Compstat, NYPD, & Mad Data




S

candal
erupted in the New York City Police Department in March 2004 when
the Police Benevolent Association (PBA), a police officers union,
accused various precinct commanders and other officials of “cooking
the books” on crime statistics. One of the most publicized
cases was the 50th

Precinct in the Bronx. According to

News


day

, when
the deputy inspector left for a more esteemed position in Brooklyn,
questions arose about how crimes had been recorded under his leadership.
While the precinct had seen a 26 percent drop in crime during the
deputy inspector’s tenure (a probable reason for his promotion),
there was an 11 percent upsurge in the weeks following his departure. 


In
addition to police officer denunciations, an array of community
members, from residents to delivery people, have come forward to
describe incidents where they were turned away at the police station
when they sought to file theft or other reports. In addition to
these “non-reports,” officers from the 50th claim that
crime was often mis-catalogued and understated. In one instance,
a case of grand larceny—a felony—was reported as a petty
larceny misdemeanor. A rape was logged as an “inconclusive”
incident. Officers also claim that sometimes aggressive supervisors
would personally review crime complaints to determine whether they
were really felonies or could be downgraded. These kinds of problems
occurred throughout the city, prompting denunciations not only in
the Bronx, but in other boroughs as well. 


While
the PBA brings forth these charges to demand growth in their numbers
and their pay, the proposed solutions mischaracterize the problem.
As a range of commentators from police officers to civil rights
advocates to scholars have pointed out, the problem might be rooted
in Compstat, a computerized system used to track crime trends. Compstat,
short for “computer statistics,” was first used in 1994
by then police Commissioner William Bratton and newly hired Deputy
Commissioner Jack Maple. An idea based on the old pushpin crime
maps, Compstat keeps statistical and spatial records of what kinds
of crimes are occurring and where. In the 1990s Compstat was lauded
as having played a central role in the much publicized decline of
crime in New York City (the causes of which continue to be disputed),
bringing fame to “tough on crime” Mayor Rudy Giuliani,
Commissioner Bratton, and the NYPD in general. Today Compstat technology
is used in police departments across the nation and throughout the
world, often peddled by consulting firms headed by Giuliani and
various ex- NYPD top brass. 


Initially,
Compstat tracked only the seven major crimes that municipalities
must report to the Federal Bureau of Investigation—murder,
rape, robbery, assault, burglary, grand larceny, and grand larceny
auto. However, it has evolved since its first application and now
tracks a large amount of information, including all categories of
crimes, misdemeanors, and even the status and whereabouts of parolees.
Compstat relies on a Geographic Information System (GIS) especially
tailored for police departments by GIS software giant, ESRI Inc.
The GIS can quickly compute, graph, and map information, and can
be used to overlay spatially displayed crime records with information,
such as proximity to schools or population densities. The result
is a data-driven, cost-benefit analysis where police forces are
targeted to the areas deemed the most needy. However, problems arise
as precinct commanders’ “efficiency” is converted
to statistics and often marketed as well. A central aspect of Compstat
is regular meetings, often held at the Command and Control Center
in lower Man- hattan, where precinct commanders are grilled by NYPD
top brass. Precinct commanders must explain their statistics and
strategies under the glare of giant monitors replete with graphs,
charts, and maps displaying their success—or lack of it. These
meetings are reported to be very tense. The rhetoric and procedures
center on the idea of “accountability” and precinct com-
manders are rewarded  or punish- ed, depending on their per-
formance. 


As
odd as it may seem, there are striking similarities between this
policing strategy and current national trends in K-12 education.
The rhetoric and ideology of “accountability” are paramount
in both these trends and they both rely on standardized measures
of achievement and on the visibility of performance directly tied
to rewards and punishments. Much like school principals, precinct
commanders are under tremendous pressure to generate new numbers
and, much like teachers, the discretion of officers is taken away
as the job becomes standardized. While in light of repeated police
abuse, it may seem beneficial to remove officers’ individual
discretionary power, the decision-making power is not given to residents
or independent review boards, but is instead shifted to the precinct
commanders and administrators who are no less likely to violate
the rights and dignity of the people they police. 


Through
Compstat, the importance of keeping crime statistics was shifted
from a clerical task to a central administrative obligation around
which data-driven decisions are made. This has led to fraud by precinct
commanders and to a “lean and mean” style of management
that often ignores the needs of the community. Furthermore, Comp-
stat data can lead to causal connections where they may not exist,
leading, for example, to a crack down on porn shops if these are
correlated with high crime areas. It is important to understand
that a data-driven system is only as good as its data; that which
does not get recorded does not get analyzed.





R. Valeria Treves
is a graduate student in geography at Hunter College, CUNY.