Under the stewardship of Police Commissioner Howard Safir, the NYPD began analyzing daily crime statistics collected from its 40,000 officers throughout the five boroughs of New York City and generating computer modeling of crime trends in a system dubbed CompStat that allowed the accurate identification of crime trends with pinpoint accuracy, often permitting nearly as precise predictive modeling via extrapolation (Safir, 2003).
The other main benefit and purpose of CompStat was that is enabled police administrators to grade the performance of every precinct according to any criteria defined by policy considerations. That aspect of CompStat is relied upon heavily by NYPD administration to the extent that Commissioner Safir reassigned, removed, or demoted fifty-four precinct commanders during his tenure as commissioner between 1996 and 2000 (Safir, 2003). Naturally, the technical means of data collection and analysis techniques differ quite profoundly from those available to previous generations of police administrators, but the actual methodology of the CompStat system is very similar, in principle, to the original pin mapping in use for almost a century and a half throughout modern municipal police organizations worldwide (Schmalleger, 2008).
The Problem with Evaluation via Objective Crime Statistics:
Ideally, crime mapping and predictive modeling also provide police administrators with purely objective data enabling evaluations that are immune from subjective contamination to whatever extent those data are relied upon for evaluations.
Whereas the principle weakness of narrative supervisory evaluations is their subjective nature, the corresponding weakness inherent in the objective results-oriented evaluation made possible by computer-enhanced, up-to-date, pinpoint analyses of crime in each precinct or command area lies in the manipulation of data prior to its entry into the system (Conlon, 2004).
For example, the fact that police administrators used CompStat to identify poorly performing precincts led to numerous instances of deliberate data manipulation, such as in the form of enforcement instructions issued to patrol officers. Specifically, if borough commanders or precinct captains faced negative evaluations and supervisory criticism by virtue of CompStat data indicating high rates of robbery in their areas of command, they learned to tailor their enforcement commands to manipulate the data in their favor (Conlon, 2004).
Typical examples included borough-wide instructions to apply very stringent technicalities to differentiate robbery from petit larceny based on statutory minutia that allowed crimes that had previously almost always been categorized as robberies to be entered into the system as lesser offences. In many cases, outright thefts from persons were downgraded to instances of "abandoned property" infuriating crime victims whose wallets had been snatched from a park bench, but without explicit threats of violence or intimidation that would have met the strictest technical definition of "robbery." The practice become so rampant that NYPD officers came to refer to the CompStat system as "ConStat"(Conlon, 2004).
Because of the potential inadequacies and error inherent in both narrative supervisory review and objective crime mapping methodologies, their advantages are best employed in conjunction with each other rather than relied upon exclusively.
OUTCOMES (FBI LEB, 2007)
In that regard, a fundamental approach detailed by Dr. W. Dean Lee, the current head of the Organizational Program Evaluation and Analysis Unit at the Federal Bureau of Investigations (FBI) headquarters in Washington, DC. According to Dr. Lee, the most reliable approach to program evaluations consists of a complex analysis of all the elements represented in the figure below, but from right to left, in order to identify specific program and performance goals in terms of their ultimate objectives (Lee. 2007).
The advantage of the comprehensive evaluation system detailed by Dr. Lee is that it addresses every conceivable variable while incorporating the advantages of the other models of police and criminal justice administration. Ultimately, the hope is that the model advanced by Dr. Lee will eventually be combined with the beneficial elements of narrative supervisory review and predictive modeling via crime mapping into computerized formats that allow for the seamless integration of different evaluation methodologies.