“I Don’t Believe Anything They Say Anymore!” Explaining Unanticipated Media Effects Among Distrusting Citizens

The erosion of political and societal trust, polarization, and the omnipresence of disinformation may undermine the per‐ ceived trustworthiness of established sources of information. Yet, many forced exposure media effect studies in the field of political communication studying polarizing issues such as disinformation and populism assume a baseline level of trust among participants exposed to seemingly neutral information. This neglects long‐standing issues of distrust in the press and trends toward increasing distrust among growing segments of the population. Resistance toward established infor‐ mation presented as news may result in unanticipated findings, as a substantial part of the population may not accept these sources as trustworthy or neutral. To enlighten confusion, this article relies on two different experiments ( N = 728 and N = 738) to explore how citizens with low levels of trust and high dissatisfaction with the established order respond to information from established information sources. Our main findings indicate that participants with higher levels of pop‐ ulist attitudes, media distrust, and fake news perceptions are more likely to find established information untrustworthy. They are also less likely to agree with the statements of such content. These findings indicate that media effect studies assuming univocal acceptance of seemingly neutral information may fall short in incorporating problematic trends toward factual relativism in their design.


Introduction
Issues related to distrust in the mass media, science, and other institutions have existed way before the advent of digital media. As illustrated by Bennett et al. (2007), the mass media does not always operate as an independent fourth estate, acting as a watchdog of powerful institutions and providing citizens with a critical outlook on socio-political issues. Yet, issues related to declining trust in the established order may have been accelerated and amplified by the affordances of social media. In digital information contexts, a plethora of alternative counter-factual narratives competes for the audience's attention and legitimacy (Waisbord, 2018).
As a consequence, growing segments of the population may not know whom to trust or systematically circumvent elite sources altogether. At the same time, political movements that cultivate distrust in the established order-such as radical right-wing populists-mobilize and amplify sentiments of disenchantment among citizens (e.g., Hameleers, Bos, Fawzi, et al., 2018). Together, these developments coincide with an information era in which facts have become relative and subject to distrust and "fake news" accusations (Van Aelst et al., 2017). In this setting, the perception that fake news is everywhere may dramatically decline people's trust in authentic information. Against this backdrop, we argue that long-standing issues related to distrust in the media may have taken on a different shape in the context of current developments toward mis-and disinformation and weaponized applications of these terms in a digital media landscape.
These trends are problematic for democracy, as a citizenry that disagrees on basic facts cannot make wellinformed political decisions (e.g., Arendt, 1967). Distrust and disenchantment may also impact the conclusions we draw from empirical evidence on media effects, as the assumption that all people are equally willing and able to accept the information sources we (forcefully) expose them to may lead to inconsistent conclusions. In this setting, we need to resolve confusion about contradictory media effects in a communication setting of polarization, distrust, and factual relativism. The question central in this article is therefore whether citizens' disenchantment and distrust result in disagreement with and the reduced credibility of information that is presented as authentic, neutral, and factually correct. Here, we specifically focus on the field of political communication that has dealt with issues related to declining trust in (established) information sources (e.g., Fawzi, 2019;Schulz et al., 2020) or scientific elites (e.g., Mede & Schäfer, 2020), especially among citizens with more pronounced populist attitudes (e.g., Schulz et al., 2020).
Against this backdrop, this article uses insights from two different experiments to explore how citizens with low levels of trust and high dissatisfaction with the established order respond to information coming from established information sources. Hence, most media effect studies in political communication research rely on forced exposure designs that may not sufficiently take into account some people's experienced distrust in elite information. Research on the effects of disinformation and corrections, for example, mostly used a forced exposure design to present people with fact-checks from allegedly neutral sources (e.g., Nyhan & Reifler, 2010). In a similar vein, most research on the effects of populist communication used a forced exposure design in which populist messages are presented as seemingly neutral news messages (e.g., Bos et al., 2019; see also Müller et al., 2018). If we take into account that citizens supporting a populist ideology would normally selectively avoid or severely distrust the sources referred to in such experiments (Schulz et al., 2020), how can we validly assess the effect of such polarizing content among disenchanted segments of the audience?
Considering the findings of the Reuters Institute Digital News Report 2021 (Newman et al., 2021) that only 44% of people trust the news most of the time, this has far-reaching consequences for the conclusions we draw about media effects. For example, null effects or contradictory findings in experimental research on populism or disinformation (e.g., Hameleers et al., 2020) may partially be driven by distrust in information sources presented to participants, rather than the actual failure of the stimulus to activate attitudes in line with the predictions. As a main contribution, this article explores the impact of distrust and dissatisfaction with the estab-lished order on media effects surrounding polarizing issues in political communication by relying on two different experimental studies using different samples, designs, and issues. It herewith aims to enlighten the confusion of unanticipated findings in media effect studies that either find null effects or contradictory patterns for some segments of the population.

Truthfulness in an Era of Post-Factual Relativism
Although mis-and disinformation are by no means novel phenomena, the affordances of digital information ecologies have been associated with the amplification and acceleration of disinformation (e.g., Van Aelst et al., 2017;Waisbord, 2018;Zhang et al., 2021). The digital information environment hosts many nonprofessional communicators who can communicate with audiences directly and circumvent traditional journalistic routines and gatekeepers. This has arguably led to a fragmented information ecology where a plethora of alternative narratives compete for legitimacy and the audience's attention (Waisbord, 2018). In this setting, verified factual information may be dismissed as opinions or politicized as biased content, whereas conspiracy theories and disinformation are presented as truthful interpretations of reality. This can confuse news users about the epistemic status of factual knowledge and empirical evidence. In addition, many (political) actors use their direct communication channels to de-legitimize established facts, mainstream media, or expert sources, accusing them of spreading "fake news" (e.g., Egelhofer & Lecheler, 2019;Waisbord, 2018). The ongoing legitimization of alternative anti-establishment narratives may cause a downward spiral of distrust: The antagonistic construction of "the truth"-fueled by the delegitimizing discourse of radical right-wing populists-may amplify existing levels of distrust in the established political and media order (e.g., Van Aelst et al., 2017;Waisbord, 2018).
Disinformation-which we can define as fabricated, doctored, or manipulated information that is made and disseminated to achieve certain political goals (e.g., )-may be spread to raise cynicism in the established political order and fuel polarized divides in society (e.g., Bennett & Livingston, 2018). There is ample evidence that disinformation may succeed in this goal. Using an experimental study, Vaccari and Chadwick (2020), for example, found that deepfakes do not directly mislead recipients. Rather, its delegitimizing discourse resulted in lower trust in the (digital) news environment. If we consider the fact that disinformation is thriving around key events such as the Covid-19 pandemic, where uncertainty is remarkably high, the existence of many counterfactual narratives, disinformation, and conspiracy theories in people's newsfeeds may have cultivated existing levels of distrust in the established order. But what are the consequences of distrust in the established order for mapping the effects of mainstream media coverage on citizens and public opinion?

The Consequences of Declining Trust and Increasing Dissatisfaction for Media Effect Studies
Increasing distrust and dissatisfaction with the established order may have severe ramifications for how we perceive and study media effects. Here, we define trust in the broadest sense of the concept: an individual's evaluation or judgment of the likelihood that a trustee (i.e., the media, the political establishment) can fulfill the expectations of a trusting actor (i.e., a news user or citizen; Baier, 1986). In a well-functioning democracy, news users should expect the media and political elites to inform them of key developments in an accurate, complete, honest, and transparent manner. People who distrust the media cast doubt on the extent to which the news media are capable of fulfilling these role perceptions (e.g., Brosius et al., 2021). In today's information setting, in which the aforementioned developments of post-factual relativism, fake news accusations and disinformation take center stage, these role expectations are under fierce attack (e.g., Tamul et al., 2020). Arguably, news users may not systematically hold the evaluation that the news media and political elites can fulfill their democratic roles, resulting in a lack of trust or distrust in the media and political institutions (Hameleers et al., 2020). This perception may either be experienced as skeptical attitudes (i.e., a critical attitude towards the established order and the media) or cynicism (i.e., a more systematic rejection of the established order or the media as an information source; see, e.g., Pinkleton et al., 2012).
Why is it relevant to consider these developments in media effects studies? Importantly, for people to be influenced by the media, they have to accept the message as truthful (see e.g., Schaewitz et al., 2020). High levels of distrust or existing disagreement with the foundations of a message can result in reactance, avoidance, or the rejection of a message's arguments. In line with this, the high levels of audience fragmentation in the digital age correspond with (partisan) selective exposure and minimal persuasive media effects (e.g., Bennett & Iyengar, 2008). Yet, this conceptualization of minimal media effects is not uncontested. In a response to Bennett and Iyengar, Holbert et al. (2010) argue that we need to regard persuasion as something more compelling than changes in attitudes. More specifically and related to the fragmented and high-choice information ecology, media effects should be understood as the formation and reinforcement of attitudes and beliefs too. Here, a reinforcing spiral model of media selection and effects is especially worthwhile to consider (Slater, 2007): Media effects can best be understood as the consequence of an over-time process in which selection and effects are entangled into a mutually reinforcing mechanism that leads to attitude reinforcement over time.
This understanding of media effects has probably increased in relevance amidst increasing concerns about the relative status of untruthfulness and post-factual relativism (Van Aelst et al., 2017;Waisbord, 2018). In digitized media environments, multiple alternative truth claims, conspiracies, and counterfactual narratives compete for the audience's attention (Waisbord, 2018). There is no singular truth that is accepted across audience segments, and the high-choice setting of social media allows citizens to select the version of reality that best fits their existing beliefs or (partisan) identities, a development that is further amplified by algorithms and the social embedding of disinformation (Lukito et al., 2020;Zhang et al., 2021). This setting of high choice and competing claims on truthfulness may not only promote the selection of attitude-reinforcing content but also engenders audience distrust in the establishment's version of the truth and factual reality (Van Aelst et al., 2017).
High levels of distrust are reflected in the increasing salience of disenchantment in the form of populist attitudes and political cynicism (Hameleers et al., 2020;Schulz et al., 2018), as well as overall low trust in the news and online media (Newman et al., 2021). Populist attitudes map the perceived divide between the ordinary people as an in-group and the allegedly corrupt elite as an out-group that fails to represent the common people (Akkerman et al., 2014;Schulz et al., 2018). Such attitudes may play a crucial role in how (mainstream) information is perceived. People with stronger populist attitudes tend to believe that the media disseminate fake news (e.g., Fawzi, 2019;Schulz et al., 2020) and that most news media spread disinformation . In addition, populist attitudes often have an anti-expert or anti-media dimension: Experts, established facts, and scientific knowledge are severely distrusted among populist segments of the audience (Mede & Schäfer, 2020). Thus, media distrust, populist attitudes, and perceptions of fake news may all correspond to growing levels of disenchantment with the established order and information.
Taken together, shifts in the audience's interpretation of (un)trustworthiness in a digital age of fragmentation have important implications for how we may understand media effects (see also Van Aelst et al., 2017). Citizens with stronger populist attitudes, fake news perceptions, or other distrusting and disenchanted views on media and society may systematically reject or counterargue information that comes from mainstream media or established information channels (Fawzi, 2019;Schulz et al., 2020). Taking into account that relatively high proportions of the audience hold (moderate) populist beliefs (e.g., Schulz et al., 2020) or fake news perceptions , we assume that such audience segments are also represented in public opinion research aiming to measure the impact of media content on society, potentially resulting in an unmeasured bias in the estimation of effect sizes.
We, therefore, argue that a failure to detect anticipated direct effects resulting from exposure to a media stimulus may in part be driven by distrust and cynicism toward such content among audience segments who no longer accept established information sources as trustworthy. People accepting the message and source (i.e., people who do believe that the mainstream media and established information sources are trustworthy and credible sources of information) may display anticipated effects, whereas reactance by distrusting segments may cancel out effects, leading to an underestimation of the potential effects of media content. As many media effect studies rely on source cues and stimuli reflecting everyday formats used by established information channels and news sources, we expect that disenchantment and distrust directed at such elite channels (i.e., fake news perceptions or populist attitudes) may play a key biasing role in the assessment of media effects.
Against this backdrop, we postulate the following central hypotheses: H1: People with more pronounced populist attitudes are more likely to rate established information as uncredible or disagree with its arguments compared to people with less pronounced populist attitudes.
H2: Participants with more pronounced levels of media distrust are more likely to rate established information as uncredible or disagree with its arguments compared to more trusting participants.
H3: People with more pronounced fake news perceptions are more likely to rate established information as uncredible or disagree with its arguments compared to people with less pronounced fake news perceptions.

Methods
We rely on two different data collections that vary in terms of topical scope and panel composition. Specifically, we rely on one experiment measuring participants' responses to corrective information in the US and one experiment that looks at responses to episodic and thematic frames in the US. Altogether, we capture variety in panel compositions (samples were recruited via different means and panel companies) and topics (climate change and immigration). The consistent part across the data collections is that the stimuli are presented as neutral sources of information that were allegedly published in recent US news coverage, a scenario that is also used in many media effects studies in the field. This allows us to explore to what extent and how participants indicating to have lower media trust and higher levels of dissatisfaction with the establishment respond differently to dependent variables aiming to measure (a) the credibility of the stimuli and (b) agreement with the positions forwarded in it.

Theory on Misinformation and Corrective Information
The first study focuses on misinformation and corrective information. For this study, we define misinformation as an umbrella term for information that is factually incorrect or not based on relevant expert knowledge and/or empirical evidence (e.g., Vraga & Bode, 2020). It may refer to both the dissemination of unintentionally false information and doctored, fabricated, or manipulated information disseminated with the intention to deceive or mislead-also known as disinformation (e.g., . In response to the alleged uncontrolled dissemination of misinformation, numerous interventions to pre-or de-bunk false information have been introduced. In this study, we specifically focus on corrective information presented after exposure to misinformation: fact-checks (see also Nyhan & Reifler, 2010;Wood & Porter, 2018). Fact-checks are typically short, factual messages that check the veracity of statements to arrive at a verdict of the (un)truthfulness of information. They may be effective as they rely on short, simple, and factual messages that forward an unequivocal conclusion about truthfulness (Lewandowsky et al., 2012). Although some studies have indicated that such messages can lead to reactance (e.g., Thorson, 2016), more recent and meta-analytic research has shown that fact-checks overall have a positive effect on correcting misinformation (e.g., Chan et al., 2017).
For the first study, we look at the effects of exposure to both misinformation and corrective information presented in response to false information. As citizens with more pronounced levels of distrust are more likely to accept misinformation (e.g., Zimmermann & Kohring, 2020) and in line with findings that citizens with populist attitudes are more likely to distrust established information (e.g., Schulz et al., 2020), we expect that misinformation forwarding an anti-establishment narrative has the strongest effects for distrusting and populist audience segments. Here, we look at the effects on both the credibility of and agreement with false statements. In line with Schaewitz et al. (2020), we understand credibility as the assessment of the "truth value" of a (news) item (see also Lewandowsky et al., 2012). Although the concept of credibility is multifaceted, as it may involve a complex interaction between evaluations of the source, recipient, and message characteristics (Wathen & Burkell, 2002), we aim to measure credibility as the overall evaluation of the credibility of the news item shown to participants. The level of agreement, the second dependent variable, was measured to map the effects of (un)corrected misinformation on message-congruent beliefs. In line with previous research, we expect that misinformation may mostly influence the beliefs of recipients with congruent prior perceptions (e.g., Schaewitz et al., 2020).
In this case, as the misinformation message used in this study forwards a populist anti-establishment narrative, higher levels of populist attitudes, fake news perceptions, and media distrust should make misinformation more persuasive. In line with research demonstrating that these audience segments are more likely to reject or counter-argue fact-checks (Nyhan & Reifler, 2010), the opposite may be expected for exposure to fact-checks. We specifically introduce the following hypotheses: H1: People with more pronounced levels of populist attitudes (a), media distrust (b), and fake news perceptions (c) are more likely to agree with misinformation than accurate information.
H2: People with more pronounced levels of populist attitudes (a), media distrust (b), and fake news perceptions (c) are less likely to be affected by factchecks in their credibility ratings and agreement with statements emphasized in misinformation than people with less pronounced populist attitudes, distrust, and fake news perceptions.

Data Collection
This study relies on a survey-embedded experiment in the US for which data collection was outsourced to an international research agency. The design can be summarized as a 2 (misinformation: present versus absent) × 2 (fact-checking: present versus absent) between-subjects factorial design. Participants were recruited by Kantar Lightspeed, an international research agency with a large and diverse global database of survey participants. A total of 728 participants completed the study. The composition of the sample closely reflects the US population in terms of age, gender, education, region, and political preferences (differences between sample and population composition fall within a 10% deviation).
First, all participants saw misinformation on crime rates and immigration (i.e., stating that crime rates were increasing due to rising immigration), and, depending on the condition they were randomly allocated to, saw a fact-check that corrected the misinformation (the fact-check came from an independent established source, PolitiFact). The misinformation connected immigrants to alleged rising crime rates (this was false information, as crime rates in the US were decreasing at the time of data collection) and stated that violent crimes increased rapidly due to the threat coming from immigrants. The misinformation condition falsely depicted this situation as a threat to the native population. The fact-check used factual evidence, objective knowledge, and expert analyses to refute this misinformation. Relevant to this study, the fact-check can be regarded as an established source of information: It comes from an independent source that is part of elite media. In this study, we thus contrasted alternative information (disinformation framed with a clear par-tisan de-legitimizing agenda) to established information (the fact-check). Item measures for the two dependent variables credibility and issue agreement are included in Section A of the Supplementary File. The conceptualization and measurement of the moderators are also described in the Supplementary File.

Findings of Study 1
We conducted OLS-regression models in which we assessed the direct and interaction effect of misinformation and exposure to the fact-check (versus unrefuted misinformation) and the three perceptions on (a) issue agreement with misinformation and (b) the credibility of misinformation. The central expectation is that people with more pronounced antagonist beliefs related to established information and elite sources would (a) perceive misinformation as relatively more credible and established (authentic) information as less credible whilst they (b) resist the corrective information coming from a fact-checking source.
For issue agreement, the findings indicate that people with higher levels of populist attitudes (B = 0.23, . This does not offer support for H1, H2, and H3. However, in line with the general thesis that disenchantment corresponds to a lower tendency to accept established information, we do find that these distrusting beliefs increase the credibility of false information compared to factually accurate information from established information sources.
Turning to our second dependent variable-the credibility of the misinformation article-we find exactly the same: Populist attitudes (B = 0.11, = 0.12, SE = 0.04, p = 0.005), media distrust (B = 0.14, = 0.17, SE = 0.06, p = 0.024), and fake news perceptions (B = 0.18, = 0.21, SE = 0.06, p = 0.004) are all related to a higher credibility of misinformation compared to authentic information. Hence, in support of H1, H2, and H3, disenchanted citizens (i.e., those with populist attitudes or fake news perceptions) find established sources of information less credible. However, there are again no significant two-way interaction effects between exposure to fact-checks and populist attitudes (

Conclusion of Study 1
The central expectation of this study was that media effects and responses to stimuli coming from established sources of information are perceived differently by people who distrust or oppose established information compared to people with more trust in elite sources. We only find partial support for this expectation in the context of corrected misinformation on immigration and crime rates. Ceteris paribus, we found that higher levels of populist attitudes, media distrust, and fake news perceptions resulted in lower levels of credibility and issue agreement with authentic information presented as coming from an established source. Misinformation, however, was more credible for participants that distrusted the established political or media order. Our findings do not support the expectation that fact-checking information responding to misinformation is rejected by citizens with higher levels of populist attitudes, media distrust, and fake news perceptions. This is in line with recent empirical evidence showing that corrective information can work across the board and even persuade strong partisans (e.g., Nyhan et al., 2019). Adding to this literature, we show that different indicators of disenchantment are not causing resistance to fact-checking information, revealing the potential of corrective information among different segments of the population.
Yet, these conclusions have to be interpreted with care. In an experimental and short-term set-up, factchecks may simply be accepted as they give a direct indication and instruction to participants, who are asked to evaluate the information only minutes after reading a correction. In addition, this first study showcased a highly polarized topic-immigration and crime rates-for which people may already have formed strong opinions that are difficult to alter by exposing them to just one or two messages. It is interesting to assess to what extent the findings of this study are transferable to a "most different" topic. For this reason, we will focus on an issue owned by the left-wing in the second study: climate change. In addition, we will use a less strong manipulation of the independent variable. Rather than contrasting misinformation to authentic content and corrected to uncorrected falsehoods, we simply manipulate the type of generic news frame used to cover climate change: a thematic versus episodic frame.

Framing Effects Theory
Just like misinformation is regarded as a problematic trend in society, climate change denialism and resistance toward interventions intended to fight global warming is an alarming development. The media play a role in cultivating support for or opposition to climate change interventions by the framing of these issues (e.g., Feldman & Hart, 2018). Here, we understand framing as patterns of interpretation or organizing ideas that guide recipients' interpretation of events by offering a specific framework for interpretation (e.g., de Vreese, 2005;Entman, 1993;Scheufele, 1999). The term "framing" can be used in two different ways: It can refer to the patterns of interpretation in texts (frames in communication) or individual frames held by recipients (frames in thought; see Chong & Druckman, 2007;Scheufele, 1999). Framing effects can generally be understood as the influence of frames in communication on frames in thought (Chong & Druckman, 2007). Hence, when exposure to patterns of interpretation in a communication text influences people's understanding of a given situation and their attitudes toward the situation, we can speak of framing effects (Druckman, 2001). For this study, we specifically focus on the effects of two generic frames: episodic versus thematic framing. Such frames differ in the emphasis on individual-level cases versus more generic information (Iyengar, 1991). Specifically, episodic frames may focus more on exemplars, individual cases, or personal stories that would exemplify broader issues. Thematic frames, on the other hand, offer more abstract background information and give insights into wider trends and the overall socio-political embedding of issues (e.g., Gross, 2008;Iyengar, 1991). Both of these frames are commonly used in the news reporting of established outlets, which makes it a relevant case to consider in light of this article: Would people who oppose or distrust established media and institutions also be more likely to reject the different emphasis made in these frames, which may explain contradictory findings based on these different treatments found in extant research?
Extant literature suggests that thematic frames promote more society-level responsibility attributions because of their emphasis on society-wide implications and embeddings of issues, whereas episodic frames that showcase individuals and exemplars promote responsibility attributions on the individual level (Iyengar, 1991). However, it should be regarded that the evidence supporting this thesis is not convincing (see also e.g., Springer & Harwood, 2015). A lack of support for differential framing effects in replications of Iyengar's original experiments may be due to confounding factors in the experimental design or the conditionality of effects (Springer & Harwood, 2015). In line with this latter explanation, this study aims to establish whether inconsistent effects of thematic versus episodic framing effects can be explained by people's overall levels of distrust and cynicism toward the established media and climate change. Similar to the first study, we aim to explore whether individual-level indicators of disenchantment and distrust toward the source and content of the message may explain inconsistent effects resulting from experimental research that exposes participants to seemingly authentic news messages. In line with framing effects literature, we generally expect that framing effects are strongest when the frame in communication is more mentally accessible, relevant, and applicable for certain individuals (Chong & Druckman, 2007). Extending this argument, climate change communication focusing on individual cases or statistical information promoting a pro-climate change narrative may be less relevant and mentally accessible for recipients with skeptical beliefs. Hence, such communication does not resonate with their prior beliefs. Skeptical and distrusting participants should therefore be less likely to believe frames emphasizing that climate change is an urgent threat. To measure this mechanism, we focus on two different outcome variables of framing effects: credibility and issue agreement. We specifically expect that conspiracist thinking, distrust, and skepticism related to the issue of climate change make frames on this issue less personally relevant. This consequentially should result in a weaker affinity between frames in communication and frames in mind, which we operationalize as the agreement with the presented frame and the credibility of the communication text. Using slightly different indicators of disenchantment and distrust related to the topic of the second study-climate change communication-we expect that participants with more pronounced levels of conspiracy beliefs (H1), media distrust (H2), and climate skeptic beliefs (H3) report lower levels of credibility and are less likely to agree with media content presented as established information than people with less pronounced cynical or distrusting beliefs.

Data Collection
We rely on an experimental dataset based on an online survey experiment among US participants (N = 738). Data were collected by the international research agency Dynata. As part of the experiment, participants were randomly exposed to either an episodic frame of climate change developments (i.e., focusing on exemplars and individual cases of a community severely hit by the consequences of global warming) or a thematic frame (i.e., focusing on statistics and contextual base rate information to depict the problematic trend of global warming's consequences). The different conditions were kept as similar as possible regarding all other factors. The two dependent variables and moderators of Study 2 are included and explained in Section B of the Supplementary File.

Results of Study 2
First of all, we assessed whether the effects of thematic versus episodic frames on agreement with the news media's message would be contingent upon conspiracy beliefs (H1) or climate change denialism (H3) and media distrust (H2). We expected weaker effects and a lower credibility rating among participants with more cynical or distrusting views, which we measured as lower levels of agreement and perceived accuracy/trustworthiness. In line with our expectations, we found that the more participants supported denialism and conspiracies related to climate change the more they rejected the arguments of the news message by indicating lower levels of agreement with the core statements made in the message (B = −0.21, = −0.21, SE = 0.03, p < 0.001). This is in line with H1 and H3. In addition, higher levels of mainstream media trust corresponded to more acceptance of the message (B = 0.42, = 0.47, SE = 0.03, p < 0.001). In support of H2, the more people distrusted the established media the more they rejected the arguments of the message. Contradicting the tentative expectation that this pattern would be reversed for alternative media trust, we find similar results for alternative media distrust, albeit with smaller effect sizes (B = 0.15, = 0.16, SE = 0.04, p < 0.001).
If we focus on the interaction effect between exposure to a thematic (versus episodic) frame and climate change denialism and conspiracies, we do not find a significant effect (B = 0.13, = 0.13, SE = 0.07, p = 0.073). Yet, we do see that higher levels of distrust in established information channels correspond to higher levels of message rejection than lower levels of distrust (B = −0.14, = −0.17, SE = 0.07, p = 0.047). This supports H2. Although the effects are reversed for trust in alternative media, the interaction effect between trust in alternative media and exposure to thematic frames is not significant (B = 0.14, = 0.15, SE = 0.09, p = 0.106).
Turning to our second dependent variableperceived credibility of the news item/trustworthinesswe see a strong relationship between support for conspiracies/denialism and the perceived trustworthiness/credibility of the news item (B = −0.30, = −0.37, SE = 0.03, p < 0.001). This means that, in support of H1 and H3, the more participants perceive that climate change is a hoax or non-issue the more likely they perceive the news message as inaccurate, deceptive, or even fake news. We also find support for H2: The more participants distrust the mainstream media as a source of information, the less they perceive the news message as authentic or accurate (B = −0.33, = −0.45, SE = 0.03, p < 0.001). We do not find such a relationship for trust in alternative sources of information (B = 0.06, = 0.08, SE = 0.03, p = 0.057).
The findings do not offer support for significant interaction effects between exposure to thematic versus episodic frames and climate change denialism/ conspiracies (B = 0.05, = 0.07, SE = 0.06, p = 0.354) or media (dis)trust (B = 0.09, = 0.13, SE = 0.06, p = 0.111). Contrary to our expectations, existing levels of cynicism related to the content (denialism/conspiracies) or source of the message (trust in established media) did not moderate the effects of differential framing conditions on the perceived accuracy or trustworthiness of the message.

Conclusion of Study 2
We found support for this article's thesis that existing levels of distrust and cynicism related to the source and the content of information corresponds to lower levels of perceived credibility and agreement. This means that stimuli presented as mainstream or established news may be rejected by participants who do not support the perspective of the message or distrust the source of information it allegedly comes from.
These patterns are not consistently found when we also take the type of manipulation into account. In the context of this study, we do not find that the effects of exposure to thematic versus episodic framing are different for participants with a tendency to oppose the mainstream media or the dominant consensus framing of climate change communication. This suggests indifference among segments of the audience that do not support the message's arguments or source: They already show a stronger tendency to find the message incredible, inaccurate, and untrustworthy, which may also indicate that they are not sensitive to nuances in the message's framing.

Conclusions
The current information ecology has been connected to worrisome developments such as mis-and disinformation, polarization, and increasing distrust in established media and information sources (e.g., Bennett & Livingston, 2018;Van Aelst et al., 2017;Waisbord, 2018). Arguably, omnipresent concerns about false information and accusations of fake news in the political domain (e.g., Egelhofer & Lecheler, 2019) have contributed to eroding levels of trust in sources of information that are assumed to be impartial, neutral, and independent. Against this backdrop, media effect studies conducted in communication science and adjacent fields that expose people to (manipulated) information coming from allegedly neutral sources may face an important challenge: These sources may not be regarded as trustworthy by all participants, which may result in unanticipated findings and contradictory conclusions.
To better understand how increasing levels of distrust and disenchantment may explain unanticipated findings in media effect studies, we relied on two experimental studies in which we mapped the biasing impact of disenchantment on credibility and agreement with established information sources. Considering that populist attitudes have been associated with distrust and avoidance of established information (e.g., Fawzi, 2019;Müller & Schulz, 2021), and taking into account that fake news perceptions may lead to the rejection of established information (e.g., Hameleers et al., 2021), we zoomed in on (a) populist attitudes, (b) media distrust, and (c) fake news perceptions or related conspiracy beliefs as attitudinal filters that can lead to the rejection of established information.
Based on Study 1-an experiment investigating the effects of misinformation and fact-checking-we find mixed support for our general expectation. In line with previous findings, we show that false information is relatively more persuasive and credible for distrusting news users (Zimmermann & Kohring, 2020). Yet, we do not find that corrective information is processed differently by disenchanted segments of the audience: Fact-checks are equally effective for people with more or less pronounced populist attitudes, fake news perceptions, and media (dis)trust. This confirms extant literature demonstrating that fact-checks can correct misperceptions across the board (e.g., Nyhan et al., 2019) but contradicts research pointing to strong conditional effects of such corrections based on confirmation biases (e.g., Thorson, 2016). Based on these findings, we show that disenchanted segments of the population may not clearly distinguish between authentic and deceptive information and that they are less resilient to misinformation. However, they do not reject corrections coming from established sources of information and are thus open to communication that challenges their distrusting views on the media and society. This is in line with earlier research demonstrating that fact-checks have an effect in lowering misperceptions, even among partisan audiences (e.g., Wood & Porter, 2018).
In a different context of climate change information, Study 2 on the effects of thematic versus episodic framing confirms these findings: We find that conspiracy beliefs, climate change denialism, and media distrust correspond to lower levels of credibility and agreement with established information sources, which indicates that these indicators of disenchantment correspond to a higher likelihood for citizens to resist information presented as published by mainstream news sources. However, people supporting conspiracies or distrusting the media did not respond differently to the stimuli than more trusting news users, which indicates that disenchantment may not result in unanticipated findings in the context of an emphasis framing study.
How can these findings enlighten confusion about media effects in an era of factual relativism? First of all, we should not assume universal levels of credibility, trustworthiness, and message acceptance for media effect studies that use (representative) panels of respondents. When exposing people to seemingly neutral "news" stories or messages, it is important to consider that distrusting and disenchanted segments of the population may find them less trustworthy and neutral than others. Controlling for this factor or acknowledging these individual differences can help to explain why (mis)information is found credible by some news users but rejected or counter-argued by others. Disenchantment may especially be an important factor to consider when comparing information coming from different sources (i.e., a mainstream versus alternative news source), or when assessing differences between authentic and factually correct information versus inauthentic and incorrect information.
On a more optimistic note, varying the frame of communication does not result in backfire effects of communication among distrusting segments of the audience. Hence, although trends toward misinformation, fragmentation, and eroding trust in established information are problematic for society at large, these perceptions may not lead to strong reactance when it comes to framing effects. At the very least, null effects of framing manipulations found in this study were not driven by existing levels of disenchantment among participants. Although it reaches beyond the scope of the empirical data presented here, we can also interpret these findings in light of overall fatigue and lack of systematic processing in media effect studies: When participants are forcefully exposed to information that they may not consume in real life, we may fail to accurately simulate the conditions under which people process information in real life. Another way to enlighten confusion, then, is to conduct media effect studies in more realistic information settings, taking into account people's biases, motivation, and personalized selective exposure environments.
The current study and the cases explored here come with limitations. First, we only looked at two cases in partisan US settings, and it remains to be tested how these findings are transferable to other settings, such as multiparty systems in which partisan cleavages and media distrust are less prevalent. Second, we operationalized disenchantment mostly in the context of factual relativism and a right-wing populist way of rejecting information from established institutions (e.g., Fawzi, 2019;Müller & Schulz, 2021;Schulz et al., 2020). There may be different dimensions of disenchantment and distrust that we did not explore in this article, but which would also be relevant to consider when explaining unanticipated findings in media effect studies. Third, our studies did not explicate the source of (established) information, but only very generally mentioned that participants were shown information published recently in the news (the source was not mentioned). It may be the case that using more explicit source cues from actual news media channels may cause more resistance among disenchanted and distrusting audience segments. However, we did not find this for the fact-check manipulation, as distrusting and trusting participants were equally likely to accept corrective information. Despite these limitations, this study illustrates how problematic trends related to eroding trust and misinformation in digital democracies may spill over to media effect studies that may operate under an assumption that is no longer valid in a post-truth world: That all media users are equally likely to accept the authenticity and trustworthiness of seemingly neutral information.