Credit where credit is due
I am often critical of libertarian economist Donald Boudreaux. This morning, I sent him an email criticizing a post of his in which he cited a study claiming that pcr tests for covid have a false positive rate of 80% or more. I pointed out that this is obvious nonsense, since overall test positivity rates, which include both false and true positives, have generally been in the 4% to 10% range. I also criticized him quite harshly for passing this (mis)information on in an uncritical manner. To his considerable credit, he published my comment in full on his blog and removed the link from his initial post.
Eric:
Making a note of Eric being kind to Libertarian.
Since the link is now censored, I can’t answer my own question: is the study truly that absurdly wrong (if so how did it get published?) or is there a repertorial misunderstanding of Bayesian probability?
To wit: if at any given moment 1% of people have Covid, and the test is 96% accurate, then testing 1000 people will give around 50 positive results (roughly 1% + 4%) . The odds that someone who tests positive really does *not* have Covid will be around 40/50, or 80%.
To a lay reporter that will feel like an 80% false positive rate although technically I used a 4% rate for the sake of argument.
So I’m left to wonder what exactly went on.
PaulS,
Nice. It has been said that the US has a problem with STEM education, but thanks to our problem in language education then it is not said nearly often nor clearly enough.
PaulS: Boudreaux linked to and quoted from a web article, not the underlying study. The passage he quoted suggested that pcr tests would yield between 80 and 97% “false positive rates”. It’s certainly possible that the reporter made the mistake you outline. The issue is why Boudreaux, a Ph.D. economist who has spent the last 6 months railing about covid, would not see the problem – or at least be suspicious! – and refrain from circulating this inflammatory nonsense. FWIW, a very brief web search (by me) suggests that the pcr false positive rate is about 0.5%.