Technical/Rational Perspective

According to the technical model, an organization's environment is characterized by precisely specified products or services. These are exchanged in a market, and organizations operating in these environments are rewarded for effective and efficient performance (Reference ScottScott 1987:31). In response to these technical pressures, the organization develops formal structures to organize work processes rationally: positions, policies, programs, and procedures (Reference Ritti and MastrofskiRitti & Mastrofski 2002).Footnote ³

COMPSTAT's implementation in the NYPD closely resembled the technical perspective's conception of technical or functional rationality. When Bratton assumed command, his vision for the organization appeared to be predicated on the assumption that the department operated in a market economy and was impelled to be as productive as possible by market mechanisms:

COMPSTAT was the centerpiece of his reform efforts (Reference SilvermanSilverman 1999). Bratton focused on developing structures and practices to enable his organization to maximize efficiency and drive crime down. Along with his advisors, he thoroughly assessed the agency's problems by reviewing existing policies and operations, and by identifying key issues and obstacles through focus groups and surveys. Next he assembled management teams to gather and evaluate all this information before making recommendations (Reference HenryHenry 2002:187–234).

To optimize operations, Bratton borrowed state-of-the-art management doctrines from organizational development experts in the private sector, Michael Hammer and James Champy (Reference Bratton and PeterBratton 1998:224; Reference SilvermanSilverman 1999:82). The thrust of their “reengineering” approach is that successful organizations respond to uncertain environments by reprioritizing their goals and revamping core structures for the accomplishment of results. This involves several principles: developing a management commitment and capacity to establish priorities, securing the allegiance of employees, finding novel ways for accomplishing tasks, and using information scientifically to drive decisionmaking (Reference Hammer and ChampyHammer & Champy 1993:32–47, 85). Bratton based his COMPSTAT strategic problem-solving model on these principles (Reference Bratton and PeterBratton 1998:224; Reference SafirSafir n.d.).

Institutional Perspective

Developed from an open systems conception of organizations and from Reference Berger and LuckmannBerger and Luckmann's (1966:53–67) emphasis on institutions as products of human activities and commonsense knowledge, the institutional perspective challenges the view that organizational structures emerge from rational processes (Reference Meyer and RowanMeyer & Rowan 1977; Reference ScottScott 1995). According to this model, the structures and practices of organizations are not only influenced by rational calculations and technical imperatives, but also by the cultural features of their environments (Reference ScottScott 2004:6). Certain organizations operate in environments that are not well-developed technically: their products or services are not well-specified, methods for their production are not well known, and competition is weak or nonexistent. Many public service organizations (Reference WilsonWilson 1989), including the police, tend to fall into this category (Reference CrankCrank 2003; Reference Mastrofski and BrodeurMastrofski 1998).

Rather than being driven to perform efficiently, these organizations are judged by how well they respond to wider beliefs about what such organizations should look like and what work they should be doing. Meyer and Rowan refer to the taken-for-granted cultural expectations that suffuse the organization's institutional environment as “rationalized myths.” These myths stipulate the “social purposes” of the organization and operate as formal prescriptions for their accomplishment (1977:343–4). Since these social constructions of reality may emanate from and be enforced by politically salient actors or “sovereigns” in the external environment (typically the press, community groups, and public officials), organizations are under considerable pressure to incorporate them into their formal structures and activities (Reference Crank and LangworthyCrank & Langworthy 1992:342). By closely integrating their internal structural arrangements to myths of what is the “proper” or “natural” way to behave, institutionalized organizations gain legitimacy, thereby improving their prospects for resources and survival (Reference Meyer, Meyer, Scott, Deal and RowanMeyer et al. 1983:46; Reference Suchman and EdelmanSuchman & Edelman 1997:919; Reference Tolbert and ZuckerTolbert & Zucker 1983:25).

This is not to suggest that highly institutionalized organizations experience only one set of pressures through which structural conformity (isomorphism) occurs. Reference DiMaggio and PowellDiMaggio and Powell (1983:150–2) distinguish three mechanisms of isomorphism: mimetic, normative, and coercive. Organizations confronting a great deal of uncertainty in their environments are especially prone to mimetic isomorphism. To help ensure their survival, they copy other organizations that have received recognition and support for appearing effective. Alternatively, an organization may seek approval and resources by adopting measures that rest on the authority of professionals and experts in the field (normative isomorphism), or by putting certain structures in place in response to pressures from another organization (coercive isomorphism).

Whatever their source, the cultural forces exerted on organizations can result in the implementation of structures that conflict with technical demands for efficiency (Reference Meyer and RowanMeyer & Rowan 1977:356). Thus, unlike the technical model, the institutionalized organization benefits from decoupling its structures from its core routine tasks so that its structures can be more closely aligned with institutional values (Reference Meyer, Meyer, Scott, Deal and RowanMeyer et al. 1983:47). This allows the organization to focus on its functional activities while simultaneously employing means that have little do with its actual day-to-day work (Reference Maguire and KatzMaguire & Katz 2002).

We have distinguished between the specific aspects of each type of environment, but it is important to remember that the technical and institutional perspectives are not opposites on a continuum but rather different conceptual dimensions (Reference ScottScott 1987:126). Organization structures, such as those that constitute COMPSTAT, may respond either to technical or institutional pressures, or to both simultaneously (Reference Meyer, Meyer, Scott, Deal and RowanMeyer et al. 1983:61–2). “The fact that a structure is perceived to produce a technical/rational result, does not preclude it from serving an institutional function as well, and vice versa” (Reference Mastrofski and RittiMastrofski & Ritti 1996:293). At issue here is how much structures are directed toward the achievement of a valued goal (“rational”), and how much they are understood to produce the desired results (“technical”). Just as decoupling suggests an institutional function, so too does an organization's unwillingness or inability to assess critically existing organizational factors and the promise of enhanced technical performance.

There is a small but growing literature on the applicability of institutional theory to police organizations (Reference CrankCrank 2003), but much of it asserts the superiority of institutional theory in accounting for police structures and practices without conducting an empirical analysis that compares the strength of this perspective to the technical model. The institutional perspective has been applied to explaining certain police organization structures and practices, the emergence of community policing and racial profiling, the practices of anti-gang units, and accounting for the effects of police training on drunk-driving enforcement (see Reference CrankCrank 2003 for a review). While this literature offers an intriguing and provocative perspective of the structures and practices of policing, it does not offer thorough and conclusive empirical tests of the superiority of institutional theory. Indeed, the popularity of the technical model as a driving force among researchers is evidenced in the proliferation of studies designed to test the effects of police practices on crime and disorder, and to identifying “best practices” through scientific study (Reference CrankCrank 2003:189). A survey of police research during the latter half of the twentieth century found that by the 1990s, evaluations of the effectiveness of police strategies had come to dominate police research, especially research that was government-funded (Reference Wesley and KathleenNational Research Council 2004: Ch. 2). COMPSTAT is currently one of the most touted crime control police innovations, yet its dynamics remain untested. Because it is far from obvious how technical and institutional processes may influence the general direction of organizational change (Reference ScottScott 1987:156–7), our comparative empirical analysis of these two models better explains COMPSTAT and also assesses the utility of these theoretical approaches.

Mission Clarification

For police agencies to be strategic, COMPSTAT assumes that they must have a clear organizational objective. This runs counter to the ethos of another powerful police reform already well under way: community policing. Community policing seeks to complicate the mission of the police—to pursue diverse goals and constituencies simultaneously (Reference Greene and SkoganJack Greene 2004; Reference Moore, Tonry and MorrisMoore 1992). However, COMPSTAT seeks to narrow the organization's goal to a sharp focus on crime control, to set a rational measure or benchmark for success, and to establish a single belief system to which all members are strongly committed (Reference BrattonBratton 1996, Reference Bratton and Peter1998). When he assumed command of the NYPD, Bratton announced the objective to reduce crime by 10 percent his first year, and 15 percent the next (Reference Bratton and PeterBratton 1998:202). In addition to stating one core value and promoting high expectations, the power of the mission as a belief system rests on making its message highly visible and underpinning its symbolic importance with practical mechanisms for its achievement (Reference Bratton and PeterBratton 1998:202; Reference SilvermanSilverman 1996:4).

According to the technical model, patrol officer commitment to the organization's core objective of fighting crime should have been highest in Minneapolis, where a specific strategy was used to guide decisionmaking. The MPD set a precise crime reduction goal to decrease crime by 10 percent in 2000, announced this goal in daily roll calls and weekly COMPSTAT meetings, publicized it in the local press, and reinforced its importance with a day of departmentwide training on COMPSTAT. The mission was further supported with the widespread implementation of a specific strategy: zero-tolerance policing facilitated through the practice of directed patrol. In Minneapolis, line officers were expected to spend their uncommitted time patrolling problem areas identified by the district commander. They were directed to enforce the law for minor offenses in order to reduce serious crime. Although establishing high expectations for performance, Lowell's mission to become the “safest city of its size in the United States” was more visionary and less specific than Minneapolis's. Newark's chief did not establish a specific benchmark, preferring a more general announcement that everyone in the organization was to focus on “crime and service citizen's complaints,” and that COMPSTAT was a means of refocusing attention on this “core responsibility.”

Although Minneapolis's mission statement was “concise, value-laden, and inspirational” and buttressed with a variety of goal-directed mechanisms to ensure that it could not be dismissed as an empty slogan (Reference SimonsSimons 1995:82), rank-and-file officers showed the weakest buy-in to the department's redefined mission. Only about half said the organization's crime reduction goal was “very important” to COMPSTAT, compared to more than three-fourths in the other two cities (see Table 1). Institutional theory can help explain this paradox by anticipating conflict between COMPSTAT's simple crime-fighting mission and the diverse set of values and approaches embraced by community policing (Reference Roth and SkoganRoth et al. 2004:20).

Table 1. The Importance of Reducing Violent Crime to the Department's COMPSTAT Strategy

All three departments supported community policing, but Minneapolis had allocated the greatest share of department resources to this approach. This included decentralizing its crime prevention bureau to the districts and implementing a program to make management responsible for officers' relationships with community members (Reference WillisWillis et al. 2004a). Thus, Minneapolis did the most to infuse both community policing and COMPSTAT into its structures and practices, and to the extent that it did so, it created an environment where the competing approaches would create conflict and dissension. Comments from our line officer survey illustrated this conflict. Some officers felt that COMPSTAT devalued important police services, such as fulfilling citizen requests for help. As one officer put it, “Answering 911 calls is not a priority,” while another stated, “[COMPSTAT] focuses on the minor offenses hoping for it to turn into something bigger … wasting time to stop people for loitering and not answering calls.” Others felt COMPSTAT denied residents equal access to police services by allocating more patrols to high crime areas, thereby reducing patrols in other areas.

These statements reveal how community partnerships and service competed with crime control through surveillance and law enforcement, and how answering calls for service, though valuable to community policing, conflicted with directed patrol. Hence, the MPD sent conflicting signals. By contrast, neither Lowell nor Newark had made a significant attempt to reallocate resources to crime control and continued to stress the general service function. By decoupling COMPSTAT from other core operations, the rank-and-file at these other two sites experienced less disconnect between community policing and COMPSTAT.

The undesirable outcome in Minneapolis and the absence of significant change in Lowell and Newark both reveal the limitations of a technical explanation for what we observed. COMPSTAT's emphasis on goal specificity and formalization ignores powerful political and cultural trends valorizing a multifaceted police mission (Reference Mastrofski, Weisburd and BragaMastrofski 2006). In its narrow restructuring of the police mandate, COMPSTAT clashed with community policing. Moreover, the strength of the institutional environment helps explain why Lowell and Newark failed to change existing structures to support COMPSTAT's crime-fighting focus. These departments coped with the pressure of having to provide a wide variety of police services by decoupling COMPSTAT's focus on crime fighting from the “realities of the workplace” (Reference Wasserman and MooreWasserman & Moore 1988). In doing so, their officers were buffered from the same kinds of competing demands that were a potent source of frustration in Minneapolis.

Internal Accountability

Bratton reasoned that for a department's mission statement to be effective, there needed to be a mechanism to hold police officers responsible for accomplishing the goals that it espoused (Reference Bratton and PeterBratton 1998). Regular COMPSTAT meetings provide a rational inspection and evaluation system for identifying middle managers' primary responsibilities, monitoring their activities, and providing positive or negative consequences for their performance (Reference McDonaldMcDonald et al. 2002).

Our fieldwork showed that regular COMPSTAT meetings did make district commanders feel directly accountable for being well-informed about crimes in their beats and taking action. Across sites, these meetings closely resembled the NYPD. The top brass asked probing questions about crime problems and expected immediate answers. District commanders acknowledged that it was their responsibility to come to COMPSTAT fully prepared with responses to the chief's queries. For example, a district commander in Lowell commented that COMPSTAT was a way of “keeping them honest,” since “having things up there on a map can show you how bad things are, and you cannot say, ‘Ooh, I missed those reports; I did not see them.’” District commanders at other sites told us they spent several hours poring over police reports and maps in preparation for the upcoming meeting—a significant change from past practices. Our observations also suggested that COMPSTAT's efficacy as an accountability mechanism depended upon district commanders not wanting to perform poorly before their peers and other high-ranking personnel.

These changes notwithstanding, the technical model's promise of sweeping reform was unfulfilled. COMPSTAT's accountability mechanism was limited in its scope and effect on the quality of police work. Its exclusive focus on middle managers, especially at COMPSTAT meetings, meant that lower-ranking officers did not experience a similar degree of accountability. About 40–50 percent of patrol officers responded that holding officers (as opposed to district commanders) accountable for crimes in their beats was not at all important to their department's COMPSTAT program (see Table 2). That this feature was significantly more important to patrol officers in Lowell may be attributed to the department's size. Here patrol officers were more likely to feel visible to the top, hence personally responsible for good performance, than in the two larger departments where responsibility was shared by many.

Table 2. The Importance of Holding Officers Accountable for Crimes in Their Beat to the Department's COMPSTAT Strategy

^** Does not sum to 100 percent due to rounding

Furthermore, in none of these departments did COMPSTAT reforms include efforts to get the rank-and-file to respond to the direction of middle managers. It is here in the less visible aspects of organization structure that the accountability chain remained unchanged and tenuous. When asked “How often does your supervisor discuss what happened at COMPSTAT meetings?” 61 percent of those officers we surveyed in Lowell responded “never” (43 percent) or “every few months” (18 percent); for Minneapolis, 34 percent responded “never” (14 percent) or “every few months” (20 percent), and for Newark it was 42 percent (8 and 34 percent, respectively).

In addition to its limited range, internal accountability was a dull instrument for encouraging major improvements in police performance. District commanders were assessed primarily on their ability to provide a response to the chief's inquiries, not on the quality of that response. They demonstrated their competence by providing details of individual crime problems, including where and when crime occurred, and the age, race, and sex of any suspects. In comparison, they spent little time explaining the decisionmaking process that led them to choose a particular crime strategy. Indeed, most of the middle managers focused on having facts and figures at their fingertips for the formal meeting, rather than finding ways to prevent and control crime in their districts on a day-to-day basis. Furthermore, COMPSTAT did not strengthen control over lower-ranking officers who continued to exercise the same high level of discretion long recognized as a characteristic of police work (Reference MuirMuir 1977; Reference ReissReiss 1971).

The technical model cannot easily account for these deficiencies, or each department's lack of attention toward making accountability more effective. At minimum, it appears that a rational response to improving control over the quality of police work would require the systematic collection and analysis of crime data. These could then be used as specific performance indicators of the effectiveness of a particular district commander's or police officer's response. However, we did not observe any site using this more rigorous form of assessment. From an institutional perspective, this makes sense. As highly institutionalized organizations, police departments are likely to minimize inspection and evaluation of their work and output. A more rigorous assessment mechanism could well uncover inconsistencies and failures that undermine the organization's legitimacy (Reference Meyer and RowanMeyer & Rowan 1977:359). Moreover, the discrepancy between internal accountability and actual day-to-day activities illuminates COMPSTAT's considerable ceremonial value, or “dramaturgic” appeal (Reference ManningManning 2003:123). While none of the respondents at the three sites could recall a district commander actually losing his or her position for subpar performance, this did little to assuage the perception among all ranks that COMPSTAT meetings were punitive. In reality, COMPSTAT meetings were rarely confrontational, but COMPSTAT's reputation sent a powerful message to the community, just as it did with the rank-and-file, that key personnel were being held strictly accountable, even though the results of their efforts were unclear (Reference ManningManning 2001).

Geographic Organization of Operational Command

To speed communication and to minimize coordination problems between functionally specialized units, COMPSTAT delegates primary decisionmaking to commanders with territorial responsibility (districts). District commanders are also given a larger share of the department's resources to control. Specialist units are placed under the district commander, or arrangements are made to facilitate their responsiveness to the commander's needs. The empowerment of district commanders to make decisions about how best to reach objectives and use resources to “customize and implement their own crime-fighting plans” is a significant change (Reference SilvermanSilverman 1999:85). Under the traditional model of police organizations, commanders of functionally specialized units have tended to operate independently, with organizational resources divided between them (Reference Reiss, Tonry and MorrisReiss 1992).

All three departments had already decentralized geographically after adopting community policing. A commander exercised 24-hour responsibility over each district. However, in the MPD, COMPSTAT resulted in districts being further subdivided into sectors, each of which was headed by a sector lieutenant. Although Minneapolis had attempted to devolve decisionmaking authority farther down the chain of command, district commanders, as with the other sites, were ultimately responsible for overseeing operations and settling disputes within their districts.

All the district commanders believed that they possessed significant autonomy. They were ultimately responsible for choosing and implementing crime strategies, and they rarely had to justify their decisions. In Lowell, the district commanders were largely responsible for redrawing their beat boundaries during the department's bid process, and one district commander described his overall sense of independence as feeling “like the captain of my own ship.” Our observations suggested that this sentiment applied more generally to the MPD and NPD.

Though noteworthy, these attempts at streamlining decisionmaking did little to change the existing command-and-control structure of each department. This is not surprising: strong institutional pressures validate the existing organizational hierarchy, particularly at the highest levels. Because elected officials and the public hold top management accountable for subordinates' actions, administrators are extremely reluctant to cede too much control. Doing so increases the possibility that middle managers will engage in risky behavior, jeopardizing the well-being of the organization (Reference SimonsSimons 1995). Across sites, district commanders were encouraged to take initiative, but top management was willing to exercise decisionmaking authority over them, especially when it came to making critical decisions about resources. For example, Lowell's chief, under pressure from the community and the city to increase the total number of walking routes, overturned a district commander's decision to cut one of his officer's walking times in half. Similarly, officers walking the beat were sufficiently important to Newark's top brass that they made it almost impossible to cancel a walking post. In Minneapolis, only the chief or deputy chief had authority to move each district's community response team across districts.

Similar institutional pressures constrained the geographic redistribution of resources and at least in one department contributed to much resistance to change. In Lowell, Newark, and Minneapolis. some detectives were reassigned to the districts where they fell under a district commander's direct command, but nearly all the specialized units, including the central investigations units (CIUs), continued to operate physically out of headquarters and independently from the district commanders. In fact, in Lowell, detectives had successfully stymied department efforts to decentralize the CIU to the districts. The ensuing compromise had three to four detectives reassigned to each district, while approximately 11 detectives maintained their traditional autonomy and prestige by continuing to handle the most serious cases out of headquarters.

One explanation for why Lowell's CIU was able to resist decentralization is obvious: detectives were interested in preserving their independence. Group interests were undoubtedly an important motivating factor, but these alone do not provide a satisfactory reason for why a small unit could oppose a well-respected, resourceful chief with a track record of reform (Reference LehrerLehrer 2001; Reference ThacherThacher 1998). It seems likely that the power of Lowell's detectives to resist change derived from a culturally contingent concept of detective units as autonomous and centralized. There are no studies linking traditional investigative structures and practices to crime reduction (Reference EckEck 1983; Reference GreenwoodGreenwood et al. 1977; Reference Weisburd and EckWeisburd & Eck 2004:50), but criminal investigation units are infused with values about what criminal investigations should look like. Getting rid of them makes police organizations more vulnerable to claims that they are “negligent” or “irrational” (Reference Meyer and RowanMeyer & Rowan 1977:350). Thus whatever technical benefits decentralization might have offered (such as improving communication between investigators and patrol officers), institutionally derived pressures about the “appropriateness” of a centralized CIU did not favor change in Lowell.

In Minneapolis, a greater shift toward geographic management had been achieved, but not by reducing the number of specialist units (as predicted by the technical model). Minneapolis's shift toward decentralization came in the form of increasing the level of functional specialization. Under COMPSTAT, the department's CCP/SAFE unit (“Community Crime Prevention, Safety for Everyone”), responsible for problem-solving and community-building, was decentralized to the districts. Furthermore, each district was provided with two new community response teams (CRTs) tasked with tackling prostitution, narcotics offenses, and other quality-of-life crimes. The creation of extra units did not reduce the amount of attention district commanders had to pay to coordination issues. Now they had to pay added attention to coordinating crime responses within their district with the work of their new CRT and CCP/SAFE teams. In addition to increasing potential coordination problems, intradistrict specialization diverted resources from other core functions, especially patrol. Several of the rank-and-file suggested that giving a larger share of the department's resources to the districts undermined organizational performance. One officer commented that COMPSTAT failed to reduce crime since “more and more officers [were] taken off the street for ‘special’ units,” while another stated simply, “Too many specialized units—911 calls go unanswered.”

The failure across sites to shift away from a functional division of labor indicates the power of institutional over technical pressures on organizational change. The only department to improve significantly the resources available to district commanders did so by increasing the role of specialized units, albeit at the district level. Such an approach to the geographic organization of operational command failed to reverse the kinds of coordination and performance problems COMPSTAT claims to overcome. This anomaly can be explained by the powerful attraction of specialized units as a highly touted aspect of police work. These units signal to the wider community that police are taking a particular problem seriously, thus playing an important role in conferring legitimacy on police departments (Reference Crank and LangworthyCrank & Langworthy 1992:343–4; Reference KatzKatz 2001).

Organizational Flexibility

COMPSTAT promises a rapid and effective response to incipient problems. Since these are unpredictable and a department's resources are insufficient to staff for all contingencies, the organization needs to be adaptable to changes in its environment. This requires that the organization develop the capacity to change or disrupt existing structures and routines in response to nonroutine work demands.

Our three police agencies showed significant variation in the levels of organizational flexibility, but in general they struggled with the need to create a more flexible organization. In Lowell, the most common practice of reallocating resources outside of normal patterns was for district commanders to do so on an ad hoc or informal basis. This minimized disruptions to department routine. The most likely occurrence was for the district commander to direct a patrol officer or detective away from more routine activities during their shift and to ask them to pay particular attention to a specific area.

This ad hoc reallocation was also a common feature in the NPD and MPD, but these departments had taken greater strides to increase organizational flexibility through reliance on task forces and “taxi squads” (portable in where and when they work). The NPD and MPD's increased capacity for organizational flexibility can be attributed to their larger size, thereby having more resources to allocate to emerging crime problems. In Newark, the department constantly formed and disbanded special task forces to respond quickly to major problems as they arose. The chief and his deputies, not the district commanders, directed these task forces. However, flexibility was greatest in Minneapolis, which had facilitated flexibility through the assignment of community response teams to the districts, and district-level directed patrol.

Each district was permanently assigned a community response team that did not have a specific geographic assignment but was available to the district commander to use when and where she or he pleased in the district. Freed from responding to 911 calls (unless specifically requested to do so), the CRT provided each district commander with, as one officer stated, “fluidity of resources.” Similarly, an important feature of the department's COMPSTAT program was directed patrol. District commanders responded to crime patterns by assigning their patrol officers to these problem areas and encouraging them “to vigorously enforce misdemeanor (quality-of-life) crimes.” Given the high priority accorded directed patrol within the department, district commanders were less constrained by the “tyranny of 911” (Reference SparrowSparrow et al. 1990). Consequently, they were able to allocate more patrol resources to proactive and preventive strategies than under more traditional policing models.

These changes indicate, as the technical model suggests, that some organizational routines were changed and structures implemented as a rational response to technical or crime pressures. Following this reasoning, we would expect each organization to have used COMPSTAT in the way that it was intended, namely as a rational mechanism for the more effective allocation of personnel, overtime funds, and equipment between districts. In this way, up-to-date crime data would have driven resource allocation. According to our observations, however, this was not the case. In Lowell and Newark, department resources were divided between districts roughly in proportion to their population and/or the area they covered, while in Minneapolis, multiple measures (calls for service, reported crime, and 911 calls) were used in allocating staff. Since COMPSTAT comprises crime data, Minneapolis was the only department to use, in part, COMPSTAT data to draw beat boundaries.

Nonetheless, none of the departments relied upon COMPSTAT data available to determine the resource allocation process among districts. When a Lowell commander suggested that COMPSTAT be used more “scientifically” (that is, rationally) as a means of allocating manpower, he provoked little response. Our observations suggested that district commanders favored workload equality over the rational, yet discriminatory, allocation of resources in pursuit of the crime control mission. More important, shifting significant portions of the organization's resources would play havoc with the stability of units, the established hierarchical structures, and the efficient delivery of noncrime services important to the public. Consequently the sharing of officers between districts was infrequent and planned well in advance.

Organizational flexibility was not only hindered by conventional considerations about the distribution of officers to patrol, but it was also shaped by other nontechnical criteria. City politics powerfully influenced officer deployment. A Lowell city councilor expressed concern about drug-related crimes in one area. He argued for increased police visibility in this neighborhood and told the chief and city council that he wanted to see “more cops walking.” Placing a higher priority on walking routes reduced the number of cruisers available in other areas of the city where crime was higher. In addition to the city council's position on foot patrol, the police department was under pressure from powerful sovereigns, such as the editor of the local newspaper and local business owners, to maintain a high level of police visibility in the downtown area. In Minneapolis, the district that served the downtown business area received a similar level of resources as more crime-ridden districts. We can speculate that this was due to pressures on the department from businesses, politicians, and other special interests. Although we did not observe this directly, one commander was clearly disappointed that the reallocation of resources was so dependent upon these concerns. He could not understand why these decisions relied on “taxpayers as opposed to victims or people who live on a block with drug dealers.”

In sum, random preventive patrol and the calls-for-service apparatus that dominate the organization of patrol work in most urban police departments were largely unchanged by COMPSTAT reform. Even the creation of CRTs in the MPD only afforded a modest increase in flexibility, given how small a percentage of the uniformed force these in fact represented. In order to fulfill its pledge to increase organizational flexibility, COMPSTAT needs to do much more than just decentralize some resources to district commanders. It requires a fundamental change in deeply held beliefs about the efficacy of preventive patrol and standard workload measures that largely determine who will work where and when. It also requires that powerful sovereigns no longer constrain police managers to allocate resources to places and jobs that would not have received priority according to the dictates of the COMPSTAT process. At these sites, COMPSTAT's technical features did nothing to buffer the organization from the pressure that sovereigns exerted to affect the distribution of services. Absent these changes, institutional objectives supersede the technical requirements of police reformers.

Data-Driven Problem Identification and Assessment

Crime statistics are crucial to COMPSTAT as a more efficient way to structure management decisionmaking, inasmuch as they shift the focus from selective, anecdotal accounts of individual cases (Reference SafirSafir n.d.:2). This is accomplished through the collection and analysis of timely crime data. These data are also used to reassess police strategies. Underlying this element is the belief that crime can be reduced more effectively through proactive policing and an attack on underlying sources of criminal activity than through arresting perpetrators after a crime has occurred. Data-driven decisionmaking is a highly touted feature of COMPSTAT and one that appropriates elements of another widely recognized and accepted policing reform—problem-oriented policing (POP) (Reference GoldsteinGoldstein 1990; Reference SilvermanSilverman 1999:123–4).

Before COMPSTAT, these departments merely conducted an annual review of index crime rates collected for the FBI's UCRs. Compared to this practice, all three departments made significant advances in the speed that data were organized, aggregated, and analyzed. The primary sources of crime data were police incident and arrest reports and CAD (computer-aided-dispatch) data. Members of each department's Crime Analysis Unit (CAU) selected the data on those crimes regularly presented at COMPSTAT meetings and entered them into a database using a data management program. These data were generally available the following day. A day or two before a COMPSTAT meeting, the CAU distributed summary tallies and computerized maps to the district commander to help prepare for the upcoming meeting.

The crimes presented at COMPSTAT meetings varied across sites, but they generally included a range of the most serious index violent and property offenses. The data entered by crime analysts included field report number, district, day of the week, and location. Additional information that could also be useful in identifying crime patterns was also entered (e.g., point of entry for a burglary). Minneapolis differed from Lowell and Newark in that within 24 hours of entering a police report, the department mapped the location of the incident on the department's mainframe. District commanders could then access these maps and try to discern crime patterns or hot spots (concentrations of crime in specific geographic areas) as they emerged. COMPSTAT did produce substantial changes in the timeliness by which crime data were made available to police in all three locations, and it heightened sensitivity to the geographic distribution of crime, but it did not otherwise alter the nature of that information in ways that would meet the needs of POP. Crime classifications remained based on statutory definitions rather than those that would facilitate identifying the causes of a problem and devising the most effective interventions (Reference GoldsteinGoldstein 1990:38, 67). Meeting those expectations would have required profound changes in the ways that the organizations gathered and classified crime information, and that would have also undermined the standardized approach that COMPSTAT took to ensuring internal accountability at its periodic review sessions.

Timely crime data are intended to improve the technical environment of police, moving departments toward an empirically driven problem-solving approach. Still, the capacity of this data-driven approach to produce results does not depend on the availability and impressive presentation of crime data, but on how much these data are used to restructure daily decisionmaking. Police managers should thus know how to analyze crime data; if not, they should be trained to do so. According to COMPSTAT's supporters, learning how well and why a particular strategy works allows the organization to become smarter, more adept at reaching its objectives (Reference SilvermanSilverman 1999:192). And systematic follow-up helps the organization use its resources more intelligently because it can redeploy resources quickly once a problem has abated (Reference SafirSafir n.d.:2).

Our fieldwork showed that COMPSTAT data did improve the speed and focus of each organization's response to emerging problems but did not result in the police doing things they had rarely or never done. Under COMPSTAT, district commanders, now highly sensitized to knowing what was going on in their areas, read officer reports daily to familiarize themselves with crime problems, identify crime trends, and mobilize resources. But they did not use the more sophisticated tools available to search for and characterize patterns in crime; they continued to rely on the traditional tools of the trade—personal knowledge of people and places, serially reviewing individual officer reports, and forming impressions directly from these undifferentiated data (Reference RubinsteinRubinstein 1973; Reference Van and JacobVan Maanen 1974). Daily crime maps were not available in Lowell and Newark, but this did not trouble the district commanders. Many responded that they could map crimes in their heads as they were occurring. Even in Minneapolis, where daily crime maps were available, we observed only one district commander use them frequently.

Taken-for-granted organizational routines and practices were not the only limitations on crime analysis. Middle managers were not recruited based on their analytical skills, and those who were selected were not given much training in crime analysis, nor were they provided with support staff trained in criminology, research methods, statistics, or crime mapping: they were simply expected to learn a complex, new technical task on the job. This explains why we did not observe any district commanders conducting an in-depth analysis of crime data to determine the underlying causes of problems, or how to mobilize (that is, deciding exactly what to do). The tendency toward superficial crime analysis that we observed at these three sites was replicated in 16 other police departments observed by the Police Foundation team in brief site visits (Reference GreenspanGreenspan et al. 2003) and has also been reported in an in-depth case study of problem-oriented policing (Reference Cordner and BiebelCordner & Biebel 2005) and in a national survey (Reference O'Shea and KeithO'Shea & Nicholls 2003).

Finally, contrary to the technical model, follow-up was not intensive. All three sites relied on using maps or crime statistics at COMPSTAT meetings to identify whether crime activity had stopped or diminished. If a crime pattern disappeared, it was considered resolved. Identifying the short-term disappearance of a crime pattern and conducting simple pre-post comparisons of crime data did not provide rigorous assessment of why a particular strategy worked (Reference Mastrofski and BrodeurMastrofski 1998:176). This lack of follow-up across sites greatly impeded the capacity of each organization to learn much about the long-term effects of any intervention.

In many ways COMPSTAT had just shifted traditional reactive policing to a higher level. Police did not respond to individual crimes, but they reacted to crimes in the aggregate, producing a “whack-a-mole” phenomenon of placing highest priority on doing something quickly whenever a crime spike surfaced (Reference WillisWillis et al. 2004a). The knowledge-based approach, held by COMPSTAT's architects as essential to its effectiveness, would have taken each organization far from customary practice. Although more police are becoming college-educated, the majority are not, and regardless, the vast majority do not have extensive training in crime mapping (Reference Wesley and KathleenNational Research Council 2004:90). Significant time and resources would have to be invested in developing managers' analytic skills for any chance at realizing the vision of COMPSTAT's supporters. This extensive training regime would take district commanders from their other numerous and necessary tasks, placing a considerable burden on the organization. And it would disrupt traditional career advancement avenues that place a high value on learning the craft on the street before advancing to higher rank.

As a result of these limitations, the institutionalized response of all three departments was only a weak alignment of data structures with actual activities. This lack of coordination allowed our three sites to use data in ways that corresponded with the powerful institutional myth, one accepted and promoted generally by police themselves, that the police are highly trained and effective crime fighters (Reference ManningManning 2003). Improvements in data management systems, an intense focus on serious crime and noncrime problems, and the appearance of expert crime analysis signified that COMPSTAT was a sophisticated means of tracking and solving crime when in reality little had been done to transform existing decisionmaking processes.

Innovative Problem-Solving Tactics

The application of innovative responses to crime problems is arguably the most highly publicized feature of COMPSTAT and one that also overlaps with POP (Reference GoldsteinGoldstein 1990; Reference SilvermanSilverman 1999:123–4). Under COMPSTAT, middle managers are expected to select crime solutions that offer the best prospects of success, not because they are what has always been done, but because a careful consideration of a number of alternatives has showed them to be the most effective (Reference ShermanSherman 1998). This is a radical departure from the traditional model of policing, which is almost wholly reliant on random patrol, rapid response to calls for service, and a standard set of law enforcement powers (Reference Wesley and KathleenNational Research Council 2004). The best police response might include traditional enforcement tactics, but this should only be determined after considering other possibilities, such as altering the physical environment, mediation, mobilizing the community for social control, and so on. Thus under COMPSTAT police are expected to solve the underlying causes of crime problems by looking beyond their own experiences, by drawing upon knowledge gained in other departments, and by applying innovations in theory and research about crime control and prevention.

The technical model could not explain our findings that COMPSTAT had done very little to change existing crime-fighting strategies. We did witness some innovation, such as Lowell's successful use of a problem-oriented approach to shutting down a dilapidated, crime-ridden rooming house, but this was the exception, not the norm. The vast majority of problem-solving approaches identified at COMPSTAT meetings relied on traditional police strategies—in particular, asking patrol officers to identify suspects and keep an eye on things, saturate an area, and increase arrests. This is consistent with recent findings on problem-solving in San Diego (Reference Cordner and BiebelCordner & Biebel 2005) and a recent national survey of COMPSTAT (Reference WeisburdWeisburd et al. 2001). For example, during a COMPSTAT meeting in Lowell showing that the street-side windows of several parked cars had been smashed, the chief asked, “What kinds of things have we done in the past?” His deputy suggested they had clamped down on motor vehicle violations: “You know, chief, sometimes you just get lucky. You catch a kid and they just talk.” Similarly, when we asked a district commander in Newark what he had done regarding a spate of violent crimes, he told us, “Good old police work,” which included increasing police visibility.

There appeared to be some small but observable differences between these organizations regarding their capacity to facilitate innovation. Lowell's chief strongly encouraged his command staff to share ideas on crime strategies, and Newark task forces provided a similar forum. Opportunities for innovation were most limited in Minneapolis, where the importance accorded the combination of directed patrol and zero-tolerance policing, and the department's focus on crime numbers (not strategies), stifled the exploration of problem-solving alternatives. Elsewhere it has been suggested that this lack of innovation can be attributed to COMPSTAT itself, most importantly its accountability mechanism (Reference WeisburdWeisburd et al. 2003; Reference WillisWillis et al. 2004b). This pressures district commanders to identify crime problems quickly and to implement speedy responses. Such an approach discourages the careful sifting of and deliberation about patterns of crime, and a careful review and discussion of the benefits and drawbacks of various approaches.

However, our comparison between the technical and institutional models provides a broader theoretical explanation for the disparity between COMPSTAT's prescriptions and how it was practiced. From a technical perspective, the reluctance of middle managers to use evidence-based strategies instead of those based on their own experiences could be attributed to the underdevelopment of policing's technical environment. Because there is much uncertainty about the effectiveness of various crime control techniques, middle managers would be unlikely to implement the more innovative ones. The trouble with this explanation is that it fails to account for middle managers' failure to take advantage of advances in departmental crime analysis and consult published research on promising alternatives. Mobilizing this knowledge may have provided a better basis for choosing the strategy most likely to produce desirable results than drawing conclusions based on personal experiences. A growing body of evidence suggests that the standard police responses are generally less effective than focused strategies tailored to specific problems (Reference Weisburd and EckWeisburd & Eck 2004).

Pressure from the highly institutionalized environments of police organizations provides a more compelling explanation for why innovation was so rare. Politicians and local citizens expect the police to respond to crime and disorder immediately, with appropriate, “effective” methods that have received widespread legitimacy regardless of their actual efficacy. Meanwhile, powerful beliefs among police about the value of personal experiences hinder the systematic mobilization of information to guide police practices (Reference Bayley and BittnerBayley & Bittner 1984). In this context, the time-consuming search for equally unproven but less legitimate alternatives is an extremely unlikely outcome. A police department that pursues experimentation and accepts the risk of failure associated with such trial and error seriously jeopardizes its external support. Thus the current institutional environment undermines the kind of experimentation on which the development of long-term knowledge about what works depends.