What counts as evidence? Universal face mask use in the COVID-19 crisis

On April 3, 2020, the United States Centre for Disease Control (CDC) changed its advice on the universal use of face masks by citizens in public. While the CDC had been discouraging universal use of face masks, it now recommends it. On April 4, the World Health Organization (WHO) made a similar about-face. Around the world, there have been noticeable differences in official policy on this issue: mass use of masks has been a feature in the response of places like Hong Kong and South Korea, which were initially held up as successful models. In general, the response in Europe and the US has been striking for the way it avoided mass adoption of face masks, even though this is not a consistent pattern: the Czech Republic and some small parts of Germany have mandated mask use for citizens in public, for example.

The CDC now recommends the use of cloth masks by the public. Image by Christo Anestev from Pixabay

All of these different governments say that they are following expert advice. And if we look at what the experts are saying as they are being reported in the media, it is clear that there are radically different positions being adopted.  At the same time, these experts are reaching their conclusions based on scientific data and scientific modes of reasoning. How is it possible for members of the scientific community to reach such radically different positions? How do those experts justify such different stances?

The role of language and discourse

As someone who is interested in language and communication, it’s natural for me to approach this by looking at the way that experts express themselves, that is, the language that they use. In this post, I’m not trying to say that one side is right and one side is wrong. I’m interested in the way that experts make and present their arguments and recommendations. One thing that strikes me is that the way that experts on different sides of this debate talk about uncertainty and limitations of scientific knowledge can be revealing.

I’d like to take a look at a couple of articles that appeared in the Guardian on April 2 and 3, 2020, and that featured the views of expert professionals. A bit of background about these articles:

Knowns and unknowns (Howard)

In the April 2 article, Howard is asked, ‘Is there any way to disaggregate the widespread use of face masks from other measures – like social distancing – to understand how effective they are?’ Here is his full response:

“We absolutely can’t. There are four measures – rigorous testing, contact tracing, quarantine of potentially infected persons and universal mask-wearing – that represent a known good recipe. We don’t know exactly which combination of those things works and how important each one is. We’ve just got to do what works.”

Example 1

The response references both ‘knowns’ and ‘unknowns’. He talks about a ‘known good recipe’ that includes universal mask use but also notes that ‘we don’t know’ how important, say, universal face mask use is as an ingredient in that mix. He omits to provide any evidence for the ‘known good recipe’, either taking this for granted or assuming that readers will supply it for themselves.

Absence of evidence (Heymann)

In the April 3 article, Heymann provides a longer, more nuanced take on available scientific knowledge. The article begins by reviewing what we know we can do in COVID-19, then reviews what the evidence says about face masks, and concludes by considering what we should now do. In this last section, he lists one possible reason for the public wearing masks (‘as a precaution’) and four possible reasons why the public might not.

With the COVID-19 crisis, expert professionals like David Heymann have been called on to explain the situation to the public. Image (CC BY 2.0) by Don Pollard/Council on Foreign Relations, Chatham House on Flickr

In this article, the notion of ‘evidence’ plays an important role. For example, here is how the writer characterises existing measures, not including mask use:

“We know from scientific evidence, as well as what we have learned from other countries further ahead in their epidemics, that these things work.”

Example 2

Here is how he presents knowledge about face mask use:

“There is a lack of good, robust evidence on the effectiveness of standard face masks worn by the public.”

Example 3

We don’t even have good, case-controlled studies about how effective face masks are at preventing the spread of influenza, which is the model for respiratory virus diseases. There have been some studies comparing a group of people who got flu with a group of people who didn’t get flu, which asked retrospectively whether they wore face masks, but they don’t convincingly tell us that it was the face mask, rather than something else, that was effective in preventing transmission of flu.”

Example 4

Example 4 is also interesting because the writer also notes limitations in existing scientific studies, noting that the evidence is not convincing.

Different standards of evidence?

Howard and Heymann seem to be using different standards of evidence and this leads them to adopt a fundamentally different stance on the issue of universal face mask use. Heymann makes the standard explicit when he says:

“One of the best forms of evidence in medical research is a randomised controlled trial… We would have difficulty doing this kind of trial of face masks, because it’s impossible for participants to be unaware of whether or not they’re wearing a mask.”

Example 5

Even though both scientists make claims about what is known and unknown, their basis for doing so would seem to be different. While not made explicit, Howard is apparently relying on the experience to date of countries fighting the coronavirus to support the need for universal mask use. Meanwhile, Heymann relies on the lack of randomised controlled trials to establish that there is an absence of ‘robust evidence’ for this kind of mask use. Yet, at the same time, Heymann does cite ‘what we have learned from other countries further ahead in their epidemics’ when approving other methods like social distancing. There is at least the possibility that ideological considerations could influence the standard of evidence that is applied. In the political arena, we might therefore see different kinds of evidence being deployed to justify different political positions.

Even if Heymann’s stance in the April 3 article did not seem very sympathetic to the universal use of face masks, this stance may have changed. In an April 2 BBC interview, he appears more open to the possibility that masks could be effective, saying ‘it might be that wearing a mask is equally as effective or more effective than distancing, provided that mask is worn properly and provided that people don’t infect themselves when they’re taking the mask off and touch an outer surface which may be contaminated.’ Here it’s interesting how carefully he presents this position, using a raft of what applied linguists call ‘hedging devices’ (e.g. ‘it might be’, ‘provided that’) to refrain from committing entirely to a position. Heymann is careful here to present the state of knowledge as uncertain and avoid a black-and-white ‘answer’.

Presenting policy arguments

Ultimately, a lack of consensus in expert knowledge is nothing new. New knowledge emerges and is ‘perfected’ when scientists identify and address pre-existing gaps by applying rigorous scientific procedures. When it comes to presenting convincing policy arguments though, scientists’ training to acknowledge limitations in claims and understandings works against them. This goes well beyond COVID-19 and applies to things like the climate crisis too. Policymakers and the public often think they need absolute certainty rather than contingent advice. However, this is something that can be very difficult for experts to provide.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: