On Thursday last week, The Guardian published an article claiming that a new global study had demonstrated that mask-wearing cuts COVID incidence by 53%.
In my small sample size, my experience is that you're often stuck either way. When studies that overstate the effect are used, the error is obvious and undermines the point being made. However when studies that show the smaller real effect are used, they are discounted because the effect size is too small.. and knowing that it's possible to overstate the effect, these get discounted as evidence without biased individuals feeling the need to find a flaw in the study that backs up their view that they are overstating the effect. In many ways, when trapped in that sort of circular loop, the ones that overstate the effect come out better.. because even though their error is obvious, they have the defense that while it's may not be 53%.. surely there must be 10% left!
In my small sample size, my experience is that you're often stuck either way. When studies that overstate the effect are used, the error is obvious and undermines the point being made. However when studies that show the smaller real effect are used, they are discounted because the effect size is too small.. and knowing that it's possible to overstate the effect, these get discounted as evidence without biased individuals feeling the need to find a flaw in the study that backs up their view that they are overstating the effect. In many ways, when trapped in that sort of circular loop, the ones that overstate the effect come out better.. because even though their error is obvious, they have the defense that while it's may not be 53%.. surely there must be 10% left!