COVID-19: Examining the Roles of Traditional and Social Media Attention in the Amplification of Risk

Abstract

This study used a nationally representative survey of U.S. residents (N = 1,969) to examine whether attention to information about COVID-19 in traditional news media sources and on social media correlated with a higher perceived risk of personal and public harm. As anticipated, we found a positive association between attention to COVID-19 information in traditional news media sources and the perceived risks of COVID-19. We also found a positive association between attention to COVID-19 social media posts and perceived risks but only among Republicans. Other predictors of increased risk perception included age, being female and awareness of a local stay-at-home order.

Keywords

COVID-19 risk perceptions social amplification of risk social media traditional media political ideology

In March 2020, the United States declared COVID-19 a national emergency, and attention to news spiked. A poll conducted in late March found that the number of Americans paying a ‘great deal’ of attention to the news reached levels not seen in 3 years, with attention to local and national news increasing by 22 and 13 percentage points between December 2019 and March of 2020 (Ritter, 2020a). Meanwhile, COVID-19 dominated Americans’ news diets and social media platforms. In late March, 92% of Americans said they were following COVID-19 news fairly or very closely (Jurkowitz & Mitchell, 2020a), and in late April, 47% of social media users in the United States said most or almost all of what they were seeing on social media was COVID-related (Ritter, 2020b).

Past research suggests exposure to media coverage of disease outbreaks tends to amplify perceptions of risk (e.g., Thompson et al., 2017; Young et al., 2008). Similar research findings are beginning to emerge regarding the exposure to disease information through social media (Choi et al., 2017; Yoo et al., 2020). Based on previous research, we might expect to find a positive relationship between individuals’ exposure to COVID-19 information through traditional or social media and perceptions of risk. Compared with previous disease outbreaks, however, the U.S. experience with COVID-19 stands out in ways that could have important implications for perceptions of risk and the role of media in shaping those perceptions. In general, risk messages can raise public awareness, but they can also oversimplify or misrepresent risks (Smith & McCloskey, 1998). The unfolding of the COVID-19 pandemic in the United States triggered unprecedented but inconsistent government response measures muddled by political division (e.g., Coppins, 2020; Motta et al., 2020). Communication about the disease has been similarly inconsistent with mixed messaging about risks and efficacy of protective measures (e.g., Watson, 2020). Conflicts were apparent among different federal and state government press briefings and other sources, perhaps making it difficult for the public to assess the potential risk. News coverage and social media posts about the pandemic were also awash with mixed messages about the risks of COVID-19.

In the United States, perceptions about the risks of COVID-19 quickly became defined along partisan lines. A national poll conducted in late March 2020 found that Republicans were less likely than Democrats to view the outbreak as a major threat and more likely to view people as overreacting. Meanwhile, Democrats were more likely to think people across the country were not taking the pandemic seriously enough (Green & Tyson, 2020). These partisan differences in concern about the risks associated with the virus would continue to widen over the following few months (Pew Research Center, 2020).

The purpose of this research project was to examine how traditional and social media and political affiliation have shaped people’s perceptions of the risks of COVID-19. Our examination of these factors is guided by the social amplification of risk framework (SARF) (Kasperson et al., 1988). The framework suggests the process of communicating about risk events involves several factors that can contribute to amplification or attenuation of risk perceptions (Kasperson & Kasperson, 1996). We elaborate on this framework, the COVID-19 media landscape and explain hypotheses and research questions in the following literature review.

Literature Review

Social Amplification of Risk Framework

Kasperson et al. (1988) introduced the SARF to help us think about how elements in the message system might amplify or attenuate public perceptions of risk. Perceptions of risk involve public assessments about the potential to experience a range of negative outcomes, such as injury, illness, disease and death (Slovic, 2000). Risk events are recognised as having a combination of objective and subjective qualities (Renn et al., 1992). The subjective nature of risk means risk can be communicated and experienced differently because communication is informed by culture, institutions, social groups, and individual psychological and social characteristics.

We are exposed to some risk events directly. However, many risk events are experienced indirectly through electronic media and actors in our social networks (Kasperson & Kasperson, 1996). The SARF refers to sources of indirect experiences as amplification stations, which include experts such as scientists, public institutions, social organisations, reporters within media companies and peers (Kasperson et al., 1988). People’s risk assessments are also influenced by factors such as the amount of media coverage given to an event, the amount of information provided about the event and the way communicators frame the event (e.g., Kasperson et al., 1988; Rosa, 2003; Wardman, 2008).

Two of the most common channels through which people receive information about risk are mass media (such as television and newspaper organisations) and social networks through which people can share information with people who have varying degrees of direct connections with one another (e.g., Pew Research Center, 2019). Increasingly, an important way in which we exchange information with our social networks is through social media. We will address these channels starting with the traditional media.

Traditional Media

Engagement with traditional media can play a key role in the public’s risk assessment process. Traditional media organisations facilitate public communication in a couple of ways. First, they scan the informational horizon and select topics to share with the public while applying news judgement about the salience of the topic (e.g., McCombs, 2006). The topics deemed more salient are given greater prominence in the media (such as through lead stories in televised newscasts and front pages of papers). Second, the way the media frame the issue in the coverage of issues highlights perspectives (e.g., Price et al., 1997). Through these roles, the media influence the public debate on the issues (e.g., Hornig, 1993).

Many studies have linked media coverage and attention to media to amplified perceptions of risk, such as the perceived risk of genetically modified foods (e.g., Frewer et al., 2002), environmental risks (e.g., Zhao et al., 2011), as well as the perceived risks of disease threats (e.g., Ackerson & Viswanath, 2010; Hart et al., 2011; Thompson et al., 2017; Young et al., 2008). With respect to the perceived threat of diseases, for example, researchers have linked greater attention to environmental stories on TV news and health media to a greater perceived risk of zoonotic disease (Ackerson & Viswanath, 2010; Hart et al., 2011) and attention to Ebola-related media and the perceived risk of contracting Ebola (Thompson et al., 2017). Similarly, Young et al. (2008) found perceptions of diseases that are frequently covered in media are perceived as ‘worse’ than diseases that do not receive much media coverage, even if the inverse is true. In general, attention to media is a relevant factor to examine because it reflects the roles of information seeking and scanning that are associated with health risks, particularly threats from infectious diseases (e.g., Pew Research Center, 2016; Ruppel, 2016). Additionally, attention to media has been observed to have direct effects on risk perceptions while media exposure did not have such effects (e.g., Slater et al., 2007, 2009). With this prior body of empirical evidence in mind, we propose the following hypothesis:

H₁: Greater attention to traditional media for information about COVID-19 will be positively associated with higher levels of perceived risk.

Note that all hypotheses and research questions we present in this literature review will be tested while controlling for all other variables included in this study and which will be discussed in more detail below. These include six demographic variables, political affiliation, government policy and attention to social media.

Social Media

The role of the media in shaping public perceptions of risk has become increasingly complex with the growth of social media. In its early stages, the online information environment was a place where users searched for and consulted information (Web. 1.0). Since then, it has evolved into a world where users generate content themselves (Web. 2.0) (Sutton & Veil, 2017). The traditional one-way flow of information has been replaced by a communication environment in which social media allow individuals to participate in an online ‘public square’, where users interact with one another to disseminate information and generate their own messages (Rutsaert et al., 2013; Sutton & Veil, 2017).

Individuals are more likely to encounter non-official information through social media platforms, where social networks tend to be dominated by friends and family (Rutsaert et al., 2013). In traditional media, a formal system of decision-making influences how and what information is distributed to the public. Information in social media is distributed through networks of people, each with varying degrees of power to boost attention to risk-oriented messages (e.g., Chew & Eysenbach, 2010; Choi et al., 2017). For example, Zhang and Cozma (2022) found that trust in the Twitter accounts of regular people predicted the sharing of COVID-19 related information, whereas trust in traditional media accounts did not. And unlike traditional media, social media lacks the same level of gatekeeping, and therefore can be a greater source for misinformation (Ng et al., 2018; Wang et al., 2019).

Furthermore, traditional information sources lose control of how their messages are shared and interpreted, as users spread messages with minimal contact with the message sources. While news producers such as journalists use Twitter as a tool for sharing their work (Broersma & Graham, 2012), those messages become interspersed among messages from political advocates or candidates and are interpreted through the observations and reactions of independent individuals (Graham et al., 2016; Newman et al., 2012). Therefore, in the parlance of SARF, engagement in social media leads to a wider variety of potential amplification stations (Fellenor et al., 2018).

Like traditional media, recent research suggests social media may also serve to amplify risk-oriented communications (e.g., Ng et al., 2018; Oh et al., 2020; Yang et al., 2016; Yoo et al., 2020). In a study on food safety issues, for example, Yang et al. (2016) found a positive link between greater attention to food safety issues on social media and greater perceived risks associated with food safety. In a study on carcinogenic hazards, Yoo et al. (2020) found a link between attention to information about carcinogenetic hazards on social media and risk perceptions associated with these hazards, but this link was limited to social media that primarily connect users based on shared interests (e.g., Twitter and Instagram). Finally, with respect to perceptions of risks associated with pathogenic disease, Oh et al. (2020) found a positive correlation between the perceived risk of contracting Middle East Respiratory Syndrome (MERS) coronavirus and self-reported exposure to MERS information on social media. Given these findings on social media and risk perceptions, the following hypothesis is proposed:

H₂: Greater attention to information about COVID-19 in social media will be positively associated with higher levels of perceived risk.

Amplification and Political Affiliation

Although media practitioners and the channels through which messages are delivered are important aspects in the SARF, other factors such as individual and social values can also exert influence on each part of the framework. When people, whether experts or non-experts, communicate about risks, their characterisations and analyses are influenced by their personal opinions and values (e.g., Hansson, 2010). For example, when people who are members of social and/or political organisation communicate about a risk event, those affiliations can influence how the risks are defined as well as the way the messages are distributed. When social or political organisations are in conflict over a risk event, that conflict can increase media coverage and influence interpersonal discussions within and outside social media (Frewer et al., 1993). Additionally, an individual’s affiliation with social and political organisations can influence how risk messages are interpreted, thereby reflecting group values that can amplify or attenuate perceptions of risk (e.g., Kasperson & Kasperson, 1996; Nordenstedt & Ivanisevic, 2010).

Examinations of public responses to epidemics have revealed differences between people with different political associations. For example, Johnson (2017) observed that identifying as Democrat was a significant predictor of attention to Ebola news but not perceptions of personal risk. On the contrary, identifying as Republican was not a factor in attention to Ebola news but was a predictor of perceptions of personal risk of the virus.

From the beginning of the pandemic, Americans were exposed to contradictory information (e.g., Zhao et al., 2020). In a 4 March interview with Fox News’ Sean Hannity, for example, then President Trump dismissed WHO estimates of a 3.4% global mortality rate, suggesting actual mortality rate was likely to be ‘way under 1 per cent’ and that the symptoms are ‘very mild’ and people ‘will get better very rapidly’ (Trump, 2020). On the same day, National Institute of Allergy and Infectious Diseases Director Anthony Fauci testified that the case and fatality numbers in China suggested the death rate there was about 2% (National Institutes of Health, 2020). In addition, numerous authorities questioned the accuracy of government records used to track the number of cases and fatalities within the United States (Banco & Suebsaeng, 2020). Conflicting assessments can undermine the management of the risk event and lead to polarisation (Busby & Onggo, 2013; Horlick-Jones, 1998). Political leaders and media on the left and right have conveyed conflicting messages about the risks and severity of the COVID-19 pandemic with messages from Republican leaders sometimes contradicting health experts (e.g., Peters & Grynbaum, 2020).

Partisan media messaging has been identified as a potential explanation for why Republicans, as compared to Democrats, perceive COVID-19 as less risky (e.g., Allcott et al., 2020; Barrios & Hochberg, 2020; Xu, 2023). During the coronavirus pandemic, traditional news media representing the poles of the political spectrum have been found to deliver different information in relation to COVID-19 (e.g., Coppins, 2020; Motta et al., 2020). Motta et al. (2020) compared mainstream media coverage (including the New York Times and USA Today) of COVID-19 in February and March to coverage of right-leaning media (including Fox News and Breitbart). They found that right-leaning outlets had more than twice as many references to misinformation. The researchers also looked at beliefs in misinformation, such as misinformation about the origins of the disease and development of a vaccine. People who consumed conservative-leaning media were twice as likely to believe misinformation as people who did not. Furthermore, people with misinformed views were more likely to downplay the severity of the pandemic (Motta et al., 2020). Research has also linked differences in concern about the virus to attention to particular partisan media outlets. A poll conducted in late March 2020 found that 79% of Fox News viewers believed the media had exaggerated the risk of the coronavirus compared to 54% CNN viewers and 35% of MSNBC viewers (Jurkowitz & Mitchell, 2020b).

Similarly, in comparing COVID-19 misinformation on social platforms, Cinelli et al. (2020) found that disinformation was both more likely to be present and more likely to be amplified on Gab (which is popular among conservatives) than it was on other social media platforms. Gab amplified unreliable posts by 400%. By comparison, Twitter neither amplified nor downplayed disinformation, whereas YouTube reduced it by 10% and Reddit downplayed it by 50%. The visibility of different assessments can be an impediment to managing the risk event and contribute to polarisation (Busby & Onggo, 2013; Horlick-Jones, 1998). Given the prior observations of a relationship between political ideology and perceptions of risk, the following hypothesis and research questions are posed:

H₃: Compared with Republicans, Democrats will perceive COVID-19 as riskier.

RQ1: Will the relationship between attention to COVID-19 information in traditional media and perceived risk depend on political affiliation?

RQ2: Will the relationship between attention to COVID-19 information on social media and perceived risk depend on political affiliation?

Awareness of a Stay-at-home Order

Implementation of policies to control the spread of COVID-19 varied from state to state and sometimes from one county or city to the next. Many, but not all, states implemented stay-at-home measures to curb the spread of the disease. And in regions without state-wide stay-at-home orders, many local governments implemented their own. Some scholars theorise that the actions of those responsible for managing a risk event can send strong signals that could amplify or, conversely, attenuate perceptions of risk (e.g., Blevins et al., 2018; Khatapoush & Hallfors, 2004; Liu et al., 2017). We, therefore, include in our analysis a measure of whether there was a stay-at-home order that the respondent is aware of to account for the potential impact.

Methods

Sample

This study employed a nationally representative sample of U.S. residents (N = 1,969) for an online survey. The survey was conducted in cooperation with the professional market research firm Kelton Global. U.S. residents age 18+ (with Census based, interacting quotas for gender, age groups, ethnicity and region) were recruited from a market research panel. Regional quotas sampled participants from the U.S. Northeast (17%), Midwest (21%), South (37%) and West (24%). Ethnic quotas included White/Caucasian (65%), Hispanic or Latino (15%), Black or African American (11%), Asian (6%) and multiracial/other (2%). Inattentive respondents were removed by excluding speeders and straight-liners who gave identical values to most survey questions. All survey questions used to test hypotheses and research questions were forced answer, so there was no missing data. The final sample had an average age of 47.74 (SD = 16.86) and was 48% male.

Measures

Attention to Traditional Media

Attention to traditional media as a source for news about COVID-19 was measured as the average response to two question items. Respondents were asked how much they pay attention to (i) national news reports and (ii) local news reports as a source of news about COVID-19. Responses were measured on a 5-point scale ranging from 1 (not at all) to 5 (a great deal) (M = 3.64, SD =1.41, Pearson’s r =.62).

Attention to Social Media

Attention to social media as a source for news about COVID-19 was measured as the average response to two question items. Respondents were asked how much they pay attention to (i) social media news reports from news organisations and (ii) social media comments posted by friends and family as a source for news about COVID-19. Responses were measured on a 5-point scale ranging from 1 (not at all) to 5 (a great deal) (M = 2.66, SD =1.41, Pearson’s r = 0.77).

Political Affiliation

Political affiliation was measured using a single item asking respondents to indicate whether they were Democrat (n = 761), Republican (n = 531) or independent/other (n = 677), which we dummy coded with Republicans as the comparison group.

Awareness of a Stay-at-home Order

Awareness of a stay-at-home order was measured as a single item asking respondents to report whether they lived in a region in which there is a stay-at-home order. Respondents indicating that there was a stay-at-home order in their area (71%) were coded as ‘1’ and those that did not indicate that there was any stay-at-home order in their area that they were aware of coded ‘0’.

Perception of Risk

Perception of risk was measured as the average of four items of which they indicated their agreement on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). The statements included: ‘Not enough is being done to fight the spread of coronavirus (COVID-19).’; ‘We have been over-reacting to coronavirus (COVID-19)’ (reverse coded); ‘I am worried about the number of Americans that may become sick with coronavirus (COVID-19)’; and ‘I am worried that I could become sick with coronavirus (COVID-19)’ (M = 3.61, SD = 0.97, α = 0.76). We created these items specifically for this study, drawing on some of the discourse that was unfolding around COVID-19 at the time to enhance the ecological validity of our measure. Our items are consistent with measurements of perceived risk in similar U.S.-based COVID-19 survey studies (e.g., Barber & Kim, 2021; Eroglu et al., 2022).

Control Variables

Our study controlled for six demographic variables that have been found to account for some of the variance in perceived risk or concern about COVID-19—age, gender, race, education, income and rural (vs urban) residence (Chauhan et al., 2021; Lewis & Duch, 2021; Niño et al., 2021). Age was measured as a continuous variable, and Female gender was coded as a dichotomous variable (female = 1; male = 0). Race, included five dummy coded categories (White, Black, Asian, Hispanic, other), with White as the reference category. Education was coded as a dichotomous variable with those having only some college or less coded as ‘0’ and those with at least a college degree (51%) coded as ‘1’. Income was also coded as a dichotomous variable with individuals earning $49,999 or less coded as ‘0’ and those earning $50,000 or more (59%) coded as ‘1’. Rural was coded as a dichotomous variable with those respondents living in small municipal or rural areas coded as ‘1’ and those urban or suburban areas coded as ‘0’ (69%).

Procedure

The survey was fielded from 19 to 21 May 2020, a period in which we anticipated attention to COVID-19 coverage would be high among U.S. residents. By 19 May, it had been just two and a half months since COVID-19 was declared a national emergency, less than three weeks since Trump announced Operation Warp Speed to fast-track the development of a vaccine, and American deaths due to the disease were approaching 100,000. Across the three days, we fielded the survey, an average of 22,262 Americans per day tested positive for the virus (Centers for Disease Control and Prevention, 2020). By the end of the month, there would be 1.8 million cases and 103,000 deaths in the United States since the start of the outbreak. On the final day of data collection, worldwide cases hit 5 million, up by more than 1 million in less than two weeks.

Data Analysis

To test our hypotheses and research questions, we ran a hierarchical ordinary least squares regression model, in which we entered variables in blocks based on their assumed causal order. Blocks were entered in the following order: demographics, stay-at-home order, political affiliation, media attention variables and finally a block including interactions between partisanship and media attention variables. All continuous independent variables were standardised, and all dichotomous variables were contrast coded for centring. All analyses were conducted with R (Version 3.4.4; http://www.r-project.org/).

Results

Overall, the regression model explained 27% of the variation in perceived risk (see Table 1). Among the blocks included in the final regression model, the political affiliation block and media attention block explained the greatest percentage of variance in perceived risk (10% and 12%, respectively). Results for variables entered in the media attention block provide support for hypothesis 1 but not hypothesis 2. As predicted by hypothesis 1, attention to COVID-19 news in traditional media, positively correlated with perceived risk (β = 0.37, P < 0.00). However, we did not detect a significant correlation between perceived risk and attention to social media as a source of COVID-19 news (H₂). Results in the political affiliation block provides support for our final hypothesis (H₃). As hypothesis 3 predicted, the perceived risk of COVID-19 was higher among Democrats than among Republicans (β = 0.71, P < 0.00),

Table 1.

Hierarchical Regression Model Predicting Perceived Risk of COVID-19 (N = 1,969).

	Upon Entry	Final β
Block 1: Demographics
Age	0.09***	0.04
Gender (female)	0.10*	0.10**
Race (reference category = white)
African American	0.15*	–0.11
Asian	0.18	0.04
Hispanic	0.24***	0.02
Other	–0.11	–0.20
Education	0.08	–0.03
Income	0.00	0.02
Rural (vs urban)	–0.09	0.03
Adj. inc. R² (%)		1.94***
Block 2: Government policy
Awareness of a stay-at-home order	0.27***	0.17***
Adj. Inc. R² (%)		1.53***
Block 3: Political affiliation
Democrat (vs Rep.)	0.81***	0.71***
Other (vs Rep.)	0.47***	0.49***
Adj. inc. R² (%)		10.05***
Block 4: Media attention
Traditional media	0.37***	0.37***
Social media	–0.03	–0.03
Adj. inc. R² (%)		12.48***
Block 5: Interaction terms
Traditional media × Democrat (vs Rep.)		0.04
Traditional media × Other (vs Rep.)		0.04
Social media × Democrat (vs Rep.)		–0.28***
Social media × Other (vs Rep.)		–0.21***
Adj. inc. R² (%)		1.12***
Total adj. R² (%)		27.12***

Note: ***P < 0.001; **P < 0.01; *P < 0.05. Betas in the second column represent final standardised regression coefficients, except for blocks 3 and 4 which report coefficients estimated in the final model before entering block 5 to reduce the influence of multicollinearity on interpretations of product terms included in the interactions.

The final block of the model tested interactions in response to our two research questions (R1 & R2), exploring whether the influence of our media attention variables might depend on an individuals’ political affiliation. The results did not reveal a significant interaction between attention to traditional media and political affiliation. However, the results did reveal a significant interaction between attention to COVID-19 information on social media posts and political affiliation (Democrats v. Rep., β = –0.28, P < 0.001; other v. Rep., β = –0.21, P < 0.001). Attention to social media was associated with greater risk perceptions among Republicans, but not among Democrats or individuals that do not identify as either Republicans or Democrats (see Figure 1).

Figure 1.

Interactive Influence of Attention to Social Media Posts on COVID-19 and Respondents’ Affiliation with the Democratic, Republican or other Political Party on Perceived Risk.

Discussion

The purpose of this project was to examine the role attention to traditional and social media play in amplifying or attenuating perceptions of risk related to COVID-19. Furthermore, we explored how a person’s political affiliation interacted with their attention to social media regarding COVID-19 to be positively associated with risk perceptions. The regression model indicates traditional and social media played distinctive roles in perceptions of risk associated with COVID-19. Analyses conducted to test our hypotheses and research questions revealed three significant results.

First, political affiliation was an important contributor to risk perceptions. Our findings indicate political affiliation was a strong contributor to perceptions of risk and are consistent with other published observations finding that Republican respondents were less concerned about risks of COVID-19 than their independent/other and Democrat counterparts (e.g., Pew Research Center, 2020). The politicisation of COVID-19 could be an outgrowth of positions taken by political figures (McCarthy, 2020). For example, from January through the beginning of March 2020, President Trump’s public statements reflected low levels of perceived risk (Watson, 2020). There is some evidence that Trump supporters may have adopted this attitude toward the pandemic. For example, counties with higher percentages of Trump supporters were less likely to search the internet for information about the coronavirus and engaged in fewer mitigation behaviours (Barrios & Hochberg, 2020). Conflicts over trust in science or between the amount of value placed on health or economic concerns could also contribute to the differing perceptions of risk (e.g., Barry et al., 2020). Second, we found that, regardless of political affiliation, people who paid more attention to COVID-19 via traditional news media perceived more risk.

Several studies suggest some conservative media downplayed COVID-19 risks or spread misinformation about the pandemic (Jurkowitz & Mitchell, 2020b; Motta et al., 2020). Consequently, one might expect a positive correlation between people’s attention to traditional media coverage of COVID-19 and perceptions of risk, but one that is attenuated by Republican respondents whose exposure would likely lead to less concern. One possible explanation for why we did not observe this relationship in our data is that partisan audiences may rely on a more varied media diet than what might be widely assumed. A Gallop poll conducted in March 2020 found many Republicans and Democrats were consuming news from a mix of media sources with a greater share of Republicans consuming news from sources more representative of other parties’ view or that cut across party lines. The poll found 46% of Republicans consume a ‘conservative news diet’, while the rest consume a ‘mixed’ (32%), ‘undesignated’ (15%) or ‘liberal’ (7%) news diet (Ritter, 2020b). Meanwhile, 58% of Democrats consume a ‘liberal news diet’, while the rest consume a ‘mixed’ (32%) or ‘undesignated’ (9%) news diet. Perhaps due to these mixed media diets, increased attention to news among partisan groups does not necessarily amplify partisan views.

Finally, attention to COVID-19 information through social media was not a significant predictor of risk in our examination of main effects. However, in this study, we took our analysis one step further by examining whether the relationship between attention to COVID-19 messaging in social media and risk perception might be moderated by individual-level value orientations. The most noteworthy finding in this study relates to how political affiliation interacts with attention to social media to contribute to perceptions of risk. Republican respondents who reported more attention to COVID-19 social media posts also reported higher levels of perceived risk. This finding may suggest that social media engagement is not taking place in as much of a partisan bubble (i.e., among people with shared views and values) as is often assumed. Additionally, if social media messaging is more likely to feature framing related to consequences that messaging may also have been particularly vivid, enhancing accessibility and persuasiveness (Tversky & Kahneman, 1973). However, we did not observe a similar boost among Democrats who similarly used social media. It may be that a ceiling effect impeded the potential social media effect among Democrats.

Before we conclude, there are some limitations in this study that should be noted. First, this project reflects a snapshot of the evolution of perceived risks associated with COVID-19. This cross-sectional approach does not convey the context of a dynamic health event that changes over time. The characteristics of this health event have similarities with prior diseases such as MERS and avian flu, there are unique attributes associated with positions of policy makers, media, as well as government and international agencies that may make the results difficult to generalise. Second, our data collection and analysis methods predict risk perceptions. Our findings do not establish causation and should be interpreted as correlational. Third, measures of attention to COVID-19 through traditional and social media required respondents to provide a heuristic judgement that could lack accuracy. And even to the extent our measures are accurate, they do not distinguish between exposure to specific social media platforms. We join Cascini et al. (2022) in calling for measuring attention to specific social media platforms because there is reason to believe impact may vary.

The SARF identified multiple layers and components that make up a complex messaging system involving risk events. This study highlights ways in which the increasing complexity of the media system is associated with the need for a more nuanced understanding of which components are most important in different risk contexts. The strongest components in the model examined were political affiliation and attention to social media messaging about COVID-19. However, today’s polarised and complex media environment calls for deeper investigation. The interplay between traditional media messaging and social media interactions is an area well-suited for further research. The social construction of risk evident in this study indicates the challenge of communicating about health or other risks in a manner that uses traditional and social media venues to define and respond to a hazard. The discourse in social media played a role in achieving some semblance of shared perspective of the COVID-19 risk. This study helps clarify components that had an impact on the construction of COVID-19 risk perceptions and can inform researchers, policy analysts and others about how media converge to inform risk perceptions and potentially guide health communicators in the development and distribution of risk messages.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

References

Ackerson

L. K.

, & Viswanath

(2010). Media attention and public perceptions of cancer and eastern equine encephalitis. Journal of Community Health, 35(4), 409–416.

Allcott

, Boxell

, Conway

, Gentzkow

, Thaler

, & Yang

(2020). Polarization and public health: Partisan differences in social distancing during the coronavirus pandemic. Journal of public economics, 191, 104254. 10.1016/j.jpubeco.2020.104254

Banco

, & Suebsaeng

(2020, May 13). Team Trump pushes CDC to revise down its COVID death counts. Daily Beast. https://www.thedailybeast.com/team-trump-pushes-cdc-to-dial-down-covid-death-counts?ref=home

Barber

S. J.

, & Kim

(2021). COVID-19 worries and behavior changes in older and younger men and women. The Journals of Gerontology: Series B, 76(2), e17–e23.

Barrios

J. M.

, & Hochberg

Y. V.

(2020). Risk perception through the lens of politics in the time of the COVID-19 pandemic. University of Chicago, Becker Friedman Institute for Economics, Working Paper 2020-32. https://bfi.uchicago.edu/wp-content/uploads/BFI_WP_202032.pdf

Barry

, Han

, & McGinty

(2020, June 17). Trust in science and COVID-19. School of Public Health Experts Insights, John’s Hopkins University. https://www.jhsph.edu/covid-19/articles/trust-in-science-and-covid-19.html

Blevins

C. E.

, Marsh

, Banes

K. E.

, Stephens

R. S.

, Walker

D. D.

, & Roffman

R. A.

(2018). The implications of cannabis policy changes in Washington on adolescent perception of risk, norms, attitudes, and substance use. Substance Abuse: Research and Treatment, 12, 1178221818815491.

Broersma

, & Graham

(2012). Social media as beat: Tweets as news source during the 2010 British and Dutch elections. Journalism Practice, 6(3), 403–419.

Busby

J. S.

, & Onggo

(2013). Managing the social amplification of risk: A simulation of interacting actors. Journal of the Operational Research Society, 64(5), 638–653. 10.1057/jors.2012.8

10.

Cascini

, Pantovic

, Al-Ajlouni

Y. A.

, Failla

, Puleo

, Melnyk

, Lontano

, & Ricciardi

(2022). Social media and attitudes towards a COVID-19 vaccination: A systematic review of the literature. EClinicalMedicine, 101454. 10.1016/j.eclinm.2022.101454

11.

Centers for Disease Control and Prevention. (2020). Cases in the U.S. https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/cases-in-us.html

12.

Chauhan

R. S.

, da Silva

D. C.

, Salon

, Shamshiripour

, Rahimi

, Sutradhar

, Khoeini

, Mohammadian

, Derrible

, & Pendyala

(2021). COVID-19 related attitudes and risk perceptions across urban, rural, and suburban areas in the United States. Findings, 23714. 10.32866/001c.23714

13.

Chew

, & Eysenbach

(2010). Pandemics in the age of Twitter: Content analysis of tweets during the 2009 H1N1 outbreak. PLoS One, 5. 10.1371/journal.pone.0014118

14.

Choi

D.-H.

, Yoo

, Noh

G.-Y.

, & Park

(2017). The impact of social media on risk perceptions during the MERS outbreak in South Korea. Computers in Human Behavior, 72, 422–431. 10.1016/j.chb.2017.03.004

15.

Cinelli

, Quattrociocchi

, Galeazzi

, Valensise

C. M.

, Brugnoli

, Schmidt

A. L.

, Zola

, Zollo

, & Scala

(2020). The COVID-19 social media infodemic. Scientific reports, 10(1), 1–10.

16.

Coppins

(2020, March 11). Trump’s dangerously effective coronavirus Propaganda. The Atlantic. https://www.theatlantic.com/politics/archive/2020/03/trump-coronavirus-threat/607825/?utm_source=share&utm_campaign=share

17.

Eroglu

S. A.

, Machleit

K. A.

, & Neybert

E. G.

(2022). Crowding in the time of COVID: Effects on rapport and shopping satisfaction. Journal of Retailing and Consumer Services, 64(4), 102760.

18.

Fellenor

, Barnett

, Potter

, Urquhart

, Mumford

J. D.

, & Quine

C. P.

(2018). Ash dieback and other tree pests and pathogens: Dispersed risk events and the social amplification of risk framework. Journal of Risk Research, 22(12), 1459–1478. 10.1080/13669877.2018.1485170

19.

Frewer

L. J.

, Miles

, & Marsh

(2002). The media and genetically modiﬁed foods: Evidence in support of social ampliﬁcation of risk. Risk Analysis, 22(4), 701–711. 10.1111/0272-4332.00062

20.

Frewer

L. J.

, Raats

, & Shepherd

(1993/1994). Modelling the media: The transmission of risk information in the British quality press. Journal of the Institute of Mathematics and its Applications to Industry, 5, 235–247.

21.

Graham

, Jackson

, & Broersma

(2016). New platform, old habits? Candidates’ use of Twitter during the 2010 British and Dutch general election campaigns. New Media & Society, 18(5), 765–783. 10.1177/1461444814546728

22.

Green

T. V.

, & Tyson

(2020). Five facts about partisan reactions to COVID-19 in the U.S. Pew Research Center. https://www.pewresearch.org/fact-tank/2020/04/02/5-facts-about-partisan-reactions-to-covid-19-in-the-u-s/

23.

Hansson

S. O.

(2010). Risk: Objective or subjective, facts or values. Journal of Risk Research, 13(2), 231–238. 10.1080/13669870903126226

24.

Hart

P. S.

, Nisbet

E. C.

, & Shanahan

J. E.

(2011). Environmental values and the social amplification of risk: An examination of how environmental values and media use influence predispositions for public engagement in wildlife management decision making. Society and Natural Resources, 24(3), 276–291.

25.

Hornig

(1993). Reading risk: public response to print media accounts of technological risk. Public Understanding of Science, 2(2), 95–109.

26.

Horlick-Jones

(1998). Meaning and contextualization in risk measurement. Reliability Engineering and System Safety, 59, 79–89.

27.

Johnson

B. B.

(2017). Explaining American’s responses to dread epidemics: An illustration with Ebola in late 2014. Journal of Risk Research, 20(10), 1338–1357. 10.1080/13669877.2016.1153507

28.

Jurkowitz

, & Mitchell

(2020a). Older Americans continue to follow COVID-19 news more closely than younger adults. Pew Research Center. https://www.journalism.org/2020/04/22/older-americans-continue-to-follow-covid-19-news-more-closely-than-younger-adults/

29.

Jurkowitz

, & Mitchell

(2020b). Cable TV and COVID-19: How Americans perceive the outbreak and view media coverage differ by main news source. Pew Research Center. https://www.journalism.org/2020/04/01/cable-tv-and-covid-19-how-americans-perceive-the-outbreak-and-view-media-coverage-differ-by-main-news-source/

30.

Kasperson

R. E.

, & Kasperson

J. X.

(1996). The social amplification and attenuation of risk. The Annals of the American Academy of Political and Social Science, 545, 95–105. 10.1177/0002716296545001010

31.

Kasperson

R. E.

, Renn

, Slovic

, Brown

H. S.

, Emel

, Goble

, Kasperson

J. X.

, & Ratick

(1988). The social ampliﬁcation of risk: A conceptual framework. Risk Analysis, 8(2), 177–187. 10.1111/j.15396924.1988.tb01168.x

32.

Khatapoush

, & Hallfors

(2004). Sending the wrong message: Did medical marijuana legalization in California change attitudes about and use of marijuana? Journal of Drug Issues, 34(4), 751–770. 10.1177/002204260403400402

33.

Lewis

, & Duch

(2021). Gender differences in perceived risk of COVID-19. Social Science Quarterly, 102(5), 2124–2133.

34.

Liu

, Busby

, & Onggo

(2017). Modeling social risk amplification during recall crises: Simulation series. Paper presented at the Simulation Series, Society for Computer Simulation.

35.

McCarthy

(2020, March 29). Disunited states of America: Responses to coronavirus shaped by hyper-partisan politics. The Guardian. https://www.theguardian.com/us-news/2020/mar/29/america-statescoronavirus-red-blue-different-approaches.

36.

McCombs

(2006). A look at agenda setting: Past, present and future. Journalism Studies, 6(4), 543–557.

37.

Motta

, Stecula

, & Farhart

(2020). How right-leaning media coverage of COVID-19 facilitated the spread of misinformation in the early stages of the pandemic in the U.S. Canadian Journal of Political Science, 53(2), 335–342. 10.1017/S000842392000039

38.

National Institutes of Health. (2020). Fiscal year 2021 budget request. U.S. Department of Health and Human Services, National Institutes of Health. https://officeofbudget.od.nih.gov/pdfs/FY21/br/1OverviewVolumeSingleFile-toPrint.pdf6

39.

Newman

, Dutton

W. H.

, & Blank

(2012). Social media in the changing ecology of news: The fourth and fifth estates in Britain. International Journal of Internet Science, 7(1), 6–22.

40.

Y. J.

, Yang

Z. J.

, & Vishwanath

(2018). To fear or not to fear? Applying the social amplification of risk framework on two environmental health risks in Singapore. Journal of Risk Research, 21, 1487–1501. 10.1080/13669877.2017.1313762

41.

Niño

, Harris

, Drawve

, & Fitzpatrick

K. M.

(2021). Race and ethnicity, gender, and age on perceived threats and fear of COVID-19: Evidence from two national data sources. SSM Population Health, 13, 100717.

42.

Nordenstedt

, & Ivanisevic

(2010). Values in risk perception: Studying the relationship between values and risk perception in three countries. Journal of Disaster Risk Studies, 3(1), 335–345. DOI: 10.4102/jamba.v3i1.34

43.

S. H.

, Lee

S. Y.

, & Han

(2020). The effects of social media use on preventive behaviors during infectious disease outbreaks: The mediating role of self-relevant emotions and public risk perception. Health Communication, 36(8), 972–981. 10.1080/10410236.2020.1724639

44.

Peters

J. W.

, & Grynbaum

M. M.

(2020, March 11). How right-wing pundits are covering coronavirus. The New York Times. https://www.nytimes.com/2020/03/11/us/politics/coronavirusconservative-media.html

45.

Pew Research Center. (2016). Most in U.S. public think the Zika virus poses some level of threat. https://www.pewresearch.org/science/2016/07/08/half-of-americans-say-threats-from-infectious-diseases-are-growing/#:~:text=58%25%20of%20Americans%20believe%20the,U.S.%20population%20as%20a%20whole

46.

Pew Research Center. (2019). 2. Social media users more likely to interact with people who are different from them. https://www.pewresearch.org/internet/2019/08/22/social-media-users-more-likely-to-interact-with-people-who-are-different-from-them/

47.

Pew Research Center. (2020). Republicans, democrats move even further apart in coronavirus concerns. https://www.people-press.org/2020/06/25/republicans-democrats-move-even-further-apart-in-coronavirus-concerns/

48.

Price

, Tewksbury

, & Powers

(1997). Switching trains of thought. Communication Research, 24(5), 481–506. 10.1177/009365097024005002

49.

Renn

, Burns

W. J.

, Kasperson

J. X.

, Kasperson

R. E.

, & Slovic

(1992). The social amplification of risk: Theoretical foundations and empirical applications. Journal of Social Issues, 48(4), 137–160. 10.1111/j.15404560.1992.tb01949.x

50.

Ritter

(2020a). Amid pandemic, news attention spikes; Media favorability flat. Gallup. https://news.gallup.com/opinion/gallup/307934/amid-pandemic-news-attention-spikes-media-favorability-flat.aspx

51.

Ritter

(2020b). Americans use social media for COVID-19 info, connection. Gallup. https://news.gallup.com/poll/311360/americans-social-media-covid-information-connection.aspx

52.

Rosa

E. A.

(2003). The logical structure of the social amplification of risk framework (SARF): Metatheoretical foundations and policy implications. In: Pidgeon

, Kasperson

R. E.

, & Slovic

(Eds.), The social amplification of risk (pp. 47–79). Cambridge University Press.

53.

Ruppel

E. K.

(2016). Scanning health information sources: Applying and extending the comprehensive model of information seeking. Journal of Health Communication, 21, 208–216. 10.1080/10810730.2015.1058438

54.

Rutsaert

, Regan

Á.

, Pieniak

, McConnon

Á.

, Moss

, Wall

, & Verbeke

(2013). The use of social media in food risk and benefit communication. Trends in Food Science & Technology, 30(1), 84–91.

55.

Slater

M. D.

, Goodall

C. E.

, & Hayes

A. F.

(2009). Self-reported news attention does assess differential processing of media content: An experiment on risk perceptions utilizing a random sample of U.S. local crime and accident news. Journal of Communication, 59(1), 117–134. 10.1111/j.1460-2466.2008.01407.x

56.

Slater

M. D.

, Hayes

A. F.

, & Ford

V. L.

(2007). Examining the moderating and mediating roles of news exposure and attention on adolescent judgments of alcohol-related risks. Communication Research, 34(4), 355–381. 10.1177/0093650207302783

57.

Slovic

(2000). The Perception of Risk. Routledge. 10.4324/9781315661773

58.

Smith

, & McCloskey

(1998). Risk and crisis management in the public sector: Risk communication and the social amplification of public sector risk. Public Money and Management, 18(4), 41–50. 10.1111/14679302.00140

59.

Sutton

, & Veil

(2017). Risk communication and social media. In Kasperson

R. E.

(Ed.), Risk conundrums: Solving unsolvable problems (pp. 96–111). Routledge.

60.

Thompson

R. R.

, Garfin

D. R.

, Holman

E. A.

, & Silver

R. C.

(2017). Distress, worry, and functioning following a global health crisis: A national study of Americans’ responses to Ebola. Clinical Psychological Science, 5(3), 513–521.

61.

Trump

(2020, March 4) Trump blasts Biden’s record in ‘Hannity’ exclusive interview/Interviewer: S. Hannity. Sean Hannity, Fox News.

62.

Tversky

, & Kahneman

(1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5, 207–232.

63.

Wang

, McKee

, Torbica

, & Stuckler

(2019). Systematic literature review on the spread of health-related misinformation on social media. Social Science & Medicine, 240, 112552. 10.1016/j.socscimed.2019.112552

64.

Wardman

J. K.

(2008). The constitution of risk communication in advanced liberal societies. Risk Analysis: An International Journal, 28(6), 1619–1637.

65.

Watson

(2020, April 3). A timeline of what Trump has said on coronavirus. CBSNews.com https://www.cbsnews.com/news/timeline-president-donald-trump-changing-statements-on-coronavirus/

66.

(2023). Examining U.S. newspapers’ partisan bias in COVID-19 news using computational methods. Communication Studies, 74(1), 78–96. 10.1080/10510974.2023.2169729

67.

Yang

, Chen

, & Feng

(2016). Risk perception of food safety issue on social media. Chinese Journal of Communication, 9(2), 124–138.

68.

Yoo

, Paek

H. J.

, & Hove

(2020). Differential effects of content-oriented versus user-oriented social media on risk perceptions and behavioral intentions. Health Communication, 35(1), 99–109. doi: 10.1080/10410236.2018.1545169

69.

Young

M. E.

, Norman

G. R.

, & Humphreys

K. R.

(2008). Medicine in the popular press: The influence of the media on perceptions of disease. PLoS One, 3(10), e3552.

70.

Zhang

X. A.

, & Cozma

(2022). Risk sharing on Twitter: Social amplification and attenuation of risk in the early stages of the COVID-19 pandemic. Computers in Human Behavior, 126, 106983.

71.

Zhao

, Wu

, Crimmins

E. M.

, & Ailshire

J. A.

(2020). Media trust and infection mitigating behaviours during the COVID-19 pandemic in the USA. BMJ Global Health, 5(e003323), 1–10. 10.1136/bmjgh-2020-003323

72.

Zhao

, Leiserowitz

A. A.

, Maibach

E. W.

, & Roser-Renouf

(2011). Attention to science/environment news positively predicts and attention to political news negatively predicts global warming risk perceptions and policy support. Journal of Communication, 61(4), 713–731.