I've seen this a few times today, so I wanted to check it out. There's something to it, but not quite as dramatic as this would suggest. 1/
I took the case data from COVID Tracking over the past 60 days, as I've been doing for my weekly charts. I then took each state as an observation, and looked at the percentage increase in the number of cases in each state over that 60 day period. 2/
There's some evidence that being in a state with no mask mandate is associated with a higher increase in cases over that 60 day period (p = 0.07). Note the causal claim is harder to make, because there's other things that could be driving this, e.g., 3/
temperature in the sun belt could be driving more people indoors, states with masks were hit harder so immunity levels may be higher or people may be more reluctant to venture out, etc. Regardless, 4/
we lack evidence that, over this 60 day period mandatory mask wearing provides any benefit over requiring both certain employees and customers to wear masks (p = 0.68) or certain employees (p = 0.65). 5/
In other words, there may be less intrusive means of accomplishing this goal. Now, over the past two weeks, it *is* a different story. The evidence is pretty strong that there is at least an association b/w mask vs. no mask states. 6/
States with a mask mandate saw about a 25 percentage point lower increase in cases over this time period (p < 0.001) vis-a-vis states with no mandate. 7/
But again, the evidence is weaker for states that just required employees to wear masks -- about a 10 percentage point lower increase in states that had mandatory masks everywhere vs. "certain employees," and the p-value is 0.11 8/
Of course, in both cases the r-square is terrible (0.02 for the 60-day model and 0.2 for the 14-day model) suggesting either a *lot* of randomness or other factors were aren't considering (or some combination). 9/
So, I don't think this advances the ball that much. I still wear a mask when I go out, and you probably should too. It's a minor inconvenience with potentially major benefits. But the evidence for the impact of government policy seems surprisingly weak. 10/10
