Re: Covfefe-19: We Can Handle Slight Inconveniences. Part 8.
I'm a little concerned about the antibody test results Cuomo is touting. First, I don't know a lot about antibody testing but I would be very cautious if they exclude a couple things:
1. Specificity of the tests (essentially gives the number of false positives). This is especially important given that we don't seem to know if someone who had a different coronavirus that's essentially harmless but has not been infected by COVID-19 could give a positive result on these tests.
2. How they selected the samples
3. A caution that a positive test result doesn't mean you're immune, even if it's not a false positive. We don't know that yet. We can predict, but we don't know.
4. Relatedly, if the false positive rate is fairly high, that's... not good.
5. How did they process the data.
6. Whether they are using resolved cases that correspond to the sampling dates to determine any kind of fatality rate
Regarding point 4, there's a great short video from Pro Publica (I really, really need to subscribe/donate to them) that helps explain why proportion of FP to total positives is important:
<blockquote class="twitter-tweet"><p lang="en" dir="ltr">In some situations, antibody tests generate more false positives than real ones. It’s because of ... math. Here's what you need to know. <a href="https://t.co/jyertTc60Z">pic.twitter.com/jyertTc60Z</a></p>— ProPublica (@propublica) <a href="https://twitter.com/propublica/status/1256568874365063168?ref_src=twsrc%5Etfw">May 2, 2020</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
For point 5, here's a fairly good but dense thread on why there are some issues with the studies coming out of Santa Clara right now and why they might be important for the NY testing results.
<blockquote class="twitter-tweet"><p lang="en" dir="ltr">The statistical analysis in the Bendavid et al. revision still seems to have at least one fatal bug.<a href="https://t.co/bLqgRdflkj">https://t.co/bLqgRdflkj</a></p>— Will Fithian (@wfithian) <a href="https://twitter.com/wfithian/status/1256308404194144256?ref_src=twsrc%5Etfw">May 1, 2020</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>