From the end of World War II until the mid-1970s, the proportion of Americans in prison each year never exceeded 120 per 100,000
That's a funny way of saying 0.12%. Is there a reason for this? It sure doesn't make it easy to compare the numbers they're giving with other numbers given as percentages.
I guess if you're considering a sufficiently small population you could go from ~600,000 people in Vermont * 120/100,000 -> ~720 imprisoned people in Vermont trivially, but we're the second smallest state. This certainly doesn't scale to cities over a million. At least I'd start having to think harder about it.
It also lets you abstract away or compare to stats that are scaled to population but might not be 1:1 with a person, e.g. "thefts per 100,000 population per year" where one person might either commit or be the victim of multiple thefts in a year.
120 per 100,000 includes significant digits. 0.12% could be anywhere from 120-124 per 100,000. You'd really want 0.120%, but that's confusing for different reasons.
Worse would be 1,000 per 100,000, which is 1% but there's no way to tell that it's not rounded or truncated.
I would presume, perhaps incorrectly, that “120 per 100,000” has 3 significant digits and “12 per 10,000” has 2.
I’ve never seen a period used like that in census data. It seems like a conscious choice because the period is confusing when used in the middle of a phrase. 12E1 makes more sense but is abnormal notation for many people.
I guess if you're considering a sufficiently small population you could go from ~600,000 people in Vermont * 120/100,000 -> ~720 imprisoned people in Vermont trivially, but we're the second smallest state. This certainly doesn't scale to cities over a million. At least I'd start having to think harder about it.