Impact on COVID-19 of Wearing Masks, Population Density and Mass Gatherings
I offer the following as a step forward in measuring the value of wearing masks in public. Although one's choice to wear a mask actually serves a proxy for a whole set of behaviors. So, it's the entire set of behaviors that is actually being measured.
One of the first steps in any analysis should be to find simple relationships within the data. This must be done before one starts tackling complex data sets with overwhelming computer power.
In this case, I'm trying to predict Pandemic severity (COVID-19 patients per 100k state residents) using three variables for each of 16 contiguous Midwestern states:
--% of citizens wearing masks
--Adjacency to South Dakota (impact of Sturgis Rally)
--State population density
The three variables explain 69% of the variance in the states' current Pandemic Severity. Each variable and the intercept's statistical significance exceeds 97%.
The impact of each variable on current hospitalizations per 100k state residents follows:
--1% greater mask wearing: ( -2 )
--Adjacent to South Dakota: +10
--10 more people per sq mi +0.64
This analysis isn't definitive. It is based on a few states at a specific point-in-time. But its simplicity is evidence the relationships among the data are real and should be considered in any analysis determining the reasons for the differing severity of the pandemic among states and nations.