Article in the Observer from two well-know statisticians, Sylvia Richardson and David Spiegelhalter, summarizing the trickiness of interpreting the data on covid because of all of the variation in testing and reporting, amongst other complications. On big source of variabilities in covid deaths is that some countries are reporting presumed covid deaths if the patient died of what looked like covid-related lung disease or organ failure while other countries only report as covid death if there was a confirmed positive test, which would undercount given the known undertesting problem.
