Maybe I am wrong but shouldn't you normalize (in a Bayesian way) to the percentage of vaccinated people in the sample? I.e. we should look at the probability of

Maybe I am wrong but shouldn't you normalize (in a Bayesian way) to the percentage of vaccinated people in the sample? I.e. we should look at the probability of contracting serious COVID for a vaccinated person vs a unvaccinated one. Since 60-plus people are very likely to be vaccinated the effect of the vaccine could be much lower than reported. It's the textbook case of confusing conditional probabilities and while physicians surely know what they do it is simple for published data to be in a form that is misleading for the average reader. EDIT: My bad, the data is already this way BUT if you look at https://datadashboard.health.gov.il/COVID-19/general https://datadashboard.health.gov.il/COVID-19/general (Hebrew) you see how the effect of the vaccine has decreased Probability of contracting COVID for an unvaccinated 60+ y.o. was 15x than a vaccinated one in February, now it's more like 3x. Probability of getting a serious case for an unvaccinated 60+ y.o. was 30x than a vaccinated one in February, now it's more like 8x.