Rudolph Giuliani, who became mayor of NYC in 1993, and William Bratton, the New York Police Department (NYPD) Commissioner, jointly revolutionized policing in New York. Their approach was based on actively preventing crimes, rather than reacting to criminal activity. The model they implemented, CompStat (COMPuter STATistics or COMParative STATistics), strategically uses data to fight against crime. By implementing this approach, the yearly crime rate was reduced from 700,000 complaints in 1994 to only 100,000 complaints in 2013. Thus, the New York model is analyzed as a best practice example for the field of smart public safety. The implementation, approach and successes of CompStat were discussed in a group setting, as well as in a subsequent interview with Dennis Smith. Smith is a professor of public policy at New York University, and together with William Bratton, he published an analysis of CompStat in 2001 (Police Executive Research Forum 2013).
The history of the new approach
Systematic data collection began long ago in NYC. In the 1970’s, the first Mayor’s Management Report had already been published. This report collects and distributes statistics about all public offices in New York twice a year, in order to measure and understand their performance. However, the fact that these values were not, or not sufficiently, analyzed and that no corresponding strategies for improvement were developed, continued to be a problem. Additionally, there were very few, if any, results-oriented indicators. For example, the central values of New York’s police department were based on reactions to crime: how quick was the response to calls, was an arrest made, how many officers are on patrol each day etc.
Before William Bratton introduced CompStat in New York in 1994, the general belief was that the police is unable, or barely able, to exert any influence on the magnitude of crime within a city. Correspondingly, the reduction of crime was not seen to be the responsibility of the police. Instead, the police’s main task was considered to consist of reacting to crimes when they occurred and, most importantly, reacting fast.
Since the 1970s, when radio communication and 911 systems were implemented in the United States, 911 calls had been distributed to patrol cars in a random manner, in such a way that they could be responded to as quickly as possible. This is known as the Random Patrol model. This approach was based on the assumption that citizens, when they are threatened or involved in a crime, will call the police and that the police will then arrive on the scene as soon as possible, and – ideally – arrest the perpetrator. However, studies conducted during 1979 and 1980 showed that Random Patrol had no significant effect on either the reduction of criminal activities or on the feelings of safety amongst the population.
Progress was first initiated by David Dinkins, NYC’s mayor before Giuliani, through the introduction of a new concept: Community Policing. Policing was re-thought in such a way that police officers now entered the districts and came into direct contact with the residents living there. This allowed the crime rate, which had been continually increasing until then, to be halted.
These developments were further developed once, in 1994, William Bratton was given the position of Commissioner of the NYPD by the mayor of the time, Rudolph Giuliani, and the two jointly implemented CompStat. Their goal was to use this approach to increase the city’s safety. During a press conference they announced that a 10% reduction in crime was expected in New York, and by the end of 1994, crime had already decreased by 13%. In the following year, a decrease of 17% was achieved.
The CompStat approach is about a change in how police work is managed. This includes the central elements “Decentralized commands”, “Results-oriented activities” and “A focus on the reduction of crime”.
Thus, a shift from reactive policing to a focus on the reduction of crime ensued. In order to achieve the goal of preventing criminal activity, the following methods were used in New York: Evidence-based Policing, Hot Spot Policing und Community Policing.
The central aspects of evidence-based policing include the use of available data in order to develop an effective strategy and continual follow-up and evaluation.
Yearly, approximately five million radio announcements were made and followed up on by the police. It was only upon the introduction of CompStat that patterns were able to be identified and it was observed that 40% of these calls came, repeatedly, from the same addresses. This meant that a police patrol would be sent to an address to which officers had already been sent several hours, or days, earlier. In doing so, none of the information generated during the previous visit was passed on to the officers now going to the same address. Thus, police were sent to this address, but the problems there were not being solved. One reason for this was that the pressure to respond to calls quickly can lead to officers concluding visits as fast as possible in order to be available for the next call. The new approach change this by trying to assign not different police officers to the area but if possible the same.
Hot spot policing
Bratton used the large volumes of available data in order to analyse where crimes occur and where criminal activities are most concentrated. Corresponding to the results, police were stationed in higher numbers in these areas. There was no concern that, in doing so, criminality would be pushed from one location to another. Smith explains this by pointing out that one cannot assume that individuals who commit crimes are inherently criminal. When people are provided with different ways of earning money, they will choose the best option. Through the NYPD’s new approach, the risks associated with illegal businesses increased dramatically, and thus, this line of business was simply no longer worth pursuing. Therefore, Hot Spot Policing works with the premise that by focusing police presence in certain areas in which criminal activities are taking place, the rate of crime within the entire city will be reduced. The key point is stationing officers where crimes are occurring.
Community Policing began in 1989 while David Dinkin was mayor of NYC. The police went into the districts and approached residents in order to speak with them and to obtain information on crime within their neighbourhood. The residents of a district often notice conflicts, crime, and especially the potential for crime, before the police ever would. Therefore, the idea was to begin working together with residents. Smith makes reference to a case in which an older lady would no longer go into the park with her grandchild, because drugs were being sold there. From her window, she was able to see where the drugs and the money were hidden, and passed this information on to the local police. This was possible because the local police made regular contact with residents of the district a foundational part of their work. Therefore, the police was able to find both the money and the drugs. If this starts occurring regularly, the drug dealer will most likely choose to end his activities in this area.
Before 1994, all the of the NYPD’s information management was centrally visualized and it took three months to process the data accrued during one month. Bratton began his mission to reduce the crime rate by providing each police station with a computer. Each station assigned one individual who was then responsible for entering and analyzing data concerning all crimes and reports. Thus, for the first time, police commanders played an important role in the fight against crime. Additionally, weekly CompStat meetings were held in which crime patterns, strategies to fight against these patterns within the different police districts and evaluations of these strategies were discussed.
Starting in 1994, a strong focus was initially placed on the reduction of drug-related crimes. Smith told about one CompStat meeting in which Jack Maple, the Deputy Police Commissioner for strategies to fight against crime, asked the police commanders when drug-related crimes tend to occur most. The officers answered that this type of crime usually occurs in the evening, on the weekends and in clubs. Maple then informed them that narcotics officers work between nine and five from Monday to Friday, and asked: shouldn’t we be deploying officers to where and when crimes are happening? The key CompStat element is about deploying police to places where crime is happening. This concept was already applied within Hot Spot Policing. An additional field of application is in the localization and mapping of data. Based on this, 4000 additional police officers were deployed to Brooklyn-North, as this was the area characterized by the highest concentration of drug-related crimes.
Another example for the strategic application of CompStat was within efforts to reduce car theft, which was one of the biggest criminal businesses in New York. The strategy did not focus on predicting individual crimes, but rather on understanding the system as a whole. In order to increase profits, the stolen cars were taken to so-called Chop Shops, where they are taken apart into pieces and then re-sold. Therefore the police looked for and then shut down the Chop Shops, and arrested their owners. Another possibility for turning stolen cars over for a good profit was to ship these vehicles elsewhere. The Piers from which this was possible were also closed down. As a result, car theft was no longer profitable, and since 1990 it has therefore been reduced by 94% in New York.
Funding and Partners
Giuliani, NYC’s mayor, and Bratton, Commissioner of the NYPD, worked together very closely during the introduction of CompStat, and together, they implemented the strategy to fight against crime. After CompStat’s introduction, cooperation between these two positions continued to be important for the success of the approach.
Another key factor is the collaboration between different city offices. The city’s mayor should therefore set related priorities and goals, and, based upon these, decide which offices and departments must work together towards achieving the objectives.
One example that illustrates the importance of these types of collaborations is the construction of social housing. Smith spoke of a police commander who had explained that before the apartments could be moved into, building inspectors and fire inspectors had to check and approve the building. In contrast, these residential locations are never inspected by a police or security officer. Correspondingly, problems relating to crime are often one of the first things to show up. In order to prevent this, a police representative could evaluate the building from the perspective of public safety, and identify danger areas that could then be avoided.
Smith calls one of the challenges presented by the CompStat approach to data analysis the problem of ‘what gets counted counts’. Those factors which are measured and observed will have the greatest impact on the management approach that results from this analysis. Therefore, the danger exists that only some of the key factors are analyzed and other very important factors are not taken into consideration. Thus, many factors must be studied, and there must be constant feedback on whether changes in management actually have a positive effect. Before CompStat, this was the case, since the focus was placed primarily on the minimization of response times and the number of arrests. It is therefore necessary to implement a process of continual learning, such as, for example, the implementation of the weekly CompStat strategy meetings.
An additional measurement problem results from the fact that it is difficult to find an opportunity to measure crime prevention – to determine a value that depicts how many crimes it was possible to prevent.
In order to achieve the goal of preventing crimes in New York before they occur, people who behave in any way that seems at all suspicious are stopped by the police and questioned. Since the people who are questioned are consciously selected, this leads to the challenge that these individuals may get the feeling they are being unjustly suspected.
In this context, a major area of criticism in New York is concerned with the fact that an over-proportional number of African Americans, as well as people from Latin-American descent, are stopped for questioning. 88% of all victims and 90% of perpetrators in New York are either African-American or of Latin-American descent and the districts in which a large number of these population groups live have the highest crime rates in NYC. By using the approach of deploying police officers to those areas in which crime occurs, the majority of the ‘stops’ take place in these districts. Correspondingly, a small proportion of New York’s population, living in certain districts, receives the majority of the police presence and has the highest number of interactions with the police.
This leads to the danger of the relationship between the citizens in these areas and the police becoming fraught with negativity. Thus, the challenge is to increase the citizen’s awareness that the police are there to protect them, and it becomes particularly important that during direct interactions the police behave with the utmost professionalism. In addition, technologies which may assist in improving this process would be very useful. For example, detectors can be used during searches conducted on passers-by, in order to prevent unnecessary physical contact. Additionally, the police’s interactions with the citizens must be able to be scrutinized, which can be accomplished through the increased use of video recordings and careful note-taking during questioning.
Impact and transferability
The city does not stand on its own, and during the development of strategies the environment – and the context in which these strategies will be applied – must also be considered. The key factor for successfully reducing crime is not being able to predict individual criminal activities. Rather, it is about understanding the thing as a whole, such as for example car theft, and developing strategies that effectively interfere with this business, such as closing down the Chop Shops. Through the analysis of data, the idea is to identify which tactics are effective in different contexts.
An additional important factor is the collaboration between different public offices. These must share their experiences with one another, and in particular, should check on the interplay of different sectors and their effects on specific strategies. The citizens of the city should also be included in such processes.
The foundational idea behind CompStat is to change the management of police work from reactive to preventive. Important features of this type of management are decentralized leadership, focusing on results and the goal of reducing the crime rate. A critical factor in doing so is that data are not only collected, but also utilized and analyzed. They should be used to develop strategies and also to monitor whether these strategies are, in fact, contributing to the achievement of set objectives. Therefore, it is important that this approach is conducted as a continual process of learning. Of additional importance is the mapping of data and the use of these to fine-tune strategies accordingly.
In order to transfer the evidence-based approach used in New York to other cities, the first step would be determine what data is available. Each city’s police department collects certain data during their daily operations, which could be used within this approach. If this is not already available, initial data should be collected in order to determine in which areas an over-proportional number of crimes tend to occur. The strategies of the local police must be adapted to correspond to this data.
One example of the successful transfer of this approach to another city is the case of Los Angeles, where William Bratton, after leaving NYC, became Chief of Police. Once again, Bratton was able to dramatically reduce the rate of car theft; however, he accomplished this using an entirely different strategy than what he had used in New York. The key to successfully transferring CompStat is the identification of the problem as a whole and the development of a corresponding strategy which is appropriate for solving the problem. Continually testing, evaluating and adapting the strategy is also a critical component of its success.
Dennis Smith talked about a Crime Strategy Meeting held by Bratton during the early part of his work in Los Angeles and during which different categories of crime were discussed – amongst these was car theft, since this represented a major problem within the city. Around 11:30 in the morning, Bratton asked ‘so, where are your Chop Shops?’ and his employees thought he was announcing the lunch break, since, in the South of the United States the term ‘Chop Shop’ refers to a store in which items such as steaks can be purchased. Following this, Bratton explained how car theft was organized in New York, and that Chop Shops functioned as an essential part of the system, used to take apart cars and re-sell their individual parts. His police officers answered: ‘here in Los Angeles we get the cars back; they are not taken apart. We have a terrible public transit system here in New York, and when young people go out in Los Angeles, and then need to get home, they break into a car and use it to drive themselves home. They are often drunk and crash into things or are involved in accidents, but we get the car back.’ Thus, Bratton was not able to apply the same strategy he had used in New York, and was forced to develop an adapted approach. The method used in Los Angeles was composed of analyzing where and when certain types of cars were stolen. Then, the police placed ‘bait cars’ of the particular make at the appropriate locations and observed these closely. With a high certainty, young people would come up to the cars and attempt to steal them, at which time it was possible to arrest them. Soon, it was known all over Los Angeles that the police were cracking down on car theft, and the rate of these types of crimes was reduced considerably.
Therefore, a generalizable strategy can be formulated which states that the problem must be understood as a whole, and aspects must be identified through which patterns can be disrupted.
Smith, D. C. (2009). Making Management Count. Journal of Policy Analysis and Management, 28(3), 496–496. doi:10.1002/pam.20443
Smith, D. C., & Bratton, W. J. (2001). Performance Management in New York City. In D. Forsythe, Quicker, Better, Cheaper (pp. 1–30). Albany: Rockefeller Institute Press.