I'm surprised at that. Way back at first someone offered that a 200K sample would be sufficient to predict 360M. Well, we now have 3 mil tested, or 1 in 360. So shouldn't we be able to predict infection rate from the results of 3M?
“I’m surprised at that. Way back at first someone offered that a 200K sample would be sufficient to predict 360M. Well, we now have 3 mil tested, or 1 in 360. So shouldn’t we be able to predict infection rate from the results of 3M?”
Not really. It is multiple populations and many are tainted samples. For example, in many places you don’t get tested without symptoms, in others they track down everybody your grandmother ever spoke to since 1963.
All we can really confirm is that not “everybody has already had it” as the tainted samples run 10%+/- or so positives (a week ago, statistic) but are heavily weighted toward active areas. Looks like the numbers are bigger now, though. In the hot spots they are behind in testing and everywhere else is low volume.
If you sampled 10,000 people totally at random you would have a sample more accurate than your typical election poll. 200,000 would probably nail it to well less than 1% margin of error (not doing the math right now, though).