A team of economists has published a National Bureau of Economic Research (NBER) working paper that accuses Fox News host Sean Hannity of killing thousands of people through “misinformation” about the coronavirus pandemic.
The paper, “Misinformation During a Pandemic,” was funded by the Becker Friedman Institute at the University of Chicago, but largely consists of partisan hackery dressed in statistical models to smear a popular pro-Trump voice.
To begin with, at least one of the authors of the paper, David Yanagizawa-Drott, appears to dislike President Donald Trump intensely. He complained in 2017 that then-New England Patriots quarterback Tom Brady supported the president. He also encouraged protests against the Trump administration in 2017, and again in 2018.
— D. Yanagizawa-Drott (@YanagizawaD) January 30, 2017
Hoping for some external validity. #FamiliesBelongTogether
— D. Yanagizawa-Drott (@YanagizawaD) June 30, 2018
That would not, by itself, invalidate the analytical conclusions of the paper. But the introduction to the paper reveals more about its partisan bias.
The first footnotes are references to articles in The Washington Post and the New York Times attempting to claim that Fox News had put Americans in “danger” through “misinformation.”
The paper itself focuses on Hannity. It uses survey data and economic models in an attempt to argue that because fellow Fox News host Tucker Carlson focused on the coronavirus pandemic from an early stage — in late January — and Hannity did not devote much attention to it until weeks later, there are observable effects of that editorial difference in the personal behaviors and health outcomes of the audiences watching each show on the network.
They say: “Consistent with a persuasive effect of content on behavior, we find that viewership of Hannity is associated with changing behavior four days later than other Fox News viewers; while viewership of Tucker Carlson Tonight is associated with changing behavior three days earlier (controlling for demographics and viewership of other shows and networks” (original emphasis).
Watching Hannity, they say, “is associated with a greater number of COVID-19 cases and deaths.”
The authors present a detailed, sophisticated set of analytical tools, perhaps designed to impress the causal reader with the trappings of intellectual rigor.
But a closer look at the study reveals glaring errors.
First of all, the authors begin measuring “misinformation” by counting the number of times the coronavirus was mentioned on each show. Hannity did not mention it much during February, while Carlson mentioned it often.
That is not “misinformation”; that is a decision to express opinions on other news topics.
Another tool the authors use to measure “misinformation” is to code — “yes” = 1 or “no” = 0 — based on whether each mention of the coronavirus on each show described it as a sufficiently dangerous threat.
The data for most of January and February are useless as a basis for comparison, because Hannity was not mentioning the coronavirus much at all.
Toward the end of February, both shows mention the virus with the same frequency. There is a gap between how seriously each show portrays the virus, but that gap only lasts until mid-March.
What the authors omit is that during that same period, some scientists — including Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases, were also telling the public that the risk to Americans was low. Public health officials gave similar advice well into March — not just on Fox News, but everywhere. Fauci even said on March 9 that it was safe for otherwise healthy people to go on cruises or political rallies.
The authors do not examine coverage on left-leaning networks, such as MSNBC and CNN, which focused obsessively on the partisan impeachment trial during the 20 days after the coronavirus first reached the U.S. on January 15.
It would make as much sense — perhaps more — to blame those networks for “misinforming” the public. Once those networks turned to the coronavirus, ostensibly providing more reliable information, they had already alienated many viewers.
The authors also used a survey to determine whether their behavior was affected by watching either Fox show. The survey asked over 1,000 respondents to provide the exact date (month and day) when they “first significantly change[d]” any personal behavior in response to information about the coronavirus pandemic.
The authors acknowledge that respondents might “systematically misremember that they changed their behavior earlier or later than they actually did.”
It is not clear how the authors deal with that problem.
What is clear is that they rely heavily on “qualitative evidence” — a few selective quotes from Hannity and Carlson that reflect as much about the bias of the authors as they do about the opinions of the anchors.
For example, the authors cite Hannity complaining that Democrats were “weaponizing” and “politicizing” the coronavirus — a perfectly defensible opinion that does not preclude the seriousness of the virus itself.
That pseudo-analysis forms the basis for accusing Hannity of killing thousands of people.
No doubt the paper will be used as part of an ongoing campaign to attack Hannity and encourage advertising boycotts, part of a partisan effort to remove one of the most popular pro-Trump pundits from the airwaves just months before the presidential election.
But it brings to mind the saying popularized by Mark Twain: “There are three kinds of lies: lies, damned lies, and statistics.”
Breitbart News submitted the following list of questions to the paper’s authors:
1. Who funded the research? Was it the BFI alone, or was the BFI deriving funds for the study from an outside party? Were there other sources of funds?
2. Given that the paper effectively accuses Sean Hannity of killing people, do the authors perceive any ethical problems with their study?
3. Is it appropriate for an economics paper to follow the lead of the Washington Post and New York Times in blaming Fox News for misinforming the public, when both of these publications are left-leaning competitors of that network? (Both have also published information about the pandemic that later turned out to be incorrect.)
4. At least one of the authors, Yanagizawa-Drott, has a number of anti-Trump social media posts. To what extent did your own political bias drive your research?
5. Why did the paper not consider the effects of other, “mainstream” sources discrediting themselves through prior false reporting — such as years of obsession with a “Russia collusion” conspiracy theory that was subsequently debunked?
6. Why did the paper not consider the extensive focus of the U.S. media on the partisan impeachment trial during the first 20 days in which the coronavirus had arrived in the U.S.?
7. How did the paper deal with other influences — such as advice from Dr. Anthony Fauci, as late as March 9, that it was safe to attend rallies or travel on cruises?
8. Your analysis excludes New York. Given that the bulk of cases and deaths in the period studied were in the heavily-Democratic New York metropolitan area, does a perceived difference between two television shows primarily watched by Republicans tell us anything meaningful about the pandemic?
9. Why does a paper on “misinformation” study two opinion shows, rather than news programs?
10. In the “qualitative” section, why does the paper consider a statement of opinion — such as the supportable claim that Democrats were “politicizing” the virus — to be “evidence” of misinformation?
11. What precise “misinformation” did Hannity provide about the virus? The only example given is a comparison to other causes of death.
No response was received, despite multiple attempts.
Joel B. Pollak is Senior Editor-at-Large at Breitbart News and the host of Breitbart News Sunday on Sirius XM Patriot on Sunday evenings from 7 p.m. to 10 p.m. ET (4 p.m. to 7 p.m. PT). His new book, RED NOVEMBER, is available for pre-order. He is a winner of the 2018 Robert Novak Journalism Alumni Fellowship. Follow him on Twitter at @joelpollak.