The Ioannidis videos posted in this thread are interesting. He comes across as very calm and reasoned guy and he obviously should have the science down to pat given his background but (there is always a but), in the first video posted back in March, every one of his assumptions about the disease were talking down the risk. He was quite right that we didn't (and still don't) have enough data, but it was obvious that his gut feeling was that it wasn't as serious as feared.
The new video references the Stanford study which makes very bold claims but we've already had a number of critiques of this work. Specifically, they are assuming that the specificity of their test is incredibly high and has very few false positives. They have also made some extrapolations from their data which don't match the population spread of the Santa Clara County where the work took place. Their selection criteria was problematic as well (asking for volunteers in Facebook!). This link has probably been posted earlier in this thread and provides what is pretty much a takedown of the statistical analysis of the study:
https://statmodeling.stat.columbia....-in-stanford-study-of-coronavirus-prevalence/
I can't help but feel that the study began with a belief that it was massively more widespread and has worked towards that conclusion.
Hopefully, similar studies can be carried out elsewhere with a solid statistical analysis available so we have some sort of a comparison to these Stanford figures. Perhaps ones which are more widespread with random sampling taking place.
The new video references the Stanford study which makes very bold claims but we've already had a number of critiques of this work. Specifically, they are assuming that the specificity of their test is incredibly high and has very few false positives. They have also made some extrapolations from their data which don't match the population spread of the Santa Clara County where the work took place. Their selection criteria was problematic as well (asking for volunteers in Facebook!). This link has probably been posted earlier in this thread and provides what is pretty much a takedown of the statistical analysis of the study:
https://statmodeling.stat.columbia....-in-stanford-study-of-coronavirus-prevalence/
I can't help but feel that the study began with a belief that it was massively more widespread and has worked towards that conclusion.
Hopefully, similar studies can be carried out elsewhere with a solid statistical analysis available so we have some sort of a comparison to these Stanford figures. Perhaps ones which are more widespread with random sampling taking place.