EyeRunMD wrote:
Thom wrote:
ike wrote:
Two-state comparisons have little value. It’s usually just data mining or cherry picking in support of a conclusion. There are so many possible two-state combinations that one can almost always find some pair that supposedly proves something. And another pair that proves the opposite. Why not compare ND to ME: they are both rural and cold, yet had very different policies and very different results.This is at least the 3rd time cherry picking the CA/FL data has come up on this forum. At least twice I've pointed out the data mining aspect. You could easily compare NY and Florida and prove how effective mitigation efforts work. The comparisons raise some interesting questions, but not really reliable conclusions. Real data scientists have built models on mask wearing and restaurant closings that have shown to be pretty damn accurate.
I’m at work so I don’t have time right now to look for these models but I find it hard to imagine a model is a good comparison of modeled data vs real world actual data. Maybe the models are based on actual data (like I said, I can’t look them up right now).
There are also case examples showing masks don’t make a big difference (ie, some of the cruise ship case studies) vs others that do show the effectiveness of masks use. It just goes to show the info is all over the place when evaluating COVID.
I think the models accurately tell you what the model tells you. It can compare, for example, masks and restaurants at half capacity to no masks and restaurants at full capacity. When doing a state by state comparison, (ie "real world actual data") you involve population density, weather, rules compliance, activity types, age, health factors, etc. IOW, you can't look at Florida and assume that the same rates of death and infection would occur in New York City if the same rules were applied.
Both statistical modeling and real world data has to be considered.
-----------------------------Baron Von Speedypants
-----------------------------RunTraining articles here:
http://forum.slowtwitch.com/...runtraining;#1612485