...I understand what per-100,000 means, but I also understand that not all groups of 100,000 people are the same; removing a large sub-population of people that doesn’t exactly match the overall population’s average will result in a change to the overall population average.
If you have a total population (T), and you are measuring the rate of an event (E), then E / T gives your average event rate for the total population, which you can then normalize to a per-X number. For example: T = 1000 people E = 10 incarcerations. 10 / 1000 = .01, normalized to per 100 capita would be 1 per 100 people on average, from the total population.
If you have a sub-demographic in that population (Ts), and it has a different rate of an event (Es) then its rate is also Es / Ts. For example: Ts = 100 poor-people Es = 5 incarcerations. 5 / 100 = .05, normalized to a per 100 capita would be 5 people per 100 on average, for that sub-population.
If you suddenly remove that sub-population, what happens to the rate of the overall population? That’s easy to calculate: (E - Es) / (T - Ts) (10 - 5) / (1000 - 100) = 5 / 900 = .0055, normalized to a per 100 capita would be .55.
Suggesting that a sub-demographic doesn’t perfectly match the per-capita average of an entire population and that removing them would change the overall per-capita rate isn’t nonsense.
To be fair, China is 2nd in overall prison population by country globally, so it's not like their numbers are complete bs. I'm sure there is some fudging in what constitutes as a "prisoner" when they ~~have~~ had "re-education camps" though. That said, the US's numbers are fucking insane.