Home > Uncategorized > End Broken Windows Policing

End Broken Windows Policing

July 14, 2016

Yesterday I learned about Campaign Zero, a grassroots plan to end police violence. The first step in their plan is to end Broken Windows policing. Here’s their argument:

A decades-long focus on policing minor crimes and activities – a practice called Broken Windows policing – has led to the criminalization and over-policing of communities of color and excessive force in otherwise harmless situations. In 2014, police killed at least 287people who were involved in minor offenses and harmless activities like sleeping in parks, possessing drugs, looking “suspicious” or having a mental health crisis. These activities are often symptoms of underlying issues of drug addiction, homelessness, and mental illness which should be treated by healthcare professionals and social workers rather than the police.

Having studied the effects of uneven policing myself, especially how it pertains to the data byproduct of “police events,” I could not agree more.

There was a recent New York Times article that got people’s attention. It claimed that there was no bias in police shootings of blacks over whites. What it didn’t talk about – crucially – was the chance that a given person would end up in an interaction with the police in the first place.

It’s much more likely for blacks, especially young black men, to end up in an interaction with cops. And that’s due in large part to the broken theory of Broken Windows policing.

New York City’s version of Broken Windows policing – Stop, Question, and Frisk – was particularly vile, and was eventually declared unconstitutional due to its disparate impact on minorities. The ACLU put some facts together when Stop and Frisk was at its height, including the following unbelievable statistics from 2011:

  1. The number of stops of young black men exceeded the entire city population of young black men (168,126 as compared to 158,406).
  2. In 70 out of 76 precincts, blacks and Latinos accounted for more than 50 percent of stops, and in 33 precincts they accounted for more than 90 percent of stops. In the 10 precincts with black and Latino populations of 14 percent or less (such as the 6th Precinct in Greenwich Village), black and Latino New Yorkers accounted for more than 70 percent of stops in six of those precincts.

What happens when this kind of uneven policing goes on? Lots of stupid arrests for petty crimes, for “resisting arrest,” and generally for being poor or having untreated mental health problems. About 1 in 1000 such stops are directly linked to a violent crime.

And again, since those stopped are overwhelmingly minority, it means that when City Hall decides to use predictive policing based on this data, they end up over policing the same neighborhoods, creating even more uneven and biased data. That continuing stream of data even ends up in sentencing and paroling algorithms, making it more likely for those same over-policed populations to stay in jail longer.

It’s high time we get rid of the root cause, the theory of Broken Windows, which was never proven in the first place, which optimizes on the wrong definition of success, and which further undermines community trust in the police.

Categories: Uncategorized
  1. July 14, 2016 at 10:24 am

    This practice, particularly in the case of New York City, has been well-known and watched by some statisticians. In fact, it is (unfortunately) a staple of statistics textbooks along with the contingency table analysis showing predominantly white juries are much more likely to make death penalty convictions for blacks than for whites.

    In this case, for instance, Gelman, Carlin, Stern, Dunson, Vehtari, and Rubin in their Bayesian Data Analysis (3rd edition, 2014) have section 16.4 where their illustrate hierarchical Poisson regression using data for police stops in New York City. For the aggregate data for a 15 month period in 1998-1999, the basic facts are that “[b]lacks and hispanics represented 51% and 33% of the stops, respectively, despite compromising only 26% and 24%, respectively, of the population of the city.” They go on to check whether the claims that these minorities are responsible for a greater number of crimes (as measured by a proxy of number of arrests), and controlling for precincts, as an indicator of more crime-prone regions, that is, the “broken windows” policy.

    Unfortunately, the authors conclude, or, rather, in my opinion, rationalize “Does the overall pattern of disproportionate stops of minorities imply that the NYPD was acting in an unfair or racist manner? Not at all. It is reasonable to support that effective policing requires many people to be stopped and questioned in order to gather information about any given crime. In the context of some difficult relations between the police and ethnic minority communities in New York City, it is useful to have some quantitative sense of the issues under dispute. Given that there have been complaints about the frequency with which polices have been stopping blacks and hispanics, it is relevant to know that this is indeed a statistical pattern. The police department then has the opportunity to explain its policies to the affected communities.”. Hmmm.

    I much prefer the presentation of the same study and data in Gelman and Hill, Data Analysis Using Regression and Multilevel/Hierarchical Models (2007), culminating in their section 15.1, but visited twice earlier in sections 1.2 and 6.2. They present the odds that blacks and hispanics are stopped for violent crimes 2.5x and 1.9x as often as whites, and for weapons crimes, 1.8x and 1.6x as often as whites. For property crimes and drug crimes, whites were stopped slightly more often for the former, and more often for the latter. There Gelman and Hill present the quantitative conclusions without editorializing.

    The original study was A. Gelman, J. Fagan, A. Kiss, “An analysis of the NYPD’s stop-and-frisk policy in the context of claims of racial bias”, Journal of the American Statistical Association, 2006.

    The death penalty question is presented as an exercise 27 in Chapter 13, “The analysis of categorical data”, within Rice’s Mathematical Statistics and Data Analysis (3rd edition, 2007), and as the example of section 19.1.2 of Ramsey and Schafer’s The Statistical Sleuth (2nd edition, 2002).


    • July 14, 2016 at 10:30 pm

      I had a chance to look into this a bit more. The Hill and Gelman text essentially reproduces the paper of Gelman, Fagan, and Kiss. Moreover, that paper is the source of the quote I initially disagreed with in the above, described in the paragraph beginning “Unfortunately …”. Rather the explanation or interpretation which appears in Gelman, Carlin, et al is taken directly from the Gelman, Fagan, and Kiss paper and, given that it is not an afterthought, even if I am familiar with that study, it is wrong for me to second guess the primary authors. So I withdraw that “Unfortunately”.

      There have been other statistical studies and discussions which have emerged amongst all this, notably the one by Freyer, a counterpoint by Feldman (which I have some quibbles with because I judge it is distorting and cherry-picking), one by Sadler, Correll, Park, and Judd, and an interesting one by Ross which uses methods I use day to day, professionally.


    • July 15, 2016 at 7:27 am

      I will have to respectfully disagree with Gelman and his co-authors here. They treat crime as a well-defined data generating mechanism, which is far from true. Most crimes never get found, and many police stops associated to crimes are “found” crimes – like drug possession – or even manufactured ones – like resisting arrest. That’s why whites and blacks smoke pot at the same rate but blacks are much more likely to get in trouble for it (at least until recently).

      What we need is to sort out which crimes are actually reported to the police and which crimes are essentially created by the police, and only after that can we understand which populations and neighborhoods “deserve” extra scrutiny from police.

      Liked by 1 person

  2. G.
    July 14, 2016 at 10:31 am

    Well what is the presumed solution?

    Whichever way it goes, the outcome will most probably be more centralization of power in the state, not to mention use of robots in civil coercion.


  3. zamanskym
    July 14, 2016 at 10:57 am

    Funny how closely this //s those misguided, amazingly harmful education policies such as charters / high stakes testing / Vam.

    It’s also interesting that both broken windows and school reform have moved away from what I feel is fundamental to both – community.

    Just as school reformers have destroyed community schools and community voice in education broken windows policing has removed community policing. One of Dinkins big moves, for which he gets no credit, was IMO a big part of the start of the drop in crime that Guiliani gets credit for with his broken windows was Dinkins move to get the cop out of the car, back on the beat on his feet as a member of the community he’s there to protect.


  4. July 15, 2016 at 7:06 am

    I vote for Columbia stats, journalism and econometrics combined, including Gelman and mathbabe. Local knowledge really helps interpretation of statistics!


  5. July 15, 2016 at 7:17 am

    Reblogged this on Mooms's Blog and commented:
    Great lessons in conditional probability to be learned from these data in Criminology


  6. July 16, 2016 at 12:31 am
  7. July 17, 2016 at 12:25 am

    The Dallas police reportedly had a new de-escalation training program which had allegedly dropped police violence issues by about 90% over the last year or two. I believe this was covered in Bloomberg. There are some obvious issues with number reliability.


  1. No trackbacks yet.
Comments are closed.
%d bloggers like this: