Thread by @JohnFPfaff

So I deleted my earlier tweet abt shootings being down in the NCVS. I misread the chart, which was about ANY gun use (including displaying it), not just shooting.

Those are, obviously, two different things, and addressing it downthread seemed insufficient.

Still, some thoughts:

That gun-involved violence falls in the NCVS is still somewhat surprising, since total gun purchases appear to be way up.

Could reflect something about the NCVS, but it could also tell us something about how those guns were being used that was otherwise not immediately obvious.

There's still some intriguing back of the envelope math:

UCR says ~40% of robberies involve a gun.
NCVS20 has a ~97K drop in robberies.
If 40% holds, that's ~40K fewer armed robberies.
Total gun drop for NCVS20 is 132K.

Still ~80K fewer gun events outside robbery.

It's true that ShotSpotter data suggests a sharp rise in shootings.

Which again: could point to issues with the NCVS, but it isn't like ShotSpotter isn't without ITS concerns too: vice.com/en/article/qj8…

All our data is... messy. On a good day.

It could also be that the police shooting data (which we don't have from a lot of places) and the NCVS are completely consistent: fewer people shot at, more shooting in general (perhaps even shooting at ppl who didn't realize it, assumed someone else was the target).

But it's not just something we can ignore. The NCVS and UCR disagreed this yr, on what is perhaps the second-most salient offense (agg assault), and closely-related gun use.

It's something that demands consideration, even if the implications are genuinely unclear right now.