3 ms·

> unless you posit that there actually is an effect but a confounder perfectly balances it out in the opposite direction. No need for an exact balance; ther st

> unless you posit that there actually is an effect but a confounder perfectly balances it out in the opposite direction. No need for an exact balance; ther study found that there was a positive correlation between mask-wearing and infection. That suggests there was a confounder of some kind, and and that it was more than sufficient to overcompensate for any protective effect. The only explanation I've seen offered for the positive correlation is that people wore masks more when infection rates were known to be high; that explanation is obviously a confounder.

native_samples 5y ago
It found such a correlation only sometimes, and at any rate such a correlation is either irrelevant or harmful to the actual claim of interest (that mask mandates have a negative effect on COVID). The correlation between masking and reduction of COVID is zero. It certainly doesn't imply the existence of a confounder that takes real effectiveness and reduces it somehow past zero. Once you try to explain the (inconsistent) positive correlation between masks and COVID you're back in the territory where "correlation != causation" is a valid counter argument. As you note, such a correlation could be driven entirely by the false perception of effectiveness, without any actual effectiveness. But it can also be for other reasons e.g. places that were tougher about enforcing mask mandates had generically less competent governments that were more likely to do harmful things like flushing hospital patients into care homes. More likely: such correlations are spurious and there's no real causality there at all. The fact it doesn't show up consistently would certainly imply that. Also possible that the IHME mask usage data is bogus in some countries, or potentially lots of other things. We have no evidence that can support any explanation over any other.
- denton-scratch 5y ago
  I agree with most of that, I think. Especially that the correlation between masking and protection isn't supported (I don't agree that it's zero; it looks to me like there's scant evidence either way). I agree that a counfounder can't reduce effectiveness below zero.

native_samples 5y ago

It found such a correlation only sometimes, and at any rate such a correlation is either irrelevant or harmful to the actual claim of interest (that mask mandates have a negative effect on COVID). The correlation between masking and reduction of COVID is zero. It certainly doesn't imply the existence of a confounder that takes real effectiveness and reduces it somehow past zero. Once you try to explain the (inconsistent) positive correlation between masks and COVID you're back in the territory where "correlation != causation" is a valid counter argument. As you note, such a correlation could be driven entirely by the false perception of effectiveness, without any actual effectiveness. But it can also be for other reasons e.g. places that were tougher about enforcing mask mandates had generically less competent governments that were more likely to do harmful things like flushing hospital patients into care homes. More likely: such correlations are spurious and there's no real causality there at all. The fact it doesn't show up consistently would certainly imply that. Also possible that the IHME mask usage data is bogus in some countries, or potentially lots of other things. We have no evidence that can support any explanation over any other.

denton-scratch 5y ago
I agree with most of that, I think. Especially that the correlation between masking and protection isn't supported (I don't agree that it's zero; it looks to me like there's scant evidence either way). I agree that a counfounder can't reduce effectiveness below zero.