Hacker News Reader

Comment by ZeroGravitas | Hacker News Reader

I feel I'm missing some paranoid back story here, but the Asians, Native Americans, Hawaiian's and others are included in the "other" category.

Good to see crime continuing its long decline though.

Looking up the report the table is a summary of lists the Asian group out seperatly.

Seems like the numbers for "Asian" went down and "other" went up so they combined them for the summary.

https://bjs.ojp.gov/content/pub/pdf/cv19.pdf

Looks like it declined again in the more recent report

https://bjs.ojp.gov/library/publications/criminal-victimizat...

edit: worth also mentioning, this is crime, not just "hate-crime" data.

View on HN

I don't see in the 2020 report the same chart, breaking down race for victim and offender. Am I missing it?

It's there (table 5); it's just formatted differently. Presumably the original poster decided not to include it, as it added the footnote "Victims of other races are not shown separately due to small numbers of sample cases", which doesn't fit his framing.

What document are you talking about here? In the "full report" [1] table 5 is "Rate of victimization reported and not reported to police, by type of crime, 2019–2020", and has nothing about race.

This document doesn't contain any equivalent of the cross-tabulations contained in the tweet, as far as i can tell.

[1] https://bjs.ojp.gov/sites/g/files/xyckuh236/files/media/docu...

The report's been split into two docs. The referenced table is in the Supplemental Statistical Tables by the same authors: https://bjs.ojp.gov/content/pub/pdf/cv20sst.pdf

At least in the screenshots posted, "other" was only an option for offenders, not victims.

Yeah, it's a summary and “Asian, Native American, Native Alaskan, 2 or more races, Pacific islander, etc." didn't have very many samples so they grouped them.

There's more detail in the reports I linked, both offender and victim.

Notably one of the items has this warning note:

"! Interpret with caution. Estimate is based on 10 or fewer sample cases, or coefficient of variation is greater than 50%."

As this is survey data so like polling they have to multiply responses to get the final answer and once you start subdividing by victim demographic and crime type and offender demographic, you're hitting the limits of what useful info you can extract without it just being random noise.

Reply on news.ycombinator.com