3 ms·
In general, only if you can be very specific about exactly how underestimated they are, which is equivalent to the problem of knowing the correct values in the
by jerf 7y ago
In general, only if you can be very specific about exactly how underestimated they are, which is equivalent to the problem of knowing the correct values in the first place, so, basically, no.
(Sometimes statistics can help you out, but I think that's best understood as a set of fairly specific exceptions with very strong conditions on them, rather than the rule.)
- bordercases 7y agoHere's a paper that explores where underestimates may exist: https://www.medrxiv.org/content/10.1101/2020.02.04.20020495v1.article-metrics https://www.medrxiv.org/content/10.1101/2020.02.04.20020495v...