Replies to post #231115 on Biotech Values

[jq1234]

Again I would caution conclusion of that result. They tested 3,330 with 50 positive, resulting 1.5%. This wasn’t random sample testing. To get that 50-80 fold numbers they depended on double adjustments of both sensitivity/specificity and population tested vs population from the county.

Test kit estimate “sensitivity of 80.3% (95 CI 72.1-87.0%) and specificity of 99.5% (95 CI 98.3-99.9%)”, but they gave strong disclaimer and warning about their final conclusion could be substantially off, due to test kit actual sensitivity/specificity and population selection bias I highlighted:

https://twitter.com/jq1234t/status/1251198958195654658?s=21

https://twitter.com/jq1234t/status/1251207334002638858?s=21

Their population adjustment was even more questionable.

[DewDiligence]

Re: (Unreliable) Santa Clara study

Along with the links furnished by jq1234, this Twitter thread should convince you that the Santa Clara study is unreliable:

What do I see as the major problems?

1. Poorly chosen, non-representative samples

2. Weakness at thinking through important Bayesian principles

3. Inconsistency with other, more verifiable results

4. Attachment to a particular idea (slow willingness to update on evidence)

— Alan Cole (@AlanMCole) April 18, 2020

It’s a long thread, but reading it is well worth the effort.