182. Numbers racket

ABC News' website ran a story on Thursday headlined, "Startling New Stats Show Cross-Country Crime Spike."  Now, that's a good headline.  The alliteration is satisfying, and beginning with "startling" and ending with "spike" produces a pleasing symmetry, not just of sibilants but of bookended promises of drama.  And "cross-country crime," with its echo of that evergreen the cross-country crime spree, adds a subtle overtone of menace.

Anyway, here's one of the startling spike stats: "Murder was up in 26 of 53 cities".  Now, I agree that rising murder rates anywhere are worrisome.  But ABC's stat tells us that murder rates stayed the same or dropped in 27 cities.  If I were to tell you that 26 cities saw an increase in the number of Italian restaurants, while in 27 cities the number of Italian restaurants stayed the same or dropped, what would you conclude about the nationwide trend in Italian restaurants?

Crime statistics are always subject to intentional massaging, as the New York Times reported recently in a story headlined: "A Very Violent School, or Just Very Honest?", reporting that Rome Free Academy in upstate New York was listed as one of the 10 most dangerous schools in Gotham's state, which is just laughable.   Apparently, only one principal in the state failed to get the memo about reporting violent incidents.

A teacher at a notoriously violent high school in Albuquerque once told me the administration had established a firm policy that teachers were not to  break up fights.  I used to assume that policy was intended to avoid injury to the teachers, but I now suspect it also serves the purpose of permitting the administration to pretend it was unaware of the incident when compiling statistics.

But given a couple grains of salt, changes in crime statistics tell us a lot.  The problem is that it's not always easy to be sure what, exactly.  Arguments about whether this or that initiative "works" or not look at single variables, and of course there's no single variable that explains more than part of the story.

The nationwide drop in murder rates during the 1990s and continuing for the first few years of the new century was certainly welcome, but it was just a drop from a hideous high.  Washington had 81 homicides in 1960 and 482 in 1991, even as the city's population dropped by about 165,000 - and despite concurrent advances in emergency response services and medical care that vastly reduced the "lethality rate" - the percentage of shooting and stab victims who die from their wounds.  2000's number of homicides (241) was just half the 1991 figure - but it was also three times the 1960 figure.

One obvious reason why the murder rate dropped in the years following the peak - no more than part of the story, but a part I haven't seen emphasized anywhere - is that by the late 1990s, so many dangerous people were dead.  At any given time, in any given population, there is only a small number of people - or, rather, teenaged boys and young men - who are prepared to kill, or to participate in the kill-or-be-killed gang culture.  A city can't lose several thousands of them in just a few years without some effect.

If there's any validity to that partial explanation, you'd expect the murder rate to start creeping up again as a new generation of dangerous people grows into - well, not into maturity, exactly.  Interestingly, the District itself doesn't show an uptick, according to these statistics, but Maryland's Prince George County, just across the border, certainly has.  Last year its number of recorded homicides topped the all-time record set in 1991.

Comments

[michael]

I think you also need to take into account the factor discussed in the book Freakanomics - by the mid to late 1990's, the first generation of children of poor single women who would have been born before Roe v. Wade were not born but aborted. Right when this generation was coming of age - a large portion of kids who would have been in the crime prone segment simply never existed.