You see the what, but not the when

[brin_bellway]

[cw: illness]


Saw an interesting graph recently, courtesy of Tomas Pueyo. Yellow bars show Hubei's COVID-19 cases by date of *diagnosis*; blue bars show cases by date of *when those later-diagnosed people first showed symptoms*.

Comments

[yvannairie]

*puts hand up* I'm bad at statistics -- why is this interesting?

[brin_bellway]

It visually shows how important it is not to assume that the curve of *official* new cases represents the curve of *true* new cases: by the time you *find out* that things are bad, they have been bad for some time. Particularly notable are days like January 23rd, where the number of people who *became symptomatic* that day--the *true* new cases, more or less--was more than *five times* the number of *official* new cases (who are people who caught it a while ago and are now coming to attention).

(In this particular case, thanks to actions taken when things *appeared* to be relatively not that bad, reality was *already on the mend* by the time the official stats really caught on to how bad things were. But the first paragraph is what interests me most.)

[yvannairie]

Thank you ^-^)