). Controls The design of the forced exposure conditions assured a random assignment to treatment and control conditions, assignment thus not resulted in differential attrition (Mutz & Pemantle, 2011). A between-conditions randomization check on the variables of gender, age, education, ideological extremity, and news exposure1 yielded non-significant differences across experimental conditions. Pilot study The manipulations and stimuli were extensively pre-tested in two pilot studies: one pilot study among a convenience sample of university students (N = 52) and one among a diverse sample of Dutch citizens (N = 137). Participants rated the stimuli as credible (M = 4.62, SD = 1.65) and similar to everyday media content (M = 5.09, SD = 1.31). All manipulations succeeded. Participants found it more likely that the article was published on an online news website (M = 5.56, SD = 1.74) than any of the other outlets. For this reason, we framed the articles as online news stories in both studies. Results of Study 1 Drivers of selective exposure to media populism To assess the likelihood of self-selection of populist media content, we compared the six different experimental conditions to the control group (see Table 2). The results of the logistic regression analyses indicate that those participants with stronger perceptions of relative deprivation were more likely to self-select into populist media content than participants with weaker perceptions of relative deprivation. The other variables do not play a significant role in driving selection into populist content. Hypothesis 1 can thus be supported: the more people experience feelings of relative deprivation, the more they are inclined to self-select populist media content. Table 2 Binary Logistic Regression Model Predicting Drivers of Selective Exposure to Anti-Elites Populist and Non-Populist Content Conditions  Populist  Non-Populist  Counter-Blame  Counter–No Blame  Variables  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  (constant)  −4.14 (1.36)**    −2.35 (1.27)    −1.12 (1.69)    −1.15 (1.34)    Age  0.01 (0.01)  [0.98, 1.03]  0.01 (0.10)  [0.99, 1.03]  −0.01 (0.01)  [0.97, 1.01]  −0.01 (.01)  [0.97, 1.01]  Gender  0.18 (0.30)  [0.66, 2.15]  0.52 (0.28)  [0.98, 2.91]  −0.33 (0.41)  [0.32, 1.62]  −0.36 (0.32)  [0.37, 1.30]  Education (higher)  −0.73 (0.39)  [0.23, 1.03]  0.03 (0.30)  [0.57, 1.86]  0.07 (0.45)  [0.45, 2.60]  0.34 (0.33)  [0.73, 2.71]  Deprivation  0.61 (0.15)***  [1.36, 2.49]  −0.09 (0.13)  [0.72, 1.17]  0.26 (0.20)  [0.88, 1.91]  −0.03 (0.15)  [0.73, 1.30]  Issue Agreement  0.05 (0.16)  [0.77, 1.42]  0.07 (0.14)  [0.82, 1.41]  −0.14 (0.20)  [0.59, 1.28]  −0.01 (0.16)  [0.73, 1.35]  Governmental trust  −0.06 (0.12)  [0.75, 1.19]  −0.03 (0.12)  [0.77, 1.23]  0.07 (0.16)  [0.78, 1.48]  0.10 (0.13)  [0.86, 1.42]  Political cynicism  −0.12 (0.15)  [0.67, 1.19]  −0.02 (0.13)  [0.76, 1.28]  −0.25 (0.19)  [0.53, 1.13]  0.02 (0.16)  [0.75, 1.39]  Nagelkerke R2  0.17  0.03  0.04  0.03  χ2 (df)  37.90 (7)***  5.65 (7)  5.32 (7)  5.29 (7)  Conditions  Populist  Non-Populist  Counter-Blame  Counter–No Blame  Variables  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  (constant)  −4.14 (1.36)**    −2.35 (1.27)    −1.12 (1.69)    −1.15 (1.34)    Age  0.01 (0.01)  [0.98, 1.03]  0.01 (0.10)  [0.99, 1.03]  −0.01 (0.01)  [0.97, 1.01]  −0.01 (.01)  [0.97, 1.01]  Gender  0.18 (0.30)  [0.66, 2.15]  0.52 (0.28)  [0.98, 2.91]  −0.33 (0.41)  [0.32, 1.62]  −0.36 (0.32)  [0.37, 1.30]  Education (higher)  −0.73 (0.39)  [0.23, 1.03]  0.03 (0.30)  [0.57, 1.86]  0.07 (0.45)  [0.45, 2.60]  0.34 (0.33)  [0.73, 2.71]  Deprivation  0.61 (0.15)***  [1.36, 2.49]  −0.09 (0.13)  [0.72, 1.17]  0.26 (0.20)  [0.88, 1.91]  −0.03 (0.15)  [0.73, 1.30]  Issue Agreement  0.05 (0.16)  [0.77, 1.42]  0.07 (0.14)  [0.82, 1.41]  −0.14 (0.20)  [0.59, 1.28]  −0.01 (0.16)  [0.73, 1.35]  Governmental trust  −0.06 (0.12)  [0.75, 1.19]  −0.03 (0.12)  [0.77, 1.23]  0.07 (0.16)  [0.78, 1.48]  0.10 (0.13)  [0.86, 1.42]  Political cynicism  −0.12 (0.15)  [0.67, 1.19]  −0.02 (0.13)  [0.76, 1.28]  −0.25 (0.19)  [0.53, 1.13]  0.02 (0.16)  [0.75, 1.39]  Nagelkerke R2  0.17  0.03  0.04  0.03  χ2 (df)  37.90 (7)***  5.65 (7)  5.32 (7)  5.29 (7)  **p < .01; ***p < .001. Note. N = 346. CI = confidence interval; OR = odds ratio. Two-tailed tests. Unstandardized regression weights. Standard errors reported between brackets. The reference category for the reported conditions is the control category. Freedom of selection In a second step, we explored the relationship between the freedom of selection and participants’ populist attitudes. A one-way ANOVA comparing treatment and control conditions between forced and selective exposure environments indicated that the selective exposure control condition differed significantly from all other conditions (F[3, 554] = 8.92, p < .001, partial η2 = .05). In this condition, populist attitudes are relatively low (M = 3.59, SD = 1.06). For all other conditions, the mean scores of populist attitudes are consistent and significantly and substantially higher (M = 4.51, SD = 1.28). This can be interpreted as evidence that those participants who self-selected a control condition have significantly lower populist attitudes than those who self-selected treatment conditions or those who were forced into treatment or control conditions (also see Table 3). Table 3 Effects of Populist Blame Attribution to Elites on Anti-Establishment Populist Attitudes   Model I (n = 558)  Model II (n = 558)  Model III (n = 558)  Model IV (n = 558)  Model V (n = 558)  B  SE  β  B  SE  β  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.37  .12    4.65  .16    4.65  .16    4.65  .15    4.65  .15    Selective Exposure  −0.22  .12  −.04  −0.42  .19  −.16*  −0.52  .19  −.19**  −0.38  .19  −.14*  −0.38  .19  −.14*  Populist Message  0.34  .15  .11*  −0.36  .24  −.11  −0.44  .25  −.14  −0.72  .24  −.23**  −0.79  .25  −.25**  Balanced Message  0.18  .17  .05  −0.22  .24  −.06  −0.22  .24  −.06  −0.22  .23  −.06  −0.22  .23  −.06  Counter-Blame Message  −0.04  .17  −.01  −0.11  .23  −.03  −0.11  .23  −.03  −0.10  .23  −.02  −0.10  .23  −.03  Selective Control  −0.76  .19  −.19***  −0.65  .19  −.16**  −0.55  .20  −.12*  −0.69  .19  −.16**  −0.69  .19  −.16**  Populist × Selection        1.12  .30  .29***  1.13  .30  .29***  0.69  .31  .15  0.86  .36  .18  Balanced × Selection        0.65  .34  .11  0.76  .34  .14*  0. 62  .34  .11  0.62  .34  .11  Counter × Selection        −0.05  .35  −.01  −0.17  .35  −.03  0.67  .48  .11  0.89  .48  .13  Attitudinal Congruence              0.33  .15  .14*  −0.11  .20  −.04  −0.12  .20  −.04  Pop × Congruent                    1.47  .31  .33***  1.76  .39  .40***  Counter-Blame × Congruent                    −1.42  .53  −.19*  −1.44  .56  −.19*  Pop × Congruent × Selection                          −0.49  .48  −.10  Counter × Congruent × Selection                          0.18  .32  .14  Adjusted R2  0.05  0.07  0.08  0.13  0.13  F  6.34***  6.25***  6.11**  8.76***  7.51***  F for Change in R2    5.82**  4.69*  18.89***  0.69    Model I (n = 558)  Model II (n = 558)  Model III (n = 558)  Model IV (n = 558)  Model V (n = 558)  B  SE  β  B  SE  β  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.37  .12    4.65  .16    4.65  .16    4.65  .15    4.65  .15    Selective Exposure  −0.22  .12  −.04  −0.42  .19  −.16*  −0.52  .19  −.19**  −0.38  .19  −.14*  −0.38  .19  −.14*  Populist Message  0.34  .15  .11*  −0.36  .24  −.11  −0.44  .25  −.14  −0.72  .24  −.23**  −0.79  .25  −.25**  Balanced Message  0.18  .17  .05  −0.22  .24  −.06  −0.22  .24  −.06  −0.22  .23  −.06  −0.22  .23  −.06  Counter-Blame Message  −0.04  .17  −.01  −0.11  .23  −.03  −0.11  .23  −.03  −0.10  .23  −.02  −0.10  .23  −.03  Selective Control  −0.76  .19  −.19***  −0.65  .19  −.16**  −0.55  .20  −.12*  −0.69  .19  −.16**  −0.69  .19  −.16**  Populist × Selection        1.12  .30  .29***  1.13  .30  .29***  0.69  .31  .15  0.86  .36  .18  Balanced × Selection        0.65  .34  .11  0.76  .34  .14*  0. 62  .34  .11  0.62  .34  .11  Counter × Selection        −0.05  .35  −.01  −0.17  .35  −.03  0.67  .48  .11  0.89  .48  .13  Attitudinal Congruence              0.33  .15  .14*  −0.11  .20  −.04  −0.12  .20  −.04  Pop × Congruent                    1.47  .31  .33***  1.76  .39  .40***  Counter-Blame × Congruent                    −1.42  .53  −.19*  −1.44  .56  −.19*  Pop × Congruent × Selection                          −0.49  .48  −.10  Counter × Congruent × Selection                          0.18  .32  .14  Adjusted R2  0.05  0.07  0.08  0.13  0.13  F  6.34***  6.25***  6.11**  8.76***  7.51***  F for Change in R2    5.82**  4.69*  18.89***  0.69  *p < .05; **p < .01; ***p < .001 Note. The reference category is the forced control condition. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. The effect of populist cues To explore the impact of populist messages, we compared forced and selective exposure to populist, balanced, and counter-blame messages to the forced exposure control condition (Table 3). The results show that participants in the counter-blame and balanced treatment conditions did not have significantly higher populist attitudes than those in the control condition (Model I). Yet, exposure to a populist message has a positive and significant effect on anti-establishment populist attitudes. This provides support for H2. The role of attitudinal congruence The two-way interaction effect of selective exposure and populist blame attribution (Table 3, Model II) illustrates that populist messages had the strongest effect on populist attitudes when they were self-selected. However, once attitudinal congruence is included in Model IV, the interaction effect of self-selection and exposure to a populist message is no longer significant. Specifically, as indicated by the significant and positive two-way interaction effect of exposure to a populist message and congruence (Table 3, Model IV), participants exposed to congruent populist content had significantly higher anti-establishment populist attitudes than those exposed to incongruent populist messages. Selective exposure did not further specify this relationship (Table 3, Model V), which indicates that congruent populist messages do not have a different effect when self-selected. In support of hypothesis 3, participants exposed to attitudinally congruent elitist blame attributions are persuaded most by populist messages. Study 2 Our first study revealed that the effects of exposure to populist media content are driven by attitudinal congruence: only those participants exposed to messages congruent with their priors are persuaded by populist attributions of blame. By explicitly taking attitudinal congruence into account in the design of the second experiment, we are able to more precisely test how attitudinal biases drive the effects of populist communication. Method Design The design of the second experiment concerns a 2 (attitudinal stance: pro-attitudinal vs. counter-attitudinal) × 2 (choice: populist vs. non-populist framing) + 3 (Forced exposure: populism vs. counter-no-blame vs. control) between-subjects factorial design (see Table 1 for conditions). Participants were divided into groups that viewed congruent versus incongruent framings of the issue immigration based on their pre-exposure scores on a measure of perceived relative deprivation connected to the out-group threat depicted in the stimuli. Moderates were randomly assigned to congruent versus incongruent stimuli. In line with Arceneaux et al. (2012), we controlled for them in the analyses. Sample The experiment was completed by 558 participants (completion rate 72.4%) with a mean age of 49.15 years (SD = 16.09). 47.5% were male and 52.5% were female. 23.7% of the participants were lower educated, 32.8% were higher educated, and 43.5% had a moderate level of education. Procedure The procedure regarding the pre-test was similar to Study 1. After the pre-test, participants completed the following item on a 7-point scale before being randomly allocated to attitudinal incongruent versus incongruent conditions: “People who are not originally from the Netherlands, are profiting more from all kinds of benefits in society than Dutch citizens” (M = 4.58, SD = 1.90). The sample was divided into pro-attitudinal publics (scores ranging from 5 through 7); counter-attitudinal publics (scores ranging from 1 through 3) and moderates (scoring the midpoint on the scale). All these groups were randomly assigned to the pro- and counter-attitudinal forced and choice conditions. Independent variables and stimuli The second experiment focused on increasing crime rates in the light of immigration, a salient monocultural, right-wing populist issue. Selective exposure was manipulated by presenting participants with a cover story describing how people may prefer some news items above others available to them. Participants were told that they should select one preferred article out of two alternative online articles. One article was framed in a populist way, whereas the other article was framed in a non-populist way (see Table 1 and examples in Appendix A and Supplemental Material). In the forced exposure conditions, participants were randomly assigned to either populist or non-populist stimuli. Informed by the results of Study 1, attitudinal stance was manipulated into a pro-attitudinal versus a counter-attitudinal framing of the issue, either highlighting that migrants or Dutch people were associated with the increasing crime rate. Based on their scores on the relative deprivation measure, we assigned participants to congruent and incongruent pro- and counter-attitudinal conditions at a 1:1 rate. Manipulation checks The manipulation of populist blame attribution to migrants was successful (F[1, 553] = 157.25, p < .001). Participants in the blame attribution conditions were significantly more likely to believe that the article attributed responsibility to migrants (M = 5.68, SD = 1.23) than participants in the other conditions (M = 3.71, SD = 1.77). The manipulation of selective exposure also succeeded (F[1, 553] = 87.44, p < .001). Participants in the selective exposure conditions were significantly more likely to perceive they were offered the choice to self-select an article (M = 4.81, SD = 1.58) than participants in the forced exposure conditions (M = 3.50, SD = 1.72). Measures We used the same measures as reported in Study 1. Again, the CFA for populist attitudes demonstrated good model fit (χ2[27] = 43.87, χ2/df = 1.63, p = .021; RMSEA = .03, 90% CI [.01, .05]; CFI = .99). The correlation between factors was .69 and model fit declined substantially and significantly when the correlation between factors was constrained to one. Participants’ populist attitudes were measured on a 7-point exclusionist populist attitudes scale (Cronbach’s α = .94, M = 4.14, SD = 1.71). The correlation between populist attitudes and perceived relative deprivation was (r = . 56). The model fit for the CFA with one factor, including items for populist attitudes and perceived deprivation, was significantly and substantially worse than a model that treats populist attitudes and relative deprivation as separate scales (∆χ2 [1] = 19.01, p < .01). Results of Study 2 First, we compared mean scores on exclusionist populist attitudes between the control, selective treatment, and forced treatment conditions. The one-way ANOVA yielded non-significant differences in populist attitudes between conditions (F[2, 555] = 1.89, p = .151, partial η2 = .001). This indicates that, ceteris paribus, exclusionist populist attitudes were similar across the control and both the forced and selective treatment conditions. Selective exposure to attitudinal congruent populist versus non populist cues As illustrated in Table 4, participants who self-selected into congruent populist blame attributions scored higher on exclusionist populist attitudes (M = 5.64, SD = 1.02) than participants who selected non-populist content (M = 4.85, SD = 1.13). In support of hypothesis 2, these results indicate that people exposed to populist cues have higher populist attitudes than people exposed to non-populist cues. Table 4 The Effects of Populist and Non-Populist Stimuli on Anti-Establishment Populist Attitudes at Different Levels of Attitudinal Congruence (Monocultural Media Populism)   Lower in Relative Deprivation  Higher in Relative Deprivation  F  Partial η2  Congruent  Incongruent  Congruent  Incongruent      Counter-Blame  Counter–No Blame  Populist  Non-Populist  Populist  Non-Populist  Counter-Blame  Counter–No Blame      Choice  2.74a, x (1.48)  2.35a (1.14)  2.35a, x (1.19)  2.39a (1.22)  5.64b, x (1.02)  4.85c (1.13)  5.27b, x (1.49)  5.10c (1.13)  24.04*  .38  Forced Exposure  2.12a, y (1.13)  -  2.82b, y (1.29)  -  5.09c, y (1.50)  -  5.04c, x (1.37)  -  25.78*  .28  Control  2.27a, (0.97)  5.09b (1.47)        Lower in Relative Deprivation  Higher in Relative Deprivation  F  Partial η2  Congruent  Incongruent  Congruent  Incongruent      Counter-Blame  Counter–No Blame  Populist  Non-Populist  Populist  Non-Populist  Counter-Blame  Counter–No Blame      Choice  2.74a, x (1.48)  2.35a (1.14)  2.35a, x (1.19)  2.39a (1.22)  5.64b, x (1.02)  4.85c (1.13)  5.27b, x (1.49)  5.10c (1.13)  24.04*  .38  Forced Exposure  2.12a, y (1.13)  -  2.82b, y (1.29)  -  5.09c, y (1.50)  -  5.04c, x (1.37)  -  25.78*  .28  Control  2.27a, (0.97)  5.09b (1.47)      * = p < .001 Note. N = 555. Standard deviations are reported in parentheses below the means. Means with differing first subscripts within rows and differing second subscripts within columns differ significantly at the p < .05 level based on post hoc independent samples t-test. From top to the bottom, the F-values represent: (1) the main effect of the experimental conditions in the selective exposure conditions; (2) the main effect of the experimental conditions within forced exposure. Attitudinal congruence and monocultural media populism In Table 5, we see a significant and positive two-way interaction effect of the populist stimuli and attitudinal congruence. This indicates that, in the forced exposure conditions, exclusionist populist attitudes are significantly higher among participants exposed to congruent than incongruent populist blame attribution. The results further indicate a significant and negative two-way interaction effect of exposure to counter-blame stimuli and attitudinal congruence. Participants at lower levels of relative deprivation exposed to a message that blames the native people for the increasing crime rate have lower exclusionist populist attitudes compared to participants who are exposed to incongruent populist messages (see Table 4). Table 5 Effects of Populist Blame Attribution to Migrants on Exclusionist Populist Attitudes Within Forced Exposure Conditions   Model I (n = 277)  Model II (n = 277)  Model III (n = 277)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.13  .12    4.20  .15    4.08  .15    Populist  0.34  .23  .08  0.38  .24  .09  −1.26  .36  −.28*  Counter-Blame  0.14  .22  .03  0.21  .29  .05  1.25  .26  .30**  Moderates        −0.80  .56  −.08  −1.73  .51  −.16*  Attitudinal Congruent        −0.14  .19  −.04  0.10  .22  .03  Populist × Congruent              2.18  .45  .43**  Counter-Blame × Congruent              −3.31  .44  −.45**  Adjusted R2  .01  .01  .21  F  1.13  1.17  18.14**  F for Change in R2    1.20  51.48**    Model I (n = 277)  Model II (n = 277)  Model III (n = 277)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.13  .12    4.20  .15    4.08  .15    Populist  0.34  .23  .08  0.38  .24  .09  −1.26  .36  −.28*  Counter-Blame  0.14  .22  .03  0.21  .29  .05  1.25  .26  .30**  Moderates        −0.80  .56  −.08  −1.73  .51  −.16*  Attitudinal Congruent        −0.14  .19  −.04  0.10  .22  .03  Populist × Congruent              2.18  .45  .43**  Counter-Blame × Congruent              −3.31  .44  −.45**  Adjusted R2  .01  .01  .21  F  1.13  1.17  18.14**  F for Change in R2    1.20  51.48**  *p < .01; **p < .001 Note. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. As can be seen in Table 6, the two-way interaction effect of populist blame attribution to migrants and attitudinal congruence is also positive and significant in the selective exposure conditions. This means that when participants higher in relative deprivation are exposed to populist attributions of blame to migrants, their exclusionist populist attitudes are stronger than in the other conditions. Table 6 Effects of Populist Blame Attribution to Migrants on Exclusionist Populist Attitudes Within Selective Exposure Conditions   Model I (n = 281)  Model II (n = 281)  Model III (n = 281)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.38  .13    4.52  .17    4.82  .16    Populist  0.39  .30  .07  0.38  .23  .07  −2.47  .47  −.39**  Counter-Blame  −0.22  .33  −.04  −0.29  .33  −.05  0.69  .41  .11  Non-Populist  −0.75  .26  −.15*  −0.81  .25  −.16*  −0.81  .25  −.16*  Counter–No Blame  −0.20  .24  −.04  −0.29  .25  −.06  −0.31  .23  −.07  Moderates        −0.91  .54  −.09  −0.97  .51  −.9  Attitudinal Congruent        −0.19  .19  −.05  −0.32  .19  −.09  Populist × Congruent              4.06  .56  .62***  Counter-Blame × Congruent              −2.22  .59  −.24**  Non-Populist × Congruent              2.23  .44  .38**  Counter–No Blame × Congruent              −1.94  .43  −.26**  Adjusted R2  .02  .02  .28  F  3.11  2.63*  18.53**  F for Change in R2    1.65  48.43**    Model I (n = 281)  Model II (n = 281)  Model III (n = 281)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.38  .13    4.52  .17    4.82  .16    Populist  0.39  .30  .07  0.38  .23  .07  −2.47  .47  −.39**  Counter-Blame  −0.22  .33  −.04  −0.29  .33  −.05  0.69  .41  .11  Non-Populist  −0.75  .26  −.15*  −0.81  .25  −.16*  −0.81  .25  −.16*  Counter–No Blame  −0.20  .24  −.04  −0.29  .25  −.06  −0.31  .23  −.07  Moderates        −0.91  .54  −.09  −0.97  .51  −.9  Attitudinal Congruent        −0.19  .19  −.05  −0.32  .19  −.09  Populist × Congruent              4.06  .56  .62***  Counter-Blame × Congruent              −2.22  .59  −.24**  Non-Populist × Congruent              2.23  .44  .38**  Counter–No Blame × Congruent              −1.94  .43  −.26**  Adjusted R2  .02  .02  .28  F  3.11  2.63*  18.53**  F for Change in R2    1.65  48.43**  *p < .01; **p < .001 Note. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. In the selective exposure conditions, the analyses further point to a significant, negative two-way interaction effect of the selection of counter-blame content and attitudinal congruence (see Table 6). This indicates that people who self-select into counter-blame content that fits their prior attitudes of lower relative deprivation have weaker populist attitudes compared to participants that self-select into incongruent counter-blame (see Table 4 for mean score differences). In support of H3, participants exposed to attitudinal congruent monocultural media populism have stronger exclusionist populist attitudes than participants exposed to incongruent populist communication (see Figure 1). Figure 1 View largeDownload slide Marginal effect plots for interaction attitudinal congruence and populist blame attribution in forced and selective exposure conditions. Note. Dots represent regression weights and lines represent 95% confidence intervals. N = 558. Figure 1 View largeDownload slide Marginal effect plots for interaction attitudinal congruence and populist blame attribution in forced and selective exposure conditions. Note. Dots represent regression weights and lines represent 95% confidence intervals. N = 558. Discussion In the midst of the rise of persuasive populist ideas throughout the globe, a growing body of literature has pointed to the key role of the media in disseminating the populist message to the people (Krämer, 2014; Mazzoleni, 2008; Mudde, 2004). This literature has also assumed that those people who selectively expose themselves to populist messages form a specific type of audience. Specifically, those attracted to populist communication are described as a discontented (Mazzoleni, 2008) and relatively deprived group of citizens (Elchardus & Spruyt, 2016). Yet, at the same time, the scarce body of literature that has studied the effects of populist communication has done so in a forced-exposure environment (e.g., Bos et al., 2013). Therefore, an important discrepancy in populism research is that, despite the assumptions about the peculiarities of the audience susceptible to persuasion by populist communication, there are no studies that actually simulate a selective exposure media environment. Advancing this line of research, this paper presents two experiments that do take selective exposure and attitudinal congruence into account. In line with recent literature showing that citizens who feel relatively deprived are more likely to hold populist perceptions (Elchardus & Spruyt, 2016), our results show that these citizens are also more likely to be drawn to populist content. Addressing the ordinary people as a blameless, relatively deprived in-group juxtaposed to a causally responsible enemy makes populist discourse stick. In the midst of an increasingly more fragmentized media environment (e.g., Stroud, 2008), the findings of this paper demonstrate that people who self-select into populist communication differ substantially from those who circumvent such content. In line with recent literature, it is indeed the citizens who feel relatively deprived that expose themselves to populist messages (Elchardus & Spruyt, 2016). Citizens with weaker perceptions of deprivation avoid populist content. It was, however, not selective exposure but attitudinal congruence that conditioned the effects of exposure to media populism on populist attitudes. In both forced and selective exposure media environments, populist blame attributions only affected the populist attitudes of people for whom the message was congruent with their priors. Based on the evidence from the experiments presented in this paper, the role of giving people the freedom to self-select into media exposure is outweighed by exposure to a message that is in line with their prior attitudes. Hence, our results raise the question of whether future research should take the effort to methodologically simulate a selective exposure environment if dividing the sample into congruent and incongruent issue publics has the same effect on the outcomes. This methodological note does not mean that selective exposure does not play a role in conditioning the effects of media populism on the populist perceptions of the electorate. Backed up by the empirical evidence from our first experiment, exposure to attitudinal congruent populist media content is most likely to occur among citizens higher in relative deprivation. For these citizens, media populism confirms their priors of being deprived by some culprit other opposed to ordinary people. This finding is in line with the theory on motivated reasoning (Lodge & Taber, 2013). Outside of the experimental setting, then, it is unlikely that people for whom the populist message is counter to their priors actually expose themselves to media populism. This key finding has important implications for the polarizing effect of exposure to attitudinal congruent populist content. People who avoid populist content because it counters their priors are least likely to be affected by populist communication. When these people are forced into exposure to populist content, the message can even have a backlash. In that case, their negative priors towards populist interpretations are activated by the incongruent populist message. Those citizens that are higher in relative deprivation already have higher populist attitudes. Yet exposure to congruent populist content activates their priors, resulting in a bolstering of their existing populist attitudes on both the anti-establishment and exclusionist dimension. We interpret this as evidence for the polarizing potential of media populism in a fragmentized media environment (Stroud, 2008). Indeed, exposure to messages that articulate the causal and moral divide between us and them actually widens gaps in society, pitting those with populist interpretations against those who feel distant from such worldviews. The findings of this paper indicate that selective exposure and attitudinal congruent persuasion have important consequences for media effect studies beyond the U.S. context of partisanship and (affective) polarization (e.g., Iyengar & Hahn, 2009). Specifically, to better understand how media content affects public perceptions of salient socio-political issues—such as populism—selection should be taken into account (also see Bennett & Iyengar, 2008). In the current era of choice and fragmentation, it is crucial to determine who avoids or selects political communication biased toward a certain interpretation of reality. Moving beyond selection as dependent variable, our findings indicate that persuasion is contingent upon attitudinal congruence, and that self-selection does not further bolster the effectiveness of polarized media content. Advancing the literatures on selective exposure and motivated reasoning, these findings indicate that people are affected most when political communication confirms their priors, and that the freedom to select reassuring content does not necessarily makes it more persuasive. The underlying information processing mechanism can be understood as a two-step model of selection and attitudinal congruent persuasion. First, people select media content that reassures their existing interpretations of reality. Second, only those with congruent priors are positively affected by such content, making them more aligned with populist worldviews. People who disagree with such worldviews are not likely to select populist content, and thus not likely to be persuaded as long as they can avoid such content. Indeed, people who are confronted with incongruent content may defend their prior attitudes to counter-argue attacks on their existing beliefs. Because avoidance has crucial ramification for media effects, an important avenue for selective exposure research is to assess the extent to which the audience is actually able to avoid certain political content in their daily media environment. These findings have important practical implications. Because perceived relative deprivation is the main driver of attitudinal congruent persuasion, future interventions should focus on a correction of these deprived beliefs among these citizens. Selective exposure plays a key role in people’s everyday media environment. If populist messages are difficult to avoid in daily life, and if these messages counter people’s priors, populist communication may have a backlash, reinforcing citizens’ already negative prior attitudes toward populist interpretations. In that sense, it may be more important to base interventions on lowering perceived deprivation than exposing people to different content. Specifically, political campaigns should explicitly relativize causes and consequences of people’s deprivation, for example by showing people that they are actually not worse off than other groups in society and that refugees are facing more challenges than native people. If external causes for the people’s crisis forwarded by populist parties—such as “profiting” migrants—can be effectively refuted in political campaigns, the electorates’ perceptions of relative deprivation can be lowered further. Here, it is important that misinformation on mainstream political parties’ and immigrants’ involvement in causing various socio-political problems is effectively discredited. Current evidence regarding the effectiveness of corrective attempts is, however, still mixed (see for example Thorson, 2016). Discrediting fake news on blame and deprivation could also be regarded as a task for the media, such as via fact checker websites. By indicating which attributions of blame are actually correct or not, the electorate will be better able to judge who is responsible for their experienced problems. Our studies bear some limitations that may be dealt with in future research. First and foremost, we could not randomly assign participants to the selective exposure conditions, which limits the causal conclusions that can be drawn from the choice conditions. Against this backdrop, we have to acknowledge that choice experiments result in more realistic treatment effects at the cost of introducing a self-selection bias because randomization is not achieved. Improving external validity thus comes at the cost of internal validity (e.g., Gaines & Kuklinski, 2011). However, as we accounted for the most important factor that drives selective exposure to media populism—relative deprivation—in addition to a post hoc randomization check, we do believe that we have at least partially dealt with this concern. An important consideration in the experimental design of selective exposure studies concerns the number of options offered to participants (see Feldman et al., 2013, for a comprehensive overview). To more closely simulate a real-world high-choice media environment, future research may include different options in the selective exposure environment relevant to the effects of media populism, such as offering a choice between tabloid and broadsheet media outlets (e.g., Mazzoleni, 2008). This may also allow for a further exploration of why people choose populist content in a fragmented media environment with a plethora of alternatives. Another limitation concerns the potential of ceiling effects found after exposure to populist messages among the public. Independent of the stimuli, people with similar perceptions of relative deprivation have stronger populist attitudes (in line with Hameleers et al., 2017 and Elchardus & Spruyt, 2016). On the measured scales, there was not much room for the stimuli to further boost their populist attitudes. Populist attitudes and relative deprivation can also be interpreted from an alternative, reciprocal causal framework. People may select populist content because they feel relatively deprived (Elchardus & Spruyt, 2016). Populist content hereby responds to people’s perceived deprivation, making people more aligned with an antagonistic, populist interpretation of society. In a second step of this mechanism, people may feel even more deprived as a result of like-minded populist media framing. As a consequence of this, they may seek out even more populist content. It may be an interesting avenue for future research to tease out such a reinforcing spiral of populism and deprivation. Future research could, for example, rely on pre-and post-exposure measures of relative deprivation and populist attitudes to more precisely investigate how these perceptions were primed differentially. Despite these limitations, the studies presented in this paper are the first to demonstrate how the effects of media populism are driven by attitudinal congruence. In showing that populist communication is only selected by and persuasive for those higher in relative deprivation, this paper provides an important contribution to the literature that has argued that those citizens that populist messages appeal to form a specific type of audience that can be distinguished from those who circumvent populist coverage. Hence, in the midst of the revival of the populist zeitgeist, this paper demonstrates which part of the electorate is most likely to expose itself to populist messages to further bolster their priors, with eventually paramount societal and democratic implications in terms of a polarization between the populist and non-populist electorate. Supporting Information Supporting information is available at Journal of Communication online. Note 1 A series of chi-squared tests revealed no significant differences across conditions for age (forced: χ2 [6] = 3.32, p = n. s.; selective: χ2 [2] = 2.41, p = n. s), gender (forced: χ2 [3] = 8.62, p = n. s.; selective: χ2 [1] = .35, p = n. s.;), education (forced: χ2 [8] = 8.73, p = n. s.; selective: χ2 [2] = .19, p = n. s.), ideological extremity (forced: χ2 [3] = 1.22, p = n. s; selective: χ2 [1] = .07, p = n. s), and news exposure (forced χ2 [24] = 31.22, p = n. s.; selective: χ2 [7] = 5.02, p = n.s.) References Akkerman, A., Mudde, C., & Zaslove, A. ( 2014). How populist are the people? Measuring populist attitudes in voters. Comparative Political Studies , 47( 9), 1324– 1353. doi:10.1177/0010414013512600 Google Scholar CrossRef Search ADS   Arceneaux, K., Johnson, M., & Murphy, C. ( 2012). Polarized political communication, oppositional media hostility, and selective exposure. The Journal of Politics , 74( 1), 174– 186. doi:S002238161100123X. Google Scholar CrossRef Search ADS   Bennett, W. L., & Iyengar, S. ( 2008). A new era of minimal effects? The changing foundations of political communication. Journal of Communication , 58( 4), 707– 731. doi:10.1111/j.1460-2466.2008.00410.x Google Scholar CrossRef Search ADS   Bos, L., van der Brug, W., & de Vreese, C. H. ( 2013). An experimental test of the impact of style and rhetoric on the perception of right-wing populist and mainstream party leaders. Acta Politica , 48( 2), 192– 208. doi:10.1057/ap.2012.27 Google Scholar CrossRef Search ADS   Caiani, M., & della Porta, D. ( 2011). The elitist populism of the extreme right: A frame analysis of extreme right-wing discourses in Italy and Germany. Acta Politica , 46( 2), 180– 202. doi:10.1057/ap.2010.28 Google Scholar CrossRef Search ADS   Canovan, M. ( 1999). Trust the people! Populism and the two faces of democracy. Political Studies , 47, 2– 16. doi:10.1111/1467-9248.00184 Google Scholar CrossRef Search ADS   Dixon, T. L. ( 2008). Crime news and racialized beliefs: Understanding the relationship between local news viewing and perceptions of African Americans and Crime. Journal of Communication , 58( 1), 106– 125. doi:10.1111/j.1460-2466.2007.00376.x Google Scholar CrossRef Search ADS   Elchardus, M., & Spruyt, B. ( 2012). The contemporary contradictions of egalitarianism: An empirical analysis of the relationship between the old and new left/right alignments. European Political Science Review , 4( 2), 217– 239. doi:1.1017/S1755773911000178 Google Scholar CrossRef Search ADS   Elchardus, M., & Spruyt, B. ( 2016). Populism, persistent republicanism and declinism: An empirical analysis of populism as a thin ideology. Government and Opposition , 51( 1), 111– 133. doi:10.1017/gov.2014.27 Google Scholar CrossRef Search ADS   Engesser, S., Ernst, N., Esser, F., & Büchel, F. ( 2016). Populism and social media: How politicians spread a fragmented ideology. Information, Communication & Society , 20( 8), 1109– 1126. doi:10.1080/1369118X.2016.1207697 Google Scholar CrossRef Search ADS   Entman, R. M. ( 1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication , 43( 4), 51– 59. doi:10.1111/j.1460-2466.1993 Google Scholar CrossRef Search ADS   Feldman, L., Stroud, N. J., Bimber, J., & Wojcieszak, M. ( 2013). Assessing selective exposure in experiments: The implications of different methodological choices. Communication Methods and Measures , 7( 3-4), 172– 194. doi:10.1080/19312458.2013.813923 Google Scholar CrossRef Search ADS   Festinger, L. ( 1957). A theory of cognitive dissonance . Evanston, IL: Row, Peterson & Company. Gaines, B. J., & Kuklinski, J. ( 2011). Experimental estimation of heterogeneous treatment effects for treatments related to self-selection. American Journal of Political Science , 55( 3), 724– 736. doi:10.1111/j.1540-5907.2011.00518.x Google Scholar CrossRef Search ADS   Hameleers, M., Bos, L., & de Vreese, C. H. ( 2017). “They did it”: The effects of emotionalized blame attribution in populist communication. Communication Research , 44( 6), 870– 900. doi:10.1177/0093650216644026 Google Scholar CrossRef Search ADS   Hawkins, K. A. ( 2009). Is Chávez populist? Measuring populist discourse in comparative perspective. Comparative Political Studies , 42( 8), 1040– 1067. doi:10.1177/0010414009331721 Google Scholar CrossRef Search ADS   Hewstone, M. ( 1989). Causal attribution: From cognitive processes to collective beliefs . Oxford, UK: Blackwell. Hobolt, S. B., & Tilley, J. ( 2014). Blaming Europe? Responsibility without accountability in the European Union . Oxford, UK: Oxford University Press. Google Scholar CrossRef Search ADS   Hogg, M. A., Meehan, C., & Farquharson, J. ( 2010). The solace of radicalism: Self uncertainty and group identification in the face of threat. Journal of Experimental Social Psychology , 46( 6), 1061– 1066. doi:10.1016/j.jesp.2010.05.005 Google Scholar CrossRef Search ADS   Iyengar, S. ( 1989). How citizens think about national issues: A matter of responsibility. American Journal of Political Science , 33( 4), 878– 900. doi:10.2307/2111113 Google Scholar CrossRef Search ADS   Iyengar, S., & Hahn, K. S. ( 2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication , 59( 1), 19– 39. doi:10.1111/j.1460-2466.2008.01402.x Google Scholar CrossRef Search ADS   Jagers, J., & Walgrave, S. ( 2007). Populism as political communication style: An empirical study of political parties’ discourse in Belgium. European Journal of Political Research , 46( 3), 319– 345. doi:10.1111/j.1475-6765.2006 Google Scholar CrossRef Search ADS   Krämer, B. ( 2014). Media populism: A conceptual clarification and some theses on its effects. Communication Theory , 24, 42– 60. doi:10.1111/comt.12029 Google Scholar CrossRef Search ADS   Kriesi, H.-P., Grande, E., Lachat, R., Dolezal, M., Bornschier, S., & Frey, T. ( 2006). Globalization and the transformation of the national political space: Six European countries compared. European Journal of Political Research , 45( 6), 921– 956. doi:10.1111/j.1475-6765.2006.00644.x Google Scholar CrossRef Search ADS   Laclau, E. ( 1977). Politics and ideology in Marxist theory . London, UK: New Left Books. Lodge, M., & Taber, C. S. ( 2013). The rationalizing voter . Cambridge, UK: Cambridge University press. Google Scholar CrossRef Search ADS   Marsh, M., & Tilley, J. ( 2010). The attribution of credit and blame to governments and its impact on vote choice. British Journal of Political Science , 40( 1), 115– 134. doi:0.1017/S0007123409990275 Google Scholar CrossRef Search ADS   Matthes, J., & Schmuck, D. ( 2017). The effects of anti-immigrant right-wing populist ads on implicit and explicit attitudes: A moderated mediation model. Communication Research , 44( 4), 556– 581. doi:10.1177/0093650215577859 Google Scholar CrossRef Search ADS   Mazzoleni, G. ( 2008). Populism and the media. In D. Albertazzi & D. McDonnell (Eds.), Twenty-first century populism: The spectre of Western European democracy  (pp. 49– 64). Basingstoke, UK: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Mudde, C. ( 2004). The populist zeitgeist. Government and Opposition , 39( 4), 542– 564. doi:10.1111/j.1477-7053.2004.00135.x Google Scholar CrossRef Search ADS   Mutz, D. C., & Pemantle, R. ( 2011, July). The perils of randomization checks in the analysis of experiments. Paper presented at the Annual Meeting of the Society for Political Methodology, Princeton, NJ. Pappas, T. S. ( 2014). Populist democracies: Post-authoritarian Greece and post-communist Hungary. Government and Opposition , 49( 1), 1– 23. doi:10.1017/gov.2013.21 Google Scholar CrossRef Search ADS   Richey, S. ( 2012). Campaign advertising and the stimulation and activation of the authoritarian personality. Political Communication , 29( 1), 24– 43. doi:10.1080/10584609.2011.616874 Google Scholar CrossRef Search ADS   Rooduijn, M. ( 2014). The mesmerising message: The diffusion of populism in public debates in Western European media. Political Studies , 62( 4), 726– 744. doi:10.1111/1467-9248.12074 Google Scholar CrossRef Search ADS   Spruyt, B., Keppens, G., & van Droogenbroeck, F. ( 2016). Who supports populism and what attracts people to it? Political Research Quarterly , 69( 2) 335– 346. doi:10.1177/1065912916639138 Google Scholar CrossRef Search ADS   Stroud, N. J. ( 2008). Media use and political predispositions: Revisiting the concept of selective exposure. Political Behavior , 30, 341– 377. doi:10.1007/s11109-007-9050-9 Google Scholar CrossRef Search ADS   Taggart, P. ( 2000). Populism.  Buckingham & Philadelphia: Open University Press. Tajfel, H. ( 1978). Social categorization, social identity, and social comparisons. In H. Tajfel (Ed.). Differentiation between social groups  (pp. 254– 267). London, UK: Academic Press. Thorson, E. ( 2016). Belief echoes: The persistent effects of corrected misinformation. Political Communication , 33, 460– 480. doi:10.1080/10584609.2015.1102187 Google Scholar CrossRef Search ADS   Waisbord, S., & Amado, A. ( 2017). Populist communication by digital means: Presidential Twitter in Latin America. Information, Communication & Society , 20( 9), 1330– 1346. doi:10.1080/1369118X.2017.1328521 Google Scholar CrossRef Search ADS   © 2018 International Communication Association http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Communication Oxford University Press

Selective Exposure to Populist Communication: How Attitudinal Congruence Drives the Effects of Populist Attributions of Blame

Loading next page...
 
/lp/ou_press/selective-exposure-to-populist-communication-how-attitudinal-Hdvo6iqkNT
Publisher
Oxford University Press
Copyright
© 2018 International Communication Association
ISSN
0021-9916
eISSN
1460-2466
D.O.I.
10.1093/joc/jqx001
Publisher site
See Article on Publisher Site

Abstract

Abstract Although it has been argued that populist communication only appeals to a specific audience, extant research has exclusively studied the effects of populist communication in forced-exposure media environments. Responding to this discrepancy, we conducted two experiments (N = 562 and N = 558) in which we manipulated the core idea of populist messages—attributing blame—in forced and selective exposure media environments. Our results demonstrate that citizens higher in relative deprivation are most likely to select populist messages. Irrespective of selective exposure, citizens’ populist attitudes are only positively affected if the populist message is congruent with their prior feelings of relative deprivation. These results provide important insights for the polarizing potential of media populism in a fragmented media environment. Populism has become influential in the political landscape of many democracies around the world. In recent years, the populist ideas expressed by a variety of actors and movements all around the globe have affected politics and society in important ways, potentially leading to a polarized electorate (e.g., Pappas, 2014). In populist discourse, the ordinary people are addressed as the “good” in-group. This in-group should consequentially be the focal point of political decision-making (Canovan, 1999; Mudde, 2004). The elites are constructed as a corrupt enemy that deprives the people from their deserved earnings, and are accused of failing to represent the people’s will. The core idea of populism thus revolves around the moral and causal divide between blameless ordinary people and culprit others (e.g., Hameleers, Bos, & de Vreese, 2017; Mudde, 2004). The persuasiveness of such populist blame attributions forms the backdrop of this paper: what are the effects of populist messages, and which citizens are most likely to select and be persuaded by them? It has been demonstrated that exposure to populist communication affects citizens’ political perceptions (e.g., Bos, van der Brug, & de Vreese, 2013). Extant literature further pointed to the media as an important supply-side factor facilitating the dissemination of populist ideas in society (e.g., Bos et al., 2013; Mazzoleni, 2008; Waisbord & Amado, 2017). Two explanations of the media’s role in the spread of populism have been forwarded. First, driven by media logic, the media are said to offer favorable attention to the newsworthy ideas of populist politicians (e.g., Mazzoleni, 2008). Second, journalists are themselves assumed to actively frame issues along the lines of a populist distinction in society (e.g., Krämer, 2014; Mazzoleni, 2008). This journalistic practice has been described as populism by the media or as media populism (Krämer, 2014; Mazzoleni, 2008). This implies that the media, and not only politicians, engage in populist communication by actively stressing the moral and causal divide between the blameless people and the culprit others. Backed up by empirical evidence, we argue that such mediatized attributions of blame can be regarded as a persuasive frame that activates populist schemata among receivers (Hameleers et al., 2017; Krämer, 2014). Such populist blame attributions may not convince all citizens. In line with selective exposure theory, people are only expected to select populist content if they have similar worldviews. Outside of experimental settings, people are only expected to expose themselves to populist content if their prior beliefs are congruent with it (Festinger, 1957). Therefore, traditional experiments that force participants into viewing messages that they would normally not select themselves do not reflect the fragmented media environment of citizens (Bennett & Iyengar, 2008). But what prior beliefs drive people’s selection of populist media content? What are the consequences of being forced to read incongruent populist messages? To answer these questions, we conducted two online survey experiments (N = 562 and N = 558) in which selective exposure was linked to different types of media populism framing different societally relevant issues. This allowed us to investigate whether it is indeed the case that populist messages are most persuasive among the part of the electorate that self-selects these messages, motivated by attitudinal congruence. Populist attributions of blame in media populism Populism revolves around the construction of a societal divide between ordinary citizens as a blameless in-group versus the “corrupt” elites as a culpable out-group (e.g., Caiani & della Porta, 2011; Mudde, 2004; Taggart, 2000). Populist discourse addresses the in-group of the people as being victimized by the corrupt elites. These elites do not represent ordinary citizens’ will and deprive them from cultural and material resources. The elites are accused of favoring other groups in society instead of their “own” people (Elchardus & Spruyt, 2016). Populist discourse presents issues in “black and white” or “us versus them” interpretations, hereby simplifying complex political issues into an antagonistic societal divide (Taggart, 2000). By repeatedly exposing people to such simplified messages that reinforce a positive image of the deprived people and a negative image of the elite and other out-groups, populist discourse is argued to be highly persuasive (e.g., Rooduijn, 2014). Attributing blame to the elites or societal out-groups forms the ideational core of populist messages (e.g., Hameleers et al., 2017). Populist ideas articulate the divide between us and them by scapegoating various out-groups for causing the crisis while absolving ordinary people of blame (e.g., Taggart, 2000). Populist communication provides citizens with an influential heuristic cue that helps them to translate complex socio-political issues into black and white terms (Hewstone, 1989; Iyengar, 1989). Attributions of responsibility hereby fulfill an important role in citizens’ political decision making, as they can heuristically use responsibility cues to punish the culprits (Marsh & Tilley, 2010). Attributions of blame guide citizens’ political opinions in important ways (Hobolt & Tilley, 2014; Iyengar, 1989). Specifically, if certain politicians are attributed blame, citizens evaluate them in more negative terms and are less likely to vote for them (Marsh & Tilley, 2010). It has been demonstrated that the media play an important role in the effects of blame attributions, as the framing of societal issues in terms of “who has done it” influences citizens’ own perceptions of blame (e.g., Iyengar, 1989). It must be noted here that attributing blame to political opponents is a common political phenomenon. Populist blame attribution differs from general attributions of responsibility by emphasizing both an antagonistic causal and moral divide between the good people and the culpable other. Beyond highlighting which actors are responsible, populist blame attributions emphasize that the elites are failing to represent the people’s will, depriving them from what they deserve. Populist messages attribute blame to different out-groups deemed responsible for ordinary people’s problems (Jagers & Walgrave, 2007). First, blame is attributed to the elites (e.g., Mudde, 2004; Rooduijn, 2014). The most salient elitist enemies are the politicians in government, who are accused of not representing ordinary people and their will (e.g., Hawkins, 2009; Mudde, 2004). Right-wing populist messages can additionally emphasize a societal divide between the native people as a blameless in-group and horizontally opposed others as culpable out-groups (Jagers & Walgrave, 2007). This populist divide entails the interpretation that refugees, immigrants, ethnic minorities, welfare-state profiteers, and all other out-groups living amongst the people are responsible for causing the problems of the native ordinary citizens, such as an increasing crime rate or a declining economy. The populist core idea of in-group favoritism and out-group hostility lies at the heart of media populism (Krämer, 2014; Mazzoleni, 2008). Media populism can be defined as journalists’ active reconstruction of populist ideas, acting independently of the political actors associated with populism (Krämer, 2014). These populist ideas are most saliently reflected in the journalistic interpretation of a divide between the “silenced majority” of ordinary citizens and the corrupt elites or culpable out-groups (e.g., Caiani & della Porta, 2011). This implies that, beyond disseminating the hard facts of events, journalists who engage in media populism actively emphasize or frame an interpretation of the background, causes, moral evaluations, and treatment recommendation when reporting on the news (e.g., Entman, 1993). Specifically, media populism attributes causal responsibility for negative developments while emphasizing the moral divide between the “good” people and the “evil” others. The treatment recommendation is to remove the elites or out-groups. The effects of populist communication on populist interpretations Populism is not only communicated by politicians or journalists. Citizens can interpret issues along the lines of populist distinctions themselves as well (e.g., Akkerman, Mudde, & Zaslove, 2014; Rooduijn, 2014). Such individual-level populist attitudes tap into citizens’ sentiments of a causal and moral divide between ordinary people and culprit others. The process through which populist messages affect populist attitudes can be described as “media-based othering” (Krämer, 2014). As a result of exposure to messages that attribute blame to elites or societal out-groups, citizens may interpret issues in congruent us versus them frames as well. This effect can be interpreted in light of the psychological mechanisms of social identity and stereotyping (e.g., Tajfel, 1978). These mechanisms postulate that positive images of the in-group and negative stereotypes of the out-group are primed and chronically accessible among receivers (e.g., Dixon, 2008). Exposure to populist messages may thus stimulate and activate populist attitudes among the public, which is in line with the literatures on trait activation and cognitive priming (e.g., Richey, 2012). Specifically, exposure to populist communication primes congruent us versus them cognitive clusters or schemata among receivers. As a consequence of exposure to populist communication, people’s negative stereotypes of the out-group and positive stereotypes of the in-group may be activated. Such stereotypical representations are reflected in populist attitudes that tap into the perceived opposition between the good people and the evil others (Akkerman et al., 2014). This corroborates a recent line of empirical research that has causally related exposure to populist communication to populist attitudes (e.g., Matthes & Schmuck, 2017). Selective exposure to media populism In the era of increasingly more fluid and user-controllable media diets, exposure to specific media messages cannot be taken for granted (e.g., Bennett & Iyengar, 2008; Stroud, 2008). Responding to this changing media landscape, a growing body of literature points to selective exposure as an important factor that needs to be taken into account in media effect studies (e.g., Arceneaux, Johnson, & Murphy, 2012; Iyengar & Hahn, 2009). Selective exposure can be defined as the guiding influence of people’s prior beliefs on their selection of media content (Stroud, 2008). Recent literature contends that, when offered the choice, people are most likely to expose themselves to pro-attitudinal as opposed to counter-attitudinal content. Media effects should consequentially occur most among people that actively choose to expose themselves to specific media content. The psychological underpinnings of selective exposure are rooted in theories of cognitive consistency (Iyengar & Hahn, 2009). In line with this, people are expected to select and filter novel information based on their prior beliefs (Festinger, 1957). Doing so, people pay attention to information that confirms their prior attitudes and beliefs while circumventing information that counters their priors. In line with this, people’s desire to adhere to their prior beliefs weighs more than getting it right, or making the most accurate decision based on an elaboration on all the factual information available to them (Lodge & Taber, 2013). These selection biases are psychologically explained by people’s intrinsic motivation to reduce cognitive dissonance. Although not necessarily conscious, people engage in selective processing strategies that avoid or counterargue inconsistent information (Festinger, 1957). Selective exposure can result in a polarization of beliefs (e.g., Stroud, 2008). The argumentation behind this view is as follows: as people with certain viewpoints expose themselves to congruent information, their beliefs will be reinforced and driven by the search for cognitive consistency. In addition, people become immune to attitudinal change by incongruent information. Although people may already hold a certain belief, deliberate exposure to congruent views may thus bolster their priors (e.g., Iyengar & Hahn, 2009). Aligning ourselves with this view on the polarizing potential of selective exposure, we expect that people who self-select attitudinal congruent populist content are those most persuaded by populist attribution of blame. How perceptual biases drive the effects of populist communication While some have argued that populism appeals to people that can be categorized as the “losers of modernization” (e.g., Kriesi et al., 2006) based on their social characteristics, many scholars have more recently gravitated towards an understanding that populist messages appeal most to people who perceive themselves to have lost out more than others in society. These vulnerable people should be those most attracted to populist sentiments that voice their grievances of being worse off than others (Elchardus & Spruyt, 2016). In line with this reasoning, the attitudinal filter of selective exposure to media populism can be described as relative deprivation: the perception that the out-group opposed to the people’s in-group unfairly receives economic and cultural resources at the cost of the in-group of ordinary citizens (Elchardus & Spruyt, 2012, 2016). Put differently, relative deprivation entails the perception of an unjust distribution of the society’s wealth: ordinary people “like us” never get what they deserve from society, whereas others always seem to profit (e.g., Hogg, Meehan, & Farquharson, 2010). Relative deprivation has been causally related to populism (e.g., Elchardus & Spruyt, 2012, 2016; Spruyt, Keppens, & van Droogenbroeck, 2016). As demonstrated by Elchardus & Spruyt (2016), populist perceptions are the consequence of an inherently negative perspective on the evolution of society and of the feeling of belonging to a group of people that is unfairly treated by society. Populist discourse is known to respond to a variety of grievances among society (Laclau, 1977). Populist discourse addresses these “victimized” people by claiming that their grievances are not heard by the corrupt elites. Against this backdrop, people who feel vulnerable and deprived have a higher tendency to be attracted to populist discourse, which claims to voice the grievances of the deprived people (Elchardus & Spruyt, 2016). Relative deprivation can be regarded as a salient prior belief driving people’s selection of attitudinal-congruent populist media content. H1: The more people experience feelings of relative deprivation, the more likely they are to self-select into exposure to congruent pro-attitudinal populist media content. It has been argued that populist cues that highlight the opposition between us and them are highly persuasive because they simplify complex societal and political issues into matters of who has done it, while absolving ordinary people of any blame (e.g., Mudde, 2004;,Rooduijn, 2014). The in-group can thus maintain a positive self-concept by attributing blame to others. This persuasive appeal has also been identified in studies on responsibility attributions (e.g., Hobolt & Tilley, 2014). H2: Exposure to populist content leads to stronger populist attitudes than exposure to non-populist content. We postulated that prior beliefs of relative deprivation drive the selection of populist content. In line with the premises of selective exposure theory, this implies that the media effects of populist blame attributions should occur most for those people who feel most deprived. For these people, the populist message is congruent with their priors. Motivated by a desire to avoid cognitive dissonance, their populist attitudes should be bolstered most (Festinger, 1957). H3: People exposed to attitudinal congruent populist content are persuaded most by media populism. Study 1 Method Design To investigate who selects populist content and how selective exposure and attitudinal congruence drive the effects of populist communication, we employed two experiments. The first experiment focuses on anti-elites media populism. Before exploring the role of attitudinal congruence, this experiment investigates who is most likely to select populist content. The design concerned a 2 (forced vs. selective exposure to media populism) × 3 (pro vs. counter. vs. balanced attitudinal stance) + 3 (pro vs. counter. vs. balanced self-select non-populist attitudinal stance) + control between-subjects factorial design. Because of the asymmetry in the number of cells, we allocated relatively more participants to the choice than the forced exposure conditions (Feldman, Stroud, Bimber, & Wojcieszak, 2013). Within the forced exposure conditions, participants were randomly assigned to one of the four experimental groups. In the choice conditions, participants self-selected into one of the six experimental groups or the control group. Sample The survey experiments were carried out on a diverse sample of Dutch citizens recruited by an international polling agency. Eligible participants in the company’s database (over 18 years, living in the Netherlands) were randomly selected and approached via e-mail. The samples are by and large representative of the voting population in terms of age, gender, and level of education. 759 individuals accessed the survey link. The study was completed by 562 participants (completion rate 74.0%). Their mean age was 49.29 years old (SD = 14.84). 46.7% were male and 53.3% were female. 24.5% of the participants were lower educated, 31.2% were higher educated, and 44.3% had a moderate level of education. Procedure The experiment was conducted online. After the informed consent procedure, participants completed a pre-test that included measures for demographics, moderating variables, and control variables. Participants were randomly allocated to either the forced or selective exposure media environment. In the selective exposure environment, participants were allowed to select one of the seven alternative news items. Next, they were exposed to the stimulus they selected themselves. In the forced exposure conditions, participants were randomly assigned to watch one of the three populist stimuli or the control stimulus. The selected or forced news item was visible for at least 30 seconds. This minimum exposure time was informed by the pilot study and was thoroughly pre-tested by the researchers. After 30 seconds, a “next” button appeared on the lower-left corner of the screen. After reading the news item, participants were forwarded to the post-test survey. This survey contained measures for the dependent variables and manipulation checks. The average response time was 15.64 minutes (SD = 20.28), of which 54.26 seconds were devoted to reading the stimuli (SD = 31.29). Once participants completed the survey, they were directed to the panel company environment, where they received incentive vouchers. Independent variables and stimuli In the selective exposure conditions, choice was manipulated by presenting participants with a cover story describing that people may prefer some news items above others available to them. To simulate their everyday media environment, participants were told that they should select one preferred article out of a list of alternative online articles. By clicking on a link showing the article’s headline, they were forwarded to their preferred article. In the forced exposure conditions, the cover story explained that people’s everyday media environment exists of a plethora of alternative news stories, and that they were exposed to one of these many options. The topic of the stimuli was the decreasing health care budget. In line with extant research on selective exposure, the articles’ attitudinal stance was manipulated into pro-attitudinal, counter-attitudinal, and balanced content (see Feldman et al., 2013). Pro-attitudinal messages connected the negative development to the elites, whereas counter-attitudinal messages connected it to the in-group (see stimuli in Appendix A and Table 1 for an explication of conditions in the various congruency groups). Table 1 Experimental Conditions and Attitudinal Congruence Attitudinal Congruence Stimuli  Congruent Stimulus  Incongruent Stimulus  Populist  People higher in perceived deprivation are exposed to a populist stimulus blaming the out-group  People lower in perceived deprivation are exposed to a populist stimulus blaming the out-group  Counter-Blame  People lower in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  People higher in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  Non-Populist  People higher in perceived deprivation are exposed to a stimulus connecting the situation to the elites  People lower in perceived deprivation are exposed to a populist stimulus connecting the situation to the elites  Counter–No Blame  People lower in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  People higher in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  Attitudinal Congruence Stimuli  Congruent Stimulus  Incongruent Stimulus  Populist  People higher in perceived deprivation are exposed to a populist stimulus blaming the out-group  People lower in perceived deprivation are exposed to a populist stimulus blaming the out-group  Counter-Blame  People lower in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  People higher in perceived deprivation are exposed to a populist stimulus blaming the ordinary people  Non-Populist  People higher in perceived deprivation are exposed to a stimulus connecting the situation to the elites  People lower in perceived deprivation are exposed to a populist stimulus connecting the situation to the elites  Counter–No Blame  People lower in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  People higher in perceived deprivation are exposed to a stimulus connecting the situation to ordinary people  Populist blame attribution was manipulated by emphasizing the ideational core of populism in the article’s framing of the news event: the causal and moral opposition between the good, blameless ordinary citizens and the culprit, evil elites (e.g., Mudde, 2004; Jagers & Walgrave, 2007). To do so, the populist stimuli, for example, referred to “our” own people in need and “our” tax money when referring to the blameless in-group. The elites in government were attributed blame for ordinary people’s problems by stressing how they are corrupt, self-interested, and unable and unwilling to care for ordinary people. The balanced stimuli articulated a mixed interpretation: both the Dutch people and the government were connected to the development (non-populism) or blamed for it (populism). All other factors were kept constant across conditions. The same topics and problem interpretations were displayed across conditions. We randomized the display order of all headlines in the selective exposure conditions. In that sense, the bias of incidental selection was constant across conditions. A growing body of literature has pointed to the fragmented spread of populism in the media (e.g., Engesser, Ernst, Esser, & Büchel, 2016). This implies that the different elements of populist content may not be present at the same time in one line of text. To make sure that people could intentionally self-select populist content, the Manichean construction of us against them was represented in both the texts and headlines of the populist stimuli. Manipulation checks The manipulation of populist blame attribution was successful (F[1, 226] = 71.45, p < .001). This means that participants in the elitist blame attribution conditions were significantly more likely to believe that the article attributed responsibility to the elites (M = 5.39, SD = 1.34) than participants in the other conditions (M = 3.74, SD = 1.60). Participants in the counter blame conditions were more likely to believe that ordinary people were blamed (M = 4.76, SD = 1.81) than in the other conditions (M = 3.23, SD = 1.84). The manipulation of selective exposure also succeeded (F[1, 553] = 206.21, p < .001), which indicates that participants in the selective exposure conditions were significantly more likely to perceive they had the choice to self-select an article (M = 5.49, SD = 1.45) than those in the forced exposure conditions (M = 3.53, SD = 1.72). Measures Unless explicitly described otherwise, all items were measured on 7-point scales (where 1 = completely disagree and 7 = completely agree). Populist attitudes We used confirmatory factor analyses (CFA) to estimate a two-dimensional structure of populist attitudes, consisting of anti-establishment and exclusionist populist attitudes. The two-dimensional model fitted the data well (χ2[28] = 35.80, χ2/df = 1.28, p = 0.148; RMSEA = .02, 90% CI [0.001, 0.040]; CFI = 0.99)). All standardized regression weights pointed to a satisfactory convergent validity. The model’s discriminant validity was also satisfactory, and the correlation between both factors was .58. Model fit declined substantially and significantly when the correlation between factors was constrained to one. In the first experiment, participants’ populist attitudes were measured on a four-item anti-establishment populism scale (Cronbach’s α = .82, M = 4.37, SD = 1.29; see Appendix B for items). Relative deprivation We expected that perceived relative deprivation would drive participants’ selection of populist media content (also see Elchardus & Spruyt, 2016; Spruyt et al., 2016). We measured perceived relative deprivation on a five-item seven-point scale (see Appendix B; Cronbach’s α = .92, M = 4.33, SD = 1.42). In addition, we included measures of alternative explanations, such as political cynicism (Cronbach’s α = .88, M = 4.82, SD = 1.42), governmental trust (Cronbach’s α = .92, M = 3.02, SD = 1.44), and issue agreement (Cronbach’s α = .64, M = 5.42, SD = 1.21). We assessed whether relative deprivation and populist attitudes tapped into different underlying constructs. The CFA indicated that model fit declined significantly when the correlation between the two constructs was constrained to the value 1 (∆χ2 [1] = 4.83, p = .020). The correlation between both factors was reasonably high (r = .74). Still, based on the assessment of discriminant validity, the correlation was too weak to merge both factors. Populist attitudes and perceived relative deprivation should therefore not be considered as the same construct. We constructed a variable that indicates whether the stimulus chosen by or shown to the participant was congruent with his or her prior perceptions of relative deprivation. For someone lower in relative deprivation, for example, counter attitudinal messages were coded as congruent with this participant’s priors (Table 1). Controls The design of the forced exposure conditions assured a random assignment to treatment and control conditions, assignment thus not resulted in differential attrition (Mutz & Pemantle, 2011). A between-conditions randomization check on the variables of gender, age, education, ideological extremity, and news exposure1 yielded non-significant differences across experimental conditions. Pilot study The manipulations and stimuli were extensively pre-tested in two pilot studies: one pilot study among a convenience sample of university students (N = 52) and one among a diverse sample of Dutch citizens (N = 137). Participants rated the stimuli as credible (M = 4.62, SD = 1.65) and similar to everyday media content (M = 5.09, SD = 1.31). All manipulations succeeded. Participants found it more likely that the article was published on an online news website (M = 5.56, SD = 1.74) than any of the other outlets. For this reason, we framed the articles as online news stories in both studies. Results of Study 1 Drivers of selective exposure to media populism To assess the likelihood of self-selection of populist media content, we compared the six different experimental conditions to the control group (see Table 2). The results of the logistic regression analyses indicate that those participants with stronger perceptions of relative deprivation were more likely to self-select into populist media content than participants with weaker perceptions of relative deprivation. The other variables do not play a significant role in driving selection into populist content. Hypothesis 1 can thus be supported: the more people experience feelings of relative deprivation, the more they are inclined to self-select populist media content. Table 2 Binary Logistic Regression Model Predicting Drivers of Selective Exposure to Anti-Elites Populist and Non-Populist Content Conditions  Populist  Non-Populist  Counter-Blame  Counter–No Blame  Variables  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  (constant)  −4.14 (1.36)**    −2.35 (1.27)    −1.12 (1.69)    −1.15 (1.34)    Age  0.01 (0.01)  [0.98, 1.03]  0.01 (0.10)  [0.99, 1.03]  −0.01 (0.01)  [0.97, 1.01]  −0.01 (.01)  [0.97, 1.01]  Gender  0.18 (0.30)  [0.66, 2.15]  0.52 (0.28)  [0.98, 2.91]  −0.33 (0.41)  [0.32, 1.62]  −0.36 (0.32)  [0.37, 1.30]  Education (higher)  −0.73 (0.39)  [0.23, 1.03]  0.03 (0.30)  [0.57, 1.86]  0.07 (0.45)  [0.45, 2.60]  0.34 (0.33)  [0.73, 2.71]  Deprivation  0.61 (0.15)***  [1.36, 2.49]  −0.09 (0.13)  [0.72, 1.17]  0.26 (0.20)  [0.88, 1.91]  −0.03 (0.15)  [0.73, 1.30]  Issue Agreement  0.05 (0.16)  [0.77, 1.42]  0.07 (0.14)  [0.82, 1.41]  −0.14 (0.20)  [0.59, 1.28]  −0.01 (0.16)  [0.73, 1.35]  Governmental trust  −0.06 (0.12)  [0.75, 1.19]  −0.03 (0.12)  [0.77, 1.23]  0.07 (0.16)  [0.78, 1.48]  0.10 (0.13)  [0.86, 1.42]  Political cynicism  −0.12 (0.15)  [0.67, 1.19]  −0.02 (0.13)  [0.76, 1.28]  −0.25 (0.19)  [0.53, 1.13]  0.02 (0.16)  [0.75, 1.39]  Nagelkerke R2  0.17  0.03  0.04  0.03  χ2 (df)  37.90 (7)***  5.65 (7)  5.32 (7)  5.29 (7)  Conditions  Populist  Non-Populist  Counter-Blame  Counter–No Blame  Variables  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  B (SE)  95% CI OR  (constant)  −4.14 (1.36)**    −2.35 (1.27)    −1.12 (1.69)    −1.15 (1.34)    Age  0.01 (0.01)  [0.98, 1.03]  0.01 (0.10)  [0.99, 1.03]  −0.01 (0.01)  [0.97, 1.01]  −0.01 (.01)  [0.97, 1.01]  Gender  0.18 (0.30)  [0.66, 2.15]  0.52 (0.28)  [0.98, 2.91]  −0.33 (0.41)  [0.32, 1.62]  −0.36 (0.32)  [0.37, 1.30]  Education (higher)  −0.73 (0.39)  [0.23, 1.03]  0.03 (0.30)  [0.57, 1.86]  0.07 (0.45)  [0.45, 2.60]  0.34 (0.33)  [0.73, 2.71]  Deprivation  0.61 (0.15)***  [1.36, 2.49]  −0.09 (0.13)  [0.72, 1.17]  0.26 (0.20)  [0.88, 1.91]  −0.03 (0.15)  [0.73, 1.30]  Issue Agreement  0.05 (0.16)  [0.77, 1.42]  0.07 (0.14)  [0.82, 1.41]  −0.14 (0.20)  [0.59, 1.28]  −0.01 (0.16)  [0.73, 1.35]  Governmental trust  −0.06 (0.12)  [0.75, 1.19]  −0.03 (0.12)  [0.77, 1.23]  0.07 (0.16)  [0.78, 1.48]  0.10 (0.13)  [0.86, 1.42]  Political cynicism  −0.12 (0.15)  [0.67, 1.19]  −0.02 (0.13)  [0.76, 1.28]  −0.25 (0.19)  [0.53, 1.13]  0.02 (0.16)  [0.75, 1.39]  Nagelkerke R2  0.17  0.03  0.04  0.03  χ2 (df)  37.90 (7)***  5.65 (7)  5.32 (7)  5.29 (7)  **p < .01; ***p < .001. Note. N = 346. CI = confidence interval; OR = odds ratio. Two-tailed tests. Unstandardized regression weights. Standard errors reported between brackets. The reference category for the reported conditions is the control category. Freedom of selection In a second step, we explored the relationship between the freedom of selection and participants’ populist attitudes. A one-way ANOVA comparing treatment and control conditions between forced and selective exposure environments indicated that the selective exposure control condition differed significantly from all other conditions (F[3, 554] = 8.92, p < .001, partial η2 = .05). In this condition, populist attitudes are relatively low (M = 3.59, SD = 1.06). For all other conditions, the mean scores of populist attitudes are consistent and significantly and substantially higher (M = 4.51, SD = 1.28). This can be interpreted as evidence that those participants who self-selected a control condition have significantly lower populist attitudes than those who self-selected treatment conditions or those who were forced into treatment or control conditions (also see Table 3). Table 3 Effects of Populist Blame Attribution to Elites on Anti-Establishment Populist Attitudes   Model I (n = 558)  Model II (n = 558)  Model III (n = 558)  Model IV (n = 558)  Model V (n = 558)  B  SE  β  B  SE  β  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.37  .12    4.65  .16    4.65  .16    4.65  .15    4.65  .15    Selective Exposure  −0.22  .12  −.04  −0.42  .19  −.16*  −0.52  .19  −.19**  −0.38  .19  −.14*  −0.38  .19  −.14*  Populist Message  0.34  .15  .11*  −0.36  .24  −.11  −0.44  .25  −.14  −0.72  .24  −.23**  −0.79  .25  −.25**  Balanced Message  0.18  .17  .05  −0.22  .24  −.06  −0.22  .24  −.06  −0.22  .23  −.06  −0.22  .23  −.06  Counter-Blame Message  −0.04  .17  −.01  −0.11  .23  −.03  −0.11  .23  −.03  −0.10  .23  −.02  −0.10  .23  −.03  Selective Control  −0.76  .19  −.19***  −0.65  .19  −.16**  −0.55  .20  −.12*  −0.69  .19  −.16**  −0.69  .19  −.16**  Populist × Selection        1.12  .30  .29***  1.13  .30  .29***  0.69  .31  .15  0.86  .36  .18  Balanced × Selection        0.65  .34  .11  0.76  .34  .14*  0. 62  .34  .11  0.62  .34  .11  Counter × Selection        −0.05  .35  −.01  −0.17  .35  −.03  0.67  .48  .11  0.89  .48  .13  Attitudinal Congruence              0.33  .15  .14*  −0.11  .20  −.04  −0.12  .20  −.04  Pop × Congruent                    1.47  .31  .33***  1.76  .39  .40***  Counter-Blame × Congruent                    −1.42  .53  −.19*  −1.44  .56  −.19*  Pop × Congruent × Selection                          −0.49  .48  −.10  Counter × Congruent × Selection                          0.18  .32  .14  Adjusted R2  0.05  0.07  0.08  0.13  0.13  F  6.34***  6.25***  6.11**  8.76***  7.51***  F for Change in R2    5.82**  4.69*  18.89***  0.69    Model I (n = 558)  Model II (n = 558)  Model III (n = 558)  Model IV (n = 558)  Model V (n = 558)  B  SE  β  B  SE  β  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.37  .12    4.65  .16    4.65  .16    4.65  .15    4.65  .15    Selective Exposure  −0.22  .12  −.04  −0.42  .19  −.16*  −0.52  .19  −.19**  −0.38  .19  −.14*  −0.38  .19  −.14*  Populist Message  0.34  .15  .11*  −0.36  .24  −.11  −0.44  .25  −.14  −0.72  .24  −.23**  −0.79  .25  −.25**  Balanced Message  0.18  .17  .05  −0.22  .24  −.06  −0.22  .24  −.06  −0.22  .23  −.06  −0.22  .23  −.06  Counter-Blame Message  −0.04  .17  −.01  −0.11  .23  −.03  −0.11  .23  −.03  −0.10  .23  −.02  −0.10  .23  −.03  Selective Control  −0.76  .19  −.19***  −0.65  .19  −.16**  −0.55  .20  −.12*  −0.69  .19  −.16**  −0.69  .19  −.16**  Populist × Selection        1.12  .30  .29***  1.13  .30  .29***  0.69  .31  .15  0.86  .36  .18  Balanced × Selection        0.65  .34  .11  0.76  .34  .14*  0. 62  .34  .11  0.62  .34  .11  Counter × Selection        −0.05  .35  −.01  −0.17  .35  −.03  0.67  .48  .11  0.89  .48  .13  Attitudinal Congruence              0.33  .15  .14*  −0.11  .20  −.04  −0.12  .20  −.04  Pop × Congruent                    1.47  .31  .33***  1.76  .39  .40***  Counter-Blame × Congruent                    −1.42  .53  −.19*  −1.44  .56  −.19*  Pop × Congruent × Selection                          −0.49  .48  −.10  Counter × Congruent × Selection                          0.18  .32  .14  Adjusted R2  0.05  0.07  0.08  0.13  0.13  F  6.34***  6.25***  6.11**  8.76***  7.51***  F for Change in R2    5.82**  4.69*  18.89***  0.69  *p < .05; **p < .01; ***p < .001 Note. The reference category is the forced control condition. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. The effect of populist cues To explore the impact of populist messages, we compared forced and selective exposure to populist, balanced, and counter-blame messages to the forced exposure control condition (Table 3). The results show that participants in the counter-blame and balanced treatment conditions did not have significantly higher populist attitudes than those in the control condition (Model I). Yet, exposure to a populist message has a positive and significant effect on anti-establishment populist attitudes. This provides support for H2. The role of attitudinal congruence The two-way interaction effect of selective exposure and populist blame attribution (Table 3, Model II) illustrates that populist messages had the strongest effect on populist attitudes when they were self-selected. However, once attitudinal congruence is included in Model IV, the interaction effect of self-selection and exposure to a populist message is no longer significant. Specifically, as indicated by the significant and positive two-way interaction effect of exposure to a populist message and congruence (Table 3, Model IV), participants exposed to congruent populist content had significantly higher anti-establishment populist attitudes than those exposed to incongruent populist messages. Selective exposure did not further specify this relationship (Table 3, Model V), which indicates that congruent populist messages do not have a different effect when self-selected. In support of hypothesis 3, participants exposed to attitudinally congruent elitist blame attributions are persuaded most by populist messages. Study 2 Our first study revealed that the effects of exposure to populist media content are driven by attitudinal congruence: only those participants exposed to messages congruent with their priors are persuaded by populist attributions of blame. By explicitly taking attitudinal congruence into account in the design of the second experiment, we are able to more precisely test how attitudinal biases drive the effects of populist communication. Method Design The design of the second experiment concerns a 2 (attitudinal stance: pro-attitudinal vs. counter-attitudinal) × 2 (choice: populist vs. non-populist framing) + 3 (Forced exposure: populism vs. counter-no-blame vs. control) between-subjects factorial design (see Table 1 for conditions). Participants were divided into groups that viewed congruent versus incongruent framings of the issue immigration based on their pre-exposure scores on a measure of perceived relative deprivation connected to the out-group threat depicted in the stimuli. Moderates were randomly assigned to congruent versus incongruent stimuli. In line with Arceneaux et al. (2012), we controlled for them in the analyses. Sample The experiment was completed by 558 participants (completion rate 72.4%) with a mean age of 49.15 years (SD = 16.09). 47.5% were male and 52.5% were female. 23.7% of the participants were lower educated, 32.8% were higher educated, and 43.5% had a moderate level of education. Procedure The procedure regarding the pre-test was similar to Study 1. After the pre-test, participants completed the following item on a 7-point scale before being randomly allocated to attitudinal incongruent versus incongruent conditions: “People who are not originally from the Netherlands, are profiting more from all kinds of benefits in society than Dutch citizens” (M = 4.58, SD = 1.90). The sample was divided into pro-attitudinal publics (scores ranging from 5 through 7); counter-attitudinal publics (scores ranging from 1 through 3) and moderates (scoring the midpoint on the scale). All these groups were randomly assigned to the pro- and counter-attitudinal forced and choice conditions. Independent variables and stimuli The second experiment focused on increasing crime rates in the light of immigration, a salient monocultural, right-wing populist issue. Selective exposure was manipulated by presenting participants with a cover story describing how people may prefer some news items above others available to them. Participants were told that they should select one preferred article out of two alternative online articles. One article was framed in a populist way, whereas the other article was framed in a non-populist way (see Table 1 and examples in Appendix A and Supplemental Material). In the forced exposure conditions, participants were randomly assigned to either populist or non-populist stimuli. Informed by the results of Study 1, attitudinal stance was manipulated into a pro-attitudinal versus a counter-attitudinal framing of the issue, either highlighting that migrants or Dutch people were associated with the increasing crime rate. Based on their scores on the relative deprivation measure, we assigned participants to congruent and incongruent pro- and counter-attitudinal conditions at a 1:1 rate. Manipulation checks The manipulation of populist blame attribution to migrants was successful (F[1, 553] = 157.25, p < .001). Participants in the blame attribution conditions were significantly more likely to believe that the article attributed responsibility to migrants (M = 5.68, SD = 1.23) than participants in the other conditions (M = 3.71, SD = 1.77). The manipulation of selective exposure also succeeded (F[1, 553] = 87.44, p < .001). Participants in the selective exposure conditions were significantly more likely to perceive they were offered the choice to self-select an article (M = 4.81, SD = 1.58) than participants in the forced exposure conditions (M = 3.50, SD = 1.72). Measures We used the same measures as reported in Study 1. Again, the CFA for populist attitudes demonstrated good model fit (χ2[27] = 43.87, χ2/df = 1.63, p = .021; RMSEA = .03, 90% CI [.01, .05]; CFI = .99). The correlation between factors was .69 and model fit declined substantially and significantly when the correlation between factors was constrained to one. Participants’ populist attitudes were measured on a 7-point exclusionist populist attitudes scale (Cronbach’s α = .94, M = 4.14, SD = 1.71). The correlation between populist attitudes and perceived relative deprivation was (r = . 56). The model fit for the CFA with one factor, including items for populist attitudes and perceived deprivation, was significantly and substantially worse than a model that treats populist attitudes and relative deprivation as separate scales (∆χ2 [1] = 19.01, p < .01). Results of Study 2 First, we compared mean scores on exclusionist populist attitudes between the control, selective treatment, and forced treatment conditions. The one-way ANOVA yielded non-significant differences in populist attitudes between conditions (F[2, 555] = 1.89, p = .151, partial η2 = .001). This indicates that, ceteris paribus, exclusionist populist attitudes were similar across the control and both the forced and selective treatment conditions. Selective exposure to attitudinal congruent populist versus non populist cues As illustrated in Table 4, participants who self-selected into congruent populist blame attributions scored higher on exclusionist populist attitudes (M = 5.64, SD = 1.02) than participants who selected non-populist content (M = 4.85, SD = 1.13). In support of hypothesis 2, these results indicate that people exposed to populist cues have higher populist attitudes than people exposed to non-populist cues. Table 4 The Effects of Populist and Non-Populist Stimuli on Anti-Establishment Populist Attitudes at Different Levels of Attitudinal Congruence (Monocultural Media Populism)   Lower in Relative Deprivation  Higher in Relative Deprivation  F  Partial η2  Congruent  Incongruent  Congruent  Incongruent      Counter-Blame  Counter–No Blame  Populist  Non-Populist  Populist  Non-Populist  Counter-Blame  Counter–No Blame      Choice  2.74a, x (1.48)  2.35a (1.14)  2.35a, x (1.19)  2.39a (1.22)  5.64b, x (1.02)  4.85c (1.13)  5.27b, x (1.49)  5.10c (1.13)  24.04*  .38  Forced Exposure  2.12a, y (1.13)  -  2.82b, y (1.29)  -  5.09c, y (1.50)  -  5.04c, x (1.37)  -  25.78*  .28  Control  2.27a, (0.97)  5.09b (1.47)        Lower in Relative Deprivation  Higher in Relative Deprivation  F  Partial η2  Congruent  Incongruent  Congruent  Incongruent      Counter-Blame  Counter–No Blame  Populist  Non-Populist  Populist  Non-Populist  Counter-Blame  Counter–No Blame      Choice  2.74a, x (1.48)  2.35a (1.14)  2.35a, x (1.19)  2.39a (1.22)  5.64b, x (1.02)  4.85c (1.13)  5.27b, x (1.49)  5.10c (1.13)  24.04*  .38  Forced Exposure  2.12a, y (1.13)  -  2.82b, y (1.29)  -  5.09c, y (1.50)  -  5.04c, x (1.37)  -  25.78*  .28  Control  2.27a, (0.97)  5.09b (1.47)      * = p < .001 Note. N = 555. Standard deviations are reported in parentheses below the means. Means with differing first subscripts within rows and differing second subscripts within columns differ significantly at the p < .05 level based on post hoc independent samples t-test. From top to the bottom, the F-values represent: (1) the main effect of the experimental conditions in the selective exposure conditions; (2) the main effect of the experimental conditions within forced exposure. Attitudinal congruence and monocultural media populism In Table 5, we see a significant and positive two-way interaction effect of the populist stimuli and attitudinal congruence. This indicates that, in the forced exposure conditions, exclusionist populist attitudes are significantly higher among participants exposed to congruent than incongruent populist blame attribution. The results further indicate a significant and negative two-way interaction effect of exposure to counter-blame stimuli and attitudinal congruence. Participants at lower levels of relative deprivation exposed to a message that blames the native people for the increasing crime rate have lower exclusionist populist attitudes compared to participants who are exposed to incongruent populist messages (see Table 4). Table 5 Effects of Populist Blame Attribution to Migrants on Exclusionist Populist Attitudes Within Forced Exposure Conditions   Model I (n = 277)  Model II (n = 277)  Model III (n = 277)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.13  .12    4.20  .15    4.08  .15    Populist  0.34  .23  .08  0.38  .24  .09  −1.26  .36  −.28*  Counter-Blame  0.14  .22  .03  0.21  .29  .05  1.25  .26  .30**  Moderates        −0.80  .56  −.08  −1.73  .51  −.16*  Attitudinal Congruent        −0.14  .19  −.04  0.10  .22  .03  Populist × Congruent              2.18  .45  .43**  Counter-Blame × Congruent              −3.31  .44  −.45**  Adjusted R2  .01  .01  .21  F  1.13  1.17  18.14**  F for Change in R2    1.20  51.48**    Model I (n = 277)  Model II (n = 277)  Model III (n = 277)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.13  .12    4.20  .15    4.08  .15    Populist  0.34  .23  .08  0.38  .24  .09  −1.26  .36  −.28*  Counter-Blame  0.14  .22  .03  0.21  .29  .05  1.25  .26  .30**  Moderates        −0.80  .56  −.08  −1.73  .51  −.16*  Attitudinal Congruent        −0.14  .19  −.04  0.10  .22  .03  Populist × Congruent              2.18  .45  .43**  Counter-Blame × Congruent              −3.31  .44  −.45**  Adjusted R2  .01  .01  .21  F  1.13  1.17  18.14**  F for Change in R2    1.20  51.48**  *p < .01; **p < .001 Note. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. As can be seen in Table 6, the two-way interaction effect of populist blame attribution to migrants and attitudinal congruence is also positive and significant in the selective exposure conditions. This means that when participants higher in relative deprivation are exposed to populist attributions of blame to migrants, their exclusionist populist attitudes are stronger than in the other conditions. Table 6 Effects of Populist Blame Attribution to Migrants on Exclusionist Populist Attitudes Within Selective Exposure Conditions   Model I (n = 281)  Model II (n = 281)  Model III (n = 281)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.38  .13    4.52  .17    4.82  .16    Populist  0.39  .30  .07  0.38  .23  .07  −2.47  .47  −.39**  Counter-Blame  −0.22  .33  −.04  −0.29  .33  −.05  0.69  .41  .11  Non-Populist  −0.75  .26  −.15*  −0.81  .25  −.16*  −0.81  .25  −.16*  Counter–No Blame  −0.20  .24  −.04  −0.29  .25  −.06  −0.31  .23  −.07  Moderates        −0.91  .54  −.09  −0.97  .51  −.9  Attitudinal Congruent        −0.19  .19  −.05  −0.32  .19  −.09  Populist × Congruent              4.06  .56  .62***  Counter-Blame × Congruent              −2.22  .59  −.24**  Non-Populist × Congruent              2.23  .44  .38**  Counter–No Blame × Congruent              −1.94  .43  −.26**  Adjusted R2  .02  .02  .28  F  3.11  2.63*  18.53**  F for Change in R2    1.65  48.43**    Model I (n = 281)  Model II (n = 281)  Model III (n = 281)  B  SE  β  B  SE  β  B  SE  β  (Constant)  4.38  .13    4.52  .17    4.82  .16    Populist  0.39  .30  .07  0.38  .23  .07  −2.47  .47  −.39**  Counter-Blame  −0.22  .33  −.04  −0.29  .33  −.05  0.69  .41  .11  Non-Populist  −0.75  .26  −.15*  −0.81  .25  −.16*  −0.81  .25  −.16*  Counter–No Blame  −0.20  .24  −.04  −0.29  .25  −.06  −0.31  .23  −.07  Moderates        −0.91  .54  −.09  −0.97  .51  −.9  Attitudinal Congruent        −0.19  .19  −.05  −0.32  .19  −.09  Populist × Congruent              4.06  .56  .62***  Counter-Blame × Congruent              −2.22  .59  −.24**  Non-Populist × Congruent              2.23  .44  .38**  Counter–No Blame × Congruent              −1.94  .43  −.26**  Adjusted R2  .02  .02  .28  F  3.11  2.63*  18.53**  F for Change in R2    1.65  48.43**  *p < .01; **p < .001 Note. Two-tailed tests. Unstandardized (B) and standardized (β) regression weights. In the selective exposure conditions, the analyses further point to a significant, negative two-way interaction effect of the selection of counter-blame content and attitudinal congruence (see Table 6). This indicates that people who self-select into counter-blame content that fits their prior attitudes of lower relative deprivation have weaker populist attitudes compared to participants that self-select into incongruent counter-blame (see Table 4 for mean score differences). In support of H3, participants exposed to attitudinal congruent monocultural media populism have stronger exclusionist populist attitudes than participants exposed to incongruent populist communication (see Figure 1). Figure 1 View largeDownload slide Marginal effect plots for interaction attitudinal congruence and populist blame attribution in forced and selective exposure conditions. Note. Dots represent regression weights and lines represent 95% confidence intervals. N = 558. Figure 1 View largeDownload slide Marginal effect plots for interaction attitudinal congruence and populist blame attribution in forced and selective exposure conditions. Note. Dots represent regression weights and lines represent 95% confidence intervals. N = 558. Discussion In the midst of the rise of persuasive populist ideas throughout the globe, a growing body of literature has pointed to the key role of the media in disseminating the populist message to the people (Krämer, 2014; Mazzoleni, 2008; Mudde, 2004). This literature has also assumed that those people who selectively expose themselves to populist messages form a specific type of audience. Specifically, those attracted to populist communication are described as a discontented (Mazzoleni, 2008) and relatively deprived group of citizens (Elchardus & Spruyt, 2016). Yet, at the same time, the scarce body of literature that has studied the effects of populist communication has done so in a forced-exposure environment (e.g., Bos et al., 2013). Therefore, an important discrepancy in populism research is that, despite the assumptions about the peculiarities of the audience susceptible to persuasion by populist communication, there are no studies that actually simulate a selective exposure media environment. Advancing this line of research, this paper presents two experiments that do take selective exposure and attitudinal congruence into account. In line with recent literature showing that citizens who feel relatively deprived are more likely to hold populist perceptions (Elchardus & Spruyt, 2016), our results show that these citizens are also more likely to be drawn to populist content. Addressing the ordinary people as a blameless, relatively deprived in-group juxtaposed to a causally responsible enemy makes populist discourse stick. In the midst of an increasingly more fragmentized media environment (e.g., Stroud, 2008), the findings of this paper demonstrate that people who self-select into populist communication differ substantially from those who circumvent such content. In line with recent literature, it is indeed the citizens who feel relatively deprived that expose themselves to populist messages (Elchardus & Spruyt, 2016). Citizens with weaker perceptions of deprivation avoid populist content. It was, however, not selective exposure but attitudinal congruence that conditioned the effects of exposure to media populism on populist attitudes. In both forced and selective exposure media environments, populist blame attributions only affected the populist attitudes of people for whom the message was congruent with their priors. Based on the evidence from the experiments presented in this paper, the role of giving people the freedom to self-select into media exposure is outweighed by exposure to a message that is in line with their prior attitudes. Hence, our results raise the question of whether future research should take the effort to methodologically simulate a selective exposure environment if dividing the sample into congruent and incongruent issue publics has the same effect on the outcomes. This methodological note does not mean that selective exposure does not play a role in conditioning the effects of media populism on the populist perceptions of the electorate. Backed up by the empirical evidence from our first experiment, exposure to attitudinal congruent populist media content is most likely to occur among citizens higher in relative deprivation. For these citizens, media populism confirms their priors of being deprived by some culprit other opposed to ordinary people. This finding is in line with the theory on motivated reasoning (Lodge & Taber, 2013). Outside of the experimental setting, then, it is unlikely that people for whom the populist message is counter to their priors actually expose themselves to media populism. This key finding has important implications for the polarizing effect of exposure to attitudinal congruent populist content. People who avoid populist content because it counters their priors are least likely to be affected by populist communication. When these people are forced into exposure to populist content, the message can even have a backlash. In that case, their negative priors towards populist interpretations are activated by the incongruent populist message. Those citizens that are higher in relative deprivation already have higher populist attitudes. Yet exposure to congruent populist content activates their priors, resulting in a bolstering of their existing populist attitudes on both the anti-establishment and exclusionist dimension. We interpret this as evidence for the polarizing potential of media populism in a fragmentized media environment (Stroud, 2008). Indeed, exposure to messages that articulate the causal and moral divide between us and them actually widens gaps in society, pitting those with populist interpretations against those who feel distant from such worldviews. The findings of this paper indicate that selective exposure and attitudinal congruent persuasion have important consequences for media effect studies beyond the U.S. context of partisanship and (affective) polarization (e.g., Iyengar & Hahn, 2009). Specifically, to better understand how media content affects public perceptions of salient socio-political issues—such as populism—selection should be taken into account (also see Bennett & Iyengar, 2008). In the current era of choice and fragmentation, it is crucial to determine who avoids or selects political communication biased toward a certain interpretation of reality. Moving beyond selection as dependent variable, our findings indicate that persuasion is contingent upon attitudinal congruence, and that self-selection does not further bolster the effectiveness of polarized media content. Advancing the literatures on selective exposure and motivated reasoning, these findings indicate that people are affected most when political communication confirms their priors, and that the freedom to select reassuring content does not necessarily makes it more persuasive. The underlying information processing mechanism can be understood as a two-step model of selection and attitudinal congruent persuasion. First, people select media content that reassures their existing interpretations of reality. Second, only those with congruent priors are positively affected by such content, making them more aligned with populist worldviews. People who disagree with such worldviews are not likely to select populist content, and thus not likely to be persuaded as long as they can avoid such content. Indeed, people who are confronted with incongruent content may defend their prior attitudes to counter-argue attacks on their existing beliefs. Because avoidance has crucial ramification for media effects, an important avenue for selective exposure research is to assess the extent to which the audience is actually able to avoid certain political content in their daily media environment. These findings have important practical implications. Because perceived relative deprivation is the main driver of attitudinal congruent persuasion, future interventions should focus on a correction of these deprived beliefs among these citizens. Selective exposure plays a key role in people’s everyday media environment. If populist messages are difficult to avoid in daily life, and if these messages counter people’s priors, populist communication may have a backlash, reinforcing citizens’ already negative prior attitudes toward populist interpretations. In that sense, it may be more important to base interventions on lowering perceived deprivation than exposing people to different content. Specifically, political campaigns should explicitly relativize causes and consequences of people’s deprivation, for example by showing people that they are actually not worse off than other groups in society and that refugees are facing more challenges than native people. If external causes for the people’s crisis forwarded by populist parties—such as “profiting” migrants—can be effectively refuted in political campaigns, the electorates’ perceptions of relative deprivation can be lowered further. Here, it is important that misinformation on mainstream political parties’ and immigrants’ involvement in causing various socio-political problems is effectively discredited. Current evidence regarding the effectiveness of corrective attempts is, however, still mixed (see for example Thorson, 2016). Discrediting fake news on blame and deprivation could also be regarded as a task for the media, such as via fact checker websites. By indicating which attributions of blame are actually correct or not, the electorate will be better able to judge who is responsible for their experienced problems. Our studies bear some limitations that may be dealt with in future research. First and foremost, we could not randomly assign participants to the selective exposure conditions, which limits the causal conclusions that can be drawn from the choice conditions. Against this backdrop, we have to acknowledge that choice experiments result in more realistic treatment effects at the cost of introducing a self-selection bias because randomization is not achieved. Improving external validity thus comes at the cost of internal validity (e.g., Gaines & Kuklinski, 2011). However, as we accounted for the most important factor that drives selective exposure to media populism—relative deprivation—in addition to a post hoc randomization check, we do believe that we have at least partially dealt with this concern. An important consideration in the experimental design of selective exposure studies concerns the number of options offered to participants (see Feldman et al., 2013, for a comprehensive overview). To more closely simulate a real-world high-choice media environment, future research may include different options in the selective exposure environment relevant to the effects of media populism, such as offering a choice between tabloid and broadsheet media outlets (e.g., Mazzoleni, 2008). This may also allow for a further exploration of why people choose populist content in a fragmented media environment with a plethora of alternatives. Another limitation concerns the potential of ceiling effects found after exposure to populist messages among the public. Independent of the stimuli, people with similar perceptions of relative deprivation have stronger populist attitudes (in line with Hameleers et al., 2017 and Elchardus & Spruyt, 2016). On the measured scales, there was not much room for the stimuli to further boost their populist attitudes. Populist attitudes and relative deprivation can also be interpreted from an alternative, reciprocal causal framework. People may select populist content because they feel relatively deprived (Elchardus & Spruyt, 2016). Populist content hereby responds to people’s perceived deprivation, making people more aligned with an antagonistic, populist interpretation of society. In a second step of this mechanism, people may feel even more deprived as a result of like-minded populist media framing. As a consequence of this, they may seek out even more populist content. It may be an interesting avenue for future research to tease out such a reinforcing spiral of populism and deprivation. Future research could, for example, rely on pre-and post-exposure measures of relative deprivation and populist attitudes to more precisely investigate how these perceptions were primed differentially. Despite these limitations, the studies presented in this paper are the first to demonstrate how the effects of media populism are driven by attitudinal congruence. In showing that populist communication is only selected by and persuasive for those higher in relative deprivation, this paper provides an important contribution to the literature that has argued that those citizens that populist messages appeal to form a specific type of audience that can be distinguished from those who circumvent populist coverage. Hence, in the midst of the revival of the populist zeitgeist, this paper demonstrates which part of the electorate is most likely to expose itself to populist messages to further bolster their priors, with eventually paramount societal and democratic implications in terms of a polarization between the populist and non-populist electorate. Supporting Information Supporting information is available at Journal of Communication online. Note 1 A series of chi-squared tests revealed no significant differences across conditions for age (forced: χ2 [6] = 3.32, p = n. s.; selective: χ2 [2] = 2.41, p = n. s), gender (forced: χ2 [3] = 8.62, p = n. s.; selective: χ2 [1] = .35, p = n. s.;), education (forced: χ2 [8] = 8.73, p = n. s.; selective: χ2 [2] = .19, p = n. s.), ideological extremity (forced: χ2 [3] = 1.22, p = n. s; selective: χ2 [1] = .07, p = n. s), and news exposure (forced χ2 [24] = 31.22, p = n. s.; selective: χ2 [7] = 5.02, p = n.s.) References Akkerman, A., Mudde, C., & Zaslove, A. ( 2014). How populist are the people? Measuring populist attitudes in voters. Comparative Political Studies , 47( 9), 1324– 1353. doi:10.1177/0010414013512600 Google Scholar CrossRef Search ADS   Arceneaux, K., Johnson, M., & Murphy, C. ( 2012). Polarized political communication, oppositional media hostility, and selective exposure. The Journal of Politics , 74( 1), 174– 186. doi:S002238161100123X. Google Scholar CrossRef Search ADS   Bennett, W. L., & Iyengar, S. ( 2008). A new era of minimal effects? The changing foundations of political communication. Journal of Communication , 58( 4), 707– 731. doi:10.1111/j.1460-2466.2008.00410.x Google Scholar CrossRef Search ADS   Bos, L., van der Brug, W., & de Vreese, C. H. ( 2013). An experimental test of the impact of style and rhetoric on the perception of right-wing populist and mainstream party leaders. Acta Politica , 48( 2), 192– 208. doi:10.1057/ap.2012.27 Google Scholar CrossRef Search ADS   Caiani, M., & della Porta, D. ( 2011). The elitist populism of the extreme right: A frame analysis of extreme right-wing discourses in Italy and Germany. Acta Politica , 46( 2), 180– 202. doi:10.1057/ap.2010.28 Google Scholar CrossRef Search ADS   Canovan, M. ( 1999). Trust the people! Populism and the two faces of democracy. Political Studies , 47, 2– 16. doi:10.1111/1467-9248.00184 Google Scholar CrossRef Search ADS   Dixon, T. L. ( 2008). Crime news and racialized beliefs: Understanding the relationship between local news viewing and perceptions of African Americans and Crime. Journal of Communication , 58( 1), 106– 125. doi:10.1111/j.1460-2466.2007.00376.x Google Scholar CrossRef Search ADS   Elchardus, M., & Spruyt, B. ( 2012). The contemporary contradictions of egalitarianism: An empirical analysis of the relationship between the old and new left/right alignments. European Political Science Review , 4( 2), 217– 239. doi:1.1017/S1755773911000178 Google Scholar CrossRef Search ADS   Elchardus, M., & Spruyt, B. ( 2016). Populism, persistent republicanism and declinism: An empirical analysis of populism as a thin ideology. Government and Opposition , 51( 1), 111– 133. doi:10.1017/gov.2014.27 Google Scholar CrossRef Search ADS   Engesser, S., Ernst, N., Esser, F., & Büchel, F. ( 2016). Populism and social media: How politicians spread a fragmented ideology. Information, Communication & Society , 20( 8), 1109– 1126. doi:10.1080/1369118X.2016.1207697 Google Scholar CrossRef Search ADS   Entman, R. M. ( 1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication , 43( 4), 51– 59. doi:10.1111/j.1460-2466.1993 Google Scholar CrossRef Search ADS   Feldman, L., Stroud, N. J., Bimber, J., & Wojcieszak, M. ( 2013). Assessing selective exposure in experiments: The implications of different methodological choices. Communication Methods and Measures , 7( 3-4), 172– 194. doi:10.1080/19312458.2013.813923 Google Scholar CrossRef Search ADS   Festinger, L. ( 1957). A theory of cognitive dissonance . Evanston, IL: Row, Peterson & Company. Gaines, B. J., & Kuklinski, J. ( 2011). Experimental estimation of heterogeneous treatment effects for treatments related to self-selection. American Journal of Political Science , 55( 3), 724– 736. doi:10.1111/j.1540-5907.2011.00518.x Google Scholar CrossRef Search ADS   Hameleers, M., Bos, L., & de Vreese, C. H. ( 2017). “They did it”: The effects of emotionalized blame attribution in populist communication. Communication Research , 44( 6), 870– 900. doi:10.1177/0093650216644026 Google Scholar CrossRef Search ADS   Hawkins, K. A. ( 2009). Is Chávez populist? Measuring populist discourse in comparative perspective. Comparative Political Studies , 42( 8), 1040– 1067. doi:10.1177/0010414009331721 Google Scholar CrossRef Search ADS   Hewstone, M. ( 1989). Causal attribution: From cognitive processes to collective beliefs . Oxford, UK: Blackwell. Hobolt, S. B., & Tilley, J. ( 2014). Blaming Europe? Responsibility without accountability in the European Union . Oxford, UK: Oxford University Press. Google Scholar CrossRef Search ADS   Hogg, M. A., Meehan, C., & Farquharson, J. ( 2010). The solace of radicalism: Self uncertainty and group identification in the face of threat. Journal of Experimental Social Psychology , 46( 6), 1061– 1066. doi:10.1016/j.jesp.2010.05.005 Google Scholar CrossRef Search ADS   Iyengar, S. ( 1989). How citizens think about national issues: A matter of responsibility. American Journal of Political Science , 33( 4), 878– 900. doi:10.2307/2111113 Google Scholar CrossRef Search ADS   Iyengar, S., & Hahn, K. S. ( 2009). Red media, blue media: Evidence of ideological selectivity in media use. Journal of Communication , 59( 1), 19– 39. doi:10.1111/j.1460-2466.2008.01402.x Google Scholar CrossRef Search ADS   Jagers, J., & Walgrave, S. ( 2007). Populism as political communication style: An empirical study of political parties’ discourse in Belgium. European Journal of Political Research , 46( 3), 319– 345. doi:10.1111/j.1475-6765.2006 Google Scholar CrossRef Search ADS   Krämer, B. ( 2014). Media populism: A conceptual clarification and some theses on its effects. Communication Theory , 24, 42– 60. doi:10.1111/comt.12029 Google Scholar CrossRef Search ADS   Kriesi, H.-P., Grande, E., Lachat, R., Dolezal, M., Bornschier, S., & Frey, T. ( 2006). Globalization and the transformation of the national political space: Six European countries compared. European Journal of Political Research , 45( 6), 921– 956. doi:10.1111/j.1475-6765.2006.00644.x Google Scholar CrossRef Search ADS   Laclau, E. ( 1977). Politics and ideology in Marxist theory . London, UK: New Left Books. Lodge, M., & Taber, C. S. ( 2013). The rationalizing voter . Cambridge, UK: Cambridge University press. Google Scholar CrossRef Search ADS   Marsh, M., & Tilley, J. ( 2010). The attribution of credit and blame to governments and its impact on vote choice. British Journal of Political Science , 40( 1), 115– 134. doi:0.1017/S0007123409990275 Google Scholar CrossRef Search ADS   Matthes, J., & Schmuck, D. ( 2017). The effects of anti-immigrant right-wing populist ads on implicit and explicit attitudes: A moderated mediation model. Communication Research , 44( 4), 556– 581. doi:10.1177/0093650215577859 Google Scholar CrossRef Search ADS   Mazzoleni, G. ( 2008). Populism and the media. In D. Albertazzi & D. McDonnell (Eds.), Twenty-first century populism: The spectre of Western European democracy  (pp. 49– 64). Basingstoke, UK: Palgrave Macmillan. Google Scholar CrossRef Search ADS   Mudde, C. ( 2004). The populist zeitgeist. Government and Opposition , 39( 4), 542– 564. doi:10.1111/j.1477-7053.2004.00135.x Google Scholar CrossRef Search ADS   Mutz, D. C., & Pemantle, R. ( 2011, July). The perils of randomization checks in the analysis of experiments. Paper presented at the Annual Meeting of the Society for Political Methodology, Princeton, NJ. Pappas, T. S. ( 2014). Populist democracies: Post-authoritarian Greece and post-communist Hungary. Government and Opposition , 49( 1), 1– 23. doi:10.1017/gov.2013.21 Google Scholar CrossRef Search ADS   Richey, S. ( 2012). Campaign advertising and the stimulation and activation of the authoritarian personality. Political Communication , 29( 1), 24– 43. doi:10.1080/10584609.2011.616874 Google Scholar CrossRef Search ADS   Rooduijn, M. ( 2014). The mesmerising message: The diffusion of populism in public debates in Western European media. Political Studies , 62( 4), 726– 744. doi:10.1111/1467-9248.12074 Google Scholar CrossRef Search ADS   Spruyt, B., Keppens, G., & van Droogenbroeck, F. ( 2016). Who supports populism and what attracts people to it? Political Research Quarterly , 69( 2) 335– 346. doi:10.1177/1065912916639138 Google Scholar CrossRef Search ADS   Stroud, N. J. ( 2008). Media use and political predispositions: Revisiting the concept of selective exposure. Political Behavior , 30, 341– 377. doi:10.1007/s11109-007-9050-9 Google Scholar CrossRef Search ADS   Taggart, P. ( 2000). Populism.  Buckingham & Philadelphia: Open University Press. Tajfel, H. ( 1978). Social categorization, social identity, and social comparisons. In H. Tajfel (Ed.). Differentiation between social groups  (pp. 254– 267). London, UK: Academic Press. Thorson, E. ( 2016). Belief echoes: The persistent effects of corrected misinformation. Political Communication , 33, 460– 480. doi:10.1080/10584609.2015.1102187 Google Scholar CrossRef Search ADS   Waisbord, S., & Amado, A. ( 2017). Populist communication by digital means: Presidential Twitter in Latin America. Information, Communication & Society , 20( 9), 1330– 1346. doi:10.1080/1369118X.2017.1328521 Google Scholar CrossRef Search ADS   © 2018 International Communication Association

Journal

Journal of CommunicationOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off