Baltimore's post about the Ab false positive actually helps solve a problem I was puzzling over. In the German study, 2% of the subjects were positive with the virus, but 14% with the antibody. Germany, like Austria, has close to an equal number of active cases and recovered, so the 2% figure can be doubled to 4%. This is the fraction of the population (in that town) who have tested positive and shown symptoms.
But that still indicates that asymptomatics are about 70% of the infected people: (14-4)/14. If we asssume 4.4% of the population are false positives, though, the true proportion of Ab positives is 9.6%. The proportion of asymptomatics drops to 5.6/9.6 = 58%. This is much closer to the 50% figure emerging from other data, such as Iceland.
However, this number would have to be revised upwards somewhat to account for the recent negatives that haven't yet developed Abs. For example, if half of the currently negative people haven't developed Abs. then the proportion of asymptomatics would be 6.6/10.6 = 62%. In fact, this may be the assumption, because according to one report, the 14% became 15% by "eliminating overlap between the two groups", IOW, they assumed that !% of the population tested negative but did not have Abs yet. I'd have thought that number was lower, but anyway, probably 60% is about as high a proportion of asymptomatics as this study would imply.
This makes me wonder how a mortality rate of 0.37% was calculated. Did they take into account some false positive rate? I assume they would, but if they did, the 14% or 15% rate reported would be inaccurate and misleading--unless the actual rate was higher than that, and they got that after subtracting false positives. But I haven't been able to find any information on this.
Edit: Yes, it turns out that some experts have questioned the sensitivity (false negatives) and selectivity (false positives) of the Ab test used in Germany. This suggests to me that they didn't make any correction for false positives. I don't even know what that 4.4% figure I used is based on. I went back to Levi's original tweet, and he throws it out as though it were some fixed constant. It depends on the Ab, and probably, the person doing the assay. Maybe he got it from the German group, but then again, why didn't the group point that out when they were reporting to the media?
https://www.sciencemediacentre.org/...v-2-virus-in-residents-of-gangelt-in-germany/
I also wonder what they used as the mortality rate prior to taking into account the Ab information. According to a story in German, they used a Johns Hopkins estimate of 1.98%. But that is very worrisome, because it's very close to the naive mortality rate (deaths/cases), which is known to be an underestimate. For April 7, that rate was 1.87%; on April 8 it was 2.07%, so they do seem to be using that. But then how do they go from there to 0.37%?