About . . . . . . Classes . . . . . . Books . . . . . . Vita . . . . . . . Links. . . . . . Blog

by Peter Moskos

March 10, 2015

Value Over Replacement Cop

This was gonna be my idea! "Bobbies and Baseball Players: Evaluating Patrol Officer Productivity Using Sabermetrics." So kudos to Luke Bonkiewicz because he actually researched and wrote the article and I didn't. Here's the abstract from the current issue of Police Quarterly (2015, Vol. 18(1) 55–78):
Police officer productivity is an understudied topic in police research. Prior studies on productivity have primarily relied on rudimentary statistics, such as calls for service and arrests. A more advanced method for evaluating productivity should (a) account for the diverse activities of patrol officers, (b) weight different productivity outputs, (c) evaluate officers in terms of available minutes for self-initiated activities (productive time), and (d) offer agencies the flexibility to select, prioritize, and weight patrol activities most relevant to their jurisdictions. Borrowing from a baseball sabermetric called Value Over Replacement Player, we create and test an innovative statistic called Value Over Replacement Cop. This metric analyzes 12 patrol activities and generates a single number by which to quantify and evaluate a patrol officer’s productivity. Using data from a midsize U.S. Police Department (325 sworn officers), we find strong support for the validity of this new metric.
This is a good start. But the problem is that this measure doesn't take into account crime, the prevention of which is the primary purpose of police. Crime needs to be the main variable, not indicators of police officer "productivity" (which aren't unimportant, but still).

8 comments:

Noumenon said...

Value over Replacement Player counts runs and outs -- the things you produce -- not wins, the ultimate metric of baseball. Maybe crime should go into evaluating the department but not the individual cops?

Peter Moskos said...

But I *do* want to link crime prevention to individual cop. Nothing else matters.

Well, a few other things matter: reduced public fear; any "setting straight" of wayward youths; being accessible and helpful to the good people; treating victims with dignity and respect. But the most important thing, "the principal object to be attained is 'the Prevention of Crime.'"

(I'm talking about patrol here, not detectives and specialized divisions.)

Unlike baseball, police will never "win" the game. Crime is ongoing work. It's not a game to win.

Cops do need to make arrests. But I've never been convinced that more arrests are better. In baseball terms arrests are more like an outfielder catching fly balls. Most are hit right at you. A few times a good outfielder makes a difference... a bit faster or dives to make a catch (though often that same aggressive outfield lets one get by for a 3-base error). As to cops being like outfielders, I'd prefer a slow but steady outfielder who hits 300 and has some power.

Anonymous said...

I'm a police officer and I find this idea intriguing.

First, I think that VORC can be used to see how cops and their productivity affect crime. So I think that crime is still the main variable of interest; VORC is just a new way to examine the relationship between cops and crime.

Second, the article says that arrests aren't the only outputs that can be measured and used in the equation. What about POP projects or specific patrols? I think the point of this article is to give agencies a flexible method for evaluating cops, right? Not sure, though.

Finally, Peter, in some places (ahem, Baltimore), most arrests might stare you in the face, but in my opinion, there are many other places in which officers have to work harder to make arrests. Here in Podunk Town, USA, for instance, we have a lower crime rate and I feel like I earn most of my arrests through field contacts or traffic stops.

Anyway, take it for what it's worth. Enjoy your blog, by the way.

kevinkarpiak said...

I don't know, this seems like another in a long line of "police science" using statistical lights ang fog to avoid the real tricky--and philosophical/ethical/moral--question of what the ends of police should be.

In Game Theory this is pretty easy, since the goal is pretty clear. But I'm not sure we even have the intellectual tools to begin a discussion about what police are for; and I'm certain the answer to that is neither straightforward or obvious.

SupportYourLocalSheriff said...

Uh, so how does VORC capture a cop's ability to interview people, interact with the public, and problem-solve community problems? Another worthless idea from an ivory tower nerd. You might want to work the streets for a little while before applying sports stats to crime stats.

Peter Moskos said...

Wow. That was sweet and unnecessary.

Your point is actually good. You describe exactly what I'm trying to measure.

And that's why, from my time on the street, I'm trying to figure out how to do so (no thanks to you), you county fuck. You're so right, except for the fact you're a dick.

Good luck in your next election.

CJstudent said...

Hey, SupportYourLocalSheriff, you might want to search both Moskos and Bonkiewicz. Moskos worked in Baltimore. Bonkiewicz, according Google, is still a street cop (curiously, patrolling and researching at the same time). Cops who write stuff about cops for cops are usually spot on, and I think Moskos would probably agree.

Anonymous said...

Jeez, there sure are a lot of negative comments. It's an article that explains a way for police administrators to more validly and reliably evaluate patrol officers. Yes, there are drawbacks, but it makes some major improvements over just measuring a cop's arrests. You'd think the guy wrote something about the relationship between race and crime.