We can’t trust them! The effects of populist blame attributions to political and media elites on perceived factual relativism

Populist attributions of blame in a post-truth world

Essentially, populism can be understood as the emphasis on a central cleavage between ordinary people and corrupt elites (e.g., Mudde, 2004). Populist ideas emphasize that the ordinary people share a universal will, and that politicians and other elite actors are failing because they do not represent the will of the homogenous people. This ideational core of populism entails a blame shifting discourse: The ordinary people as an in-group are referred to as powerless, silenced or deprived, whereas the out-group of elite actors is blamed for being self-interested, short-sighted, corrupt, or hiding reality from the people (e.g., Busby et al., 2019; Hameleers et al., 2017; Vasilopoulou et al., 2014).

Populist communication may not only target political elites as a scapegoat for the problems experienced by the ordinary people. Especially in times when expert knowledge and empirical evidence are frequently disputed, relativized, and subject to partisan attacks (e.g., Van Aelst et al., 2017; Waisbord, 2018), populist messages often delegitimize conventional knowledge. In line with populism’s antagonistic anti-establishment narratives, mainstream media or other conventional sources of knowledge may be framed as a corrupt out-group that is opposed to the honest people (e.g., Egelhofer et al., 2022; Fawzi and Krämer, 2021; Waisbord, 2018). The delegitimization of established knowledge, expert analyses and scientific reasoning is in line with the essence of populism’s antagonistic narrative, as it emphasizes a divide between ‘honest’ and ‘pure’ ordinary people with common sense versus the ‘deceptive’ and ‘corrupt’ narratives disseminated by the elites (Egelhofer and Lecheler, 2019).

Essentially, we expect that populist worldviews correspond with a lower acceptance of evidence-based and factual knowledge conveyed in established media discourse. Hence, populism’s anti-establishment perspective may entail the delegitimization of journalists, established media, and experts as part of the ‘corrupt elite’ (Fawzi and Krämer, 2021; Krämer, 2018). As such, the media logic of established journalistic content that emphasizes expert analyses, empirical facts, fact-checking and balance is at odds with populism’s focus on people centrism, common sense, and an opposition to the distant and elitist narratives central in established media (Hameleers, 2020). Hence, populism should correspond with a preference for people-centric realities and experiences, whilst circumventing elite experts. Therefore, alternative media that resonate with populism’s antagonistic worldview should appeal most to people with stronger populist attitudes (e.g., Müller and Schulz, 2021).

Populism’s epistemic antagonism stressing a divide between the honest people and deceptive or lying elites can come in different forms. Mede and Schäfer (2020) have used the term ‘science-related populism’ to explain how populism can cultivate a divide between honest and good ordinary people versus scientific elites that allegedly do not represent the people’s truth. Similarly, Fawzi and Krämer (2021) refer to anti-media populism to explain the delegitimization of the press from a populist perspective. In line with the emphasis on blame-shifting in populist communication, scientific or media elites are held responsible for deceiving or lying to the people, and for not being willing or able to offer an unbiased and true account of reality. In a similar vein, the ‘fake news’ labels often used by right-wing populist actors target the established media for deceiving the people (e.g., Egelhofer and Lecheler, 2019; Schulz et al., 2020). The populist delegitimization of established knowledge, expertise and truthfulness can thus be targeted at different actors, and the media are not the only scapegoat that is blamed for deceiving the people in a populist framework.

However, to date, most experimental research has either looked at the effects of populist blame attribution that targets political elites (e.g., Busby et al., 2019), scientists (Mede and Schäfer, 2020) or media elites (Egelhofer et al., 2022). We currently lack an understanding of how populist blame attributions either targeted at the established media or elite actors in politics affect people’s perceptions of factual relativism.

In this paper, we understand perceptions of factual relativism as beliefs related to the biased, ideologically colored and relative status of factual knowledge and truths (also see Van Aelst et al., 2017). Perceptions of factual relativism, then, relate to the idea of a post-factual information era in which factual knowledge is constantly debated, delegitimized and subjected to counter-factual interpretations. By explicitly measuring perceptions of factual relativism as a dependent variable, this paper moves beyond studies on the effects of disinformation labels or populist blame attributions on general political attitudes. As such, we aim to establish how the emphasis on the causal responsibility of either political elites or the mainstream media for misrepresenting reality or not representing the people’s will (the independent variable related to populist blame attributions) affects beliefs related to the epistemic status of reality, and the extent to which truth and objectivity exist. As such, we aim to explore the epistemic consequences of populist blame attributions.

The epistemic consequences of populist blame attributions

In this paper, we look at two variants of a populist blame attribution message. First, a populist delegitimizing message in which political elites are blamed for being dishonest, corrupt, and failing to respond to the reality and will of the ordinary people. This corresponds to a more traditional understanding of media populism, which has been understood as the media’s framing or emphasis on a central divide between ordinary people and corrupt elites (e.g., Krämer, 2014). We contrast this with an anti-media populist condition in which not the political elites, but the media elites are framed in opposition to the ordinary people (also see Krämer, 2018). Although we regard both as variants of a populist blame attribution frame, there are noteworthy differences. Specifically, anti-media populism blames the media for spreading disinformation, and for not living up to their normative role of informing the people in an honest manner. This aligns with literature on the ‘fake news’ label often associated with populist communication (Egelhofer and Lecheler, 2019). The delegitimizing message in which political elites are blamed refers more centrally to a normative and causal opposition between political elites and ordinary citizens, who are allegedly not represented by the ‘lying and dishonest’ political establishment.

For both forms of populism, the delegitimization of established institutions of knowledge dissemination is central. This resonates with the practices of alternative and conspiracy media that often challenge or attack the mainstream knowledge spread by established media or political elites (e.g., Heft et al., 2019). Hence, alternative or hyper-partisan media that offer an anti-establishment perspective on reality often include delegitimizing messages that blame the established media and political elites for spreading false information or inaccurate reporting (see e.g., Holt, 2018). As such, populist blame attributions are likely to be encountered in anti-establishment alternative media sources (also see Hameleers & Yekta, 2023).

Previous research has indicated that populist messages that emphasize a divide between ‘us’ and ‘them’ can be persuasive, and have an effect on people’s blame attributions, populist attitudes, and negative evaluations toward the elites (e.g., Bos et al., 2020). Populist ideas may be persuasive as they simplify complex political and societal issues in a comprehensible divide between in-groups and out-groups, whilst absolving the in-group of responsibility. By offering a credible scapegoat for the people’s problems, populist messages may restore a consistent and positive image of the self (e.g., Bos et al., 2020).

These effects can be understood within the framework of social identity framing (see e.g., Bos et al., 2020). Specifically, in line with the premises of social identity framing, populist messages that create a salient social identity of deprived ordinary people threatened by corrupt elites may motivate people to act on behalf of their deprived in-group (Van Zomeren et al. 2008). Thus, populist messages that emphasize an in-group threat whilst forwarding credible scapegoats may strengthen people’s closeness to the in-group as well as their opposition to the blamed out-group. This may also apply to populist messages that attribute blame to the media or politicians for spreading disinformation: Such accusations may strengthen the belief that the people are trustworthy, whereas targeted elites in media and politics cannot be trusted. To restore the people’s status, the out-group may be regarded as culpable, deceptive, and incapable of reporting on facts accurately.

Just like populist communication blaming elite politicians, fake news labels and other disinformation attributions are found to be persuasive (Egelhofer et al., 2022). Egelhofer et al. (2022) used an experiment to show that accusing the media of spreading disinformation affects the trust people have in the attacked outlet. Even more so, for people with stronger populist attitudes, exposure to accusations of disinformation lowers general media trust beyond the outlet that is delegitimized. Accusations of disinformation can lower media trust as they delegitimize information by casting doubts on both its facticity and honesty (e.g., Anspach and Carlson, 2020; Egelhofer et al., 2022).

In line with this reasoning, populist blame attributions that cast doubt on the media’s democratic role and the intentions of politicians and the media to correctly inform the public may thus undermine trust by highlighting that news users cannot rely on the media to fulfill their role of informing them in an accurate and complete manner. As such, we expect that beliefs related to objectivity and the (subjective) truth value of information captured in perceived factual relativism (Van Aelst et al., 2017) are affected by populist blame attributions emphasizing distrust in elites.

In line with the aforementioned epistemic consequences of populist blame attributions, we further expect that populist attributions of blame can result in the avoidance of established information sources. As indicated by Müller and Schulz (2021) populist ideas may resonate with specific media behaviors and preferences. More specifically, the more people perceive a binary divide between ordinary people and corrupt elites, the more they may avoid established media and approach alternative media sources. Thus, populism’s anti-establishment narrative may correspond to anti-establishment media preferences. Considering that disinformation accusations delegitimize mainstream elites and enhance distrust in established sources of information (Egelhofer et al., 2022), we expect that exposure to disinformation accusations also increases the likelihood that people avoid the mainstream media sources scapegoated.

Against this background, we introduce the following hypotheses.

H1a: Exposure to populist blame attributions of disinformation to media and politicians results in stronger perceptions of factual relativism compared to messages in which populist blame attributions are absent.

H1b: Exposure to populist blame attributions of disinformation to media and politicians results in a higher likelihood to not select established information sources compared to messages in which populist blame attributions are absent.

As part of the analyses, we will compare whether the effects of populist blame attributions differ for the targets that have blame attributed to them. Essentially, literature on the fake news label (e.g., Egelhofer and Lecheler, 2019) and anti-media populism (Fawzi and Krämer, 2021) have mostly regarded the media and journalists as part of the elitist out-group responsible for deceiving the ordinary people. However, populist communication can also blame political elites for spreading disinformation or fake news, especially when the truth claims of political opponents do not align with the issue positions of the attacker (Farkas and Schou, 2018).

Here, we postulate that an accusation aimed at the news media by populist communication is potentially more effective in enhancing factual relativism than a blame attribution aimed at political actors. Compared to political actors, citizens may be more likely to perceive the media as having a neutral role as ‘truth crusaders’ or disseminators of objective information (Skovsgaard et al., 2013). Citizens may, however, be less likely to associate the role perceptions of political actors with accuracy and truth telling. As such, we expect that delegitimizing accusations targeting the media are more likely to affect factual relativism and the avoidance of established media than political elites, who are less likely to be associated with, and held accountable for, truth-telling and the objectivity norm. The hypothesis reads as follows:

H1c: Exposure to populist blame attributions of disinformation to the media has stronger effects on perceptions of factual relativism and the likelihood to select established information sources than exposure to populist blame attributions to political elites.

The effects of anti-media populism

Along with general populist blame attributions, we expect that attributions of blame to the media, in particular, correspond to disinformation perceptions and a decreased likelihood to approach mainstream media sources. Thus, although we did not distinguish between both variants of populist blame attributions under H1, we expect that the difference becomes more important, when we shift our focus to other outcome variables that directly correspond to perceived disinformation and the avoidance of established media. Hence, the anti-media populism condition explicitly stresses that mainstream media misrepresent reality and deceive the ordinary people, which should make the association of lying and fake media more salient among recipients (also see Egelhofer et al., 2022).

Populist attributions of blame to the media can be understood as an expression of anti-media populism, which expresses a divide between honest ordinary people and ‘bad’ media elites and journalists allegedly conspiring with elite actors (Fawzi, 2019; Krämer, 2018). This populist accusation links up with a hostile media bias that accuses the mainstream media of being a mouthpiece of elite and corporate interests, instead of striving towards an honest and truthful coverage of reality (Fawzi, 2019). Thus, different than anti-political elites populism, anti-media populism contains a fake news accusation that casts doubt on the independence, facticity, and honesty of the mainstream media.

Based on this conceptualization, we specifically expect that there is a strong resonance between populist blame attributions to the media and perceived disinformation among recipients. Populist blame attributions that attack the mainstream media emphasize that the media are intentionally lying to the people, which may cultivate the impression that established media sources cannot be trusted (e.g., Egelhofer et al., 2022). As a consequence of this delegitimization, recipients may question both the intentions and facticity of media coverage (Anspach and Carlson, 2020). We therefore expect that these blame attributions, more than populist blame attributions targeting politicians, result in perceptions of disinformation associated with mainstream media. Perceived disinformation is defined as the belief that the media are intentionally spreading false information. Although closely connected to perceived factual relativism, perceptions of disinformation tap the specific media evaluation that established sources of information are intentionally deceptive.

As part of H1b, we expected that populist blame attributions in general may result in a higher likelihood to avoid established information sources. Arguably, more than anti-political elite populism, anti-media populism delegitimizes and attacks established media that are blamed for spreading fake news, and deceiving the people. This variant of media populism explicitly blames established media sources for deceiving ordinary people, and for hiding the reality from them through misleading news coverage. We therefore introduce the following hypotheses on the effects of anti-media populism.

H2a: Participants exposed to anti-media populist messages are more likely to associate the media with disinformation than participants not exposed to anti-media populist messages.

H2b: Exposure to populist messages delegitimizing the mainstream media results in a higher likelihood to not select established information sources.

The moderating role of media distrust and political cynicism

Extant research has indicated that the effects of populist communication are stronger for people with higher levels of political cynicism (e.g., Bos et al., 2013). Political cynicism here refers to the systematic distrust people have in the ability and intentions of political actors to represent voters in an accurate and honest manner. More cynical recipients may be more susceptible to populist communication because their prior beliefs on politics resonate with populism’s core message that stresses that the elites cannot be trusted and are primarily self-interested. Considering that populist ideas appeal to disenchanted voters that have lost their faith in political elites (Jagers and Walgrave, 2007), it can be expected that populist communication has the strongest effects on segments of the population with more cynical beliefs about politics and politicians. For these people, blame attributions may resonate most with their existing distrust of the established order.

Although this mechanism has not been tested outside of the political realm, we expect that media distrust plays a similar role for anti-media populism. People with higher levels of media distrust are likely to believe that the media are unable to fulfill the expectation of informing the public in an accurate and honest manner (e.g., Kohring and Matthes, 2007). Populist messages that blame the media for spreading ‘fake news’ may be congruent with this negative evaluation of the media’s credibility and trustworthiness. Especially given that, in this study, the populist attributions of blame were delivered by a (fictitious) media outlet resembling the framing of alternative media, it can be argued that pre-treatment levels of overall trust in the media play a role in the effects of anti-media populism. Taken together, we introduce the following hypothesis on the moderating role of political cynicism and media distrust for populist blame attributions to the media (anti-media populism) and political elites (anti-political elites populism).

H3: The effects of populist blame attributions on (a) perceived factual relativism; (b) perceived disinformation and (c) the selective avoidance of established information sources are strongest for participants with more pronounced levels of media distrust and political cynicism.

Sample

The recruitment of respondents in the US was outsourced to Dynata. This international research company relies on voluntary opt-in databases that seek to be representative of the U.S. population regarding core demographics, political attitudes, region, and voting behavior. From this database, participants were invited to contribute via e-mail or the digital contributor platform of Dynata. The company ensured that participants could not participate in different surveys at the same time, and panelists had a maximum number of weekly projects they could participate in.

Of all the panelists invited, 428 participants completed the full survey experiment (completion rate of 87.5 %). The final sample size was informed by a power analysis: Based on existing research, we expected small effect sizes. With three conditions and two issues to rate, we aimed for 125 completes per condition to achieve a power of .80. In the final sample, female participants were slightly overrepresented (49.8 % female, 47.0 % male, 1.1 % other or preferred not to say). More lowly-educated participants were underrepresented (no education at all or only primary education) at 4.5 %, but we obtained a good balance between more-highly educated (51.4 %) and moderate/lower levels of education (42.0 %). In terms of ideology, the final sample distribution shows a normal distribution in which left-wing (38.1 %) and right-wing participants (42.2 %) are equally represented. The mean age of participants was 50.16 years (SD = 19.85). Randomization checks that assessed whether distributions on key demographics were similar across the conditions succeeded (see Appendix B).

Independent variables and stimuli

All stimuli were formatted as online news articles (see Appendix A for the stimuli texts). As a control condition, we used real news articles on immigration and climate change. We slightly changed the wording and lay-out in order to circumvent familiarity and prior exposure.

For the populist manipulations of the immigration article, a populist anti-establishment interpretation was connected to the statements voiced in the control condition. For the political elite blame attribution, the following narrative was made central: “The immigrants allowed to enter our country illegally are increasingly allowed to profit from our welfare. The failing policies of our current government allow immigrants to receive more welfare than native U.S. citizens who need to receive support most in times of the crisis we are facing.” In the anti-media populist condition, the attribution of blame was targeted at the media, and the accusation specifically emphasized that the mainstream media were manipulating and disinforming the people: “At least, the established media want us to believe that immigration is declining. They deliberately hide the fact that immigrants allowed to enter our country illegally are increasingly allowed to profit from our welfare. They conceal the fact that immigrants receive more welfare than native U.S. citizens, who need to receive support most in times of the crisis we are facing.”

For the climate change conditions, we also used an explicit populist framework in both the political and media elite blame attribution conditions. Because the control condition used as a starting point for the manipulations already referred to the potentially misleading nature of official information, the blame attributions in both populist conditions more specifically referred to a (deliberate) misrepresentation of reality by the political and media elites, respectively. In the political elite blame attribution condition, the main narrative explicated that “They deliberately mislead the ordinary people for electoral gain and profit – creating threats that in reality are not happening.” In the conditions emphasizing that the media were to blame for deceiving the people, the following interpretation was central: “The established press deliberately misleads the ordinary people for profit – creating threats that in reality are not happening.” (see Appendix A for all stimuli).

We made sure that all stimuli, apart from the different levels of the independent variables, were kept similar within topics. We also used exactly the same lay-out, sourcing, and image across conditions. The articles were also equal in length, arousal and valence (with the exception of the control versus treatment conditions, that were more negative and emotional as they emphasized a populist narrative that is inherently negative and emotional in style, see e.g., Engesser et al., 2017). Qualitative pilot testing of the stimuli and the post-hoc manipulation checks (see section on Manipulation Checks) further confirmed that the manipulations were perceived as intended. Hence, all stimuli were seen as equally credible/realistic and representative of actual news coverage on the depicted issue. Specifically, the conditions in which populist blame attributions were included (M = 3.86, SD = 1.75) and the control conditions in which they were absent (M = 4.02, SD = 1.78) were comparable in perceived credibility on a 7-point scale (p > .05).

Dependent variables

After exposure to the stimuli and a short open-ended thought listing task that aimed to distract people from the news articles, the dependent variables were measured. Perceived factual relativism was measured using seven statements (i.e., “It is hard to tell which sources of information can be trusted”, “Facts are mostly used to justify political perspectives”, “The truth is in the eye of the beholder”), which were averaged into a scale (M = 4.45, SD = 1.42, Cronbach’s α = .888). All items were measured on 7-point scales (1 = completely disagree, 7 = completely agree). The items were developed for this study, but based on the conceptualization of post-factual relativism (see Van Aelst et al., 2017).

For the likelihood to select established information sources, we asked participants to indicate the likelihood that they would select information on the issues of climate change and immigration after exposure to messages on these two issues. Specifically, the following question wording was used: “After reading this message, could you please indicate to what extent you would prefer to read additional news on the topic of climate change/immigration from the following information sources? (1 = I would not like to read follow-up information from this source at all, 7 = I would prefer to read follow-up information from this source”). Intentions to select follow-up information were indicated for both topics separately and included the following sources: The New York Times, an independent fact-check platform, BBC News, Breitbart, Telegram, Fox News, Twitter, a mainstream media source, an alternative news website. The selection was based on diversity and balance in partisan viewpoints and familiarity. In addition, for theoretical reasons, we aimed to include hyper-partisan and alternative media that are likely to align with populist viewpoints, such as Breitbart (e.g., Müller and Schulz, 2021).

We used two alternative approaches to compute the likelihood to avoid established information sources. First, we used an average scale of the likelihood to select information from established news sources (New York Times, BBC News, Fox News, M = 2.47, SD = 2.32, Cronbach’s α = .962). Overall, the average score of 2.47 on a 7-point scale indicates a low likelihood to select mainstream information as a follow-up of exposure to the conditions. As robustness check, we also ran the analyses with all different media outlets separately, which did not change the results. In addition, as news preferences are likely to be driven by partisanship and existing habitual patterns, we re-ran the analyses using just the one-item measure of intentions to select “a mainstream media source” (M = 3.78, SD = 2.61). As the findings are again similar to the more extensive battery of intentions to use specific news sources, we use the average scale of intentions to select established information sources in the main analysis, which was reverse-coded to indicate a lower likelihood to select established information sources after exposure to delegitimizing populist content. Although it should be acknowledged that selective exposure and avoidance can be regarded as separate concepts theoretically, we aim to assess the likelihood of selecting information from various sources. As such, we rather look at the extent to which the likelihood of exposure to mainstream media is lower when people are exposed to delegitimizing content.

As a final dependent variable, we measured perceptions of disinformation related to the news media (also see Hameleers, 2020). This latent scale was constructed as an average of seven different statements capturing people’s cynical evaluations of the news media (i.e., “The news media are deliberately lying to the people”, “The news media do not accurately report on facts that happened” and “The news media only serve their own interests”). On a 7-point scale, the mean score on disinformation perceptions related to the news media was 4.48 (SD = 1.53, Cronbach’s α = .914).

Moderators: Media trust and political cynicism

Media trust was measured based on existing validated measures pointing to different dimensions of credibility and trust (e.g., Kohring and Matthes, 2007). We specifically used a five-item average scale of media trust (all items measured on 7-point completely disagree-completely agree scales): “The news media are fair when covering the news”, “The news media tell the whole story when covering the news”; “The news media are unbiased when covering the news”, “The news media are accurate when covering the news” and “The news media separate facts from opinions when covering the news.” (M = 3.57, SD = 1.82, Cronbach’s α = .950). These items have been used in other studies that capture media (dis)trust as an dependent variable, and mostly correspond to the accuracy, completeness, and fairness dimensions of news media trust (e.g., Kohring and Matthes, 2007). Using an alternative one-item measure of media trust “I think you can trust the media most of the time” yields the same results as the scale constructed using different statements.

For political cynicism, a validated scale consisting of five items was used (see e.g., Bos et al., 2013). Measured on the same scale as the media trust items, we used the following statements: “Politicians are generally driven by self-interest”, “Political parties are only interested in my vote, and not my opinion”, “Politicians do not know what is going on in American society”, “People like me have no interest in what the government does” and “Politicians talk too much and take too little action” (M = 5.30, SD = 1.29, Cronbach’s α = .859).

Procedures and manipulation checks

Participants entered the survey through the digital invitation sent by Dynata. When they entered, they completed the informed consent procedures (all procedures have been approved by the University’s ethical review board under number [anonymized for peer review]). Contingent upon accepting the terms and conditions of the study, participants first answered a block of pre-treatment questions (i.e., age, gender, education) and the questions used as moderators (the batteries of items measuring media (dis)trust and political cynicism). In the next step, they were exposed to the treatments. A short block of questions followed after each article, including questions on the intention to select or avoid a variety of media sources as a follow-up of the information they had just read. After reading the messages and answering specific questions related to these (manipulated) messages on climate change and immigration, participants answered questions for the general dependent variables: Perceived factual relativism and disinformation beliefs related to the news media. More details on the procedures are offered in Appendix B.

Finally, participants answered questions on the manipulation checks and received a careful debriefing. All manipulations succeeded, and participants who failed the manipulation checks were not excluded from the study (see Appendix B for details). It has to be noted here that participants in the control condition also perceived some degree of blame attribution, which may be due to the fact that – although populism was absent in the control conditions – the same threats and developments were mentioned to allow for more consistency across conditions. In addition, we see some overlap in the perception of blame across the anti-media and anti-political elites blame attribution conditions. Yet, robustness checks excluding participants that clearly did not differentiate between conditions did not affect any of the conclusions reported in the paper.

The effects of delegitimizing populist messages on factual relativism and media selection

We first of all hypothesized that exposure to a populist delegitimizing message – irrespective of the target of blame attribution – would result in more pronounced perceptions of factual relativism (H1a) and a higher likelihood to not select established information sources (H1b). We tested this hypothesis using independent samples t-tests (we looked at the two blame attributions together, and contrasted this with the control condition). For perceptions of factual relativism, we see a non-significant effect of exposure to the populist conditions versus the control (t(426) = .128, p = .865, 95 %CI[-.25, .15]). We thus find no support for H1a. Looking at the effect of exposure to populist blame attributions on the likelihood to select mainstream media (H1b), we ran the analyses for the two topics separately as this dependent variable was measured after each message (on climate change and anti-immigration). For climate change information, participants exposed to a populist blame attribution were not more likely to avoid mainstream media than participants exposed to the control condition (t(214) = –.36, p = .361, 95 %CI[-.60, .42]). The same results were found for the anti-immigration messages (t(244) = .370, p = .712, 95 %CI[-.42, .62]). H1c is also not supported: Both forms of populist blame attribution do not significantly affect perceived factual relativism or mainstream media selection. Against this backdrop, our results offer no support for H1: Exposure to populist blame attributions do not affect perceptions of factual relativism or the selection of mainstream media.

H2a postulates that anti-media populist messages activate perceived disinformation. Based on a one-way ANOVA in which the conditions (populist blame attribution to politicians versus populist blame attribution to the media versus control) were included as independent variable and perceptions of disinformation as dependent variable, we can first of all see that the overall model is not significant (F(2, 428) = .071, p = .492, partial η² = .003). Inspecting the corrected mean score comparisons, we can see that participants exposed to messages blaming the media (M = 4.34, SD = 1.60) hold similar levels of disinformation perceptions as participants exposed to blame attributions to politicians (M = 4.54, SD = 1.56) or the control (M = 4.51, SD = 1.41). We therefore do not find any support for H2a.

For H2b, which postulates that exposure to populist messages delegitimizing the mainstream media results in a higher likelihood to avoid established information sources, we ran ANOVAs in which the different topics were taken into account. Again, the overall model is not significant (F(5, 462) = 1.56, p = .171, partial η² = .017). The corrected pairwise mean score comparisons are also non-significant, although one finding is worth mentioning here. Concretely, exposure to populist messages blaming politicians for immigration problems resulted in a higher likelihood to select mainstream media (M = 4.14, SD = 1.86) than exposure to populist messages blaming media elites (M = 3.45, SD = 2.11). Although the difference is not significant by conventional thresholds, it does show that targeting mainstream media with fake news accusations can demotivate the selection of mainstream media, whereas this is not the case for populist messages blaming political elites.

The moderating effects of trust

We expected that the effects of delegitimizing populist labels are strongest for participants with higher levels of distrust in the media (H3). The regression models summarized in Appendix C, Table 1 (for perceived factual relativism) and Table 2 (for perceived disinformation) offer some support for this hypothesis. First of all, the significant interaction effects between populist blame attributions to the media and politicians indicate that participants with higher levels of distrust in the media are most likely to have stronger perceptions of factual relativism after exposure to a populist blame attribution versus the control message (Table I, Model III). Second, as can be seen in Table 2 of Appendix C (Model III), the significant interaction effect between populist blame attributions to the media and media distrust indicates that participants with higher levels of media distrust were more likely to perceive after exposure to a populist message that the media disinform the public.

Looking at Table 3 (Appendix C), which reports the regression models for the effects of delegitimizing populist messages on mainstream media avoidance, we find some additional support for H3. Hence, there is a significant two-way interaction effect between exposure to anti-media populism and media distrust on the avoidance of established media sources (B = .17 SE = .08, β = .18, p = .043). More specifically, participants with higher levels of distrust in established media are most likely to not select established media sources after exposure to populist messages that blame the media for spreading fake news (versus participants exposed to the control condition without populist content). This effect was not found for exposure to populist messages that blame political elites for disseminating fake news.

Our findings offer no support for the moderating role of political cynicism (see Tables 1 through 3 in Appendix C). More specifically, none of the interaction effects between exposure to populist delegitimizing labels and political cynicism on (a) perceived factual relativism, (b) perceived disinformation or (c) the selective avoidance of mainstream media are significant. Thus, although participants with higher levels of distrust in mainstream media were more likely to be affected by anti-media populist messages, more cynical participants were not more vulnerable to delegitimizing content.