The Rare Norm last week Share Tweet Pin Share On the other hand, it is quite possible that the currently reported death count is still lower than actuality without knowing the inclusion criteria that Puerto Rican officials used to assess if an individual died from the hurricane. Note that studies such as the one mentioned above are often conservative in nature, due to a variety of biases and other methodological quirks. They essentially conducted a poll of a small sample of households It’s a representative random sampling of more than 3,000 households aimed at reaching previously determined power and significance thresholds, and using post-hoc analyses making sure that the study population had similar demographic characteristics with census information. The sample size is hardly small, and most likely took a significant amount of time. Picking a number right in the middle of your giant range isn’t what I would call precise calculations. The authors were precise; they gave the 95% CI because that is the norm in medicine to provide statistical context towards the strength of their estimate. If they doubled the sample size, the rage of the CI would shrink, but this does not in any way invalidate their findings. This study was published in NEJM, the best medical/epidemiological journal in the world. Studies such as these are “imprecise” not due to the wide range provided by the 95% CI, but due to their assumptions. Reading over their protocol and methodology, I actually don’t see any assumptions that are incorrect to make the study imprecise. I would be interested to hear what you and other Redditors believe is imprecise about their methodology or protocol — to me it seems like a fairly straightforward and thorough study, but I may have missed something. We like to laugh about “polls” in this political climate, but any study published in the NEJM deserves some attention. There are few journals as prestigious, difficult to publish in, more widely read, and have more of an impact than the NEJM.