NYC's newest data tool

In the final days of his mayoral term, Mike Bloomberg has announced the rollout of a new database available to the public, known as the Data Analytic Recidivism Tool, or DART.  DART compiles "recidivism" data on all offenders arrested in NYC in 2009.  I place the word "recidivism" in quotes because the DART database tracks all individuals who were initially arrested in 2009 in NYC, including those who were not necessarily convicted of anything.  Accordingly, the city appears to be labeling (untentionally?) as "recidivist" any person who was arrested initially in 2009 and then re-arrested again within that year.  This is not the definition of recidivism used by the Bureau of Justice Statistics, which tracks recidivism among released offenders.  

NYC's new DART database ostensibly enables researchers to establish correlations between repeate arrests and certain crimes and offender characteristics.   As explained by the official press release:

DART enables users, including criminal justice professionals, program planners and researchers to select a group of defendants based on factors like age, prior criminal history and details about their criminal cases. DART will then produce a graph displaying three different one-year re-arrest rates for the selected group, including the percentage re-arrested for any crime within a year; the percentage re-arrested for a felony within a year; the percentage re-arrested for a violent felony within a year; and a comparison to the citywide average.

Out of curiosity, I decided to visit the DART website and try the database.  Here's what I discovered:

First, the age groupings seem large and somewhat arbitrarily constructed.  You cannot run a report on "21 year olds arrested once" but instead must run a report for individuals falling within various age groups such as 16-19, 20-24, 25-29, 30-34, 35-39, and 40+.  Are we really to believe that recidivism breaks according to these very neat, five year groupings?  Moreover, is there any good reason (other than expediency) for lumping everyone over the age of 39 in the same category?  Were I a decision-maker (ie, sentencing judge, probation officer etc), I might worry that my data has been unnecessarily skewed by these groupings. For example, 40 year old criminals may be receiving a slight benefit, if we assume that a 40 year old's likelihood of recidivism is, on average, higher than that of a similarly situated 50 or 60 year old.  And, for all I know, there may be a world of difference in terms of recidivism rates between a 16 year old and a 19 year old.

Then I tried to run a report. After playing with the query terms for a few minutes, I could see how this might provide an interesting idea for a future research project; beyond that, I didn't find DART very illuminating. First, on several occasions, I tried to come up with a report for a specific type of offense and offender (males, between the ages of 16-19, arrested in Brooklyn for sale of a controlled substance, released on bail, subsequently convicted etc etc), and ran into the problem that I had crafted a category that was too narrow and therefore failed to produce a report.   So, in order to generate a report, I needed to reduce specificity.  But of course, the broader the search terms became, the less information the "tool" provided.  

For example, thanks to DART, I now know that of the approximately 8,300 16-19 year olds arrested in Brooklyn in 2009, "47.2% were arrested for any crime within one year."  This handy graph also breaks down the info in terms of nonviolent and violent felonies.  What do we make of this information?  Does the data tell us something important about the police, or about the individuals who were arrested? 

And while we're at it, for whom is this "tool" useful?  The city's Criminal Justice Coordinator, John Feinblatt, advises:

“Prosecutors and defense attorneys can use facts about recidivism to assess plea and sentencing policies, and policy makers can use this tool to help allocate scarce criminal justice dollars where they are needed most.”  

Really?  I'm a big fan of data, and I'm all for using empirical analysis to guide policy, but right now, this looks a lot like a cutesy tool prone to manipulation and misinterpretation.  Frank Zimring,(who compliments the tool here because it improves the public's access to data) made the following comment to the Wall Street Journal:

"If you are asking for probation for your client, you can produce outcomes for your guy and create averages and compare it to others in that category," said Franklin Zimring, a law professor at the University of California-Berkeley, who was among a group of people asked by the city for preliminary feedback on the database. "If you're a prosecutor, you can use the same data, but obviously with a different spin." 

So if both sides can spin the data, what is its value?

A final point: as I noted above, the Bureau of Justice Statistics also maintains a database on recidivists. Like DART, the BJS database also maintains the five-year age groupings and the 40+ category, and BJS' data is based on a sampling of offenders released from prisons in 15 participating states in 1994, a full generation ago. According to the BJS website, a database containing updated information is on its way.  Were I a criminal defense attorney (or prosecutor, for that matter), I would watch very carefully for the updated BJS website, as its information may turn out to be far more illuminating than DART's pretty bar graphs.