Wow, yes, we completely disagree. Learning about the prevalence of cases in regional & demographic subpopulations will help us find more true positives. Assuming testing ramps up over time, the tests later are better deployed with knowledge of where to deploy them.
-
-
Replying to @sinanaral @GidMK
Testing convenience samples helps those convenience samples who get treatment, but not the millions who could learn from a statistically valid sample. I'm not saying test randomly. I'm saying test smartly so we learn something.
1 reply 0 retweets 0 likes -
Replying to @sinanaral
But we learn a lot! As I said, we know enormous amounts about the disease, we just don't have a ~perfect~ sample and the certainty that comes from it. The issue with Ioannidis is that he's essentially arguing that without perfection, we know nothing
1 reply 0 retweets 0 likes -
Replying to @GidMK
Have you seen the confidence intervals?? Our estimates of mobility and mortality, transmission and prevalence are so wide we know very little. I don't think we need a perfect sample, but a little more precision is warranted.
1 reply 0 retweets 1 like -
Replying to @sinanaral
CIs are mostly a function of sample size, not sampling accuracy. I do agree we don't know a lot, but I would strongly disagree that we know very little. We have a very good idea of the R0, confident estimates of hospitalization rate, and pretty decent predictions of morbidity
1 reply 0 retweets 0 likes -
Replying to @GidMK
I pretty much disagree with everything in this tweet.
3 replies 0 retweets 6 likes -
Replying to @sinanaral
I never said "perfect" estimates, but the range of plausible R0s, death rates, etc has shrunk significantly. It's incredibly unlikely, for example, that the death rate is anywhere near as low as Ioannidis hypothesizes
1 reply 0 retweets 0 likes -
Replying to @GidMK
I'm sorry, I really don't mean to be semantic but you literally said "we just don't have a ~ perfect ~ sample"... That's the only reason I wrote that reply... I would never get into these words games unless the straw man was actually advanced as argument.
1 reply 0 retweets 0 likes -
Replying to @sinanaral
I'm not sure I understand. My point is that we have a reasonable understanding of what the morbidity/mortality is likely to be - it may not be perfect, but it is also not nothing
1 reply 0 retweets 0 likes -
Replying to @GidMK
We disagree on the uncertainty in those estimates of population morbidity/mortality, transmission & prevalence. Yes, it's not perfect & it's not nothing but as I said in my very first tweet it's "very little." The error bars are wide & they would be less so if we sampled smartly.
1 reply 0 retweets 0 likes
It's not very little at all. For example, I'd say from the data we have now that the CFRs for 60+ are consistent and likely to be within relatively small margins of error
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.