How the ‘Lockdown Sceptics’ misquote research in their quest to appear scientific,Danny Bradley,The Skeptic

I learned a new word the other week. Meretricious. Defined by Oxford Languages as ‘apparently attractive but having no real value’. If I didn’t believe in coincidences, I might well imagine that the gods of fate had meant for me to discover this lovely new word in the same week that Will Jones’ article ‘COVID-19: Just The Facts’ was published by the website LockdownSceptics.

The website was founded in April 2020 by ex-Daily Mail journalist Toby Young, who wrote in June 2020 that the second wave in the UK had ‘refused to materialise’ and that the virus had ‘all but disappeared’. (He later apologised, saying ‘hands up, I got that wrong’). Will Jones’ piece, like much of LockdownSceptic’s output, casts aspersions on the efficacy of masks, lockdowns, social distancing measures, COVID-19 vaccines & the overall severity of the virus itself, all of which is commonly tread ground in Covid-sceptic circles.

Many scientists may feel an urge to write off a piece like this without engaging with it. In fairness, I can see why. However, I think that would be a big mistake. Jones’s article, unlike some conspiracy theorist content fermenting in the dark corners of YouTube, claims to be ‘fully referenced from peer-reviewed research and leading authorities’. And, at first glance, this does appear to be true.

References range from randomised control trials (RCT’s) published in reputable scientific journals like The American College of Physicians, PLOS One, and the American Journal of Infection Control, to comparative studies and reviews in The Lancet. To non-academic-science literate people, which is almost all of us, this looks incredibly convincing. Especially when we are all pandemic-fatigued, and sick to death of the harmful, contradictory advice given by the US and UK governments since February 2020. Therefore, I believe it is not only worthwhile but absolutely vital to address the kinds of arguments Will Jones is making here. And to point out, as is the case in his piece, that almost none of the research he links to actually supports his position.

Jones routinely makes the fatal error of quoting selectively from the conclusions of the research he references, often omitting crucial information which would otherwise undermine his entire argument. The neglecting of inconvenient data is a pattern that runs throughout his piece, as we will see.

I believe that Jones’s article can be truly useful, not only because it contains many of the most common Covid-sceptic talking points in their most persuasive form, but also because it can serve as a valuable education in how some pundits are taking genuine scientific research and contorting it via sleight of hand to fit their own agendas.

The article’s claims are too numerous to address one by one, so I will be focusing on the claims Jones makes about one aspect of the debate: masks. Let’s see how well the literature he cites actually buttresses his arguments.

‘Do masks work?’

This section links extensively to studies in reputable journals, which makes the overlying arguments seem compelling. That is, of course, until you read the studies. It then becomes slowly apparent that Jones has a sneaky penchant for not reading the fine print.

Jones cites a randomised control trial from 2013-15 which he claims ‘concludes that face masks did not seem to be effective against laboratory-confirmed viral respiratory infections, nor against clinical respiratory infections.’ But the trial’s conclusion notes state the following:

‘This trial was unable to provide conclusive evidence on face mask efficacy against viral respiratory infections, most likely due to poor adherence to protocol…’

The full conclusion goes on to say:

‘This trial failed to provide definitive evidence for the effectiveness of facemasks… This was likely due to poor compliance with facemask use. We report difficulties in implementing a large cRCT, evaluating the effectiveness of facemasks against viral respiratory infections including participants’ poor compliance with the protocol…’

In other words, some participants did not follow the instructions on wearing masks properly, or even wore them for the full length of time that they were supposed to, so the masks didn’t help. Either Jones did not read the study he is referencing in full, or he did and is ignoring the part of the conclusion that disagrees with him. Let us pay him the compliment of assuming it is the former.

Jones links to a comparative study from 1992 to support the claim that ‘even surgical masks do not filter out enough infected aerosols to be considered respiratory protection devices’. The use of ‘even’ in this sentence is particularly misleading, as it implies that surgical masks are the best, most protective masks available. The implication being that if ‘even surgical masks’ aren’t very protective, then surely no masks are particularly helpful. (The word ‘surgical’ perhaps offering connotations of serious doctors in hospitals, wearing lab coats, relying on the use of these masks). This is not true. N95 masks are the most protective masks available, as they have a 95% filtration efficiency. That’s why they are called that.

This is why N95’s are favoured and prioritised for frontline workers. Even notable, prominent Covid sceptics like Alex Berenson do not dispute that N95’s significantly protect hospital workers. If frontline workers were using surgical masks instead, it’s because they had to. Because they ran out of N95’s. Or because the government didn’t supply them with enough (and surgical masks are probably slightly better than nothing at all). The only thing that Jones’s studies show so far is that some masks work better than others, and we should want hospital workers to use equipment that will protect them the most. Most experts agree with this already.

One of the main studies that Jones gives as evidence that masks don’t work very well is an oft cited Danish RCT which Jennifer Handsel deftly untangles at length in this Current Affairs piece. There are quite a few problems with the trial, not least of which is the fact that it doesn’t seem to be measuring what Will Jones says it’s measuring. As Handsel writes,

‘The study did not look at masks’ protection to others, or the aggregate effect of widespread mask usage’.

The trial’s stated statistical power also appeared to rely on the false assumption that PCR tests are 100% accurate, which they are not.

Many pre-pandemic mask studies, like the ones Jones links to, have similarly failed to find that masks offer significant protection. The New England Complex Systems Institute published a systematic review In February 2021 called ‘Unmasking The Mask Studies’, which found that many of these studies were significantly lacking in statistical power. As it happens, some of the very studies that Jones cites in his piece actually appear in this NECSI review:

‘False positives were not accounted for in the [Denmark] study’s statistical analysis or conclusions… [this] may limit statistical power’.

The fact that Jones didn’t seem to spot that his sources had gaping methodological flaws (especially when even the sources themselves sometimes flag them up!) ought to raise serious questions about the bar for scientific rigour that he deems sufficient. A recent evidence review by PNAS found that ‘The preponderance of evidence indicates that mask wearing reduces the transmissibility per contact by reducing transmission of infected droplets’, and recommended that ‘public officials and governments strongly encourage the use of widespread face masks in public.’

Jones also makes the point that ‘additionally, aerosols routinely escape with breath around the sides of the mask’. This is not an argument that masks don’t work. It is an argument that masks aren’t perfect, and some masks are even less perfect than others. This fact is uncontroversial in the global medical community — this is why it is recommended that masks be used in conjunction with other safety measures. A Science Magazine report, published in May 2021, found that ‘face masks effectively limit the probability of SARS-CoV-2 transmission’, especially when used ‘in combination with other preventative measures like ventilation & distancing’.

Just because protection is incomplete does not mean it is useless. Masks slow down exhaled air, which makes water droplets containing the virus less likely to reach people around us, even if some aerosols escape around the sides of the mask. Quantitative analyses exist which demonstrate that ‘wearing a mask will offer substantial, but not complete, protection to a susceptible person by decreasing the number of foreign airborne sneeze and cough droplets that would otherwise enter the person without the mask’.

Jones cites another review which he says shows that ‘surgical masks…’ (again, surgical masks) ‘… provide the wearer with protection from only 6% of infections’, and that ‘the same study’s review of RCT’s for masks as source control finds no evidence above low quality’. However, under the section ‘Quality of evidence’, the paper reveals that:

‘Many of the included RCT’s reported that participants did not follow instructions about wearing facemasks. Several reported that some controls wore face masks during the monitoring period, while many participants did not wear their face masks the majority of the time. All of the RCTs included in our review provided specific face masks (usually surgical grade, rarely P2 or equivalent grade respirator) with instructions on how to wear the face masks, how often they should be changed and how to hygienically dispose of used face masks. No information was reported about the types of face masks that (contrary to protocol) some controls in RCTs used. Very few of the observational studies collected information about what type of face covering was used. Several studies highlight potential problems of recall bias.’

The review also concludes that ‘The quality of the evidence is problematicour best estimate is between 6%-15% reduction in disease transmission’.

This review not only goes to great lengths to disclose the data’s shortcomings (and why we cannot glean much as a result of them) but also affirms that which we already know; surgical masks aren’t brilliant, better masks are better, and if participants in an experiment remove their masks whenever they feel like it, it tends to undermine the validity of the results. As the NECSI review concludes, ‘the studies that did not find statistically significant effects prove only that masks cannot offer protection if they are not worn.’

Despite cobbling together a semi-successful veneer of scientific credibility, not all of Jones’s references are to reputable journals. In trying further to demonstrate that masks don’t work, Jones links to a tweet by entrepreneur Yinnon Weiss who plots a series of graphs which purport to show how mask mandates ‘did not in fact contribute to altering the size or course of covid epidemics in countries & states all over the world’. Every single one of the graphs plots Covid infections against time, with a single mark plotted to show the point at which a mask mandate was implemented. They are presented to give the impression that mask mandates did not affect the spread of COVID-19.

N95 mask, left. Surgical mask, right.

This is called a univariate analysis. This is the opposite of a multivariate analysis. This tells us almost nothing. The fact that there could be a myriad of other factors contributing to the result is not even entertained, let alone accounted for; relative compliance levels between countries are unaccounted for. When or at what point lockdowns began or ended is unaccounted for. Whether or not social distancing was implemented (and how effectively it was enforced) is unaccounted for. The scale & range of anti-mask movements is unaccounted for. Lockdown fatigue is unaccounted for. The differing kinds of lockdowns are unaccounted for (How long, how harsh, how enforced, etc).The fact that several countries happened to implement mask mandates right as they were heading into a sharp spike anyway is unaccounted for (but made to look as though it were evidence of masks failing to curb a surge).

Univariate analyses are seldom useful. They only account for one variable. The exact same graphs could be plotted to show how jacket use contributed to covid surges, or ice cream consumption, or how many infected citizens are fans of Dan Brown novels. Correlation is not causation. This is the opposite of science. This is cherry picking data to prove a point you’ve already decided you’re going to make.

Here is a much better case study examining the effects of a mask mandate between two counties in the same US state. It accounts for many additional variables, unlike Weiss’s Twitter thread. The two counties share a state and therefore have similar laws, infrastructure, and cultural norms. And, as the report says, it’s ‘findings are consistent with declines in COVID-19 cases observed in 15 states and the district of Columbia, which mandated masks, compared with states that did not have mask mandates’. A recent CDC report also found that mask mandates in schools contributed to significantly lower incidents of COVID-19 (even more so when in conjunction with improved ventilation).

We could continue in this vein for a long time, as every other section in Jones’s article is similarly omissive of basic information. I encourage readers to entertain themselves by combing through the piece, and clicking on any bit of hyperlinked research they fancy. Again and again, the evidence does not support the argument, and in some cases even demonstrates the precise opposite.

It seems, then, that many of the arguments made by Covid sceptics, even in their most robust form, are unable to exist without overlooking a considerable amount of key information. Whether or not they are doing this on purpose, I do not know. Perhaps Will Jones sincerely did not read more than a few lines of the literature he cites. Or maybe he did, and is attempting to pull the wool over your eyes. Either way, one thing ought to be clear by now; Jones’s arguments are entirely unsupported by good data. Absolutely nobody should be convinced by them.

However, perhaps more importantly, refusing to engage with these arguments will not make them go away. Real scientists may feel, perhaps understandably, that addressing this type of chicanery is a waste of time. However, most people are not scientists, or even particularly science-literate. And as a population of tired, hurting people become ever more bruised by government u-turns and accusations of corruption, they will inevitably become increasingly susceptible to the siren call of false answers. Especially when those false answers are dressed up to look like science.

The fact that websites like LockdownSceptics appear to be alarmingly efficient at reaching a jaded general public should be a patent cause for concern. We ought to be doing everything in our power to counter their effect. After all, we are not writing to try and convince Will Jones. We are writing to try and persuade any unsure, fence-sitting readers who find his work attractive. In this exhausting age of confusion, uncertainty, and an endemic distrust in experts, meretricious science is more harmful than ever.

The post How the ‘Lockdown Sceptics’ misquote research in their quest to appear scientific appeared first on The Skeptic.

Even the most robust arguments of the so-called ‘Lockdown Sceptics’ rely on quoting selectively from the studies they reference
The post How the ‘Lockdown Sceptics’ misquote research in their quest to appear scientific appeared first on The Skeptic.