In recent weeks, three studies have focused on conservative media’s role in fostering confusion about the seriousness of the coronavirus. Taken together, they paint a picture of a media ecosystem that amplifies misinformation, entertains conspiracy theories and discourages audiences from taking concrete steps to protect themselves and others.
The end result, according to one of the studies, is that infection and mortality rates are higher in places where one pundit who initially downplayed the severity of the pandemic — Fox News’s Sean Hannity — reaches the largest audiences.
“We are receiving an incredible number of studies and solid data showing that consuming far-right media and social media content was strongly associated with low concern about the virus at the onset of the pandemic,” said Irene Pasquetto, chief editor of the Harvard Kennedy School Misinformation Review, which published one of the studies.
Misinformation and conspiracy theories
In April, Kathleen Hall Jamieson of the Annenberg Public Policy Center and Dolores Albarracin of the University of Illinois at Urbana Champaign published a peer-reviewed study examining how Americans’ media diets affected their beliefs about the coronavirus.
Administering a nationally representative phone survey with 1,008 respondents, they found that people who got most of their information from mainstream print and broadcast outlets tended to have an accurate assessment of the severity of the pandemic and their risks of infection. But those who relied on conservative sources, such as Fox News and Rush Limbaugh, were more likely to believe in conspiracy theories or unfounded rumors, such as the belief that taking vitamin C could prevent infection, that the Chinese government had created the virus, and that the U.S. Centers for Disease Control and Prevention exaggerated the pandemic’s threat “to damage the Trump presidency.”
These findings held even after controlling for viewers’ political affiliation, education, gender and age.
The authors pinpoint several examples of misinformation circulating in conservative media. For example, on March 6, Fox medical contributor Marc Siegel stated on Hannity’s popular evening program that “the virus should be compared to the flu. Because at worst, at worst, worst-case scenario it could be the flu.”
Similarly, in February, Sen. Tom Cotton (R-Ark.) appeared on Fox’s “Sunday Morning Futures” to suggest, without evidence, that the virus was connected to a research laboratory in Wuhan, China.
Beliefs lead to actions
A working paper posted by the National Bureau of Economic Research in May examined whether these incorrect beliefs affected real-world behavior.
The authors used anonymous location data from millions of cellphones to explore how the popularity of Fox News in a given Zip code related to social distancing practices there. By March 15, they found, a 10 percent increase in Fox News viewership within a Zip code reduced its residents’ propensity to stay home, in compliance with public health guidelines, by about 1.3 percentage points.
Given total stay-at-home behavior increased by 20 percentage points during the study period, that effect size is “pretty large,” said Andrey Simonov, the study’s lead author. It’s comparable to Fox’s persuasive effect on voting behavior, as identified in a 2017 paper by a different team.
“The effect that we measure could be driven by the long-term message of Fox News, which is that the mainstream media often report ‘fake news’ and have a political agenda,” Simonov said. “This could result in lowering trust in institutions and experts, including health experts in the case of the pandemic.”
It’s plausible, of course, that this difference in behavior could be attributed to other characteristics of Fox viewers, such as their age or political ideology. To control for these factors, the authors used “the quasi-random assignment of each news channel’s relative position across cable markets” as an instrumental variable.
Because TV stations with lower channel numbers tend to be watched more than those with higher numbers, and because those numbers are assigned in a more-or-less random way across TV markets, the authors were able to measure exposure to Fox News driven not by personal tastes, but by differences in channel lineups. Rather than asking “How does Fox News viewership affect social distancing,” the analysis asked “How does viewership of Fox News due to low channel position affect social distancing.”
While a big step beyond simple correlation, this type of analysis was nonetheless unable to definitively say that Fox programming caused viewers to be less compliant with social distancing guidelines.
How conservative media consumers’ behavior could worsen the pandemic
Another recent working paper, by economists at the University of Chicago and other institutions, similarly finds that Fox News viewers are less likely to comply with public health guidelines than consumers of other media. But their paper takes the analysis two steps further: It finds that Fox viewers aren’t a monolith, with fans of some media personalities acting distinctly from others. It also provides evidence that those behavioral differences are contributing to the spread of the coronavirus and mortality rate of covid-19 the disease it causes, in certain areas.
The study focuses specifically on the differences in how Fox’s Tucker Carlson and Hannity discussed the pandemic during its early days.
“Carlson warned viewers that the coronavirus might pose a serious threat from early February,” the researchers wrote, “while Hannity first ignored the topic on his show and then dismissed the risks associated with the virus, claiming that it was less concerning than the common flu and insisting that Democrats were using it as a political weapon to undermine the president.”
To quantify this claim, the authors recruited people via Amazon’s Mechanical Turk [MTurk] service to rate the seriousness with which the hosts discussed the virus. The coders consistently rated Carlson’s coverage as significantly more serious than either Hannity’s or other Fox hosts, particularly in February. The tones began converging in March; by the middle of the month, Hannity, Carlson and other Fox hosts were discussing the outbreak in similar terms.
Then they identified media markets where Hannity is more popular than Carlson to determine whether this coverage influenced people’s behavior. Like Simonov and his colleagues, they used an instrumental variable to sidestep the inherent demographic differences in the show’s two audiences: in this case, variations in local sunset times, which influence how likely people are to stay up later to watch “Hannity,” which typically airs later than “Tucker Carlson Tonight.”
They then fielded a survey to Fox News viewers 55 and older to see how their behavior differed in areas where Hannity’s show is more or less popular than Carlson’s program.
They found that Hannity viewership was associated with changing pandemic-related behaviors (like hand-washing and canceling travel plans) four days later than other Fox News viewers, while Carlson viewership was associated with changing behaviors three days earlier.
Given the importance of individual behavior in curbing the spread of the coronavirus, it stands to reason that places where people were slower to take preventive steps might see more severe outbreaks. That’s exactly what the final step of their analysis shows.
“Our results indicate that a one standard deviation increase in relative viewership of Hannity relative to Tucker Carlson Tonight is associated with approximately 32 percent more COVID-19 cases on March 14 and approximately 23 percent more COVID-19 deaths on March 28,” the authors write. They further note that those differences fade beyond March, as the two hosts’ coverage had largely converged by then.
In a statement, provided to news outlets writing about these studies, a Fox News spokesman said: “as this timeline proves, Hannity has covered Covid-19 since the early days of the story. The ‘study’ almost completely ignores his coverage and repeated, specific warnings and concerns from January 27-February 26 including an early interview with Dr. Fauci in January. This is a reckless disregard for the truth.”
In response, the study’s lead author Leonardo Bursztyn noted that their analysis encompasses the entirety of Fox’s prime-time coverage through the end of March, including Hannity’s interview with Fauci. “There’s no ‘cherry-picking’ possible because our independent MTurk coders read every transcript between late January and late March.”
* * *
“If the results hold, the research demonstrates the influence that broadcast media can have on behavior,” said Jevin West, director of the Center for an Informed Public at the University of Washington.
It’s important to stress that the Simonov and Bursztyn papers have not yet gone through a formal peer-review process, which typically assesses whether the evidence presented in the research supports the claims made. And as with much social science research, definitively proving causality — that Fox News coverage caused people to change behavior, which in turn caused the pandemic to spread more — is beyond the scope of the methods used.
“The results are suggestive of this but not conclusive,” West said. “It could be, for example, that long-standing mistrust of government or influence from current political leaders were the main drivers.”
However, the studies took a number of steps — including controlling for demographic factors, using instrumental variables to sidestep known and unknown confounders, and running tests for alternative explanations — to strengthen their claims. The University of North Carolina’s Zeynep Tufekci, who has written extensively on the American response to the pandemic, praised the Bursztyn study for its rigor.
Nevertheless, not all researchers who have seen these studies have been persuaded. The University of Chicago’s Anthony Fowler has written an op-ed critical of one of the study’s methods, and said in an email that he’s “skeptical” of the findings of the other two papers.
“I think there are good studies showing that different news outlets do have different ideological slants and that these slants can affect beliefs and behaviors,” he said. But he said he’s not convinced by these particular ones.
In the view of Pasquetto, the Harvard editor, the balance of evidence presented in the recent studies is strong. “Given all the data we have seen, and all the studies we are reviewing, we can say that empirical evidence clearly shows that this social group [those who routinely watch, read, and follow far-right media and social media] tended to take the disease less seriously and delayed their own response to the virus,” she said.