[Goose]

As we know, all models are only as good as their data. The dropping Ro rates could be a function of degrading testing capacity. Apparently the re-agents to do testing are understandably in short supply because the whole world wants them.

I have a hard time believing Arkansas, with no stay at home orders, has a lower Ro than CT with such orders, and about 2x the testing per capita.

Arkansas has a population density of 58 persons per square mile. CT has a population density of 736 persons per square mile. That might have a lot to do with it. The number of people in Arkansas that commute to areas of even higher density is also much less than CT. CT will therefore "need" a stay at home order much more than Arkansas. Probably lots of other behavior factors involved. It is almost certainly true that one size doesn't fit all.

There is also clearly significant "noise" in the measurements using to compute this value. If testing is skewed by different availability in different areas for example, even within the state this estimate is GIGO anyway. If one looks at reported cases by day from various states, there is frequently (but not always) a 7 day periodicity in reported cases present in the data. That is shown in the "wiggles" of the Rt estimates.

[Goose]

As we know, all models are only as good as their data. The dropping Ro rates could be a function of degrading testing capacity. Apparently the re-agents to do testing are understandably in short supply because the whole world wants them.

I have a hard time believing Arkansas, with no stay at home orders, has a lower Ro than CT with such orders, and about 2x the testing per capita.

Arkansas has a population density of 58 persons per square mile. CT has a population density of 736 persons per square mile. That might have a lot to do with it. The number of people in Arkansas that commute to areas of even higher density is also much less than CT. CT will therefore "need" a stay at home order much more than Arkansas. Probably lots of other behavior factors involved. It is almost certainly true that one size doesn't fit all.

There is also clearly significant "noise" in the measurements using to compute this value. If testing is skewed by different availability in different areas for example, even within the state this estimate is GIGO anyway. If one looks at reported cases by day from various states, there is frequently (but not always) a 7 day periodicity in reported cases present in the data. That is shown in the "wiggles" of the Rt estimates.