Read on Twitter
Data and statistics around the #coronavirus are riddled with problems and often create confusion. That& #39;s why I& #39;ve written a short paper explaining the most common pitfalls, published open access by @Intereconomics_: (1/n) https://www.intereconomics.eu/contents/year/2020/number/3/article/common-pitfalls-in-the-interpretation-of-covid-19-data-and-statistics.html">https://www.intereconomics.eu/contents/...
Pitfall 1: Measurement of lethality. Case fatality rate (CFR), infection fatality rate (IFR), and mortality rate (MR) (or death rate) are different concepts. They shouldn& #39;t be mixed up. Rule of thumb regarding their magnitude: CFR>IFR>MR. (2/n)
Pitfall 2: Comparability: Demographics play a huge role for the lethality of COVID-19, so check the age distribution behind the cases. Some readers will notice the similarity of this graph to the one in my @Medium story from March ( https://medium.com/@andreasbackhausab/coronavirus-why-its-so-deadly-in-italy-c4200a15a7bf).">https://medium.com/@andreasb... (3/n)
Pitfall 3: Death count. Are we over-/undercounting COVID-19 deaths? Look at the deviation of all deaths in 2020 from the previous average. How much excess mortality is explained by COVID-19? @jburnmurdoch, @OurWorldInData, @TheEconomist have graphs. (4/n) https://www.ft.com/content/6bd88b7d-3386-4543-b2e9-0d5c6fac846c">https://www.ft.com/content/6...
Pitfall 4: Reporting lags. The newest public data releases are often incomplete and subject to major revisions. Don’t rely on them for far-fetched conclusions. Statistics Sweden is very transparent about this: (5/n)
Pitfall 5: Sample selection bias. Think about why a person is included in a sample. Often, we see the sample is not representative of the population. We can still compute statistics, but they don’t generalize. @CT_Bergstrom has threads and courses: (6/n) https://callingbullshit.org/ ">https://callingbullshit.org/">...
Pitfall 6: Endogeneity of policy interventions. Cross-country comparisons of lockdown effects are biased. Countries that locked down hard also had many COVID-19 deaths early on. This determined both the lockdown policy and the later death toll. More: (7/n) https://p-hunermund.com/2018/10/15/sample-selection-vs-selection-into-treatment/">https://p-hunermund.com/2018/10/1...
Some of the great researchers whose works appear in the paper: @drjenndowd @c_dudel @HelenBranswell @AdamJKucharski @christianbaye13 @kuhnmo Many thanks also to @chris_breu @ZBW_news @Zeitschrift_WD @JifferBourg1 @CEPS_thinktank! (8/n)
