Physician Trust in the News Media and Attitudes toward COVID-19

Abstract

Context: Previous research has established the importance of primary care physicians in communicating public health directives. The implicit assumption is that, because of their expertise, doctors provide accurate and up-to-date information to their patients independent of partisan affiliation or media trust.

Methods: The authors conducted an online survey of 625 primary care physicians and used the results to test (1) whether physician trust in media outlets is consistent with their political partisanship, and (2) whether trust in media outlets influences (a) personal concern that someone in their family will get sick, (b) perceptions about the seriousness of the pandemic as portrayed in the media, and (c) trust in federal government agencies and scientists.

Findings: Physicians are better positioned to critically evaluate health-related news, but they are subject to the same biases that influence public opinion. Physicians' partisan commitments influence media trust, and media trust influences concern that a family member will get sick, perceptions regarding the seriousness of the pandemic, and trust in federal government agencies and scientists.

Conclusions: Physician trust in specific media outlets shapes their understanding of the pandemic, and—to the extent that they trust conservative media outlets—it may limit their effectiveness as health policy messengers.

Previous research has established the important role that primary care physicians play in communicating health-related messages to the public (Elledge et al. 2008; Taha, Matheson, and Anisman 2013). Because patients trust their primary care physicians (Fiscella et al. 2004; Gabay 2015; Findling, Blendon, and Benson 2022), they are more receptive to health-related messages when their physician is the source of the information (Bhat et al. 2021; Clarke et al. 2015). In a recent survey,¹ for example, 78% of Americans said they trusted their personal physician while expressing considerably less trust in the health care system as whole. A similar survey conducted by the Pew Research Center found that 74% reported a “mostly positive” view of their doctor, and only 8% had a mostly negative view (Funk and Gramlich 2019). In this sense, primary care physicians are often seen as allies working to support their patients within a broader and less caring health care system.

For these reasons, primary care physicians are frequently identified as critical to successful health communication strategies and effective health policy implementation (Pepe et al. 2015), particularly when there is a need to encourage hesitant patients to comply with public health directives (Patashnik, Gerber, and Dowling 2017). Throughout the COVID-19 pandemic, for example, the public has been asked to wear masks and maintain social distance. With the successful development and testing of COVID-19 vaccinations, health authorities subsequently encouraged the public to get vaccinated against the virus to protect themselves and others. Based on previous research, we know that primary care physicians play an important role in encouraging patient compliance with such directives, reassuring reluctant patients of the safety and necessity of public health measures, and confirming that public health recommendations are supported by the best available science (Jallinoja et al. 2007; Shen and Dubey 2019). In recognition of the potential importance of physicians to health communication strategies, scholars, elected policy makers,² and health officials³ have advocated that physicians play a more active role on social media to communicate health messages (Fatollahi et al. 2020; Furstrand et al. 2021). There is some evidence of this occurring. Recent research finds that primary care providers regularly use social media to encourage preventive health behaviors, correct misinformation, and encourage vaccination (Wahbeh et al. 2020).

The implicit assumptions underlying these models are (1) because of doctors' expertise, they will provide accurate and up-to-date information to their patients, and (2) because patients trust their primary care physicians, patients will be more likely to follow their advice. Previous research, however, has shown that this may not always be the case and that a nontrivial subset of physicians and health care workers express doubt about vaccines and/or are unlikely to encourage vaccination (Verger et al. 2015; Lucia, Kelekar, and Afonso 2021; Paterson et al. 2016; Callaghan et al. 2022). Perhaps even more telling is research indicating that physician care may depend, in no small part, on a physician's partisan political affiliation (Hersh and Goldenberg 2016), that physicians actively engage in the political process by contributing to partisan political candidates (Bonica, Rosenthal, and Rothman 2014, 2015), and that there is ideological and geographic sorting among physicians such that partisan physicians may be interacting most frequently with physicians who share their partisan commitments (Bonica et al. 2020). While these studies do not speak directly to media trust, they do suggest that we might reasonably expect that physicians trust those media outlets that most reflect their earlier partisan commitments.

In this respect, primary care physicians and other front-line health care workers may not be entirely immune to a broader communication environment that is increasingly defined by partisan polarization and ideologically driven news. Nor are they likely to be entirely removed from the effects that partisan predispositions have on political attitudes and behaviors or the role that motivated reasoning plays in decisions to seek out information that largely confirms one's earlier beliefs (Brodie et al. 2019; Bonica, Rosenthal, and Rothman 2015; Gusmano 2019; Petrova et al. 2019; Lupia 2013).⁴ While this is true even in “normal” times, it may be particularly important during public health crises when new data and evidence are being rapidly updated and, because of polarization, are being interpreted through a partisan lens. In the case of COVID-19, the public health response has been intentionally and explicitly politicized to deflect causal and treatment responsibility for the pandemic (Gollust, Nagler, and Fowler 2020; Graham and Singh 2022).

Within this polarized environment, news coverage of the pandemic by conservative media outlets has differed sharply from news coverage by mainstream media outlets, with important consequences for public opinion (Motta, Stecula, and Farhart 2020; Hoewe, Brownell, and Wiemer 2020).⁵ Specifically, individual exposure to right-wing media outlets increased misinformation about the origins of the virus and perceptions that the mainstream media were overstating the threat of COVID-19 (Motta, Stecula, and Farhart 2020; Hamel et al. 2021).⁶ Perhaps as a result, we see sharp partisan differences in preventive behaviors (e.g., vaccination rates, social distancing, and mask wearing) and COVID-related beliefs and attitudes (Gadarian, Goodman, and Pepinsky 2021; Allcott et al. 2020; Wu and Huber 2021; Milosh et al. 2021; Bursztyn et al. 2020). In this article, we ask whether the same forces that influence public attitudes and preventive behaviors also shape physician beliefs. First, to what degree do partisan commitments influence physician trust in conservative vs. mainstream news media? Second, does physician trust in partisan media affect their beliefs about the COVID-19 pandemic? For example, are physicians who place greater trust in Fox News (relative to the New York Times or CNN) less worried that someone in their family would become sick with COVID-19? Are they more likely to believe the scope and severity of the pandemic has been exaggerated by the news media? And do they report less trust in federal government agencies and, more generally, in scientists to effectively handle the pandemic? In exploring these questions, we fill an important gap in the existing literature and offer a word of caution for future health communication efforts. Physician trust in specific media outlets shapes their understanding of the COVID-19 pandemic and—to the extent that physicians trust conservative media outlets—may limit their effectiveness as health policy messengers.

Media Coverage of Public Health and Its Effects on Primary Care Physicians

While one might hope that media coverage of a pandemic would assist public and expert news audiences in accurately understanding the risks of infection and the steps necessary to avoid infection, health news is often sensationalized and exaggerated (Haglin et al. 2020; Smith 2006; Rezza et al. 2004; Wilson, Thomson, and Mansoor 2004; Cortiñas-Rovira, Pont-Sorribes, and Alonso-Marcos 2015; Kilgo, Yoo, and Johnson 2019). This is a fact not lost on news audiences. In public opinion surveys, members of the general public say they see the news media as exaggerating health-related risks (Taha, Matheson, and Anisman 2013). For example, early coverage of the COVID-19 pandemic focused on infection and death rates, providing less information about how to prevent transmission (Basch et al. 2020). As a result, increased attentiveness to health news does not necessarily lead to more informed audiences, and in fact, in some cases it can lead to greater endorsement of misinformation. A Rutgers/Eagleton poll in 2014, for example, found that greater attention to news about the Ebola outbreak was associated with being less informed about how easily the disease was spread.⁷ It would seem there is good reason for news audiences—and especially for medical-expert audiences such as primary care physicians—to be skeptical of mainstream news coverage during a crisis (Miller and Goidel 2009; Pan and Meng 2016).

The increase of ideologically driven news adds an additional reason to worry about both the content and the effects of pandemic news coverage. Baum (2011), for example, observed more than a decade ago that media fragmentation and user selection of news sources led to partisan differences in H1N1 vaccination rates. These differences, he argued, were the result of Fox News and Republican elite skepticism regarding vaccine safety. More recent research indicates that Fox News provided more skeptical coverage of the COVID-19 pandemic, which in turn affected preventive behaviors in areas with larger Fox viewership (Ash et al. 2020; Ananyev, Poyker, and Tian 2021; Motta, Stecula, and Farhart 2020; Allcott et al. 2020). Subsequently, one might fairly suspect that these differences in content would affect pandemic-related attitudes and behaviors. First, physicians' personal concern about the risk of becoming infected by COVID-19 would likely be affected by how much trust physicians place in partisan news outlets. Second, individual attitudes toward the seriousness of the pandemic and whether the pandemic is being exaggerated by the news media should be consistent with the attitudes on display in the content presented in trusted media outlets. Third, trust in federal government agencies and in scientists may also be affected by media trust.

In this respect, news content can be shaped by the sources used in news stories, including the presence or absence of specific partisan and/or expert cues and by the frames adopted to help audiences make sense of news content. A pandemic-related news story that includes mostly partisan sources, for example, signals that the news is inherently political, thus polarizing public perceptions of the pandemic (Hubner 2021). According to a recent study, President Trump's endorsement of hydroxychloroquine led to an increase in hydroxychloroquine prescription rates across the country, especially in areas where Trump received larger shares of the two-party vote (Madanay, McDevitt, and Ubell 2022). While Madanay and colleagues do not directly connect the increase in prescriptions to physician partisanship, they do argue that physicians may have determined that the potential benefits of hydroxychloroquine outweighed the risks.

News stories that include more medical and scientific experts, in contrast, signal that a story's content is technical or scientific. During the early phase of the pandemic, Fox News downplayed the seriousness of the pandemic, relied on partisan sources in its coverage, and gave more airtime to misinformation (Gollust, Nagler, and Fowler 2020; Motta, Stecula, and Farhart 2020). In an early study of pandemic-related news coverage, Motta, Stecula, and Farhart (2020) reported that 59% of primetime Fox News shows used experts as sources compared to 81% on MSNBC primetime shows. Other research has shown that Fox News viewers were more likely than other news viewers to believe misinformation (Hamel et al. 2021).

Because the pandemic was an emerging health issue, primary care physicians would, by necessity, rely on media reporting to help them make sense of it (Fowler et al. 2012), including learning about the effectiveness of various preventive behaviors and new medications and treatments (Campion 2004; Phillips et al. 1991). Even if news media content might not matter (or might matter less) when there is little or no temporal urgency and physicians have time to carefully evaluate news content, it would likely matter in the context of a public health crisis. In this respect, while physicians may rely on more medically oriented sources of information (e.g., medical journals, training in medical school and residency) to learn about established treatments and protocols during “normal” times, during a crisis they are likely more dependent on the media to disseminate updates about the rapidly changing nature of a crisis's public health risks and impact (Kiernan 2003). Potentially even more disconcerting, partisan news sources may make viewers less trusting of these other sources of news and information. Previous research on climate change, for example, shows that conservative media use leads to declining trust in scientists (Hmielowski et al. 2013). More recent research on COVID-19 connects Fox News viewership at the local level with reduced compliance behaviors (Simonov et al. 2021).

Unlike the general public, primary care physicians should have the knowledge and expertise to critically evaluate the accuracy of health-related news and to be less influenced by news content (Zaller 1992). According to the receive-accept-sample model, a physician's knowledge and expertise provide a basis for arguing against media content that contradicts established medical facts (Zaller 1992). As a result, we might expect to find little association between media trust in reporting on COVID-19 and other political attitudes because primary care physicians should have the expertise necessary to independently evaluate the evidence at hand. This idea is well captured by Edward Campion (2004) in an editorial written for the New England Journal of Medicine:

Yet there is perhaps equally good reason to suspect that the reverse might be true. Physicians might selectively seek out ideologically congenial information about health crises and allow it to influence their relevant public health attitudes and behaviors. This could be the case with respect to the COVID-19 pandemic for several reasons. First, the pandemic emerged quickly and evolved rapidly, meaning physicians had to think fast and make quick decisions about the credibility of new information (Kahneman 2011). Second, we do have evidence suggesting physicians are influenced by news content. In one study, for example, an increase in news stories about “flesh-eating bacteria” was associated with an increase in emergency room testing for the bacterium, even when there was no corresponding increase in patients displaying symptoms (Sharma et al. 2003). Third, greater knowledge is often associated with greater directional (or motivated) reasoning, particularly on issues where one has strong, well-grounded previous attitudes (Bolsen and Druckman 2018; Taber and Lodge 2006; Kraft, Lodge, and Taber 2015; Kahan 2017; Guay and Johnston 2022). Expertise does not make one immune to cognitive biases and can in fact make one more confident in their conclusions. In addition, superior information and capacities to counterargue may help experts interpret sensational, misinformed, or uninformed news coverage in a way that reinforces their existing beliefs and perceptions (Zaller 1992).

We should note one issue we cannot address with the available data: the question of whether physicians choose to view media they find to be ideologically congruent with their own beliefs (via motivated reasoning and confirmation bias) or whether they are influenced by the coverage provided by a specific outlet they deem credible (or not). Recent research suggests that directionally motivated reasoning may be less important than differences in the perceived credibility of various media sources (Bayes and Druckman 2021; Druckman and McGrath 2019). Trying to accurately understand the risks associated with COVID-19 by evaluating information obtained from a source one considers credible may yield results similar to beginning with a strong previous attitude and seeking out information to confirm that opinion.

Based on this review of the literature, we propose the following hypotheses:

• H1: Physician trust in specific media outlets will reflect partisan affiliation. Republican physicians will place greater trust in conservative media outlets, and Democratic physicians will place greater trust in mainstream media outlets.

• H2: Physicians' partisan affiliation will influence attitudes about the risk of COVID-19 infection, the seriousness of the COVID-19 pandemic as reflected in news coverage, and trust in federal government institutions and in scientists to handle the pandemic. Republican physicians will express less concern about contracting COVID-19, be more likely to believe the pandemic has been exaggerated by the news media, and report less trust in federal agencies and in scientists to handle the pandemic.

• H3: Trust in media outlets will influence attitudes about the risk of COVID-19 infection, the seriousness of the COVID-19 pandemic as reflected in news coverage, and trust in federal government agencies and in scientists to handle the pandemic. Physicians who trust mainstream media outlets will express more concern about contracting COVID-19, be less likely to believe the pandemic has been exaggerated by the news media, and report more trust in federal agencies and in scientists to handle the pandemic.

• H4: The effects of partisan affiliation on the perceived seriousness of the pandemic will be contingent on media trust. Republican physicians who place more trust in mainstream news sources will be more concerned about the risk of COVID-19 infection, will be less likely to believe news coverage of the pandemic is exaggerating its seriousness, and will report greater trust in federal government agencies and in scientists to handle the pandemic.

Data and Methods

To test these hypotheses, we used an original online survey given to 625 primary care physicians in the United States. The survey research firm Dynata administered the survey from May 14 to May 25, 2021. Dynata is a widely respected survey research firm frequently used in public health research (Callaghan et al. 2022; Clipman et al. 2020; Luchetti et al. 2020; MacIntyre et al. 2021). In constructing the sample, Dynata draws from existing panels of research participants identified as primary care physicians based on the participants' listed occupation. We focus on primary care physicians for reasons outlined above. Patients place high levels of trust in them, and they can be an effective source for health messages encouraging compliance with public health directives (Findling, Blendon, and Benson 2022). Qualified panel members are randomized for inclusion in the study and are provided an incentive for participating and completing the survey questionnaire. Dynata provides a number of security checks to ensure the accuracy and reliability of survey responses, including digital fingerprints, screening questions, and removal of participants who provide incomplete or inconsistent responses or who speed through the survey. Of the 737 respondents initially identified by Dynata as potential participants, 625 qualified as primary care physicians working in internal medicine or family medicine or as general practitioners, and they completed the survey.⁸ On average, survey respondents took 15 minutes to complete the survey.

While our sample of physicians is national in scope, it is technically and necessarily a nonprobability sample. Using population benchmarks from the American Medical Association Masterfile, which is made available by the American Association of Medical Colleges, as well as income data from Wilcox (2021), we find that our sample is similar to critical population benchmarks for primary care physicians. Specifically, table 1 shows that our sample approximates population benchmarks for the proportion of primary care physicians who are Asian or Hispanic, and for physician salary. We do see some deviations between our sample and population benchmarks for race and gender, meaning we undersample Blacks and women and oversample whites and men. These differences, however, are small in magnitude, and even with these limitations, our study provides timely and representative data to investigate physician trust in news media reporting on COVID-19.⁹

Partisan Affiliation

Our theoretical interest in this article centers on partisan affiliation and its consequences for trust in media outlets and attitudes toward the COVID-19 pandemic. For these purposes, we measure partisan affiliation using a 7-point scale ranging from strong Democrat (1) to strong Republican (2) (M = 3.99; SD = 2.02). In figure 1, we present the distribution for partisan affiliation for our sample of primary care physicians. Reflective of results from previous physician-based research (Hersh and Goldenberg 2016), a plurality of physicians identify as Independent (41.9%), and there is an almost even split between physicians who identify as Democrats (29.2%) and physicians who identify as Republicans (29.1%).¹⁰ Hersh and Goldenberg (2016) reported that 35.9% of the sample registered as a Democrat, 31.5% as a Republican, and 32.6% as Independent or third party. Notably, this earlier work is based on party registration rather than party identification. Several other studies have presented estimates of physician partisan affiliation. According to a 2016 Gallup poll, 35% of doctors identified as Democrats, 27% as Republicans, and 36% as Independents (Adamy and Overberg 2019). According to this study, physicians, once reliably Republican, were increasingly identifying as Democrats. A 2015 Kaiser Family Foundation study found that 27% of doctors identified as Democrats, 27% as Republican, and 26% as Independent.¹¹ Finally, in a 2019 study on physicians' attitudes toward the Affordable Care Act, 34% of doctors identified as Democrats, 29% as Republicans, and 28% as Independents (Riordan et al. 2019).

Over time, physicians may have become more Democratic and less Republican in their partisan attachments. Bonica, Rosenthal, and Rothman (2014, 2015), for example, found changes in patterns of political contributions, with physicians increasingly giving to Democratic candidates. In a more recent study, Bonica et. al. (2020) observed that older physicians (who graduated in 1960) were more likely to be registered Republicans, while younger physicians (who graduated in 2012) were more likely to be registered Democrats (Bonica et al. 2020).

In addition to partisan affiliation, we include control variables for physician race (Black = 1), ethnicity (Hispanic and Asian), sex (female = 1), income, and age (in years).

Media Trust

We gauge physician trust in specific news media outlets by asking: “How much trust do you have in each of the following media outlets to provide accurate information about the COVID-19 pandemic?” We included a range of media outlets, including cable news (CNN, MSNBC, Fox News, and OANN), national newspapers (New York Times [NYT], Washington Post [WP], and Wall Street Journal [WSJ]), and local news. Responses were coded from 1 (indicating “none at all”) to 4 (indicating “a great deal”). For the purposes of assessing baseline levels of media trust, figure 2 presents the means and confidence intervals for our measures of media trust. As figure 2 reveals, physicians place relatively more trust in national newspapers (NYT, WP, and WSJ) and in local news sources. They place relatively less trust in mainstream (left-leaning) cable news outlets (CNN and MSNBC) and even less trust still in conservative cable news networks (Fox News, OANN). While trust in conservative media is significantly lower than trust in mainstream news outlets, a subset of our sample of physicians regard these outlets as a trusted source of news, and (as illustrated below) this trust influences their health attitudes and behavior.

In our first set of analyses, we examine the effect of partisan affiliation on trust in specific media outlets separately for each outlet. This allows us to examine patterns across media outlets and to better understand how partisan affiliation affects media trust. In our second set of analyses, we combine our measures of media trust into indices representing mainstream media outlets (NYT, WP, CNN, MSNBC, WSJ, and local news) and conservative media outlets (Fox News and OANN). This is a necessary (and, we think, parsimonious) data-reduction technique given the strong correlations in trust across these specific outlets, particularly among mainstream media outlets (CNN, MSNBC, NYT, and WP).

This decision was also guided by an exploratory factor analysis that revealed a two-factor solution with strong loadings for mainstream media outlets (CNN, MSNBC, NYT, and WP), the Wall Street Journal, and local news on the first factor and strong loadings of Fox News and OANN on a second factor (see table 2). Exploratory factor analysis groups variables together based on common (or shared) variance. To the extent that a set of variables measure the same underlying latent construct (media trust), they should load on a single factor. If, however, trust in media outlets differs in important ways depending on the type of media outlet, the media trust variables might load on two or more factors. The initial factor solution was rotated using Varimax rotation, which assumes orthogonal factors and helps to clarify the relationship between variables.

As can be seen in the analysis presented in table 2, two factors emerge from the data suggesting two underlying independent latent constructs. We label these constructs “mainstream media trust” and “conservative media trust,” reflecting trust in more explicitly ideological conservative news outlets. We should acknowledge that the labels here are subject to interpretation. We label the first factor “mainstream media trust” because it includes media outlets often perceived as liberal (e.g., NYT and MSNBC) as well as local news and the Wall Street Journal. The second factor, which includes Fox and OANN, is more clearly ideological (conservative) and partisan (Republican).¹² Based on the results from this analysis, we constructed an index for mainstream media trust based on the average trust in the media outlets that load strongly on factor 1 (M = 2.51; SD = 0.81; α = .91) and an index for conservative media trust based on the average trust placed in Fox News and OANN (M = 1.88; SD = 0.85; α = .76).¹³ The decision to include the Wall Street Journal in our index of mainstream media trust is also supported by previous research. According to a study conducted by the Pew Research Center, 35% of consistent liberals trusted the Wall Street Journal compared to 30% of consistent conservatives (Mitchell, Gottfried, and Matsa 2014). In that study, the Wall Street Journal was the only news outlet for which trust exceeded distrust across ideological classifications.¹⁴

Outcome Variables

To gauge the effects of media trust on COVID-related attitudes, we included a wide range of measures capturing physicians' personal concerns (or worry) about contracting the virus; perceptions regarding whether the seriousness of COVID-19 was being exaggerated by news media coverage of the pandemic; and trust in federal government agencies and, more generally, in scientists. We briefly describe each of these measures below.

COVID-19 worry: To determine whether media trust affected personal concerns about the virus, physicians were asked the following question: How worried, if at all, are you that you or someone in your family will get sick from COVID-19? Responses were coded from 1 indicating not at all worried to 4 indicating very worried (M = 2.86; SD = 0.82). Among the physicians included in our survey, 15.8 percent reported they were very worried that someone in their family would get sick from COVID-19, 44.4 percent reported they were somewhat worried, 32.9 percent said they were not too worried, and 6.9 percent said they were not worried at all. We expect that Democratic physicians and physicians who place greater trust in mainstream media will express greater concern that either they or a member of their family will get sick from COVID-19. Republican physicians and physicians who express greater trust in conservative media, in contrast, should report being less worried about contracting COVID-19.

Media portrayal of COVID-19 seriousness: To determine whether media trust affects perceptions of whether the pandemic has been exaggerated by the news media, we asked physicians the following: “Thinking about what is said in the news, in your view, has the seriousness of COVID-19 been generally exaggerated, generally correct, or generally underestimated?” Responses were coded 1 for physicians who thought the pandemic was “generally exaggerated,” 2 for physicians who thought coverage was “generally accurate,” and 3 for physicians who thought coverage was “generally underestimated” (M = 1.99; SD = 0.61). Among the physicians included in our survey, 19.1% thought coverage was generally exaggerated, 63.3% thought coverage was generally correct, and 17.6% thought coverage was generally underestimated. Democratic physicians and physicians who express greater trust in mainstream media should be less likely to believe the pandemic has been exaggerated by the news media. Republicans physicians and physicians who express greater trust in conservative media, in contrast, should be more likely to believe the pandemic has been exaggerated by the news media.

Trust in the Centers for Disease Control and Prevention (CDC) and scientists: To gauge institutional trust in the CDC and collective trust in scientists, physicians were asked: How much trust do you have in the following when it comes to handling the COVID-19 pandemic? Among the response options provided were leading federal agencies (the CDC and the National Institutes of Health [NIH]) and scientists. Because physician trust in the CDC and the NIH were highly correlated (r = 0.80), they were combined into a single measure gauging trust in federal government agencies. Trust in federal agencies ranges from 1, indicating no trust in either the CDC or the NIH, to 7, indicating a great deal of trust in both the CDC and the NIH (M = 5.08; SD = 1.62). The second measure used here gauges trust in scientists on a 4-point scale ranging from no trust at all to a great deal of trust (M = 3.45; SD = 0.66). Overall, physicians express a great deal of trust in scientists: 54% of physicians say they have a great deal of trust in scientists to handle the pandemic, 38.7% say they have a fair amount of trust, 6.5% report not having much trust, and 1.0% say they have no trust at all. We expect that Democratic physicians and physicians who express greater trust in mainstream media will also express greater trust in the ability of federal government agencies and scientists to handle the pandemic. We also expect that Republican physicians and physicians who express greater trust in conservative media outlets will express less trust in federal agencies and in scientists.

Results

Trust in Media Outlets to Provide Accurate Information

Before presenting the results of our full models, we begin examining differences in media trust by partisan affiliation. Figure 3 presents the mean level of trust that primary care physicians hold in specific media outlets by partisan affiliation. Consistent with H1, physician trust in various media outlets is strongly influenced by partisan identification. Democratic physicians are significantly more trusting of mainstream media outlets, such as the New York Times, Washington Post, CNN, and MSNBC. Republican physicians, in general, are less trusting of news media than Democrats. While Republicans place greater trust in Fox News than Democrats do, they do not express the same level of trust in conservative media outlets as Democratic physicians do in mainstream outlets. They also express greater trust in mainstream media outlets than Democrats express in conservative outlets. Two outlets—the Wall Street Journal and the local news—blur these partisan differences, presumably because the Wall Street Journal focuses more on financial news, and local news varies with local contexts and audience (or consumer) characteristics (Gentzkow and Shapiro 2006, 2010). Perhaps surprisingly, Democrats place greater trust in the Wall Street Journal, though the partisan differences for this outlet are muted relative to the partisan differences observed for news outlets such as the New York Times or CNN.

This pattern is similar to studies of media credibility and trust focusing on the American public (Stroud and Lee 2013; Coe et al. 2008; Ladd and Podkul 2018; Grossmann 2018), including studies focusing specifically on trust in media reporting on COVID-19 (Zhao et al. 2020). Such studies not only reveal partisan differences in trust across media outlets but also show that trust is associated with preventive behaviors (Zhao et al. 2020). The results presented in figure 3 reveal that, although the partisan effects are not quite as strong among primary care physicians, trust in media outlets is consistently correlated with partisan affiliation. That is, physicians' trust in media reflects their partisan commitments.

In table 3, we present a series of ordinal logistic regression models estimating the effect of partisan affiliation on trust in media outlets to provide accurate information about COVID-19. The results show a consistent effect of partisan affiliation on trust in media even after controlling for other demographic characteristics. Consistent with H1, we find that a physician's partisan affiliation influences how much trust they have in particular media outlets, with liberals displaying more trust in mainstream media outlets and conservatives expressing more trust in conservative outlets. While we see less of an effect of partisan affiliation on trust in the Wall Street Journal and local news outlets, partisanship remains as an important predictor of media trust in these outlets as well, with Democratic physicians expressing slightly greater trust in the Wall Street Journal and in local news than Republican physicians. In addition to the partisan effects, we find that older physicians are more trusting of the Wall Street Journal, Fox News, and their local news media. Overall, physicians are not unlike the general population. Trust in media outlets to provide accurate information about COVID-19 is largely a function of partisan affiliation.

Media Trust and COVID-Related Attitudes

How do the level of trust physicians place in various media outlets and physician partisanship influence their pandemic-related attitudes? In figure 4, we present the frequencies for physicians' COVID-19 related attitudes by partisan affiliation. Across each of these indicators, we see partisan effects. As expected, Republican physicians are less concerned that someone in their family might get sick, are more likely to believe the news media have exaggerated the pandemic's effects, and have less trust in federal government agencies and scientists to effectively handle the pandemic. For example, 22% of Democrats were very worried someone in their family might get sick compared to 10% of Republicans. Similarly, 66% of Democratic physicians said they trusted scientists a great deal when it comes to handling the pandemic compared to 46% of Republican physicians. And 30% of Republican physicians thought the seriousness of the pandemic was being exaggerated compared to 5% of Democratic physicians.

As anticipated in H3 and H4, these patterns are associated with the trust that physicians place in different media outlets. Table 4 presents ordinal logistic regressions of the relationship of physician beliefs about whether the seriousness of COVID-19 was exaggerated with partisan affiliation and trust in media outlets. In terms of modeling, we wanted to first test the effect of partisan affiliation (model 1) so we could then test the effects of mainstream and conservative media trust while controlling for partisan affiliation (model 2)¹⁵ and then finally test whether the effects of media trust are contingent on partisan affiliation.

Examining model 1 (partisan identification), we find (consistent with figure 3) that partisan affiliation consistently influenced attitudes related to COVID-19. Republican physicians were less worried about getting sick, more likely to believe the pandemic was being exaggerated, and less trusting of federal government agencies and scientists. It is worth keeping in mind that these are physicians who, as experts, might be expected to be less susceptible to the influence of partisan affiliation; yet, consistent with H2, partisan affiliation had consistent effects on their understanding of the pandemic.

In model 2 (media source), we consider how entering media trust into the model alters the effect of partisan affiliation on beliefs about media portrayals of the seriousness of the COVID-19 pandemic. First, as expected, physicians who place greater trust in mainstream media are more likely to express concern that someone in their family would get sick, more likely believe the seriousness of COVID-19 has been understated, and more trusting of federal government agencies and scientists. Physicians who place greater trust in conservative media outlets, in contrast, express less concern about contracting the virus, believe the seriousness of the pandemic has been exaggerated by the news, and are less trusting of federal agencies and scientists. Second, once we include our measures of media trust in the model, partisan affiliation no longer has a significant effect on these COVID-19-related attitudes. This tell us the effects of partisan affiliation are indirect, meaning they influence media trust (table 2), which, in turn, influences beliefs about COVID-19 and “explain away” the effect of partisan affiliation. This is consistent with theories of motivated reasoning as well as more recent work by Druckman and others who argue that partisan differences in media use may reflect less on directionally motivated reasoning than on perceptions of source credibility (Bayes and Druckman 2021; Bolsen and Druckman 2018; Druckman and McGrath 2019).

Finally, in model 3 (interactive model), we test whether the effects of media trust are contingent on partisan affiliation. In all but one of the models, we find significant interaction effects between mainstream media trust and partisan affiliation, suggesting that partisans react differently depending on their level of trust in mainstream media outlets. Because interaction terms can be difficult to interpret, we present a visualization of these effects in figure 5. For the purposes of this analysis, we present the marginal effects for self-identified strong Republicans and strong Democrats.

As figure 5 demonstrates, the contingent effects of partisan affiliation and media trust primarily affect Republican primary care physicians. Republican physicians who express little trust in mainstream media worry less about getting sick, believe that the seriousness of the pandemic is being exaggerated by the news media, and report less trust in federal government agencies. As trust in mainstream media outlets increases, Republican physicians look more like Democratic physicians. They worry more that either they or someone in their family will get sick, they are less likely to believe media coverage is exaggerating the pandemic, and they report more trust in federal government agencies to handle the pandemic. While we fail to find a similar effect on trust in scientists, keep in mind that both partisan affiliation and trust in media influence trust in scientists. The effects of media trust are not, however, contingent on partisan affiliation. Independents follow a similar path, although the effects are much smaller (not shown). There is no similar consistent effect for Democratic physicians and trust in conservative media outlets.

Conclusions

Primary care physicians are a trusted source of information for the general public and play a particularly important role during public health crises, such as the COVID-19 pandemic. We typically assume that physicians, because of their expertise, will be able to quickly and effectively filter through the noise and confusion of breaking news to provide patients with the most up-to-date and accurate health information possible. Yet, even in the best of times and with a consensus on the appropriate action, the news media can sensationalize health news, subsequently misinforming their audiences. Within this environment, primary care physicians can help patients understand the importance of preventive health behaviors and encourage compliance with public health directives.

COVID-19, however, presented a unique challenge in that there was no political consensus regarding the best course of action for combating the pandemic. These differences were reflected in news coverage of the pandemic, with Fox News and other conservative media outlets presenting a much different picture of the risks of infection, the pandemic's severity, the need for adopting preventive health behaviors, and the credibility of federal government agencies and scientists (Motta, Stecula, and Farhart 2020). How did these differences affect physician attitudes?

First, partisan affiliation influenced the level of trust physicians placed in specific media outlets. Media trust, in turn, influenced how worried primary care physicians were that either they or someone in their family would get sick with COVID-19, whether they believed the pandemic was being exaggerated in the news media, and their trust in federal government agencies and scientists. Physicians who placed trust in mainstream media outlets expressed greater concern someone in their family would get sick, were more inclined to believe the pandemic was being portrayed accurately, and reported greater trust in federal agencies and, more generally, in scientists. Physicians who trusted conservative media outlets, in contrast, expressed less concern someone in their family would get sick, were more likely to believe the media were exaggerating the seriousness of the pandemic, and expressed less trust in federal agencies. In many respects, physicians' responses to news about the pandemic mirrored the public. Partisanship mattered a great deal. As medical experts, physicians may be more resistant than the general public to the influence of partisanship and partisan news, but they are not entirely immune.

This is no trivial matter. If primary care physicians worry less about infection, believe the pandemic is being exaggerated, and have less trust in scientists and federal agencies, they may have failed to encourage patients to take preventive actions such as wearing masks, social distancing, or getting vaccinated, or they may have even actively discouraged patients from taking these actions. However, we also find evidence that the effects of partisanship on pandemic-related attitudes may be overcome, provided physicians have a base level of trust in mainstream media outlets. That is, it appears that physicians are subject to the same political forces as the general public; yet because of their expertise, they are also better positioned to resist these forces provided they have trust in mainstream news sources.

Given the polarized political environment and the fragmentation of news audiences, it is not clear whether (or how) the mainstream news media might increase physician trust in its reporting and, in doing so, advance the implementation of public health directives. Future research should investigate the conditions under which physicians who consume news from nonmainstream sources might be more likely to consume news from more mainstream outlets. One possibility is that Republican physicians may place higher levels of trust in mainstream media when they discuss politicized issues without direct reference to political disagreements or conflict more broadly (i.e., without emphasizing which side is “winning” and “losing”). We suspect that politicized content does little to inspire physician trust. As evidence of this point, we would note that physician trust in the Wall Street Journal and local news was less partisan, perhaps because these sources eschew the sort of conflict-driven narratives that drive web traffic, outrage, and distrust. Replacing those conflict-driven narratives with a facts-only (or facts-first) approach could be crucial to establishing trust in public health reporting.

We would be remiss if we did not conclude by noting the limitations of our study. To our knowledge, our sample of primary care physicians is the best sample available for understanding how doctors learned about and responded to the pandemic. Even so, it is a convenience sample. While we are able to provide some evidence that the sample mirrors physician demographics, we cannot be certain that our sample reflects the larger population of primary care physicians. Second, our data were necessarily collected at a single point in time (May 2021). As a result, we cannot evaluate how physician attitudes changed over the course of the pandemic. It is possible that physician trust in media outlets shifted over time, which, in turn, influenced perceptions regarding the seriousness of the pandemic. Indeed, we would be surprised if this were not the case. The more agreement there is across media outlets regarding the seriousness of the pandemic and the need for preventive behaviors, the fewer differences we should observe. Even with these limitations, however, we are confident in the core findings.

While primary care physicians may be more resistant than the general public to media messages, they are not immune to the broader political environment. Even among physicians, partisan affiliation influences trust in media outlets, and media trust influences attitudes toward the pandemic. As a result, Republican primary care physicians who trust conservative media outlets respond very differently from Democratic primary care physicians who trust mainstream outlets. The news matters, and it matters in ways that affect public health and the conversations being had in the doctor's office.

Notes

References