Home > Uncategorized > OIG Report: Broken Windows doesn’t work

OIG Report: Broken Windows doesn’t work

June 23, 2016

The Office of the Inspector General for the New York Police Department (OIG-NYPD) issued a report yesterday which used statistical analysis to demonstrate that the “Broken Windows” theory of policing is flawed. From their Recommendations, page 72:

OIG-NYPD found no evidence that the drop in felony crime observed over the past six years was related to quality-of-life summonses or quality-of-life misdemeanor arrests. This suggests that there are other strategies that may be driving down crime. Between 2010 and 2015, quality-of-life enforcement rates – in particular, quality-of-life summons rates – have dramatically declined, but there has been no commensurate increase in felony crime. While the stagnant or declining felony crime rates observed in this six-year time frame may perhaps be attributable to NYPD’s other disorder reduction strategies, OIG-NYPD finds no evidence to suggest that crime control can be directly attributed to quality-of-life summonses and misdemeanor arrests. Whatever has contributed to the observed drop in felony crime remains an open question worthy of further analysis.

The report goes on to say that the NYPD should take a more data driven approach to deciding what’s actually working and what isn’t, and should “conduct an analysis to determine whether quality-of-life enforcement disproportionately impacts black and Hispanic residents, males aged 15-20, and NYCHA residents.”

Very happy about this report, it’s been a super long time coming. The NYPD has said the report is flawed, and will come back with a response within 90 days. I’m looking forward to that as well.

Categories: Uncategorized
  1. June 23, 2016 at 8:17 am

    Crime is reduced when people have jobs and are working. Not eliminated, but reduced. People that are concerned about things such as 401k contributions, college savings, resume updates, and so forth don’t often commit crimes.

    The root of crime always has been poverty.

    Unfortunately, the NYPD better enjoy this little run while it lasts. The first totally autonomous taxi cab is being tested there now.


    Liked by 1 person

  2. Jesse
    June 23, 2016 at 10:43 am

    I love to see fascist-leaning police strategies getting discredited, but:

    “Between 2010 and 2015, quality-of-life enforcement rates – in particular, quality-of-life summons rates – have dramatically declined, but there has been no commensurate increase in felony crime.”

    seems lacking because, like new-deal stimulus spending, the concept was (I thought) to sort of “jolt” the system from one steady state to another.

    So it sounds to me a little like: “We stopped doing WPA projects after the depression, but the economy grew anyway-therefore the WPA didn’t do anything.”

    I might be missing something though.

    Liked by 1 person

    • JR
      June 23, 2016 at 2:49 pm

      I was thinking the same thing. However, this is a valid argument against those who think that under current conditions broken windows policing is an effective strategy.


  3. tenaflyelks
    June 23, 2016 at 6:14 pm

    NYPD’s same day response: study is Flawed but we’ll need 90 days to figure out exacltly how


    • Jonathan Marcus
      June 24, 2016 at 3:35 pm

      My thought as well. It’s wrong. We don’t know how or why, but we’re sure it’s wrong.


  4. June 26, 2016 at 2:40 pm

    Maybe it was stop-and-frisk which kept guns away in a city with strict gun control laws. Oh wait, that would be so not politically correct.


  5. Stuart Buck
    June 28, 2016 at 10:42 am

    This report is all correlation, and expressly admits on occasion that it cannot say anything about causality. The report has almost no bearing on whether quality-of-life arrests have any effect on other crimes.

    If someone want to do some “statistical analysis” that would actually be relevant to the important causal questions here, they could look for exogenous events or shocks that made enforcement go sharply up or down in various precincts, and then do a diff-in-diff analysis or perhaps comparative interrupted time series.

    Otherwise, page upon page of bar charts doesn’t tell us much of anything.


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