Xan wrote: ↑Sun Aug 08, 2021 8:48 pm
KayFaybe, your math is solid. There isn't a "Bayesian" or "Monte Carlo" improvement to be made. I would say it's likely to be the upper bound of the actual number, because Covid deaths aren't distributed evenly through the population, which is an underlying assumption of your analysis.
Thank you, Xan.
Yes, I recognize that was an assumption of my analysis.
I also disagree with those (I think Some Dude and janalong) who wrote that 16% is not in the same ballpark as 29%. I think if you consider that I used very minimal information to come up with my number, 16% is actually quite accurate.
I don't think the others are (for the most part) challenging the math so much as other things about the number. jalanlong is saying that the question itself was designed to come up with a high number, just to generate a high number and cause fear. That may be so. The definition of "family or close friend" is definitely hard to pin down; the number only appears to be reasonable if just about EVERYBODY that a person knows is counted as family or close friend.
It does seem that it might be difficult to pin down without knowing exactly how the question was worded. The link you provided indicated that it was on-line survey, so it would probably difficult to coach the person on exactly how to restrict "people I know".
So what's going on, then? Is everyone, including a large proportion of Trump voters, inflating their "know people who died of Covid" status, and if so, why?
This is a great question. It was 26% of Trump voters vs. 29% overall. These are statistically identical.
SomeDude wrote: ↑Sun Aug 08, 2021 11:06 pm
US pop = approx 330M
29% of US pop = approx 95M
Alleged covid deaths = approx 600k (includes car accident victims, people falling off ladders, term ill from cancer and heart disease and basically anyone who died who tested positive for at least the first couple months of the ordeal).
95M divided by 600k = approx 160 people that on average each covid victim would need as a close friend or relative.
Yours is a fine estimate, though it assumes a person can only know 1 person who died from COVID, so your estimate is low. I would estimate that the number of "close friends / relatives" needed is:
N = ln(1 - 0.29)/ln(1 - death rate) = ln(0.71)/ln(0.9982)
N = 190
Your extremely quick method is reasonably close because the death rate is so small.