Hacker News new | ask | show | jobs
by Bud 1630 days ago
I suggest reading up on Simpson's paradox, which keeps showing up in these kinds of data evaluations regarding covid.

https://en.wikipedia.org/wiki/Simpson%27s_paradox

1 comments

In this case it's not relevant, because it couldn't explain these similar magnitude trend lines adding up to something so tremendously different.

The effect is caused by the age data ending in mid-October, before the spike we're looking at in mid-December. So, the age-segregated data couldn't show that spike at all.