Frisks and Risks

May 13, 2010
Posted by Jay Livingston

In New York, a city of roughly 8 million people, the police stopped and frisked over half a million people* last year. (The Times story is here.)

“We are saving lives, and we are preventing crime.” said the NYPD spokesman. If you don’t believe him, just look at this pie chart of the yield of weapons that the searches turned up.

(Click on the graph for a larger view, but you still won’t be able to see that line for guns.)

Guns are the thin red line, so thin that it’s all but invisible – with 762 guns out of 570,000 stops, it’s hard to make it look like any thicker. (A post last year had a pie chart showing the proportion of stops that led to any official action, a larger slice of the pie than the slivers representing guns and other weapons.)

We have known the number of stop-and-frisks only since 2003.** In the next two years, the number doubled. In 2009 police made nearly three times as many stops as they had in 2003. Has this dramatic increase taken a bite out of crime? Let’s ask the experts.
Heather Mac Donald, a research fellow at the Manhattan Institute who has spoken to police officials about the tactic, said there was no question it had an effect on crime.
Ms. Mac Donald is apparently the Times’s go-to conservative on crime issues,*** and she must know. After all, as the Times says, she has spoken with police officials. I don’t have any contacts among the NYPD brass, so I had to look at the available crime statistics. I chose murder. It’s the crime where statistics are the most accurate. It’s also the crime most likely to be reduced by the cops taking guns away from bad guys on the street. So I expected a sharp decrease in the years following 2003.

Hmm. The trend is downward, not dramatic but gradual, and it seems to be a continuation of a trend that started before the big increase in stop-and-frisks. There’s also that rise in murders in 2006, when the number of stop-and-frisks also increased by about 25%, roughly from 400,000 to 500,000. (For a line graph showing the rise in the number of stops, see the Times article)

My analysis is just a quick-and-dirty. To draw a credible conclusion, you’d have to take several other variables into account. A good multivariate model might find that the effect of stop-and-frisk was greater than it appears – maybe Ms. Mac Donald knows of such studies and even mentioned them to the reporter, and he just left them out of his story. Or maybe those high-quality studies, if they exist, found no effect. But just looking at the basic data on the two variables – stops and murder– makes it hard to say that “there was no question of a deterrent effect.”  But Ms. MacDonald said it anyway.

* Or rather they made 570,000 stop-and-frisk searches. Since the usual suspects may have had more than one such encounter, we don’t know how many individuals were stopped. But we do know that 490,000 of them were black or Hispanic, 53,000 were white. Those numbers, while they do not reflect the population of New Yorkers, may reflect the population of street criminals.

** The police agreed to make the data public as part of the settlement of a lawsuit. Four cops stopped a man and wound up firing 41 bullets at him, killing him. They thought he had a gun. In fact, he was unarmed and innocent of any crime. Needless to say, the victim was black, and now the cops have to keep records of stop-and-frisks, including the race of the stop-and-friskee.

This post cites her view of the salutary effects of harsh drug laws, a view she supports with evidence comparable to that mentioned in the current article.

1 comment:

PCM said...

It's a shame the Times article pushed the racial-disparity angle. That seems to me to be the lesser story since the rate correlates with violent crime rates.

The real question is does the policy work--do stop and frisks prevent or lower crime. We really do not know.

It's hard to imagine that 500,000 frisks do not have some deterrent effect on criminals' willingness to carrying weapons. But indeed, it's pretty clear doubling the number of frisks has not had much impact on crime.

And there is indeed a negative impact in conducting hundreds of thousands of stop and frisks on innocent people. That is what needs to be debated. Given all the frisks and the size of the city, you'd think it would be pretty to set up an experiment in two precincts...