Google is involved in the conspiracy against him?? That's concerning. Or, I get the feeling it wasn't sensored, more like it was going through the correct steps before finalised.. (Submitted: 13 May 2020 – Revised version received: 13 September 2020 – Accepted: 15 September 2020 – Published online: 14 October 2020) ' Also, your comment "when people started to ask questions about why the entire world was in lock down for a virus that has a lower death rate than the flu.". If they knew, what's the point of his paper?
I had a read of the paper, again, not worth my effort let alone to read his sources. I have issues with the paper, for sake of you disregarding my comments below, one of my majors in one of my degrees was economic analysis, specifically focused on modelling (econometrics was the bane of my uni experience, though I do miss those days playing around with models and data). I have no doubt you'll shrug off my comments, as I don't provide a fix, but frankly I have issues with his initial objective and methodology. That being said, I'm not sure how this could be improved, but I'm not the one paid to so I won't spend much time on it.
That being said I have issues with his modelling, the way he wrote this paper, and most importantly his affinity for un-evidenced assumptions "medium IFR might even be substaintially lower than the 0.23% observed in my analysis". Straight away in the first paragraph of the paper he suggests that non-pharmaceutical interventions are unneeded, or at least not worth the cost. Which, he should stick to the objective of the paper and estimate infection-to-fatality (IFR), not comment on what he deems to be unnecessary.
His model is the following: IFR = cumulative death count / (population * seroprevalence). Death count is questionably accurate (as posters here like to point out) Seroprevalence is only an estimate, and a very rough one at that.
My main issues are as follows: - Seroprevalence is completely disregarding migration bias. It's essentially pegging a percentage figure onto an estimate and extrapolating it. The flaw can be demonstrated where using an example with invented figures. You have the total infection count of 10,000 with 100 deaths in Melbourne, you now have 5,000 seropositive individuals shifting around the country to Melbourne. You've just increased the seroprevalence by 50% and almost halved the IFR.
- Ignoring inherent bias of seroprevalence studies. You'll end up with healthy people more willing to be tested for seropositivity, those infected will be picked up through hospital reporting, but there is inherent risk in under or over reporting those who've recovered, or opt not to be tested.
- At the foundation the IFR is stating that if a person randomly selected is infected, what is the probability they die? It doesn't take into account the severity of cases, nor impacting factors like health of the population, vulnerability of population, or mitigating factors (population density, access to health care).
I'll end it there, I've wasted another hour on this. I have no issue with the author, and I respect his attempt at coming up with an accurate figure, but I have too many issues and flaws in the study that leave be unable to treat the results as conclusive.
