I agree with your overall point but I'm really curious about your finding of no correlation between median county age and mortality. I've been doing a lot of analyses of the data from just the state of Ohio and within Ohio, a linear regression (weighted by county population) between county death rate and percent of county population > 65 years old has a correlation with R^2 = 0.63 (p < 1e-15). (The correlation is considerably stronger with percent over 65 than with median age but also quite strong with median age.) By contrast, death rate vs. Trump vote percentage has R^2 = 0.18 (p < 1e-4). This makes me wonder if when looking across different regions of the country there are other confounding factors that are masking the relationship between county age and mortality. Because within the single state I'm analyzing (which might eliminate some of the confounding factors) the relationship between the age of a county's population and its COVID death rate is just incredibly strong.
But if we limit the analysis to deaths since the end of June 2021, the results flip - the correlation with percent over 65 has R^2 = 0.16 (p = 0.00013), with Trump vote percentage R^2 = 0.43 (p < 1e-11). In a multivariate linear regression against both, Trump vote has p < 1e-8 and percent over 65 has p 0.086. Supporting that since the vaccine rollout, the major determinant of county death rate switched from population age to Trump support.
what about Nursing home deaths? At the beginning those were one of the biggest sources of deaths, after they instituted rigorous NPI's it went way down, and with vaccines even more so but how big was it initially?
I agree with your overall point but I'm really curious about your finding of no correlation between median county age and mortality. I've been doing a lot of analyses of the data from just the state of Ohio and within Ohio, a linear regression (weighted by county population) between county death rate and percent of county population > 65 years old has a correlation with R^2 = 0.63 (p < 1e-15). (The correlation is considerably stronger with percent over 65 than with median age but also quite strong with median age.) By contrast, death rate vs. Trump vote percentage has R^2 = 0.18 (p < 1e-4). This makes me wonder if when looking across different regions of the country there are other confounding factors that are masking the relationship between county age and mortality. Because within the single state I'm analyzing (which might eliminate some of the confounding factors) the relationship between the age of a county's population and its COVID death rate is just incredibly strong.
But if we limit the analysis to deaths since the end of June 2021, the results flip - the correlation with percent over 65 has R^2 = 0.16 (p = 0.00013), with Trump vote percentage R^2 = 0.43 (p < 1e-11). In a multivariate linear regression against both, Trump vote has p < 1e-8 and percent over 65 has p 0.086. Supporting that since the vaccine rollout, the major determinant of county death rate switched from population age to Trump support.
what about Nursing home deaths? At the beginning those were one of the biggest sources of deaths, after they instituted rigorous NPI's it went way down, and with vaccines even more so but how big was it initially?