In re: Hurricane Maria, what do we all know a bit over 5 years later?
Principally, don’t shoot from the hip when you don’t know statistics, and might’t distinguish survey information from administrative information, preliminary from revised. Think about the next graph.
Determine 1: Cumulative extra deaths from September 2017, for easy time dummies OLS mannequin (blue), OLS mannequin adjusting for inhabitants (inexperienced), and Quantile Regression mannequin adjusting for inhabitants (pink) all based mostly on methodology on this put up, Milken Institute level estimate (black sq.) and 95% confidence interval (grey +), Santos-Lozada, Howard letter (chartreuse triangle), Cruz-Cano and Mead (pink squares), Kopits “agency” most (teal triangle). Mild inexperienced band: Kopits estimate of 300-400 for October 2018. Supply: writer’s calculations, Milken Institute (2018), Santos-Lozada and Howard (2018), Cruz-Cano and Mead (2019), and Kopits (2018a), Kopits (2018b).
From Sandberg, et al. (July 2019) in Epidemiology:
Mr. Kopits’ ire was centered on the Kishore research; whereas the purpose estimate for end-2018 (4545) is increased than the others, the boldness interval encompasses the decrease estimates. Mr. Kopits’ advert hoc vary of 5/31 encompasses no different. (It have to be admitted that Mr. Kopits didn’t on the time perceive what a confidence interval was).
