Comment by gpderetta
1 year ago
Yes, I did a bit of investigation and I commented on it the few times this article made the rounds on HN: https://news.ycombinator.com/item?id=20633769
I don't think it is just bad statistics, it is very poor data extractions.
Just an example:
"Like the ‘blue zone’ islands of Sardinia and Ikaria, Okinawa also represents the shortest-lived and second-poorest region of a rich high-welfare state"
Sardinia[1], at 83.8, had in 2018 one of the EU highest life expectancies, certainly higher than the rest of Italy (83.4). Like the rest of Italy it was badly hit by COVID in 2020. Life expectancy at 55 is 30.6 vs 30.1 for the rest of Italy. I don't know how to match it with their Figure 2 that shows the all Sardinian provinces being extreme outliers in negative other than they completely misinterpreted the data. Also the same graph shows 7 blue dots for Sardinian provinces, historically Sardinia had only 4 provinces and has had 8 only for a short period in the mid 2000s.
[edit: The newer version of the paper[2] is different and doesn't have figure 2]
[1] https://ec.europa.eu/eurostat/databrowser/view/demo_r_mlifex... (Sardegna In the table).
Figure 2 is now Figures S2 and S3 in the newer paper. Table S1 is also relevant: all four Sardinian provinces that appear in that table have existed only from 2005 to 2016. The other 4 historical provinces do not appear. I can't help but think that they didn't somehow account for that and it messed up their data.
Although the fact that those four provinces stick out as extreme outliers in their graph should have clued them that something was wrong.