1 Introduction

In our current high-choice media ecology, the concept of media repertoires – the patterns of using various types of media found in single individuals (Hasebrink and Popp, 2006; Webster, 2014; Mourão, Thorson, Chen, and Tham, 2018) – has become an important lens through which media use and news consumption are studied. But while this work has generated large insights in the diversification (e. g., Bos, Kruikemeier, and De Vreese, 2016) as well as the social stratification of media use (e. g., Lindell, 2018), it has at least two gaps. First, with the exception of political participation, there has been relatively little research on how media repertoires relate to outcome variables (but see Castro et al., 2021). This paper aims to make advancement in this regard by focusing on institutional trust.

Trust in social institutions such as politics, journalism, and science is in decline in most Western societies, which is often considered detrimental to the cohesion of society (Armingeon and Guthmann, 2014; Putnam, 2000; Norris, 2000; Twenge, Campbell, and Carter, 2014). Institutions comprise mechanisms of social order that rely on consensus, shared understandings of how individuals should behave, and ultimately on a form of belief among citizens that these institutions are functioning appropriately. Media have been mentioned as possible factors influencing trust (Cappella and Jamieson, 1997; Putnam, 2000; Norris, 2000), as the reception of media content affects the way audiences perceive the trustworthiness of institutions and their agents – for example in the political domain (Strömback, Djerf-Pierre, and Shehata, 2016), or regarding science (Hmielowski, Feldman, Myers, Leiserowitz, and Maibach, 2014). At the same time, media are themselves institutions that seem to suffer from loss in trust (Tsfati and Ariely, 2014; Yamamoto, Lee, and Ran, 2016). How trust and media repertoires are related has not yet been studied.

Second, despite preliminary indications that socialization is an important factor in shaping media and news repertoires (Lindell and Sartoretto, 2018), this has not been further examined. In this paper, the role of parents and the social network is highlighted. Whereas parental socialization has traditionally been important in reproducing lifestyle practices (Notten, Kraaykamp, and Konig, 2012), the increasing power of social media and the generational gaps in its usage (Andersen, 2021) may give leverage to social networks. Comparing the influence of these social contexts provides us with evidence on whether (certain types of) media use can be stimulated in a time where the legitimacy of media agents is decreasing (Carlson, 2017). From a broader societal perspective it is important to examine how social inequalities in access to and actual use of media and communication are developing (Lindell, 2018), to ensure that citizens remain informed on relevant social issues.

Concretely, the paper examines how media repertoires are associated with institutional trust, and adds parental socialization and social network as explanatory variables of both membership of repertoires and institutional trust. To do so, a three-step approach is applied: first, mapping repertoires via Latent Class Analysis; second, conducting explanatory analyses of repertoire membership with parental and social network indicators as predictors; and, third, doing distal outcome analyses of institutional trust in which repertoire membership is a predictor along with other explanatory variables. This analytical strategy enables us to estimate clusters in media usage, and subsequently relate these clusters to external variables (Vermunt and Magidson, 2016). The data concern survey data from a representative sample in the Netherlands.

2 Media usage versus political and institutional trust

Since media are among the main providers of information on how institutions, and the actors therein, perform, the relationship between media use and political (institutional) trust has been studied elaborately in the past decades. Some scholars explicitly point at the negative role of media with regard to trust in politics: Too much negativity in the media may lead to cynicism and lower levels of trust (Cappella and Jamieson, 1997; Mutz and Reeves, 2005). This is the “media malaise” hypothesis, which essentially predicts that larger exposure to media will lead to lower levels of political trust. Others have emphasized that media consumption and political trust are reciprocal: Citizens who look for political information in the media tend to be more socially engaged and politically interested at the outset, but will cultivate this further while using the media. Such interaction makes that they are generally more trustful of their government. This is the “virtuous circle” hypothesis, which predicts that increasing media usage will lead to greater trust (Norris, 2000, 2011).

Despite these seemingly distinctive and incompatible hypotheses, empirical tests often yield more nuanced outcomes. For example, it depends on which media are consulted: Public TV broadcasters tend to sort more positive effects on political involvement and trust than commercial TV broadcasters (Aarts, Fladmoe, and Strömback, 2012). Comparing newspaper and television consumption during a US presidential election, Avery (2009) found only limited support for the reciprocal relationship between political trust and media exposure; the news source, but also the characteristics of news consumers moderate this relationship. The cross-national study by Curran et al. (2014), analyzing TV news and its impact, emphasizes how greater levels of representativeness in the media are beneficial for the efficacy of news consumers. Ceron (2015) shows that consumption of news from news websites is associated with more trust than news from social media. While pluralism in media exposure thus appears to attenuate cynicism and cultivate trust, most previous studies into political trust have either compared different media usages (e. g., newspapers versus television), or sought varieties within media types (public versus commercial television broadcasters). Less attention has been paid to how combinations of media usages within individuals are associated with trust. Also, the majority of studies focuses on trust in political institutions.

3 Bringing in media repertoires

The proliferation of media types and outlets in the past decades has redirected research attention from how much audiences use specific types to the patterns in which audiences combine such types into single repertoires (Van Eijck and Van Rees, 2000; Webster, 2014). This shift in focus is not only instigated by the sheer increase in the amount of media, but also by the rising share of media that incorporate user-generated elements and/or alternative takes on information production (e. g., citizen journalism). Outcomes of media repertoires studies and the degree to which repertoires are embraced by various social groups has yielded different results, mainly because of the variations in items that were used. Appendix A1 provides an overview of studies that have mapped media repertoires in the last 15 years. Most studies find “minimalists” (who use very little media and/or focus on one single type of media, often television) on one side of the spectrum, and “omnivores” on the other side. The other categories often depend on the items used. Bos et al. (2016) find a distinction between public news consumers and popular news consumers. Strömback, Falasca, and Kruikemeier (2018) report similar categories, but in addition also find social media news consumers and local news consumers. Other studies point at the existence of conservative media users, “news junkies” (Mourão et al., 2018), or “curated news only” consumers (Edgerly, Vraga, Bode, Thorson, and Thorson, 2018).

Not all studies distinguish categorical classifications of media usage. Webster and collaborators have studied the formation of media repertoires in various designs, often relying on meters or server-generated data (Taneja, Webster, and Malthouse, 2012; Webster, 2014). In an analysis that mainly focused on television and internet, they find considerable “crosscutting exposure to news” (Webster, 2014, p. 119–127). That is, most audience members combine various television channels and internet platforms, including those of different ideologies and news types. However, their analysis does not consider the degree to which media outlets are consulted on a regular basis.

4 Influences on media repertoires

Despite the variety in labels and classifications in media repertoire studies, there are a number of recurring results in the explanations of membership (which often focus on socio-demographic variables and political interest). It appears that most findings can be interpreted in terms of effects of cultural capital, political engagement, and digital proficiency. For example, minimalists or news avoiders tend to be low-educated, and have lower political interest, which has been associated with less affinity with “legitimate” cultural expressions (cultural capital), and self-perceived positioning in class structures discouraging them to keep up to date with politics (Lindell and Sartoretto, 2018). This group contrasts with omnivores and clusters that prioritize information seeking, who tend to be higher-educated and have stronger incentives to keep politically up to date (Bos et al., 2016; Lindell, 2017; Vandenplas and Picone, 2021). The main other rift that emerges from explanatory analyses concerns the degree to which individual are capable or willing to use digital media. Here, age is obviously the main predictor. Older individuals are more often found in “traditional” media repertoires consisting of television and newspapers, as well as public news sources (Bos et al., 2016; Edgerly et al., 2018; Vandenplas and Picone, 2021).

These results underscore the conclusions by Lindell (2018; Lindell and Sartoretto, 2018), who argues that socialization in the family and in the school is an important resource for shaping news and media repertoires. Similar to other lifestyle characteristics and taste preferences (see Bourdieu, 1984), individuals are heavily influenced by the social contexts in which they are raised, since these provide them (or not) with the dispositions and cultural capital needed to appreciate certain media content (Lindell, 2017; Van Eijck and Van Rees, 2000). Still, explicit tests of socialization effects on media repertoire membership have (as far as I know) not been conducted. Previous studies did show that the degree to which parents are higher-educated and set the example of using “legitimate media” content, increases the chances that their children will use “legitimate media” later in life (Notten et al., 2012; Willekens and Lievens, 2014). It should be noted that “legitimate media” here refers to a socially constructed, and not necessarily stable way of viewing media. Legitimation of cultural content, such as knowledge or information, that is considered valuable or important (in other words, legitimate) in society, has traditionally been connected to those having the “appropriate” cultural capital in higher social classes in modern Western societies.

In a similar vein, higher levels of political trust are found in families where parents have higher educations and more frequently discuss politics (signaling cultural capital) (Marien, 2017). In the digital age, where social media have become quite ubiquitous, particularly among the younger generation, the social network (peers) may have become more important than the role of parents (Burgess, Marwick, and Poell, 2018). Research here is limited, but for a sample of adolescents, Shehata and Amnå (2019) still see a stronger impact of parents than of peers for shaping news media use and political interest. How these influences of social contexts play out when compared and related to repertoires is not clear yet. Therefore, the following research question is posed: How are media repertoires predicted by parental socialization and social networks?

5 Effects of media repertoires

Only about half of the studies mentioned earlier (and summarized in Appendix A1) aim to assess the impact of media repertoires on some other external variable. Wolfsfeld, Yarchi, and Samuel-Azran (2016) – who construct four types based on high/low participation in traditional and social media – show that eclectics have most political interest, but traditionalists more political knowledge. Social media users lag slightly behind. Strömback et al. (2018) focus on online and offline political participation, and find social media news consumers significantly more engaged than other groups. Edgerly et al. (2018) compare the mean participation in four types (volunteering, offline political, online political, lifestyle), and find omnivores to be consistently most active, and news avoiders least active. Mourão et al. (2018) analyze media trust and show that – for the specific period of the first year of Donald Trump’s presidency in the USA – conservative news consumers have the lowest levels of trust in media, while news junkies expressed the highest levels of trust. Castro et al. (2021) show – for news consumers across Europe – that traditionalists and online news seekers display more political knowledge than news minimalists, social media news users, and hyperconsumers of news. As far as I know, the relationship between media repertoires and political or institutional trust has not yet been studied. Therefore, the following research question is asked: How do media repertoires predict institutional trust (net of socio-demographic, parental and social network variables)?

6 Method

To answer the research questions, survey data are used that were collected in May and June 2015 among more than 800 individuals in the Netherlands. Compared to other European countries, the Netherlands has relatively high levels of both online and offline media usage: In 2016 they ranked third, first, and third among EU countries in reading the written press, using the internet, and using online social networks, respectively (European Commission, 2016: 12 ff). At the same time, the Netherlands also has relatively high levels of institutional trust: Within Europe, it ranks just behind the Nordic countries (OECD, 2015). The survey was distributed in the LISS panel of the research agency CenterData. This online panel is based on a true probability sample of households drawn from the population register by Statistics Netherlands, and can be considered representative of the Dutch population. Individuals without internet access participated in the panel using a specific electronic device (for more details: https://www.lissdata.nl/lissdata/). 1,095 persons in the panel aged 16 or older were approached via random selection, and 858 participated, which means the response rate is 78.4 %. In the realized sample, older respondents are slightly overrepresented (27 % was over 65 years old), while the categories 45–64, 25–44, and 16–24 more or less resemble the population with 36 %, 27 % and 9 %.^[1] 53 % of the sample were female. Level of education is fairly representative of the population: almost 8 % for primary school, 22 % lower vocational training, 36 % middle secondary school or vocational training, 23 % higher vocational training, and 11 % university education.^[2]

The survey questionnaire comprised various questions on media usage, cultural socialization, and the individual’s network. Media usage was assessed via the question “How often do you read or use the following media?” (five answer categories ranging from 0 (never), 1 (less than once a month), 2 (a couple of times a month), 3 (a couple of times a week), to 4 ((almost) daily)). The first five media were explicitly in printed format: (a) quality newspapers, (b) popular and regional newspapers, (c) opinion magazines, (d) special-interest magazines (e. g., on sports, cars, or science), and (e) general magazines (e. g., focused on male or female interests). The next three media outlets concerned news websites: (f) websites of quality newspapers, (g) websites of popular and regional newspapers, (h) other news sites. Finally, the questionnaire asked about four online platforms: (i) Facebook, (j) Twitter, (k) Wikipedia, and (l) YouTube. Note that for all newspapers and magazines examples were given to clarify what, for example, constituted “quality” newspapers. In addition, the survey inquired about television and radio usage via the question: “How often do you use the following media to inform yourself on current issues?”. There were three answer categories here: 1 (never), 2 (every now and then), and 3 ((almost) daily).

Institutional trust is measured by asking for seven institutions – three of which are media-related – how much trust respondents have in them on a scale from 0 (no trust) to 3 (a lot of trust). The institutions are: written press, radio, television, political parties, the government, the health system, and science. Reliability is good (Cronbach’s alpha = .811). The mean score was calculated (M = 2.71; SD = 1.44). Also separate variables for media trust (press, radio, television) (Cronbach’s alpha = .823; M = 1.58; SD = .55) and other social institutions (Cronbach’s alpha = .741; M = 2.32; SD = .52) were computed, as there can be differences between institutions (Hudson, 2006). Indeed, a factor analysis disclosed two factors (see Table A2).

Parental socialization is measured via two indicators. Newspaper reading by the parents was measured via retrospective questions to the respondent about the parents at the time she or he was about 12 years old. They were asked separately for the father and mother on a 3-point scale (1 = never, 2 = every now and then, 3 = often). The answers were averaged (M = 2.52; SD = .65). The second measurement concerns the highest educational level of the parents, which provides an indication of the cultural capital available at home. The survey asked about the educational level of both the father and the mother in nine categories; the answers were averaged (M = 3.29; SD = 2.02).

Questions on the social network were asked in two steps. The survey first asked for the size of the social network, focusing on the most intimate friends. The question, adopted from the General Social Survey (2002), was: “Not counting family and a possible partner, how many people would you say you feel really close to, that is, close enough to discuss personal or important problems?” To avoid outliers, a ceiling score of 25 was set. For the sake of interpretation, the variable was rescaled between 1 and 10 (M = 4.86; SD = 2.89). The survey also contained a name generator in which respondents were asked to name their closest friends (up to three persons), and answer two questions for these friends. First, the survey asked for the highest level of education of these friends. Since respondents may not know the precise type of education, only three answer categories were used: 1 (primary school or lower vocational training), 2 (middle vocational training or a couple of years of secondary education), and 3 (higher forms of education (including university)). The mean score was calculated (M = 2.24; SD = 59). The second question concerned the degree to which these friends read newspapers (on paper and/or online). Answer categories are: 1 (never), 2 (every now and then), 3 (quite regularly (but not every day)), and 4 ((almost) every day). The mean score was calculated and rescaled between 0 and 10 (M = 6.83; SD = 2.91). Note that 94 respondents did not report having friends; these respondents will be left out of the explanatory analyses.

The survey also contained questions on socio-demographic background: sex (coded as female), age, and educational level of the respondent (originally six categories).^[3] Finally, it was taken into account whether respondents had a migrant background or not.

7 Results

Based on the criterion of relative model fit due to a limited decrease in Bayesian (BIC) and Akaike information criteria (AIC), as well as interpretability of the model (Collins and Lanza, 2009), the five-cluster model is selected (see Appendix A3). These clusters are shaped based on combinations of various types of media usage. Still, not all forms of media usage are distinctive. Some are so widely spread that they contribute to (almost) all clusters (e. g., television, radio, Facebook usage), while others have much lower frequencies of usage, and therefore hardly show up in the repertoires (e. g., special-interest magazines). Figure 1 presents a summary of clusters using mean scores per media type; the full models can be found in Appendix A4.

The first and largest cluster is labeled cross-media news omnivores (26 %). Individuals in this category distinguish themselves by having relatively high probabilities of reading both quality newspapers and popular/regional newspapers, both in print and online. They also consult other online news websites, and regularly use social media platforms. They are the most avid users of Twitter, Wikipedia, and YouTube, and only second to cluster 5 in using Facebook. The second cluster comprises persons who are very likely to read popular/regional newspapers in print, occasionally read general magazines, and use radio and television. This group is not likely to consult online media. Only a small group of people in this cluster uses Facebook. Therefore, this cluster can be labeled the traditional non-digital media consumers. Almost a quarter of the respondents (23 %) falls into this category.

Figure 1:

Mean scores of media use types per cluster.

The third cluster (21 %) has the highest probabilities of watching television and listening to the radio. Use of these media is combined with reading popular/regional newspapers (both in print and online). People in this cluster share with cluster 2 the avoidance of quality newspapers, but the main difference is that people in this cluster are very likely to consult other news websites and use Facebook. The other online platforms are less popular. This cluster is labeled popular media omnivores.

Cluster 4 is labeled the quality news consumers and concerns about 15 % of the sample. These consumers are most likely to read quality newspapers in print. In addition, they regularly read opinion magazines, as well as the online version of quality newspapers. Although they also use other media, they are not as prolific as people in cluster 1 in using online media, and less likely than people in cluster 3 to use television and radio.

Finally, cluster 5 – which is the smallest cluster (14 %) – comprises individuals who show the lowest probabilities of reading newspapers and magazines, both online and offline. They occasionally consult other news websites, but their media diet mainly consists of watching television, using Facebook, and using YouTube. These individuals can be labeled as non-print-oriented.

8 Conclusion and discussion

This paper contributes to the study of news and media repertoires by deepening our understanding of how repertoires are shaped through parental socialization and social networks, and extending the associated outcomes to the notion of institutional trust. Most previous studies into the relationship between media use and trust disregard the increasingly more complex ways in which different types of media are combined in single repertoires. Empirically, the paper employs the Step-3 approach in Latent Class Analysis on representative survey data from the Netherlands.

Five distinct media repertoires are found among Dutch citizens between the ages of 16 to 92: cross-media news omnivores, traditional non-digital media users, popular media omnivores, quality news consumers, and non-print-oriented. Whereas previous media repertoire analyses reported relatively sharp distinctions between traditional and online news (Bos et al., 2016; Strömback et al., 2018), this research finds cross-media news consumers as the biggest cluster, suggesting that more subtle categorizations are emerging. Importantly, the repertoires found are socially structured. Younger individuals are more likely to be cross-media news consumers or non-print-oriented media users, depending on their own level of education and that of their parents: The more of these resources, the higher the likelihood that someone is a cross-media news omnivore. If persons in the social network do not read newspapers, chances increase that someone is non-print-oriented. Older citizens are most likely to belong to one of two repertoires: traditional non-digital media users (mostly lower-educated and a low-educated network) or quality news consumers (mostly higher-educated). Again, parental cultural capital differs considerably between these groups, although this concerns only parental education; I find no effects of parental newspaper reading. These findings underscore the continuing importance of social contexts (Lindell, 2018): Parents exert influence via education, social networks via newspaper reading.

The five media repertoires are significantly associated with degree of institutional trust. Popular media omnivores display the highest level of institutional trust, followed by quality news consumers. The outcome that institutional trust differs across media repertoires suggests that the way citizens inform themselves on societal issues matters for the trust they develop in social institutions. On the one hand, diversity in media content consumed (omnivorism) is associated with more trust – confirming the theoretical idea that being exposed to various sources and perspectives fortifies trust in how society functions (Norris, 2000). This finding seems in line with findings that individuals with more varied media diets tend to be more engaged (Edgerly et al., 2018) and avoid echo chambers (Dubois and Blank, 2018). On the other hand, relying mainly on high-quality information sources appears to be advantageous as well. The role of “legitimate media” is thus not over. The low level of institutional trust of the non-print-oriented is somewhat worrying. This group consists of relatively young individuals who do not use a lot of news sources. They may encounter news via Facebook: Our survey did not ask about that option. It should be noted that the analyses of media repertoires and trust control for parents and social network. I find few direct relationships with institutional trust, implying that their influence mainly runs via media repertoires. The exception concerns the size of the social network, which is positively related to trust, as also found in previous studies (e. g., Putnam, 2000).

What are the broader implications of these results? On the one hand, we know that social background and socialization remains an important predictor for how people view news media and develop their own news habits (Lindell and Sartoretto, 2018). The results from this study suggest that the influence of socialization is more related to cultural capital in general (parents’ education levels) than setting specific examples (parents’ newspaper reading). We also find evidence that social networks matter for how media repertoires develop (and eventually how trust grows). When it comes to thinking about how the role of news media among younger cohorts can be restored this may offer possibilities. A recent study by Centola (2018) showed the importance of strong ties in networks – family, friends, people one trusts – for creating behavioral change. Rather than looking for opinion leaders or influencers, his work suggests that change should be sought in strategically addressing various persons in close contact with each other in order to increase credibility and legitimacy. For young people, this could perhaps be pursued in school contexts (e. g., with classes on media literacy).

Future research could pursue in at least two directions. One direction would be more fine-grained analyses of how media repertoires are precisely shaped within social networks. This could be done from a more qualitative perspective, which would enable a more in-depth examination of how various identity-building factors interact. Another direction would be to bring the study of media repertoire impacts closer to work on political trust and attitudes that examines the current rise of populist movements in relation to mistrust of institutions (Norris and Inglehart, 2019). In these studies, the role of media is often not well represented, despite social media having been associated with populism (Engesser, Ernst, Esser, and Büchel, 2017). How the results should be interpreted from a broader comparative perspective is harder to tell: Although the Netherlands has traditionally been a “high-trust” country, the rise of social media – and potentially a larger share of the non-print-oriented repertoire – may pose a challenge to this status in the long run.

There are, like in any study, certain limitations to this research project. The paper uses cross-sectional data, which implies that relationships can only be assumed to be causal based upon theoretical arguments. Certain media repertoires more often coalesce with higher levels of trust than others, but whether media usage stimulates trust, or trust steers individuals towards certain media outlets is unclear. A second point to note is that the measures of media usage did not specifically ask about news consumption, which some scholars may consider a limitation. Yet one could argue that a specification is increasingly problematic in the age of media convergence. Also, our media repertoire results are fairly in line with previous studies, implying that more specific phrasings perhaps do not bring much added value. It is also acknowledged that the indicators for social networks and parental socialization could have been more elaborate. Socialization questions were only measured on a 3-point scale, because they involved retrospective assessments. The social network question unfortunately only asked about newspaper reading, and no other forms of media usage. It would be worthwhile to improve these measures in follow-up studies. Future research could also be more elaborate in using media use indicators; one critical point of attention would be how embedded media (e. g., media feeds on social media platforms) and alternative devices (e. g., television broadcasters putting content on YouTube) can be measured in individual usage patterns.

To conclude, this paper shows how media repertoires maintain a relevant lens to investigate the way individuals in the early 21^st century find information and display engagement with society. Whereas previous studies often focused on political participation, this paper analyzed the relationship between media repertoires and institutional trust. Media repertoires differ considerably between social groups – particularly between the young and the old, and those with low and high levels of education and cultural capital. But media repertoires also invoke different levels of institutional trust: Persons who shy away from online and offline newspapers, magazines, and even television, to mostly spend time on Facebook and YouTube, have significantly lower levels of trust in social institutions than persons in other media clusters.