Ed Bassin- Ph.D. Sociology & Statistics Former Statistician with over 25 years in healthcare
Some things to consider:
The solution to items 1–3 is to begin random sampling of the general population on an ongoing basis. We don’t need huge numbers of test cases, but we do need to get an accurate count of the number of people infected and the mortality rate. Also, random sampling eliminates reporting and test availability issues. If one’s goal is to scare people with numbers, the current approach works very well. If one wants to get to the truth, well, things have to change.
- We have no idea how many people have been infected with the disease. We look at the number of confirmed cases, but that doesn’t show how many people had mild disease and were advised not to “waste” a test or were asymptomatic. Recent antibody tests suggest our counts are off by at least 2000%.
- Without an accurate count of how many people have been infected, we have no idea what the true mortality rate is of the disease. Our leaders are spewing figures with decimal point precision, creating the false impression that we know. In all likelihood, the numbers they use vastly overestimate how lethal the virus is.
- Our stats are confounded by test availability and reporting. Recently, Massachusetts reported a spike in the number of new cases after a few days of decline. It turned out that a large commercial lab company had a backlog of cases. Oops. There may never have been a decline and there surely was no spike.
- We are counting all who die with Covid-19 infection as being caused by Covid-19. Recently where I live, 2 people in hospice care (for other reasons) were infected and died. Were those deaths from Covid-19, another cause, or multiple causes?
- News reports highlight deaths among young people, but the age breakdown shows that Covid-19 mostly affects people over the age of 70. We don’t show the age breakdowns often enough.
P.S. I expect that some reading this answer will say something to the effect that “random sampling didn’t predict that Trump would be elected?” The results were within the 95% confidence interval, aka, the “margin of error”. They were hardly off by 2000%!