Thursday, September 07, 2017

Year-to-date crime is down, but should reformers tout it?

A new analysis finds that year-to-date murder rates are down for 2017 - see a report from the Brennan Center and coverage from Vox. According to the think tank:
The 2017 murder rate is projected to be 2.5 percent lower than last year. This year’s decline is driven primarily by decreases in Detroit (down 25.6 percent), Houston (down 20.5 percent), and New York (down 19.1 percent). Chicago’s murder rate is also projected to fall, by 2.4 percent. The 2017 murder rate is expected to be on par with that of 2009, well at the bottom of the historic post-1990 decline, yet still higher than the lowest recorded rate in 2013.
The news will be widely heralded as a correction to slight upticks in violent crime in 2015 and 2016, casting doubt on the notion that the last couple of years represent some new wave of violence that justifies references to "American carnage," etc.. So perhaps Brennan's analysis will provide an antidote to those misguided memes.

However, crime data such as these can only be meaningfully interpreted over the long term; short term blips tend not to tell us much, whether crime is going up or down. Grits feels the same way about Brennan's analysis as I did about a similarly themed Houston Chronicle story in July, which inspired a blog post titled: "Of journalists, drunks, lamp posts, and Year-To-Date crime data." Read that for more on my complaints about YTD crime data.

Bottom line: Criminologists have shown what veteran crime reporters have known for years: that such short-term analyses of crime data don't tell us very much and are often wrong. They remain popular, though, because daily media outlets have news holes to fill, August is a slow news month, and crime reporting remains a staple. It may not be probative data, but it's new. And new fulfills a certain need for content producers in both traditional and digital media.

In a sense, I'm glad Brennan has taken on this project. Doing it every year provides some uniformity to the analyses over time, whereas otherwise the data are framed by whichever reporter from some random news outlet thinks to do the story first. But I still question the overall value of year-to-date crime data analyses, and worry that Brennan's project improperly validates their use.

Reformers should take opportunities like this one to critique misguided uses of crime data during periods when the data "help" us - as when crime is declining - because bringing it up for the first time when crime rises will raise understandable skepticism as to motive.

Grits has been at this for quite a while and, in my experience, error helps nobody in the long run. The benefits that come from relying on bad data which appear to "help" my side are far outweighed by the cost of relying on it when it runs the other direction. Crime data are used by demagogues to inspire panic, and insisting on their proper, in-perspective use, whether that helps or hurts one's "cause," in the long run will improve the culture surrounding these debates and lead to better outcomes.

Critiquing the misuse of data only when it damages your "side" reduces one's credibility. Criminal-justice reformers would be wise not to fall into that trap.

RELATED: From Slate/Fair Punishment.

No comments: