“A/B testing” the impact of school closures using crime data

This weekend, the Knight Lab is sponsoring a hack day focusing on Chicago Crime Data as made available by the Tribune’s crime data portal and API.

I’m a little wary of crime data first because crime data does not equal a resident’s experience of safety.  It’s easy to think of situations where crimes go unreported, or where increased community cohesiveness might lead to an increase in crime reports.  Second, the way crime stats are framed and parsed by Chicago residents often seems to be alarmist and often further stresses racial and economic tensions in gentrifying communities rather than offering a space for increased community collaboration or developing progressive solutions to neighborhood safety.

Are there uses of crime data that contribute to a different civic discourse? One idea that came to me is based on this current moment where Chicago Public Schools (CPS) is slated to close a number of schools.  One issue raised by critics is the safety of students who may have to cross gang boundaries to reach their new “welcoming school”.  CPS’ proposal to provide resources for students who must attend a new school after their school has been closed, includes an expansion of the “Safe Passage” program which partners with neighborhood organizations to help increase safety for students on the way to and from school. From my knowledge as a caregiver of CPS students and as a frequent news consumer, I don’t have much sense of how successful this program has been so far.  After the closures happen, how will CPS and city residents know how school closures affected students on their way to and from school?

I hypothesize that we might be able to use crime data as one way to see changes in communities after schools have been closed.  I also think this is a general case of “how does crime change along with some policy event”.  I imagine a web platform where residents can define an “experiment” by looking at a specific geography, types of crime and time period.  Crime data would then be compared before and after the test time period to see how crime changed.

In general, I think it’s important to frame these experiments as “what changed” instead of “did this work” because I think the crime data set probably isn’t enough on its own to determine

Questions:

  • What kind of crimes would be indicators of school commute safety? Or, should we look at crimes from specific time periods before and after school?
  • What methods do sociologists use to do these kinds of comparisons?
  • Which schools/communities currently participate in the “Safe Passage” program

Other use cases:

  • Neighborhood cleanups
  • Proposed city legislation targeting liquor stores
  • “Positive loitering”
  • Negative outcomes for heightened targeting of youth by police