BuzzH
Well-known member
I've created, seen, and analyzed a lot of practical data. You can mess with %'s to make them say nearly anything you want. Leave out the sample size, C.I., and model assumptions and it's very nearly a useless number. Sample size alone doesn't fix scope of inference, confounding variable, and model assumption problems.
I'm very much a data guy, but in this case I feel the "art" of applying medical science and first hand accounts need to carry more weight than normal.
The cost:benefit of the current closures is a completely separate matter that's theoretically informed by data, but it's more like a broken link at present. It seems to me like that link will continue to be broken until testing capacity is vastly increased.
I just spent the last 6-8 weeks writing a report on sample size, confidence interval, sampling error, standard error, data collection errors, etc. on data we collect and you're exactly right.
Everyone should read SnowyMountaineer's first paragraph a couple times...its a great summary to the complexity of data analysis and how we model data.
Its not easy to draw absolute conclusions, even from very good data sometimes, even with large sample sizes, etc....seen it.