Compstat design.

The Compstat process, as described in the first part of this article, hinges on four crime-reduction principles: accurate and timely intelligence, effective tactics, rapid deployment of personnel and resources, and relentless follow-up and assessment. (1) Coupled with these are accountability and discretion at all levels of the law enforcement agency. The design of the Compstant model creates and atmosphere where both officers and executives can remain focused on the core mission of the agency, protecting the members of their community.

[ILLUSTRATION OMITTED]

THE COMPSTAT DESIGN

The chief is absolutely critical to Compstat's design and success. He must sponsor and champion the process with the command staff. (2) "Sponsoring [Compstat] and championing it are different; sponsorship is necessary to provide legitimacy to the process, while championing provides the energy and commitment to follow through. [Compstat] does not just happen--involved, courageous, and committed people make it happen. The department's leaders must serve as process champions. These people must believe in the [Compstat] process and be committed to it." (3)

When designing the Compstat model for the organization, those involved in the process must sort through a few administrative details. These include organizational placement, required attendees, the facilitator, the facility, the equipment, and, most important, data collection, analysis, and presentation.

Organizational Placement

As a managerial function, Compstat should appear at the top of the organization. Data must flow to the chief and the executive staff without delay. As few lines of reporting as possible should exist between the Compstat unit (or the individual responsible for collating the material) and the chief. This will ensure that the unit collates, analyzes, and delivers the data to the chief for preliminary review before preparing the final version for publication. The Compstat unit should not have to negotiate several organizational layers before handling the chief the completed work, particularly as the material is time sensitive.

Required Attendees

As a managerial process, Compstat employs managers to assess the operational effectiveness of the department and how those entrusted with geographic or organizational command perform in response to a set of given conditions (i.e., the Compstat data). Those required to attend Compstat include the department's executives (chief, deputy chiefs, captains, and division/section commanders) and those decision makers responsible for developing deployment strategies or committing resources (personnel or materiel).

Command-rank personnel must answer for the state of their commands, including how well they perform individually and compared with other commands and the total crime picture within their commands. Commanders of speciality divisions (e.g., narcotics, warrants, robbery, and homicide) must report on the level of support they have committed to an area to reduce a problem, such as arrests effected, canvasses conducted, street surveillances performed, warrants served, and cases cleared, along with their command's overall performance.

All commanders receive support from their staffs--the executive officer, detective squad supervisor, and crime control officer (4)--during Compstat meetings. Commanders sit at the conference table with their support staffs directly behind them. The support staffs provide the commanders with notes, charts, statistics, and performance data on the strategies undertaken since the last Compstat session. The support staffs, particularly the executive officers, must throughly attune themselves to their commanders' intentions and presentations.

[ILLUSTRATION OMITTED]

The Facilitator

The chief or his executive-level designee must act as the Compstat facilitator. The chief sets the tone: if Compstat is important enough for him to take time out of his schedule, then participants should respect the process and take it seriously. In the chief's absence, an uninvolved member of the command staff, such as the chief of staff, assistant chief, or deputy commissioner, must assume the role. Other command staff officers have a biased interest because they are the ones facing critique. The facilitator moves Compstat along, questioning the commanders, helping devise solutions, ordering information for the recap (outstanding issues requiring follow-up), and issuing censure when necessary. To achieve success, the facilitator should understand patrol and investigative strategies, know how to interpret statistics, and possess analytical skills concerning linking conditions, performance, and outcome.

The Compstat Facility

The Compstat facility need not be elaborate, merely large enough to comfortably accommodate all of the required personnel and, preferably, guests. The room should have audiovisual capabilities, such as an overhead projector, a projection screen, computers, and an amplification system. Each commander should have printed copies (the Compstant book) of all of the visual aids (e.g., charts, graphs, and data). This ensures that everyone can follow along and remain attentive during the discussion.

Data Collection, Analysis, and Presentation.

Compstat, a process grounded in data, begins with collecting, analyzing, and mapping crime occurrences. Usually, the person or unit designated to gather and collate FBI Uniform Crime Reporting (UCR) and performance data handles these tasks and prepares the Compstat book (a collated, printed version of the activity occurring in the previous Compstat period normally beginning on Monday at 12:01 a.m. and ending the following Sunday at 11:59 p.m., except for speciality divisions, such as narcotics and criminal investigations, which extend the period to 2 weeks to produce a better trend analysis). (5) The Compstat book will vary in size based upon the jurisdiction and how each chief wishes to display the material. The best guiding principle for designing the Compstat book is whatever issues are prevalent must be captured and made part of the book. A typical Compstat book includes certain elements for the Compstat period, with the sections separated by numbered, lettered, or named tab dividers (e.g., precincts and divisions), such as--

* cover page with Compstat period dates and conference number and name of featured command;

* recap of previous Compstat period's notes and issues;

* numerical crime summary of citywide data, such as aggregate number of incidents, average number of incidents, percentage change, rate of violent and nonviolent crime, and corresponding charts (e.g., pie chart representing crime rates for Part I offenses), sorted by offense;

* violent crime summary by offense in detail, sorted by precinct and weapon involved;

* property crime summary by offense in detail, sorted by precinct;

* precinct profile with identity of commanding officer, executive officer, detective squad supervisor; demographics; personnel strength by rank; and aggregate and disaggregate crime data by precinct;

* weekly sector analysis of crime data and performance comparisons across categories and citywide analysis by current week versus previous week, aggregate difference, and percentage change;

* crime maps with density comparisons and thematic layers with separate maps for each corresponding crime;

* narrative crime summary of every incident under investigation for Compstat period, sorted by data and described by sector, complaint number, date, time, location, type of premises, and means of attack (in speciality commands, more data specific to command or operation, such as make, model, color, caliber, and serial number of recovered guns or packaging description and "brand name" of illegal drugs);

[FIGURE 1 OMITTED]

* pattern crimes with confirmed or emerging crime patterns (e.g., commercial robberies, residential burglaries, or sexual assaults of college students);

* performance indicators with type of arrests by patrol and detectives; summonses issued, both moving and parking, by type; field interviews by sector and precinct; arrest and search warrants issued and served; clearance rates for detective squads and individual detectives; sick time by precinct, tour, rank, and illness with ratio of sick to well officers; investigations and complaints against personnel by division, assignment, rank, sex, and tour with ratio of investigations to complaints; overtime by category, sorted by division; accidents with city vehicles by division, tour, and contributing circumstances; response time by precinct and tour with priority code, number of calls dispatched and self-initiated, queue goals and average queue time, and travel on-scene and service times; and personnel grievance by division, rank, and category;

* optional data, such as abandoned/unsecured buildings, vacant lots, confirmed gangland areas, "top ten lists," truancy and curfew violations, found property lists, offenders' residences (burglars, auto thieves), and sex-offender registrants, that can help commanders identify a nexus between noncrime conditions and crime, antecedents to existing problems, and who to enlist to control them;

* specialty commands, such as narcotics, traffic enforcement, special investigations, and task force operations, with data depicting their level of performance germane to their command and pertinent to the Compstat process; and

* special programs, such as grant-funded initiatives, with explanations of performance measures specified in the program for monitoring purposes.

The best method of capturing calls-for-service and incident-report data involves a computer-aided dispatch (CAD) and a records-management system (RMS). A robust CAD and RMS can produce most of the Compstat book at the touch of a button. Otherwise, a data-entry clerk or crime analyst must enter the details of every incident report, arrest effected, summons issued, case cleared, and other pertinent information into a spreadsheet or database to produce the reports. Essentially, every piece of data to be presented must be collated in an easy-to-read format, organized in a logical order, and assembled into a coherent book.

Absent CAD or RMS reports, each command must capture the essential data elements each week and report them to the designated person or unit who arranges the style and format of the book. Desktop software applications with a suite of products, such as spreadsheets, databases, and word processing programs all in one, make collating and analyzing the information quite easy. This, coupled with the use of special statistical programs for more complex analysis, works well for data reduction.

Describing the Data

In displaying the information, the first step involves describing the data so that readers can quickly understand relevant information. (6) Descriptive statistics give commanders a comprehensive overview of how their command is faring (see Appropriate Statistics chart).

The second type of descriptive statistics is designed to help [commanders] understand the relationship between two or more [crimes]. These statistics are called measures of association, and they enable [crime analysts] to quantify the strength and direction of a relationship. These statistics are very useful because they enable a [crime analyst] to investigate two matters of practical importance: cause and prediction. These techniques help a [crime analyst] disentangle and uncover the connections between crimes. They help trace the ways in which some [crimes] might have causal influences on others, and, depending on the strength of the relationship, they enable a [crime analyst] to make predictions. The measures of association cannot, by themselves, prove that two [crimes] are causally related. However, these techniques can provide valuable clues about causation and are therefore extremely important for [commanders to test their beliefs about crime at certain locations].

For example, suppose a [crime analyst] was interested in the relationship between [calls for drug sales] and [aggravated assault shootings] and had gathered the appropriate data. By calculating the appropriate measure of association, the [crime analyst] could determine the strength of the relationship and its direction. Suppose the [crime analyst] found a strong, positive relationship between these two variables. [A commander] would infer that [calls for drug sales] and [aggravated assault shootings] were closely related [strength of the relationship] and that as one increased in value, the other also increased [direction of the relationship]. [A crime analyst] could make predictions from one variable to the other, for example, [the more calls for drug sales, the higher the shootings].

[FIGURE 2 OMITTED]

Now, as a result of finding this strong, positive relationship, [a commander] might be tempted to make causal inferences. That is, [a commander] might jump to the conclusion that [a high number of calls for drug sales leads to (causes) more shootings]. Such a conclusion might make a good deal of common sense and would certainly be supported statistically. However, the causal nature of the relationship is in no way proven by the statistical analysis. Measures of association can be taken as important clues about causation, but the mere existence of a relationship never should be taken as conclusive proof of causation. (7)

Comparing the Data

The next step involves comparing the data, which enables the chief and command staff to gauge progress and adjust or compensate for shifts in trends or patterns. Appropriate charts should display the information for each precinct, as well as city-wide, including the aggregate difference and percentage change in reported incidents (see Data Comparison chart). These charts should be disaggregated for each crime, and a chart depicting the temporal (time) distribution also should be included so commanders can see when crimes occur.

Crime Mapping

The final step, spatial analysis or crime mapping, has gained popularity over the last 10 years as an inexpensive, valuable resource for law enforcement agencies to identify and plot the occurrence of crime. Crime analysts can detect "nodes," "paths," and "edges" along which criminals travel. (8) They can create overlays of calls for service versus arrests effected, unsolved burglaries with known burglars' residences, or calls for service with abandoned buildings. Analysts can create speciality maps, such as locating sex offenders' residences (Megan's Law registrants), recovered guns, recovered stolen autos, and thefts of auto headlights. Most important, they can display aggregate data to show relationships among offenses in time space, also known as "hot-spot" analysis (see Figure 1). Once agencies conduct spatial and temporal analysis, they can develop intervention strategies (see Figure 2). (9)

A variety of commercial mapping software applications exist that integrate easily with the spreadsheet and database applications that harness the raw data. Personnel simply can import the data into the mapping program and run the reports to create the desired maps. (10)

CONCLUSION

The design and success of the Compstat model rests with the commitment level of the leaders of the law enforcement agency. They must sponsor and champion Compstat to their employees. They also must ensure that all administrative details are handled effectively and efficiently to produce the most important aspect of the process: data collection, analysis, and presentation. Next month, the FBI Law Enforcement Bulletin will feature the final part of this article in which the author will discuss the implementation and adaptability of the Compstat model.

Appropriate Statistics

                  Maximum      Minimum     Mean  Median         Mode
Descriptive       Violent and nonviolent   Aggregate increase  Standard
analysis          crime summary            or decrease         deviation
Frequency         Percentage   Proportion  Ratio and rates
distribution      increase or  across      Incidents to population
                  decrease     categories  Performance to police
                                           officers
Organization      Pie charts (for          Bar charts (for aggregate
and presentation  percentage of total)     data or rate; e.g., incidents
of data                                    per 100,000 people)
                  Line charts (frequency   Histograms with a    Crosstab
                  polygon); add trend      normal curve (e.g,   charts
                  lines to establish       response time
                  direction                analysis)

Data Comparison

                                                            Diff.  % +/-
Day to day              One chart for each week of the
                                Compstat period
Week to week            Current week   vs.  Previous week    +5     +3%
Month to month          March 2003     vs.  April 2003      -18    -27%
Quarter to quarter      Jan, Feb, Mar  vs.  Apr, May, Jun   +32    +44%
Half year to half year  1st 6 months   vs.  2nd 6 months    -63    -40%
Year to year            2002           vs.  2003            -27     -2%
Year to date            January 1, 2003 to present date     Aggregate
Last 12 months          March 15, 2002 to March 14, 2003    Aggregate
Custom date             Any time period (days, weeks, months, quarters,
                          years, decades)

Comparisons for each period against the prior period
Week       Current week 2003             vs.  Same week 2002
Month      Current month 2003            vs.  Same month 2002
Quarter    Jan, Feb, Mar 2003            vs.  Jan, Feb, Mar 2002
Half year  1st 6 months 2003             vs.  1st 6 months 2002
Year       Jan 1, 2003 to present        vs.  Jan 1, 2002 to present
12 months  Jan 18, 2002 to Jan 17, 2003  vs.  Jan 18, 2001 to Jan
                                                17,2002
Custom     Any custom date period compared with the prior date period

Endnotes

(1) Jon Shane, "Compstat Process," FBI Law Enforcement Bulletin, April 2004, 12-21.

(2) For illustrative purposes and to maintain clarity, the author refers to the leaders of law enforcement organizations as chiefs and employs masculine pronouns for these individuals, as well as other command-level personnel, throughout the article as needed.

(3) John M. Bryson, Strategic Planning for Public and Nonprofit Organizations: A Guide to Strengthening and Sustaining Organizational Achievement (San Francisco, CA: Jossey-Bass, 1995), 57.

(4) The crime control officer (CCO) is a division/district-based position that monitors crime trends and patterns on a daily basis. The CCO advises the commander of conditions on a daily basis, captures crime data, analyzes trends and patterns, and makes recommendations on deployment strategies and tactics.

(5) Some departments use a 2-week Compstat period with a 1-week overlap for all commands because they feel that this improves the ability to gauge trends and emerging patterns.

(6) J.F. Healey, Statistics: A Tool for Social Research, 6th ed. (Belmont, CA: Wadsworth, 2002).

(7) Ibid., 8-9.

(8) Paul and Patricia Brantingham have conducted work on the "geometry of crime," suggesting that "each offender tends to be somewhat lazy, sticking close to known places and routes" in Marcus Felson, Crime and Everyday Life, 2d ed. (Thousand Oaks, CA: Pine Forge Press-Sage Publications, 1998), 58, See also P.J. Brantingham and P.L. Brantingham, Environmental Criminology (Prospect Heights, IL: Waveland, 1991); "Environment, Route and Situation: Toward a Pattern Theory of Crime," in R. V. Clarke and M. Felson, eds., Routine Activity and Rational Choice: Advances in Criminological Theory (New Brunswick, NJ: Transaction Books, 1993); and "Criminality of Place: Crime Generators and Crime Attractors," European Journal of Criminal Policy and Research 3, (1995): 5-26.

(9) J.H. Ratcliffe, "The Hot-Spot Matrix: A Framework for the Spatio-Temporal Targeting of Crime Reduction," Police Practice and Research, 5 (2004); and as an unpublished paper presented at the 11th International Symposium on Environmental Criminology and Crime Analysis/ECCA, June 20, 2003, Cincinnati, OH; retrieved on June 26, 2003, from http://www.jratcliffe.net/conf/Ratcliffe%20(2004)%20Hotspot %20matrix%20final%20draft.pdf.

(10) The Police Foundation's Crime Mapping Laboratory in Washington, D.C., can provide excellent technical and crime-mapping assistance. The foundation "is an independent and unique resource for policing. the Police Foundation acts as a catalyst for change and an advocate for new ideas, in restating and reminding ourselves about the fundamental purposes of policing, and in ensuring that an important link remains intact between the police and the public they serve"; access the Police Foundation at http://www.policefoundation.org. For additional crime-mapping resources, see http://www.cslj.net/links/CrimeMapping.htm.

The author thanks his friend and colleague Chief Anthony F. Ambrose of the Newark, New Jersey, Police Department for his inspiration and insight concerning this article.