Yes I remember that one so well. I had nothing better to do so I ran my own crack Bayesian analysis at home and found not only was their estimate way off (by many X) compared to the expectation value but the spread in the posterior distribution was enormous. But of course the right wing mediasphere absolutely ran with it.
It's been a while obviously but I'll try to see if I can dig it up. I'm also not inclined to give these authors the benefit of the doubt, meaning, their biases simply blinding them from seeing defects in study design/analyses and accounting for them. This is just too egregious. I think this is deliberate intellectual dishonesty, scientific fraud perpetrated in order to give certain people the results they wanted (business people on the one hand, conservative religious people on the other).
I agree. They are at best completely incompetent, but I actually spoke with him last summer and he still insisted the study was right. That is willful ignorance. Not forgivable.
Yeah. There's no way somebody trained at this level doesn't know you need to correct for test specificity in order to get from test positive fraction to true prevalence. I'm going to look up what I did; it was back-of-the-envelope and not rigorous but enough to tell me the paper was a, well, you-know-what.
Surely testing for antibodies requires that the patient has been infected for at least 7 days and is unwell? It doesn’t catch people who are infected with mild symptoms as there’s no requirement for a significant numbers of antibodies as the innate immune system is handling the infection. Which means that antibody numbers will reduce once the infection is cleared.
Yes I remember that one so well. I had nothing better to do so I ran my own crack Bayesian analysis at home and found not only was their estimate way off (by many X) compared to the expectation value but the spread in the posterior distribution was enormous. But of course the right wing mediasphere absolutely ran with it.
Would love to see your analysis. I have been collecting critiques of this horrible study like it's a hobby.
It's been a while obviously but I'll try to see if I can dig it up. I'm also not inclined to give these authors the benefit of the doubt, meaning, their biases simply blinding them from seeing defects in study design/analyses and accounting for them. This is just too egregious. I think this is deliberate intellectual dishonesty, scientific fraud perpetrated in order to give certain people the results they wanted (business people on the one hand, conservative religious people on the other).
I agree. They are at best completely incompetent, but I actually spoke with him last summer and he still insisted the study was right. That is willful ignorance. Not forgivable.
Yeah. There's no way somebody trained at this level doesn't know you need to correct for test specificity in order to get from test positive fraction to true prevalence. I'm going to look up what I did; it was back-of-the-envelope and not rigorous but enough to tell me the paper was a, well, you-know-what.
Surely testing for antibodies requires that the patient has been infected for at least 7 days and is unwell? It doesn’t catch people who are infected with mild symptoms as there’s no requirement for a significant numbers of antibodies as the innate immune system is handling the infection. Which means that antibody numbers will reduce once the infection is cleared.
A reference list related to which issue?
Unfortunately, there are about 450,000 papers on COVID. A bibliography would be an overwhelming project.