How researchers get heard
Abstract lines

Unmasking Your Priors

OK: which of you is wearing a mask now or about to start? And why?

The US government seems on the verge of reversing its position on whether all of its citizens should now wear some sort of covering over those noses and mouths while in public.

Some of the credit for that reversal has to go to Zeynep Tufekci, who wrote a New York Times op-ed two weeks agoarguing that a) wearing masks is clearly protective against the virus that causes COVID-19 and b) the US government lied to its citizens about that efficacy to prevent a run on the scarce supply of masks, thus saving that supply for US health care workers.

Tufekci’s column was shared widely on social media; it was spread further through promotion by elite media types (see, for instance, David Leonhardt’s “The Mask Fiasco: Next time, how about the truth?”). Ben Thompson of Stratechery called it “probably the most important piece in terms of changing the narrative around masks.”

Tufekci’s piece is heavy on rhetoric and assumptions and, arguably, lighter on evidence. She asserts that mask-wearing “has always been advised as part of the standard response to being around infected people.” As evidence for that assertion, she links to an archived, un-updated CDC interim recommendation web page made during the 2009 H1N1 outbreak. Her Times’ op-ed successfully links to only one study — a 2013 paper on surgical masks and aerosolized influenza virus that concluded such masks have “demonstrated limitations” although “are to some extent protective.” Tufekci also links to a number of news reports and Twitter feeds with Asian public health experts asserting that the use of masks was an essential part of their battle against the COVID-19 virus. Out of this collection, she makes her argument about government miscommunication and deception.

Tufekci isn’t an epidemiologist or a virologist. She teaches information science at the University of North Carolina, and has written extensively about the role of social media in social uprisings. As part of that work, she’s done a lot of research in the midst of such uprisings in the Middle East and Asia. As she told Thompson in an interview he published yesterday for Stratechery subscribers, she saw widespread mask-wearing in Asia during the SARS outbreak there and ordered a supply for herself in January when she first heard about the COVID-19 outbreak. She also told Thompson that, not being an expert, she was reluctant to write about the government’s messaging on masks, but after her op-ed was published she got “lots of doctors, medical people, epidemiologists, people from our government, other governments, just writing me, thanking me, telling me that they just didn’t dare say this, they couldn’t go against the CDC.”

So here we have a research authority writing outside her lane for an elite outlet with an argument that many casual observers find persuasive; that some experts find gives voice to an opinion they can’t dare express within the confines of their discipline’s power structure; and that links up with an existing media narrative about the untrustworthiness of those in power. A thought leadership success story, of sorts.

Scott Alexander has also written a blog post that disagrees with Tufekci’s conclusion about the government lying.

Alexander isn’t a public health expert, either. He’s a psychiatrist and the author the long-running rationalist blog Slate Star Codex. His blogging often has him doing copious lit reviews on topics of moment.

So: What Alexander does in his post that Tufekci doesn’t is systematically review a great deal of medical literature on the efficacy of masks.

In that post, Alexander reviews as much of the research as he can get his hands on about whether mask-wearing is protective and for whom. He also talks about his own experience trying to put on a mask correctly as a doctor working in a hospital, and how difficult it is to use an N95 respirator correctly. His conclusion:

Please don’t buy up masks while there is a shortage and healthcare workers don’t have enough.

If the shortage ends, and wearing a mask is cost-free, I agree with the guidelines from China, Hong Kong, and Japan – consider wearing a mask in high-risk situations like subways or crowded buildings. Wearing masks will not make you invincible, and if you risk compensate even a little it might do more harm than good. Realistically you should be avoiding high-risk situations like subways and crowded buildings as much as you possibly can. But if you have to go in them, yes, most likely a mask will help.

In low-risk situations, like being at home or taking a walk, I mean sure, a mask might make you 0.0001% (or whatever) less likely to get infected. If that’s worth it to you, consider the possibility that you might be freaking out a little too much about this whole pandemic thing. If it’s still worth it, go for it.

You are unlikely to be able to figure out how to use an N95 respirator correctly. I’m not saying it’s impossible, if you try really hard, but assume you’re going to fail unless you have some reason to think otherwise. The most likely outcome is that you have an overpriced surgical mask that might make you incorrectly risk-compensate.

If you are a surgeon performing surgery, bad news. It turns out surgical masks are not very useful for you! You should avoid buying them, since doing so may deplete the number available for people who want to wear them on the subway.

Alexander also links to exactly the same archived CDC link that Tufekci’s piece does — pointing out a table on that page that advises almost every group not to wear a mask during the H1N1 outbreak.

Two observations:

  1. The power of already established authority plus platform plus resonant argument trumps evidence. This point might be obvious to everyone except scientists. Alexander has a lot of authority with the small, rationalist audience for his blog, but doesn’t do interviews or podcast or content for elite media. Tufekci writes frequently for Wired and the Times and is very active on Twitter and connected with media there — so, has a potentially Pinkeresque kind of authority when she ventures out of her lane to make this argument. She also has the cachet of the Times behind her. And she’s also making an argument that resonates with a frightened, authority-mistrustful readership.
  2. We’re on our own when it comes to figuring this out. There is no research body or consortium we can turn to that has assembled the best available evidence on mask efficacy in similar contexts. So — like Tufekci and Alexander — we have to assemble it ourselves, and (if we’re not scrupulous, and most of us aren’t) that assemblage is subject to our priors.

BTW: Alexander also reviews what he thinks is the only published RCT study on the efficacy of improvised cloth masks to block virus particles. That study concludes cloth masks “cause an increase in infection risk in HCW (health care workers).” In other words: don’t use them.

Alexander thinks the study is flawed, and asks for his community’s help in reviewing it. Here’s his conclusion:

Somebody with more experience and statistical knowledge than I have should be looking this over with a fine-toothed comb and trying to figure out what we should do.

Until then, should people stay away from cloth masks? I’m not sure, and this is so not a recommendation, but I lean toward no. The prior that they should work or at least be neutral is too high for a study this weak to convince me otherwise. More important, this study only examines incoming pathogens. Even if they are harmful for blocking incoming pathogens, there are still reasons to think they are helpful for blocking outgoing ones. If I had to hang out with a coronavirus patient for a while, and I had to choose between both of us wearing cloth masks, or neither, I would go with the masks. Only until we could get real surgical masks, which are much better. But I’d go with the cloth ones instead of nothing.

But right now that’s a gut judgment, and the evidence says I’m wrong. This is one of those times where people have to make a life-or-death decision in conditions of high uncertainty, and it really sucks.

Is he right? Who knows? The government and the research community — inexcusably — aren’t summing this up for us.

Instead, we’re looking to non-expert pundits and bloggers. And finding through their priors maybe what we wanted to find all along.