Don't shoot the messenger - I was merely reporting the thinking of some of my daughter's generation. In any event, I would think that your comments underline the point - that, despite it being more catastrophic than Covid has been, the world did not 'shut down' back then.
In any event, the fact that the world did not 'shut down' makes the comparison very difficult. We have just passed 4 million Covid deaths globally. Had no country 'taken any measures' the figure would probably be at least an order of magnitude higher, quite possibly a lot more. Hence I don't think it is a foregone conclusion that Covid would have been much less of a global catastrophe than was Spanish 'Flue had we done as little 'shutting down of the world' as happened back then.
As I've said, that ideal is not always attainable - for example, in this case, in which I believe that undertaking a trial that could produce 'robust evidence' would essentially be impossible. The absence of such evidence doesn't necessarily stop us - where, for example, is the 'robust evidence' to support convicting someone of a criminal offence for driving at 80mph on a nearly empty motorway under good road/weather conditions?
There are essentially two ways in which "failing to reject" (i.e. being consistent with) a null hypothesis can arise. Firstly, most obviously, if there is no true difference between the things being compared. Secondly, if, despite their being a 'true difference', the trial was under-powered (too small) - what is known in the trade as a Type II Error (failing to detect, as 'significant', a true difference which is actually present). All of the mask trials I've seen so far have been pitifully under-powered (too small), hence it is very possible that many of the ';non-significant' results are simply Type II Errors.
We've been through this, and you're really underlining my point. The result is "not statistically significant" because, as you say, the difference between groups only amounted to 11 subjects, and that's because the study was so pitifully small to detect (as 'statistically significant') the very small observed difference in infection reates. Had the study been of the size required to provide adequate statistical power (as I said, around 100,000 subjects), rather than just under 5,000, and if the proportions had remained the same, the difference in number of infections would have been somewhat over 220, and the difference would have been highly statistically significant.
I can tell you that if it had been considered to be a 'pilot study' (which is all it deserves to be regarded as, since it only had about 5% of the subjects required for an adequately-powered trial) then, in the world in which I function, it would have been considered as a 'very promising result', adequate to justify proceeding to a full-sized study.
I can see what they probably did, and this is a problem I very often find myself having to deal with. I would suggest that, in order to achieve a sample size that they considered to be 'do-able' (and affordable), they designed (the sample size of) a study which would only be powerful enough to detect (as 'statistically significant') a totally unrealistic effect of mask-wearing (a 50% reduction in infections) - I cannot believe that anyone in their right mind would think that wearing masks could halve the number of infections, particularly in the wearers of the masks. Calculation is consistent with their having done that - if they assumed a 2% infection rate in non-mask-wearing people then to have a 90% power (probability) to detect (as statistically significant' at p=0.05 level) a 50% reduction (i.e. to 1%) in infection rate in mask-wearers, that would require 3,018 subjects per group - extremely close to the 3,030 + 2,994 that they recruited.
Thanks. I'm not sure that I'll have time to plough through many of them - particularly given that, as I thought we had agreed, mask-wearing is a pretty minor part of the big picture. For what it's worth, in terms of Covid, I don't think I've seen any (of the almost all unsatisfactory) trials which have given much of a suggestion that (Covid) 'outcome' is worse in people wearing masks.
... I only mentioned that as something which was only 'a few steps ahead' of the buck-passing that the government seems to be doing. In any event, the numerical speed limits are, themselves, obviously arbitrary (and do not take account varying road/traffic conditions etc.). The only 'robust evidence' is that, in general, lower speeds result in less deaths and serious injuries, but there is no 'robust evidence' of which I am to support any particular numerical limit (rather than a similar, but different, one). As you imply, we already rely on driver judgement to drive at speeds below the legal maximum when conditions require that, but we don't allow the same judgement (to drive faster that the arbitrary limit) when conditions render that 'appropriate' (e.g. on an almost-empty motorway in good weather conditions).
Don't get me onto seatbelts (or crash helmets)
I have always been a passionate advocate and user of seat belts - back in the 60s, I installed and them in my first car at a time that it was not even a requirement to have them, let alone mandatory to wear them. However, I've also always been equally passionate in my 'opposition' to the legal requirement for drivers and adult front-seat passengers to wear them, because one has to really scrape the barrel to think of ways in which not wearing a seatbelt will harm anyone other than the person concerned any more than would countless other activities which are totally legal!You seem to be speaking for my daughter and her contemporaries! In the manner they suggest, history might look back and point out that if the world had simply 'ignored' Covid (and allowed it to 'kill however many people it was going to kill'), then none of that 'collateral damage' (consequent upon the world's response to the virus, not the virus itself) would have arisen. It's certainly one possible approach, and I've heard a good few 'significant' people suggesting that it is what we should have done.
As I commented earlier, I don't think anyone understands what has been happening in Sweden, at any stage. As you imply, despite their relative lack of control meaures, for most of the pandemic they have had similar numbers of deaths per capita as the UK, but for much of the past ~4 months their figures (during our 'lockdown') became much higher than ours.
Kind Regards, John
