I posted recently about the discussion between Phil Magness and Jeremy Horpedahl about mask mandates to deal with COVID-19. My sometimes co-author and former student Charley Hooper wrote the following on masks in a recent email. He’s given me permission to share it. The bottom line: the evidence in favor of masks, let alone mandates, just does not seem to be there.
Here’s Charley:
The only randomized controlled trials conducted to study the effects of wearing masks and washing hands show that those two preventative techniques don’t significantly reduce the spread of the influenza virus. In some studies they help a bit. In other studies, they hurt a bit.
“Although mechanistic studies support the potential effect of hand hygiene or face masks, evidence from 14 randomized controlled trials of these measures did not support a substantial effect on transmission of laboratory-confirmed influenza.”
[Source: https://wwwnc.cdc.gov/eid/article/26/5/19-0994_article]
Why did I mention influenza and not COVID? COVID-19 is supposed to be transmitted by the same mechanism as influenza and influenza has been around long enough to be better studied.
The Xiao study referenced above contains this shocking admission: “It is essential to note that the mechanisms of person-to-person transmission in the community have not been fully determined. These uncertainties over basic transmission modes and mechanisms hinder the optimization of control measures.” Scientist don’t have a handle on how the flu transmits throughout the community. If you don’t know that basic fact, it’s pretty hard to effectively prevent the transmission of influenza! By extension, I think it’s safe to say that scientists don’t understand how the SARS-CoV-2 virus is transmitted.
The influenza virus can last about five minutes on a human hand. (“Virus survived on hands for up to 5 min after transfer from the environmental surfaces.”) I suspect that the SARS-CoV-2 virus lasts about the same length of time on hands.
[Source: https://pubmed.ncbi.nlm.nih.gov/6282993/]
Therefore, if you don’t wash your hands but also don’t touch your eyes, nose, or mouth shortly after touching an infected surface. you should be fine.
There was one randomized controlled trial of the use of face masks to prevent COVID-19. The study was conducted in Denmark in April and May 2020. The results were not statistically significant but showed that the mask group suffered a 1.8% infection rate while the control group suffered a 2.1% rate (95% confidence intervals = 46% reduction to 23% increase due to masks). In other words, masks helped but were not a panacea.
[Source: https://www.acpjournals.org/doi/full/10.7326/M20-6817]
Other studies show the benefits of wearing masks, but they are correlational, population-based studies that identify relationships but don’t necessarily prove cause and effect. I’m not saying that masks don’t work. I’m instead highlighting some of the scientific uncertainty around the use of masks. With this uncertainty, COVID absolutism is unjustified and harmful.
The rule seems to be that the less people understand about something, the more adamant their beliefs.
READER COMMENTS
AMT
Feb 17 2021 at 6:48pm
“wearing masks and washing hands show that those two preventative techniques don’t significantly reduce the spread of the influenza virus”
Ok, you can stop washing your hands then.
“the mechanisms of person-to-person transmission in the community have not been fully determined”
You need to have enough intelligence to understand what this means. Not “FULLY determined” means we aren’t sure if we know absolutely everything there is to know. Maybe, there MIGHT, theoretically, be some other additional avenue of transmission that we haven’t thought of. This doesn’t mean washing your hands and wearing a mask is useless. “By extension, I think it’s safe to say that scientists don’t understand how the SARS-CoV-2 virus is transmitted.” Wrong, and a terrible argument. Why don’t you argue economists have no idea how prices are set because we don’t “fully” understand everything about psychology and therefore how people may or may not deviate from homo economicus? Because that is an equally stupid argument.
Jon Murphy
Feb 17 2021 at 9:46pm
You’re strawmanning Charley’s point. He’s not saying that masks are useless, that washing one’s hands is pointless, or that we have no clue how COVID is transmitted. What he is saying is the data are not as clear as Fauci and many others claim. On the margin, hand-washing and mask wearing may help some. But mask mandates and the like are unlikely to prevent spread of the disease.
I’ve had many interactions with Charley. “Stupid” and “unintelligent” are not how I’d classify him. You ought to be very careful here…you’re making the mistakes you’re accusing him of making.
AMT
Feb 17 2021 at 11:59pm
Not strawmanning his point: “I think it’s safe to say that scientists don’t understand how the SARS-CoV-2 virus is transmitted.” I will very confidently and safely stand by my “stupid argument” comment.
Jon Murphy
Feb 18 2021 at 6:47am
Ok. Now you’re strawmanning my comment.
Now, you can stand by your comment all you want. But know that you are ignoring what the studies say to do so.
AMT
Feb 18 2021 at 11:23pm
LOL!
Jon Murphy
Feb 18 2021 at 8:33am
“Despite extensive clinical experience and decades of research on influenza, it is still not fully understood how influenza is transmitted among humans” (Smieszek, Lazzari, and Salathe. Nature 9, no. 2185).
Charley Hooper
Feb 19 2021 at 3:06pm
AMT,
Scientists recognize that they don’t fully understand how influenza is spread but you somehow think that they do understand exactly how SARS-CoV-2 is spread, even though influenza has been studied for 100 years and SARS-CoV-2 has been studied for one year?
How likely is that?
AMT
Feb 19 2021 at 8:30pm
Quote for me where I said that. I never said scientists fully understand how either illness is spread. I simply criticized that the hedge word, “fully,” was grossly misinterpreted by you (and others). Which is now very unsurprising, based on how you have also misinterpreted me. Not to mention you couldn’t figure out that a study on the effects of recommending people wear masks doesn’t actually study the effect of people actually wearing masks. Let’s also pretend we have done a study on the effects of diet and exercise, based doctor recommendations that their overweight patients improve their diets and exercise more.
If, for example, 99.999999% of transmission is spread through bodily fluids, and we are unaware of how that remaining proportion is being transmitted, we don’t “fully” understand the transmission. But then you take the word “fully” and erroneously claim “Scientist [sic] don’t have a handle on how the flu transmits throughout the community.” This is purely disingenuous. That’s why I gave an analogy.
It is simple common sense that if you know of avenues of transmission, even if not 100% of them, you can implement ways to reduce the transmission. If there is any “absolutism [that] is unjustified and harmful,” it is your demand for full knowledge of all avenues of transmission before accepting even the most obvious methods of reducing known transmission should be utilized.
suddyan
Feb 18 2021 at 7:13am
[This doesn’t mean washing your hands and wearing a mask is useless.]
It also doesn’t mean washing your hands and wearing a mask is eminently useful.
Justin
Feb 17 2021 at 7:45pm
I don’t find Hooper’s statements persuasive at all. Actually, they are highly misleading. For example, his discussion of the Denmark RCT seems to imply that what is being measured is the treatment effect of wearing masks. In fact, the treatment in the study he is referring to (Bundgaard et al)y was merely the distribution of masks and a recommendation to wear them. They make some efforts to tease out the effect the actual mask, but that’s extremely difficult in practice for obvious reasons given that they can’t actually observe whether and to what extent the treatment group followed through with the recommendation. Even if masks did have a positive effect, I wouldn’t expect this study to pick it up.
He dismisses much the literature by referring to the studies as “correlational, population-based studies that identify relationships but don’t necessarily prove cause and effect.” But the studies go way beyond correlation. Of ones that I’m aware of, Chernozhukov et al and Karaivanov et al, make a serious effort to estimate causal effects from observational data and do find substantial effects. Sure, you can dismiss them as observational. But there is no dispute about whether it is possible to draw causal effects from observational data under certain assumptions – rather the questions always surround whether those assumptions are met in practice. If he’d like to explain why the strategies these papers use to estimate the causal effects are flawed, he’s welcome to do so and perhaps he’d be correct. But dismissing them as merely correlational is not giving them an honest reading. At the end of the day, these are the best stud
Jon Murphy
Feb 17 2021 at 9:56pm
That’s precisely Charley’s point. At best, the effect size seems to be so small as to be unmeasurable.
Justin Max
Feb 17 2021 at 10:42pm
That may be Charley’s point but it’s not mine. My point is that even if the effect size of wearing masks were large, the study probably wouldn’t capture it because the study was not designed to estimate that particular effect. The fact that a study finding Y has a small effect size tells us very little about the effect size of X.
mm
Feb 17 2021 at 8:15pm
Why use influenza data when we have data for other coronaviruses-and that data clearly supports the use of masks:
“Eight of nine retrospective observational studies found that mask and ⁄ or respirator use was independently associated with a reduced risk of severe acute respiratory syndrome (SARS). ”
https://onlinelibrary.wiley.com/doi/epdf/10.1111/j.1750-2659.2011.00307.x
Chris
Feb 17 2021 at 9:41pm
Why indeed. It’s almost as if Charley has a conclusion and is in search of premises:
https://www.econlib.org/my-bet-on-covid-19-and-why-i-might-lose/
Jon Murphy
Feb 17 2021 at 10:06pm
Because that’s the method used by the WHO, CDC, and other organizations:
“COVID-19 and influenza viruses have a similar disease presentation“
Mm
Feb 18 2021 at 9:06am
At this point relying on data from influenza when we have data on coronaviruses is akin to conflating stroke and heart attack data- they are both vascular diseases- but there is no need to do so. We have data from SARS and COVID now so we should use it- just last week:
https://jamanetwork.com/journals/jama/fullarticle/2776536?utm_source=silverchair&utm_campaign=jama_network&utm_content=covid_weekly_highlights&utm_medium=email
Another recent review
https://www.pnas.org/content/118/4/e2014564118
Most of the data is for coronavirus in the above review they only noted one controlled study of mask use in the community for influenza and found “protective efficacy in excess of 80% against clinical influenza-like illness.”
almost every review of the data on masks finds some protective effect- the data is not conclusive but a) it is the data we have and b) we have little else to reduce spread of the virus while we get vaccinated (or the other choices are more disruptive and more expensive). Statements like “ Eight of nine retrospective observational studies found mask &/or respirator use was INDEPENDENTLY associated with a reduced risk of severe SARS.” are hard to ignore given the risks we face and the data we have.
Jon Murphy
Feb 18 2021 at 10:49am
Well, part of the problem is we don’t actually have data. Most of the numbers being generated are quite meaningless. Flu still is the better comparison because the issues plaguing (pun intended) COVID and SARS have been somewhat dealt with. These issues include (but are not limited to):
-Small sample sizes
-Biased reporting
-Inconsistent reporting (meaning we often cannot add daily figures).
J.P. Ioannidis and co-authors have a forthcoming paper where they document a lot of these issues.
Mm
Feb 18 2021 at 3:34pm
Many of the same problems with the flu data as well
Jon Murphy
Feb 18 2021 at 3:37pm
There certainly are many similar problems, but the n is much, much larger. That helps. And reporting is fairly standardized. Flu numbers from season to season are roughly comparable. COVID-19 numbers are not comparable day to day, never mind county to county, state to state, or country to country.
Alan Goldhammer
Feb 17 2021 at 9:40pm
As one who read all of the copious literature, bot pre-prints and published articles, Mr. Hooper’s comments are misleading at best. There have been a number of studies on mask wearing both in the laboratory and in real life. I documented these in the newsletter that I published for eight months and those interested can browse through all of the work I did. The best real world study was done in Jena Germany where they implemented a mask regulation. Case reports of COVID-19 dropped to practically zero while a neighboring town of equal size that waited several weeks to put in place a mask mandate saw a continuation of COVID-19 cases. There are also some local school studies here in the US that yield similar studies.
One can be a libertarian and argue against mandates; one cannot be a scientist and argue that they do not work.
Jon Murphy
Feb 17 2021 at 9:48pm
I didn’t realize the CDC and WHO were populated by non-scientists.
Alan Goldhammer
Feb 18 2021 at 8:44am
What a flippant comment here and elsewhere on this thread. If you can’t take time to read all the literature that goes against Mr. Hooper’s comment, why bother posting multiple replies? I spend considerable time reading everything that was available through to the point when the Pfizer vaccine was authorized. There are numerous papers from aerosol researchers along with a bundle of real world observational data pointing to the value of masks. This information is solid and if you don’t like it, that is your privilege.
I’ll leave it at that.
Jon Murphy
Feb 18 2021 at 10:25am
Flippant is as flippant does. If you didn’t want a flippant response, why make a flippant comment?
The point is simple: the WHO and the CDC, even as recently as the early days of the pandemic, advised against mask mandates. If one cannot be a scientist and oppose mask mandates, then the WHO and CDC are not scientists, QED.
zeke5123
Feb 18 2021 at 12:29pm
The comment actually is right on the point. In a very short period of time (i.e., not enough time to actually do any serious study), the WHO and CDC settled on using masks (i.e., about a month).
At that time, masks became political. Given (i) the data supporting mask use is weak because the studies are weak, (ii) replication crisis and all the ways people can “hack” results, (iii) the political aspect of it (i.e., very quickly there was one acceptable view and contrary views would not be tolerated — hell see the fake study published about another political topic in hydroxychloroquine), I don’t think it is realistic to believe the papers on this topic represent disinterested scientific opinion. But of course that’s what happens when (i) journals don’t require pre-research methods, (ii) and we don’t call people “deniers” for going against the political consensus.
Jon Murphy
Feb 17 2021 at 10:26pm
In a more general comment on the philosophy of this absolutist statement, is not the attitude that “a scientist does not question” distinctly anti-science? I mean, let’s reframe the statement in a context from about two centuries ago:
“Dr. John Snow, you can claim to be a liberal* and be opposed to miasma-related regulations. You cannot be a scientist and argue they do not work.”
And yet, Dr. Snow did just that. And discovered germ theory (not to mention radically revolutionized public health and statistics as we know it). And he did it by arguing against the status quo (miasma theory). A status quo, I hasten to add, which had more evidence backing it than mask mandates.
Now, I may just be a simple economist who knows a little statistics. Mayhap I simply do not understand science. But it seems odd to me that conclusions cannot be questioned. Indeed, I was under the impression that the entire point of the journal system, conferences, and academia in general was that people could share their conclusions and debate them publicly.
*I substituted “liberal” here because the word “libertarian” did not exist in the 1850s and “liberal” is more accurate anyway.
Jon Murphy
Feb 17 2021 at 9:54pm
Charley cites are just the tip of the iceberg. A 2019 Nature article declarers: “Despite extensive clinical experience and decades of research on influenza, it is still not fully understood how influenza is transmitted among humans” (Smieszek, Lazzari, and Salathe. Nature 9, no. 2185). There are countless more studies.
At best, the benefits to mask-wearing are swamped by other issues. That’s why they’re not picked up in studies.
As an aside, I’ve spent much of the past year digging into both COVID-19 research and broader public health research. So many of the studies are poorly done it’s not even funny.
HH
Feb 18 2021 at 1:10am
Perhaps it would be worth noting that the various measures – hand washing, masks, distancing, etc – have brought the flu to nearly zero last year while Covid-19 continues to spread, so maybe treating then as interchangeable isn’t very informative.
Jon Murphy
Feb 18 2021 at 6:50am
Assumes facts not in evidence, HH. Some folks speculate that the actions reduced flu figures. But there’s no evidence to support that. Just conjecture.
Indeed, it’s far more likely that many flu cases were classified as COVID-19 as per CDC regulations
HH
Feb 18 2021 at 9:48am
There are other countries, Jon.
Jon Murphy
Feb 18 2021 at 10:25am
Indeed. And as the Nature study I cite above and below show, as well as the WHO studies cited, the non-results hold across countries.
HH
Feb 19 2021 at 2:03am
And the flu is down almost everywhere. Are you seriously saying that Australia has a 98% decline in flu deaths because the US CDC is classifying deaths differently? (No.)
More generally, every country has had a huge decline in flu cases, all at the same time, in the year they also took Covid precautions. I don’t need a randomized controlled trial to see the connection. Feel free to call it conjecture.
BC
Feb 18 2021 at 6:00am
Instead of discussing government mask guidance and mandates, suppose we were discussing whether for-profit companies should be allowed to advertise their masks as “scientifically proven to reduce the spread of Covid-19” or “medically effective against Covid-19 and other viruses”. I notice that many masks, even 3-layer surgical masks and 4-layer KF94 masks (Korean comparable to N95), are labeled as protecting against household dust or for “non-medical use”, I assume for regulatory or legal liability reasons. I get the sense that many of the same people that argue for mask mandates, or at least strong norms in favor of masks, would object to marketing and advertising of those very same masks as medical devices that have been proven safe and effective, especially say homemade cloth masks sold on Etsy.
Of course, government mandates are different from advertising. Even if a mask is advertised as “scientifically proven effective”, it doesn’t mean that someone has to buy it. Government mandates are non-voluntary. But, that would seem to argue in favor of a tougher standard for government mandates compared to the standard for being allowed to market and sell a medical device. Yet, I get the feeling that the hurdle for approving vaccines and rapid tests has been much higher than for issuing guidance and mandates on masks.
Why would the answer to the question, “Does the scientific evidence demonstrate that masks are effective against Covid-19?” depend on whether the context is government mandate or approval for marketing as a medical device?
Charley Hooper
Feb 18 2021 at 1:32pm
Good point.
The FDA would never allow for-profit companies to advertise face masks for the prevention of COVID-19 infections, for the following reasons: (1) the required studies don’t exist today, (2) the necessary studies would be difficult to conduct in a manner that would avoid confounding factors, and (3) I suspect the results would be too lackluster to “prove” efficacy. And I should like to add (4), which is that the FDA’s processes typically take so long that the pandemic would likely be over.
Andre
Feb 18 2021 at 6:44am
First, an observation: were this post on one of the major social media sites, it would most likely have a warning banner suggesting it contravenes “consensus science,” with links redirecting you to one of the approved sources of information (CDC, WHO). In some circumstances, the entire page or channel might be given a strike and taken down temporarily or permanently. We now have truth gatekeepers.
Second, isn’t the underlying point that masks are not a reliable way of preventing infection spread? It’s not that they don’t help limit spread, it’s just that they help primarily with droplets, but SARS-CoV-2 is clearly aerosol, and masks won’t do diddly for that. Being indoors with a mask around anyone infected (or who had been in the room, and was infected, per the CDC) puts you at risk of transmission.
People are under the mistaken impression that masks help prevent transmission, but compared to not going somewhere in the first place, they are terrible. Now, I understand that some people have no choice but to go out – work requires it, or they may have no one to rely on. And many have adjusted to this.
However, a huge chunk of the population is under the impression that masks are some kind of magical protection that prevent spread – they don’t. They primarily limit droplet spread. Not only will they not do much against indoor spread, since SARS-CoV-2 is aerosol, they end up putting people directly in harm’s way. I daresay a majority of spread happens during human movement that could have been avoided entirely in the first place. I’d wager the vast majority.
If the public were told that the number one rule should be don’t go to indoor places where strangers are if you can help it at all, ever, this would have spread much more slowly. But they are being told primarily to “mask up to save lives.” However, if using a mask puts people in proximity with each other, it does not save lives, compared to the often available and perfect option of not going out in the first place.
One last thing: a non-trivial portion of the population have piggish hygiene, either for lack of knowing or caring. Add to these the folks whose priorities conflict with the safety of others – they want to work, need to work, don’t care about others, think the mask will prevent others from catching whatever they have, etc. In these cases all, masks provide a ready excuse for the person, and make it less obvious to others that they are sick/symptomatic. All of this contributes to spread as well, compared to the alternative of not having a mask norm, which might keep them home or subject them to the wrath of others.
To me, it’s obvious that pushing masks creates several incentives for the disease to spread. That doesn’t mean that masks are useless or counterproductive, just that their benefits and costs are situational, and pushing masks as an incontrovertible good is bad public health policy.
Also, what do I know, I’m just a dude in an chair who has been following this with reasonable attention to evidence and argument.
Thomas Hutcheson
Feb 18 2021 at 7:55am
I think it is proper for a pubic health official to use vey high-noise, evidence to recommend a low cost NPI. It there were some way for commercial firms to make a similarly nuanced, tentative claim that would be OK, too.
I agree that the mask recommendation has been misunderstood as self-protecting rather than other-protecting (how much of a Flynn Effect there is, is more doubtful), but that is common across other recommendations.
In some the damage is less because the benefits are symmetric: distancing protects the distancer as much as it protects the distancee (although the failure to take into consideration the benefit to others leads, given ones estimate of risk, to too little distancing). Vaccinations, however, are another intervention that have not been adequately messaged as being for the protection of the non-vaccinated as for the person being vaccinated.
Nick
Feb 18 2021 at 7:01pm
Not sure about the “clearly aerosol” bit.
I haven’t read the literature, am just asking why SARS-CoV-2 can’t be spread both by aerosols and droplets?
Also don’t understand the commenters who (unlike the original poster, who creates a proper context and makes proper observations) leap from “it’s not proven” to “it isn’t the case.” Also, does anyone remember AIDS? It took a looong time to figure that one out, right?
Andre
Feb 18 2021 at 9:38pm
I haven’t read the literature, am just asking why SARS-CoV-2 can’t be spread both by aerosols and droplets?
It can. Didn’t mean to imply otherwise. Read it as, “Masks help stop the droplet spread, but since SARS-CoV-2 is aerosol also, masks are inadequate to prevent exposure in indoor exposure settings.”
Thomas Hutcheson
Feb 18 2021 at 7:28am
It was exactly the lack of anything like RCT on how much wearing a mask protected people around the mask wearer that lay behind the initial recommendation NOT to wear (then scarce) surgical or N95 masks. How could one set up a trial in which one group was around masked people and another around statistically similar maskless people?
Instead, public health officials did with mask wearing what they did not do with so many other recommendations, go with the view that putting some cloth around the nose and mouth of a potential transmitter of and airborne virus was such a low cost intervention that it passed a “common sense” cost benefit test even without being able to reject the null hypothesis that just as many virus articles reached the airways of potential victims when the transmitter did not wear a mask as when they did.
As for whether wearing a mask might be recommendable to protect others from transmission of flu from an asymptomatic person with flu, there are factors that would reduce the net benefits:
1.) There is (probably) a longer period when a person with COVID can transmit the disease without themselves being asymptomatic than with flu.
2) With vaccines available the probabilities of someone being an asymptomatic transmitter of flu are lower than being a transmitter of COVID
3) COVID has more serious health risks per transmission than flu.
Frank Berger
Feb 18 2021 at 9:32am
I haven’t read all the comments so sorry if this has been said. I suspect that masks and hand washing are to covid spread as condoms and other birth control measures are to pregnancy. If you uses the tools correctly they are highly effective. The problem is that in many or maybe most cases they aren’t. How often do you see a person with a mask on but his nose is exposed? Does he even think he is protected at all? You get in the card, remove your mask and scratch you nose. Remember the spokesperson early on on TV talking about not touching your face and she did exactly that multiple times during her short talk?
The point is that a given person is careful about mask-wearing, hand-washing, not face-touching and social distancing, I think they have a very high degree of safety. The problem is that most of us can’t do that, so at the population level these techniques don’t seem to work very well.
Jon Murphy
Feb 18 2021 at 11:16am
Just about all the negative commenters here are missing Charley’s point. Charley writes:
He is absolutely right here. The evidence is very murkey and the absolutist statements made here (eg “One can be a libertarian and argue against mandates; one cannot be a scientist and argue that they do not work”) are simply not justified by the data.
Additionally, I have spent much of the past year digging deeply into COVID research. From a statistical POV, much of it is awful. I was a forecaster for years, and the major models (Imperial College, IHME, etc) don’t even follow best-practices or rudimentary statistical methods. For example, IHME’s confidence intervals bizarrely shrunk the further out the forecast was. There’s no reason to think the model is more accurate forecasting two weeks out than 24-hours out. And the CI’s were hilariously huge as well. For example, a 24-hour-out prediction of hospital beds by the IHME model had a range of 13,000-192,000. One day out. Tell me: even if we ignore all the other issues with the model, how is anyone supposed to speak with any confidence with a range that huge?
The advocates who are speaking with such absolute confidence in their recommendations ought to be ashamed of themselves.
Charley Hooper
Feb 18 2021 at 2:03pm
Jon,
That’s spot on. I, too, noticed the problem with the confidence intervals coming out of the IHME models. The fact that no one involved with the those models noticed the problem is shocking.
We have historical data and we have future projections from models. Many treat the two as if they are equally valid. But they aren’t! Models have so many ways to be wrong and I’ve seen so many bad models during my decades in the pharmaceutical industry. Bad models are more common than good models. One of the worst aspects of modeling is that we often won’t know until years or decades have passed whether the predictions are correct or not.
In 2020 we suffered through bad COVID and election predictions. In earlier decades, to give but one example, we were misled by the Club of Rome models. We were assured that they were completely scientific!
How credulous should we be to the predictions that emerge from the models of really complex phenonemon, such as the climate?
Jon Murphy
Feb 18 2021 at 2:47pm
I think there is an explanation for that. When one is building a forecast, once can sometimes get swept up in it and forget the purpose of the model. Thus, to an outside observer, the range is odd but to the forecaster, it seems normal.
For example, when I was a young forecaster (22, just on the job), we gave a presentation to a company on a forecast of their sales data. I’ll never forget the conversation because it was formative. (For confidentiality reasons, the numbers change been changed, but the idea is the same).
We had just finished the presentation. The VP I was working with asked me a simple question: “What am I supposed to do with this information?”
I said: “What do you mean?”
He said: “You’re telling me that a year out, my sales could be flat or they could be up 6%. How am I supposed to plan for that? That’s a difference of hundreds of millions of dollars.”
My boss jumped in to explain the forecasting method.
The VP responded: “You’re forecasts are philosophically interesting. But tactically useless.”
I had never been so embarrassed in my life. But the lesson stuck with me: what good is a model if it provides no useful insight? As a forecaster, I had forgotten that. My models were very pretty. But they provided no useful insight. As an expert, I had failed.
So, my point is, I think the IHME, Imperial College, etc., folks lost the goal of their models, which was to try to provide actionable information. They claimed it was actionable, but it wasn’t.
Charley Hooper
Feb 19 2021 at 3:41pm
I would say that your forecast was useful. If the range was flat to 6% growth, then plan on 3% growth and realize that sales could be higher (up to 6% growth) or lower (down to flat). There’s always uncertainty and your forecast conveyed the level of uncertainty.
AMT
Feb 18 2021 at 2:53pm
If you say “it’s pretty hard to effectively prevent the transmission” you ARE saying masks don’t work to prevent the transmission. Let’s not blame others because Hooper is making contradictory statements.
Jon Murphy
Feb 18 2021 at 3:34pm
Again, that’s a pretty strawmanned version of the quote in question. Think on the margin, bro.
AMT
Feb 18 2021 at 11:14pm
I’m not sure why you think anyone would believe you when you immediately resort to saying I have strawmanned the argument. I mean, I literally quoted him, and there is nothing marginal about that statement. Try making an actual, relevant criticism instead of jumping to an obviously false counter.
Jon Murphy
Feb 19 2021 at 1:48pm
Because you have. You keep turning probabilistic statements (like the one you quote) into absolutist statements. That is an example of strawmanning.
Thomas Hutcheson
Feb 19 2021 at 6:51am
If masks are mildly effective (in preventing the wearer from becoming infected + >”mildly” in preventing infecting others) and tying a piece of cloth around your mouth and nose is almost zero cost, that gives a pretty high expected value for the NPV of wearing a mask with a pretty small tail in the negative portion of the distribution. If calling a mask recommendation in those circumstances “absolutist,” maybe it’s OK to be “absolutist.”
Charley Hooper
Feb 19 2021 at 3:48pm
But it’s not zero cost. It’s harder to breathe–which is an issue if you are doing any physical work–glasses fog up, people get acne, it’s harder to communicate, masks sometime need to be repositioned, masks leave lines on faces, and my mask occasionally makes me sneeze.
Michael
Feb 18 2021 at 1:39pm
I think it is a mistake to rely too heavily on influenza data to assess the value of masks for SARS-CoV2 for the simple reason that influenza is far less infectious than SARS-CoV2.
If we were speculating about the effectiveness of masks in influenza based on robust data that they are ineffective in SARS-CoV2, I would find that a lot more compelling an argument.
HH
Feb 19 2021 at 2:09am
Very much agree with this. The anti-Covid measures we’ve taken the past year – including distancing & masks – have reduced flu cases by an order of magnitude. Covid, as we all know, is still spreading with comparative ease. Not sure we can precisely assign credit to which measures had the biggest impact on reducing the spread of flu, but it seems obvious that flu is not a useful proxy for Covid, given that one of them has very strongly responded to our countermeasures and one much less.
Vivian Darkbloom
Feb 19 2021 at 8:21am
“The anti-Covid measures we’ve taken the past year – including distancing & masks – have reduced flu cases by an order of magnitude.”
I don’t think this is necessarily the conclusion to draw. From what I have read, while possible, it is extremely rare for two respiratory viruses to infect a person at the same time. Late last summer, over at MR, I raised the issue as to what effect Covid would have on the flu season. Is it possible, I conjectured, that the Covid virus would “trump” influenza thus reducing the number of flu cases? It appears to me that this is a more likely explanation than that masks and other measures have reduced the incidence of influenza during the 2020/2021 season.
In fact, I draw three tentative conclusions for the extremely low number of flu cases reported this year:
Covid 19 does indeed “trump” influenza and if one is exposed to both, Covid takes over;
Some influenza infections/deaths are being improperly registered as Covid infections/deaths;
While I am not suggesting there is no effect, the low incidence of influenza is not necessarily a result of mask-wearing and social distancing. If masks and social distancing were that effective against influenza, one would expect an similarly, even if not equally, robust protection against Covid. If masks and social distancing *is* the (sole) explanation for the low incidence of flu, this actually cuts against the argument that those same measures are effective against Covid. “Contagiousness”, without more, is not for me a very useful argument. Either those masks prevent transmission or they don’t.
Regarding the last point: When faced with a claim that some action has an alleged effect, I think it is useful to ask what the mechanism is that creates the connection between alleged cause and effect (for example, fiscal or monetary policy, such as QE). With respect to masks, it is allegedly that masks filter out droplets and perhaps (but not as likely) aerosols. Most experts seem to agree that if there is a positive effect, the mechanism is that masks prevent the infected person from exhaling droplets or aerosols from reaching a non-infected person. It is not clear to me what happens to those exhaled viruses. Are they trapped in the mask where they die? Do they drop to the ground where they die?
I’ve read articles suggesting that mask-wearing is not harmful to the wearer in the sense that one is not inhaling excessive amounts of CO2. However, there’s one issue that I’ve not seen discussed, to wit, is mask-wearing harmful to someone who is already infected? If those masks are actually serving as a barrier to those droplets, aerosols and virus RNA’s, are they not serving to have the infected person constantly re-inhaling them? I can’t imagine that this inevitable increase in viral load is very healthy for the wearer. Also, I’ll take judicial notice of the fact that those virally infected masks are constantly being taken off and handled by the person wearing them. They then touch other objects. Could it be those masks are actually causing more transmission issues and serious illnesses from all this, particularly when worn in situations where transmission is less of an issue, such as outdoors? Let’s also keep in mind that in the real world we are not talking about technically superior masks worn perfectly in laboratory conditions.
Bottom line: My jury is still out on the issue of masks.
HH
Feb 19 2021 at 3:27pm
” If masks and social distancing were that effective against influenza, one would expect an similarly, even if not equally, robust protection against Covid.”
ONLY if you already assume the transmission mechanisms are similar. But that’s the question, isn’t it? The obvious alternative to my hypothesis (Covid prevention measures work much much better on flu than on Covid) is yours (Covid dominates flu). Superficially the numbers make me think mine is more likely (I’d bet on it, Caplan style) but at least these are testable hypotheses. But declaring the flu as the same as Covid for modeling purposes is way premature. And that’s what and to be happening here.
Vivian Darkbloom
Feb 20 2021 at 3:39am
“ONLY if you already assume the transmission mechanisms are similar. But that’s the question, isn’t it?”
Well, why isn’t this a valid assumption regarding masks? You’ve failed to even posit a hypothesis, much less a supportable one, that the transmission mechanism for Covid is different from influenza. But, you didn’t hesitate to make an assumption that it was because of masks (and social distancing) that we’ve had record low influenza rates this year. You’ve implicitly made an assumption regarding the transmission mechanism!
As I indicated above, the protection that masks allegedly afford is that they prevent the transmission of droplets and aerosols that carry either type of virus. If you don’t think that this is the transmission mechanism, what would be the use of masks to protect against either?
HH
Feb 20 2021 at 8:14am
My stated hypothesis is only that the transmission mechanisms between the flu and covid vary enough that one isn’t a useful proxy for the other.
Vivian Darkbloom
Feb 20 2021 at 9:33am
Excuse me, HH, but I think you are avoiding the issue. Once again, the mechanism that masks are allegedly preventing the transmission of viruses are airborne droplets and aerosols. Can you come up with any other hypothetical mechanism that masks would conceivably prevent the transmission of a virus?
Scott Sumner
Feb 18 2021 at 1:43pm
When people cite the Danish study as evidence that masks don’t work I tend to discount everything else they say. The Danish study did not even test the effectiveness of masks in preventing the spread of Covid.
Jon Murphy
Feb 18 2021 at 1:47pm
As Charley, I, and others (see The Price of Panic by Axe, Briggs, and Richards) discuss, there are other studies (including by the CDC and WHO) that read the same conclusion of the murkiness of masks and mask mandates.
Daniel Kling
Feb 18 2021 at 2:02pm
Bayes’ theorem to the rescue!
Charley Hooper
Feb 18 2021 at 2:14pm
Scott Sumner,
What did I say that was misleading or wrong about the Danish study?
I mentioned that it was the only randomized controlled trial to test the effectiveness of masks for the prevention of COVID-19. And then I summarized the results.
I didn’t want to exclude it lest people would say that I missed a data point.
der Wienermeister
Feb 18 2021 at 5:51pm
The Danish study was a study of a mask recommendation, not directly a study of masks.
AMT
Feb 18 2021 at 11:10pm
If you can’t understand that a study on the effectiveness of recommending mask usage is not a study on the effectiveness of actually wearing masks…
Scott Sumner
Feb 20 2021 at 11:44am
The Danish study did not examine the question of whether masks prevented the transmission of Covid-19. I’d encourage you to look at the many critiques of the study.
David Henderson
Feb 18 2021 at 11:27pm
Hmmm. Interesting methodology, Scott.
Dylan
Feb 19 2021 at 8:26am
I’m going to come at this from the other direction. I’ve been reading here long enough to have read Charley multiple times on subjects I know at least a little about. During that time, I’ve found enough cases where an argument rests (at least partially) on information that is devoid of context and used in misleading ways. That has caused me to approach his writing even more skeptically than I normally do, even when it is in an area (like this one) where I think we’re in rough agreement. Referencing the Danish study, without mentioning the obvious drawbacks (study was designed to look at protectiveness of masks for the wearer. and not the impact of transmission in the community as a whole) is consistent with this pattern.
Art K
Feb 18 2021 at 2:42pm
Refutation of the Danish Mask Study:
https://fooledbyrandomnessdotcom.wordpress.com/2020/11/25/hypothesis-testing-in-the-presence-of-false-positives-the-flaws-in-the-danish-mask-study/
Alan Goldhammer
Feb 19 2021 at 8:39am
Here is the link to the Jena mask (the paper was expanded since the pre-print appeared to cover some other German regions) study that I mentioned in my initial response. Scott Sumner’s point on the Danish study is also correct. The confirmation bias of may respondents above is quite striking and not unlike the anti-mask debates that took place broadly across the country. There have been a number of aerosol scientists such as Lindsey Marr at Virginia Tech who have also written about the utility of masks.
Masks are coupled with physical distancing can protect most individuals against COVID-19. Addressing Vivian Darkbloom’s point about influenza levels, they are down this flu season based on national surveillance. However, because of COVID-19 the statistics must be interpreted with caution. You can always get the weekly reports from the CDC if interested in tracking this.
Jon Murphy
Feb 19 2021 at 1:53pm
This is unintentionally ironic given it’s coming from the guy who claimed absolutely “You cannot be a scientist and oppose [mask mandates].”
No one has denied that. Where the question has been all along is the marginal effects. Do masks reduce spread sufficiently to justify mandates? The WHO and CDC did not think so. The recent evidence of COVID-19 do not indicate so. The evidence that masks reduce transmission significantly is very, very murky. To pretend otherwise is…well, confirmation bias.
Comments are closed.