Thread by @JohnFPfaff

🚨 The 2018 Uniform Crime Reports are finally up🚨

1. In what will surely be a blow to the Trump admin’s messaging, the 2018 UCR is out and… except for rape, all violent crime is down:

Murder: -6%
Robbery: -12%
Agg Assault: -0.3%

Rape: +3%

ucr.fbi.gov/crime-in-the-u…

2. All property crimes are down as well, but they have basically been falling all along.

Burglary: -12%
Larceny/theft: -6%
Auto theft: -3%

3. Obviously the biggest immediate question is how the rape data in the UCR compares to the NCVS, which showed that huge jump in 2018.

UCR: rape is +3%.

In NCVS, *reported* sexual assault (a BROADER definition) is +16%

4. For the NCVS definition that comes closer to the UCR’s (“serious violence”), there is no comparable rate of reporting, but if we use the same rate as that for the broader definition, then the increase here is +10%.

So NCVS is moving diff than the UCR for rape. Don’t know why.

5. As is always the case, our clearance rates remain pretty low, with no real perceptible change.

Always important to note: what really deters is detection, not punishment. Low clearances, esp. for serious crimes, are serious failings.

6. Worth noting that it looks like, in general, non-rape violence fell across all pop sizes (major to minor cities, metro to non-metro areas); and rape generally rose in all.

(Tho some patterns may emerge w more inspection.)

ucr.fbi.gov/crime-in-the-u…

ucr.fbi.gov/crime-in-the-u…

7. Total arrests are down -2%.

Total index violent arrests are up…ish, at +0.4% (which = ~0% given all the noise).

Total index property arrests are -7%.

Total non-index assaults held steady.

Total drug arrests are UP, but by +1.3% (which could also be noise).

8. Not much changed with those drug arrests. From 2017-2018:

Share for traffic/poss went from 15/85 to 14/86 (likely noise).

Marijuana’s share went from 40.4 to 40.1 (surely noise).

So… not much change for arrests when it comes to drugs, along any margin.

9. Now to head off an annual canard.

There were NOT more marijuana arrests than arrests for violence.

There WERE, as in previous years, more MJ arrests than for INDEX violence.

MJ arrests: ~ 663K
Index viol arrests: ~521K
Non-index assault arrests: ~1.06M

(That’s still bad.)

10. Back to reported crimes, robbery and agg assault #s in the UCR moved differently than the NCVS’s as well.

While robbery events fell in the NCVS, reporting went up, so REPORTED robbery (what the UCR measures) was +18%

NCVS reported agg assault was +12% (with incidents +7%).

11. Now, to be clear, NCVS and UCR are two VERY DIFFERENT ways of measuring crime, and they’ve often moved in different directions (in fact, while the UCR rose over the 1960s and 1970s for violence, the NCVS consistently fell, tho they’ve tracked much better in 1990s-now).

12. NCVS is a survey of people, UCR is an (approximate) counting of police-recorded events.

Given those differences, they may reach diff pops in imp ways: the NCVS, for ex, may have a hard time reaching pops that are wary of surveys—which may also be more heavily policed.

13. The goal, then, is not to figure out which one is “right” or “better,” but to appreciate that they are indirect and complementary efforts to track a common hard-to-measure concept.

My gut is divergence like this means: be cautious with either “trend” (NCVS up, UCR down).

14. A few points to highlight about homicides:

As always, our homicide rate—low by our standards, still about 5x higher than in the EU—is a guns issue. Almost 3/4 of all homicides involved guns.

15. The offender-victim race/sex/age table is currently a blank; it’s a key one for dispelling the “black-on-black homicide problem” by showing that most homicides are intra-racial.

Am sure we’ll see similar results in 2018, but annoying it’s blank right now.

16. Also worth noting, nationwide, a majority of people know their killer. Of the 14,123 homicides w data (out of 16,214), only 1,392 (10%) involved a stranger, tho 6,992 (50%) were “unknown.”

Unless ~100% of the unknowns are “strangers,” then majority of victims knew killer.

17. I’m sure more interesting details will emerge as people look over this closely, but my immediately takeaway is that there’s not much to see here…

… which may, really, be the story itself, in complicated ways.

In the midst of all these reform efforts… not much changed.

18. OT1H, that’s good for reform: despite all these changes, crime hasn’t changed much, and in fact is down-ish, sorta.

OTOH, the arrest data hasn’t changed either. So a decade in, and the front door to the system looks… about the same.

19. Given how noisy UCR data is—that it reports crimes out to the ones places is basically statistical malpractice—I wonder what is gained arguing about the changes all that much (tho those rape numbers do demand some).

It may be about the STABILITY, and what THAT means.

20. Okay, that’s enough for now. Looking forward to seeing what other people dig up in this tomorrow.