Media Use and Societal Perceptions: The Dual Role of Media Trust

How citizens’ perceptions of societal problems are shaped by media use has been a critical question in media effects research for decades. This study addresses a specific puzzle concerning media effects in contemporary fragmented media environments: the dual role of media trust as both (a) an antecedent variable guiding news selection and (b) a modera‐ tor variable conditioning the effects of news use on perceptions of societal problems. Building upon the differential sus‐ ceptibility to media effects model, we analyze the role of media trust for citizens’ orientation towards mainstream and alternative news media—and how such usage influences perceptions of two major societal issues: health care and school. Findings from a four‐wave panel survey conducted in Sweden suggest that public service and alternative news use mat‐ ter for citizens’ perceptions of societal problems and that media trust influences news choices and may, partly, condition media effects.


Introduction
Over the last decades, two closely intertwined changes affecting democracies worldwide are digitalization and the transformation from low-to high-choice media environments. Among other things, these changes have resulted in a greater abundance of information and different types of media than ever, including political alternative media, and increasing selectivity in citizens' media use (Van Aelst et al., 2017).
Two potential outcomes of these processes are an increasing divergence in worldviews and societal perceptions as well as a growing prevalence of misperceptions (O'Connor & Weatherall, 2019;Strömbäck et al., 2022;Vosoughi et al., 2018). Numerous studies also show that societal perceptions partly can be explained by citizens' media use and thus should be conceptualized as a media effect (Damstra et al., 2021;Meltzer & Schemer, 2021;Ridout et al., 2008). Societal (mis)perceptions have furthermore been linked to the use of political alternative media (Garrett et al., 2016;Hmielowski et al., 2014;Hmielowski et al., 2020). This suggests that the use of political alternative media may influence societal perceptions more broadly, but also that the differential use of mainstream and political alternative media may lead to increasing perception gaps.
At the same time, the effects of political alternative media versus mainstream media use on societal perceptions should depend on both context and individual-level factors. One potentially important individual-level factor is trust in mainstream news media (which we will refer to as general media trust). To begin with, the greater media choice there is, the more selective people have to be, and the more selective people have to be, the more important media trust should become. Previous research has also found that there is a relationship between media trust and selective media use (Fletcher & Park, 2017;Kalogeropoulos et al., 2019;Strömbäck et al., 2020;Tsfati & Cappella, 2003). Beyond influencing media use, media trust may also moderate the impact of media use on societal perceptions, as credible sources are generally more persuasive than less credible ones (Eagly & Chaiken, 1993;Pornpitakpan, 2004) and as studies have found media trust to moderate other media effects (Damstra et al., 2021;Miller & Krosnick, 2000).
That said, there is still confusion surrounding the role of general media trust as a factor in the media effects process: Is media trust operating as a predictor of media use, as a moderator of media effects-or both? In addition, most research has been done in the US, which is an atypical case considering its media and political system, high degree of political polarization, and low level of media trust (e.g., Hallin & Mancini, 2004;Hanitzsch et al., 2018;Hopkins & Sides, 2015). The generalizability of findings is thus unclear. Furthermore, there is only scant research on media trust using longitudinal data and on whether general media trust influences the effects of media use on societal perceptions.
Against this background, the purpose of this article is to investigate the dual role of general media trust when explaining the use of mainstream and political alternative media and the effects thereof on societal perceptions. Theoretically, we will depart from the differential susceptibility to media effects model (DSMM; Valkenburg & Peter, 2013) and conceptualize media trust as both a predictor of media use and a moderator conditioning the effects of media use on societal perceptions. Empirically, we study citizens' perceptions regarding two issues-health care and school-using a multi-wave panel survey collected in Sweden.

Theoretical Review
Over the last decades, it has become increasingly established that there are no such things as universal media effects (Slater, 2015;Valkenburg & Peter, 2013). Instead, all types of media effects should be understood as conditional, meaning that they depend on both systemic factors, such as the media system and the supply and character of media content (Castro et al., 2021;Shehata & Strömbäck, 2011), and individual-level differences such as gender or political interest Slater, 2015). With respect to individual-level factors, they may influence not only people's selective media use, but also the direction and/or strength of media effects (moderators) and provide the causal link explaining media effects (mediators; Baron & Kenny, 1986).
One model taking this into account is the DSMM (Valkenburg & Peter, 2013). According to this model, differential susceptibility variables usually function as both predictors and moderators. These variables can in turn be dispositional (i.e., they predispose the selective use of and responsiveness to media), developmental (i.e., the selective use of and responsiveness to media are due to cognitive, emotional, and social development), or social (i.e., social context-factors that influence the selective use of and responsiveness to media). In line with the reinforcing spirals model (Slater, 2015;Slater et al., 2020), it also proposes that media effects are transactional in the sense that media might have an effect on certain attitudes, beliefs, or behaviors, which in turn might have effects on subsequent media use. This may hold true in particular in high-choice media environments, as these provide greater opportunity structures for selective exposure (Prior, 2007;Skovsgaard et al., 2016).

The Transformation of Media Environments and Rise of Political Alternative Media
A key aspect of digitalization and the transformation from low-choice to high-choice media environments is the increasing prevalence of what is variously called political alternative media, partisan media, or ideological media, although the supply and prominence of such media vary across countries (Heft et al., 2020). What these terms have in common is that they refer to media that are guided by political rather than journalistic values and norms (Benkler et al., 2018;Holt et al., 2019), which sets them apart from mainstream media. According to Holt (2018, p. 51), political alternative media are typically "created and run in opposition to what is perceived as a dominant discourse in traditional media." In contrast to mainstream news media, which display great similarities across media in terms of how they operate and their routines, norms, and values (Cook, 2005), there are great differences across political alternative media in terms of, among other things, their political leaning, their degree of alternativeness, how closely linked they are to political parties or other movements, and ultimately their content (Benkler et al., 2018;Holt et al., 2019;Müller & Freudenthaler, 2022).
Since a common denominator of political alternative media is that they are guided by political values and norms, their coverage can be expected to differ from that of mainstream news media in terms of what issues they cover and how they frame issues or events (Benkler et al., 2018;Holt, 2018;Müller & Freudenthaler, 2022). More specifically, research suggests that political alternative media compete by seeking to provide information that confirms the worldviews and attitudes of their targeted audiences. That may hold in particular for right-wing alternative media, where research suggests they display a greater degree of alternativeness and hostility toward mainstream media than left-wing alternative media (Benkler et al., 2018;Figenschou & Ihlebaek, 2019;Ihlebaek & Nygaard, 2021). This implies that the main effect of political alternative media might not be that they influence people's attitudes as much as their perceptions of different issues and events-including their societal perceptions. However, it could also depend on levels of media trust among different groups.

Media Trust as a Predictor of Media Use
Following the DSMM, there are theoretical reasons to assume that general media trust is one key dispositional differential susceptibility variable that influences media use. This holds in particular in high-choice media environments where people are not constrained to using mainstream media and have greater opportunities than ever to seek out political alternative media with content that is attitude-congruent (Strömbäck et al., 2022).
Conceptually, media trust broadly refers to a relationship where people expect that interaction with the media will lead to gains rather than losses and that media will perform in a satisfactory manner (Fawzi et al., 2021;Strömbäck et al., 2020). While media trust can be located at different levels of analysis, in this study we are focusing on general trust in mainstream news media.
Since media trust involves a relationship where people expect some kind of gain, one reason why media trust should function as a predictor of media use is simply that it is most rational for people to select media that they trust (Tsfati & Cappella, 2003), although their media use is also constrained by structural, habitual, and situational factors (Webster, 2014). A second reason is that general media trust may function as a heuristic when people face a choice between using different media (Webster, 2014). In addition, a key motivation for media use is to get informed and satisfy one's cognitive needs (Rubin, 2009;Ruggiero, 2000), and that presumes that people trust the media. Consequently, a number of studies have shown that there is a link between media trust and media use (Fletcher & Park, 2017;Kalogeropoulos et al., 2019;Ladd, 2012;Tsfati & Cappella, 2003).
More specifically, in contexts where people can choose between mainstream and political alternative media, a key reason why people replace or complement the use of mainstream news media with political alternative media may be that they do not trust mainstream news media or perceive these as hostile (Ladd, 2012;Perloff, 2015;Tsfati & Cappella, 2003). This, in turn, might be explained by the fact that people tend to prefer information and information sources that confirm their already held beliefs and attitudes (Kunda, 1990). Studies also show that counter-attitudinal news reporting is likely to induce hostile media perceptions (Arceneaux et al., 2012), and such reporting is more likely in mainstream media than in political alternative media that compete by reaffirming their audiences' political beliefs and attitudes (Benkler et al., 2018). Numerous studies, albeit predominantly from the US, have also found evidence for political selective exposure, meaning that people seek out media that can be expected to provide attitude-congruent information (Arceneaux & Johnson, 2013;Dahlgren et al., 2019;Garrett et al., 2013;Stroud, 2011). This may hold in particular for those leaning to the right ideologically or sympathizing with right-wing populist parties, as they generally trust mainstream media less than others (Andersson, 2021;Fawzi, 2019;Gottfried et al., 2019;Strömbäck & Karlsson, 2017). This may in turn moderate the effects of media use.

Media Trust as Moderator of Media Effects
In line with the DSMM, differential susceptibility variables such as general media trust can be expected to function not only as predictors but also as moderators. Most importantly, how people interpret and process whatever information they are exposed to depends on the extent to which they find the source trustworthy and credible (Eagly & Chaiken, 1993;Ladd, 2012;Pornpitakpan, 2004). Furthermore, both trustworthiness and credibility are closely intertwined with trust (Kohring & Matthes, 2007;Metzger et al., 2003;Strömbäck et al., 2020). Thus, whereas those who trust a certain media type are likely to accept the information provided, those who distrust them are more likely to engage in counter-arguing or discount the information altogether (Kunda, 1990;Ladd, 2012;Lodge & Taber, 2013).
The motivations for taking part in different types of media are thus likely to differ depending on whether people trust them or not (Ladd, 2012). Whereas those who trust mainstream news media are likely to use them to get informed and satisfy their surveillance needs (Rubin, 2009), those who distrust them are more likely to use them out of curiosity, to find counterarguments, or because they do not feel they have a choice. For example, before the rise of political alternative media, those who distrusted mainstream news media did not have much of a choice if they wanted to know what was going on in society, but in contemporary media environments, they can find political alternative media that they may trust more. That said, those with a high need for cognition have however been found to consume media even if they distrust them (Tsfati & Cappella, 2005). In contrast, those who use political alternative media are more likely to do it not only to get informed, but because they anticipate that they will get their worldviews and attitudes confirmed (Benkler et al., 2018).
In line with this, previous research has found that "those who distrust the media update their beliefs [about societal conditions] less in response to events, instead relying more on their partisanship," and that "(t)hose who distrust the institutional media resist new information from the mainstream media [and are] more likely to utilize alternative, partisan media outlets" (Ladd, 2012, pp. 149, 195). Following Hall (1980), the moderating role of media trust may thus be described as a matter of how those who trust versus distrust the media decode the media content, where those who distrust mainstream media are more likely to engage in a negotiated or oppositional interpretations of the media content. Through this vein, trust in mainstream media may moderate the effects of media use on societal perceptions.

Media Effects on Societal Perceptions
Broadly, societal perceptions can be defined as beliefs about the current state or development of societal affairs, for example, the state or the development of the national economy or crime. Perceptions thus involve (more or less correct) knowledge and refer to how things are rather than how they ought to be (Ajzen, 2005;Eagly & Chaiken, 1993). This separates perceptions from attitudes, that per definition involve an evaluative component.
Although societal perceptions and attitudes are conceptually distinct, they are closely intertwined. Consequently, research has shown that societal perceptions matter greatly to people's attitudes and behaviors. For example, studies show that perceptions of the national economy influence voting (Lewis-Beck & Stegmaier, 2007), that perceptions of the size of the immigrant population influence opposition to immigration (Sides & Citrin, 2007), and that perceptions of crime influence feelings of fear (Ambrey et al., 2014). The underlying reason can be traced back to Lippman (1997), who argued that "the pictures in our heads" may matter more than reality per se, since the "pictures in our heads" are what we ultimately have access to.
Problematic in this context are signs that misperceptions have become more common and that there is an increasing divergence in societal perceptions (Kavanagh & Rich, 2018;O'Connor & Weatherall, 2019). For example, clear differences in perceptions have been found with respect to issues such as whether there is a process of anthropogenic global warming (Krosnick & MacInnis, 2020), whether there were weapons of mass destruction in Iraq before the invasion in 2003 (Gaines et al., 2007), and the origins of the Coronavirus (Douglas, 2021).
Such perception gaps are problematic not only because they run counter to the notion that a wellfunctioning democracy requires reasonably informed citizens (Dahl, 1998). They may also cripple meaningful political debates, as such require common ground and a large body of shared facts. As noted by Rosenfeld (2019, pp. 173-174), "democratic debate is premised from the start on every opinion being informed by some shared body of facts.'' As noted above, a significant body of research simultaneously suggests that (mis)perceptions and perception gaps at least partly can be explained by media use, and that political alternative media often trade in politically framed (mis)information (Benkler et al., 2018;Garrett et al., 2016;Glogger & Shehata, 2022;Hmielowski et al., 2014). Most research has however been done in the US, implying that the generalizability of findings is unclear. The same holds for the mechanisms by which the effects occur and the role of media trust.

Hypotheses
Based on the review above and the DSMM in particular, our general expectation is that general media trust will both predict and moderate the effects of using mainstream news media versus political alternative media on societal perceptions. Thus, while we expect that the use of mainstream and alternative media has effects on perceptions of societal problems, general media trust is likely to both guide news choices and condition the relationship between these news choices and societal perceptions over time. Hence, our hypotheses are: H1: General media trust both predicts use of mainstream news and political alternative media (H1a) and moderates the effects of these news sources on perceptions of societal problems (H1b).
H2: Use of mainstream news (H2a), left-wing (H2b), and right-wing (H2c) political alternative media have differential effects on citizens' perceptions of societal problems.
In addition, it might be the case that general media trust is associated with citizens' perceptions of societal problems. To explore this, we ask: RQ: What is the relationship between general trust and citizens' perceptions of societal problems?

The Case, Data, and Methodology
To investigate the hypotheses, this study focuses on citizens' perceptions concerning two societal issues: health care and school. These issues represent two critical areas of the welfare state which almost every citizen has personal experience of. At the same time, they are both contested politically and generally salient on the political, media, and public agendas, with ongoing framing battles over how to perceive current conditions and trends. Thus, we regard these as two similar cases in terms of their basic issue characteristics. The key question then is how the use of mainstream news media on the one hand, and political alternative media on the other, is related to such perceptions.
To explore the dual role of general media trust in media effects on societal perceptions, we use data from a four-wave panel survey conducted in Sweden during 2020-2021. The data collection was administered by the Laboratory of Opinion Research (LORE) at the University of Gothenburg. A probability-recruited sample of 3,327 web survey participants aged 18-80 was invited to take part in the study. The sample was pre-stratified on gender, age, and education. The first wave was fielded on March 17, 2020, the second on October 26, 2020, the third on April 19, 2021, and the fourth on October 25, 2021. The net participation rate was 65% (W1), 57.6% (W2), 55.3% (W3), and 53.9% (W4). The sample is broadly representative in term of gender (50% female), age (13% < 30 years, 15% 30-39, 18% 40-49, 17% 50-59, 20% 60-69, and 17% > 70), and education (23% with more than three years at university).

Societal Perceptions
This study focuses on perceptions of societal conditions relating to Swedish health care and schools. For both issue domains, we use a battery of three items following the survey question:

General Trust in Mainstream News Media (General Media Trust)
Four items were used to tap general trust in mainstream news media following the survey question "There are different views in society on news coverage in Swedish media. To what extent do you agree with the following statements? The traditional news media in Sweden…." The statements were: (a) Don't tell the truth about important societal issues; (b) let all important voices be heard in the discussion; (c) provide a one-sided perspective on important issues; and (d) provide the best and most reliable information about politics and society. Response scales ranged from 1 (completely disagree) to 7 (completely agree), and items (a) and (c) were reversed before averaged into a media trust index (W1 Cronbach's = 0.85, M = 4.32, SD = 1.52). We acknowledge that this measure of media trust deviates from more common operationalizations (see Strömbäck et al., 2020, for a review), but it has previously been found appropriate  and was the measure that the panel survey included.

News Media Use
The study distinguishes between three types of news use: mainstream news media, left-wing alternative media, and right-wing alternative media. With respect to mainstream news media, we focus on public service media. The rationale is that public service media can be described as both the most mainstream and most salient of mainstream news media. Public service news consumption was measured as the number of days in the past month the respondents had followed news on Sveriges Radio Ekot (SR), Aktuellt (SVT) and Rapport (SVT)-which correspond to the main public service radio and television news programs in Sweden (W1 Cronbach's = 0.71, M = 3.66, SD = 1.48; W2 = 0.68, M = 3.48, SD = 1.45; W3 = 0.66, M = 3.47, SD = 1.43; W4 = 0.67, M = 3.47, SD = 1.44). The use of alternative left-wing and right-wing media was measured similarly using a list of 13 online outlets (seven left-wing and six right-wing sources). Given the relatively low frequency of use of these outlets, our final measures are based on the one outlet that a respondent used most frequently. Response categories ranged from 1 (daily) to 6 (never) but were reversed before combined into indices.

Data Analysis and Control Variables
To address our hypotheses, we estimate structural equation models (SEM) where general media trust predicts our three forms of news media use, which, in turn, predict societal perceptions. To capture change over time, we estimate three models per issue with perceptions at W2, W3, and W4 serving as the final outcome variable, controlling for lagged perceptions from the previous panel wave (W1, W2, and W3). Since the survey items for news use are retrospective (usage in the past month), the models include news use measures from the same panel wave as the outcome variable (instantaneous effects). Media trust from W1 is used in all models. While the perceptions equations control for the lagged dependent variables (t −1 ), all news use equations control for gender, age, political interest, and ideology. To test the moderating role of media trust, we furthermore use multiple group comparison of coefficients across three levels of media trust-low, medium, and high-by dividing the sample into three approximately equally sized groups using the media trust scale. This approach allows us to simultaneously test the conditional effects of three forms of news use across levels of media trust. The analysis thus addresses media trust as an antecedent (H1a) and as a moderator (H1b) variable. Bivariate correlations between all key variables are available in Table A1 in the Supplementary Material.

Findings
Before testing our hypotheses, we will address our research question about the relationship between general media trust and citizens' perceptions of societal problems (RQ). Towards that end, Figure 1 presents descriptive trends regarding citizens' perceptions of health care and school, for three levels of general media trust. Higher values represent a more positive perception of current conditions and developments.
A few things are worth noting. First, Swedish citizens appear to have a more negative view of the performance of the school system than the health care system. This difference is evident already in W1 but the "gap" remains and even increases over time. Second, media trust is related to perceptions. Citizens with higher trust in the media tend to see societal conditions in a more positive light. This is the case for both health care and school perceptions. This is also captured by the cross-sectional correlations between general media trust and health care perceptions across the waves (W1: Pearson's r = 0.33, p < 0.001; W2: r = 0.36, p < 0.001; W3: r = 0.33, p < 0.001; W4: r = 0.30, p < 0.001) on the one hand, and school perceptions (W1: Pearson's r = 0.40, p < 0.001; W2: r = 0.39, p < 0.001; W3: r = 0.40, p < 0.001; W4: r = 0.37, p < 0.001) on the other. Third, although perceptions appear rather stable over time, there are some changes as well. While perceptions of Swedish health care become somewhat more positive following W1-increasing from a mean value of 3.75 in W1, through 3.97 in W2, to 4.02 in W3, before becoming more negative again-an opposite trend emerges for school perceptions, displaying a gradual increase in negative perceptions from 3.62 in W1 to 3.30 in W4. Both these general trends are also statistically significant compared to baseline values from W1 (the time trends were tested using wave dummy variables in a random effects panel model, with W1 operating as the category of reference). Next, we address our hypotheses concerning the dual role of media trust as (a) an antecedent factor explaining news media use and (b) a moderator variable conditioning the relationship between news media use and societal perceptions. Structural equation models and multiple group comparisons are used to test these hypotheses. Figure 2 presents results from the first unconditional model focusing on health care perceptions and provides an overall picture of the key relationships of interest. Each arrow shows the estimated effects at three occasions separately-W2, W3, and W4. For instance, general media trust has a positive effect on use of public service news in W2 (b = 0.11, p < 0.001), W3 (b = 0.09, p < 0.001), and W4 (b = 0.07, p < 0.01), controlling for gender, age, political interest, and ideology. Thus, citizens with higher general media trust are more likely to use public service media. The opposite is true for right-wing alternative media, which displays a consistent negative effect (W2: b = −0.24, p < 0.001; W3: b = −0.25, p < 0.001; W4: b = −0.23, p < 0.001). There are however no relationships between media trust and the use of left-wing alternative media. These findings lend support to H1a.
Turning to the relationship between news media use and health care perceptions, we see that use of public service has a positive effect (W2: b = 0.10, p < 0.001; W3: b = 0.09, p < 0.001; W4: b = 0.02, p > 0.05), controlling for lagged health care perceptions (not illustrated in Figure 2). This means that higher use of public service news is related to an increase in positive health care perceptions over time. Use of right-wing alternative media, however, increases negative perceptions (W2: b = −0.05, p < 0.05; W3: b = −0.08, p < 0.001; W4: b = −0.06, p < 0.01). Use of left-wing alternative media displays no relationship with health care perceptions (see Figure A1 in Supplementary Material for graphical display of these unconditional effects based on OLS models).
While the findings in Figure 2 suggest general effects of some forms of news media use on health care perceptions, they do not address H1b concerning the moderating role of media trust. Therefore, Table 1 presents findings from a multiple group comparison of the models illustrated in Figure 2. Again, three models were estimated to represent the different time points. Table 1 displays coefficients across three levels of media trust-low, medium, and high. Bolded coefficients highlight effects that are statistically different between the three groups. The only difference found here relates to W1-W2 estimates for use of right-wing alternative media. These findings suggest that the effect is significantly stronger among citizens with medium-level trust in traditional news media (see Figure A2 in the Supplementary Material for graphical display of the conditional marginal effects across different levels of media trust). Figure 3 presents results relating to school perceptions. The main findings concerning general media trust as an antecedent of news media use are no different from the previous model: Trust is positively related to the use of public service news, but negatively related to right-wing alternative media (H1a) and unrelated to left-wing alternative media-controlling for gender, age, political interest, and ideology. With respect to school perceptions, the use of right-wing alternative media is the only consistent predictor of changes in school perceptions. More specifically, the results show that more frequent use of these is related to more negative perceptions over time (W2: b = −0.10, p < 0.001; W3: b = −0.08, p < 0.001; W4: b = −0.08, p < 0.001). These results lend partial support for H1b and support H2c. Table 2 presents findings from the corresponding multiple group analyses focusing on H1b and the conditional relationship between news media use and school perceptions. Although use of right-wing alternative media appears to significantly increase negative school perceptions only among the low-trusting group (b = −0.09, p < 0.01; b = −0.10, p < 0.001; b = −0.06, p < 0.05), the differences across groups are not significant. The only significant group difference relates to public service news, which displays a stronger negative impact among medium-level trustors in wave 3 (b = −0.08, p < 0.01). (See Figure A3 in supplementary Notes: Media trust is divided into three approximately equally sized groups of low trust (N = 748), medium trust (N = 632), and high trust (N = 679); bolded coefficients represent effects that are statistically significant across trust groups; * = p < 0.05; ** = p < 0.01; *** = p < 0.001.  Notes: Bolded coefficients represent effects that are statistically significant across trust groups; * = p < 0.05; ** = p < 0.01; *** = p < 0.001. material for graphical display of the conditional marginal effects across different levels of media trust).

Conclusion
The purpose of this study was to investigate the dual role of general media trust when explaining the use of mainstream versus political alternative media and the effects on perceptions of the development of Swedish health care and Swedish schools. Summing up, one key takeaway is that media trust clearly predicts the use of mainstream versus right-wing-but not left-wing-alternative media. Importantly, this holds even when controlling for ideology. Across waves, general media trust positively predicts the use of public service media and negatively predicts the use of right-wing alternative media, while there are basically no relationships between media trust and the use of left-wing alternative media. These findings lend general support to H1a on the role of media trust as an antecedent variable explaining differential news use. A second takeaway is that the use of mainstream versus right-wing-but again, not left-wing-alternative media have differential effects on societal percep-tions, lending overall support to H2. More specifically, right-wing alternative media use is consistently related to more negative perceptions of Swedish health care and school, while the opposite holds true for use of public service media with respect to health care perceptions. Taken together, these findings suggest that right-wing alternative media display a greater alternativeness than left-wing alternative media, not only in the US as suggested by previous research (Benkler et al., 2018), but also in Sweden.
With respect to the potential dual role of media trust as a factor behind media effects, our findings suggest that trust may be more important as an antecedent guiding news choices than as a moderator of media effects on societal perceptions. Most findings supported universal, rather than conditional, effects across trust groupslending limited support to H1b. This is a tentative conclusion, however. While the findings point in this direction our analyses cannot finally determine the precise causal relationships at work. The bivariate correlations between media trust and news use vary from weak (left-wing alternative media), to moderate (right-wing alternative media), which together with a lagged dependent variable and relatively few users of alternative media, reduces the information available to fully test the conditional effect of media trust. More research is therefore needed, including experimental designs, to disentangle the dual role of media trust. Although the focus of this study has been on media trust as an antecedent and moderator variable, other aspects of the DSMM could also be addressed, such as exploring mediating and transactional effects of media trust in greater detail.
Over time, however, the end result is likely an increase in perception gaps across groups depending on their general media trust and their use of mainstream versus right-wing alternative media. In light of this, there are strong theoretical as well as societal reasons to further disentangle the dual role of general media trust when explaining media use and the effects thereof on societal perceptions. Future research is thus encouraged to investigate this dual role in the context of other issues and other contexts.