Lies, Damned Lies and CompStats

Have New York City cops manipulated data to make the statistical evidence of decreasing crime in the Big Apple seem better than it really is? "Most certainly," a new survey of retired NYPD cops suggests, as summarized in a recent New York Times article.

The NYPD, naturally, responded to the news by attacking the survey's methodology and findings in an effort to defend the integrity of its signature CompStat program. But the NYPD's arguments are flawed and beside the point.

Have CompStat's books been cooked? There's no debate about the fact that CompStat -- which was pioneered in New York, and is now widely replicated in other U.S. cities -- has had its issues. Accusations of tampering have accompanied the CompStat franchise's spread to places like Atlanta, Baltimore, Dallas, New Orleans and Washington. Meanwhile, once faced with manipulation accusations in the past, the NYPD refused to cooperate with a city commission (chaired by a federal prosecutor) that sought to investigate such charges. Even while resisting external investigations, though, the NYPD internally meted out discipline in 11 proven cases of data manipulation -- with four cases directly implicating precinct commanders. The idea that the NYPD caught all wrongdoers, or that they acted alone, is simply not credible.

These manipulations happen for structural reasons. After all, the CompStat system works by (1) precisely mapping crime trends, precinct by precinct; and (2) punishing or rewarding precinct commanders based on how reported crime rates in their areas rise or fall. Such a system inherently creates an incentive for commanders to show decreases in the number of reported crimes -- even if the number of actual crimes remains steady. Sometimes it's just easier to fudge data on the books rather than confront real problems in the field.

This phenomenon isn't unique to law enforcement. Progressive educator Diane Ravitch, for example, identifies a similar trend that happens when the federal government uses test scores as a metric to base school funding on. Pressure to increase scores causes teachers to simply teach to the test, and even manipulate results. Chalk it up to Campbell's Law, which states that the more any quantitative indicator is relied on for decisions, the "more subject it will be to corruption pressures, and the more apt it will be to distort and corrupt the social processes it is intended to monitor."

I don't mean to argue that the CompStat program hasn't helped reduce crime, or that the NYPD's crime data is entirely suspect. The point is that we need to put the numbers in perspective, rather than having our perspective defined by one narrow set of numbers. A better perspective wouldn't just track the number of larcenies or murders in a neighborhood -- it would also track relative rates of unemployment, literacy or access to health care. Rather than encouraging the cops to view crime victims as statistics, a progressive perspective would reward cop commanders for establishing meaningful human relationships within the communities they police.

Photo Credit: SpecialKRB