🇳🇱
@yaksha
Headlines like this are so dishonest because they rely on people not understanding statistics.
If everyone gets vaccinated, and the vaccination is 99.9999% effective, you would be telling the truth to say "100% of all new covid cases are vaccinated people"... but despite being true it is highly manipulative since people dont understand that it isnt really saying anything of value, its just meant to sound scary and say nothing.
@freemo @yaksha @wa__em Interestingly, this still paints a pretty poor picture of vaccine efficacy though…
61% vaccinated, 57% fully vaccinated, so vaccinated people are only slightly under-represented in the cases. Obviously there are all sorts of confounding variables (like people who estimate their risk of covid exposure being more likely to get the vaccine, etc) or important things beyond infection like disease severity, but at a glance, doesn’t seem great…
Source: https://www.nytimes.com/interactive/2021/world/covid-vaccinations-tracker.html
the devil is in the details.. you cant draw any meaningful conclusions by just looking at national vaccination rates. You'd have to do a proper study on vaccine effectiveness, which ultimately shows relatively high rates, even on the delta variant (which it has 88% - 94% efficacity on depending on the vaccine you get).
@wa__em @yaksha
@freemo @wa__em @yaksha I’m glad you posted this before I shared my calculation based on those numbers heh. I did skim an efficacy paper and I thought that was precisely what you did though, you look at your case population and compare it to the wider population to calculate to what extent a population is over/under represented in the case population, and from that you get a relative risk?
odds_ratio_vaccinated = (vaccinated_cases / vaccinated_pop) odds_ratio_unvaccinated = (unvaccinated_cases / unvaccinated_pop)
relative_risk = or_vaccinated / or_unvaccinated
efficacy = (1 - relative_risk) * 100
?
I only ever took the one undergrad stats course, was mediocre at it, and promptly forgot everything, so I’m happy to be corrected.
Your logic isnt too far off, the devil though is in the details.
For example, whats your sample size? How were the sampled from the population? Was it double blind randomly checking members from the population or inferred by just those who show up in a hospital? What about the people who got sick but were so minor they never went to the hospital? etc.
The simple numbers we can run on the back of a napkin is fine and all, but it isnt a replacement for a properly run study.
@wa__em @yaksha
@freemo @ademan @wa__em @yaksha
"Risk isn't just about the probabilities, it's also about the consequences..." https://youtu.be/YjvRRy-rm6E?t=1000