How the public understands news media trust: An open-ended approach

Abstract

Despite the central role that ordinary citizens play as ‘trustors’ (i.e. the actor that places trust) in the literature on news media trust, prior quantitative studies have paid little attention to how ordinary citizens understand and define news media trust. Here, trust tends to be studied from a researcher-defined – rather than an audience-defined – perspective. To address this gap, we investigate how the public describes news media trust in their own words by asking them directly. We analyse 1500 written responses collected through a Norwegian online probability-based survey, here using a semisupervised quantitative text analysis technique called structural topic modelling (STM). We find that citizens’ own understanding of news media trust can be categorised into four distinct topics that, in some instances, are comparable to academic and professional discourse. We show that citizens’ written descriptions of news media trust vary by many of the same variables that prior research has found to be important predictors of levels of trust. Respondents’ written descriptions of news media trust vary by education and satisfaction with democracy but not other known predictors of trust, such as ideological self-placement and political preferences.

Keywords

Bias credibility ideology open-ended survey question structural topic model trust in media

Introduction

A considerable body of research has investigated the causes and effects of trust in news media (e.g. Engelke et al., 2019; Kiousis, 2001; Ladd, 2012; Prochazka and Schweiger, 2019; Tsfati and Ariely, 2014), successfully tapping into the multiple dimensions of trust derived from theory (e.g. Appelman and Sundar, 2016; Kohring and Matthes, 2007; Metzger et al., 2003; Roberts, 2010). This strategy has enhanced our understanding of how news media trust and credibility should be conceptualised and measured. Yet because the majority of quantitative research on news media trust deals with responses to close-ended questions along a scale or from pre-established categories from a researcher-defined – rather than an audience-defined – perspective, we still lack quantitative data on how citizens understand and describe news media trust, how such understandings vary and how different factors drive such differences.

To address this gap, we contribute to the growing body of journalism research focusing on audiences (Swart et al., 2017) by examining what words people draw on when asked to explain what trust in news media means to them and, in extension, the reasons they offer regarding trust and mistrust in journalism. As such, we build on what Nielsen (2016) has called folk theories of journalism, that is, peoples’ perceptions of what journalism is and should be. In doing so, we glean a small part of ‘the culturally available symbolic resources that people used to make sense of journalism’ (Nielsen, 2016). These theories are informative of how citizens make sense of the institution of news media in their everyday lives.

The present study provides the first in-depth, large-scale analysis of an open-ended question about how the public understands trust in media (for a notable study exploring the neighbouring aspects of trust, see Newman and Fletcher, 2017), here by using a semisupervised quantitative text analysis technique called structural topic modelling (STM) to analyse the open-ended survey responses and how they vary systematically with respondent characteristics.

Trust in news media

The concept of trust describes the relations between a trustor, the actor who places trust, and a trustee, the actor being trusted. A widely used definition of trust in the literature is ‘the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control the other part’ (Mayer et al., 1995: 712).

In the context of trust in media and journalism, news consumers are (ideally) trusting newsrooms, editors and journalists to gather useful and correct information that the consumers may use to orient themselves in the world. People are taking a risk by not double-checking or doing the information gathering themselves. Thus, Hanitzsch et al. (2017) define media trust as ‘the willingness of the audience to be vulnerable to news content based on the expectation that the media will perform in a satisfactory manner’.

As noted by Kohring and Matthes (2007: 238), ‘when there is nothing at stake, trust is not needed’. Thus, trust in news media entails a certain risk in which news consumers place trust in journalists and editors of the news media and in the news media as an institution (e.g. Tsfati and Ariely, 2014), specific sources (e.g. Hovland et al., 1949) or a journalistic message (e.g. Appelman and Sundar, 2016).

Typically, trust is measured using a single question in a survey, asking directly to what degree people trust or distrust the news media or ‘the press’ in general as a form of institutional trust (e.g. Hanitzsch et al., 2017). Alternatively, news media trust and credibility can be measured by incorporating a scale of questions measuring factors such as accuracy, selectivity of facts, selectivity of topics and journalistic assessments (see Kohring and Matthes, 2007) or factors such as ‘fair’, ‘unbiased’, ‘objective’ and ‘delivering a diversity of opinions’ (perceived trustworthiness), along with ‘written by professional journalists’ (perceived expertise) (see, e.g. Chung et al., 2012), or ‘fair/unfair’, ‘unbiased/biased’, ‘tells the whole story/doesn’t tell the whole story’, ‘accurate/inaccurate’ and ‘can be trusted/can’t be trusted’ (the believability index) (Meyer, 1988: 574). Although not used as an index scale, some studies have also investigated attitudes about news media that are related to news media credibility, such as ‘news organizations tend to favor one side when presenting the news on political and social issues’ (Barthel and Mitchell, 2017).

Research on trust in news media has almost solely emerged under the label of media credibility (for an overview, see Kohring and Matthes, 2007; Metzger et al., 2003; Roberts, 2010). As noted by Kiousis (2001) and Kohring and Matthes (2007), the terms trust and credibility are used interchangeably within the literature. Furthermore, the findings related to media generally and news media specifically are also entwined (Fisher, 2016).

In the current paper, we draw on scholarship from both the tradition of media trust research and credibility research. However, we view trust as a higher order concept, of which the term credibility can speak to some aspects. As such, we follow what Engelke et al. (2019) posit as the most common position, here treating credibility as ‘one of several antecedents of the more complex concept of journalistic trustworthiness’ and viewing credibility as a form of believability, that ‘only partially captures the concept of trust’. Kohring (2019) argues that people tend to look for clues and signs that can legitimise their own trust or distrust in the media, here seeing as they do not have the capabilities to fully judge the reliability of the news they use. In this view, perceptions of media credibility can ‘be conceptualized as encapsulating the clues that people use to evaluate their trust in media’ (Strömbäck et al., 2020).

Kohring and Matthes (2007) identify three different methodological approaches for the measurement of credibility within this body of research: (a) research on source credibility (e.g. Hovland et al., 1949), (b) research on the comparative credibility of newspapers, radio and television (e.g. Greenberg and Razinsky, 1966; Roper, 1985; Shaw, 1973) and (c) research seeking to identify the dimensions of credibility by a factor analysis (see, e.g. Gaziano and McGrath, 1986; Kohring and Matthes, 2007; Meyer, 1974 for an overview). Although Kohring and Matthes (2007) acknowledge an increasing precision in measurement in this body of research, they argue that these studies lack a clear conceptual basis for the operationalisations of the concept of media credibility.

In an attempt to alleviate this concern, Kohring and Matthes (2007) have developed and validated a multidimensional scale; their study is based on survey items that have been fielded in Germany. Through several pilot studies, the authors develop a battery of survey questions that taps into four theoretically informed dimensions of news media trust: ‘selectivity of topics’ (emphasising the adequate attention and status of topics), ‘selectivity of facts’ (emphasising inclusion of essential points, important facts and information and different points of view), ‘accuracy’ (placing emphasis on verifiable information, truth and correctness) and ‘journalistic assessment’ (consisting of notions that ‘criticism is expressed in an adequate manner’, that ‘the journalists’ opinions are well-founded’, that commentaries consist of ‘well-reflected conclusions’ and that the journalistic assessments are ‘useful’).

The approaches described above are examples of researcher-defined understandings of trust. However, we propose an alternative and supplementary approach towards understanding citizens’ trust in news, one where we start inductively by directly asking people what comes to their mind when they think about trust in media without stipulating anything beforehand. We attempt to answer what people are thinking about when asked about ‘trust’ in a single question.

The public’s understanding of news media trust

The vast majority of quantitative research on news media trust deals with responses to close-ended questions that use a scale and pre-established categories. However, as Fisher (2016) and Daniller et al. (2017) point out, close-ended question items about ‘trust in media’ are not enough to unveil what people actually think and how they interpret the concept. To elucidate whether citizens’ interpretation and perception of trust in the media is separate from scholarly conceptualisations, we need to use an approach that fully captures citizens’ perception of trust.

Although one obvious route to shed light on this question would be qualitative studies, for instance, to perform focus groups to gain rich and nuanced descriptions of perceptions of trust (see Metzger et al., 2010), we choose a method that allows us to consider qualitative data in a generalizable environment representative of a population. This has the added benefit of allowing for analyses of heterogeneity in citizens’ understanding of trust, here by exploring whether demographic factors such as age and education vary systematically.

This does not exclude the use of focus groups in future research, but because this is an empirical question with profound implications for what we know about trust gathered from the realm of surveys, it is only natural that we begin untangling this problem by understanding what people actually think about when they answer surveys about the concept of trust. Thus, the main contribution of our study is that we can speak to how different understandings covariate with individual-level characteristics.

Our solution is to let citizens freely formulate an answer to an open-ended question about news media trust. Close-ended survey questions have long been considered to lack depth (Lazarsfeld, 1944), and researchers have argued for open-ended approaches for measuring public opinion (Krosnick, 1999; Lazarsfeld, 1944). In many ways, open-ended survey questions are superior for studying spontaneous responses and avoiding bias that can result from fixed response options (Reja et al., 2003; Schuman and Presser, 1979). However, the majority of survey analyses still focuses on closed-ended questions. From a respondent perspective, open-ended questions in population-based mail surveys have been regarded as being too demanding and resulting in an increase in nonresponse. From an analytic perspective, open-ended answers have been regarded as too time-consuming to analyse. However, the digitalisation of surveys, as well as recent developments within semisupervised quantitative text analysis techniques, has sparked renewed interest in the method (Roberts et al., 2014). An open-ended survey question in an online population-based survey provides the opportunity for an analysis that is both qualitatively thick and quantitatively broad in scope.

Prior research

Newman and Fletcher (2017) provide a notable example of how an open-ended approach can bring more depth to questions about news media trust. Based on data from Newman et al. (2017), they explore citizens’ ability to distinguish fact from fiction across nine countries; this was done by using open-ended survey responses. The respondents were asked whether they agreed with the following statement: ‘The news media does a good job in helping me distinguish fact from fiction’. The respondents who indicated a preference (N = 7915) – in terms of not choosing the middle option (‘neither agree nor disagree’) – were asked to give a reason in an open-ended text box. Based on human coding of the responses, the authors found that ‘a significant proportion of the public feels that powerful people are using the media to push their own political or economic interests, rather than represent ordinary readers or viewers’ (Newman and Fletcher, 2017: 5). On the other hand, those who do trust the news media’s ability to distinguish between fact and fiction typically responded that journalists do a good job of checking sources, providing evidence to back up claims and verifying facts. Although the study certainly provides valuable methodological and empirical insights into the scholarly literature on why citizens trust or distrust the news media, it is specifically related to the question of whether ‘the news media does a good job in helping me distinguish fact from fiction’ (Newman and Fletcher, 2017). Thus, Newman and Fletcher’s study is particularly concerned with a specific dimension of news media credibility (see, e.g. Gaziano and McGrath, 1986: 455 for the dimension ‘Does or does not separate fact and opinion’).

The present study extends these insights and breaks new ground by analysing a large number of responses to an open-ended question specifically about trust in media. How would we expect the public to describe news media trust when asked to do so in their own words? Previous research tells us not to expect dramatic differences from the professional and academic discourse. For instance, Gladney (1996) finds newspaper readers and newspaper editors to agree on the importance of seven out of nine organisational characteristics, including traits such as integrity, impartiality, editorial independence, editorial courage and staff professionalism. Others have found an overall agreement in attitudes when comparing the two groups (Heise et al., 2014; see also Abdenour et al., 2020 for more of an overview and Lacey and Rosentiel, 2015 for elucidations on various definitions of quality journalism).

Research questions

We seek to address three overarching questions: (1) How do people describe and understand news media trust? (2) How do these understandings vary? (3) What factors drive these differences? To address the first two questions, we analyse the respondents’ open-ended responses through a semisupervised quantitative text analysis technique named STM, which ‘allows the researcher to discover topics from the data, rather than assume them’ (Roberts et al., 2014: 1066). Because this semisupervised procedure is inductive and empirical in nature and because this analysis is the first of its kind in terms of an open-ended question about media trust, we ask the following (Research question (RQ1)): Which topics can be induced through STM when citizens are asked to define trust in media using their own words?

To address the third question, we include a selection of known predictors of news trust as the variables in the STM model described above. As a theoretical backdrop for this model, we build on a growing body of cross-national research explaining why and how news media trust varies (Hanitzsch et al., 2017; Suiter and Fletcher, 2020; Tsfati and Ariely, 2014). Although this stream of literature on the predictors of media trust points the way in terms of what to expect, it is important to stress that we do not know the extent to which and how predictors of news trust matter for how people understand and define trust.

First, we aim to explore how and the extent to which understandings of news trust vary by age, gender and education. Although gender and education consistently predict news trust in prior cross-national studies (Hanitzsch et al., 2017; Suiter and Fletcher, 2020; Tsfati and Ariely, 2014), the correlation between age and trust is inconsistent (significant predictor in Hanitzsch et al., 2017; Suiter and Fletcher, 2020; but not in Tsfati and Ariely, 2014). We ask the following (Research question (RQ2)): How and to what degree do understandings of media trust vary by age, gender and education?

Second, we use the literature on ideological and political differences in media trust (e.g. Hanitzsch et al., 2017; Suiter and Fletcher, 2020) as a starting point for our theoretical expectations. Hanitzsch et al. (2017) find that ideological extremism predicts lower trust, and Suiter and Fletcher (2020) find that left-right-partisans trust news less than non-left–right partisans. Building on these insights, we seek to explore the following (Research question (RQ3)): How and to what degree do understandings of media trust vary by ideology and partisanship?

Third, we explore the possibility that people’s understanding of news media trust vary based on their news consumption of traditional media (i.e. mainstream television, newspapers and radio). Cross-national evidence shows that the more mainstream news individuals consume, the more they tend to trust the news in general (Hanitzsch et al., 2017; Tsfati and Ariely, 2014; see also Suiter and Fletcher, 2020, who used ‘Interest in news’ as a predictor). This leads us to ask the following (Research question (RQ4)): How and to what degree do understandings of media trust vary by how important traditional news is as a source of news? Fourth, we rely on cross-national evidence (Tsfati and Ariely, 2014) showing that interest in politics is a positive predictor of news media trust, so we ask the following (Research question (RQ5)): How and to what degree do understandings of media trust vary by political interest?

Finally, prior studies have found a strong link, a so-called ‘trust nexus’ between trust in the news media and citizens’ approval of political institutions (Gronke and Cook, 2007; Hanitzsch et al., 2017). The argument is that citizens’ trust in the news media is related to the way they look at and trust other political institutions and in democracy in general (e.g. Hanitzsch et al., 2017; Tsfati and Ariely, 2014; Tsfati and Cappella, 2005). Citizens who are less satisfied with democracy or disdain other political institutions also tend to trust the news to a lesser degree (Hanitzsch et al., 2017). Building on these insights, we ask the following (Research question (RQ6)): How and to what degree do understandings of media trust vary by satisfaction with democracy?

Methods

Context

Our survey is set in Norway, a multiparty system and welfare state characterised by a highly educated population, a large population of newspaper readership and a ‘corporate democratic’ media system: a strong institutionalised professionalism (Hallin and Mancini, 2004), with extensive schemes for press support from the state. Traditional news media institutions remain strong across platforms: online newspapers with ‘print mothers’, along with the publicly funded broadcasting institutions, which are the top online brands in terms of weekly use in Norway (Newman et al., 2017).

Norwegian citizens have a relatively high trust in the news media and institutions in general (Newman et al., 2017). The country has a history of partisan newspapers with tight affiliations to political parties; these affiliations began dissolving in the 1970s when newspapers increasingly pursued the role of the ‘independent watchdog’, ideally holding those in power to account on behalf of the people. The Scandinavian news media occupies a middle position between the Southern European and North American media systems. While the Southern European news media and journalists are more likely to adhere to a partisan role, the news media and journalists in North America are more prone to take on a neutral, nonpartisan role (Hallin and Mancini, 2004).

Data

The data for the current study were collected from a probability-based online national survey conducted by the Norwegian Citizen Panel (NCP). The NCP’s respondents were gathered through the postal recruitment of 25,000 individuals over the age of 18 years. These individuals were randomly selected for recruitment from Norway’s National Registry, a list of all individuals who either are or have been a resident in Norway; this registry is maintained by the official Tax Administration. For details about response rates or other methodological matters, we refer to the NCP methodology report (Skjervheim and Høgestøl, 2017). The present study was fielded within the NCP in May 2017 with a randomly drawn subsample of 2086 Norwegian citizens (Norwegian Citizen Panel Round 9, 2017). The respondents were asked the following: ‘There has been a lot of talk about “trust in media.” In your own words, what does trust in the media mean to you?’ Below the vignette, there was a large text box. Here, 1898 respondents wrote one word or more, of which 1500 responses were used in the estimation. Additional details about the data, including the response length and missing data, are shown in the Supplemental Appendix.

Research design

To conduct the analyses, we made use of a semisupervised quantitative text analysis technique named STM (see Roberts et al., 2014). The STM technique helps uncover topics using a clustering algorithm that is based on the co-occurrence of words across the respondents’ written responses. Here, the researcher chooses the number of topics to estimate. The STM then provides the prevalence of each uncovered topic for each response, along with other quantities of interest. We focus on prevalence, that is, how much each respondent writes about each topic. The STM technique often produces several proposed topics that can be regarded as optimal solutions in numerical terms (Roberts et al., 2014). We ran and manually evaluated models showing between 3 and 12 topics.

What separates STM from other LDA-based topic models is that the researcher can provide priors, such as demographic variables, for the topic prevalence to improve the estimation (Roberts et al., 2014). We have included the respondents’ age, education, gender, satisfaction with democracy (1 ‘Not satisfied at all’ to 5 ‘Very satisfied’), interest in politics (1 ‘Not interested at all’ to 5 ‘Very interested’), interest in political news (1 ‘Not interested at all’ to 5 ‘Very interested’), interest in sports and entertainment news (1 ‘Not interested at all’ to 5 ‘Very interested’), traditional media as an important source of news (1 ‘Not important at all’ to 5 ‘Very important’), social media as an important source of news (1 ‘Not important at all’ to 5 ‘Very important’), whether or not they like or dislike nine different Norwegian political parties (1 ‘Strongly dislike party’ to 7 ‘Strongly likes party’) and self-placement on the left–right ideological scale (0 ‘Left’ to 10 ‘Right’).

For each topic number, we started by performing 100 initial STM runs with a few expectation–maximisation (EM) iterations. The four models with the highest likelihood values were then fully estimated. Each of these runs were then manually evaluated and validated by all the authors, and this procedure led to the joint selection of one STM run with four topics. We describe these four topics in the next section. More details about the estimation and data are available in the Supplemental Appendix.

Results

Four topics about news media trust

RQ1 asked which topics could be induced from the open-ended survey responses about news media trust. From the selected STM run, we found that Norwegian citizens wrote about four different topics. Table 1 displays a summary of these topics’ most discriminating words in terms of exclusivity and frequency, here by using the simplified frequency-exclusivity (FREX) (see Roberts et al., 2014: 1068), our suggested labels for these topics, and the topics’ prevalence. FREX attempts to find words that are both frequent in and exclusive to a topic. To more easily refer to the topics in the results section, we have suggested labels for each topic: truthfulness, thoroughness and professionalism, bias and independence and objectivity. It is important to note that these ‘labels’ are the authors’ own ideas and are only suggestions. In Table 1, we have italicized the words that we have placed an emphasis on (in the FREX summary) when suggesting the labels. Note that some of the words that the participants used do not necessarily relate to their understanding of trust but rather to reasons or context or simply other things that the respondents were reminded of. The words displayed in the FREX summary in Table 1 are indicative of what the people wrote about in the different topics and what words were more distinct for one topic over another. In addition to the FREX summary, the STM procedure let us investigate the most representative answers for each topic. We relied on some of the words that were actually used in the most representative answers when suggesting labels for the topics. When manually reading through the entire text corpus and the most representative answers in the topics, we also observed that the written responses often were consistent with the scholarly conceptualisations of media trust and credibility, and moreover, these responses were comparable to academic definitions and professional norms of journalism. We now present the four suggested topics, giving brief qualitative explanations of them to elucidate what makes a topic distinct from the others.

Table 1.

The selected four-topic model with the most frequent/exclusive words and overall topic prevalence.

Suggested label	Topic prevalence	Most important words (FREX)
(1) Truthfulness	0.26	Write, true, truth, news, facts, good, trust, colored, honest, says, presents, news, said, source criticism
(2) Thoroughness and professionalism	0.22	Source, trustworthy, accountable, cover, rich, more, correct, convey, information, thorough, publish, clarify, report, relevant
(3) Bias	0.32	Trust, more, much, often, others, low, large, seen, thing, NRK, sees, people, person, experience
(4) Independence and objectivity	0.20	Neutral, independent, fourth, framing, agenda, viewpoint, forward, political, fact checking, should, describe, presentation, confident, governed

The table shows the topic prevalence of the four induced topics, the topic prevalence for each topic, and the most important words (as determined by the STM model). Note that the FREX list of words are translated from Norwegian.

Topic 1: Truthfulness

In Topic 1, which we have suggested as labelling ‘Truthfulness’, the respondents wrote more about news media trust in relation to ‘writing the truth’ and being trustworthy. Some also use the term ‘fake news’ in a contrasting manner. This topic seemingly relates to the respondents’ expectations that news and journalistic content should produce relatively undistorted versions of events or reality. The implication is that the domain of journalism should be that of truth and facts. As one respondent whose response were among the most representative answers in the topic model wrote, ‘Trust is, for me, equivalent to being able to trust something or someone. When I read newspapers, I expect to read news and reports about real events (. . .)’.

Topic 2: Thoroughness and professionalism

In Topic 2, which we have suggested as labelling ‘Thoroughness and professionalism’, the respondents wrote more about news media trust in relation to how thorough journalists and newsrooms are in their practice. This topic seemingly relates to the respondents’ expectation that what is presented as truth and fact is certain, that the information comes from credible sources, and that the information is validated and checked, that is, that the information presented can be trusted to be of a good quality and, thus, useful. Furthermore, the respondents also touched on the thoroughness of presentation in this topic. Claims should be substantiated. In other words, the respondents seemingly wanted a thoroughness of argumentation and information presentation and thoroughness in information gathering by individual journalists and newsrooms. As one respondent whose response was among the most representative answers in the topic model wrote, ‘That issues are presented in a trustworthy fashion, that sources are of quality and journalist exercise good professional judgment’.

Topic 3: Bias

In Topic 3, which we have suggested as labelling ‘Bias’, the respondents wrote more about mainstream media bias and a political left bias in journalism. Notably, the answers related to a structural-ideological level and were seemingly descriptive because they were claiming that a perceived state of hegemony is at play in journalism and newsrooms. In contrast to the three other topics, the word summary (FREX) (see Table 1) for this third topic does not give a clear indication of what the respondents wrote about. Thus, for the third topic, we have relied on the manual reading of the most representative answers when suggesting a label. Some answers claimed that those who work in the media have a different worldview and ideology from the respondent, indicating that the journalists do not represent some people’s worldviews and ideologies. The responses in the bias topic are not primarily centring on the respondents’ own ideological beliefs; rather, it is about the perceived ideological orientation among the news media in general and perceived lack of ideological diversity in news media in particular. As one respondent, whose response was among the most representative answers in the topic model, wrote, ‘The media influence people like the Bible. Some believe in the story while others are in doubt. Much of the press is left-wing biased. Personally, I have relatively high trust in the media’.

Topic 4: Independence and objectivity

In Topic 4, which we have suggested as labelling ‘Independence and objectivity’, the respondents wrote more about the importance of steering clear of the influence of powerful actors such as commercial (owners, advertisers) or political powers. In many ways, this mirrors the arm’s-length principle that is often held as an ideal by journalists. Importantly, the use of the word objectivity in this topic suggests a wish for impartiality. Seemingly, the answers are prescriptive for how the respondents would like to see journalists work. We can observe similarities between this topic and that of bias because they both address the problem of subjectivity in news. The main difference is that the former is prescriptive and the latter is descriptive. When writing about ‘objectivity’ in this topic, the respondents wrote more concretely about the various measures they would like journalists and newsrooms to take (at a level of everyday practice), while the responses for the bias topic were more likely to treat journalism and media as larger entities (at the institutional level). For the topic of independence and objectivity, there were even indications that a political or ideological stance in and of itself is not really a problem as long as there is transparency and openness about the issue. As one of the most response-wise representative respondents (according to the STM-model) wrote, ‘To be certain that the media acts without political, economic, or other influences. That the media does not have one agenda, and that they reproduce facts and not speculation or assumption’.

Topics and respondents’ characteristics

RQ2–RQ6 addressed how the demographic variables included in the model can explain variations in the prevalence of each topic. To answer these questions, we proceeded by regressing the four induced topics on the demographic variables included in the model (for more details about this analytical approach, see Roberts et al., 2014). For each analysis, we held all other variables included in the model constant (see the Supplemental Tables B1–B4 for the regression tables of the presented figures). We find that some of these topics vary substantially in terms of the respondents’ education, satisfaction with democracy and political interest.

Figure 1 illustrates how topic prevalence varies by the respondents’ education. Lower values (i.e. point estimates to the left of 0 in Figure 1) indicate lower education, and higher values (i.e. point estimates to the right of 0 in Figure 1) indicate higher education. The figure shows that lower education predicts higher topic prevalence of the bias topic and the truthfulness topic, while higher education predicts higher topic prevalence for the thoroughness and professionalism topic, along with the independence and objectivity topic. This means that what a respondent writes about can be predicted by his/her level of education.

Figure 1.

The relative prevalence of each topic by the respondents’ education.

Figure 2a shows the prevalence of each topic by the respondents’ satisfaction with democracy, showing statistically significant differences for the bias and truthfulness topics. More specifically, Figure 2a shows that low satisfaction predicts a higher topic prevalence for the bias topic and higher satisfaction predicts a higher topic prevalence of the truthfulness topic.

Figure 2.

The relative prevalence of each topic by different respondent characteristics.

Figure 2b shows the prevalence of each topic by the respondents’ interest in politics, showing statistically significant differences for the truthfulness topic. More specifically, Figure 2b shows that a lower interest in politics predicts higher topic prevalence of the truthfulness topic.

Figure 2c indicates a statistically significant relationship between the bias topic and importance of traditional media (television, newspapers and radio, as opposed to social media) as a source of news. Here, lower importance of traditional media predicts a higher topic prevalence for the bias topic.

Finally, Figure 2d shows the differences in topic prevalence by self-placement on the ideological scale. Strikingly, all four topics vary little with Norwegian’s political (party preference) and ideological (self-placement on the ideological left–right scale) preferences. As displayed in Figure 2d, we cannot find a clear pattern in terms of self-placement on the ideological left–right scale, indicating that the four topics are neither related to the right nor left sides of the political scale (see the Supplemental Tables B1–B4, for the lack of statistically significant differences for party preference as well).

Discussion

Despite the central role ordinary citizens play as ‘trustors’ in studies of news media trust, prior quantitative studies have paid little attention to how ordinary citizens understand and describe news media trust. To address this gap, the current study contributes with an analysis of quantitative open-ended data on how citizens understand and describe news media trust in their own words, how such understandings vary and how different factors drive such differences.

Through a STM analysis of 1500 written responses, we induced four distinct topics. We observe a form of overarching agreement between (some of the) responses in the topics discovered by our STM model and in the academic and professional understanding of quality news and news media trust. For instance, the most important words (FREX) and most representative answers in the truthfulness topic, here relating to the respondents’ expectations that news and journalistic content should produce relatively undistorted versions of reality, corresponds with the dimension described by Meyer (1988) that the news media is ‘telling the whole story’. The thoroughness and professionalism topic, which relates to the respondents’ expectation that the news media presents the truth and validates information, corresponds with factors such as ‘selectivity of facts’, ‘accuracy’, ‘journalistic assessments’ and ‘selectivity of topics’ (Kohring and Matthes, 2007), along with ‘trustworthiness’ (Chung et al., 2012). The independence and objectivity topic, which relates to objectivity and a wish for impartiality, corresponds with factors such as ‘fairness’ (including different points of view), objectivity (Chung et al., 2012) and ‘selectivity of facts’ (Kohring and Matthes, 2007). This overarching agreement mirrors previous research, finding overall agreement on many, if not all, elements (Gladney, 1996; Heise et al., 2014). One exception is the bias topic. Here, we observe explanations that are not usually seen when professional journalists, editors or journalism scholars describe news media trust. This discrepancy warrants more exploration because it points to a citizen perspective on news media trust that is rarely explicated in existing research. The contents of the bias topic in the respondents’ discourse on media trust could be taken to corroborate the findings by Newman and Fletcher (2017: 36) that ‘many people no longer see the media as representing the interests of ordinary people (. . .). For many people, the news media is seen as part of the establishment elite, biased, or just out of touch’.

Furthermore, we observe that citizens’ written descriptions of news media trust vary by many of the same variables that prior research has found to be important predictors of levels of trust. We know from previous research that levels of trust in news media typically vary along these factors; that is, people with low education and low satisfaction with democracy will typically trust news media to a lesser extent than people with a higher degree of education and higher satisfaction with democracy. Using these insights as a theoretical starting point for understanding what predicts differences in people’s understanding of news media trust, we explored how a range of different known predictors of media trust correlated with the four induced topics in the STM model. We found that satisfaction with democracy, education and interest in politics correlated with differences in what people emphasise when asked to describe news media trust in their own words.

Those who wrote more about the bias topic were more likely to report a lower satisfaction with democracy. This is partially in line with the argument by Tsfati and Cappella (2005) that media trust is related to trust in democracy and political institutions. Extending this argument, Hanitzsch et al. (2017) suggest that this link between media trust and political trust is related to antielitism, which is understood as a perceived antagonistic relationship between ‘the elite’ and ‘the people’. This link is in line with the bias topic in the present study. However, the four induced topics vary little with the respondents’ party preference and self-placement on the political left–right scale; this includes the bias topic, suggesting that the four topics in general and the bias topic specifically transcend the political left–right dimension of politics and are not limited to, for instance, the radical right of politics. We interpret this to mean that the induced topics cannot clearly be divided into differences of political preference and ideological self-placement. This is not obvious because one could speculate that the respondents writing the most about political bias (i.e. some of the most representative answers in the bias topic) would typically self-identify to the right of the political spectrum. Previous research has found lower trust in news among such groups in many different countries, including Norway (Suiter and Fletcher, 2020). Here, we find that the respondents across the whole of the political spectrum can draw on such reasons when describing what media trust means to them.

A different reading of these results could be that we – at least for some respondents, particularly those within the bias topic – captured their reasons for trust rather than their understanding of trust. Thus, our data could also be interpreted as showing that different levels of trust are based on different reasons. Although our data did not allow us to include trust in the news media as a predictor, it is reasonable to assume that the respondents in the bias topic were more likely to report lower levels of trust in the news media because they were more likely to report lower satisfaction with democracy, which prior research has shown are highly correlated with trust in journalism (e.g. Hanitzsch et al., 2017).

These results provide new insights and nuances that leave us better equipped to construct more fine-grained tools for exploring the differences in people’s trust in news media. We now know that even though people mostly agree with academic and professional discourses when writing about trust in news media, different demographic groups emphasise different perspectives based on their education and satisfaction with democracy.

Future studies should explore audiences’ understandings of trust further. One natural next step, at least in the vein of folk theories of journalism (Nielsen, 2016), could be to take the groups for which the bias topic matters and examine how they make sense of the news and institution of news media in their everyday lives, along with what symbolic and cultural resources they draw on when doing so.

Supplemental Material

sj-docx-1-jou-10.1177_14648849211005892 – Supplemental material for How the public understands news media trust: An open-ended approach

Supplemental material, sj-docx-1-jou-10.1177_14648849211005892 for How the public understands news media trust: An open-ended approach by Erik Knudsen, Stefan Dahlberg, Magnus H Iversen, Mikael P Johannesson and Silje Nygaard in Journalism

Footnotes

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by Trond Mohn Stiftelse (project number: BFS2015TMT01) and the MediaFutures centre (project number: 309339).

ORCID iD

Magnus H Iversen

Supplemental material

Supplemental material for this article is available online.

Author biographies

Erik Knudsen is a researcher at the Department of Information Science and Media Studies at the University of Bergen. Much of Knudsen’s research explores patterns and effects of news and political communication, trust in journalism and selective exposure to like-minded information. Email: Erik.Knudsen@uib.no.

Stefan Dahlberg is a professor at the Department of Comparative Politics at the University of Bergen and at the Department of Political Science at the Mid Sweden University. Email: Stefan.Dahlberg@miun.se.

Magnus H Iversen is a postdoctoral research fellow at the Department of Information Science and Media Studies at the University of Bergen. His research interests are located in the fields of political communication and rhetoric, argumentation theory and journalism studies. Email: Magnus.Iversen@uib.no.

Mikael P Johannesson is a PhD candidate the Department of Comparative Politics, and in the Digital Social Science Core Facility (DIGSSCORE), University of Bergen. Email: mikj@norceresearch.no.

Silje Nygaard is a PhD Candidate at the Department of Information Science and Media Studies, University of Bergen. Email: Silje.nygaard@uib.no.

References

Abdenour

McIntyre

Dahmen

(2020) Seeing eye to eye: A comparison of audiences’ and journalists’ perceptions of professional roles and how they relate to trust. Journalism Practice. Epub ahead of print 30 January 2020. DOI: 10.1080/17512786.2020.1716828.

Appelman

Sundar

(2016) Measuring message credibility: Construction and validation of an exclusive scale. Journalism and Mass Communication Quarterly 93: 59–79.

Barthel

Mitchell

(2017) Americans’ attitudes about the news media deeply divided along partisan lines. Pew Research Center 10: 49.

Chung

Nam

Stefanone

(2012) Exploring online news credibility: The relative influence of traditional and technological factors. Journal of Computer-Mediated Communication 17: 171–186.

Daniller

Allen

Tallevi

, et al. (2017) Measuring trust in the press in a changing media environment. Communication Methods and Measures 11: 76–85.

Engelke

Hase

Wintterlin

(2019) On measuring trust and distrust in journalism: Reflection of the status quo and suggestions for the road ahead. Journal of Trust Research 9(1): 66–86.

Fisher

(2016) The trouble with ‘trust’ in news media. Communication Research and Practice 2: 451–465.

Gaziano

McGrath

(1986) Measuring the concept of credibility. Journalism Quarterly 63: 451–462.

Gladney

(1996) How editors and readers rank and rate the importance of eighteen traditional standards of newspaper excellence. Journalism and Mass Communication Quarterly 73(2): 319–331.

10.

Greenberg

Razinsky

(1966) Some effects of variations in message quality. Journalism Quarterly 43: 486–492.

11.

Gronke

Cook

(2007) Disdaining the media: The American public’s changing attitudes toward the news. Political Communication 24: 259–281.

12.

Hallin

Mancini

(2004) Comparing Media Systems: Three Models of Media and Politics. Cambridge, MA: Cambridge University Press.

13.

Hanitzsch

Van Dalen

Steindl

(2017) Caught in the nexus: A comparative and longitudinal analysis of public trust in the press. International Journal of Press/Politics 23(1): 3–23.

14.

Heise

Loosen

Reimer

, et al. (2014) Including the audience. Journalism Studies 15(4): 411–430.

15.

Hovland

Lumsdaine

Sheffield

(1949) Experiments on Mass Communication. Princeton, NJ: Princeton University Press.

16.

Kiousis

(2001) Public trust or mistrust? Perceptions of media credibility in the information age. Mass Communication and Society 4: 381–403.

17.

Kohring

(2019) Public trust in news media. In: Vos

Hanusch

Dimitrakopoulou

, et al. (eds) The International Encyclopedia of Journalism Studies, pp.1–6.

18.

Kohring

Matthes

(2007) Trust in news media: Development and validation of a multidimensional scale. Communication Research 34: 231–252.

19.

Krosnick

(1999) Survey research. Annual Review of Psychology 50: 537–567.

20.

Ladd

(2012) Why Americans Distrust the News Media and How It Matters. Princeton, NJ: Princeton University Press.

21.

Lazarsfeld

(1944) The controversy over detailed interviews—An offer for negotiation. Public Opinion Quarterly 8: 38–60.

22.

Mayer

Davis

Schoorman

(1995) An integrative model of organizational trust. Academy of Management Review 20: 709–734.

23.

Metzger

Flanagin

Eyal

, et al. (2003) Credibility for the 21st century: Integrating perspectives on source message and media credibility in the contemporary media environment. Annals of the International Communication Association 27: 293–335.

24.

Metzger

Flanagin

Medders

(2010) Social and heuristic approaches to credibility evaluation online. Journal of Communication 60(3): 413–439.

25.

Meyer

(1988) Defining and measuring credibility of newspapers: Developing an index. Journalism and Mass Communication Quarterly 65(3): 567–574.

26.

Meyer

(1974) Media credibility: The state of the research. Public Telecommunications Review 2: 48–52.

27.

Newman

Fletcher

(2017) Bias Bullshit and Lies: Audience Perspectives on Low Trust in the Media. Oxford: Reuters Institute.

28.

Newman

Fletcher

Levy

, et al. (2017) The Reuters Institute Digital News Report 2017. Oxford: Reuters Institute.

29.

Nielsen

(2016) Folk theories of journalism: The many faces of a local newspaper. Journalism Studies 17(7): 840–848.

30.

Norwegian Citizen Panel Round 9 (2017) Data collected by Ideas2evidence for Elisabeth Ivarsflaten, the University of Bergen. Fourth NSD edition, Bergen 2021.

31.

Prochazka

Schweiger

(2019) How to measure generalized trust in news media? An adaptation and test of scales. Communication Methods and Measures 13(1): 26–42.

32.

Reja

Manfreda

Hlebec

, et al. (2003) Open-ended vs. close-ended questions in web questionnaires. Developments in Applied Statistics 19: 159–177.

33.

Roberts

(2010) Correlations among variables in message and messenger credibility scales. American Behavioral Scientist 54: 43–56.

34.

Roberts

Stewart

Tingley

, et al. (2014) Structural topic models for open-ended survey responses. American Journal of Political Science 58: 1064–1082.

35.

Roper

(1985) Public Attitudes Toward Television and Other Media in a Time of Change: The Fourteenth Report in a Series. New York: Television Information Office.

36.

Schuman

Presser

(1979) The open and closed question. American Sociological Review 44: 692–712.

37.

Shaw

(1973) Media credibility: Taking the measure of a measure. Journalism Quarterly 50: 306–311.

38.

Skjervheim

Høgestøl

(2017) Norwegian Citizen Panel Methodology Report Wave 9. Bergen, Norway: Ideas 2 Evidence.

39.

Strömbäck

Tsfati

Boomgaarden

, et al. (2020) News media trust and its impact on media use: Toward a framework for future research. Annals of the International Communication Association 44(2): 139–156.

40.

Suiter

Fletcher

(2020) Polarization and partisanship: Key drivers of distrust in media old and new? European Journal of Communication 35(5): 484–501.

41.

Swart

Peters

Broersma

(2017) Repositioning news and public connection in everyday life: A user-oriented perspective on inclusiveness engagement relevance and constructiveness. Media Culture and Society 39(6): 902–918.

42.

Tsfati

Ariely

(2014) Individual and contextual correlates of trust in media across 44 countries. Communication Research 41: 760–782.

43.

Tsfati

Cappella

(2005) Why do people watch news they do not trust? The need for cognition as a moderator in the association between news media skepticism and exposure. Media Psychology 7: 251–271.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.28 MB