The influence of negativity in news coverage on governmental trust during the COVID-19 pandemic

1 Introduction

In times of societal crisis, the success of measures taken by governments to counter the crisis depends largely on the trust citizens have in institutions to handle the problem, and the COVID-19 pandemic serves as a prime example of such a crisis where compliance with countermeasures was positively associated with trust in those issuing the measures (for an overview, see e.g., Devine, 2024). Following the outbreak, governments worldwide experienced an increase in public support, with trust and approval ratings spiking – a phenomenon commonly referred to as the rally-around-the-flag effect (Mueller, 1970; see also Baekgaard et al., 2020; Esaiasson et al., 2021; Hegewald & Schraff, 2024). However, these initially high levels of trust and approval began to decline as the crisis continued (e.g., Johansson et al., 2021; Satherley et al., 2022). In this paper, we aim to analyze how media coverage and individuals’ perceptions of the ongoing crisis contribute to the decline of positive evaluations of the government in times of persisting crises.

While the emergence of the rally-around-the-flag effect is well studied, studies on factors affecting the longevity of the effect are relatively few (Satherley et al., 2022), suggesting that waning crisis salience, increasing polarization over the issue, return of critical voices from the opposition and in media coverage, or individuals’ perceptions of the development of the ongoing crisis, could play a role. Regarding the latter, Johansson et al. (2021), for example, emphasized that perceptions of how a country had been affected by the pandemic contributed to reversing the rally effect: The more individuals perceived COVID-19 to have a negative impact on the country, the less they thought the government was “doing a good job.” How such perceptions are formed has, so far, not been explicitly considered in the context of the rally-around-the-flag effect, though. The literature on societal beliefs can inform this question as it emphasizes that societal perceptions are shaped through the interplay of interpersonal communication, ideological predispositions, and media exposure (Mutz, 1998). Building on this understanding, research has shown that it is not only exposure to the news itself but also the news media’s “emphasis upon favorable or unfavorable developments or indicators” (Shah et al., 2002, p. 342) that shapes individuals’ perceptions of issues. In other words, the tone of news coverage can influence beliefs about an issue. Research has particularly focused on negative tone in the news and found that more exposure to negative tone results in more negative perception of the issue – a finding replicated for various issues (e.g., Boukes et al., 2021; Damstra et al., 2021; Eberl et al., 2018). Against these backdrops, we suggest that trust in the government after the initial phase of the COVID-19 pandemic is shaped by exposure to negativity in media coverage, mediated by perceptions of the development of the ongoing pandemic.

At the same time, research on media effects underscores that effects of exposure to negativity do not occur in isolation but are shaped by various personal factors (Soroka and Krupnikov, 2021). One crucial moderating factor, for example, is personal experience with a societal problem and its resonance with media coverage (Gerbner et al., 1986). In this paper, we test whether personal resonance functions as “a ‘double dose’ of meaning” (Potter, 2014, p. 1019), moderating the effect of exposure to negative news coverage on the formation of perceptions of the COVID-19 pandemic, and in turn, on the trust in the government to handle the crisis.

In this study, we thus aim to address the questions of (1) how negativity in media coverage influences the development of governmental trust during a crisis through perceptions of this crisis, and (2) how personal experience with the crisis moderates this relationship. To answer the research questions, we make use of two waves of a probability-recruited panel survey, as well as an accompanying quantitative content analysis of the coverage of the most important news outlets in Sweden – a country whose government received worldwide attention for its less restrictive and voluntary-based approach to managing the pandemic (Saunes et al., 2022) and where the rally-around-flag effect was at a turning point at the beginning of our data collection (Johansson et al., 2021). The findings will help to underscore the importance of trust in government during crises, particularly in cases where government interventions diverge from conventional strategies (Johansson and Vigsø, 2021).

2 Trust in institutions during crises

When a country is affected by an external crisis, public support for its leaders often intensifies, a response commonly known as the rally-around-the-flag effect. This phenomenon describes a temporary surge in approval ratings for a leader or government as citizens unify in the face of a national crisis, conflict, or external threat (Mueller, 1970). Research on the effect has put forward two types of reasons why a surge in trust and approval of those in power can be observed during such times (Baker and Oneal, 2001). On the one hand, a crisis can activate in-group identification, leading people to rally around national symbols like the president, as they seek solidarity and reassurance. This reaction is partly driven by a heightened need for security and stability in uncertain times, when citizens view leaders as protectors. On the other hand, the role of opinion leader communication is emphasized. During the beginning of a crisis, critical voices, such as those from the opposition, tend to fall silent and avoid commenting on the government’s policies (Hetherington and Nelson, 2003). In times when criticism is lacking, the news media are also “likely to report rally events in a more one-sided fashion, one more favorable to the president, than would be the case on issues of domestic policy or routine international events” (Baker and Oneal, 2001, p. 668; in the context of COVID-19, see e.g., Hayek, 2024). As a result, the public lacks access to diverse perspectives and cues from opposing political or ideological voices, which in turn can lead them to align more readily with those in power.

While the rally-around-the-flag effects were initially described for wars and in the context of the United States (Mueller, 1970), an increase in trust toward institutions could also be observed at the onset of, for example, health crises, such as the COVID-19 pandemic. A significant increase in political trust and approval ratings for political leaders was observed across various countries (e.g., Baekgaard et al., 2020; Bol et al., 2021; Cardenal et al., 2021; Dietz et al., 2023; Hegewald and Schraff, 2024; Kritzinger et al., 2021), including Sweden – the context of our study. In this Nordic country that gained worldwide attention due to its unique approach to handling the pandemic (Johansson and Vigsø, 2021), trust in the government was generally stable and typical for a non-crisis period; yet it saw a significant increase in March 2020 as the pandemic reached Sweden, and the government, alongside the Public Health Agency, announced a response strategy that garnered initial public support: For example, Esaiasson et al. (2021) found that trust in Swedish government authorities increased significantly between the initial and the acute phase of the pandemic (see also, Johansson et al., 2023; Nielsen and Lindvall, 2021). While some argued that this effect was independent of factors such as political leaning, others found that the rally-around-the-flag impact in Sweden was influenced by political ideology from the outset (Nielsen and Lindvall, 2021) or that this moderating effect only emerged after the first few months of the pandemic (Johansson et al., 2021). By late April – the start of this study – trust remained relatively high, though emerging criticisms in public support were becoming more visible. In the following months, both approval of the government’s performance (Johansson et al., 2021) and trust in the government (Johansson et al., 2023) began to wane and the rally-around-the-flag effect disappeared.

3 The role of societal perceptions in shaping trust

The rally-around-the-flag effect has not only been convincingly described in the literature; researchers also continuously underscore that the phenomenon tends to be short-lived (Dinesen and Jæger, 2013). However, while the majority of studies have focused on describing the increase of positive evaluations of or trust in institutions and on explaining the mechanisms behind these observations, less is known about the factors that contribute to the decline of the approval levels to the baseline. Indeed, “the duration of rally effects – both in general and as they relate to COVID-19 – is relatively underexplored” (Satherley et al., 2022, p. 2). In this study, we focus on individuals’ perceptions of the development of the pandemic. Societal beliefs or perceptions are specifically a type of belief that individuals have about issues of national concern rather than about their own circumstances (Mutz, 1998; Shehata, Andersson, et al., 2021). These beliefs are often related to issues such as economic policy, public health, or education. Against these understandings, a societal belief about the pandemic could be, for example, the belief that Sweden has managed the COVID-19 pandemic better than other EU countries, or that the country had been well prepared for the crisis (Johansson et al., 2021; Shehata, Glogger, et al., 2021).

These types of beliefs can significantly influence trust in or evaluations of those in power. For example, Kritzinger et al. (2021) found that in Austria, when individuals perceived pandemic measures shifting from appropriate to too extreme, trust in the government declined. Similarly, Johansson et al. (2021) examined perceptions of how Sweden was affected by the pandemic, and their impact on governmental approval. They found that evaluations of governmental performance early in the pandemic were largely shaped by perceptions of the country’s handling of the crisis, and that “perceptions of how Sweden managed the crisis is an important independent driver of the declining rally-around-the-flag effect” (Johansson et al., 2021, p. 331, emphasis added by authors). Building on these findings, we focus on trust as a distinct yet related construct that reflects confidence in the government’s long-term capability and governance.

4 The role of media use in forming beliefs

While these studies provided us with initial ideas about the role of societal perceptions of the development of a crisis on evaluations of the government, they did not consider how these perceptions are formed in the first place. Johansson et al. (2023) point out that “perceptions of success or failure are driven by contextual factors, such as media coverage, personal media consumption, and predispositions” (p. 289). As discussed earlier, media coverage that lacks negative and critical voices has been suggested as one explanation for the rally-around-the-flag effect (e.g., Baker and Oneal, 2001; Groeling and Baum, 2008). In this study, we argue that media coverage can indirectly affect the decline of trust during a crisis by influencing individuals’ perceptions of the development of the crisis.

Indeed, this assumption is supported by the long-standing research tradition of public opinion formation which emphasizes the role of the news media in the origins of societal perceptions, since societal issues are often complex and direct personal experience can be limited (Lippmann, 1922; Mutz, 1998). Building on this, scholars have identified specific characteristics of news content that influence public belief formation (Shehata, Andersson, et al., 2021). One characteristic of news that has consistently been found to influence societal beliefs, attitudes, or behavior, is negativity. Being hardwired to focus on negative events in our environment, humans exhibit not only a negativity bias when selecting news (Soroka, 2006). They also process news with a bias (Meffert et al., 2006). While negativity can take various forms in news content, we focus on negativity as a frame-related content feature in which the negative tone of the news is determined by, for example, disastrous, threatening, or criticizing elements (for an overview, see Lengauer et al., 2012). A negative tone in news coverage affects individuals’ beliefs about a societal issue in such a way that these beliefs also become more negative (de Vreese and Boomgaarden, 2003). Comparable patterns were described for various issues, such as the economy (e.g., Boomgaarden et al., 2011; Boukes et al., 2021; Damstra et al., 2021) or immigration (Eberl et al., 2018).

News coverage in Swedish media^[1] at the beginning of the pandemic was in favor of the government’s approach (Johansson et al., 2021; Skog and Lundström, 2022). At the same time, news users perceived the coverage to be alarmist (Winters et al., 2022). This period was followed by a rise of critical voices in the news coverage (Giritli Nygren and Olofsson, 2021; Johansson et al., 2021; Skog and Lundström, 2022). This holds true also for journalists asking questions in the “question and answer” sessions after the daily press conferences by Sweden’s Public Health Agency (Dahlgren, 2021). The salience of the issue in news coverage also declined toward the end of the acute phase of the pandemic after the first wave (Esaiasson et al., 2021).

5 Personal experience as a moderator

The literature on sociotropic beliefs suggests that news media coverage plays a decisive role in shaping people’s perceptions of societal issues. Yet, such influences have increasingly been conceptualized not to be universal but rather of a moderated nature. Connecting to the growing literature on the conditionality of media effects (Valkenburg and Peter, 2013), one factor of particular interest is the concept of personal experience. Here, media effects are suggested to be moderated by the congruence between how the news media cover a problem and how individuals personally experience the issue. To be more precise, when personal experience and media coverage are congruent, media effects are expected to be stronger. In contrast, incongruence between the two can weaken the media effect (Valkenburg and Peter, 2013). The importance of personal experience as a moderator of media effects, as well as the role of congruence/incongruence, has been demonstrated in the context of several early media effect theories, such as cultivation theory through the concept of experience resonance (Gerbner et al., 1986). It refers to a situation in which individuals get a “double dose” of information leading to a reinforcement of the media effect (Potter, 2012).

The concept of issue obtrusiveness provides a complementary layer to understanding how personal experience moderates media effects. Obtrusiveness refers to the degree to which an issue can be directly experienced by individuals – obtrusive issues being those that can be directly experienced, while unobtrusive issues being those that individuals have only indirect experienced with, for example, through media coverage (Zucker, 1978). Two opposing perspectives on the role of issue obtrusiveness in media effect research co-exist (for a detailed discussion, see e.g., Lee, 2004). On the one hand, and in line with the first introduction of the concept, obtrusiveness can reduce media effects since individuals form opinions about an issue based on their direct experiences with the issue. On the other hand, obtrusiveness may strengthen media effects since the direct experience with the problems primes individuals’ attention to the issue and makes it less effortful to process the information provided in news coverage (Mutz, 1998). COVID-19 exemplified an issue that was both obtrusive and unobtrusive, depending on personal circumstances and the progression of the pandemic (Schaller et al., 2024). For those who, for example, experienced illness or job loss, the pandemic was highly obtrusive. Conversely, for individuals who for example remained healthy, continued working remotely, or had minimal personal exposure, COVID-19 was less obtrusive.

6 Research question and hypothesis

Against these theoretical and empirical backdrops, we pose the following hypothesis and ask the following research questions, which are summarized in Figure 1. First, we assume that both exposure to negativity in news coverage and perceptions of the development of the pandemic play a role in the decline of trust in the government after the initial rally-around-the-flag effect at the beginning of the crisis. We suggest that the best way to understand the relationship between media exposure and government approval is as an indirect one:

H1: The effect of exposure to negative news coverage on trust in the government is mediated by an individual’s societal beliefs about the pandemic.

Second, we address the role of personal experience with the pandemic, focusing on the importance of personal experience with a societal issue in forming beliefs about this issue, asking:

RQ1: How does experience with COVID-19 moderate the indirect effect of media exposure on governmental approval through beliefs about the pandemic?

Figure 1:

Theoretical model of indirect effects of exposure to negative media coverage on governmental trust through perception of the development of the COVID-19 pandemic moderated by personal experience.

7 Method

We rely on two waves of a probability-recruited panel survey and a content analysis of six important news outlets in Sweden. The data are part of a larger research project comprising a total of five waves that were collected during the first year of the COVID-19 pandemic in Sweden. Since we are particularly interested in the indirect effect of media use on the potential decrease in trust in the government at the beginning of the pandemic, which was characterized by belief formation and opinion shifts after a short time of unanimous support for Sweden’s handling of the crisis, we decided to focus on the first two waves of the panel. We briefly address these peculiarities of the Swedish approach toward the COVID-19 pandemic in the following, before presenting the method applied in this study.

8 Results

The starting point of our analysis was the observation that rally effects tend to diminish over time. Before testing the hypothesis and addressing the research question, we thus briefly examine the trajectory of trust in the government. As shown by the descriptive values presented in the method section, people reported lower trust in the government’s ability to handle the crisis during the second wave compared to the first. Similarly, participants indicated that COVID-19 had a lower impact on their daily lives during the second wave. To assess how exposure to negativity in media coverage, via societal perceptions, influenced the diminishing of the rally-around-the-flag effect, a mediation analysis was conducted. The analysis controlled for political leaning, gender, age, education level, personal experience with COVID, as well as previous trust in government and previous perception of COVID-19 development.

First, we found that exposure to negativity in media coverage influenced the perception of the COVID-19 development significantly (b = 0.04, p = .002). This indicates that higher exposure to negative media content is associated with a more negative perception of COVID-19 development. Second, the results indicate that the perception of COVID-19 development influenced trust in government in such a way that a more negative perception of COVID-19 development was associated with lower trust in government (b = 0.34, p < .001). Third, we found that the indirect effect of exposure to negativity in media coverage on trust in government through perception of COVID-19 development was significant (b = 0.01, BootSE = 0.005, BootLLCI = 0.01, BootULCI = 0.02). At the same time, the direct effect of exposure to negativity in media coverage on trust in government, controlling for the mediator and covariates, was not significant (b = 0.02, p = .14). These results suggest a full mediation effect, where exposure to negativity in media coverage influences trust in government indirectly through perception of COVID-19 development. Specifically, higher exposure to negative media content leads to a more negative perception of COVID-19 development, which in turn leads to lower trust in government, allowing us to confirm H1 (Table A5 in the appendix).

Figure 2:

Indirect effect of exposure to negativity in news coverage on governmental trust through perceptions of the development of the COVID-19 pandemic.

Note: Unstandardized coefficients are reported. Age, gender, education ideology, political interest, governmental trust at W1, perceptions at W1, and direct experience with COVID-19 were controlled for. Dotted lines indicate non-significant effects. *p < .05, **p < .01, ***p < .001.

Last, we present the results of a moderated mediation to answer research question RQ1, following the same modeling approach as above but including additionally an interaction term between exposure to negativity and personal experience with the pandemic. We found that the interaction between exposure to negative media coverage and direct experience with COVID-19 was not significant. When assessing how personal experience moderated the indirect effect of media exposure on governmental approval, we found the following: Comparing individuals with a relatively low direct COVID-19 experience (1 SD below the mean; b = 0.01, BootSE = 0.006, BootLLCI = –0.006, BootULCI = 0.02), those with a relatively medium direct COVID-19 experience (at the mean; b = 0.01, BootSE = 0.004, BootLLCI = 0.006, BootULCI = 0.02), and those with a relatively high direct COVID-19 experience (1 SD above the mean; b = 0.02, BootSE = 0.007, BootLLCI = 0.004, BootULCI = 0.03), we found only a significant moderation of personal experience in the two latter groups. This indicates that the moderation of the indirect effect of exposure to negativity in news coverage on governmental approval via the perception of the COVID-19 development can be found for those who have been more affected by the pandemic in their daily life. Yet, when contrasting these effects, we could not find significant differences between these conditional indirect effects. We, thus, conclude that personal experience does not moderate the indirect effect of media exposure on governmental approval via COVID-19 perception (Table A6 in the appendix).

9 Discussion

Trust in institutions is a crucial ingredient for successfully managing a crisis, such as a global pandemic (Devine, 2024). While we have a good understanding of the factors that contribute to an increase of trust at the onset of a crisis, less is known about the factors that influence the decrease of trust in institutions during an ongoing, prolonged crisis (Satherley et al., 2022). Building upon the literature on media effects and trust in the context of the rally-around-the-flag effect, we address the question of how exposure to negativity in media coverage, personal experience with the COVID-19 pandemic, and societal perceptions of the crisis contributed to the waning trust in the Swedish government after an increase at the beginning of the pandemic. We found that exposure to negative news coverage led to agreement that the development of the pandemic in Sweden was heading in the wrong direction. In turn, this perception negatively affected the trust in the government to handle the pandemic. However, personal experience with the crisis did not moderate this relationship.

Two main conclusions can be drawn from our results. First, our findings contribute to the literature on the rally-around-the-flag effect by providing insight into the factors that contribute to reversing the observed surge in trust. While other studies have focused foremost on describing the rally-around-the-flag effect and its antecedents (e.g., Baker and Oneal, 2001; Lambert et al., 2011), we found that societal beliefs, shaped by negativity in media coverage, significantly influenced the decline of trust after the initial phase of the COVID-19 pandemic. On the one hand, this connects to research that has demonstrated perception of the ongoing pandemic led to decreasing positive evaluations of the government (Johansson et al., 2021). On the other hand, our findings illustrate how such perceptions are formed by emphasizing the role of the media. This connects to the assumption that the lack of critical, negative voices against the government contributes to the rise of positive evaluations and trust in the government when a country is affected by a sudden external crisis (Hetherington and Nelson, 2003).

Second, we observed that personal experience with the COVID-19 crisis neither influenced the relationship between exposure to negativity and perceptions of the ongoing crisis, nor the overall mediation. This is contrary to other studies that found that personal experience had a significant influence on, for example, concerns about specific aspects of the development of the crisis (e.g., Schaller et al., 2024). Our findings, in contrast, point to an influence of media-driven perceptions on trust in the government to handle the crisis. One explanation for this might be that the panel study took place at the beginning of the pandemic, when life was dominated by great uncertainty (Brandt et al., 2023) – but direct, personal experience with COVID-19 as a potentially life-threatening disease was still limited. In other words, the pandemic was still an unobtrusive issue (Schaller et al., 2024; Zucker, 1978) regarding which individuals relied on media coverage and not personal experience with the issue in the formation of beliefs.

Some limitations should be considered at this point. First, while we aimed to test resonance effects, the item to measure personal experience was not worded for valence but only assessing the magnitude of the pandemic on an individual’s personal life. Even if it is highly unlikely given the number of studies showing the negative effects on various aspects of people’s lives (e.g., Gröndal et al., 2021), we cannot rule out that individuals experienced also positive effects during the pandemic (such as an increased work-life balance due to remote work, see e.g., Brulin et al., 2022). Similarly, we used only one broad item to assess personal experience with the pandemic. However, in light of findings highlighting that the effects of personal experience also depend on the specific issue area (Schaller et al., 2024), future studies should not only assess the valence of personal experiences. They should also consider different dimensions of being affected by the issue, such as personal economic circumstances or job-related impacts. Second, we included only six of the largest national news outlets in Sweden in our study, ignoring, for example, social media. This poses a limitation to our analysis, since studies on news use during the pandemic pointed out that especially social media gained importance (Van Aelst et al., 2021). Third, the results of the attrition analysis indicate that certain demographics – particularly older individuals and men – were more likely to remain in the study, leading to a potential overrepresentation of these groups in the Wave 2 sample. Even if we controlled for age and gender in the analysis, this might impact conclusions drawn about trust and perception shifts over time given that especially older individuals were more likely to be affected by the pandemic. Finally, our main dependent variable in this study, governmental trust, does not translate automatically into how people vote in elections (Fournier et al., 2003). Future studies should also study election behavior to assess whether the evaluation of government performance during a crisis could have implications for the ballot.

In terms of the generalizability of our findings, it is worth noting that Sweden is typically considered a high-trust country, particularly with respect to institutional trust (Organisation for Economic Co-operation and Development, 2024), which may play a significant role in various types of rally effect dynamics. On the one hand, this high baseline institutional trust could serve as a protective factor against the effects of negative news coverage as citizens are more likely to rely on established beliefs or institutional credibility rather than external media cues (for so-called “buffering effects” of baseline attitudes, see e.g., Jonkman et al., 2020). On the other hand, the trust dynamics documented here refer more to specific forms of trust as approval of incumbent office holders’ handling of the pandemic, than to diffuse support for institutions, which are generally more resilient to change (Norris, 2011). Thus, while it may be the case that traditionally high levels of institutional trust are less sensitive to negative news coverage, evaluations of government approval are more affected. In that sense, high institutional trust may not necessarily protect against influences of more specific forms of government trust when media coverage becomes negative. Although our findings demonstrate that exposure to negative news coverage can still shape specific forms of trust by influencing perceptions of the pandemic’s development, the effects observed here, particularly those of media coverage, are relatively small. However, they underscore the importance of media negativity in shaping trust during prolonged crises.