[T]he hard but just rule [of science] is that if the ideas didn’t work, you must throw them away.
— Carl Sagan1
Astrophysicist Carl Sagan puts into plain language the principle of falsification propounded by the philosopher of science Karl Popper. Rigorous science demands that a theory must be capable of being tested and disproven; when disproven, the theory must be discarded.2 Popper posited a modus tollens (“If P, then Q. Not Q. Therefore, not P.” This is also known as denying the consequent). If a hypothesis can have the modus tollens logic applied, then it is a science, if not it is a pseudoscience.
In evaluating scientific knowledge within health care, the gold standard is the randomized control trial (RCT).3
Physicist Denis Rancourt has examined the scientific literature on what RCTs reveal about the prophylaxis of masks and respirators against respiratory viruses. His conclusion is that masks don’t work.4 While Rancourt relies on the gold standard of scientific results, those results fly in the face of what the broad swath of mass media and the Establishment proclaim.5
This topical dissension sparked a recent debate between Denis Rancourt and philosopher David Kyle Johnson on the efficacy of wearing face masks to protect against transmission of respiratory viruses. The debate over mask wearing has taken greater significance now that some jurisdictions worldwide are punishing non-compliance with mask wearing impositions.
Johnson began the debate by downplaying Rancourt’s credentials pointing out that Rancourt is “a physicist who specialized in metals and no longer works in academia.”
This smacks of academic snobbery. Johnson’s logic seemingly posits that only people working in a specific profession can possess the expertise to speak on certain topics. If so, then one wonders why have a debate at all given that the ability of non-experts to discern factual accuracy and logic is in question. To Rancourt’s credit, he says of people: “I trust their ability to judge for themselves whether they be scientists, laypersons, and so on…”
Johnson is not a professional scientist, but he sought to accredit his expertise to participate in the debate because one area of his study includes pseudoscience, which he accuses Rancourt of promoting. In fact, Johnson claims greater expertise than Rancourt:
Although I am not an epidemiologist either, I am a logician who teaches entire courses on argumentation and medical pseudoscience. So, unlike with Rancourt, my writing of this article falls squarely within my area of expertise.6
While promoting one’s bona fides gives insight to the level of knowledge expected of a source, it is the factual accuracy and the logic of an argument that is important and not the source of the argument. To think otherwise is the logical fallacy of argumentum ex cathedra. During his summation, Johnson will correct this faux pas, although it appears he did so without realizing that he had committed such an error.
The non-scientist begins his ad hominem by denigrating the expertise of a man who was a tenured full-professor of physics — which is by most anyone’s understanding, a hard science.
Johnson focuses, not on whether masks work, but on the mechanism of how masks work, and he adamantly states that masks do work. He says evidence supports the filtration effectiveness of masks. The problem with the studies that Johnson cites are that they are correlational, non-control studies. Correlation does not lead itself to ascribing causation because not all variables are controlled. For example, if one were testing the effectiveness of mask wearing at reducing contraction of COVID-19 while other preventative measures were simultaneously being carried out, then how should one ascertain with certainty what was the cause? If mask wearing is accompanied by hand washing, social distancing, fomite disinfection, etc and a statistically significant reduction in becoming infected by a viral pathogen was found, then a question arises: was it solely due to the wearing of face masks? Or did the face masks have a nugatory effect and the significant finding was due to one or more of the other variables? Rancourt points to this in his paper: “no study exists that shows a benefit from a broad policy to wear masks in public. There is good reason for this. It would be impossible to obtain unambiguous and bias-free results.”4
Johnson accuses Rancourt of a confirmation bias through cherry-picking studies. Yet Rancourt examined all 14 RCTs7 that survive the quality barrier to be included in most systematic reviews (7 RCTs in the initial paper4 ). RCTs remove experimenter bias and allow for causal attribution. When RCTs have been carried out, what is the logic of preferring results from inferior science methodologies? Nonetheless, Johnson gives short shrift to the primacy of RCTs and relies on less rigorous observational and comparative studies to affirm the efficacy of mask wearing.
In choosing Rancourt’s number one egregious mistake, Johnson says Rancourt “quote-mined” bin-Reza et al. (2012). The quotation in question is:
There were 17 eligible studies. … None of the studies established a conclusive relationship between mask/respirator use and protection against influenza infection.8
In his blog, Johnson refines his argument against the bin-Reza quotation from the debate. It has importance because he told Rancourt,
Given that it appears in the section of his paper where he is arguing that masks don’t work, his quoting of this line from the study implies that the authors intended this statement to mean there is no benefit to wearing masks. In reality, however, the slash in the “mask/respirator” phrase is meant to indicate a comparison between the two types of facial coverings. The study is not lumping them together and declaring them both ineffective; the study actually concludes that masks and respirators are equally effective. Several of the sentences before and after the one he quotes demonstrate this. For example,
“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).”6
First, what Johnson means by “it appears in the section of his paper where he is arguing that masks don’t work” is puzzling because Rancourt’s entire paper from the title to the conclusion argues that masks don’t work.
Second, it seems Johnson has a non-mainstream grammar take on the slash/virgue. A virgule is used “between two words indicating that whichever is appropriate may be chosen to complete the sense of the text in which they occur.” Applying the typical grammar usage of a virgule to “None of the studies established a conclusive relationship between mask/respirator use and protection against influenza infection,” two renderings are possible: (1) None of the studies established a conclusive relationship between mask use and protection against influenza infection. (2) None of the studies established a conclusive relationship between respirator use and protection against influenza infection. After reading the bin-Reza et al. paper left to right, right to left, upside down, downside up, I fail to see how this is the big mistake of Rancourt. I invite readers to do so for themselves.8
Third, Johnson’s cited example is a red herring. Why inject less rigorous observational studies into the debate? Rancourt confined himself to the gold standard: RCTs. Bin-Reza et al. did not find statistical significance for a RCT investigating the effectiveness of wearing face masks; in fact, the results for all of the studies were non-significant.
Johnson does catch a mistake in Rancourt’s article wherein he stated that Jacobs et al. (2009) studied N95 wearing.9 Although Jacobs et al. did not find any statistical significant finding in support of wearing face masks, the sample size was too small to reach a definitive conclusion.
Johnson appealed to Cowling et al.’s (2010) summary wherein it was stated:
There is some evidence to support the wearing of masks or respirators during illness to protect others, and public health emphasis on mask wearing during illness may help to reduce influenza virus transmission.10
However, this summary belies what Rancourt pointed to in Cowling et al.:
None of the studies reviewed showed a benefit from wearing a mask, in either HCW or community members in households (H). See summary Tables 1 and 2 therein.11
Interested readers can go to the Cowling et al. article and see the results. What is important is the results of a study. The results speak for themselves. The numbers don’t lie or twist perceptions as is possible with words in a discussion or conclusion. This is what Rancourt pointed out during the debate.
Johnson’s twisting of language to rebut Rancourt’s article is evident:
The article is now widely cited by the “anti-mask” movement as proof that masks don’t work and thus laws requiring citizens to wear masks are ineffectual. But, to put it mildly, Rancourt’s argument is fraught with pseudoscience and logical mistakes, and it fails entirely to even provide evidence for (much less proof of) his thesis.6
The use of the wording “anti-mask” is pejorative and tendentious. Rancourt could characterize Johnson’s argument of compulsory mask wearing as anti-freedom. Laypersons may mistakenly refer to an experiment as proving something, but scientists should never make that mistake. Science disproves the alternative hypothesis (the principle of falsification); science, however, proves nothing. Johnson is not a scientist, but given that he claims expertise in the philosophy of science, he should not make such a mistake.
Characterizing Rancourt’s literature review, as any good researcher would do when doing a systemic review, as having “scoured the literature” is further evidence of injecting animus into the debate.
Johnson states that Rancourt’s position is that “COVID spreads solely via aerosols.” It is a mischaracterization, as Rancourt does not state this.
Although inconsequential to Rancourt’s systemic review of the RCT mask literature on protection against respiratory viral contraction, Johnson next takes aim at Rancourt’s positing a connection between humidity and transmission of SARS-CoV-2. Johnson refutes such a hypothesis by pointing to the spread of COVID-19 during the humid summer in Texas and Florida. What Johnson has seemingly overlooked is that COVID-19 is likelier to be contracted indoors. During the hot summer, air conditioners are in widespread use, especially in shopping malls and other places of business. AC dries out the air, thus mitigating humidity indoors. However, the current state of knowledge regarding transmission of SARS-CoV-2 is still developing.
Adding to his litany of ad hominem against Rancourt, Johnson states that Rancourt does not understand the mechanism of the N95 mask.
Johnson also takes aim at those who share the position of Rancourt regarding the paucity of data supporting mask prophylaxis (disparaged as Rancourt’s “echo chamber” and ideologues averse to scientific evidence), and says such people cannot be expected to be persuaded by his argument since “you can’t reason someone out of a position they didn’t reason themselves into.” Negative aspersions aside, this is easily refuted sophistry. For example, many people grew up as Christians (and other religious faiths) and accepted it as the truth because that was what the community and a wide segment of society believed (aka the appeal to ancient wisdom fallacy), but some people would later scrutinize the words of the religion and the archaeological evidence. Many came to the conclusion that the evidence is underwhelming and became apostates.
Johnson, who has accused Rancourt of falling prey to the all-or-nothing fallacy, finishes off with, “What I have presented is enough to persuade any fair and open-minded person that, yes indeed, public-mask mandates exist in curbing the spread of COVID-19…” Thus, Johnson can be accused of succumbing to the all-or-nothing fallacy — either one is fair and open-minded and is persuaded by Johnson’s argument or that person is unfair and close-minded and refuses his argument.
Johnson says that masks work and “the burden of proof is on Rancourt to prove otherwise.” Again, this is not how science works. Science disproves, not proves phenomena.
Later in the debate, Johnson introduces a bizarre analogy, a faulty analogy.
It’s like putting a cast on your broken leg, and your friend says “There could be asbestos in that cast, I wouldn’t do that.” [You respond] “Yeah, I guess, but until I actually have good reason to think there is an actual risk, and the risk outweighs the benefit, I’m going to stick with what has been proven to work.”
Rancourt posits two things:
1) the RCT evidence shows masks do not work
2) mask wearing is known to cause harm
Johnson’s analogy posits:
1) casts are known to help heal broken bones (he does not directly posit this, but assuredly he agrees with the premise)
2) there might be asbestos in the cast, and asbestos might be harmful
Comparing the two:
1) Rancourt points to something not working and Johnson points to something known to work (these are mirror opposites)
2) Rancourt points to known harms; Johnson points to potential harm
Perhaps the most egregious of Johnson’s arguments is that RCTs can not be done and would be unethical. Yet, the fact is, as Rancourt stated, several RCTs have been done. As for the ethical prohibition against such a RCT, that is premised on the assumption that masks are known to work and that masks are safe to be worn. In other words, the results were in before the science was carried out. That is not science.
Johnson sees no need for RCTs because observational studies are sufficient in his estimation. Johnson apparently lacks understanding in experimental design and methodology. As Rancourt argued, “RCTs are designed to remove bias … and, while not perfect, are the best way to get at the truth.”
Rancourt adds that since there is no evidence yet that masks are effective, this points to the conclusion that even if there is an effect, that “effect is too small to have been detected.”
In the follow-up, Johnson appeared flustered and referred to a collection of bad studies that find no evidence for masks working, what he calls an appeal to ignorance, aka absence of evidence is not evidence of absence. The philosopher argues that because the studies “were unable to detect any difference… that doesn’t prove they [masks] don’t work.” Sagan points out the absurdity of such an argument: “There is no compelling evidence that UFOs are not visiting the Earth; therefore UFOs exist.”12 Another problem with such an argument is that it attempts to shift the burden of proof. Yet, the onus must be on a person making a claim to support that claim.
Johnson wields ad hominem to support his case; he says Rancourt doesn’t understand, misrepresents, is a conspiracy theorist, and then attempts to paint him as a Nazi because his article was posted at a Nazi website — in other words the fallacy of guilt by association.
In the end, Johnson argues against ad hominem; he says an “argument should stand or fall based on evidence.”
To be fair, Rancourt is not pristine during the debate. He does give back later in debate. He says to Johnson, “You wouldn’t even know how to measure it [a minimally infectious dose].” He dismisses a “broad nonsensical question,” talks of “spinning and misrepresenting” and “cherry-picking.” At one point, he says to Johnson: “You’re nuts.”
And Rancourt talks about proof as well.13
Watch the video below, read the papers, and judge for yourself what is scientific, factual, and logical.
CORRECTION: “meta-analysis” is changed to “sytemic review”
- Carl Sagan, The Demon-Haunted World: Science as a Candle in the Dark (Headline Book Publishing, 1997): 34.
- See Karl Popper, The Logic of Scientific Discovery (New York: Routledge Classics, 2005).
- Lars Bondemark and Sabine Ruf, “Randomized controlled trial: the gold standard or an unobtainable fallacy?,” European Journal of Orthodontics, 37(5), October 2015: 457–461, https://doi.org/10.1093/ejo/cjv046
- Denis Rancourt, “Masks Don’t Work: A review of science relevant to COVID-19 social policy,” Technical Report · April 2020. See interview, “Do Masks and Respirators Prevent Viral Respiratory Illnesses?” 8 May 2020.
- Today, the local newspaper tells us the number of people testing positive for COVID-19 is falling “a development experts credit partly to increased wearing of masks– even as the outbreak continues to claim nearly 1,000 lives in the U.S. each day.” in “New virus cases fall in the U.S. and experts credit masks,” Times-Colonist, 26 August 2020: A9.
- David Kyle Johnson, “A (Complete) Debunking of Denis Rancourt’s Argument That ‘Masks Don’t Work,’” 28 July 2020.
- Quoted is a 2020 CDC published paper, “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.” See Todd McGreevy, “Still No Conclusive Evidence Justifying Mandatory Masks,” River Cities’ Reader, August 2020.
- Faisal bin‐Reza, Vicente Lopez Chavarrias, Angus Nicoll, & Mary E. Chamberland, “The use of masks and respirators to prevent transmission of influenza: a systematic review of the scientific evidence,” Influenza and Other Respiratory Viruses, 6(4), July 2012: 257-267.
- Jacobs JL, Ohde S, Takahashi O, Tokuda Y, Omata F, Fukui T. “Use of surgical face masks to reduce the incidence of the common cold among health care workers in Japan: a randomized controlled trial,” Am J Infect Control. 2009;37(5):417-419. doi:10.1016/j.ajic.2008.11.002
- Cowling, B., Zhou, Y., Ip, D., Leung, G., & Aiello, A. (2010). “Face masks to prevent transmission of influenza virus: A systematic review,” Epidemiology and Infection, 138(4): 449-456. doi:10.1017/S0950268809991658
- Denis Rancourt, “Masks Don’t Work: A review of science relevant to COVID-19 social policy,” Technical Report · April 2020: 2.
- Sagan, p 199.
- “We prove that the ‘COVID-peak’ feature that is present in the all-cause mortality data of certain mid-latitude Northern hemisphere jurisdictions, including France, cannot be a natural epidemiological event occurring in the absence of a large non-pathogenic perturbation.” [italics added] See Denis Rancourt, Marine Baudin & Jérémie Mercier, Abstract in “Evaluation of the virulence of SARS-CoV-2 in France, from all-cause mortality 1946-2020,” Research Gate, August 2020. This report, while likely contentious, deserves closer examination.