Posted: April 24th, 2025

Discussion 4 CJUS-4110 Senior Capstone

What is Compstat? Why was it created? What are the pros and cons? In your opinion, what could be changed to make it more effective? 

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 1 2-SEP-05 9:56

DID CEASEFIRE, COMPSTAT, AND EXILE
REDUCE HOMICIDE?*

RICHARD ROSENFELD
ROBERT FORNANGO
ERIC BAUMER

University of Missouri-St. Louis

Research Summary:
Police officials across the United States often claimed credit for crim

e

reductions during the 1990s. In this article, we examine homicide trends
in three cities that mounted widely publicized policing interventions
during the 1990s: Boston’s Operation Ceasefire, New York’s Compstat,
and Richmond, Virginia’s Project Exile. Applying growth-curve analy-
sis to data from the 95 largest U.S. cities and controlling for conditions
known to be associated with violent crime rates, we find that New
York’s homicide trend during the 1990s did not differ significantly from
those of other large cities. We find some indication of a sharper homi-
cide drop in Boston than elsewhere, but the small number of incidents
precludes strong conclusions. By contrast, Richmond’s homicide
reduction was significantly greater than the decline in other large cities
after the implementation of Project Exile, which is consistent with
claims of an intervention effect, although the effect may have been
small.

Policy Implications:
Criminologists gave police and other public officials something of a
free ride as they claimed credit for the 1990s crime drop. We propose
that researchers employ comparable data and methods to evaluate such
claims-making, with the current analysis intended as a departure point
for ongoing research. The use of common evaluation criteria is espe-
cially urgent for assessing the effects of the multiple interventions to
reduce violent crime launched under the nation’s primary domestic
crime-control initiative, Project Safe Neighborhoods.

KEYWORDS: Homicide Trends, Police Interventions, Compstat, Opera-
tion Ceasefire, Project Exile.

* Address all correspondence to Richard Rosenfeld (richard_rosenfeld@umsl.
edu). We wish to thank Jeffrey Fagan, Janet Lauritsen, Christopher Winship, and the
anonymous reviewers for helpful comments on a previous draft. This research was
supported by a grant from the National Institute of Justice (NIJ 2002-RG-CX-K005).
The points of view and conclusions are the authors’ and do not necessarily reflect those
of the funding agency or the U.S. Department of Justice.

VOLUME 4 NUMBER 3 2005 PP 419–450 R

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 2 2-SEP-05 9:56

420 ROSENFELD, FORNANGO, & BAUMER

As crime rates in the United States plummeted in the 1990s, it was inevi-
table that police and other public officials would claim credit for the
decline. How could they resist? Rates of violent crime, especially in the
large cities, were falling to levels not seen since the 1960s (Rosenfeld,
2004). Falling crime rates are good news for politicians, and for good rea-
son: “The potential political payoff is huge,” a columnist wrote during
New York’s crime decline. “If the crime numbers continue their precipi-
tous fall, no one will be able to beat Mr. Guliani when he comes up for re-
election. . .” (Herbert, 1995). In his 1996 State of the Union address, Presi-
dent Clinton embraced the nation’s crime drop and declared: “At last, we
have begun to find a way to reduce crime” (Krauss, 1996). A few months
later, Clinton greeted the release of FBI crime figures with the claim:
“Because of our tough and smart decisions to put more cops on the street
and get kids, guns and drugs off the street, we are now beginning to
reverse the trend in violent crime” (Butterfield, 1996:A1).

President Clinton joined local officials around the country during the
1990s in what one journalist called “a chorus of self-congratulation”
(Krauss, 1996). In city after city, officials attributed the falling crime rates,
in whole or part, to a new program or policy they had instituted. Austin’s
mayor linked crime reductions to community and problem-oriented polic-
ing. Buffalo’s police commissioner attributed the decrease to more aggres-
sive enforcement, targeting career criminals, and a youth curfew. In
Chicago, the addition of new officers and community policing were
credited (the claims for Austin, Buffalo, and Chicago are described in Law
Enforcement News, 1997). Detroit’s police chief highlighted his efforts to
“improve the image of the police with the community and get as many
people involved as possible” (Butterfield, 1995:10). The Mayor of East St.
Louis attributed the crime drop in his blighted city to more officers, new
equipment, and antidrug and antigang-targeted patrols (Gillerman, 1996).
Houston’s chief cited computerized crime analysis, more officers,
storefront stations in crime hotspots, and an antigang taskforce (Butter-
field, 1995:10). In Los Angeles, officials credited hot-spot and community
policing, but they also cited neighborhood crime prevention, a gang truce,
and an earthquake as reasons for the crime drop (Feldman, 1994).1

Criminologists generally greeted such claims-making with skepticism
and countered with alternative explanations for the crime drop, most fea-
turing some combination of withered drug markets, the booming econ-
omy, mass incarceration, or the “little brother syndrome” (younger

1. Not all officials were willing to claim credit for the drop in crime. LA Deputy
Police Chief Mark Kroeker told a reporter that, although the news was encouraging,
“the last thing we should all do is say, “Oh, it’s because of certain strategies we’re
talking about in police work. If you start taking credit for crime decreases, you better
start getting ready to take the blame for crime increases as well” (Feldman, 1994:B2).

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 3 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 421

adolescents were turned off by the drug use and violence of older adoles-
cents) (Lardner, 1997; see also Butterfield, 1996, 1998; Lacayo, 1996).
Some criminologists acknowledged they had no idea why crime rates were
decreasing and limited their public commentary to airy proclamations such
as “I think we can now say a trend has been established” and “Something
broad is happening” (Butterfield, 1996).2 Overviews of the crime drop
published as it was coming to an end contained elaborate statistical
descriptions of the trends but remarkably little in the way of explanatory
research on either the crime-reduction claims advanced by public officials
or the counterclaims of the “experts” (Blumstein and Wallman, 2000;
Cook and Laub, 2002; Rosenfeld, 2002, 2004). “The academic world,” one
critic observed, “has not distinguished itself in this field of inquiry”
(Lardner, 1997:54).

Evaluating the impact of policy initiatives on crime trends is challenging
under the best of circumstances; it is even more difficult to isolate
whatever additional difference policy may have made when crime rates
are already on the decline. Furthermore, most crime-control policies are
not designed for valid evaluation, and those with built-in evaluation
designs usually are intended to prevent the criminal behavior of individu-
als rather than reduce the crime rates of populations. An individualistic
bias pervades criminology. In the midst of the 1990s crime decline, the
significant advances in criminology were in the study of individual devel-
opmental patterns of delinquency and other problem behaviors (Farring-
ton, 2003); analysis of aggregate crime trends remained on the periphery
of the field (LaFree, 1998).

In this article, we seek to advance the study of policy impacts on crime
trends through an assessment of homicide trends in three cities with
widely publicized local law enforcement initiatives: Boston’s Operation
Ceasefire, New York’s Compstat, and Richmond Virginia’s Project Exile.
Each program has been credited with reducing rates of serious violent
crime, including homicide, and each has been subject to some form of out-
come evaluation. However, to our knowledge, the crime-reduction effects
of the three programs have never been evaluated with comparable data
and methods, and the existing evaluations either lack systematic compari-
sons with crime trends in other cities or fail to control for other influences
on crime trends, both of which are critical for separating the change in
crime trends associated with policy interventions from changes caused by
other factors. We use piecewise linear growth models to evaluate homicide
trends in the three cities during, before, and in the case of Boston, after
the intervention periods. Our models control for a broad range of other

2. And some criminologists, an anonymous reviewer reminds us, joined practi-
tioners in touting the crime-reduction effects of local law enforcement interventions.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 4 2-SEP-05 9:56

422 ROSENFELD, FORNANGO, & BAUMER

influences on homicide trends and are fit to data from all U.S. cities with a
1990 population greater than 175,000 (N = 95).

Let us be very clear about our objectives. We do not claim to evaluate
the three interventions per se; we have neither the data nor the methods
for doing so. Rather, we evaluate homicide trends in the three cities in and
around the respective intervention periods and draw inferences from that
evaluation about the accuracy of claims that the interventions reduced
homicide. Those inferences cannot be as strong as many program propo-
nents or critics would like, but they should be preferred over the baseless
success claims and counterclaims and handful of disparate empirical inves-
tigations that have characterized evaluations of the three policing initia-
tives to date.

THE INITIATIVES
Ceasefire, Compstat, and Exile arguably are the most important local

law enforcement initiatives intended to reduce serious criminal violence
launched during the 1990s. Certainly, they are the best known. In this sec-
tion, we describe each of the programs, their intended outcomes, claims
asserting their success, and prior research on their crime-reduction effects.

BACKGROUND AND PROGRAM LOGIC

Ceasefire, Compstat, and Exile were launched against a backdrop of
sharply escalating violent crime rates during the late 1980s and early 1990s.
Homicides in Boston among youth aged 24 years and under more than
tripled between 1987 and 1990 and remained above 1980s levels through
the mid-1990s (Kennedy et al., 2001:1). New York registered a record
2,245 homicides in 1990, well above the previous record of 1,826 set in
1981, prompting one seasoned journalist to warn: “If there are 2,00

0

murders this year, get ready for 4,000. New York is dying” (quoted in
Karmen, 2000:7). Richmond consistently ranked among the top ten U.S.
cities in homicides per capita during the late 1980s and early 1990s; in
1994, it ranked number two, behind New Orleans, with a homicide rate of
80 per 100,000 residents (authors’ calculations from the FBI’s Uniform
Crime Reports).

Although the three initiatives were similarly situated in a context of ris-
ing violent crime, they differed in problem diagnosis, design, and strategy.
From 1996 to 1999,3 Boston’s Operation Ceasefire focused on firearm vio-
lence involving youth gangs. Ceasefire’s central objective was to deter

3. Although the Boston police continue to employ certain tactics related to
Ceasefire, in its original formulation the intervention effectively ended in 2000 (Braga,
2004). In our assessment of Boston homicide trends, therefore, we define the interven-
tion period as 1996–1999.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 5 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 423

youth firearm violence through direct communication to gang youth that
firearm possession and use would not be tolerated, and all available levers
would be pulled to ensure swift and tough punishment of violators.4 The
dual strategy of “retail deterrence” and “pulling levers” was implemented
in a series of meetings with gang members involving police, youth workers,
probation and parole officials, the U.S. attorney, and the local district
attorney. “We’re here because of the shooting,” they would say. “We’re
not going to leave until it stops. And until it does, nobody is going to so
much as jaywalk, nor make any money, nor have any fun” (Kennedy et al.,
2001:27–28). The gang members were told to spread the word on the
street. Youth workers also took the message directly to other gang youth.
“Stop the Violence” posters were put up throughout areas with high levels
of gang violence. Other posters described the sanctions, including federal
charges and sentences, applied to recalcitrant gang members: “They were
warned,” the posters said. “They didn’t listen (Kennedy et al., 2001:28–41;
see also Kennedy, 1998).

A different kind of “retail deterrence” through aggressive order mainte-
nance policing is at the heart of New York’s Compstat. Begun in 1994 with
the installation of William Bratton as police commissioner and continuing
to the present, Compstat sought to restore order on the streets and
accountability for crime in the police department. The police were no
longer to tolerate minor offenses; they were to make arrests for vagrancy,
vandalism, littering, minor drug possession, prostitution, public drunken-
ness and urination, aggressive panhandling, and harassment by “squeegee
pests.” The logic of the approach is taken from the so-called broken win-
dows theory of crime causation: Minor crimes and disorder invite more
serious offending by signaling that the police and community have lost
control of the streets (Kelling and Bratton, 1998; Kelling and Coles, 1996;
Wilson and Kelling, 1982). Aggressively restoring order reverses the mes-
sage, thereby deterring more serious crime. In addition, frequent arrests
for minor offenses on occasion net bigger fish, such as persons wanted on
warrant for serious crimes or who are illegally carrying firearms or large
quantities of drugs for sale. Other enforcement changes were introduced
under Bratton, including “buy and bust” offensives against drug sellers
and anti-gun patrols, but for Bratton the “linchpin strategy” was aggres-
sive order maintenance policing (Bratton and Knobler, 1998:228).

The other component of the New York strategy is to hold commanders

4. A second objective of the Boston Gun Project, of which Ceasefire was a part,
was to trace and interdict crime guns flowing to gang members (Kennedy et al., 1996,
2001:18–20). An alternative explanation of Boston’s youth homicide decline in the mid-
1990s emphasizes the role of a coalition of clergy members in improving relations
between the police and minority residents and offering legitimacy to the Boston Gun
Project (see Fagan, 2002; Winship and Berrien, 1999).

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 6 2-SEP-05 9:56

424 ROSENFELD, FORNANGO, & BAUMER

accountable for the crime rates in their precincts. At regular meetings in
the “war room” at police headquarters, top administrators grill com-
manders on their knowledge of crime patterns and on their crime-reduc-
tion activities and plans (Karmen, 2000:92–94). A deterrence logic
underlies this approach as well, now directed at police officials rather than
at offenders: You are directly responsible for what happens under your
command. If you fail, there will be consequences.

In practice it is nearly impossible to separate the two components of
Compstat for evaluation purposes. Even in principle disagreement exists
concerning which one should be credited for New York’s crime decline
during the 1990s. Bratton believes order maintenance policing was the
“key” to falling crime rates. Mayor Giuliani, on the other hand, main-
tained that police reorganization and greater accountability were primarily
responsible (Karmen, 2000:94–96, 120). Critics contend that order mainte-
nance policing in New York departed markedly from the “broken win-
dows” model, and that whatever effect it may have had on crime was
because of aggressive stop-and-frisk tactics aimed primarily at racial and
ethnic minorities (Fagan and Davies, 2001). In any event, our analysis is
limited to the question of whether New York homicide trends differ signif-
icantly from those of other cities after Compstat was initiated. We cannot
determine which aspects of Compstat (or Ceasefire or Exile) may have
been responsible for observed differences in the trends.

A more traditional deterrence logic underlies Richmond, Virginia’s Pro-
ject Exile, which was formally initiated in February of 1997. Exile entails
sentence enhancements through federal prosecution for violent or drug
crimes involving firearms.5 Federal sentences for such crimes generally are
longer than those in state courts, bail is denied more often, and sentences
are served in federal prisons likely to be located out of state (Raphael and
Ludwig, 2003:254). By increasing the expected penalty for firearm-related
offenses, the program is intended to deter firearm carrying and criminal
use. By sentencing more violent offenders to longer prison sentences, the
program also is intended to reduce crime by incapacitating violent felons.
It is unclear whether deterrence or incapacitation is the primary objective
of Exile’s sentence enhancements. However, the broad “outreach” cam-
paign accompanying the program is intended explicitly to deter criminal
use of firearms. Exile is advertised extensively in print and electronic
media and on city buses and business cards carrying the blunt and bleak
warning that “An illegal gun will get you five years in federal prison,” the

5. Specifically, the program covers felon-in-possession-of-a-firearm cases and
drug and domestic violence cases involving firearms (Raphael and Ludwig,
2003:253–254). See the U.S. Attorney’s description of Exile at http://www.vahv.org/
Exile.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 7 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 425

mandatory minimum sentence for federal “felon-in-possession” cases
(http://www.vahv.org/Exile). There is, then, something of a “retail” com-
ponent to Project Exile’s practice of deterrence, if not as precisely targeted
to particular groups and neighborhoods as was the case in Boston’s Opera-
tion Ceasefire.

PRIOR OUTCOME EVALUATIONS

The most important obstacle in the way of rigorous outcome assess-
ments of the three programs is that each of them was implemented city-
wide, leaving thereby no like situated areas without the intervention that
might serve as within-city controls, A second obstacle to reliable evalua-
tion is that the different aspects of each program were implemented more-
or-less at the same time, so that even if a “program effect” was found, it
would not be obvious which program component should be credited (sen-
tence length? advertising? change in enforcement? change in command
accountability? which levers?). Not surprisingly, then, those few outcome
evaluations that have been conducted take the form of single-case time-
series investigations with limited comparisons with crime trends in other
cities.

New York. The single exception is Kelling and Sousa’s (2001) study of
violent crime trends during the 1990s in New York’s 76 police precincts.
Controlling for changes in borough-level unemployment, age composition,
and a measure of drug involvement, they find an association between
decreases in violent crime and increases in “broken windows” policing, as
measured by misdemeanor arrests. Kelling and Sousa (2001) interpret this
result as evidence for a key plank of Bratton’s policing reforms and against
so-called root cause explanations of New York’s crime drop and crime in
general. Although the coincidence of rising misdemeanor arrest rates and
falling violent crime rates across New York City precincts is compelling,
Kelling and Sousa (2001) do not control for several additional factors com-
monly associated with violent crime (e.g., poverty, family structure, immi-
gration trends) and fail to consider the possibility of reciprocal causation
between their policing measure and violent crime: Decreases in violent
felonies may enable the police to devote greater attention to less serious
offenses, which results in increases in misdemeanor arrests.

Other assessments of Compstat are less sanguine but also less conclusive
about its crime-reducing consequences. Several investigators have pointed
out that New York’s sharp decline in homicide and other violent crimes
during the 1990s was not unique; San Diego, San Antonio, Houston, San
Francisco, Los Angeles, and other large cities that had not implemented
similar policing reforms also registered very sizable declines (Harcourt,
2001:90–94; Joanes, 2000). In an early assessment, Fagan et al. (1998)
observed that nongun homicides had begun to fall in New York well

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 8 2-SEP-05 9:56

426 ROSENFELD, FORNANGO, & BAUMER

before the Compstat reforms were implemented in 1994. They also find,
however, that the drop in firearm homicides is more consistent with the
timing of Compstat and conclude that the policing changes may have con-
tributed to that decline. Even a harsh critic of broken windows policing
concedes that it “contributed in some degree to the decline in crime in
New York City” (Harcourt, 2001:103). Finally, the most comprehensive
assessment of the sources of New York’s crime drop concludes that
changes in policing likely accounted for some decrease, along with a
booming economy, escalation in imprisonment, shrinking drug markets,
favorable demographic trends, and possible changes in the values of ado-
lescents and young adults. However, the author could draw no conclusions
regarding the size of the policing effect (Karmen, 2000:262–266; see, also,
Conklin, 2003).

In summary, most independent assessments of the policing changes
under Bratton conclude that they probably reduced levels of violent crime
in New York over and above the effects of other factors. The most defini-
tive of the studies (Kelling and Sousa, 2001) does not compare

New York

crime trends with those in other cities. The others place New York trends
in the context of those elsewhere, but they do not employ comprehensive
samples of comparison cities or control for other determinants of violent
crime.

Boston. The Harvard researchers who participated in the development
of Operation Ceasefire compared the timing and magnitude of violent
crime trends in Boston with those in other New England cities and a
national sample of large cities (Braga et al., 2001; Kennedy et al., 2001;
Piehl et al., 2003 ). In a two-part investigation, the researchers first per-
formed a time-series analysis of monthly youth firearm homicides, gun
assaults, and shots fired calls in Boston before and during the intervention
period, controlling for changes in unemployment, size of the youth popula-
tion, adult homicide victimization trends, index crime trends, and youth
drug activity as measured by arrests. The results indicate a reduction in
firearm violence, net of the controls, coinciding with the implementation
of Ceasefire.

In the second part of the study, the researchers compared

Boston

monthly youth homicide trends with those in 29 other New England cities
and 39 large U.S. cities, which they adjusted for linear and nonlinear
trends in each series, monthly effects, an autoregressive component, and
an intervention component in each series. They found three cities in addi-
tion to Boston with significant values on the intervention component, but
either the direction or precise timing of those effects differ from Boston’s.6

6. In a separate investigation, Piehl et al. (2003) found that the maximum “break”

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 9 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 427

The Harvard team concluded that Ceasefire was associated with a signifi-
cant reduction in youth firearm violence. Their conclusion would be more
persuasive had they included additional controls in both the within- and
between-cities analyses they conducted. Nonetheless, their findings are
consistent with claims of a program effect.

Richmond. Richmond’s Project Exile has been subject to a single pub-
lished outcome evaluation. Raphael and Ludwig (2003) evaluate claims
that Exile significantly reduced firearm homicides in Richmond by com-
paring Richmond gun homicide trends during the 1980s and 1990s, Rich-
mond trends with those in other cities, and adult and juvenile homicide
arrest rates in Richmond. The first comparison shows that Richmond gun
homicide rates had increased markedly during the 1980s, peaking in the
early 1990s. Raphael and Ludwig (2003) conclude that Richmond’s gun
homicide decline during the 1990s was to be expected as part of a general
regression to the mean common across U.S. cities with high homicide
rates. They conclude from their comparison of Richmond’s gun homicide
trends with those of other cities that the proportional drop in Richmond’s
rates through the 1990s was not unusual. What was unusual was Rich-
mond’s gun homicide rate in 1997, the year Exile began. Raphael and Lud-
wig (2003) suggest that 1997 was subject to “transitory” influences because
it broke a two-year “trend” that began in 1995 and resumed in 1998.
Therefore, they dropped 1997 from their trend analyses. “Using this unu-
sual year as the base for calculating the change,” they maintain, “is bound
to inflate the apparent impact of the program” (Raphael and Ludwig,
2003:258).

To control for unmeasured influences on Richmond homicide trends,
Raphael and Ludwig (2003) compare trends in juvenile and adult homi-
cide arrest rates. The logic of the comparison is that juveniles are not sub-
ject to Exile provisions (or not to the same extent as adult criminals) but
are affected by other factors driving homicide rates. They find that juve-
nile homicide arrests increased slightly from 1995/1996 to 1998 (again,
1997 is omitted), but adult arrests increased even more over the same
period. In other large urban counties, adult homicide arrests declined at a
greater rate than did juvenile homicide arrests. Neither finding is consis-
tent with the idea that Exile reduced homicide among Richmond adults.

In summary, Raphael and Ludwig (2003) find little evidence to support
claims that Richmond’s Project Exile reduced firearm or overall homicide
rates. We regard their evaluation as suggestive but far from the final word
on Exile’s effects. We reserve our specific criticisms until we present the
findings from the current analysis.

in the Boston youth firearm homicide series occurred during the summer of 1996, just
after Ceasefire began.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 10 2-SEP-05 9:56

428 ROSENFELD, FORNANGO, & BAUMER

The results of the outcome evaluations of Compstat, Ceasefire, and Exile
are mixed. Evaluations of Compstat and Ceasefire support claims that the
interventions resulted in reductions in homicide. The single published
study of Project Exile concludes that it had little apparent effect on Rich-
mond firearm homicide rates. It is difficult to compare findings across the
evaluations of the three interventions, because they targeted differing pop-
ulation groups and homicide types, and because they are based on distinct
sources of data and methods of analysis.

We use the same data sources and methods to evaluate all three inter-
ventions. The logic of our approach assumes that an effective intervention
will produce a reduction in homicide that (1) differs significantly from cor-
responding changes in comparison cities; (2) occurs during or after but not
before the intervention period; (3) occurs in the specific type of homicide
targeted by the intervention; and (4) is independent of other influences.
By the same logic, an ineffective intervention will produce a reduction in
homicide that does not differ significantly from changes in comparison
sites or occurs before the intervention or does not occur in the homicides
targeted by the intervention or is brought about by conditions other than
the intervention. To be deemed “effective” (or not ineffective), an inter-
vention must meet all four conditions. To be found ineffective, it must fail
to meet only one of them.

ANALYTICAL STRATEGY

To assess the effects of the three interventions, we apply growth-curve
models to homicide trends in Boston, New York, and Richmond. We esti-
mate each city’s homicide trend as a function of a baseline model of
covariates fit to data for the 95 largest U.S. cities. An intervention’s effec-
tiveness is indicated by a reduction in homicide during the intervention
period that is significantly greater than the average reduction for the sam-
ple. Our strategy cannot conclusively rule out other possible explanations
for observed differences in homicide trends between cities with and with-
out the interventions. However, the absence of such differences would
place a particularly strong onus on program defenders to demonstrate that
an observed homicide reduction resulted from a particular intervention.

DATA

The homicide data for the analysis were obtained from the Supplemen-
tary Homicide Reports (SHR) for the 95 U.S. cities with a 1990 population
of 175,000 or more over the period 1992–2001. The data correspond to the
distinct type of homicide targeted by each intervention. Boston’s Ceasefire
was designed to reduce firearm homicides among the “gang age” popula-
tion of adolescents and young adults. The policing strategies in New York

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 11 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 429

were intended to reduce homicides (and other crimes) of all types, regard-
less of weapon or age of the offender. Richmond’s Exile specifically
targeted firearm homicides. Accordingly, the analysis of Ceasefire was car-
ried out with SHR data on victims of firearm homicide ages 15–24, the
Compstat analysis with data on total homicide victimizations, and the Exile
analysis with data on all firearm homicide victimizations. The SHR data
were missing for 62 of the 950 city-years in the analysis. Estimated values
for the missing data were obtained from within-city regressions of
observed values on a period variable (1992 = 0).7

To obtain unbiased estimates of the effects of the interventions on homi-
cide rates, we created baseline models incorporating well-established
covariates of homicide (Land et al., 1990; Messner and Rosenfeld, 1998).
Measures of social and economic disadvantage and population density
were obtained from the 1990 Census and the 2000 Census. The density
measure was logged to reduce skewness, and all measures were interpo-
lated between census years and extrapolated to 2001. The disadvantage
measure was derived from a principal components analysis of several
highly intercorrelated variables: percent of families with children under
age 18 headed by a female, percent black, median family income (logged),
the male unemployment rate, the poverty rate, and the Gini coefficient of
income inequality. This measure of economic and social disadvantage cor-
responds closely to that in Land et al. (1990).

Prior research has found that homicide trends are negatively related to
trends in incarceration and police density (Levitt, 2002; Marvell and
Moody, 1997; Spelman, 2000). We incorporated in our analysis the yearly
state incarceration rate corresponding to each of the 95 cities, obtained
from the Bureau of Justice Statistics, and an annual indicator of police per
100,000 city residents, computed from the number of sworn full-time
officers reported in the Law Enforcement Management and Administra-
tion Statistics (LEMAS) surveys.

Finally, several studies have shown a connection between homicide
trends and crack cocaine markets in U.S. cities (Baumer et al., 1998; Blum-
stein, 1995; Cork, 1999; Ousey and Lee, 2002). We developed a city-level

7. Data for the five Florida cities in the sample were missing for the years
1997–2001. Because they fall at the end of the series, our imputation procedure did not
yield plausible values for these years, and they were left missing. Our analytic methods
are robust to missing data (Raudenbush and Bryk, 2000:199–200).

An alternative homicide data source to the SHR is the homicide series compiled by
the National Center for Health Statistics (NCHS). However, the NCHS data are availa-
ble at the county level but not the city level, and the boundaries of several cities in our
sample are not coextensive with county lines. In addition, the NCHS homicides are
coded according to the victim’s residence rather than location of the incident as in the
SHR.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 12 2-SEP-05 9:56

430 ROSENFELD, FORNANGO, & BAUMER

proxy measure of the level of cocaine use among arrestees from the

20

cities in our sample for which yearly Drug Use Forecasting (DUF) and
Arrestee Drug Abuse Monitoring (ADAM) data were available between
1990 and 2000. The proportion of arrestees testing positive for cocaine
reported by DUF-ADAM was regressed on a composite measure of socio-
economic disadvantage,8 the percentage of city residents living in the same
house for five or more years, and the percentage of the population resid-
ing in owner-occupied housing. The fit statistics indicated that the model
could serve as a suitable proxy for city-level variation in cocaine involve-
ment. The predicted values from the equation were used to estimate the
extent of cocaine use in cities for which the DUF-ADAM data were
unavailable.9

METHOD

The primary research question is whether the three intervention sites
experienced greater declines in homicide than other cities during the inter-
vention period after adjusting for between-city differences in other deter-
minants. The analytic strategy must therefore provide estimates of both
the within-city trajectories in homicide and between-city differences in
those trajectories. We use growth-curve analysis to obtain these estimates.
Growth-curve analysis estimates the homicide trends for each city and
compares the intercept and slope parameters across cities. Traditionally,
such trends are estimated as a polynomial function of time, with single
parameters for the linear trend and change in the growth rate. Because
our interest lies with the trends during the intervention periods, we esti-
mate the linear components as a series of piecewise trends. The piecewise
strategy estimates a linear trend component for successive subperiods of
the series, in effect allowing the slope to “bend” at predetermined points.
Two linear components are specified in the piecewise models for New
York and Richmond, corresponding to the pre-intervention and interven-
tion periods. Three linear components are specified for Boston, corre-
sponding to the pre-intervention, intervention, and post-intervention
periods.

The hierarchical generalized linear model (HGLM) has become a com-
mon approach to modeling change within- and across-individuals, schools,
neighborhoods, and cities. This approach is an attractive alternative to
multivariate repeated-measures (MRM) models, which can be difficult to

8. Socioeconomic disadvantage is a principal components factor score of percent
black, percent female-headed families with children under 18, the male unemployment
rate, median family income, the poverty rate, and poverty concentration (the propor-
tion of the population residing in census tracts with a poverty rate of 40% or higher).

9. Details concerning the development and validation of the proxy measure are
available from the authors.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 13 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 431

specify when the number of cross-sectional units is relatively large and
when there are multiple time periods under investigation, as is the case in
our analysis (Greene, 2000). Because the plausible values for homicide
rates are constrained to non-negative numbers, the dependent variable in
our analysis has a truncated distribution, and a Poisson sampling model is
used with city population as the exposure variable.

Given our interest in modeling homicide trends both within and
between cities, we estimate two-level models. The annual homicide counts
are treated as repeated measures, nested within cities in the level-1 equa-
tion. Time-varying explanatory variables that may be related to change in
homicide within cities are included at level-1. In this analysis, three time-
varying covariates are included in the level-1 model: police per capita,
state incarceration rate, and estimated level of cocaine use. In a multilevel
model, the parameter estimates from level-1 become the outcomes, and
variables hypothesized to explain differences in those estimates across
units are incorporated in the level-2 model. In our analysis of homicide
trends in each of the three intervention sites, the key level-2 covariate is a
dummy variable that distinguishes the specified intervention site from
other cities in the sample. We also incorporate at level-2 the indicators of
resource deprivation and population density. We allow the level-1
intercepts and slope coefficients to vary randomly across cities. A more
detailed description of our methods is presented in Appendix I.

10

RESULTS
Figure 1 displays the average homicide trends for the 95 cities in our

sample as well as trends in the three intervention sites: Boston, New York,
and Richmond. As the figure shows, homicide rates declined in the largest
U.S. cities during the 1990s to about 13 per 100,000 population at the end
of the decade from 20 per 100,000 at the beginning. Homicide rates in
Boston fell from roughly 15 per 100,000 during the early 1990s to about 6
per 100,000 at the end of the decade, only to rise again to 11 per 100,000 in
2001. In New York, they plunged to about 8 per 100,000 from 27 per
100,000 over the same period. Richmond homicide rates roller-coastered
through 1997, reaching a peak of 80 per 100,000 in 1994, before falling to
about 36 per 100,000 at the end of the decade. Figure 1 confirms that
homicide rates dropped substantially in the three cities during their
respective intervention periods, but it remains to be seen whether these
declines were significantly greater than those observed elsewhere after
adjusting for relevant correlates of crime trends.

To examine whether the observed homicide declines in Boston, New

10. HLM 5.04 statistical software was used for all analyses (Raudenbush et al.,
2001).

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 14 2-SEP-05 9:56

432 ROSENFELD, FORNANGO, & BAUMER

FIGURE 1. HOMICIDE TRENDS IN US CITIES WITH
175,000 OR MORE RESIDENTS, 1992-2001

(N = 95)

0

10

20

30

40

50

60

70

80

90

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

H
o

m
ic

id
es

p
er

1
00

,0
00

P
o

p
u

la
ti

o
n

Boston

New York

Richmond

Average

York, and Richmond were significantly greater than those observed in
other large cities, net of other factors, we applied piecewise linear growth-
curve models to the trend in the type of homicide targeted by each inter-
vention. The results of our analysis are summarized in Table 1 and
presented in greater detail in Appendix II. Table 1 summarizes the results
of the level-1 and level-2 estimations of “unconditional” and “conditional”
models of the homicide trends before the intervention, during the inter-
vention period, and for Boston after the intervention. The results from the
unconditional models are unadjusted for the influence of the three time-
varying (incarceration rate, police density, drug use) and two time-invari-
ant covariates (resource deprivation and logged population density)
included in the analysis.11 The results from the conditional models are
adjusted for the effects of these covariates. In both cases, the deviation of
the trend for each of the three intervention sites from the sample average
or “base” trend is presented.

11. In preliminary analyses, we evaluated the effects of other time-varying influ-
ences on homicide trends, including a measure of firearm availability (percent suicides
with a firearm), concentrated poverty (percent population residing in census tracts with
a poverty rate of 40% or higher), age structure (percent population between the ages of
15 and 24), and male unemployment rate (percent males age 16 and older unemployed).
None of these measures yielded significant results or improved model fit and, with the
exception of the male unemployment rate, were dropped from the analyses shown. The
male unemployment rate was added as a time-invariant covariate to the disadvantage
measure.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 15 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 433

TABLE 1. HGLM SUMMARY RESULTS FOR CITY
HOMICIDE RATES, 1992–2001 (N = 95)

Unconditional Conditional
A. Boston Model Model

Initial (1992) Youth Firearm Homicide Rate, p0i

Base, b00 1.435*** .339
(.104) (.782)

Boston, b01 .212 −.146
(.992) (.632)

Pre-Intervention Change (1992-1994), p1i

Base, b10 −.015 .188
(.017) (.148)

Boston, b11 .006 −.008
(.149) (.161)

Intervention Change (1995–1999), p2i

Base, b20 −.151*** −.173
(.014) (.135)

Boston, b21 −.134 −.186
(.123) (.113)

Post-Intervention Change (2000–2001), p3i

Base, b30 .012 −.043
(.027) (.269)

Boston, b31 .372 .305
(.238) (.204)

B. New York

Initial (1992) Homicide Rate, p0i

Base, b00 2.697*** 2.048***
(.084) (.598)

New York City, b01 .612 −.521
(.800) (.493)

Pre-Intervention Change (1992–1993), p1i

Base, b10 .066*** .452**
(.023) (.200)

New York City, b11 −.209 .046
(.159) (.089)

Intervention Change (1994–2001), p2i

Base, b20 −.079*** .040
(.006) (.059)

New York City, b2i −.090*

.027

(.048) (.041)

C. Richmond

Initial (1992) Firearm Homicide Rate, p0i

Base, b00 2.358*** 1.658**
(.093) (.646)

Richmond, b01 1.593* .700
(.891) (.549)

Pre-Intervention Change (1992–1996), p1i

Base, b10 −.064*** .175
(.012) (.133)

Richmond, b11 .083 .113
(.107) (.111)

Intervention Change (1997–2001), p2i

Base, b20 −.103*** −.108
(.009) (.098)

Richmond, b21 −.069 −.144**
(.075) (.067)

* p < 0.10; **p < 0.05; ***p <0.01.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 16 2-SEP-05 9:56

434 ROSENFELD, FORNANGO, & BAUMER

BOSTON

Panel A of Table 1 presents the results for the trend in youth firearm
homicide rates targeted by Boston’s Ceasefire. The pertinent results are
the coefficients for the sample average (minus Boston) and Boston’s
deviation from the average during the pre-intervention (1992–1995), inter-
vention (1996–1999), and post-intervention (2000–2001) periods. The
average youth firearm homicide rate for the sample exhibits no significant
change during the pre-intervention period (b = –0.015). It dropped signifi-
cantly during the intervention period (b = –0.151, p < 0.01), but that
change becomes nonsignificant once the covariates are introduced in the
conditional model. Boston’s youth firearm homicide rate exhibits a greater
decrease than the sample average during the intervention period, in both
the unconditional and conditional specifications. Although neither differ-
ence is statistically significant, the difference between Boston and the sam-
ple average in the conditional model just misses the most permissive
significance threshold (p = 0.101). Finally, the Boston trend did not deviate
significantly from the sample average during the post-intervention period
in either model.

By exponentiating and subtracting 1, the coefficients displayed in Table
1 can be converted to proportional increases or decreases in the outcomes.
Doing so indicates that Boston’s youth firearm homicide rate dropped an
estimated 25% per year during the 1996–1999 intervention period, com-
pared with a decline of 14% in the average rate during the same period.
After adjusting the changes for the effects of the covariates, Boston’s
youth firearm homicide rate fell an estimated 30% per year during the
intervention period, and the average rate dropped 16%.12 Although the
estimated decline in Boston’s rate was nearly double that of the sample
average in the conditional model, the difference is not statistically signifi-
cant, even by a permissive standard of p < 0.10. The lack of statistical sig-
nificance reflects Boston’s low youth firearm homicide counts during the
intervention period (ranging from 21 in 1996 to 10 in 1999). Relatively
small year-to-year changes in the number of youth firearm homicides in

12. The percentage changes are derived as follows: The coefficients for the sample
average in the unconditional and conditional models during the intervention period are
–0.151 and –0.173, respectively. Exponentiating, subtracting 1, and multiplying the
result by 100 yields respective yearly percentage changes of –14% and –15.9%. Bos-
ton’s rate fell an additional –0.134 and –0.186 in the unconditional and conditional mod-
els, respectively. Summing these values with those for the average changes yields total
changes in the Boston rates of –0.285 (–0.151 + –0.134) and –0.359 (–0.173 + –0.186).
Exponentiating, subtracting 1, and multiplying the result by 100 yields yearly percent-
age changes in Boston’s youth homicide rate of –25% in the unconditional model and
–30.2% in the conditional model.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 17 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 435

Boston resulted in large percentage changes. Therefore, the prudent con-
clusion regarding Ceasefire’s impact is that it is statistically indeterminate
given the small number of youth firearm homicides in Boston.

It is possible that the age range used to define the youth firearm homi-
cide rate in our analysis is too narrow. In supplementary analysis, we rees-
timated the Boston models using a broadened victim age range of 11–24.13

Widening the age range of victims adds only three homicides in Boston
during the entire 1992–2001 period. Not surprisingly, the supplementary
estimation results differ only slightly from the original results, with a mar-
ginally significant p-value for the “Boston effect” in the conditional model
of 0.092. This result does not alter the substantive conclusion that
Ceasefire’s impact on youth firearm homicides in Boston is difficult to dis-
cern given the small number of incidents.

NEW YORK

Panel B of Table 1 shows the results for New York. The New York
homicide trend did not deviate significantly from the sample average dur-
ing the 1992–1993 pre-intervention period. The New York rate fell more
sharply than the sample average during the intervention period, although
the difference is only marginally significant at p < .10. However, when the
covariates are introduced in the conditional model, the difference between
the New York and the average homicide trend becomes nonsignificant.
Thus, after adjusting for measured differences between New York and
other large cities, the magnitude of the decline in homicide in New York
between 1994 and 2001 seems not to have been atypical.

It is possible that Compstat had a greater effect on New York’s firearm
homicide rate than the total homicide rate, as Fagan et al. (1998) have
suggested. However, substituting the firearm homicide rate for the total
rate in the conditional model yields no significant deviation of New York’s
firearm homicide trend from the sample average during the intervention
period (results not shown).

Consistent with prior research (Marvell and Moody, 1997), the growth
in incarceration during the 1990s is consistently related to the homicide
trends examined in this study. We cannot rule out entirely the possibility
that Compstat, and perhaps Ceasefire and Exile, had an indirect effect on
homicide through incarceration. Conceivably, the programs increased
state prison sentences that, in turn, reduced homicide rates.

More research is necessary to determine how much of the growth in

13. Previous studies of Ceasefire broaden the age range even further to include all
firearm homicide victims under the age of 25 (Braga et al., 2001; Piehl et al., 2003).
However, we question whether Ceasefire legitimately can be expected to have reduced
the number of infant killings or homicides involving very young children.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 18 2-SEP-05 9:56

436 ROSENFELD, FORNANGO, & BAUMER

New York’s state prison population was caused by the policing changes
under Compstat. New York’s imprisonment rate grew steadily through the
1980s and 1990s, with no evident acceleration after 1994 (Karmen,
2000:147). Moreover, the proportion of New York City felony arrests
resulting in a prison sentence peaked in 1993, the year before Compstat
was introduced, and declined thereafter (Karmen, 2000:155–156). A grow-
ing felony arrest rate could still have produced an increase in the size of
the prison population, even after the imprisonment-to-arrest ratio began
to decline. However, Compstat could not have contributed to the growth
in New York’s prison population that occurred before 1994, and it is not
likely that it was responsible for all of the growth after 1994.

RICHMOND

The Richmond story differs from those in Boston and New York. The
unconditional model shows that Richmond’s firearm homicide rate fell by
nearly 16% per year after Exile was introduced in 1997, but that decrease
is not significantly greater than the almost 10% average reduction in fire-
arm homicide for the sample. However, after controlling for other factors,
Richmond’s firearm homicide rate exhibits a 22% yearly decline, whereas
the average reduction for the sample remains about 10% per year. That
difference is statistically significant (p < 0.05).

This finding differs from those obtained by Raphael and Ludwig (2003)
in their evaluation of Exile, which concluded that the intervention had lit-
tle effect on Richmond’s firearm homicide rate. The discrepancy may be
from the use of a longer firearm homicide series in the current study.
Raphael and Ludwig (2003) analyzed firearm homicide rates through
1999, whereas the series used in this study extends the intervention period
two additional years to 2001. Raphael and Ludwig (2003) also omitted the
year 1997 from their analysis, on the grounds that the unusually high rate
of firearm homicide in Richmond that year constitutes an unreliable base
on which to gauge the effectiveness of Exile. We question this analytical
decision. It is not evident why 1997 should be dropped from the Richmond
data and not, for example, 1994, when the homicide rate spiked to 80 per
100,000 (see Figure 1; see, also, Raphael and Ludwig, 2003:262, Figure 7-
3B). Compared with other large cities, Richmond’s high level of homicide
is “unusual” in general and not just in 1997.

Nonetheless, it is true that assessments that include 1997, the year Exile
was introduced, are more apt to find a program effect than those from
which 1997 is excluded. We therefore omitted 1997 from our analysis and
replaced it with the 1996–1998 average firearm homicide count. Reesti-
mating the conditional model on the revised series reveals a marginally
significant downward departure of Richmond’s firearm homicide trend
from the sample average during the intervention period (p < 0.10; results

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 19 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 437

not shown). We also respecified the conditional model by including the
1992 and 1997 firearm homicide rates as additional covariates in the pre-
intervention and intervention estimations, respectively. Controlling for
between-city differences in initial levels of firearm homicide helps to
account for possible selection bias associated with the location of crime
control interventions in high-crime cities and minimizes omitted variable
bias by capturing unmeasured influences on homicide. No substantively
meaningful differences emerge from the reanalysis; in fact, the “Richmond
effect” becomes somewhat stronger with the initial firearm homicide rates
in the model (results available on request).

It would have been instructive had Raphael and Ludwig (2003) also
compared their results with and without Exile’s first year in the analysis.
However, the primary reason for the difference between the current
results and those from the Raphael and Ludwig (2003) study probably is
the inclusion of other determinants of homicide trends in our analysis. We
find a significant divergence between Richmond’s firearm homicide trend
and that for other large cities only after introducing the covariates in the
conditional model. Raphael and Ludwig (2003) acknowledge the impor-
tance of controlling for other influences on firearm homicide when exam-
ining the impact of Exile (p. 270):

The most important concern with our analysis is whether we are able
to distinguish the effects of Project Exile from those of other unmea-
sured factors that drive crime trends over time at the local level. Our
comparison of Richmond homicide trends to those of other cities is
intended to address this concern. However, such comparisons may be
invalid because of unobserved differences among cities in policing,
age structure, and other factors likely to influence homicide rates.

Their comparison of juvenile and adult homicide arrest trends is intended
to identify the effects of unmeasured factors on Richmond’s trend in fire-
arm homicide rates, on the assumption that juveniles are not subject to the
threat of federal imprisonment for using a gun in crime, but are affected
by other influences on firearm violence. But juveniles are not threatened
by state imprisonment either, and we find a significant effect of state incar-
ceration rates on firearm homicide trends (see Appendix II). Raphael and
Ludwig’s (2003) use of juveniles as a “control group” does not effectively
eliminate the effect of incarceration and perhaps other influences on Rich-
mond’s trend in firearm homicide.

In summary, we find evidence consistent with an intervention effect on
homicide trends for Richmond’s Project Exile. Richmond’s firearm homi-
cide rate fell more rapidly than the average firearm homicide rate among
large U.S. cities, with other influences controlled. We cannot rule out the
possibility that unmeasured factors are responsible for Richmond’s drop in

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 20 2-SEP-05 9:56

438 ROSENFELD, FORNANGO, & BAUMER

firearm homicides after Exile was introduced in 1997. However, they
apparently do not include changes over time in police size, drug involve-
ment, or incarceration rates, all prime candidates for explaining the
decline in big-city homicide rates during the 1990s (Rosenfeld, 2004). Nor
do they include between-city differences in resource deprivation, popula-
tion density, or unmeasured factors correlated with the initial (1997) level
of firearm homicide. Furthermore, unmeasured factors that may be
responsible for Richmond’s homicide reduction must have come into play
after and not before Exile was initiated, because Richmond’s firearm
homicide trend before 1997 did not deviate significantly from the sample
average. These results, although not definitive, amount to a fairly strong
circumstantial case for Exile’s impact, and shift the burden of proof back
to critics to identify factors not captured in our models that may explain
Richmond’s sizeable drop in firearm homicide after 1997.

CONFIDENCE IN THE RESULTS

We do not find evidence supportive of a program impact on homicide
trends for Boston’s Operation Ceasefire or New York’s Compstat. In both
instances, the homicide trends during the intervention did not diverge sig-
nificantly from the sample average, although the Boston results are mar-
ginally significant when the age range of the target group is expanded to
encompass 11–24 year-olds. Our methods may lack sufficient power to
detect Ceasefire’s effect on youth firearm homicide, although in the nature
of the case, it is difficult to identify with high confidence program effects
on low base-rate events. This point is illustrated nicely by comparing the
confidence intervals around the point estimates of the divergence of the
Boston, New York, and Richmond homicide trends from those of other
large cities during the respective intervention periods.14 As shown in Fig-
ure 2, despite the relatively large point estimate, we cannot confidently
conclude that a nonzero difference exists between Boston’s youth firearm
homicide trend and those of other large cities. The Richmond results
inspire somewhat greater confidence in the existence of a difference
between Richmond’s firearm homicide trend and the average trend for the
sample, although the difference may have been quite small. We are more
confident still in the results for New York, which are in line with earlier
criticisms that factors other than Compstat were responsible for New
York’s homicide drop during the 1990s (Eck and Maguire, 2000; Karmen,
2000).

14. We thank Christopher Winship for this suggestion.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 21 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 439

FIGURE 2. CONFIDENCE INTERVALS AROUND
CITY FIXED EFFECTS IN CONDITIONAL MODELS

-0.5

-0.4

-0.3

–

0.2

–

0.1

0

0.1

0.2

E
ff

ec
t

E
st

im
at

e

Richmond Model: Firearm
Homicide

-.186

.027

-.144

95% C.I. 90% C.I.

New York Model: Total Homicide

Boston Model: Youth Firearm
Homicide

DISCUSSION

The decade of the 1990s was a period of marked innovation in policing
practices in American cities. As new policies and procedures were intro-
duced, crime rates began to fall across the country, which resulted in
claims that the enforcement changes were responsible for the crime drop.
Criminologists generally met such claims with skepticism, but they could
not point to a substantial body of research of their own that refuted them.
Their criticisms often came off as harping or nit-picking, which prompted
some practitioners to boast that they would defeat not just crime but crim-
inology. A reporter quipped in 1995 that New York’s Police Commissioner
William Bratton would not be satisfied “until every last criminologist sur-
renders” (quoted in Karmen, 2000:75).

Researchers failed to provide the public or policy makers with clearly
articulated standards for judging the credibility of local officials’ success
claims. The standards are easy enough to identify; they are found in every
textbook on social research methods. Before an intervention can be
deemed effective in reducing crime, the observed reductions must be plau-
sibly linked to the characteristics of the intervention. Second, the observed
reductions must have occurred where and when the intervention was pre-
sent and not where and when the intervention was absent. Third, the
observed reductions must be shown to have exceeded the expected rate of
crime decline, the change that would have taken place absent the interven-
tion, what the textbooks term “history.” These criteria constitute the mini-
mum threshold, the prima facie case, for granting credence to claims that

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 22 2-SEP-05 9:56

440 ROSENFELD, FORNANGO, & BAUMER

an intervention has reduced crime. Insisting that they are met is, or should
be, the research community’s first response to official claims-making and
the basis on which the public is advised whether to take the claims
seriously.

Criminologists also must impose these minimum methodological stan-
dards on themselves. It is not the research community’s responsibility to
make the prima facie case for a program’s effectiveness; that is the burden
of program proponents. The researcher’s job is to subject the case to sys-
tematic empirical evaluation. What is remarkable about the criminological
research community’s response to the success stories told by local officials
during the 1990s crime drop is how little research went into it. Even widely
publicized innovations such as Ceasefire, Compstat, and Exile attracted no
more than a handful of outcome evaluations. And the evaluations that
were conducted are not easily compared with one another, because they
employed different empirical methods and data.

These considerations prompted this investigation. Our intention has
been to empirically assess claims made regarding reductions in homicide
associated with Ceasefire, Compstat, and Exile and to offer a model for
how such investigations might be undertaken in the future. We do not
claim to have presented the last word on either the specific interventions
or a general analytic strategy. On the contrary, given the paucity of
existing research, our effort must be viewed as a point of departure for
ongoing evaluation of the impact of local law enforcement interventions
on crime rates.

Few commonly accepted standards exist for undertaking statistical eval-
uations of crime-control interventions using observational data and
econometric methods, even though these are the data and methods we are
stuck with for evaluating large-scale initiatives such as those addressed in
this study. We propose that such evaluations meet the same requirements
we would impose on program officials: The evaluation strategy must sys-
tematically compare observed crime reductions with those that would
have occurred without the intervention. That in turn requires comparisons
with other places and controls for other influences based on nationally
representative samples, especially when evaluating widely known pro-
grams such as Ceasefire, Compstat, and Exile. Statistical models should be
applied that efficiently estimate differences across places in crime trends
before, during, and when possible after the intervention. We have chosen
piecewise growth models and hierarchical methods for these purposes, but
other approaches are defensible. What no longer can be defended are iso-
lated evaluations of significant crime-control initiatives that cannot be
compared with one another and do not even begin to rule out competing
explanations.

This is no mere academic requirement. Current federal crime control

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 23 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 441

policy in the form of Project Safe Neighborhoods (PSN) was inspired by
the presumed success of Richmond’s Exile and has been implemented in
every federal judicial district in the country. PSN’s goal is “to reduce the
violent crime rate in our communities,” and it promises “accountability”
by “measuring success based on ‘outcome’ rather than ‘output’” (http://
www.projectsafeneighborhoods.gov). PSN’s impact on violent crime
remains to be seen. But it establishes a rationale and national platform for
greater cooperation among program planners to ensure coordinated inter-
ventions and among researchers to devise common evaluation criteria. We
should expect to see greater uniformity and generalizability in the result-
ing evaluations than have characterized the research on the great policing
initiatives of the 1990s.

REFERENCES

Baumer, Eric, Janet L. Lauritsen, Richard Rosenfeld, and Richard Wright
1998 The influence of crack cocaine on robbery, burglary, and homicide rates:

A cross-city, longitudinal analysis. Journal of Research in Crime and
Delinquency 35:316–340.

Blumstein, Alfred
1995 Youth violence, guns, and the illicit-drug industry. Journal of Criminal

Law and Criminology 86:10–36.

Blumstein, Alfred and Joel Wallman (eds.)
2000 The Crime Drop in America. New York: Cambridge University Press.

Braga, Anthony
2004 Personal communication. July 13.

Braga, Anthony, David M. Kennedy, Elin J. Waring, and Anne M. Piehl
2001 Problem-oriented policing, youth violence, and deterrence: An evaluation

of Boston’s Operation Ceasefire. Journal of Research in Crime and
Delinquency 38:195–22

5.

Bratton, William, with Peter Knobler
1998 Turnaround: How America’s Top Cop Reversed the Crime Epidemic.

New York: Random House.

Butterfield, Fox
1995 Many cities in US show sharp drop in homicide rate. New York Times

(August 13):1,10.
1996 Major crimes fell in ’95, early data by FBI indicate. New York Times

(May 6):A1,A14.
1998 Reasons for dramatic drop in crime puzzles the experts. New York Times

(March 29):Section 1, p. 14.

Conklin, John
2003 Why Crime Rates Fell. New York: Allyn and Bacon.

Cook, Philip J. and John H. Laub
2002 After the Epidemic: Recent Trends in Youth Violence in the United

States. In Michael Tonry (ed.), Crime and Justice: A Review of Research,
Vol. 29. Chicago, Ill.: University of Chicago Press.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 24 2-SEP-05 9:56

442 ROSENFELD, FORNANGO, & BAUMER

Cork, Daniel
1999 Examining space-time interaction in city-level homicide data: Crack

markets and the diffusion of guns among youth. Journal of Quantitative
Criminology 15:379–406.

Eck, John and Edward Maguire
2000 Have changes in policing reduced violent crime? An assessment of the

evidence. In Alfred Blumstein and Joel Wallman (eds.), The Crime Drop
in America. New York: Cambridge University Press.

Fagan, Jeffrey
2002 Policing guns and youth violence. Future of Children 12:133–151.

Fagan, Jeffrey and Garth Davies
2001 Street stops and broken windows: Terry, race, and disorder in New York

City. Fordham Urban Law Journal 28:457–504.

Fagan, Jeffrey, Franklin E. Zimring, and June Kim
1998 Declining homicide in New York City: A tale of two trends. Journal of

Criminal Law and Criminology 88:1277–1323.

Farrington, David P.
2003 Developmental and life-course criminology: Key theoretical and empirical

issues. Criminology 41:221–255.

Feldman, Paul
1994 LA homicides drop 28% in 1st half along with other crimes. Los Angeles

Times (August 6):B1–B2.

Gillerman, Margaret
1996 Bush: E. St. Louis good guys winning. St. Louis Post-Dispatch (June 18).

Greene, William H.
2000 Econometric Analysis. 4th ed. Upper Saddle River, N.J.: Prentice-Hall.

Harcourt, Bernard E.
2001 Illusion of Order: The False Promise of Broken Windows Policing.

Cambridge, Mass.: Harvard University Press.

Herbert, Bob
1995 Good news for the City. New York Times (August 11):A15.

Horney, Julie, D. Wayne Osgood, and Ineke Haen Marshall
1995 Criminal careers in the short-term: Intra-individual in crime and its

relation to local life circumstances. American Sociological Review
60:655–673.

Joanes, Ana
2000 Does the New York City Police Department deserve credit for the decline

in New York City’s homicide rates? A cross-city comparison of policing
strategies and homicide rates. Columbia Journal of Law and Social
Problems 33:265–311.

Karmen, Andrew
2000 New York Murder Mystery: The True Story Behind the Crime Crash of

the 1990s. New York: New York University Press.

Kelling George and William Bratton
1998 Declining crime rates: Insiders’ views of the New York City Story. Journal

of Criminal Law and Criminology 88:1217–1232.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 25 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 443

Kelling George and C. Coles
1996 Fixing Broken Windows: Restoring Order and Reducing Crime in Our

Communities. New York: Free Press.

Kelling George and William H. Sousa, Jr.
2001 Do police matter? An analysis of the impact of New York City’s Police

reforms. Manhattan Institute Civic Report (December). Available online:
http://www.manhattan-institute.org/cr_22 (Accessed September 27,
2004).

Kennedy, David
1998 Pulling levers: Getting deterrence right. National Institute of Justice

Journal (July):2–8.

Kennedy, David, Anne M. Piehl, and Anthony A. Braga
1996 Youth violence in Boston: Gun markets, serious youth offenders, and a

use-reduction strategy. Law and Contemporary Problems 59:147–196.

Kennedy, David, Anthony A. Braga, Anne M. Piehl, and Elin J. Waring
2001 Reducing Gun Violence: The Boston Gun Project’s Operation Ceasefire.

Washington, D.C.: National Institute of Justice.

Krauss, Clifford
1996 Now, how low can crime go? New York Times (January 28):Section 4, p.

5.

Lacayo, Richard
1996 Law and order. Time (January 15):48–54.

LaFree, Gary
1998 Losing Legitimacy: Street Crime and the Decline of Social Institutions in

America. Boulder, Colo.: Westview.

Land, Kenneth C., Patricia L. McCall, and Lawrence E. Cohen
1990 Structural covariates of homicides rates: Are there any invariances across

time and social space? American Journal of Sociology 95:922–963.

Lardner, James
1997 Can you believe the New York miracle? New York Review (August

14):54–58.

Law Enforcement News
1997 UCR forecasts 5th consecutive crime dip. (February 14):5.

Levitt, Steven D.
2002 Deterrence. In James Q. Wilson and Joan Petersilia (eds.), Crime: Public

Policies for Crime Control. Oakland, Calif.: ICS Press.

Marvell, Thomas B. and Carlisle E. Moody, Jr.
1997 The impact of prison growth on homicide. Homicide Studies 1:205–233.

Messner, Steven F. and Richard Rosenfeld
1998 Social structure and homicide: Theory and research. In M. Duane Smith

and Margaret A. Zahn (eds.), Homicide: A Sourcebook of Social
Research. Thousand Oaks, Calif.: Sage.

Ousey, Graham C. and Matthew R. Lee
2002 Examining the conditional nature of the illicit drug market-homicide

relationship: A partial test of the theory of contingent causation.
Criminology 40:73–102.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 26 2-SEP-05 9:56

444 ROSENFELD, FORNANGO, & BAUMER

Piehl, Anne Morrison, Suzanne J. Cooper, Anthony A. Braga, and David M.
Kennedy
2003 Testing for structural breaks in the evaluation of programs. Review of

Economics and Statistics 85:550–558.

Raphael, Steven and Jens Ludwig
2003 Prison sentence enhancements: The case of project exile. In Jens Ludwig

and Philip J. Cook (eds.), Evaluating Gun Policy: Effects on Crime and
Violence. Washington, D.C.: Brookings Institution Press.

Raudenbush, Stephen W. and Anthony S. Bryk
2002 Hierarchical Linear Models: Applications and Data Analysis Methods. 2d

ed. Thousand Oaks, Calif.: Sage.

Raudenbush, Stephen W., Anthony Bryk, and Richard Congdon
2001 HLM 5.04: Hierarchical Linear and Nonlinear Modeling. Lincolnwood,

Ill.: Scientific Software International, Inc.

Rosenfeld, Richard
2002 The crime decline in context. Contexts 1:25–34.
2004 The case of the unsolved crime decline. Scientific American 290:68–77.

Spelman, William
2000 The limited importance of prison expansion. In Alfred Blumstein and Joel

Wallman (eds.), The Crime Drop in America. New York: Cambridge
University Press.

Wilson, James Q. and George Kelling
1982 Broken windows. Atlantic Monthly (March):29–38.

Winship, Christopher and Jenny Berrien
1999 Boston cops and black churches. Public Interest (Summer):52–68.

Richard Rosenfeld is Professor of Criminology and Criminal Justice at the University
of Missouri-St. Louis. He is co-author with Steven F. Messner of Crime and the Ameri-
can Dream (Wadsworth, Third ed., 2001) and has written extensively on the social
sources of violent crime. His current research focuses on explaining crime trends in the
United States. Professor Rosenfeld is a member of the National Academy of Sciences’
Committee on Law and Justice and a Fellow of the American Society of Criminology.

Robert J. Fornango is a Ph.D. candidate at the University of Missouri-St. Louis. His
research interests include the effects of community changes on violent crime trends,
reciprocal effects of violence on community change, and the spatial dynamics of crime.

Eric Baumer is an Associate Professor of Criminology and Criminal Justice, Univer-
sity of Missouri-St. Louis. He received his Ph.D. in sociology from the University at
Albany, State University of New York in 1999. His research is concerned primarily
with how social structural and cultural features of communities affect individual behav-
ior. He has examined this general issue empirically in multilevel studies of the influ-
ence of neighborhood characteristics on the nature of crime, the mobilization of law,
and the prevalence of various forms of non-normative behavior, in macro-level studies
of urban crime levels and trends, and in case studies of crime and social control in
Iceland and the Republic of Malta. Baumer has published widely in major sociology
and criminology journals including the American Journal of Sociology, American Socio-
logical Review, Social Forces, and Criminology.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 27 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 445

APPENDIX I

HGLM is a multilevel modeling strategy available for growth-curve
analysis. The HGLM for count data uses an overdispersed Poisson sam-
pling model at level-1 and a log link function to equate the transformed
count to a linear structural model. In this analysis, the model assumes an
expected homicide count E (Yitlit) = mitlit , where lit is the homicide rate
of city i at time t and mit is the exposure—the city population expressed in
100,000s. The expected homicide rate for a city is transformed through a
natural logarithm, such that hit = log(lit). The log-event rate, hit, becomes
the dependent variable in level-1 of the model. The linear structural model
thus becomes

q s
hti = b0i + ∑ bpi (Tti) + ∑ bri (Xti − X̄•i)

p=1 r=1

in level-1. Here,
q
∑ bpi (Tti)

p=1

is a series of individual linear growth estimates during the pre-interven-
tion, intervention, and post-intervention periods, and

s
∑ bri (Xti − X̄•i)
r=1

is a series of time varying covariate estimates that are group-centered to
measure within-city change (cf. Horney et al., 1995).

When the outcome for an observation is modeled as a function of time,
the indicators of temporal variation for that unit are included in the level-1
equation. Variables hypothesized to explain differences in trend parame-
ters across units are then entered into the model at level-2. From the
within-city change model at level-1, variations in the parameter estimates
of the intercept and time components between cities are modeled at level-
2. The level-2 intercept model is

q s
b0i = p00 + ∑ p0p (Wi) + ∑ p0r (X̄i) + p03(Di) + u0i,

p=1 r=1

where
q
∑ p0p (Wi)

p=1

includes the effects of resource deprivation and ln(population density),

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 28 2-SEP-05 9:56

446 ROSENFELD, FORNANGO, & BAUMER

s
∑ p0r (X̄i)
r=1

includes the within-city average for the time varying covariates, and
p03(Di) is the parameter estimate for a dummy variable representing the
city of interest. Between-city differences in the level-1 time trends are
modeled at level-2 as

r
bpi = pp0 + ∑ ppq (Wi) + pp2 (Di) + u1i,

q=1

where Wi and Di remain as above. Finally, the between-city variation in
parameter estimates for time varying covariates is estimated as a function
of the grand mean effect: bri = pr0 + uri. Each level-2 model is specified with
a city-level random effect.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 29 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 447

APPENDIX II. HGLM RESULTS FOR BOSTON, NEW
YORK, AND RICHMOND, 1992–2001 (N = 95)15

A. Boston
Fixed Effects Unconditional Conditional

Initial (1992) Youth Firearm Homicide Rate, p0i

Base, b00 1.435*** .339
(.104) (.782)

Boston, b01 .212 −.146
(.992) (.632)

Resource Deprivation, b02 .509***
(.096)

ln Population Density, b03 .015
(.108)

Mean Police Size, b04 .002***
(.001)

Mean Incarceration Rates, b05 .001**
(.000)

Mean Cocaine Prevalence, b06 .004
(.012)

Pre-Intervention Change (1992–1995), p1i

Base, b10 −.015 .188
(.017) (.148)

Boston, b11 .006 −.008
(.149) (.161)

Resource Deprivation, b12 −.003
(.017)

ln Population Density, b13 −.024
(.018)

Intervention Change (1996-1999), p2i

Base, b20 −.151*** −.173
(.014) (.135)

Boston, b21 −.134 −.186
(.123) (.113)

Resource Deprivation, b22 .007
(.012)

ln Population Density, b23 .006
(.016)

Post-Intervention Change (2000–2001), p3i

Base, b30 .012 −.043
(.027) (.269)

Boston, b31 .372 .305
(.238) (.204)

Resource Deprivation, b32 .059**
(.025)

ln Population Density, b33 .007

15. Conventional model fit statistics, such as the deviance statistic, are not
available for Poisson count models in HGLM (Raudenbush et al., 2001:ch. 5.1).
However, some sense of model fit can be gained by comparing the variance parameters
from the unconditional and conditional specifications of the intercept model. By this
measure, the models reported here would seem to fit the data reasonably well. For
example, the residual variance component from the unconditional intercept model for
New York is 0.6326; the residual variance from the conditional intercept model is
0.2634. The percentage reduction in error between the unconditional and conditional
specifications is, therefore, (0.6326/(0.2634 – 0.6326)) * 100 = −58.4%. The
corresponding reductions in error for Boston and Richmond are –57.7% and –53.4%,
respectively.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 30 2-SEP-05 9:56

448 ROSENFELD, FORNANGO, & BAUMER

(.032)
Police Size, p4i −.001

(.002)
Incarceration Rates, p5i −.001***

(.000)
Cocaine Prevalence, p6i .000

(.004)

Random Effects Variance Estimate Variance Estimate

Initial Homicide Rates, r0i .9511*** .4022***
Pre-Intervention Change, r1i .0163*** .0209
Intervention Change, r2i .0101*** .0128**
Post-Intervention Change, r3i .0315*** .0259
Police Size, r4i .0001
Incarceration Rates, r5i .0000*
Cocaine Prevalence, r6i .0002**

B. New York
Fixed Effects Unconditional Conditional

Initial (1992) Homicide Rate, p0i

Base, b00 2.697*** 2.048***
(.084) (.598)

New York City, b01 .612 −.521
(.800) (.493)

Resource Deprivation, b02 .363***
(.077)

ln Population Density, b03 −.039
(.082)

Mean Police Size, b04 .003***
(.001)

Mean Incarceration Rates, b05 .0001
(.000)

Mean Cocaine Prevalence, b06 .004
(.010)

Pre-Intervention Change (1992–1993), p1i

Base, b10 .066*** .452**
(.023) (.200)

New York City, b11 −.209 .046
(.159) (.089)

Resource Deprivation, b12 .013
(.018)

ln Population Density, b13 −.042*
(.024)

Intervention Change (1994–2001), p2i

Base, b20 −.079*** .040
(.006) (.059)

New York City, b21 −.090* .027
(.048) (.041)

Resource Deprivation, b22 .006
(.005)

ln Population Density, b23 −.013*
(.007)

Police Size, p3i −.002
(.001)

Incarceration Rates, p4i −.001**
(.000)

Cocaine Prevalence, p5i −.005
(.004)

Random Effects Variance Estimate Variance Estimate

Initial Homicide Rates, r0i .6326*** .2634***
Pre-Intervention Change, r1i .0241*** .0052***

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 31 2-SEP-05 9:56

POLICE INTERVENTIONS & HOMICIDE 449

Post-Intervention Change, r2i .0023*** .0046***
Police Size, r3i .0001***
Incarceration Rates, r4i .0000***
Cocaine Prevalence, r5i .0012***

C. Richmond
Fixed Effects Unconditional Conditional

Initial (1992) Firearm Homicide Rate, p0i

Base, b00 2.358*** 1.658**
(.093) (.646)

Richmond, b01 1.593* .700
(.891) (.549)

Resource Deprivation, b02 .420***
(.075)

ln Population Density, b03 −.039
(.090)

Mean Police Size, b04 .003***
(.001)

Mean Incarceration Rates, b05 .001**
(.000)

Mean Cocaine Prevalence, b06 −.002
(.010)

Pre-Intervention Change (1992–1996), p1i

Base, b10 −.064*** .175
(.012) (.133)

Richmond, b11 .083 .113
(.107) (.111)

Resource Deprivation, b12 −.003
(.013)

ln Population Density, b13 −.026
(.016)

Intervention Change (1997–2001), p2i

Base, b20 −.103*** −.108
(.009) (.098)

Richmond, b21 −.069 −.144**
(.075) (.067)

Resource Deprivation, b22 .025***
(.009)

ln Population Density, b23 .003
(.012)

Police Size, p3i −.004**
(.002)

Incarceration Rates, p4i −.001*
(.0006)

Cocaine Prevalence, p5i −.004
(.004)

Random Effects Variance Estimate Variance Estimate

Initial Homicide Rates, r0i .7806*** .3637***
Pre-Intervention Change, r1i .0106*** .0219***
Post-Intervention Change, r2i .0049*** .0041***
Police Size, r3i .0001***
Incarceration Rates, r4i .0000***
Cocaine Prevalence, r5i .0005***

* p < 0.10; **p < 0.05; ***p <0.01.

\\server05\productn\C\CPP\4-3\CPP310.txt unknown Seq: 32 2-SEP-05 9:56

450 ROSENFELD, FORNANGO, & BAUMER

April 2004

The Grow

t

h of Compstat in American Policing

By David Weisburd, Stephen D. Mastrofski, Rosann Greenspan, and James J. Willis

Introduction

Introduced as recently as 1994 by then commissioner William Bratton of the New York City
Police Department, Compstat has already been recognized as a major innovation in American
policing. In the few years since its appearance, police departments around the country have
begun to adopt Compstat or variations of it (Law Enforcement News 1997; Maas 1998;
McDonald 1998). The program has received national publicity, including awards from Harvard
University and former Vice President Al Gore, and has been credited by its originators and
proponents with impressive reductions in crime and improvements in neighborhood quality of
life (Bratton 1999; Gurwitt 1998; Remnick 1997; Silverman, 1996).

Despite the national attention that has been paid to Compstat, to date there has been little
systematic analysis of Compstat programs in policing. In fact, most of what we know comes
from those involved in its implementation (Sparrow et al. 1990; Moore 1995). With support from
the National Institute of Justice, the Police Foundation has tried to further our knowledge of
Compstat by undertaking a national study of the program (Weisburd et al. 2001; Weisburd et al.
2003). Drawing on a representative survey of American police departments, this report examines
the diffusion of Compstat and factors associated with its implementation.

The Emergence of Compstat

The particulars of Compstat’s origins have been detailed elsewhere (Bratton 1998; Kelling and
Coles 1996; Maple 1999; Silverman 1999; McDonald et al. 2001). The impetus behind Compstat
was New York City’s police commissioner William Bratton and his efforts to make a huge
organization, legendary for its resistance to change (Sayre and Kaufman 1960), responsive to his
leadership—a leadership that had clearly staked out crime reduction and improving the quality of
life in the neighborhoods of New York City as its top priorities (Bratton 1999). Based on his
belief in principles of strategic leadership (Bratton 1998; Silverman 1999) and his own

This project was supported by Grant Number 98-IJ-CX-007 by the National Institute of Justice, Office of
Justice Programs, U.S. Department of Justice. Points of view or opinions contained in this document are
those of the authors and do not necessarily represent the official position or policies of the U.S.
Department of Justice.

© 2001 Police Foundation. All rights reserved.

POLICE FOUNDATION REPORTS

The Growth of Compstat in American Policing Police Foundation Reports
Page 2 April 2004

experiences with the Boston Police Department and the New York City Transit Police, Bratton
and his lieutenants set out to disprove skeptics who claimed that the police can do little about
crime and disorder.

At the outset, Bratton and his administration’s analysis of the NYPD’s problems revealed several
deficiencies that have long been identified as forms of bureaucratic dysfunction (Merton 1940).
First, the organization lacked a sense of the importance of its fundamental crime control mission.
Second, because the NYPD was not setting high enough expectations about what its officers
could do and accomplish, a lot less was getting done than was possible. Third, too many police
managers had become moribund and were content to continue doing things the way they had
always been done rather than exploring new theories and studies for promising strategies to
reduce crime and improve the quality of life in neighborhoods. Fourth, the department was beset
with archaic, unproductive organizational structures that did more to promote red tape and turf
battles than to facilitate teamwork to use scarce resources effectively. As a result, operational
commanders were “handcuffed” by headquarters and lacked authority to customize crime control
to their precincts’ individual needs. Finally, the department was “flying blind”. It lacked timely,
accurate information about crime and public safety problems as they were emerging; had little
capacity to identify crime patterns; and had difficulty tracking how its own resources were being
used. Since middle managers were not in the habit of monitoring these processes, they served as
a weak link in the chain of internal accountability between top brass and street-level, police
employees.

Bratton used a “textbook” approach to deal with these problems, following the major
prescriptions offered by organizational development experts to accomplish organizational
change (Beer 1980). He brought in outsiders to obtain a candid diagnosis of the organization’s
strengths and weaknesses. He incorporated both top-down and bottom-up processes to
implement change (Silverman 1996). He sought and obtained early indicators of the success of
the change efforts and sought ways to reinforce the individual efforts of his precinct commanders
and the rank-and-file by using both incentives and disincentives (Bratton 1996).

Strictly speaking, Compstat refers to a “strategic control system” developed to gather and
disseminate information on the NYPD’s crime problems and to track efforts to deal with them.
As such, it addresses the problem of inadequate information described above and, in this sense, it
is a structure intended to serve the implementation of the NYPD’s crime-control and quality-of-
life strategies (Office of Management Analysis and Planning n.d.: 1). At the same time,
Compstat has become shorthand for the full range of strategic, problem-solving activities in the
NYPD. These elements of the department’s Compstat approach are most visible in the twice-
weekly Compstat “Crime-Control Strategy Meetings,” where precinct commanders appear
before several of the department’s top brass to report on crime problems in their precincts and
what they are doing about them.

The Growth of Compstat in American Policing Police Foundation Reports
Page 3 April 2004

This occurs in a data-saturated environment in which Compstat reports play a central role.
Precinct crime statistics and other information about a precinct and its problems are projected
onto overhead screens, and commanders respond to queries about what they are doing to deal
with those problems. Crime data that were once three to six months late are now available to
precinct commanders on a weekly basis for the preceding week. The report includes weekly,
monthly, and annual tallies of crime complaints, arrests, summonses, shooting incidents and
victims, organized by precinct, borough, and citywide. In addition, electronic pin maps are
generated to show how crimes and police activities cluster geographically. Hour-of-the-day
analyses and “crime spike” analyses are also carried out. The reports also profile the background
of the precinct commander, as well as other features of the precinct under his or her command,
such as demographic data, workload data, and various activities.

Compstat reports serve as the database for commanders to demonstrate their understanding of the
crime problems in their areas and discuss future strategies with the top brass and other
commanders present. Cross-unit coordination is planned if necessary and all of the plans are
thoroughly documented. When the precinct is reselected for participation in a Compstat meeting,
the commander must demonstrate that he or she has followed up on these strategies. Sometimes
commanders bring subordinates with them so that they can report on their efforts and receive
recognition. The press and other outside agencies are sometimes invited to attend these sessions
with as many as 200 people in attendance, thus providing “great theater” and developing a
greater public awareness of how the department is being managed (Bratton 1998, 296).

There are indicators that police leaders around the nation are interested in and willing to explore
Compstat but we do not know how widely Compstat models have diffused across the United
States or what types of departments are most likely to develop Compstat programs. Why are
American police departments adopting the Compstat model? Below we present answers to these
questions based on our national survey of police agencies.

Research Methods

We sent our survey to all American police agencies with over 100 sworn police officers and to a
sample of 100 agencies with between 50 and 100 sworn officers (see Weisburd et al. 2001).1
The full universe of larger departments was sampled because we believed that Compstat
programs are most appealing to such departments and thus most likely to be implemented in
them. We thought it important, nevertheless, to assess whether smaller agencies are also
beginning to develop Compstat-like programs. It would have been prohibitively costly to survey
all smaller agencies, but our random sample of agencies with 50-99 officers allows us to assess
whether Compstat programs are also influencing smaller departments. We decided not to sample
from among departments with fewer than 50 full-time, sworn officers because we thought it

1. Our instrument was reviewed by a group of academics and practitioners including Eli Silverman, Edward Maguire, Richard
Ritti, Lorraine Green Mazerolle, Roger Parks, Scott Keeter, Frank Gajewski, Christopher Tutko, and Thomas Frazier.

The Growth of Compstat in American Policing Police Foundation Reports
Page 4 April 2004

reasonable to assume that such police agencies lack the resources and organizational complexity
to implement Compstat.

At the time of our sample selection in 1999, the most complete, current listing of American
police agencies was the 1996 Directory Survey of Law Enforcement Agencies conducted by the
U.S. Bureau of the Census and the U.S. Bureau of Justice Statistics (BJS) which gave us both the
file and its documentation (Bureau of Justice Statistics 1998)2. According to the directory, there
were 515 agencies with 100 or more sworn officers, and 698 agencies with 50-99 officers. We
sent the survey instrument by mail to all of the 515 largest agencies and a random sample of 100
agencies with between 50 and 99 officers.3 This mailing included a letter asking the chief to
complete (or to delegate to a person who could reflect his/her views) the part of the survey
relevant to overall departmental policy and someone familiar with technology to complete those
sections of the survey.4 We assured the departments of complete confidentiality and included a
survey instrument with a unique identification number affixed and a stamped, addressed, return
envelope. We followed up with a series of phone calls as well as a second and third mailing. The
first mailing occurred on August 18, 1999, and the final surveys were received in January of
2000. The overall response rate of 86 percent achieved using this method was very high for a
mail survey (see Table 1).

Table 1. Response Rate for the Sample

DEPARTMENT SIZE RECEIVED/TOTAL PERCENT
Small (50-99 Sworn) 85/100 85
Large (100 + Sworn) 445/515 86.4
Total 530/615 86.2

We found no systematic reasons for non-response by selected departments. We received about
the same proportion of responses from larger departments as from smaller ones (see Table 1).
Moreover, there are relatively small differences in our response rate across regions (see Table 2),
though departments in the South and West were somewhat more likely to return the survey.
When we compare the distribution of our sample in terms of size of department to the BJS
Directory Survey in 2000 we find that our sample is representative of the population of police
agencies in the United States (see Figure 1).5

2. We thank Edward Maguire for his assistance in the selection of the study sample.
3. We conducted a pretest in fifteen departments—five small and ten large.
4. Part I of the survey was filled out by the chief executive officer (i.e., chief, director, superintendent, or
commissioner) in half of the departments.
5. The slight underrepresentation of the larger departments might be due to the increase in sworn officers that
resulted from the Violent Crime Control and Law Enforcement Act of 1994 program to hire more police on the
street.

The Growth of Compstat in American Policing Police Foundation Reports
Page 5 April 2004

Table 2. Department Response Rate by Region6

REGION RECEIVED/TOTAL PERCENT
Northeast 119/146 81.5
North Central 102/122 83.6
South 192/215 89.3
West 117/132 88.6

Figure 1. Percent of Departments of a Certain Size in the Sample as Contrasted with

the BJS Directory Survey

0 20 40 60 8

0

100-249

250-499

500-999

1000 +

N
um

be
r o

f S
w

or
n

Percent

Compstat Survey Directory Survey

6. The Northeastern region includes the states of Connecticut, Maine, Massachusetts, New Hampshire, New Jersey,
New York, Pennsylvania, Rhode Island, and Vermont. The South includes Alabama, Arkansas, Delaware, District
of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South
Carolina, Tennessee, Texas, Virginia, and West Virginia. The North Central region includes the states of Illinois,
Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and
Wisconsin. And the Western region includes Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana,
Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

The Growth of Compstat in American Policing Police Foundation Reports
Page 6 April 2004

How Widely Has Compstat Been Adopted and What Types of Departments Are
Implementing Compstat?

Our first concern is simply whether Compstat models have been adopted widely across
American police agencies. This has been the impression of commentators but has not been
backed up with hard evidence. Our study suggests that Compstat has in fact diffused widely
across the landscape of American policing (see Table 3).7 A third of departments with 100 or
more sworn officers in our study responded “yes” when asked whether they had “implemented a
Compstat-like program.”8 An additional quarter of the large departments in our survey claimed
to be “planning” a Compstat-like program. As we expected, departments in our small department
sample were much less likely to have implemented a Compstat model. Only nine departments or
11 percent of the departments with between 50 and 99 sworn officers had done so. However,
almost 30 percent claimed to be planning to implement a Compstat program. Because the
number of departments in our sample with between 50 and 99 sworn officers that have
implemented a Compstat model is small, unless otherwise noted in the tables below, we examine
characteristics of Compstat in the large department sample only.

Table 3. Has Your Department Implemented a Compstat-Like Program?

DEPARTMENT SIZE Percent
Yes

Percent No, But
Planning

Percent
No

Small (50-99 Sworn) 11.0 29.3 59.8

Large (100 + Sworn) 32.6 25.6 41.8
Due to rounding, rows may not add to 100.

We also asked departments when their Compstat program was implemented. As would be
expected, the large growth in implementation of Compstat programs occurs after New York’s
program had begun to gain wide-scale publicity (see Figure 2). Compstat implementation was
greatest in 1998. The downward trend in 1999 may be an artifact of our study, since some
departments who responded quickly to our survey may have implemented a Compstat program
later in that year.

7. To simplify interpretation, we generally do not report the N of cases in the tables that follow. Overall, there were
very few missing values associated with the items in the survey (mean=2.7%).
8. This question followed a section of the survey that provided a list of eleven “features that have been associated
with Compstat and similar programs instituted in other departments.”

The Growth of Compstat in American Policing Police Foundation Reports
Page 7 April 2004

Figure 2. The Year of Compstat Implementation

Interestingly, eighteen departments in our large agency sample report implementation before
1994, the year the NYPD introduced Compstat. How could departments claim to have
implemented a Compstat-like program before New York City coined the term? It appears that in
such cases, departments believed that they had implemented the essential elements of Compstat
even before New York City’s model had become prominent. This is illustrated in Table 4 which
reports the percentage of departments that claimed to have implemented specific features
“associated with Compstat and similar programs” at least six years before the survey, a time that
predates the creation of Compstat in New York City. Twenty-six percent of departments said that
they “set specific objectives in terms that can be precisely measured” or that they held “regularly
scheduled meetings with district commanders to review progress toward objectives.” Thirty
percent report using data to “assess progress towards objectives” before 1994.

0

10

20

30

40

50

1994 1995 1996 1997 1998 1999
Year Compstat Implemented

N
um

be
r o

f D
ep

ar
tm

en
ts

The Growth of Compstat in American Policing Police Foundation Reports
Page 8 April 2004

Table 4. Was This Feature (of Compstat) Implemented Six or More Years Ago?

Survey Item Percent
Yes

Set specific objectives in terms that can be precisely measured 26.0
Hold regularly scheduled meetings with district commanders to review progress
toward objectives 26.3

Hold middle managers responsible for understanding crime patterns and initiating
plans to deal with them 22.7

Give middle managers control over more resources to accomplish objectives 23.1
Use data to assess progress toward objectives 30.2
Develop, modify, or discard problem-solving strategies based on what the data
show 24.8

Our survey shows that larger American police agencies claim to have adopted Compstat at a high
rate and very rapidly. How does this compare with the adoption of other social or technological
innovations? In recent years, there has been growing interest in the analysis of innovation which
has been found to have a fairly consistent form called the “s” curve of innovation (Rogers,
1995). The “s” curve is developed by measuring the cumulative adoption of an innovation over
time. In Figure 3, the innovation adoption curve for Compstat-like programs in police agencies
with over 100 sworn officers is presented.

The Growth of Compstat in American Policing Police Foundation Reports
Page 9 April 2004

Figure 3. Observed Compstat Cumulative Adoption Curve Based on Survey
Findings

0

20

40

60

80

100

120

140

160

19
80

19
81

19
82

19
83

19
84

19
85

19
86

19
87

19
88

19
89

19
90

19
91

19
92

19
93

19
94

19
95

19
96

19
97

19
98

19
99

Year

N
um

be
r o

f l
ar

ge
d

ep
ar

tm
en

ts

Can we argue from this that the diffusion of Compstat-like programs suggests a rapid rate of
innovation? Arnulf Grübler (1991) provides a yardstick. He analyzes two samples of
technologies, including such areas as energy, transport, communication, agriculture, military
technologies, as well as some social changes such as literacy, in the United States for which data
on diffusion of innovation were available. He constructs a measure, delta t, which is the time
period it takes for an innovation to go from 10 percent to 90 percent of its saturation or highest
level of adoption. He finds that between 13 and 25 percent of different types of technology
progress from 10 percent to 90 percent of their saturation level within fifteen years. Another 25
to 30 percent of his samples reached this saturation level in thirty years.

Calculation of the delta t precisely for a Compstat-like program is not possible before the
saturation process is complete. However, we can estimate the cumulative adoption curve using
the data available from our survey. Rogers (1995, 257) notes that the adoption of an innovation
generally “follows a normal bell-shaped curve” when plotted over time as a frequency
distribution. In Figure 4, we develop a cumulative adoption curve based on this assumption
extrapolating from our observed data.9 Based on this distribution and allowing saturation to
include all police departments in our sample, we estimate a 90 percent saturation level between

9 In estimating the normal frequency distribution upon which the s curve is based, we relied upon the observed data
between 1995 and 1998. We excluded 1999 because of the timing of the survey, which likely underrepresented the
number of adoptions. We also excluded years before 1995 because the number of cases were relatively small and
likely to lead to unstable estimates. In developing an estimated value for the standard deviation unit of the normal
curve, we compared each year’s frequency between 1995 and 1998 and then took the average estimate gained. After
defining the normal frequency distribution, we then converted the estimates to a cumulative distribution curve.

The Growth of Compstat in American Policing Police Foundation Reports
Page 10 April 2004

0

50

100

150

200

250

300

350

400

450

500

19
74

19
81

19
83

19
85

19
87

19
89

19
91

19
93

19
95

19
97

19
99

20
01

20
03

20
05

20
07

20
09

20
11

20
13

20

15

20
17

20
19

20
21

20
23

20
25

20
27

20
29

Year

N
um

be
r o

f L
ar

ge
D

ep
ar

tm
en

ts

Predicted number of departments Observed number of departments

2006 and 2007. As a 10 percent saturation using the observed data was defined as occurring
between 1996 and 1997, our estimate of delta t is about ten years. Accordingly, if the adoption of
Compstat-like programs were to follow the growth patterns observed in our data, Compstat
would rank among the most quickly diffused forms of innovation.

Figure 4. The Extrapolated Cumulative Adoption Curve for Compstat-like
Programs

90% Saturation
predicted in 2006.6

10% Saturation
occurs in 1996.4

The Growth of Compstat in American Policing Police Foundation Reports
Page 11 April 2004

The Role of the New York Police Department in the Diffusion of Compstat-like Programs

While a number of departments claim to have implemented elements of Compstat before New
York formally introduced this model, the influence of the New York Police Department and its
centrality in the diffusion of Compstat models is reflected in the large number of police agencies
that came to New York to learn about Compstat (see Figure 5). An overwhelming number of
departments who observed a Compstat meeting or department did so at the NYPD. While
departments that have implemented Compstat-like programs have also visited Los Angeles, New
Orleans, or Broward County, Florida, all places that have well publicized Compstat programs,
New York is clearly the site where most police agencies go to learn about this innovation.

Figure 5. Where Compstat Departments Observed a Compstat Meeting

The profound influence of New York City’s promotion of Compstat becomes even more
apparent when considering the level of familiarity the surveyed departments claim to have with
New York City’s Compstat program. Table 5 shows that fully 40 percent of the smallest
agencies that had not implemented a Compstat-like program considered themselves very or
somewhat familiar with the NYPD’s program. The percentage of the non-Compstat departments
claiming familiarity increases with each size category, reaching 90 percent for the largest
departments. A similar pattern (albeit at higher levels) is shown for Compstat departments.

0 20 40 60 80

NYPD

LAPD

New Orleans

Broward Co.

Number of Departments (Compstat only)

The Growth of Compstat in American Policing Police Foundation Reports
Page 12 April 2004

Table 5. Familiarity with the NYPD’s Compstat by Department Size (Small Agency
Sample Included)

Percent Very or Somewhat Familiar

with New York City’s Compstat Program Number of Sworn Compstat-like program not
implemented

Compstat-like program
implemented

50-99 40.3 71.4
100-299 55.7 73.2
300-499 66.7 100.0

>500 90.3 97.6

Characteristics of Compstat Departments

The relationship between department size and the implementation of Compstat is not restricted
to a broad comparison between the largest and smallest departments (as was illustrated in Table
1). As Figure 6 illustrates, there is a direct linear relationship between Compstat programs and
department size across our sample. Almost 60 percent of departments with 500 or more sworn
officers claim to have implemented a Compstat-like program. Forty-four percent of departments
with between 300 and 499 sworn officers, and 31 percent of departments with between 200 and
299 sworn, say that they have established a Compstat-like program. This relationship between
department size and implementation of a Compstat-like program is strong and statistically
significant (p<.001).

Figure 6. Implementation of Compstat and Department Size

(Small Agency Sample Included)

0

15

30

45

60

50-99 100-199 200-299 300-499 500 +
Number of Sworn

Pe
rc

en
t w

ith
C

om
ps

ta
t

The Growth of Compstat in American Policing Police Foundation Reports
Page 13 April 2004

We also find a statistically significant relationship of p<.05 between geographic region and
implementation of Compstat-like programs (see Figure 7), though the relationship is not as
strong as that of size of department. Over 40 percent of the departments with over 100 sworn
officers in the South have implemented Compstat. This can be contrasted with the Northeast
where only 26 percent of departments claim that they have implemented a Compstat-like
program. We think that this distribution reflects a more general phenomenon in American
policing over the last decade. While innovation, as in the case of Compstat, may begin in older
police agencies in the East or Central regions of the country, police agencies in the South and
West are, on average, more willing or perhaps more able to adopt those innovations.

Figure 7. Compstat Departments by Region

Motivations for Adopting Compstat

While the survey did not ask respondents to indicate directly the motivations or priorities that led
to the implementation of Compstat, it affords an opportunity to observe patterns from which we
might infer such motivations. Respondents were asked to rank the top five goals that the chief
executive pursued in the previous twelve months, selecting from a list of nineteen.10 We

10. The nineteen goals in the order listed were: reduce serious crime; reduce quality of life offenses; reduce fear of
crime; reduce calls for service; increase citizen satisfaction with the police; increase service to citizens living in
high-crime areas; increase efficiency of service (reduce cost per unit of service); reduce conflict among different
segments of the community; increase citizen participation in police programs; increase citizens’ ability to make their
own neighborhoods better places to live; give citizen groups more influence over police policy and practice; improve
coordination with other public and private organizations; reduce complaints about police misbehavior; increase

0

10

20

30

40

50

South West Northeast North Central

P
er

ce
nt

w
ith

C
om

ps
ta

t

The Growth of Compstat in American Policing Police Foundation Reports
Page 14 April 2004

assigned a score of five to the top goal identified by each respondent, a four to the second
ranking goal and so on, giving all unranked goals a score of zero. Because we wanted to examine
priorities of departments close to when they implemented a Compstat program we excluded all
departments that had implemented Compstat before 1998. We compare these departments with
those that stated in the survey that they had not implemented a Compstat-like program and they
were not planning to do so.

The average ranking for the nineteen goals was .78 for the large department sample. Only four of
the nineteen items showed a statistically significant difference (p<.05) between the two groups of
departments (see Table 6). Accordingly, there is a good deal of consensus in these police
agencies regarding the priority goals for policing. However, departments that had recently
implemented Compstat tended to rank the reduction of serious crime and increasing management
control over field operations substantially higher than departments that were not planning
implementation of Compstat. Departments that were not planning to implement a Compstat-like
program tended to score much higher than departments that claimed to recently have adopted
Compstat on the ranks they assigned to improving officer policing skills and employee morale.

Table 6. Comparing Top Goals of Compstat Departments (pre-1998 implementation) with
Departments Not Planning to Implement Compstat: Statistically Significant
Differences

Average Rank of Goal
Compstat-like program

implemented
after 1997

(n=79)

Not planning

implementation
(n=178)

Reduce serious crime 3.32 2.26
Increase police managers’ control over field operations .91 .44
Improve officers’ policing skills .46 .96
Improve employee morale .28 .68

Departments that had recently implemented Compstat gave the reduction of serious crime a
priority ranking 1.5 (3.32/2.26) times that of departments not planning to implement Compstat,
and increasing management control a ranking of 2.1 (.91/.44) times that of such departments.
Similarly, though in reverse, departments not planning to implement Compstat gave priority
rankings to improving police officer skills that were on average 2.1 (.96/.46) times those of
agencies that had claimed to recently implemented a Compstat-like program, and priority

police managers’ control over actual field operations; improve officers’ policing skills; improve employee morale; be
more responsive to the priorities of individual neighborhoods; provide better service to crime victims; and improve
the physical appearance of neighborhoods.

The Growth of Compstat in American Policing Police Foundation Reports
Page 15 April 2004

rankings for improving employee morale that were on average 2.4 (.68/.28) times those of such
agencies. This pattern is consistent with the interpretation that the dominant motivations for
implementing Compstat are to secure management control over field operations that will reduce
serious crime. At the same time, focus on improving skills and morale of street level officers—
which, for example, have been high priorities in many community-policing programs—are
relatively lower priorities for recently implemented Compstat departments.

Conclusions

Our study confirms what many police observers have noted: that Compstat has literally burst
onto the American police scene. Our survey took place a few years after the development of
Compstat in New York City, but it shows that police agencies throughout the United States have
begun to adopt Compstat-like programs. We also find that the adoption of Compstat-like
programs in police agencies follows a process of diffusion of innovation that is rapid as
compared with innovations in other social and technological areas.

Our study also suggests that Compstat is being differentially implemented in police agencies.
Not surprisingly, larger police agencies are more likely to adopt Compstat-like programs. We
suspect that this is due to the relevance of Compstat for reinforcing management control in larger
police agencies, where hierarchical control tends to be more problematic. Whatever the cause,
there is a direct linear relationship in our study between the adoption of Compstat-like programs
and the size of a police agency. We also find that agencies in the South and West of the country
have been more likely to adopt Compstat-like programs, suggesting in our view the more general
level of innovation found in agencies in these parts of the country.

The specific motivations for adopting Compstat vary across police agencies, but we found that
the model of Compstat that has been touted in New York City has strongly influenced its
adoption elsewhere in the country. Moreover, our study shows that the adoption of Compstat is
strongly related to a department’s expressed desire to reduce serious crime and increase
management control over field operations. These goals for policing are much more prominent in
agencies that have adopted Compstat than those that have not. At the same time, we found that
agencies that had adopted Compstat programs are much less likely to focus on improving skills
and morale of street-level officers. This suggests that Compstat may represent not only a new
movement in police efforts to develop effective crime-control strategies, but also a departure
from the priorities of “bubble-up” community-policing programs that rely on initiative from
street-level officers.

The Growth of Compstat in American Policing Police Foundation Reports
Page 16 April 2004

REFERENCES

Beer, Michael 1980. Organizational Change and Development: A Systems View. Santa Monica,

CA: Goodyear Publishing Company.

Bratton, William J. 1999. Great Expectations: How Higher Expectations for Police Departments

Can Lead to a Decrease in Crime. In Measuring What Matters: Proceedings from the
Policing Research Institute Meetings, ed. Robert H. Langworthy, 11-26. Washington,
DC: National Institute of Justice.

Bratton, William J. 1998. Turnaround: How America’s Top Cop Reversed the Crime Epidemic.

New York: Random House.

Bratton, William J. 1996. Cutting Crime and Restoring Order: What America Can Learn from

New York’s Finest. Heritage Foundation Policy and Research Analysis, lecture no. 573,
http://www.heritage.org/Research/Crime/HL573.cfm.

Grubler, Arnulf. 1991. Diffusion and Long-Term Patterns and Discontinuities. Technological
Forecasting and Social Change 39:159-180.

Gurwitt, Rob. 1998. The Comeback of the Cops. Governing (January): 14-19.

Kelling, George L., and William J. Bratton. 1998. Declining Crime Rates: Insiders’ Views of the

New York City Story. Journal of Criminal Law and Criminology 88: 1217-1232.

Kelling, George L., and Catherine M. Coles. 1996. Fixing Broken Windows: Restoring Order

and Reducing Crime in Our Communities. New York: Free Press.

Law Enforcement News. 1997. NYC’s Compstat Continues to Win Admirers. October 13.

Maas, Peter. 1998. What We’re Learning from New York City. Parade, May 10.

Maple, Jack. 1999. The Crime Fighter: Putting the Bad Guys Out of Business. New York:

Doubleday.

McDonald, Phyllis Parshall, Sheldon Greenberg, and William J. Bratton. 2001. Managing Police
Operations: Implementing the NYPD Crime Control Model Using COMPSTAT.
Belmont, CA: Wadsworth Publishing Co.

McDonald, Phyllis Parshall. 1998. The New York City crime control model: A guide to

implementation. Unpublished manuscript. Washington, DC.

The Growth of Compstat in American Policing Police Foundation Reports
Page 17 April 2004

Merton, Robert K. 1940. Bureaucratic Structure and Personality. Social Forces 18:560-568.

Moore, Mark H. 1995. Creating Public Value: Strategic Management in Government.
Cambridge, MA: Harvard University Press.

Office of Management Analysis and Planning. n.d. The Compstat Process. New York: New

York City Police Department.

Remnick, David. 1997. The Crime Buster. The New Yorker, February 24 and March 3.

Rogers, Everett M. 1995. Diffusion of Innovations. New York: Free Press.

Sayre, Wallace Stanley, and Herbert Kaufman. 1960. Governing New York City: Politics in the
Metropolis. New York: Russell Sage Foundation.

Silverman, Eli B. 1999. NYPD Battles Crime: Innovative Strategies in Policing. Boston:

Northeastern University Press.

Silverman, Eli B. 1996. Mapping Change: How the New York City Police Department Re-

engineered Itself to Drive Down Crime. Law Enforcement News, December.

Sparrow, Malcolm K., Mark H. Moore, and David B. Kennedy. 1990. Beyond 911: A New Era

for Policing. New York: Basic Books.

U.S. Department of Justice, Bureau of Justice Statistics. 1998. Census of State and Local Law

Enforcement Agencies,1996, http://www.ojp.usdoj.gov/bjs/abstract/csllea96.htm.

Weisburd, David, Stephen Mastrofski, Ann Marie McNally and Rosann Greenspan. 2001.

Compstat and Organizational Change: Findings from a National Survey. Report
submitted to the National Institute of Justice by the Police Foundation.

Weisburd, David, Stephen Mastrofski, Ann Marie McNally, Rosann Greenspan, and James

Willis. 2003. Reforming to Preserve: Compstat and Strategic Problem Solving in
American Policing. Criminology and Public Policy 2 (3): 421-456.

About the Police Foundation

The Police Foundation is a national, nonpartisan, nonprofit organization dedicated to supporting innovation
and improvement in policing. Established in 1970, the foundation has conducted seminal research in police
behavior, policy, and procedure and works to transfer to local agencies the best information about practices
for dealing effectively with a range of important police operational and administrative concerns.

Our purpose is to help the police be more effective in doing their job, whether it is deterring robberies,
intervening in potentially injurious domestic disputes, or working to improve relationships between the
police and the communities they serve. To accomplish our mission, we work closely with police officers
and police departments across the country, and it is in their hard work and contributions that our
accomplishments are rooted.

The foundation helps police departments to acquire both the knowledge gained through research and the
tools needed to integrate that knowledge into police practices. Working with law enforcement agencies
seeking to improve performance, service delivery, accountability, and community satisfaction with police
services, the foundation offers a wide range of services and expertise. The Crime Mapping & Problem
Analysis Laboratory operates with the goals of providing practical assistance and information to police
departments and to developing the physical and theoretical infrastructure necessary for further innovations
in police and criminological theory.

The foundation has done much of the research that led to a questioning of the traditional model of
professional law enforcement and toward a new view of policing—one emphasizing a community
orientation. For example, research on foot patrol and on fear of crime demonstrated the importance to crime
control efforts of frequent police-citizen contacts made in a positive, non-threatening way. As a partner in
the Community Policing Consortium, the foundation, along with four other leading national law
enforcement organizations, plays a principal role in the development of community policing research,
training, and technical assistance.

Sometimes foundation research findings have challenged police traditions and beliefs. When police
agencies employed routine preventive patrol as a principal anti-crime strategy, a foundation experiment in
Kansas City showed that routine patrol in marked patrol cars did not significantly affect crime rates. When
police officials expressed reservations about using women on patrol, foundation research in Washington,
DC, showed that gender was not a barrier to performing patrol work. To address the intense debate about
how police should respond to incidents of domestic violence, the foundation conducted the Minneapolis
Domestic Violence Experiment—the first scientifically controlled test of the effects of arrest for any crime.
Foundation research on the use of deadly force was cited at length in a landmark 1985 U.S. Supreme Court
decision, Tennessee v. Garner. The court ruled that the police may use deadly force only against persons
whose actions constitute a threat to life.

Motivating all of the foundation’s efforts is the goal of efficient, effective, humane policing that operates
within the framework of democratic principles and the highest ideals of the nation.

1201 Connecticut Avenue, NW
Washington, DC 20036-2636
(202) 833-1460
(202) 659-9149 fax
E-mail: pfinfo@policefoundation.org
www.policefoundation.org

COMPSTAT IN PRACTICE:
AN IN-DEPTH ANALYSIS
OF THREE CITIES

James J. Willis
Stephen D. Mastrofski

David Weisburd

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

ii

The Police Foundation is a national, nonpartisan, nonprofit organization dedicated to supporting
innovation and improvement in policing. Established in 1970, the foundation has conducted seminal
research in police behavior, policy, and procedure, and works to transfer to local agencies the best
information about practices for dealing effectively with a range of important police operational and
administrative concerns. Motivating all of the foundation’s efforts is the goal of efficient, humane policing
that operates within the framework of democratic principles and the highest ideals of the nation. The
Police Foundation’s research findings are published as an information service.

This project was supported by Grant Number 98-IJ-CX-007 by the National Institute of Justice, Office of
Justice Programs, U.S. Department of Justice. Points of view or opinions expressed in this document are
those of the authors and do not necessarily represent the official position or policies of the U.S.
Department of Justice or the Police Foundation.

Additional support for the writing of this report was provided through a Faculty Scholarship Support Grant
from the University of Massachusetts, Boston.

Additional reports from the larger project from which this report is derived include Compstat and
Organizational Change in the Lowell Police Department: Challenges and Opportunities and The Growth
of Compstat in American Policing. These reports are available online at http://www.policefoundation.org,
Publications/Electronic Library.

©2003 by the Police Foundation. All rights, including translation into other languages, reserved under the
Universal Copyright Convention, the Berne Convention for the Protection of Literary and Artistic Works
and the International and Pan American Copyright Conventions. Permission to quote readily granted.

ISBN 1-884614-20-5

Police Foundation
1201 Connecticut Avenue, NW
Washington, DC 20036-2636
(202) 833-1460
Fax: (202) 659-9149
Email: pfinfo@policefoundation.org
www.policefoundation.org

http://www.policefoundation.org

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

iii

Table of Contents

List of Tables and Figures……………………………………………………………………………………………….. iv
Foreword ……………………………………………………………………………………………………………………….. v
Acknowledgments…………………………………………………………………………………………………………. vii
I. Introduction…………………………………………………………………………………………………………. 1
II. Background on Compstat and Its Core Elements ……………………………………………………… 2
III. Research Sites ……………………………………………………………………………………………………… 5
IV. Data and Methods ………………………………………………………………………………………………… 7
V. Analysis of Departments’ Experiences with Compstat ……………………………………………… 9

Mission Clarification ………………………………………………………………………………………….. 10
Internal Accountability ……………………………………………………………………………………….. 20
Geographic Organization of Operational Command ……………………………………………….. 29

Organizational structure……………………………………………………………………………. 31
Decision-making authority………………………………………………………………………… 31
Coordination issues………………………………………………………………………………….. 34
Geographic versus temporal organization …………………………………………………… 36

Organizational Flexibility ……………………………………………………………………………………. 37
Rivalry among districts …………………………………………………………………………….. 40
Manpower ………………………………………………………………………………………………. 43
Traditional approaches to resource allocation ……………………………………………… 43
City politics…………………………………………………………………………………………….. 46
Strategies to increase flexibility…………………………………………………………………. 47

Data-Driven Problem Identification and Assessment………………………………………………. 48
Data collection ………………………………………………………………………………………… 48
Data analysis and presentation…………………………………………………………………… 50
Use of data ……………………………………………………………………………………………… 54

Innovative, Problem-Solving Tactics…………………………………………………………………….. 58
Problem solving and brainstorming during Compstat meetings……………………… 59
Follow-up……………………………………………………………………………………………….. 61
Traditional responses ……………………………………………………………………………….. 63
Innovative responses………………………………………………………………………………… 64

VI. Discussion…………………………………………………………………………………………………………. 65
Mission Clarification ………………………………………………………………………………………….. 68
Internal Accountability ……………………………………………………………………………………….. 68
Geographic Organization of Operational Command ……………………………………………….. 69
Organizational Flexibility ……………………………………………………………………………………. 70
Data-Driven Problem Identification and Assessment………………………………………………. 71
Innovative, Problem-Solving Tactics…………………………………………………………………….. 72

VII. Conclusion ………………………………………………………………………………………………………… 73
Appendix: Line Officer Survey ………………………………………………………………………………………. 79
References……………………………………………………………………………………………………………………. 86
Authors………………………………………………………………………………………………………………………… 92

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

iv

List of Tables and Figures

Table 1: Crime index in Lowell, Minneapolis, and Newark (1996-2001)……………………………….. 6

Figure 1: Crime rates in three cities…………………………………………………………………………………… 7

Table 2: The importance of reducing violent crime to the department’s Compstat strategy…….. 15

Table 3: The importance of holding district commanders accountable for crimes in their districts

to the department’s Compstat strategy………………………………………………………………………. 26

Table 4: The importance of holding officers accountable for crimes in their beats to the

department’s Compstat strategy……………………………………………………………………………….. 27

Table 5: In your view, Compstat has made supervisors place too much emphasis on statistics .. 41

Table 6: In your view, Compstat has increased teamwork between your unit and specialized units

in the department……………………………………………………………………………………………………. 42

Table 7: Summary comparison of Compstat elements across sites ………………………………………. 66

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation v

FOREWORD

Technology shapes human culture and is influenced in turn by human factors. This

principle guides the history of invention and clearly bears on Compstat, the latest trend in crime
fighting. Compstat shows promise of helping departments reduce crime through systematic data
collection, crime analysis, and heightened accountability. Compstat’s potential as a tool of police
reform nonetheless depends on the extent to which it changes officers’ routine activities, the
effect it has on management practices, and its compatibility with a department’s existing
objectives. In other words, the impact of Compsat, like that of previous technology, is contingent
on how it interacts with the existing technical capacities, work practices, management styles, and
cultural values of police departments. Its future value also hinges on its compatibility with
community policing, a deeply entrenched movement in law enforcement and one that stresses the
human face of policing.

The influence of these various factors on Compstat’s implementation in three different

police departments is the subject of this study. As we shall see, the theory and practice of
Compstat diverged as the departments tried to incorporate Compstat within their existing
structures and practices. The experiences of Lowell, Massachusetts; Minneapolis, Minnesota;
and Newark, New Jersey, shed some light on the potential and limits of technology as a tool to
enhance the effectiveness of police.

Compstat’s many advocates assume the existence of several conditions that will allow

Compstat to transform inertial and conservative police departments into efficient crime-fighting
machines. There will be motivated employees who are stimulated by a rigorous system of
accountability. There will be enough organizational flexibility for commanders to reallocate
personnel and resources to areas that need them most. Decisions will reflect empirical knowledge
of emerging problems and draw heavily on sophisticated, electronic data-analysis systems.
Members of the department will engage in free exchange of ideas, and there will be an
atmosphere that nurtures innovation. The implementation of Compstat presumes, in short, that
the organizational culture will tolerate risk and encourage new approaches to persistent crime
problems.

While departments naturally want to make communities safer by reducing crime, they

cannot always provide a fertile ground for Compstat’s demanding agenda. Compstat’s sole
mission is to fight crime, but police departments have complex missions that may require police
to engage in many diverse activities. Scarcity of resources and political pressures may limit
departments’ freedom to allocate resources to specific crime problems. Police managers may
lack the training to take full advantage of Compstat data when they are making decisions.
Collaboration and free-flowing debate may be at odds with a traditional police culture that has
long socialized officers to follow orders and defer to rank. Lastly, police may be reluctant to
follow innovative approaches when the wrong decision may lead to waste of precious resources
and loss of even more precious human lives.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation vi

This dissonance between the ideals of Compstat and the realities of police work suggests
that Compstat’s architects may have failed to envision the program’s role in the context of a
department’s many duties. The discord is a pressing concern, however, for agencies that are
trying to adapt their organizational structures and cultural assumptions to the demands of the new
technology. When there is incompatibility, an uneasy tension may arise between the system’s
requirements and the way that officers define their jobs.

Much of this tension has played itself out over the issues of accountability and the patrol

officer’s role. Thus far, patrol officers have played little part in the Compstat process, and some
of them have even seen it as an obstacle to effective policing. This is because Compstat places a
heavy burden of accountability on supervisors, who have sometimes responded to the pressure
by severely limiting the discretion of their officers. One result of this development is to widen
the traditional divide between managers, who are privy to the wisdom of Compstat, and street
cops, who are largely excluded. Another is to foster resentment of the program among many
officers who believe their departments have sacrificed important aspects of police work, as well
as valuable resources, on the altar of technology and crime fighting.

These issues have converged in the interaction between community policing and

Compstat, as they set forth competing ideals of law enforcement. Compstat, as we have seen,
gives command staff most of the responsibility for reducing crime, largely excludes patrol
officers from the decision-making process, and has a relentless focus on crime reduction.
Community policing, on the other hand, devolves decision making to the level of the street,
envisions police and citizens as partners in problem solving, and makes officers responsible for
responding to a broad spectrum of community concerns. At stake here are two vastly different
conceptions of the officer’s role in modern society. In one, they simply follow their superiors’
orders, while the other grants them considerable discretion as the guardians of their individual
beat. Given the gap between the two models of policing, Compstat naturally tends to encounter
the greatest resistance in departments that are most committed to community policing.

Resolving the conflict between Compstat and community policing, based as they are on

opposing cultural assumptions, is a significant challenge facing police leaders in the twenty-first
century. It may require them to educate their officers in the basic principles of data analysis,
include the rank and file in the Compstat process, and find better ways to use resources so they
can meet the needs of the two competing programs. It is important, however, that police leaders
rise to the challenge because their success may further the objectives of both Compstat and
community policing. It is likely that officers who accept Compstat would perform better in the
service of Compstat’s crime-fighting mandate. It is also probable that officers with an analytical
understanding of crime problems and a handle on crime data would collaborate more effectively
with community stakeholders in solving problem. By reconciling these two valuable movements
in law enforcement, police leaders would ultimately fulfill a broader mandate that transcends the
individual goals of both Compstat and community policing. This fundamental mission is to serve
the members of the public who are certainly concerned about safe streets but also want a positive
and constructive relationship with the officers who patrol their neighborhoods.

Hubert Williams
President

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation vii

ACKNOWLEDGMENTS

We are deeply indebted to Superintendent Edward Davis, Chief Joseph Santiago,
and Chief Robert Olson for granting us access to their police departments and for their
support of this project. We would also like to thank the many sworn officers and civilians
who shared their experiences of Compstat. A note of appreciation also goes to Anne
Marie McNally, a former research associate at the Police Foundation and Tania
Belchinger, a former graduate assistant in the Sociology Department at the University of
Massachusetts-Boston. Their help with analyzing the survey and field work data
contributed to the overall quality of this report. Finally we wish to express thanks to
Professor John Eck who commented on an earlier draft.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

1

I. INTRODUCTION

Police organization and practice have changed throughout the twentieth century in
response to fundamental social and technological changes in U.S. society (Reiss 1992). Three
such powerful forces coincided nearly a century ago: the movement for police professionalism,
the introduction of the automobile, and the advent of modern communications technology
(Walker and Katz 2002). As part of a much larger attempt to reduce corruption and inefficiency
in municipal government, progressives and police leaders sought to transform the organization,
administration, and delivery of police services (Fogelson 1977). Modern management principles
called for centralization of command and control to improve police accountability, and
technological innovations promised a more scientific and effective approach to crime reduction.

In an attempt to distance police from the corrupt influence of neighborhood politics and
to increase executive oversight, supporters of police reform closed local precincts and shifted
decision-making power to central headquarters. The “span of administrative control” widened
further with the invention of the telephone and two-way radio (Reiss 1992, 52). Through a
network of constant communication, police administrators became more involved in the daily
activities of the rank and file. These technological innovations, along with the introduction of
the patrol car and the collection of crime statistics, also increased the capacity of the police to
identify and respond quickly to crime problems (Manning 1992). One of the most influential,
early twentieth-century police reformers, August Vollmer, praised these innovations as
“remarkable.” The “modern, scientifically oriented organization,” he wrote, reduced crime and
protected the community “with principles of military science, including those of strategy, tactics,
logistics, and communication” (Vollmer 1933, 161, 165).

Increasing crime and social disorder in the 1960s and early 1970s brought attention to the
limitations and unintended consequences of these reforms (Kelling and Moore 1988).
Centralized command and control had joined the emphasis on rapid response to crime to detach
officers from the communities they were supposed to serve and distill their diverse
responsibilities to the simple mandate of fighting crime. Community policing, the contemporary
form of police progressivism, is an ongoing attempt to remedy these problems (Sparrow et al.
1990). Informed by private sector management literature, supporters of community policing
advocate area decentralization of police services, devolution of decision making, and use of
technology to enhance the delivery of police services (Moore and Stephens 1991). They argue
that devolving power to district commanders and line officers makes police organizations more
responsive to a wide range of community concerns. They also claim that adopting sophisticated
computer technology increases organizations’ capacity to diagnose a specific crime problem and
then select the most appropriate response, whether it is problem solving, preventing crime, or
reducing social disorder.

At the turn of the twenty-first century, a new engine of police progressivism may have
arisen. Characterized as a new crime-control program, Compstat combines all of the major
prescriptions offered by contemporary organizational development experts (Beer 1980) with the
latest geographic information systems technology. It also re-engineers police management by
holding command staff directly accountable for crime levels in their beats. Furthermore, it uses
sophisticated computer maps and crime statistics to facilitate timely and targeted responses to

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

2

crime problems. In this report we provide an in-depth assessment of how Compstat worked in
three police departments—Lowell, Massachusetts (LPD), Minneapolis, Minnesota (MPD), and
Newark, New Jersey (NPD). We cannot be confident that our findings on Compstat would be the
same for a different set of departments. By extending our research beyond a single site, however,
we hope to capture a range of experiences that broaden and deepen discussion of this latest
policing innovation.

More specifically, this report serves three purposes: (1) to describe how Compstat

functioned as a specific program; (2) to examine how it changed police organization and
practice; and (3) to provide some insights into the direction Compstat is leading policing in the
United States.1 Like past attempts at reform, Compstat presents police organizations with a set of
specific challenges and opportunities. Some of the challenges arise from implementation
problems that departments can correct, but others appear because Compstat calls for the pursuit
of conflicting goals. In addition, some of the central tenets of this latest reform movement
significantly affect the existing philosophies, programs, and structures of community policing. In
highlighting some of Compstat’s challenges, complexities, and positive contributions, we hope to
illuminate its possibilities and limitations, as well as areas that police leaders might address to
achieve departmental objectives.

We note at the outset that the purpose of this report is not to render judgment on

Compstat’s effectiveness at controlling crime. Many advocates have claimed great success for it
(Bratton 1998; Silverman 1999; Kelling and Sousa 2001), while some critics have remained
more skeptical (Bouza 1997; Eck and Maguire 2000). An assessment of Compstat’s crime-
control effectiveness is sorely needed, but we should learn what Compstat is before attempting to
measure its effects. Hence, our focus is on providing a thorough, empirical evaluation of how
Compstat is changing the structure, management, and practices of police organizations.

II. BACKGROUND ON COMPSTAT AND ITS CORE ELEMENTS

Since 1994, when former NYPD police commissioner William Bratton first implemented
Compstat, the program has received hortatory statements similar to those Vollmer reserved for
police advances at the beginning of the last century. Winner of a 1996 “Innovations in American
Government” award, sponsored by the Ford Foundation and Harvard’s Kennedy School of
Government, Compstat has been described in grand terms. Recently, Professor William Walsh
referred to Compstat as an “emerging police managerial paradigm” (Walsh 2001), and Professors
George Kelling and William Sousa have described it as “perhaps the single most important
organizational/administrative innovation in policing during the latter half of the twentieth
century” (Kelling and Sousa 2001, 2). Nor has this praise been limited to scholars. Compstat has

1 There is some disagreement about what the acronym “Compstat” actually means. Former NYPD police
commissioner William Bratton suggests that it stands for “computer-statistics meetings” (Bratton 1998, 233), but
Silverman attributes the term to “Compare Stats”—a computer filename (Silverman 1999, 98). Some commentators
have collapsed these meanings and argue that Compstat refers to “computer comparison statistics” (U.S. National
Agricultural Library 1998, http://www.nalusda.gov/pavnet/iag/cecompst.htm). The Lowell Police Department
adopted the nomenclature of the NYPD, but a different name was used at the other two sites we visited.
Minneapolis’ program was called “CODEFOR” (Computer Optimized Deployment – Focus on Results) and
Newark’s “Comstat” stood for “Command Status Report.” For the sake of convenience, we use the generic term
“Compstat” to refer to these programs at the three departments.

http://www.nalusda.gov/pavnet/iag/cecompst.htm

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

3

received an extraordinary amount of attention from the press, think tanks, police professional
associations, and the leadership of some of the largest police departments in the nation. Its
sweeping popularity has been fueled by much national publicity that attributes the recent
plummet in New York’s crime rate to Compstat (Silverman 1996). The “Comstat Craze,” as one
public interest journal described it, has contributed to the program’s rapid diffusion across the
United States (Swope 1999). Our 1999 national survey reported that in the brief period since its
inception in 1994, a third of large departments—those with one hundred or more sworn
officers—had implemented a “Compstat-like program,” and 26 percent were planning to do so.2
Should the rate of adoption continue at that extraordinary pace, 90 percent of large police
departments will have adopted a Compstat program just a mere ten years after its implementation
in the NYPD. This is five years less than the fifteen-year period Arnulf Grubler predicts it takes
the most quickly diffused technologies to progress to a 90 percent saturation level (Weisburd et
al. 2001; Grubler 1991).

In the late 1980s, some scholars cautioned that the “rhetoric” surrounding community

policing, a program still in its formative stages, concealed “complex realities” that needed to be
addressed (Mastrofski 1988, 48). Similarly, there is a need for a more comprehensive and fine-
grained analysis of Compstat, a program that is also in its infancy: How does Compstat work? Is
it difficult to implement? Is it a coherent program? Is it compatible with existing department
programs and practices?

A review of the research literature suggests that the glowing accounts of Compstat’s

success are fueled mostly by studies that rely on anecdotal evidence or concentrate on the
NYPD, the nation’s largest and, by any measure, most exceptional police department.3 To date,
there has been little systematic analysis of Compstat in U.S. police departments of different size
and organization. Very little is actually known about how Compstat operates. Perhaps many feel
that the relationship between Compstat’s crime-fighting mandate and the country’s recent crime
drop is so compelling that there is little need to justify Compstat. One of the dangers of such a
glib approach is that it shrouds Compstat’s complexities and potential problems. Failure to
consider these problems carefully, as some scholars forewarned about community policing in the
late 1980s, may relegate even the best-intentioned reform attempts to the long list of failed
innovations that litter the history of U.S. policing (Greene and Mastrofski 1988).

This report, which assesses how Compstat works in three very different police

departments, is the third stage of a project, entitled Compstat and Organizational Change: A
National Assessment, funded by the National Institute of Justice and conducted by the Police
Foundation (Weisburd et al. 2001). We examined the existing literature on Compstat from those

2 Forty-two percent had not implemented a Compstat-like program. For the first detailed analysis on Compstat’s
implementation and operation in police departments across the country see Compstat and Organizational Change:
Findings from a National Survey, Weisburd et al., Police Foundation (2001).

3James L Heskett, “NYPD New,” Harvard Business School Report no. N9-396-293 (April 1996); Eli Silverman,
NYPD Battles Crime, Northeastern University Press (1999); Phyllis McDonald, Sheldon Greenberg, and William J.
Bratton, Managing Police Operations: Implementing the NYPD Crime Control Model Using COMPSTAT,
Wadsworth Publishing Co. (2001); Vincent E. Henry, The COMPSTAT Paradigm: Management Accountability in
Policing, Business, and the Public Sector, foreword by William J. Bratton, Looseleaf Law Publications (2002).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

4

who built it, who advocate it, and who claim to have made objective observations about it. By
couching this information in a larger literature on the structure and processes of organizations,
we derived the following six core elements:

Mission clarification: Top management is responsible for clarifying and exalting the core

features of the department’s mission that serve as the overarching reason for the
organization’s existence. Mission clarification includes a demonstration of management’s
commitment and states its goals in specific terms for which the organization and its
leaders can be held accountable—such as reducing crime by 10 percent in a year (Bratton
1998, 252).

Internal accountability: Operational commanders are held accountable for knowing their

commands, being well acquainted with the problems in the command, and accomplishing
measurable results in reducing those problems—or at least demonstrating a diligent effort
to learn from that experience.

Geographic organization of operational command: Operational command is focused on

the policing of territories, so central decision-making authority over police operations is
delegated to commanders with territorial responsibility for districts. Functionally
differentiated units and specialists—patrol, community-policing officers, detectives,
narcotics, vice, juvenile, and traffic—are either placed under the command of the district
commander, or arrangements are made to facilitate their responsiveness to the
commander’s needs.

Organizational flexibility: The organization develops the capacity and habit of changing

established routines as needed to mobilize resources when and where they are needed for
strategic application.

Data-driven analysis of problems and assessment of department’s problem-solving

efforts: Data are made available to identify and analyze problems and to track and assess
the department’s response. Data are made accessible to all relevant personnel on a timely
basis and in a readily usable format.

Innovative problem-solving tactics: Police responses are selected because they offer the

best prospects of success, not because they are “what we have always done.” Innovation
and experimentation are encouraged and use of the best available knowledge about
practices is expected.

Using these elements as a general framework, we analyzed how Compstat was being
implemented across the United States.

The project’s initial stage consisted of a national survey assessing how many local police
departments were using Compstat and the extent to which they had implemented its six core
elements. By capturing the resources used, the police units served, the geographic scope, and the
temporal frequency of various structures and activities, we were able to assay the dosage and the
character of the treatment for each element. For the second stage, we used the results from our

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

5

national survey to identify sixteen sites suitable for short, intensive site visits of two to three
days. Some of these departments were actively engaged in Compstat, while others indicated that
they used some of its core elements but had not implemented a Compstat program. Through
observations and interviews, we were able to describe each department’s broad range of
experiences with Compstat, to identify patterns of organizational change and program content,
and to distinguish departments’ individual successes, failures, surprises, and disappointments
with Compstat. Finally, we selected three police agencies, Lowell, Minneapolis, and Newark, as
sites suitable for extended site visits. Between October 2000 and June 2001, a full- or part-time
researcher was sent to each of these departments for a period of two to eight months to gather
detailed information on how Compstat worked at all levels of the police organization.4 These
three sites serve as the core of the analysis discussed in this report.

III. RESEARCH SITES

We selected these three sites because they had implemented Compstat; they differed in
size, organization, and crime environment; and they were receptive to having a field researcher
on site for an extended period. According to population figures in the 2001 Uniform Crime
Reports, or UCR, Lowell’s contingent of 260 sworn officers served a population of 105,668.
Minneapolis’ 919 police officers served a population of 386,726; and the 1,445 police officers in
Newark served a population of 275,823. In the 1980s and 1990s, all three police departments had
undergone significant organizational change upon implementing community-oriented policing.
In order to push decision making down the chain of command, to provide greater access to police
services, and to improve organizational flexibility, they had decentralized into districts or
precincts. Furthermore, they had fostered closer relationships with city residents and increased
neighborhood input into crime problems through a variety of programs and tactics, such as
assigning patrol officers to permanent beats, holding regular community meetings, increasing
foot patrol, and implementing community outreach. In addition, all three departments had
community response teams, or units whose primary responsibilities included responding to
specific community concerns.

Levels and types of crime differed across sites, but each city had recently experienced a

general decline in its crime rate. Between 1995 and 2001, index crimes reported in the UCR by
the FBI showed that serious crime was lowest in Lowell and highest in Minneapolis (see Table
1). In addition to differences in crime level, types of crime varied across sites (see Table 1). The
violent crime rate was highest in Newark while the property crime rate was highest in
Minneapolis. In all three sites, Compstat coincided with a general crime decrease, as it had in
New York. Following Compstat’s implementation in Newark during 1997 and in Minneapolis
during 1998, both cities experienced a general reduction in crime. Crime in Lowell, which also
implemented Compstat in 1997, followed a similar downward trend, but in 2000 and 2001 it
began to increase (see Table 1). Whether declines in the crime rate can be attributed to Compstat
awaits a more rigorous evaluation. However, we would note that the UCR trends at all three sites
are ambiguous at best. Crime was already in decline for at least a year before each city
implemented Compstat, and the rate of decline in subsequent years was the same or less steep

4 Terminology varied among the sites. To minimize confusion, we refer to each department’s chief executive officer
as “chief,” to mid-level managers as “middle managers” or “district commanders,” and to all geographic areas of
command as “districts,” unless otherwise stated.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

6

(see Figure 1). If Compstat had an impact, we would have expected the rate of decline to be
steeper after its implementation. This raises the question of whether Compstat was the force
bringing crime down, since the decline had already begun at all three sites and did not accelerate
subsequently. A few years of crime data at three sites is simply insufficient to determine the
nature and extent of Compstat’s effects on crime. We have therefore avoided making any
statements about it one way or the other.

Table 1: Crime index in Lowell, Minneapolis, and Newark (1996-2001)

Violent, property, and total index crimes per 100,000 population with
percentage change from previous year (+/-)

Year

Lowell Minneapolis Newark
property

crime
violent
crime

property
crime

violent
crime

property
crime

violent
crime

4906 1582 9567 1978 11527 3985 1995
6488 11545 15512

3826 1051 9408 1883 9803 3345 1996
4877 (-25%) 11291 (-2%) 13148 (-14%)

3409 1071 9589 1850 7993 2735 1997
4480 (-8%) 11439 (+1%) 10728 (-18%)

2911 1021 8035 1525 6466 2094 1998
3932 (-12%) 9560 (-16%) 8560 (-20%)

2480 777 7246 1389 6067 1814 1999
3257 (-21%) 8635 (-10%) 7881 (-8%)

3138 709 6341 1210 5728 1505 2000
3847 (+18%) 7551 (-12.5%) 7233 (-8%)

3464 804 5875 1060 5406 1391 2001
4268 (+11%) 6935 (-8%) 6797 (-6%)

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

7

Figure 1: Crime Rates in the three cities

These departments were similar in ways relevant to Compstat’s impact on organizational
change; differences were matters of degree. All three departments had strong professional
reputations, had received considerable publicity for being innovative, and had been running a
Compstat program for at least two years. We felt that an examination of departments that had
advanced furthest in scope and experience with their implementation of Compstat and were
widely recognized for being on the cutting-edge of policing would provide useful models from
which other police agencies might learn. Furthermore, information from our national survey and
initial site visits revealed significant variations in Compstat programs at these departments. The
latter reason was particularly compelling, since we felt an examination of the operation of
different Compstat programs would provide valuable insights into the various ways that
organizations adapt to new structures and technologies. Finally, a comparison of the three
departments, we believed, would help capture a range of opportunities and challenges that police
departments face when they attempt to change their routine tasks and activities to incorporate a
program whose supporters describe it as a major change in the management and crime-fighting
responsibilities of police.

IV. DATA AND METHODS

In conducting our intensive field research, we were especially interested in obtaining
detailed information on the role of Compstat-generated data and in learning about the “scanning”
part of the problem-identification process. How were problems analyzed, and how were tactics
reviewed and selected? What was the scope of the treatment developed to deal with a Compstat-
identified problem, in terms of the resources mobilized to deal with it? Was there a
reorganization or mobilization of resources to address the problem, or was it handled merely by
altering the routines of individuals or units already assigned to these duties? How were rank-and-
file officers involved in this process, if at all? Were specific officers or patrol units made
accountable for addressing problems? How much follow-through was there on these initiatives?

0

2000

4000

6000

8000

10000

12000

14000

1996 1997 1998 1999 2000 2001

In
de

x
cr

im
es

/1
00

,0
00

Newark
Minneapolis
Lowell

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

8

How were assessments of an intervention conducted? To what extent were district routines, such
as responding to calls for service, disrupted by efforts to deal with problems identified by the
Compstat process?

The main field research techniques we used were participation, observation, and formal

interviews with city officials and police at various levels in the chain of command. At each site,
researchers observed weekly or biweekly Compstat meetings and interviewed city and police
department personnel. These included the mayor, city manager, chief, civilian staff, middle
managers or district commanders, captains, lieutenants, detectives, first-line supervisors or
sergeants, and patrol officers. We attended thirty-six department-level Compstat meetings of two
to three hours in length and eight shorter district-level or pre-Compstat meetings, and we
conducted a total of seventy interviews. By guaranteeing interviewees anonymity, whenever
possible, and confidentiality—by refusing to act as a conduit for information within the
department, even when tested—we tried to gain the trust of department members. Despite some
initial suspicion, most of those interviewed felt sufficiently comfortable to engage in lengthy and
candid discussions about Compstat. The interviews ranged in length from twenty minutes to two
hours, with many exceeding the allotted time. We supplemented them with numerous informal
conversations with personnel of all ranks. In addition, we attended department-level command
staff, internal affairs, and criminal investigations meetings to help us situate Compstat within a
larger context.

Immediately following the department-level Compstat meetings, we attempted to debrief

the executive who ran the meeting; the presenting district commanders were also debriefed soon
after, usually within the next couple of days. The aim of our short, fifteen-to twenty-minute,
post-Compstat debriefings was to help us identify the main crime problems in each sector or
district and to track responses to these problems over time.

Researchers were instructed to type up their field notes the same day that they were

taken. These extensive narratives were then disseminated to another member of the research
team who was responsible for providing feedback on the quality and quantity of the on-site
observations and for identifying issues that warranted further exploration. We also collected any
documents that could further our understanding of Compstat, including internal department
memos, research grants, community handouts, newspaper and Web articles, and the Compstat
maps, spreadsheets, and crime analyses that were provided to district commanders.

Finally, if Compstat has the profound effects on police officers suggested by its

proponents, these changes should register with the rank and file. In order to assess the impact of
Compstat on those at the bottom of the police organization, we distributed an anonymous,
voluntary, and self-administered survey to patrol officers who regularly attended roll call on the
day, early night, and late shifts. Due to constraints of time and personnel, the survey allowed us
to obtain far more information from patrol officers than we could have gathered through
interviews. The on-site researcher in each department distributed the ten- to fifteen-minute
survey on separate days across all shifts to ensure a representative sample of patrol officers who
attended roll call. When distributing the survey, he or she informed officers that the Police
Foundation was conducting the study to learn how Compstat worked in their department and
valued any feedback they were willing to provide. Officers were asked to describe the extent of

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

9

their involvement in Compstat-generated activities, their motivation to participate in them, and
their views on these activities. The survey was distributed to a total of 450 patrol officers and we
used both simple descriptive and bivariate approaches in our analysis of their responses. 5

We promised all respondents that we would do our utmost to conceal their identity.

Obviously, when it came to the chief, city manager, or mayor, it was impossible to guard against
identification, and we made this clear. Protecting the confidentiality of those who occupied the
few specialized and mid/upper-level management positions in each police organization,
particularly the district commanders, was challenging. Consequently, we decided to omit any
identifying characteristics, such as the number of years a respondent had been on the force or
national origin, from this report. In addition, we occasionally changed a respondent’s gender, and
deleted text that could possibly breach his or her confidentiality.

V. ANALYSIS OF DEPARTMENTS’ EXPERIENCES WITH COMPSTAT

Using the six key components we identified as Compstat’s general framework, we

compared the different Compstat programs in Lowell, Minneapolis, and Newark with each other
and with data from our national survey. The latter approach helped us assess how typical each
program was of programs in other departments across the country. We also used our qualitative
data to assess how each department had adapted to Compstat, how each element worked, and
how Compstat operated in relation to some of the central tenets of the current reform movement
in U.S. policing—community policing. Before examining each of Compstat’s key elements in
detail, here is a brief summary of our findings:

Mission clarification: A mission statement establishing a specific crime-reduction goal,

reinforced through other mechanisms, and linked to specific plans of action designed for
its success, may fail to encourage widespread commitment to management’s objectives.
Additionally, the focus on crime fighting may conflict with other important goals central
to community policing, thus causing disaffection among the rank and file.

Internal accountability: Compstat meetings are a very effective mechanism for making

district commanders feel responsible for their beats. However, in the absence of a similar
structure, accountability is experienced far less intensely down the chain of command.
Consequently, Compstat concentrates decision making at the top of the organizational
hierarchy, while the primary responsibility of the rank and file is to follow orders. This
contradicts a central tenet of community- and problem-oriented policing—decentralized
decision making to the rank and file to initiate problem solving.

Geographical organization of operational command: Decision-making authority is

devolved down one or two levels in the organization to district commanders and/or sector
lieutenants. Although this authority is extensive, it is not unlimited. Top management
likes to be kept informed of district commanders’ decisions and is willing to contravene
them. Structural shifts toward geographic management vary among departments but are

5 Respondents were asked to complete the survey before leaving. Ninety-seven surveys were completed in Lowell,
136 in Minneapolis, and 217 in Newark.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

10

generally modest; only a few specialized units are placed under the direct supervision of
district commanders.

Organizational flexibility: The capacity of the organization to shift resources in response

to non-routine work demands can be achieved through a variety of means: ad hoc
decision making, formation of taxi squads and task forces, and sharing of resources
among districts. Compstat, however, presents organizations with two paradoxes: (1)
Increasing the allocation of resources to each district commander, or within-district
flexibility, simultaneously reduces the organization’s capacity to shift resources among
districts; and (2) The combination of two Compstat elements—internal accountability and
geographic organization of operational command—fosters competition that hinders
resource sharing among districts.

Data-driven problem identification and assessment: Compstat has significantly enhanced

management and analysis of data. District commanders previously relied heavily upon
reading daily crime reports to identify problems and patterns, and they supplemented this
information with personal experience and anecdotal evidence. Examining crime maps
and reviewing summary statistics was not essential to this process. Crime data helped
inform but did not drive decision making.

Innovative problem-solving tactics: Despite improvements in the gathering and

dissemination of information, departments continue to rely on traditional approaches,
such as increased traffic enforcement, to solve problems. The pressure of internal
accountability hinders careful sifting of, and deliberation about, patterns of crime; a
consideration of those patterns; and a careful review and discussion of the benefits and
drawbacks of various approaches—the very kind of analysis that Compstat claims to
encourage. District commanders feel compelled to respond to crimes quickly rather than
to pursue the most effective strategy possible.

Mission Clarification

The first element of Compstat is mission clarification. Compstat assumes that police

agencies must have a clearly defined organizational objective in order to function effectively.
Supporters of Compstat have argued that a mission statement makes an important contribution to
the overall success of the program. When Bratton assumed command of the NYPD, one of top
management’s first tasks was to clarify a mission statement that embodied the organization’s
fundamental reason for existing. Establishing a clear objective could change the department’s
“mindset” about “what the police should and could accomplish,” and it could instill a shared
sense of “belief, desire, and commitment” to the organizational changes necessary to achieve
this goal (Silverman 1996, 4). In order to convey a clear sense of the NYPD’s commitment, top
management reasoned that the mission statement must include specific terms for which the
organization and its leaders could be held accountable, such as reducing crime by 10 percent in a
year (Bratton 1998, 252). The establishment of a mission statement, therefore, helps the police
agency to function more effectively by encouraging leaders and line officers to commit to a
clearly defined goal, in this case crime reduction, that is highly valued by the department’s
leadership (Bratton 1995).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

11

Our research suggested that the importance of mission statements in establishing a
principal organizational objective might be overstated. Not only might there be more effective
means to communicate the values of the organization, but mission statements might actually
undermine widespread commitment to the organization’s goals—the opposite of what is
intended. Our data revealed: (1) Mission statements establishing explicit terms for crime
reduction are not necessarily preferable; (2) Mechanisms designed to reinforce management’s
message may be less effective than intended; (3) Changing work routines to match
organizational values can weaken commitment to these same ideals; and (4) Mission statements
that conflict with other organizational goals may generate disaffection among those they are
intended to inspire.

Although Compstat in the Lowell, Minneapolis, and Newark police departments was

closely associated with fighting crime, the departments differed markedly in how they
incorporated this goal into a clearly defined mission statement. According to its supporters, the
particular power of Compstat lies in its ability to get police leaders, managers, and line officers
to commit to a specific goal to which they feel accountable. However, our observations indicated
that line officers in departments with no Compstat mission statement might be equally, if not
more, committed to reducing crime than those in departments that establish explicit terms for
crime reduction or articulate crime control as a specific mandate. In other words, mission
statements are merely one, and not necessarily the most effective, means of creating a culture
that supports a particular set of values. For example, a clear and powerful crime-fighting vision
can also be conveyed through the general strategies and routines of a department—that is by
underscoring a particular set of means, such as rapid response time or deference to rank, rather
than a specific end for policing.

Furthermore, our data showed that mechanisms designed to reinforce the values of the
organization may not work as intended. One important approach to underline the importance of a
certain goal would be to implement a department-wide training program. In order to ensure that
training is not merely regarded as a symbolic expression of top management’s priorities, this
approach could be accompanied by several other organizational changes. These might include
convincing first-line supervisors that promotion of these goals is an important part of their job
and realigning the system of rank-and-file rewards, or assignments and promotions, so that it is
consistent with these goals. The organization might also make it clear that these goals take
precedence over other worthy objectives, such as protecting high performers who tend to neglect
other department goals (Mastrofski and Ritti 1996).

Although clear mission statements supported by other mechanisms might not ensure
widespread acceptance of an organization’s goal, it is still possible for the organization to do
other things that bring the behaviour of officers into conformance with top management’s
priorities. For example, top management might implement operational tactics targeted toward
achievement of a desired goal, such as assigning officers to foot patrol to encourage closer
police-community relations. Unfortunately, we did not have any direct measures of the extent to
which officers actually changed their behavior to be consistent with Compstat-linked values.
Consequently this discussion, using results from our officer survey, focuses on the relationship
between mission statements and the degree to which officers shared the crime-fighting values of
Compstat.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

12

Although we did not measure changes in officers’ behavior, we would expect that those
organizations that attach their goals to specific operational tactics would be most successful in
encouraging employee commitment, or a positive attitude, to a desired set of values. The
decoupling of mission statements from the realities of the workplace makes it more likely that an
organization’s mission will merely be regarded as a slogan that lacks any real substance
(Wasserman and Moore 1988). As a result, it runs a high risk of being widely ignored or
regarded with cynicism. The absence of any specific mechanisms for reinforcing the mission and
helping the organization achieve its goals may suggest that managers are not truly committed to
the values they espouse, so why should others care about them?

Finally, the simplicity and directness of a mission statement specifically focused on
reducing crime should help inspire commitment among the rank and file. But what happens
when an organization is committed to several goals that embody different values? Police
departments that have implemented community policing, for example, are supposed to accord as
much importance to community decision making, standards of integrity, problem solving,
reducing fear of crime, and improving social order as they do to fighting crime (Moore 1992).
Our research suggested that in departments that take community policing seriously, management
might be giving mixed messages about which values are really important. This gives rise to a
potential paradox: The more an organization responsible for a variety of important tasks focuses
its limited resources on only one goal, the more likely it is that the goal will be met with
disaffection among those whom it is intended to most inspire. Our research challenges
Wasserman and Moore’s contention that a coherent and predictable relationship between values
and administrative decisions “helps employees make proper decisions and use their discretion
with confidence that they are contributing to rather than detracting from organizational
performance” (Wasserman and Moore 1988, 3, emphasis added). Although individual values and
administrative decisions may be internally consistent, several of the organization’s core values
may be incompatible with one another, such as a focus on both reducing crime through zero-
tolerance policing and improving police-community relations through neighborhood programs.
Top managers might try to reduce the frustration and confusion of patrol officers who feel caught
between these conflicting goals by showing that they are in fact related. In trying to do so,
however, they might have to sacrifice the sine qua non of the Compstat mission, its clarity of
purpose.

In our comparison, the Minneapolis police department was the only department: (1) to
declare a crime-reduction goal of a specific amount; (2) to accompany its Compstat program
with department-wide training; and (3) to implement a specific set of operational tactics designed
to help the department meets its goal. In fact, the department’s top managers told us that a key to
the success of Compstat within the organization was that, “The structure exists to support the
mission.” To help the department achieve widespread acceptance of its goal of a 10 percent
reduction in Part I crimes, every officer attended a day of Compstat training prior to the
program’s implementation. In contrast, officers in Newark and Lowell received no Compstat
training. Furthermore, the MPD’s top management required district commanders to order the
rank and file to use their uncommitted time for directed patrol and to focus on making arrests for
even minor breaches of the law in order to reduce serious crime, or zero-tolerance policing.
Despite these changes, the patrol officers we surveyed in Minneapolis appeared less

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

13

committed—or at best, the most ambivalent—to the Compstat mission of crime fighting than
those in Lowell and Newark.

Their ambivalence expressed the conflict between organizational goals and reflected our

survey finding that patrol officers in departments where management was more committed to
community policing were less committed to the Compstat mission of reducing violent crime. In
Newark, where community policing, although important, was less of a priority than in the other
two sites, patrol officers were most committed to the goal of crime reduction. In Minneapolis,
which had allocated the most resources to community policing, patrol officers were the least
committed to the Compstat mission. The level of acceptance of the patrol officers we surveyed in
Lowell fell between the two.

It is possible that this latter finding was an artifact of our survey since we asked officers

how important Compstat was to the goal of reducing “violent crime” rather than all crime.
Minneapolis’ property crime rate in 2000, the year we administered the survey, was much higher
at 6,341 offenses per 100,000 people than in Lowell at 3,138 offenses per 100,000 and Newark at
5,728 per 100,000. Officers’ responses, therefore, might merely have reflected a more general
attitude that reducing violent crime was not as important to the department’s Compstat mission
as reducing property crime. This explanation is not very convincing, however, when we consider
that the goal of Minneapolis’s Compstat program—clearly displayed at every Compstat meeting,
embodied in the department’s daily routines, and required weekly of district commanders—was
to reduce all index crimes, both violent and property, by a specific percentage.

A more persuasive explanation is that management unintentionally weakened the power

of its message by reducing officers’ discretion and insisting that they spend more time on
specific crime-reduction strategies than on other tasks, such as calls for service, that the
department deemed important. In other words, a strong commitment to the department’s
Compstat goal was actually weakened by the very activities that were designed to help reinforce
its message.

Our general survey showed that of those large departments that had reported they had

implemented a Compstat program, 92 percent reported that they “set specific objectives in terms
that could be precisely measured.” In this sense, the three departments were typical of other
Compstat departments across the country. The overall missions that were clearly promulgated in
much of their literature incorporated several measurable goals—to reduce or minimize crime and
the fear of crime—within the more visionary objectives of maintaining citizens trust or fostering
positive change.

Compstat, however, demands that departments establish a clear and specific

organizational mission rather than a more general commitment to a broad set of objectives.
When we asked, “In the last 12 months has your agency publicly announced a goal of reducing
crime or some other problem by a specific number?” less than half, or 49 percent, of the
departments that claimed to have implemented a Compstat-like program had announced a goal of
reducing crime or some other problem by a specific amount. Almost a third of these departments
had focused on “many different goals” (Weisburd et al. 2001, 31).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

14

Our fieldwork supported these mixed findings from our national survey. The level of
importance placed on creating mission statements with specific goals for crime reduction varied
widely among the three departments. Minneapolis was the only department to commit to
reducing crime by a certain amount. In comparison, the mission statement of the Lowell Police
Department—“To become the safest city of its size in the United States”—stressed powerful
imagery over potentially confusing percentage reductions, and Newark’s mission statement was
distinguished by the complete absence of an explicit crime-reduction goal. In fact, the chief made
a point of not defining a goal by an “X percentage reduction” in crime, as he noted:

In some cases I set specific goals, in crime, no, never. I’m never going to say
bring down crime by 10 percent. We don’t set objectives of reducing crime by 10
percent. I mean, we have thought about that, that’s kind of difficult to do. I don’t
know how I turn around and say we’ll reduce burglary by about 5 percent … I
would love to say we can come up with a percentage but we can’t.

Our findings suggest that the argument that a clear and specific mission statement is

necessary for fostering commitment to an organizational objective is more tenuous than it may at
first appear. It may well be that there is no need for a value-laden goal to mobilize patrol
officers’ efforts to control crime. Our research also indicates that should a department adopt a
mission statement to foster commitment to reducing crime, it might be unnecessary to announce
a specific percentage goal. Any kind of Compstat-related statement underscoring the importance
of crime reduction, no matter how ambitious, vague, or implicit, will resonate equally well with
members of the organization. In fact, we observed a paradox: The more an organization that is
responsible for a variety of important tasks focuses all of its energies on only one goal, the more
likely it is that the goal will be met with disaffection among those whom it is intended to inspire.

The mission statement of the Minneapolis Police Department was identical to that of the
NYPD during Bratton’s first year of command. When we visited in 2000, the MPD had
established the explicit goal to reduce Part I crimes by 10 percent. One of the district
commanders noted that the department’s formerly wordy mission statement, posted on each
district’s wall, was now “down to two words—reduce crime.”6 We used our survey to measure
how much the patrol officers at the bottom of the police organization had adopted the “mindset”
of top management. This gave us some indication of how mission statements appealed to those
who were the primary targets of the department’s message. Did they accept their organization’s
“core value” (Simons 1995, 82-83)?

We would expect the rank and file in the Minneapolis Police Department, which
announced a highly visible and specific crime-reduction goal, to rank “reducing violent crime” as
very important to the department’s overall Compstat program. It is likely this message would
resonate less powerfully with patrol officers at Lowell given our hypothesis that the audacity of
the department’s Compstat goal—“To be the safest city of its size in the United States”—would
mitigate its impact in the department. It was just too ambitious for police officers to incorporate
into their daily operations. Finally, since Newark’s Compstat program did not establish a crime-
reduction goal, we would predict that patrol officers would be least committed to the mission of
crime reduction.

6 The Minneapolis Police Department met its goal in 2000; it reduced Part I crimes by 10.38 percent.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

15

Despite the different levels of importance that individual departments accorded mission
statements, our survey results revealed that patrol officers in all the departments generally
considered Compstat integral to fighting crime. When we gave patrol officers a list of fifteen
goals and asked them to rate them in terms of their importance to Compstat,7 approximately 90
percent of the officers considered Compstat “very important” or “somewhat important” to the
reduction of violent crime. Nevertheless, the difference in the number of those who thought it
“very” and “somewhat” important was the opposite of what we had predicted (see Table 2). In
Minneapolis, only 52 percent of patrol officers responded that reducing violent crime in the city
was “very important” to the department’s Compstat program. This contrasted sharply to Newark
where a much higher proportion of the rank and file, or 82 percent, responded that this was a
“very important” feature of Compstat. With 77 percent responding that reducing serious crime
was “very important” to the department’s Compstat mission, the response of Lowell’s patrol
officers lay between the two. Moreover, there was only a 5 to 6 percent difference across all
three departments among patrol officers who responded that the importance of reducing violent
crime to the department’s Compstat strategy was “not at all important.”

Table 2: The importance of reducing violent crime to the department’s Compstat strategy

Importance of reducing violent crime in city*
% Very

important
% Somewhat

important
% Not at all
important

% Don’t know

Lowell 77 15 6 2
Minneapolis 52 38 6 4
Newark 82 10 5 3

Chi Square DF Significance N
48.626 6 *P<0.001 446

The large amount of publicity Compstat has received in the press and within police
circles probably contributed to the broad consensus among patrol officers regarding Compstat’s
crime-control focus. For example, many highly visible public figures, including former
commissioner William Bratton, have attributed the crime drop in New York City to Compstat,
and their opinions have been widely publicized by the popular media and discussed at academic
and professional meetings attended by police leaders. Moreover, the elegant simplicity of
Compstat’s goal helps explain its widespread recognition. In contrast to the confusion
surrounding the many goals of community policing, Compstat and crime fighting are so closely
associated with one another that they are almost synonymous. Finally, since Compstat validates
the central professional role of the police—fighting crime—we would expect it to appeal to line
officers (Brown 1981).

But what might explain the differences among the departments—a difference that is the

opposite of what we would expect given the purported importance of mission statements? The
paradox that a specific crime-control goal might fail to mobilize a very deep and widespread
commitment to a department’s objective could be explained by the relative unimportance of

7 For a list of the goals, see question one of the line officer survey (Appendix).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

16

formal mission statements in promoting an organization’s core values. Newark’s Compstat
program did not have a specific crime- reduction goal. Whereas Bratton used his mission
statement to generate support among external constituents, Newark’s chief rejected this approach
to generating public interest in the department’s efforts to make the city’s streets safer.
Nevertheless, the chief did manage to convey the importance of this goal, at least to those within
the department, by exhorting his troops and using less direct means than articulating specific
terms for crime reduction.

When he first assumed command of the department in 1996, Newark’s chief marked his

arrival with a town hall meeting attended by all department members. He made it clear that he
wanted everyone in the organization to recognize “first and foremost” that they were to deal with
“crime and service citizen’s complaints,” and Compstat was a means of refocusing their attention
on this “core responsibility.” This message was not encapsulated neatly in a goal to reduce crime
by a certain percentage, but it was delivered effectively through the chief’s emphasis on
measurable results. His vision was further reinforced by the crime-fighting subculture embedded
within the department’s values, norms, structures, and daily activities.

The Newark Police Department was the most militaristic of the three we visited and paid

particular deference to rank, especially that of the highest-ranking official in the organization.
For example, when the chief entered the Compstat room his arrival was announced and all of his
officers immediately stopped talking, stood at attention, and saluted. Furthermore, the tenor of
the Compstat meetings was very serious, and the chief made it patently clear that district
commanders were directly responsible for fighting crime. In contrast, the atmosphere at
Compstat meetings in Lowell and Minneapolis, although tense, was more cooperative than
combative. Lowell’s chief argued that it was “counterproductive” to humiliate individuals during
Compstat and advocated a more humanistic approach: “Police officers are competitive by nature
and all you really need to do is give them the facts and ask them a question. They will go from
there.” Just asking district commanders a question in front of their peers, the chief said, would
make them feel accountable. Similarly, the chief of the MPD described his department’s
Compstat process as “a more cooperative learning environment” in contrast to the more
adversarial nature of the NYPD program.

In Newark, Compstat was a useful vehicle for the chief to send his command staff a

crime-fighting message that was also conveyed clearly down the chain of command. It was
sufficiently ingrained that officers knew not to bother their district commanders on the days
preceding Compstat when they were busy preparing their presentations. To do so would be to
risk a stern rebuke. Departments that are distinguished by their emphasis on formalized
hierarchical authority, discipline, and technical efficiency—in Newark’s case, an ardent
commitment to minimizing response time—tend to stress the importance of crime control. James
Q. Wilson referred to departments that adopt this style as “legalistic” (1968, 183-88). Crime
control as a core value was embodied in the mechanics of Newark’s Compstat meetings, but
more importantly, it suffused the entire police organization. Consequently, this mitigated the
need for an explicit mission statement announcing a measurable goal to inspire commitment. A
crime- fighting ethos was conveyed by the words and actions of its leader and was incorporated
within the department’s subculture.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

17

Unlike Newark, both Lowell and Minneapolis had highly visible mission statements that
were widely recognized inside and outside of their departments. Both chiefs had made great
efforts to publicize their missions in the local press in order to galvanize public support. In
Minneapolis, the 10 percent crime-reduction goal was mentioned so often within the department
that it was almost a mantra; and the “safest city” goal was widely recognized in Lowell.
However, widespread recognition of a department’s mission to reduce crime does not guarantee
overwhelming commitment to management’s core values. In Lowell, one respondent explained
that the only time he heard the mission statement within the department was when an officer at a
crime scene joked, “Uh-huh … another murder in the safest city in America.” Although, this
sardonic remark might express some indifference toward such a lofty and improbable goal, it
also supported our survey finding that Lowell’s patrol officers, at the very least, recognized and
accepted the relationship between Compstat and the department’s orientation toward fighting
serious crime.

In contrast to Lowell, our observations and interviews with rank-and-file officers in

Minneapolis suggested that they were more ambivalent or hostile toward the department’s
mission statement. They bought into Compstat, not because they believed in the value of the
mission, but because they were obliged to do so by the constant presence and pressure of a
mission statement that clearly established Compstat’s purpose. The specific crime-reduction goal
was announced at every Compstat and embodied in the maps displayed at every district’s roll
call, but some questioned the department’s commitment to this approach. It appeared that some
patrol officers regarded the means that the department was using to attain this goal as
disconnected from, and even counterproductive to, the message.

Not only might a mission statement fail to inspire a more general commitment to the
organization’s goals, it might actually undermine the message it is trying to deliver. The
likelihood of this occurring will increase if the organization tries to achieve a number of goals
that may conflict with one another. Disaffection among the rank and file was highest in
Minneapolis where the department’s leadership had placed a great deal of importance on
community-oriented policing. Moreover, the MPD, in comparison to Lowell and Newark, had
allocated the most resources to its community-policing program. The department had recently
enhanced and developed its district-level partnerships with citizen groups by decentralizing its
Community Services Bureau to provide each district with a Community Crime Prevention/SAFE
(CCP/SAFE) team of community-policing specialists. In addition, just after our site visit ended
in December 2000, the MPD initiated a community-outreach, or Community Engagement
Project (CEP). While Compstat holds sector lieutenants directly accountable for crime levels in
their beats, the CEP makes them responsible for their officers’ relationships with local citizens.
The LPD, under the leadership of its chief, was also strongly committed to community policing.
Districts doubled as storefronts, and the department had invested in a motorized command
center, “The Van,” to provide a speedy and mobile response to neighborhood concerns. Despite
the considerable success of community policing in Lowell, the department had not allocated
additional resources, in the form of specialized community-policing teams, to its districts. The
NPD committed the fewest resources to community policing, and our observations suggested
that it was not as high a priority as in Minneapolis and Lowell. For example, many of the mini-
stations the NPD had implemented in the 1980s were no longer used for neighborhood meetings;
the community response team remained centralized in department headquarters; and community

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

18

policing was not a regular topic of discussion during either department meetings or our
interviews.

The MPD’s strategy of reducing officer autonomy by mandating directed patrol was a
conscious management effort to keep officers from following reactive rather than preventive
practices. The department’s leadership felt that direction needed to come from top management
until rank-and-file culture had fully accepted the preventive approach to crime control.
Nevertheless, it appears that top management’s message about the values of the organization
unintentionally compromised rank-and-file commitment to the department’s Compstat mission.
By simultaneously emphasizing core values of both community policing and Compstat, top
management made many of the rank and file feel caught between conflicting goals. For example,
the importance of being responsive and attentive to citizens’ needs appeared to conflict with an
approach that de-emphasized calls for service. How could patrol officers provide an effective
service to all members of the community while being asked to ignore some citizens’ requests for
assistance? Similarly, there was a contradiction in encouraging patrol officers to exercise
decision-making autonomy while also limiting this autonomy by requiring that they police
specific areas through directed patrol and employ the specific strategy of zero-tolerance policing.

When Bratton took over as commissioner of the NYPD, he reinforced the importance of

the department’s mission by constantly reminding, measuring, and rewarding officers for
successfully reducing crime in their districts. This proved to be a powerful mechanism for
connecting the mission statement to the realities of police work. A similar approach was tried in
Minneapolis, but it seemed less successful. To help reduce crime, officers were issued verbal and
written instructions to use their uncommitted time patrolling hot spots and engaging in
aggressive order maintenance. District commanders underscored the importance of directed
patrol by using a variety of approaches to hold their officers accountable. Some asked their patrol
officers to mark their reports with an acronym DP, indicating that whatever action had been
taken, whether it be an arrest or a citation, was a result of directed patrol. Other district
commanders scanned their crime maps for evidence of increased patrol officer activity, such as
an increase in arrests of suspicious persons, in areas they had identified as hot spots.

These strategies would appear to demonstrate top management’s commitment to

achieving its mission, but they also contributed to confusion and resentment among patrol
officers. One officer commented that, “Answering 911 calls is not a priority. Response time to a
person that needs help is not viewed as being a crucial part of squad work anymore.” Other
officers remarked that Compstat resulted in heavier patrol for high-crime areas and reduced
patrol coverage in other neighborhoods. These responses demonstrate that Minneapolis’ police
managers matched the actions of their patrol officers with the value orientation of the
organization, but they were not entirely successful in ensuring patrol officers’ commitment to the
department’s crime-fighting goal. For some, the mission statement provoked cynicism, and for
others it caused a stronger reaction. Some respondents resented the entire Compstat program
because it devalued other important police tasks, such as fulfilling citizens’ requests for help. In
addition, it committed a significant amount of patrol resources to dealing with minor rather than
more serious crimes, though this was the very purpose that it claimed to fulfill. As one officer
put it succinctly, “You don’t get to work on real crime [because of Compstat],” while another

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

19

commented, “[Compstat] focuses on the minor offenses hoping for them to turn into something
bigger … wasting time to stop people for loitering and not answering calls.”

In sum, our data showed that officers in Lowell, Newark, and Minneapolis closely

associated Compstat with crime fighting. Whether this could be attributed to a specific mission
statement is less clear. Leadership in Minneapolis placed a premium on Compstat’s crime-
reduction goal through a highly visible mission statement that was reinforced by training and a
policy of holding officers accountable for pursuing specific crime-fighting tactics. Minneapolis
patrol officers, however, were less supportive of the department’s overall goal than those in
Newark, which had no Compstat-related mission statement, and Lowell, which had established a
less tangible goal. In Newark, the goal of fighting and reducing crime was conveyed through the
actions of a charismatic chief and resonated strongly with the department’s strong authoritarian
subculture. There was no need to formulate a crime-fighting mission statement because this core
value was already embodied in the management and operation of the department. Compstat
merely reinforced this technical orientation to reducing crime—a defining characteristic of the
traditional model of policing (Weisburd et al. 2003).

Moreover, our findings suggested that when a department does implement a mission

statement, it might be unnecessary to incorporate a specific crime-reduction goal to foster
widespread commitment among all members of the department. Lowell’s mission statement
relied more heavily upon powerful rhetoric to mobilize its officers than on a more concrete
crime-reduction goal, yet the rank and file seemed less resistant to the department’s goal than it
did in Minneapolis. Since a department’s failure to meet a specific performance goal might lead
to criticism within and outside the department, it might be in the department’s interests to
announce a more open-ended, visionary, or improbable goal that leaves some room for failure.
By allowing for the possibility of failure, a department can shield itself from public criticism.

We might expect Compstat’s narrow focus on crime fighting to conflict with an

organization’s commitment to the wider set of values associated with a community-policing
program. In contrast to Compstat’s exclusive focus on crime control, community policing gives
equal importance to police-community partnerships, improving community relations, problem
solving, and protection of citizens’ rights. These critical areas, proponents of community
policing argue, provide a more accurate representation of the complexity of the modern officer’s
role and the array of services that the public expects police to offer. Some research suggests that
street officers are resistant to community policing because it is inconsistent with their crime-
fighting subculture (Chan 1996; Greene 2000; Mastrofski et al. 2002). We have no evidence to
suggest that Compstat’s concentration on crime fighting reduces the organization’s commitment
to community policing, but our research does demonstrate that there may be some tension
between the two. In Minneapolis, which exhibited the strongest commitment to community
policing, at least in terms of resources, comments on our survey revealed that some felt very
strongly that the department’s focus on crime reduction in specific areas weakened its capacity to
fulfill other important obligations, including its ability to deliver necessary services to all city
residents. Some recognized that focusing on misdemeanor and quality-of-life crimes contributed
to reductions in more serious crimes, but others disagreed. As one officer put it, “It seems that
the ‘good’ citizens [who pay taxes and obey laws] are still pushed aside and do not get the
service they deserve.” Despite top management’s best efforts to expand the department’s goals

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

20

beyond crime control, line officers wanted to keep things simpler, since they saw a tension in
trying to focus on both community policing and crime control.

Furthermore, despite the strong support of Minneapolis’ mayor and the concerted

attempts of the MPD to gain the community’s acceptance, some city residents greeted Compstat
with a great deal of suspicion. In response to an open-ended question on our survey, a patrol
officer wrote that the, “Public was outraged at our aggressive police work; soon after [Compstat]
started the media and community started ‘police profiling.’” Members of the community,
especially African Americans, associated Compstat with the kind of aggressive law enforcement
that unfairly targets minorities. In order to lessen antipathy toward a specific mission statement
within the department and to avoid harming police-community relations, it may be necessary for
a department’s leadership to make an even more concerted effort—over and beyond training
sessions, meetings with community members, and public announcements—to convince line
officers and community members that: (a) Compstat attempts to improve the overall quality of
police service by emphasizing, not substituting, crime reduction for other worthy goals; and (b)
Compstat is a program that specifically targets crime problems, not minorities. The MPD’s
Community Engagement Program represents an additional attempt by the department to foster
closer police-community relations. Department leaders felt that strenuous community-outreach
efforts through a variety of programs were necessary for the department to successfully foster
widespread commitment to its crime-fighting mission among line officers and members of the
community.

Internal Accountability

For a department’s mission statement to be effective, officers need to be held responsible

for meeting the goals it espouses. Perhaps the most vivid Compstat element is that of holding
operational commanders accountable for knowing their command, being well acquainted with its
problems, and accomplishing measurable results in reducing them—or at least demonstrating a
diligent effort to learn from the experience. In short, Compstat makes someone responsible for
tackling and reducing crime, and those who fail to do so will suffer adverse career consequences,
such as removal from command.

Our fieldwork showed that regular Compstat meetings provided a very effective

mechanism for making district commanders in all three departments feel directly accountable for
their performance. District commanders expected the chief to ask many probing questions about
crime, and they openly acknowledged that it was their responsibility to come to Compstat fully
prepared with an array of responses. Some scholars have criticized police programs, such as
community policing, for being rhetorical or symbolic (Klockars 1988), but Compstat meetings
cannot be dismissed as scripted performances designed merely to impress visiting dignitaries, the
press, and members of the public.

The recognition that Compstat makes command staff members feel accountable does not

eliminate the possibility that this feeling is experienced infrequently, less intensely, or differently
than the program’s supporters allege. Compstat’s capacity for developing a degree of
accountability that is significantly higher than what has been achieved in the past ultimately rests
upon its claim to provide significant and predictable consequences for performance. Therefore,

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

21

we need to examine these consequences carefully if we are to make a clearer assessment of
whether Compstat is more bark than bite. How much can a chief punish those who fail to meet
the goals of the organization, including removing them from their position? How much reward
can a chief offer those who succeed?

Our observations suggested that incentives were largely limited to making public notice

of good and bad performance. When Bratton took over the NYPD he had the power to transfer
precinct commanders to less desirable posts or make their work life sufficiently unpleasant that
some decided to retire early. Our observations suggested that top executives in our three sites
shared similarly broad powers but were reluctant to use them in ways that could undermine their
leadership. Furthermore, a chief’s power is further constrained in smaller departments like
Lowell. The small pool of qualified candidates for the position of district commander lowers the
risk of losing one’s position for marginal performance. Furthermore, the magnitude of the
responsibility commensurate with the position, not to mention the administrative burdens,
imposes additional constraints. In Lowell, it was unlikely that all who qualified would choose to
serve as one of the three district commanders in the department. One interviewee pointed out that
given the significant responsibilities of the position, “Maybe four out of eight captains [total] are
willing to do this job.” Finally, in small departments there may be very few positions above the
rank of district commander, thus undercutting the incentive of promotion.

These limitations only had a small mitigating effect on the level of accountability

experienced by district commanders across all three sites. At the same time, they did help foster a
stronger impression, particularly among some of the lower ranks, that Compstat functions to
ridicule command staff and operates primarily on a symbolic level as “just a show.” The
impression that Compstat is designed to expose district commanders to harsh criticism may
weaken its appeal among the rank and file. In addition, since a defining characteristic of the
Compstat accountability mechanism is the possibility of removal from command, a chief’s
reluctance to carry out this ultimate threat may undermine the legitimacy of the program.

Part of this cynicism surrounding Compstat’s focus on accountability can be attributed to

the very minor role street supervisors and line officers play in the overall Compstat process.
Rank and file were not excluded from attending meetings, and Minneapolis, for example, patrol
officers were required to attend one Compstat meeting per year and sergeants sometimes gave
presentations. Approximately half of the patrol officers we surveyed, however, had not attended
a Compstat meeting. Since the lower ranks rarely attended Compstat meetings, they tended to
hear about how “so-and-so” was “balled out” rather than about positive features of Compstat,
such as information sharing and problem solving, that contributed to crime reduction. This
probably fed Compstat’s powerful reputation among line officers as brutal and punitive rather
than collaborative and creative. Our observations suggested that Compstat reinforced a more
general impression among line officers that police organizations are punishment-centered
bureaucracies. This is not the mindset chiefs want to foster when promoting a climate of
innovative problem solving, since innovation increases the visibility and cost, if not also the
likelihood, of failure.

More importantly, in the absence of other structures, our observations showed that line

officers experienced accountability far less acutely than the district commanders. Furthermore,

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

22

how they experienced Compstat as an overall program was opposite to that of the department’s
middle managers. Since Compstat holds command staff directly accountable for resolving crime
problems, middle managers were under a great deal of pressure to implement crime strategies. At
the weekly or biweekly Compstat meetings, it was imperative that a district commander be able
to demonstrate that he or she had identified any crime problem and had already implemented a
response. In Newark, the stock response to the chief’s question about whether a crime response
had been implemented was “as we speak.” In contrast, the primary responsibility of patrol
officers was to follow the orders that flowed from district commanders and their executive
officers down the chain of command. As a result of the pressure on district commanders,
Compstat concentrates strategic decision making at the top of the organizational hierarchy rather
than promoting initiative among the lower ranks. By limiting the decision-making autonomy of
patrol officers, Compstat contradicts one of the central tenets of community- and problem-
oriented policing—decentralized decision making resulting from a delegation of authority to
initiate problem solving at the level of the rank and file (Goldstein 1990; Trojanowicz and
Bucqueroux 1990).

Responses to our national survey and intensive site interviews showed that departments

that have implemented a Compstat-like program consider internal accountability to be a very
important feature. Certainly punishing middle managers for failing to meet the standards of
Compstat accountability is a key element of this model. Nearly seven in ten of these departments
told us that a district commander would be “somewhat” or “very likely” to be replaced if he or
she “did not know about the crime patterns” in the district. A much smaller proportion of these
departments reported that a district commander would be replaced simply because crime
continued to rise in a district (Weisburd et al. 2001). This perhaps reflects Compstat’s demand
that commanders be familiar with problems and initiate solutions to them, though they are not
held accountable for achieving outcomes that may be unresponsive to well-planned police
interventions (Bratton 1998).

While the use of “punishment,” in the form of a rebuke from the chief, to maintain

accountability was very much apparent in Compstat meetings, we found that departments were
much less likely to use tangible rewards to ensure internal accountability. If crime in a district
declined, less than a quarter of Compstat departments reported that it was “very” or “somewhat
likely” that the district commander would be rewarded with a promotion or desired job
assignment (Weisburd et al. 2001, 32-34).

Lowell, Minneapolis, and Newark fit very closely to the more general findings of our

national survey. Like police departments nationwide, internal accountability was an integral
component of their Compstat programs. Lowell’s police chief explicitly recognized the
importance of this feature when he defined Compstat as a means “to manage the police
department in a timely manner with an eye toward accountability.” Similarly, Newark’s chief
commented that in his twenty-eight years on the force middle mangers were not held accountable
for crime levels, clearance rates, response time, and police corruption. “Compstat,” he said,
“turned that upside down. We now have developed a new culture, and what you see now in terms
of Compstat today represents a total cultural change. And I think that’s the biggest part of
Compstat.” A high-ranking official in the Minneapolis Police Department stated, “When I hear
CODEFOR, I don’t think of increases in street arrests, I think of accountability.”

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

23

The district commanders we interviewed all recognized and accepted Compstat as a
process that held them accountable for whatever occurred in their beats. 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.’” This comment illustrates the central role that
Compstat meetings played in the accountability process. Since all the command staff attended
the weekly or biweekly meetings, Compstat provided top management with an ideal opportunity
to display its authority and hold district commanders publicly responsible. For example, during a
Compstat meeting in Newark, the chief ordered a sector lieutenant who failed to answer his
questions adequately to “step down” and stand behind his sergeant. This act constituted a vivid
and powerful act of shaming before the department’s Compstat audience.

Since Compstat meetings hold district commanders directly responsible for their

performance in front of their peers, they operate as a very effective accountability mechanism.
The possibility of receiving praise did not motivate district commanders nearly as much as the
possibility of public censure. This can be attributed to the very unpleasant and deeply visceral
reactions that embarrassment provokes in most individuals (Miller 1995). Compstat’s particular
power is based on this assumption. The mere possibility of censure from the department’s
highest-ranking official in front of one’s peers was motivation enough. The chance of a
“roasting” from the chief and the district commander’s reaction are the principal dramatic
elements at Compstat meetings. Everything else—regular time and place of the meeting,
recognizable presenters, similar order of events, familiar maps, and crime data—operates fairly
predictably. The potency of this effect is further enhanced by the specific dramaturgical elements
of Compstat meetings.8 With the display of sophisticated maps, an audience of top-ranking
police officials in full regalia, and a focus on crime-fighting strategies, Compstat successfully
combines drama, imagery, and rhetoric. The result is a tension-filled atmosphere that conveys a
powerful impression that crime control is the organization’s most important goal, that the means
of knowing the state of crime control are at management’s fingertips, and that the levers of
organizational action are at its disposal. At center stage in this dramaturgy are the district
commanders, who bear the burden for the organization’s success.

Our observations suggested that the impression of Compstat as a forum that punishes

rather than rewards command staff for its performance is widespread among the lower ranks.
Compstat provides a mechanism for making “people do the job they do,” but some are clearly
discouraged by the confrontational atmosphere that can characterize meetings. One officer in
Lowell described Compstat as a forum where officers had “their balls ripped off” and surmised
that this only served to make individuals “reluctant to speak up … reluctant to do their job.” A
sergeant in the same department, who had attended a handful of Compstat meetings, described
them as “nerve-racking” and remarked that they could be used “to humiliate or embarrass
people.”

The possibility of rebuke certainly appeared to make patrol officers aware that Compstat

held district commanders to a high level of accountability. The general recognition among the

8 “Dramaturgy examines how symbols, both material and nonmaterial, are selectively presented in order to impress
an audience” (Manning 2001, 317).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

24

rank and file, however, of the unlikelihood that a district commander would be replaced for
inadequate performance counteracted the potency of this mechanism.

During the first year of Compstat, 76 percent, or more than two-thirds, of all NYPD

district commanders retired. Commissioner Bratton replaced them with middle-ranking officers
who had reputations for being aggressive risk takers: “These were early and swift personnel
replacements by a commissioner who prided himself on rapid and effective change” (Silverman
1999, 86). Resistance from mid-level managers has often caused top management’s reform
efforts to fail (Kelling and Bratton 1993). Following the major prescriptions of organizational
development experts to accomplish organizational change (Beer 1980), Bratton was clearly
committed to replacing the old guard. In an interview, he emphasized the importance of
empowering a chief to replace underperforming managers:

John Timoney [formerly on Bratton’s team and the police chief of Philadelphia until
2002] has been given little power to move deadwood in Philly and there is a lot of it
there. Similar problems exist in Los Angeles. They are still doing a lot of good things in
both cities, but it slows the process down. You’ve got to move the deadwood and put in
good managers to make any COMPSTAT style process work” (Dussault 1999).

Compstat’s implementation corresponded with new leadership in all three departments,

and the appointments of new chiefs in Lowell and Newark were accompanied by early
retirements. One Lowell officer recalled that the incentive program to buy out officers was a
“sign of the times,” given the large number of older and more senior officers within the
department. In Newark the district commanders who were in place when the chief first took over
were encouraged to retire. Similarly, their replacements, or the “second Compstat generation,”
who were also members of the old guard, soon retired. This allowed the relatively new chief to
place the “third generation,” or those district commanders who were very supportive of
Compstat—three of the four had been personal assistants to the top leadership—into these key
command positions.

These examples illustrate how a chief can use his or her recent appointment as an

opportunity to change personnel by encouraging the retirement of command officers who are less
supportive of a new program. A chief can convey this message by reassigning the officers to
specialized positions that have low status and little power. Alternatively, the chief can move
members of the old guard into high status positions such as that of district commander, which
they openly resist and for which they feel ill prepared. The chief can then contribute to this
unpleasant work environment by exposing them to intense criticism. Of course, a chief can also
assign officers to newly created positions as a reward for good performance.

When Compstat was up and running, none of the respondents in the three departments

recalled a district commander being replaced for poor performance. In Lowell, some
remembered a district commander being reassigned, but that was because he was uncomfortable
with the public speaking and intense scrutiny involved in the Compstat process and asked to be
reassigned. It is obviously different, however, for a district commander to ask for a transfer than
to be removed by the chief for failing to fulfill the obligations of his or her position. Similarly,

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

25

praising a district commander during Compstat is different from offering a more tangible benefit
such as promotion or pay increase.

Top executives have significantly more leeway in reassigning or promoting district

commanders than members of the rank and file, who are protected by restrictive civil service and
union regulations. Our observations suggested, however, that chiefs were very cautious about
exercising this power. In order to maintain the support of department personnel, the selection,
punishment, or reward of middle managers must appear to be based on fair, consistent, and
objective criteria. The arbitrary exercise of authority opens the chief up to criticisms of
favoritism or injustice and may undermine the legitimacy of the department’s leadership. These
concerns decrease the likelihood that a chief will replace his district commanders very often.
Newark’s chief appeared to acknowledge these limitations when he commented that retirement
was the “easier way” of replacing those who resisted Compstat, presumably compared to
reassigning them to another command. Furthermore, the restrictions on leadership are
exacerbated in small police departments where there is only a small pool of qualified candidates,
and a limited number of slots available for promotion. Top leadership’s consequent reluctance to
reassign underperforming district commanders to less desirable positions as a form of
punishment weakens accountability under Compstat and helps create an impression among some
that Compstat is more style than substance.

This impression is probably amplified when a chief fills key positions with officers who

are widely acknowledged to be loyal to him and generally supportive of his reforms. At one of
the sites, an officer questioned leadership’s commitment to holding district commanders
accountable when he stated, “No inspector has ever been demoted.” Another respondent from
this same department commented that one district commander’s consistently poor performance
in reducing crime had merely resulted in a transfer to a less visible and less crime-ridden district,
where his performance still failed to improve. In relation to this incident, another officer
commented: “[Compstat] does not hold [district commanders] accountable … He fucked up
[district X] so badly, and all they did was move him to [district Y]. They were okay when he was
sent there, but look at them now, [district Y] is in the toilet. Is that [name of district commander]
being held responsible for doing a shitty job? I don’t think so.” While there were undoubtedly
many complex factors regarding this decision, the officer clearly associated the district
commander’s reassignment, rather than removal, with Compstat’s failure to hold middle
managers strictly accountable for their performance. A regular attendee at Lowell’s Compstat
meetings made a similar criticism of Compstat’s apparent failure as an effective accountability
mechanism. Having observed a district’s steady increase in motor vehicle crime over successive
Compstat periods, he exclaimed indignantly, “Can you tell me honestly that anything is getting
done because of Compstat?”

The failure to replace district commanders helps explain the ambivalence patrol officers

felt toward Compstat as an accountability mechanism. A higher proportion, or about a third, of
patrol officers in Newark and Lowell responded that holding district commanders responsible for
crimes in their beats was “not at all important” to Compstat compared to a quarter who
considered it “very important” (see Table 3). The fact that patrol officers in Minneapolis were
most likely to feel that holding district commanders accountable was “somewhat important” (51

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

26

percent) and the least likely to feel that this was “not at all important” (23 percent) explains the
significant difference between the sites.

Table 3: The importance of holding district commanders accountable for crimes in their

districts to the department’s Compstat strategy

Importance of holding district commanders accountable for crimes in their

districts*
% Very

important
% Somewhat

important
% Not at all
important

% Don’t know

Lowell 26 35 34 5
Minneapolis 23 51 23 3
Newark 21 35 36 8

Chi Square DF Significance N
14.560 6 *p<0.05 442

Whereas district commanders did not want to “look bad” in Compstat and felt
accountable for having answers, the lower ranks experienced accountability far less intensely. A
district commander might be rebuked for an inadequate response to the chief’s inquiry, but line
officers did not seem to feel that they were in the hot seat for crime increases. There was a
general awareness among line officers that their district commanders were being held
accountable and that “it was important to be in the right place at the right time,” as a sergeant in
Minneapolis stated. This suggests that some sense of accountability did trickle down to the lower
ranks. A district commander from the same department echoed this perspective when he noted
that district commanders were getting better at “pushing the responsibility downwards.”

Our fieldwork, however, also revealed that line officers did not feel the same kind of

pressure as district commanders for their own performance. Approximately half, or 40 to 50
percent, of the patrol officers we surveyed responded that holding officers accountable for crime
in their beats was “not at all important” to the departments’ Compstat program, although this
feature was more important to patrol officers in Lowell (see Table 4).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

27

Table 4: The importance of holding officers accountable for crimes in their beats to the
department’s Compstat strategy

Importance of holding officers accountable for crimes in their beats*
% Very

important
% Somewhat

important
% Not at all
important

% Don’t know

Lowell 18 38 38 6
Minneapolis** 7 44 47 3
Newark 11 35 44 10

Chi Square DF Significance N
15.354 6 *p<0.05 445

** Does not sum to 100 percent due to rounding

Since there was no structure comparable to Compstat meetings for transmitting
accountability to line officers, they did not experience accountability as intensely as district
commanders. 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 thirty-four percent responded “never”
(fourteen percent) or “every few months” (20 percent); and for Newark it was 42 percent (8
percent and 34 percent respectively). These findings challenge Silverman’s assertion that the
impression Compstat creates by grilling district commanders on their crime-reduction efforts
“reinforces the patrol officers desire to combat crime” (1999, 194). Our research suggested that it
was far more likely that the sense of accountability was diluted as one moved down the
command structure. As one Lowell officer put it when asked about Compstat, “If you don’t go,
you don’t know.” Similarly, a patrol officer in Minneapolis who had not attended a Compstat
meeting since receiving his initial training commented, “Compstat is a meeting for the brass. It
doesn’t have anything to do with the street.”

Since accountability relies heavily upon a public setting with high-ranking officials in

attendance, it follows that accountability is experienced less acutely outside of this setting.
Minneapolis had made the greatest attempt to push accountability down the chain of command,
since district commanders and their lieutenants would check officer reports or, less frequently,
crime maps to see if their patrol officers had engaged in directed patrol. This mechanism,
however, was limited. Unlike Compstat meetings, it did not lead commanders to track the
success of these tactics over long periods of time or assign any kind of follow-up. The primary
purpose was to ensure that patrol officers were following orders, not hold them accountable for
the quality of their response. In contrast to Minneapolis, the responsiveness of patrol officers in
Lowell and Newark depended heavily on the will and skill of district commanders to remember
whom they had assigned to specific crime problems, what tactics they had recommended, and
whether the problem had abated. Given the heavy administrative burden on district commanders,
our observations suggested that this level of scrutiny was desultory and usually afforded only the
most serious problems.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

28

Of course, Compstat might have worked indirectly to convey a sense of accountability
down the chain of command. In the MPD, top management expected middle managers to hold
rank-and-file officers accountable; and, across sites, district commanders who had been rebuked
in Compstat for inadequate strategies could always return to their districts and admonish their
line officers. The force of the message was considerably weakened, however, for three reasons:
(1) Compstat ultimately held middle managers, not line officers, accountable; (2) The message
was not delivered by the highest-ranking official in the police department, and (3) It did not
result in public censure on the same scale as embarrassing line officers in front of their peers.

It was clear that line officers experienced accountability less acutely than district

commanders, but our research also suggested that the pressure Compstat put on district
commanders to reduce crime had a significant impact on how line officers experienced
Compstat. One of the important effects of internal accountability under Compstat is to centralize
decision-making authority. Since district commanders are ultimately held responsible for
implementing crime strategies, decision-making authority is shifted away from the rank and file
toward middle management. Since one of the central tenets of problem-oriented and community-
oriented policing is to decentralize decision-making authority, Compstat would appear to
counterpoise this approach. Instituting district-level Compstats might help devolve decision-
making authority further down the organization, as it did in New Orleans. Here the department
held sector-level Compstat meetings the day before the precinct commander had to appear at the
department Compstat meeting. These could last as long as two-and-a half hours and involved
detectives, patrol supervisors, and some patrol officers (Gurwitt 1998). In Newark and
Minneapolis, however, district-level meetings were limited to the district commander, executive
officers, district detectives, and operations sergeants. Rank and file were, therefore, largely
excluded from the Compstat process, as they were at Lowell, a much smaller department that did
not hold any Compstat meetings at the level of the individual district.

Our research revealed that decision making was experienced by line officers under

Compstat as primarily a series of commands directing them what to do. It also appeared that
these commands were rarely justified or explained to line officers as part of the Compstat
process, given the generally low level of officer awareness of the specifics of Compstat. This
was especially true in Minneapolis, where rank-and-file officers were ordered to patrol problem
areas identified by Compstat when they were not responding to a call. This contrasts sharply to
community policing, which encourages the rank and file to take ownership of their beats, to
exercise considerable discretion, and to solve problems. By holding district commanders
disproportionately accountable for their areas and by not transmitting this message directly down
the chain of command, the three departments gave patrol officers no sense that, “This is my beat,
my problem, and I must find ways to solve it.” The concentration of strategic decision making
among middle managers reinforces one of the key features of the traditional policing model that
has characterized police organizations for much of the twentieth century—centralized command
and control (Weisburd et al. 2003; Willis et al. 2004a).

In this respect, Compstat works very much the way that Bratton intended. In Bratton’s

view the NYPD’s community-policing approach under Lee Brown had lost focus on the ultimate
goal of crime reduction (Kelling and Coles 1996). He believed that the young officers assigned
to community-policing beats were “unprepared and ill equipped” to handle complex community

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

29

problems (Bratton 1998, 198-99; Greene 1999, 74). Since Bratton did not trust patrol officers to
make good decisions, he decided to hold middle managers responsible for reducing crime. In
response to a question on the practice of accountability under Compstat, Bratton stated:

One of the significant drivers behind my career was working under a lot of incompetent
bastards. I worked hard to get past the incompetents. Making COMPSTAT work meant
doing our best to bring everybody to the table, giving them a chance to perform and, if
they didn’t, moving them out … where COMPSTAT isn’t having as dramatic a result you
see the opposite (Dussault 1999).

Our findings from the three sites suggested that accountability operated in a more

complex manner than Bratton’s description revealed. It was established through the risk of public
embarrassment, not removal from command. Furthermore, the severe limitations on a chief’s
capacity to punish sub-par performance in his department increased the likelihood that some
would regard accountability merely as a symbolic device. Finally, since accountability was
established through Compstat meetings that were largely restricted to members of the command
and administrative staff, Compstat had only a very limited impact on the level of accountability
that line officers felt for reducing crime in their beats. There were no similarly powerful
mechanisms for extending accountability down the chain of command or reinforcing it outside
that setting. The exclusion of first-line supervisors and patrol officers from Compstat meetings,
the uneven impact of accountability, and the concentration of decision-making authority at the
command level suggest that Compstat represents a shift toward the centralization of command
under the traditional model of policing—one that has been heavily criticized by community-
policing advocates.

Geographic Organization of Operational Command

Compstat holds middle managers to a high level of accountability, but it does not do so
without providing them with the means to carry out the Compstat mission. Mid-level managers
are empowered in this model through the concept of geographic organization of operational
command, which focuses operational command on the policing of specific territories and
delegates primary decision-making responsibility to commanders with responsibility for
territories, such as districts. That is, the organization puts a higher priority on commanders who
specialize in territory than on those who specialize in function and gives district commanders a
larger share of the department’s resources to control. Functionally specialized units—such as
patrol, detectives, school resource officers, and traffic officers—are placed under the command
of a district commander, or arrangements are made to facilitate their responsiveness to the
commander’s needs.

Results from our national survey suggested strong support for Compstat’s emphasis on

geographic organization of operational command. When we asked whether departments gave
authority to middle managers to select problem-solving strategies for low-level problems, 90
percent of Compstat-like departments claimed to have given this authority to district
commanders, line supervisors, or specialized unit commanders. This clearly demonstrated that
empowering middle managers with decision-making autonomy is an important feature of
Compstat departments (Weisburd et al. 2001, 34-35).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

30

In this section, we compare the organization and operation of operational command in the
three sites. We also assess to the extent to which district commanders had authority over
department sections and units that had traditionally operated independently of them. Our findings
demonstrate that decision-making authority had devolved furthest in Minneapolis—one level
below district commander to sector lieutenant. Furthermore, despite the extensive autonomy of
middle managers in the three sites, the nature of this decision-making autonomy was complex. In
all three departments, the chief and his deputies were willing to limit the authority of their
district commanders, but in Minneapolis the nature of decision-making authority was further
complicated by the relationship between district commanders and sector lieutenants, who both
had twenty-four-hour responsibility for geographic areas. Finally, the extent to which functional
differentiation had been replaced with “geographically based management” also varied among
departments (Silverman 1999,149). For example, none of the specialized units—those that
focused on specific crimes or vices and had a great deal of autonomy—in the Lowell and
Newark police departments were placed under the direct supervision of district commanders. In
contrast, district commanders in each of Minneapolis’ districts supervised a specialized
community response unit (CRT), a property crimes investigative unit, and a community crime
prevention or CCP/SAFE team.

Compared to the traditional model of policing, where command authority is centralized at

the top of the police organization, Compstat in the three departments had helped shift decision
making down the chain of command to middle managers. One way to quantify this change is to
consider the level or levels in which the organization vests responsibility for a space, whether it
is a beat, sector, or city. Before the advent of community policing, each of our departments was
organized temporally. A captain or lieutenant operated as a shift or watch commander and was
responsible for supervising officers on a single shift for the entire city; once the shift was over,
his or her responsibilities ceased. Thus, under the old system, the lowest-ranking officer in the
department with twenty-four-hour responsibility for a geographic area was located at the
command level—the deputy chief of operations. As a result of Lowell’s community-policing
program, this power devolved down one level to the district commanders, who were all captains
and had twenty-four-hour responsibility for their districts. The department’s Compstat program,
implemented two years after community policing in Lowell, increased the level of scrutiny and
accountability borne by district commanders. In other words, Lowell’s community-policing
program and its Compstat program had increased the decision-making authority of the district
commanders, but they had not shifted this authority any further down the organizational
structure. In contrast, the implementation of Compstat in the MPD and NPD had decentralized
decision-making power even further. During our time at Newark, one of the districts had been
selected as part of the department’s experimental Geographical Accountability Program (GAP)
and had been decentralized into three sectors, each headed by a lieutenant who had twenty-four-
hour responsibility for a sector. Since Newark’s attempt at greater decentralization was
experimental and limited to only one district, we do not focus on it here. The Minneapolis Police
Department had attempted to decentralize decision-making authority furthest within the
department. All, not just one, of its districts were subdivided into smaller geographic units or
sectors, each headed by a sector lieutenant. By extending decision-making authority beyond its
district commanders, Minneapolis had attempted to move decision-making power down an
additional level within the organization.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

31

Since Compstat encourages police departments to decentralize territorially and to
delegate greater responsibilities to the commanders of territories, we briefly describe each
department’s organizational structure before examining the nature and extent of district
commanders’ decision-making authority.

Organizational structure

In some respects the structure of all three police departments under Compstat remained
quite traditional. The police administration and most specialized units, such as criminal and
special investigations divisions, were centralized in headquarters. Nevertheless, the level of
geographic decentralization was still high since each department was partitioned into individual
districts. Lowell’s three (North, East, West), Minneapolis’ five (Second, Third, Fourth, Fifth and
Downtown Command), and Newark’s four districts (North, East, South, West) operated
relatively autonomously as command bases. For example, district commanders in all three
departments implemented strategies and directed operations from district substations.
Furthermore, district substations in Newark and Minneapolis held their own roll calls, though
this was not the case in the much smaller department of Lowell.

Decision-making authority

Decision making, like organizational structure, was significantly more decentralized in all
the departments than it had been under the traditional policing model. Before the widespread
introduction of community policing in the 1980s, police chiefs were primarily responsible for
making decisions about department policy and procedure without much input from middle
managers. Since all three departments had decentralized geographically as part of their shift
toward community policing, district commanders possessed significant decision-making
autonomy. Under Compstat, the MPD’s top management had taken this one step further by
subdividing its districts into smaller geographic units headed by lieutenants.

Despite these attempts to decentralize the decision-making process, the authority to make

decisions without supervisory approval had only moved down one or two levels in all three
departments. All the district commanders clearly believed that they possessed significant
autonomy in managing their own districts. They were ultimately responsible for choosing and
implementing crime strategies and rarely had to justify their decisions, although they tended to
report what they had implemented at Compstat. In Lowell, one district commander stated that he
very much felt, “like the captain of his own ship,” and our observations suggested that this
sentiment was also commonplace in the MPD and NPD.

Despite the extensive autonomy enjoyed by district commanders in Lowell and Newark

and, to a lesser extent, by sector lieutenants in Minneapolis, the exact nature of this autonomy
under Compstat was complex. Our national survey reported that 90 percent of departments with
Compstat-like programs gave district commanders, line supervisors, or specialized unit
commanders the authority to select problem-solving strategies for low-level problems, but this
percentage dropped to 70 percent when it came to authorizing middle managers to choose how to
deal with higher visibility problems (Weisburd et al. 2001). These results suggested that middle
managers under Compstat possessed a significant amount of decision-making autonomy, but the

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

32

chief and his deputies might sometimes curtail their authority. Responses to our national survey,
which examined the extent to which departments were willing to give middle managers greater
responsibility for determining beat boundaries or staffing levels, provided even less support for
the ideal of geographic organization of operational command. Only four in ten departments that
claimed to have implemented a Compstat-like program gave district commanders, line
supervisors, or specialized unit commanders the authority to determine beat boundaries. Even
fewer, only 18 percent, gave these commanders the authority to determine routine staffing levels
(Weisburd et al. 2001).

Our fieldwork supported this finding from our survey. District commanders did possess

significant autonomy, but headquarters still played an important role in how decisions were made
at the district level. This was especially true when it came to critical issues, such as the
distribution of departmental resources.

In Lowell, the district commanders were largely responsible for redrawing their beat

boundaries during the department’s bid process.9 During our research period, the district
commanders used their extensive familiarity with their sectors in redrawing cruiser and walking
routes. This reconfiguration involved the deputy superintendent of operations and district
commanders, with the chief rarely interfering with their decisions. This did not mean that he was
unwilling to overrule any decision that he felt was inappropriate. Under pressure from both the
community and city to increase the total number of walking routes, the chief overturned a district
commander’s decision to cut one officer’s walking time in half to four hours. The problem of
scarce resources in the face of increasing demand for beat officers contributed to the complexity
of this issue, and good arguments were presented on both sides. Similarly, officers walking the
beat were sufficiently important to Newark’s top brass that they made it almost impossible for a
district commander to cancel a walking post. Minneapolis’ top management also limited the
decision-making autonomy of the district commanders. Headquarters, not district commanders,
established the criteria determining the allocation of resources, such as assignment of additional
personnel to the districts. One district commander commented, “The department priorities are set
by the front office, so they usually shift the resources to deal with what they deem important.”
For instance, only the chief or deputy chief had the authority to move a district’s community
response team across districts. One district commander appeared unhappy about this form of
centralized decision making when he complained that Compstat issues of staffing and resources
“hadn’t kept up.” He suggested that the department needed “allocation meetings” in addition to
its regular pre-Compstat and Compstat meetings.

These scenarios illustrate that district commanders’ autonomy might have been extensive

but also operated within certain limits defined from above. Decisions about routine staffing
levels also revealed the persistence of the traditional hierarchical command structure. Should
district commanders require additional personnel, they did not have carte blanche to borrow from
elsewhere. Instead, it was customary to approach either the chief or his deputies for permission,
and they would try to be accommodating.

9 As part of their union contract, every eighteen months patrol officers were allowed to request reassignment to any
beat in the city. Priority was then given to those officers with the greatest seniority, or most years on the force.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

33

Like the district commanders, Minneapolis’ sector lieutenants played a more significant
role in the Compstat process than in Lowell and Newark. In the LPD and NPD, each district
commander was assigned a lieutenant or executive officer who provided command support. At
Lowell, the executive officer’s primary responsibilities were to read crime reports and convey
the district commander’s orders down the chain of command rather than act as a key decision
maker. In Newark, the executive officers played a similar role, but they also participated more
regularly and extensively in Compstat meetings. They were not expected to present data for their
districts unless the district commander was absent, but they were required to respond to the
chief’s questions. Furthermore, Newark’s executive officers attended their district’s pre-
Compstat meetings, where they helped prepare their district commanders for Compstat by
discussing individual cases, raising questions, and sharing ideas on crime problems and
strategies, as well as going over the crimes and patterns that would be shown on the maps at
Compstat. Unlike their counterparts in Newark, Lowell’s executive officers did not stand beside
their district commanders during formal Compstat presentations and were rarely called upon to
answer questions or provide suggestions. Given its size, Lowell did not have pre-Compstat
meetings, so the contribution of executive officers to Compstat was more informal and less
systematic. Despite these differences, executive officers in Lowell and Newark performed very
similar tasks in providing command and administrative assistance to their district commanders.

In contrast, Minneapolis’ executive officers, or sector lieutenants, were held considerably

more accountable than their counterparts in Lowell and Newark for whatever happened in their
beats. They were distinguished by the exercise of twenty-four-hour responsibility for a
geographic area. As a result of the variety of tasks and level of decision-making authority
commensurate with this position, the union had negotiated a 31 percent pay increase for the
sector lieutenants.

Despite this attempt to devolve decision making down the chain of command, the

autonomy of Minneapolis’ sector lieutenants was more restricted than the autonomy of its district
commanders. For example, district commanders, not sector lieutenants, were ultimately
responsible for overseeing operations and settling disputes within their districts. This is not to
suggest that sector lieutenants merely followed orders from above. They collaborated closely
with other lieutenants in the CRT and CCP/SAFE units. As a result, district commanders were
not required to supervise the investigative operations of these specialized units.

It is difficult to generalize about the role of the sector lieutenant since it varied according

to the particular personality of the officers who occupied the position and their relationship with
the district commander. This complexity was captured in our interviews. While the chief referred
to sector lieutenants as “the key rank” in the Compstat process, one district commander
described them as “tasking sergeants.” The difference would seem to rest upon differing
interpretations of the level of autonomy associated with this position. The chief’s
characterization implied that sector lieutenants operated as primary decision makers, while the
district commander’s remark appeared to emphasize their importance in providing invaluable
command support and conveying orders to the rank and file. The reality of the sector lieutenant’s
role probably lies somewhere between these positions. In some districts the sector lieutenants
operated with minimal supervision from their district commanders, while in others, the district
commander assumed a leadership role in advising and directing his or her sector lieutenants.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

34

Still, no matter what subtle differences existed between the levels of decision-making authority
exercised by sector lieutenants, district commanders remained ultimately responsible for
whatever decisions were made in their district. In the final instance, which was Compstat, the
buck stopped with them.

In addition to giving mid-level managers the authority to make decisions that once

belonged to top management, Compstat requires that functional differentiation, a conventional
feature of police organizations, be replaced with, or at least de-emphasized by, geographically
based management. Functionally specialized units, that focus on specific crimes or vices and
possess a great deal of autonomy, are supposed to be placed under the direct command of district
commanders or made responsive to their needs. In Lowell and Newark, Compstat had made a
very modest contribution to geographically based management. In these departments, some
detectives were assigned to the districts, where they fell under a district commander’s direct
command; but nearly all the specialized units, including the central investigations or detective
units, continued to operate physically out of headquarters and autonomously from the district
commanders. Aside from the reassignment of some detectives to the districts, it was clear that
the supervision of specialized units—such as gang, homicide, domestic violence, and community
response—had not been transferred to district commanders as a result of Compstat.

In Minneapolis, however, a greater shift toward geographic management had been

achieved. Each district commander was responsible for his or her own property crimes
investigation unit, CRT, and CCP/SAFE unit. Each district’s property crimes investigations unit
handled property crimes, including robbery, in its geographic area; and each district possessed
resources, such as fingerprinting equipment, that had tended only to be available at central
headquarters. Furthermore, each district commander directed two community response teams—
each composed of eight patrol officers supervised by one sergeant—that were assigned to either
the day or night shift. Since they were not held responsible for answering 911 calls, they could
concentrate all their efforts on tackling prostitution, narcotics offenses, and other quality-of-life
crimes. Finally, the CCP/SAFE units assigned to each district were responsible for problem
solving and community building. Formerly operating out of central headquarters, they were
decentralized under Compstat to foster closer collaboration between the police and local
community in addressing crime problems.

The nature of decision-making autonomy and the level of functional specialization may

have varied among the departments, but geographic organization of operational command had
contributed to two internal organizational challenges that were common to all three: (1) the
complexity of coordinating tasks both between central units and districts and across districts and
shifts, and (2) the disproportionate burden of responsibility placed on middle managers in an
organization that is structured both geographically and temporally.

Coordination issues

We noted earlier that these organizations were considerably more decentralized in terms
of both territory and decision making than more traditional departments. For example,
investigations were no longer the sole responsibility of a centralized detective unit operating out
of department headquarters. In all of the sites we visited, some central detectives had been

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

35

reassigned to the individual districts. These detectives were responsible for investigating all the
property crimes— both burglary and theft—and quality-of-life crimes that occurred within their
individual districts, and they operated under the direct supervision of their district commander.
Investigations of the most serious crimes—homicide, organized crime, and sex crime—remained
the responsibility, however, of each department’s centralized Criminal Investigations Division
(CID). Unlike the detectives operating out of the districts, central detectives answered to their
own commanding officer, the officer in charge of the CID. Obviously, the members of any
organization that is structured both geographically and functionally will need to find ways to
share information and coordinate their crime responses. In our three sites, Compstat meetings
facilitated information sharing and coordination among districts and also between districts and
centralized units. Minneapolis’ chief clearly felt that this was a useful function of the
department’s Compstat program. He surmised that the presentations of the department’s
centralized units at Compstat gave the impression “of [their] being available to precincts.”

Compstat may go some way toward fostering an impression of availability, but it is
clearly insufficient as a structure. To further facilitate information exchange, district
commanders, district detectives, members of the CID, or the various heads of specialized units
attended a variety of regular command-staff meetings. Nevertheless, given the sheer volume of
information generated in modern police departments, these meetings still did not suffice.
Consequently, each department supplemented these more formal mechanisms by relying heavily
upon individuals to take the initiative to contact and communicate with one another largely via
phone or email. It was customary, for example, for Newark’s district commanders to have
informal conversations to decide which district was responsible for dealing with a crime problem
in the rare event that a crime occurred on the boundary of two districts. A possible resolution
would be an informal agreement whereby each district commander committed the same amount
of resources, a patrol car for example, to deal with the problem; the chief might also choose to
intervene and assign responsibility.

In sum, despite the limitations of regular Compstat meetings as a coordinating

mechanism, the level of accountability experienced by district commanders under Compstat did
seem to have a cultural benefit. Our observations suggested that decision makers felt encouraged
or compelled to undertake self-initiated and less public communications. Should a problem fall
between the cracks, they risked censure from above.

Obviously under these circumstances the failure of an individual to take initiative could

lead to organizational dissonance. We heard gripes in all three departments about the problems
involved in sharing information between specialized units and the districts, especially that
surrounding the investigation process. At Minneapolis’ annual command meeting to evaluate
Compstat’s operation within the department, a district commander complained that the
department’s gang unit, which was located within the centralized Special Investigations
Division, did not share information about what it was doing with the districts. He noted, “That
sort of information would be helpful to me and other district commanders.” At the same meeting,
a member of the top brass remarked that, “They [the gang unit] are reluctant to share too much
because they feel that they do secret stuff.” Some respondents in Lowell similarly commented
that tension between the central and district detectives could have hampered coordination and
communication within the organization. Newark had sought to overcome some of these problems

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

36

by incorporating a more frequent and systematic communication structure into the organization.
A single central detective was assigned to each district to transmit information to and from
headquarters. This liaison was responsible for attending district-level Compstat meetings, in
addition to the regular meetings that took place four times per week between the CID and district
detectives. Despite this mechanism, however, concerns surrounding information sharing and
effective communication between the CID and the districts were frequently raised at Newark’s
department-level Compstat meetings.

Geographic versus temporal organization

The feature that contributes most significantly to coordination problems is the uneasy
relationship among members of an organization that must operate both geographically and
temporally. District commanders were held directly accountable for whatever happened in their
districts over a twenty-four-hour period, although they were only on duty for eight hours during
the day shift. Therefore, each department had to make arrangements to ensure that a district
commander’s orders were carried out while he or she was off-duty. In Lowell and Newark, this
might be the responsibility of a lieutenant or desk sergeant, but the district commander was still
ultimately responsible for making sure his or her decisions were conveyed across shifts and
actually implemented at the level of the street. In Minneapolis, the sector lieutenants shared this
responsibility, since they exercised twenty-four-hour responsibility for everything that occurred
in their areas of the city.

These coordination problems return us to the challenges associated with accountability.
Some middle managers felt that being held responsible for everything that happened in their
district or sector, even while off-duty, was demanding and somewhat unfair. In Lowell, the
district commanders were almost entirely responsible for crime in the city and had to answer to
the chief during Compstat. This, in effect, shifted a major part of the burden of managing the
police organization onto the shoulders of only three people (Willis et al. 2004b). One of Lowell’s
district commanders noted that the current system let everyone else “off the hook,” while another
respondent noted that some did not worry about conveying command decisions to the lower
ranks because “the heat falls on the district commanders.” Similar reservations were expressed
by some of Minneapolis’ sector lieutenants. One expressed concern about the lack of supervision
in a sector when its sector lieutenant was not on duty. He added, “People need direction, and
when a sector lieutenant is not around during each shift sometimes there may be issues.” In a
separate interview at the same department, a district commander offered a similar opinion when
he stated, “They [sector lieutenants] can’t be here twenty-four hours a day.”

To sum up, the geographic operation of operational command in all three departments
under Compstat did provide middle managers with considerable authority. In general, just a few
individuals had primary responsibility for controlling crime in the city, but this responsibility
had devolved furthest down the organizational hierarchy in Minneapolis. To support them in
their endeavors, middle managers possessed significant decision-making autonomy. In
comparison, Compstat’s impact upon the overall structure of these departments was more varied.
In Lowell and Newark, specialized units were theoretically available to district commanders but
did not fall under their direct supervision. District commanders continued to be responsible only
for patrol officers. In contrast, the level or dosage of geographic organization of operational

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

37

command was considerably higher in Minneapolis. District commanders were given a
significantly larger share of the department’s resources, since they were assigned several
specialized units in addition to the patrol officers in their districts.

Organizational Flexibility

Having the authority to respond to problems is a necessary but insufficient condition for

middle managers to implement their decisions effectively. They must also have adequate
resources. Compstat promises a rapid and effective response to incipient problems. Since these
problems are unpredictable, the organization needs to be flexible in its allocation of resources.
Police organizations, however, are usually designed in ways that restrict their capacity for
flexibility. First, they are highly bureaucratic and operate according to detailed rules and
regulations, many of which are imposed or embraced by police administrators themselves. This
makes departments much more successful at performing routine tasks in response to highly
predictable work demands than at developing the capacity for coping with the unexpected.
Second, administrators’ ability to make their organizations flexible is constrained by other forces
inside the organization, especially those concerned with employees’ rights and management’s
responsibility to negotiate for changes from organizational routines (Mastrofski 2002). For
example, the ability of middle managers to change work shifts and job assignments tends to be
constrained by inflexible regulations imposed by labor union contracts. Third, the department
experiences pressure from forces outside the organization—politicians, businesses, the press,
neighborhood associations, and a host of other interest groups—that want to determine just how
and where officers will spend their time. Sometimes these pressures reinforce and support what
management believes should be the department’s priorities, but often they constrain what
management may do in coping with unexpected problems and work demands.

One of the ways that police organizations have traditionally coped with unpredictable
fluctuations in work demand is to overstaff their patrol units relative to the demand they
expected on a given work shift (Reiss 1992). That is, they built in slack time during which
officers engaged in preventive patrol, though their principal function was really to be available
for work that might arise, especially that which might require an emergency response. Typically,
police departments were encouraged to build in sufficient slack to cope with significant
catastrophes that might come up. O. W. Wilson advocated systematic incorporation of this
flexible capacity into the calculation of staffing needs (Wilson and McLaren 1972). For many
years this approach was widely practiced among police departments and accepted by those who
approved their budgets, but the ethos of a more efficient, results-oriented policing has changed
all that. Community policing, problem-oriented policing, and Compstat all call for officers to
engage in more directed activities that focus on preventing and solving problems, spending time
with the community, and engaging in crime-reduction—all of which work to diminish the slack
time available through preventive patrol.

In this even more constrained environment, how can police departments become more
flexible? They can attempt to negotiate with collective bargaining units for greater flexibility in
making job and shift assignments. They can budget for overtime, which allows managers to work
around the constraints of set work shifts. They can create special units, or taxi squads, that are
not bound to a particular geographic area or work shift and can be flexible about when and where

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

38

they work. They can require or reward cooperation and teamwork within and across
organizational units. Ultimately, the flexibility of the police organization is displayed when
managers are able and willing to alter allocation structures and routines in response to non-
routine work demands.

In our national survey, organizational flexibility was measured by a department’s
response to two questions: (1) To what extent had a department reallocated resources, such as
reassignment and overtime, to the crime/disorder problem that had required the most effort in the
last twelve months?; and (2) To what extent did middle managers have general authority to
approve requests for flexible hours or mobilize SWAT units to specific operations? Although
these two items also reflected the commitment of the department to geographic organization of
operational command, they did focus directly on the extent of flexibility in the allocation of
departmental resources. In the national survey, departments that claimed to have implemented a
Compstat-like program did appear to allow a great deal of organizational flexibility. Eighty-five
percent of these departments had reassigned patrol officers to new units, areas, or work shifts to
address this primary problem. Eighty-one percent of departments had used overtime to provide
personnel to deal with the problem (Weisburd et al. 2001).

There was significant variation in the level of organizational flexibility among the
departments we observed. In Lowell, the most common way of reallocating resources outside of
normal patterns was for district commanders to do so on an ad hoc or informal basis, which
minimized disruptions to department routine. The most likely occurrence was for a district
commander to take a patrol officer or detective away from the more routine activities of their
shift and ask them to pay particular attention to a specific area. The chief might also do this. For
example, in response to the ongoing problem of motor vehicle crimes in one district of the city,
Lowell’s police chief requested that a central detective be assigned to work more closely with the
CAU. In addition, the LPD, if necessary, would pay officers overtime as part of a crime-
reduction strategy. For example, the chief responded to the brutal mugging of a woman in the
highly visible downtown area by authorizing overtime funds for additional patrol over a two-
week period.

This ad hoc reallocation of resources was also a common feature in Newark and
Minneapolis, but these departments had taken greater strides to increase organizational flexibility
by relying more on task forces and taxi squads. These departments’ greater capacity for
organizational flexibility probably reflects their size. Since they were larger than the LPD, they
had more resources available for allocation to emerging crime problems. In Newark, the
department was constantly forming and disbanding special task forces in response to major
problems that emerged. These task forces were directed by the chief and his deputies, rather the
district commanders, and operated out of an individual district or central headquarters. However,
within-district flexibility was greatest in Minneapolis. Each district was permanently assigned a
CRT that did not have a specific geographic assignment and was available to the district
commander to use when and where she or he pleased in the district. Freed from responding to
911 calls, the CRT provided all district commanders with a valuable mechanism for directing
resources quickly and efficiently to non-routine problems as they arose.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

39

Lowell, Minneapolis, and Newark might have differed significantly in the level of
organizational flexibility within their separate districts, but organizational flexibility among the
districts was underdeveloped in all three departments. Ironically, this could be attributed to two
features of Compstat: (1) Compstat’s commitment to increasing geographic organization of
operational command by increasing the allocation of resources to each district commander, or
within-district flexibility, simultaneously decreased the organization’s capacity to shift resources
among districts; and (2) Compstat’s emphasis on internal accountability fostered competition
that limited teamwork among districts.

Finally, organizational flexibility in all three departments was generally limited by
traditional internal and external challenges that included lack of manpower, conventional
approaches to resource allocation, and city politics. These challenges are very difficult to
overcome, but we observed a variety of attempts to reduce the constraints imposed by tight
budgets and insufficient manpower.

We noted above that a common way of reallocating resources within districts was to
direct a patrol officer or detective to pay particular attention to a specific problem area. Compstat
increased this capacity and put more resources at the disposal of district commanders by
encouraging departments to assign a handful of detectives to each district. Another example of
organizational flexibility was the temporary assignment of a specialist to a crime problem. In
Lowell, for example, the decision was made at a command staff meeting to move an assistant
crime analyst temporarily from the department’s CAU to the Criminal Investigations Division
(CID). In order to provide central detectives with greater access to gang information, the crime
analyst was relocated and assigned specific gang-related cases. The crime analyst stated that
there was a need for someone in the detective division to “go over patterns” relating to specific
crimes. According to our national survey, nearly three-quarters of departments responded that
they allowed district commanders, line supervisors, or specialized unit commanders to decide on
flexible hour requests, and 65 percent allowed them to mobilize SWAT teams (Weisburd et al.
2001). In line with this trend, Lowell’s district commanders were responsible for making these
kinds of decisions with little or no review by superior officers.

Between-district flexibility can be enhanced by the creation of department-level task
forces. These were the norm in Newark, where the chief formed task forces quickly and
frequently in order to increase the organization’s capacity to respond to problems as they arose.
In contrast, task forces in the MPD were organized less frequently, perhaps a handful per year,
and were usually the outcome of a lengthy planning process. Task forces were very rare in
Lowell and were reserved for the most intractable problems. Newark’s task forces were usually
composed of the heads of different specialized units and officers who had reputations for being
effective problem solvers. Often these officers would be poached from their districts to work on
task forces. As a result, the department was in a constant state of flux, with personnel constantly
being reassigned. At any given moment a host of task forces would be working on a variety of
problems.

Department-level task forces implemented on an ad hoc basis were a rarity in
Minneapolis since the department had built flexibility into its organizational structure. It tried to
do this by assigning community response teams to the districts and by implementing district-

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

40

level directed patrol. Community response teams provided district commanders with “fluidity of
resources,” as one officer put it. Since these teams were not required to respond to 911 calls,
unless specifically requested to do so, the district commander could assign them to tackle
problems as soon as they had been identified. A similarly 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, or quality-of-life, crimes.” Given the high priority accorded directed
patrol within the department, district commanders were less constrained by the “tyranny of 911”
(Sparrow, Moore, and Kennedy 1990). Consequently, they were able to allocate more patrol
resources to proactive and preventive strategies than under more traditional policing models. Our
observations suggested that many department members were pleased with this approach. One
sector lieutenant offered the following salutary comment, “You need an effective group of
flexible officers. We have our CRT and directed patrol units that focus on things other than calls
for service. These officers are necessary and really add to the amount of work that can get done
in our precinct.”

Finally, top and middle management’s commitment to community policing in Lowell
helps explain the lack of support for task forces within the department. The chief and his
command staff associated task forces with a more traditional style of policing and characterized
them as heavy-handed. The chief remarked that task forces often “stormed” into a problem “like
the Marines” and suggested that they often made the situation worse. He added that beat officers,
who were much more familiar with community members and their problems, were much better
equipped to solve any issues that arose. This perspective helps account for why we did not
witness the formation of any task forces during our fieldwork at Lowell. Six months after we left,
however, one was eventually formed to help tackle the ongoing problem of motor vehicle crimes
in one district.

Rivalry among districts

Unfortunately, the allocation of additional resources to individual districts simultaneously
reduces an organization’s capacity to shift resources across districts. This problem is exacerbated
by the lack of teamwork among districts fueled by Compstat’s emphasis on the accountability of
individual district commanders for their own districts. Since the Minneapolis Police Department
had allocated the greatest share of its resources to the districts, we would expect it to have the
fewest resources to share among them. Our observations supported this hypothesis; on balance,
we estimated that we heard more complaints about this feature in Minneapolis than in Newark
and Lowell. One officer complained that Compstat had led to, “the creation of a bunch of
specialty units like CRT and SAFE.” These units reduced the resources available for other
activities, leading the same officer to remark, “It’s hard to do anything extra when there are so
few people to get the required work done.” Our observations suggested his point had some merit.
During our fieldwork, Minneapolis was the only department that was hampered by a lack of
resources in its attempt to create a cross-district task force. Formed in an attempt to catch a
smash-and-grab burglar who was active in different areas of the city, the task force was
composed of officers from two districts. An important factor contributing to the task force’s
failure in apprehending the burglar was the lack of officers available for between-sector

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

41

collaborations. The lieutenant who arranged the special details commented, “[If] we would’ve
had ten more cops we would’ve had him.”

Obviously, limited within-district resources are likely to give rise to complaints similar to
those above. Not only will district commanders be unable to loan officers to task forces, as in the
burglar scenario, but they may also lack enough resources for an adequate response to problems
within their own sectors. In Lowell, where within-district resources were limited due to
department size, we heard frequent complaints about the need for more. For example, the lack of
available manpower made it very difficult for a district commander to arrange for the regular
deployment of a decoy car to catch perpetrators of motor vehicle crimes. He felt that the problem
of scare resources was particularly acute in his district, since he was experiencing a
disproportionate amount of motor vehicle and gang crime.

One method to improve organizational flexibility would be to facilitate teamwork and

resource sharing among the districts. Unfortunately, Compstat’s emphasis on accountability
encouraged competition, not collaboration, among those officers best equipped to foster inter-
district cooperation, district commanders.

One way competition was promoted was by the constant attention paid to each district’s

crime figures. Approximately 80 percent of the patrol officers we surveyed in each department
“strongly agreed” or “agreed” that Compstat had made supervisors place too much emphasis on
statistics (see Table 5). Moreover, none of those surveyed in Lowell and Minneapolis, and only 3
percent in Newark, “strongly disagreed” with this statement.

Table 5: In your view, Compstat has made supervisors place too much emphasis on

statistics

Compstat has made supervisors place too much emphasis on statistics*
% Strongly

agree
% Agree % Disagree % Strongly

disagree
No opinion

Lowell 51 31 9 0 9
Minneapolis 51 37 10 0 2
Newark 59 27 5 3 6

Chi Square DF Significance N
19.047 8 *p<0.05 432

A similarly large proportion (approximately 70 percent) believed that Compstat had done

very little to increase teamwork within the department and “disagreed” or “strongly disagreed”
with the statement, “Compstat has increased teamwork between my unit and specialized units in
the department” (see Table 6). The frequent formation of task forces in Newark might explain
why patrol officers from this department were significantly more likely (23 percent) to associate
Compstat with an increase in collaboration. Although the statement did not specifically address
inter-district cooperation, the many responses indicating lack of teamwork also suggested that
Compstat was not particularly helpful at fostering resource sharing within the organization.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

42

Table 6: In your view, Compstat has increased teamwork between your unit and
specialized units in the department

Compstat has increased teamwork between my unit and specialized units in the

department*
% Strongly

agree
% Agree % Disagree % Strongly

disagree
No opinion

Lowell 2 13 38 33 14
Minneapolis** 0 15 46 25 13
Newark 4 19 33 32 12

Chi Square DF Significance N
11.711 8 *p<0.05 432
** Does not sum to 100 percent due to rounding

Statements from the command staff we interviewed at the three sites supported these
survey results. In Minneapolis, one mid-level manager commented that the high standard of
accountability associated with Compstat might have caused district commanders to feel, “My
district is more important than the department as a whole.” He continued by providing a scenario
in which his district had been provided with overtime money and managed to mobilize
community support to tackle a homicide problem but still required additional manpower.
Unfortunately, “There were no personnel [available] from the other districts.” One of
Minneapolis’ district commanders stated: “In theory if [X, a district commander from another
district] needed more resources, he could call me directly and work something out. But, in
reality, district commanders are becoming a little greedy and don’t want to give up people to
help with another area’s problems when they could be focusing on their assigned area’s
problems.” Several sector lieutenants voiced similar concerns indicating that the district system
under Compstat fostered competition and made them reluctant to share resources with other
sectors. We note that some felt that this limitation was outweighed by other considerations. For
example, Minneapolis’ top leadership argued that competition resulting from Compstat was
beneficial in helping to improve overall department performance.

Competition among Lowell’s district commanders seemed less acute, and perhaps this

could be attributed to the absence of a goal for a specific percentage reduction in crime.
Nevertheless, since Compstat inevitably produced cross-sector comparisons of crime statistics, it
still helped foster rivalries. Our observations suggested that competition among the district
commanders over crime numbers was relatively subdued. Most often they expressed solidarity
and sympathy when a crime spike occurred in another person’s sector. A sensitivity to
competition was not, however, entirely absent. On one occasion a respondent said that he was
slightly uncomfortable that detective follow-ups and arrests had been displayed at a Compstat
meeting; this was not the norm. The respondent felt that this might have reflected negatively and
unjustly on the hard work that was being carried out in sectors that were understaffed at the time
due to injuries. Any measure that might reduce a district commander’s ability to curb crime
would be met with some resistance, and this perception was also present further down the ranks.
One of Lowell’s sergeants commented that the unwillingness to share “route men” between
districts was, “a turf thing.” “They [the district commanders] don’t go into each other’s turf.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

43

They want their own officers to ride around in their own district.” In response to a question
regarding the problem of understaffing and the possibility of borrowing officers from another
sector to deal with a pressing crime problem, a sergeant said, “It could happen … but everyone in
their district has their own problems they are dealing with.” He noted that it used to be easy to
get the people you needed [when the department was centralized], but not under the district
system. “With district commanders, people think of them as their own little fiefdoms and they
don’t want to let people go.” This sense of territoriality was also present in Minneapolis, where a
patrol officer told us that Compstat made officers “stay in their area more.”

Competition reduces the organization’s capacity to shift resources to where they are

needed most urgently. Three other external and internal organizational factors impose additional
limitations on organizational flexibility: lack of manpower, traditional approaches to resource
allocation, and city politics. We address these factors below before discussing some of the
strategies that were used to reduce their constraining effects.

Manpower

The paradox of resource flexibility is that the fewer resources the organization has to
devote to flexible assignments, the more it needs that flexibility. A common complaint across
departments was the need for additional patrol officers, or “more troops on the street,” as one
Lowell district commander phrased it. This sentiment appeared to be widely shared in Lowell
and Minneapolis, especially among the district commanders. Silverman writes, “To its credit,
Compstat is not tolerant when precinct commanders automatically respond ‘we need more
people’ to questions about below-par crime fighting”(1999, 202). That said, several respondents
did seem to feel that such a response was justified when discussing how to tackle crime
increases. One sergeant in Lowell believed that the department could use more patrol officers.
He queried the logic of keeping patrol officers in administrative or specialty positions, such as
computing or training, when they could be used more effectively on the street. Another line
supervisor in the same department opined that his officers felt they were understaffed but added,
“The administration doesn’t want to hear this.” Similarly, a sector lieutenant in Minneapolis
commented, “… The bottom line is that there is a limitation on resources.”

Compstat may not be particularly sympathetic to claims for more officers on the street.

By constantly bringing attention to crime problems and the need for rapid and effective
solutions, however, it also brings a sharper focus to the old bugbear of the police organization,
the lack of manpower. Since middle managers are held most accountable for crime problems in
their sectors, they are the main proponents of increasing personnel as one of the most effective
strategies for reducing crime.

Traditional approaches to resource allocation

One way to allay some of the problems associated with limited resources would be to use
Compstat as a mechanism for more effective allocation of personnel, overtime funds, and
equipment among districts. In this way, crime problems could be used to drive decision making
at the level of the department, just as district commanders use data to allocate resources within
their own districts. Two things would be necessary for this to work effectively: (1) Resource

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

44

allocation would need to depend upon crime data rather than more conventional considerations
about overall equity of work load; and (2) District commanders would need to play an important
role in deciding how resources should be shared among districts. We did not witness either of
these more innovative approaches to resource allocation in Lowell, Minneapolis, or Newark, but
the chiefs of Lowell and Newark had tried to increase resources through means that were less
disruptive to department routines. In Lowell, the chief hoped that identifying and preventing
officer absenteeism would help relieve manpower constraints, and in Newark the chief tried to
achieve a similar effect by ordering all officers who were normally assigned to administrative
duties to spend one day per week on patrol.

In Lowell and Newark, department resources were roughly divided among districts in

proportion to their population and/or the area they covered; while in Minneapolis multiple
measures—such as calls for service, reported crime, and 911 calls—were used in allocating staff.
Though Compstat is composed of crime data, Minneapolis was the only department to make
some use of Compstat data to draw beat boundaries. None of the departments, moreover, relied
upon the range of Compstat data—maps, spreadsheets, and breakdowns according to crime
categories—to habitually assess and drive the resource allocation process across districts. At
least one respondent was frustrated with the reliance upon more traditional approaches,
complaining that resource decisions should be based on Compstat crime data not population size.
For example, during a bid/reassignment meeting attended by a deputy superintendent and the
district commanders, one Lowell district commander queried why he was not being afforded an
increase in the department resources. After all, he stated, his district, which was one of the
poorest in the city, was experiencing a disproportionate amount of crime, and this was clearly on
display during the department’s biweekly Compstat meetings. When he suggested that Compstat
be used more “scientifically” as a means of allocating manpower to high-crime areas, he
provoked little response. The other district commanders in the room were largely unsupportive,
as we would expect, given that they were under-resourced and accountable for crime in their own
areas of the city. Any increase in the resources available to one district commander would
necessarily result in a decrease for the others. One district commander noted another flaw in the
concept of basing manpower on Compstat: “politics.” Given the limited supply of manpower and
overtime pay, the need to appease powerful interests by deploying police officers to certain areas
of the city competed against the flexible redeployment of officers to problem areas, as we discuss
later under “City politics.” Another explanation for these reservations was that district
commanders believed that long-term crime rates were roughly comparable across districts,
despite the upsurge in this one district’s crime rate, and they consequently felt that resource
allocations should not be changed. This is not a particularly convincing argument, however,
since the sine qua non of Compstat is the rapid allocation of resources in response to specific
short-term or emerging problems. Even if annual motor vehicle thefts were distributed evenly
among districts, a short-term crime spike in one district would warrant a shifting of resources.
After all, Compstat is intended to specifically replace annual crime statistics and other more
traditional methods as the most important means of reallocating resources for a targeted police
response.

We can attribute this reluctance to reallocate resources to districts with the most urgent

crime problems to traditional constraints and an additional factor that was both strongly
entrenched and closely related to how Compstat operated: conceptions of fairness. By holding

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

45

district commanders directly accountable for crime in their beats, Compstat may work to
undermine the rational, yet discriminatory, allocation of resources.

Bureaucratic rules and regulations

The police have long relied on crime statistics, especially those dealing with calls for
service, to help organize and allocate patrol to when and where it is needed most, but the
flexibility that this implies has been offset by departmental rules and regulations that establish
department routines. Union regulations, for example, often restrict organizational flexibility by
limiting changes to officers’ shift assignments. The limitations of this system are exposed under
Compstat when crime data are updated on a daily or weekly basis. Changing shift assignments
according to annual, or even semiannual, crime statistics is obviously less disruptive to
department routines than reallocation of resources across districts on the basis of more timely
data. The latter would demand a major change in how departments currently operate.

Conceptions of fairness

In addition to the restrictions commensurate with bureaucratic rules and regulations,
notions of fairness further limited organizational flexibility. Two Compstat features—geographic
organization of operational command and internal accountability—encouraged equitable rather
than scientific distribution of resources across districts. We observed that at least one district
commander would have preferred the reallocation of resources based upon Compstat. This
system did not have unanimous support, however, although it might appear to be fairer than
allocating resources according to population size. By failing to advocate change, some showed
their feeling that use of crime data was inappropriate. Of course, most of the police workload—at
least that experienced by patrol officers—is characteristically not conceived as crime work
(Greene and Klockars 1981), even in departments that make crime reduction the top priority.
Organizational efficiency, as well as fairness in the distribution of the workload among workers,
calls for an effort to make the allocation of resources proportionate to that workload. Where the
distribution of crime and the larger workload are not highly correlated, inequities and
inefficiencies in the distribution of resources are inevitable. Police officers are highly sensitive to
workload inequities (Rubinstein 1973; Van Maanen 1974), creating a strong pressure up the
chain of command to allocate resources in ways likely to sustain long-run equality, not short-run
flexibility.

This tendency to inertia in how the police organization allocates its resources was

reinforced by district commanders’ perceptions of the status of their commands. Since all district
commanders were under-resourced, they did not see why they should relinquish their resources
to others while still being held to the same high standard of accountability. Irrational as this
consideration might seem, geographic organization of operational command combined with
internal accountability to produce a notion of fairness based upon the equitable allocation of
resources among districts. This superseded concerns about changes in the crime rate. We
conjecture that an important component of this sense of fairness might be a more general
impression among district commanders that any district that relinquished any of its patrol
resources would inevitably experience a crime upswing. It might have been that this belief—
reinforced by a more general set of shared values about occupational solidarity and the necessity

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

46

of maximizing patrol resources—worked to undermine resource sharing. Our sense was that
district commanders felt it would be unfair to ask or expect one of their peers to share his or her
resources lest they suffer adverse consequences. Of course, a more fundamental and simpler
bureaucratic imperative might have been at work here. Since a district commander’s worth,
status, and power was directly linked to the resources he or she controlled, giving up resources
was akin to giving up these privileges.

City politics

Organizational flexibility is not only limited by internal constraints, forces outside the
department may also limit the flexible reallocation of resources. City politics can reduce
organizational flexibility by influencing officer deployment.

In Lowell, for example, a city councilor expressed concern about a spate of drug-related
crimes in one particular area of the city. He argued that there was a need for increased police
visibility in this neighborhood and he told the chief and other city councilors that he wanted to
see “more cops walking.” Lowell’s police chief was a staunch advocate of community policing
and foot patrol, but he responded that placing an even higher priority on walking routes would
reduce the number of available cruisers. Consequently, some areas of the city would receive less
patrol. The only way to avoid this problem would be to increase the city’s expenditures by hiring
more officers or increasing overtime. Since the city council was unable or unwilling to support
either one of these measures, the department’s only alternative was to come up with a plan
entitled “Advancing Community Policing” that gave priority to assigning patrol officers to all the
city’s walking routes before filling cruiser route assignments. In addition to the city council’s
position on foot patrol, influences on the police department also included local business owners
and the editor of the local newspaper, who was a powerful political figure in Lowell. They were
also keenly in favor of the department maintaining a high level of police visibility in the
downtown area where their businesses were located. In response to this political pressure, the
chief required his district commanders to maintain foot patrols in the downtown area on the day
and early night shifts. In addition, according to an agreement with the city, the chief was
committed to assigning several officers to the city Housing Authority. Similarly in Minneapolis,
the district that served the downtown business area was allocated a level of resources comparable
to that of more crime-ridden districts. We can speculate that this was on account of pressures on
the department from businesses, politicians, and other special interests. Although we did not
observe this directly, one district commander was clearly disappointed that the reallocation of
resources was so dependent upon these concerns. He told us he could not understand why these
types of decision relied more on “taxpayers as opposed to victims or people who live on a block
with drug dealers.”

The prioritization of patrol in certain areas placed considerable restrictions on the ability
of district commanders to allocate resources and select tactics. A Lowell sergeant whose territory
covered the downtown area noted how difficult it was at times for him to fill his routes. That
weekend, he had been unable to fill his cruisers because four officers had called in sick and he
had to keep “two walkers” in areas designated as a high priority by the chief and city.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

47

Strategies to increase flexibility

City politics exert a powerful influence on resource allocation in police departments, but
there are some plausible ways to free up additional patrol resources with minimal disruption to
department routines. Lowell’s chief was in the process of creating a database to record excused
and unexcused officer absences, as well as catalog citizen complaints. He hoped that this would
serve two useful purposes: (1) It would act as an early warning system for the identification of
problem officers; and (2) It would help district commanders identify officers who were
chronically absent. The increased likelihood that this system would recognize or flag officers
who were abusing their sick leave or feigning injury, the chief surmised, would help deter such
instances from occurring in the first place. He obviously intended this as a way to increase the
number of available officers.

Lowell’s chief was not keen on pulling specialists from their key positions because he felt
they fulfilled a necessary purpose where they were, according to one respondent. Newark’s chief,
however, was less opposed to this approach, and we witnessed a variant on this theme. He did
not remove specialists from their positions, but he insisted that every officer assigned to
administrative or desk duty spend one day per week on patrol. In this way, he was able to make
additional resources available where they were needed most in geographic terms, since patrol is
organized geographically.

Finally, a department can encourage teamwork in order to try to overcome the
competition fostered by Compstat. Top leadership in Lowell and Minneapolis emphasized the
importance of collaboration within their departments, but teamwork rarely occurred. When it did
take place, it usually reflected the involvement of headquarters. Task forces were emblematic of
top management’s recognition that teamwork was unlikely to occur on a voluntary basis. When a
crime problem was diffused across separate districts, headquarters often felt compelled to order
the formation of a task force to resolve the issue.

Why did these departments pay lip service to teamwork but were unable to provide
tangible rewards? Part of the explanation lies in how Compstat operates as a system. Compstat
explicitly rewards district commanders for handling their own beat problems effectively, but it
does not contain a similar mechanism for rewarding more efficient sharing of resources—even if
this could contribute to an overall reduction in crime. Without a structure that specifically
recognizes and rewards district commanders for voluntarily sharing valuable resources and
collaborating with other precincts, it is unlikely that a small police department can maximize its
organizational flexibility.

In sum, traditional internal and external constraints limited the capacity of the three
departments to reallocate resources to where they were needed most. Furthermore, two Compstat
elements—geographic organization of operational command and internal accountability—
combined to constrain organizational flexibility by fostering competition not teamwork among
districts. Top management sought to increase flexibility by: (1) maintaining its authority to
allocate resources across districts; and (2) ordering the formation of task forces. Ironically, in
concentrating decision-making power at the top of the organizational hierarchy, both these
measures undercut the ideal of delegating operational command to district commanders.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

48

Furthermore, it is likely that the allocation of resources to task forces would require reductions in
the resources available to district commanders, thereby further reducing the organization’s
commitment to geographic organization of operational command.

Data-Driven Problem Identification and Assessment

Crime statistics are crucial to Compstat as a new way to structure information for
managerial decision making, inasmuch as they shift the focus from highly selective, anecdotal
accounts of individual cases to the search for larger patterns and trends. Results from our
national survey indicated that Compstat departments have the capacity to manage and analyze
data in more sophisticated ways than non-Compstat departments. Over 90 percent of Compstat
departments claimed to conduct “crime-trend identification and analysis,” and almost 90 percent
claimed to use “database or statistical analysis software for crime analysis” (Weisburd et al.
2001, 44). Lowell, Minneapolis, and Newark were typical in this regard. Each department’s
Compstat program was designed to provide useful information for decision making in handling
crime/safety priorities. As one officer put it at Lowell, “Information is always the name of the
game.”

In this section we examine how crime data were routinely collected, analyzed, and
presented. Our discussion highlights the specific challenges associated with checking data for
accuracy and deciding which to select for Compstat. We also compare how each department
used crime data to learn about and tackle problems. Our findings suggest that decision making
was concentrated at the command level and was informed, but not driven, by data analysis.
Timely maps and reports helped middle managers identify crime problems as they arose, but
district commanders continued to rely upon their personal experience and anecdotal evidence to
assess and respond to problems. Under pressure to respond swiftly and decisively to any crime
increase, district commanders did not monitor hot spots continuously. Nor did they examine
crime data systematically, in order to discern the underlying causes of crime patterns or the
reasons for the success or failure of a particular strategy. The Compstat process, therefore,
precipitated a form of reactive policing denounced by advocates of strategic problem solving.

Data collection

All three departments depended upon police incident and arrest reports for their crime
data, but they also used computer-aided dispatch (CAD) data to help identify geographic hot
spots. In Minneapolis, officer debriefings of suspects were an additional feature of the
information/data-gathering process. The Compstat process began when an officer filled out an
incident/arrest report that he or she then gave to a sergeant or commanding officer for approval.
Once approved, a data clerk, who was located in the district, at the MPD and NPD, or in
headquarters, at the LPD, entered crime information from the reports into a records management
system, or RMS.10 Analysts in the Crime Analysis Unit (CAU) selected the data on the crimes
that were regularly presented at Compstat meetings and entered them into a computer database
using a data management program, either MS Excel or MS Access. The crimes presented at
Compstat varied among the sites depending on the priorities of the individual department, but

10 The LPD did not have an RMS due to Lucent Technology’s inability to fulfill its contract. Consequently, the CAU
was responsible for inputting data from officer reports in preparation for Compstat.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

49

they generally included a range of Part I violent and property offenses: homicide, rape, robbery,
assault, burglary, and larceny/theft, which contained a separate category for motor vehicle theft.
In all three departments, the brass identified the crimes that demanded a targeted response
because they were considered most serious or the likeliest source of concern to the community.
Additionally, Compstat sometimes included non-crime information that could serve as a useful
gauge of department performance. Data collected in Newark included records of officer response
and on-scene time, and in Newark and Minneapolis data on individual officer complaints was
gathered and presented at regular, internal affairs Compstat meetings.

In all three departments, the gathering and inputting of Compstat data into a database was
centralized in a CAU staffed by both civilians and sworn officers. Unlike Lowell and
Minneapolis, Newark’s district commanders were each assigned a crime analyst, or crime control
officer, to help with data processing. The crime control officers were supposed to conduct crime
analysis, but they spent most of their time inputting crime statistics into their district’s database
and then using it to construct reports and daily crime bulletins. Since the crime bulletins merely
listed incidents that occurred in the districts, one experienced member of the CAU dismissed
them as “useless.” He added that they did not attempt to identify crime patterns. The type of data
inputted by each department’s CAU included: field report number, shift, district, the date of the
crime, day of the week, and location. Additional information that could also prove useful in the
identification of crime patterns—the point of entry for a burglary or the make and model of a
stolen vehicle—was also entered.

Across sites, basic crime data were entered and generally available the next day, but the

degree to which the data were processed—organized, aggregated, and analyzed—by the CAU
varied among departments. Every morning in Lowell, data-entry clerks scanned the previous
day’s police reports onto the department’s mainframe where they could be accessed directly by
the district commanders and their executive officer. On a biweekly basis, during the week before
Compstat, the CAU produced additional printouts that they made available to each district. These
included MS Excel spreadsheets that were divided by city district and provided details on
individual crimes, such as the suspects involved in an assault and battery case, the case’s
disposition, and summary sheets reporting aggregates, like the total number of assaults during a
shift. The CAU did produce a quotidian newsletter that was made available at roll call and placed
in officers’ mailboxes, but it was generally a summary of the previous day’s crimes. As such it
functioned as a general information tool rather than a systematic and routine way to drive
decision making on crime problems.

In Newark descriptive information from police reports was similarly available each day.

In addition, the CAU also manufactured printouts that detailed the date, time, location, and
district where individual crimes occurred and then organized this information by the crime
categories of auto theft, burglary, theft, or violent crime. This data formed the basis for the
regular morning meetings attended by the CAU, CID, and district detectives. Every day, except
weekends and the day of the Compstat meeting, attendees at Compstat looked over events and
discussed patterns. The CAU provided printouts for the district commanders, and the district
detectives were then responsible for conveying this information back to their districts. One
captain captured the sense of importance surrounding these meetings when he compared them to
“soap operas.” If you missed a day, he said, “You did not know who married whom, who got

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

50

sick, the new characters….” All this specific crime data was then compiled in the department’s
“Compstat book,” along with a brief description of the discussions that had taken place at the
morning meetings. This book was then used as a reference for discussion during the weekly
Compstat meetings. Furthermore, in preparation for Compstat, Newark’s CAU generated maps
and summary sheets reporting crime aggregates that were then added to the Compstat book. This
occurred a day or two prior to the meeting. Since Newark’s CAU manufactured a printout of
individual crime cases to help detectives look for patterns, the department engaged in a higher
level of data processing than the daily scanning and retrieval of reports that took place in Lowell.

Minneapolis differed from Lowell and Newark. Immediately after a data clerk had

inputted information from a police report, the location of the incident was mapped into MapInfo
on the department’s mainframe, and incident case numbers became automatically available on
the map. This usually took place within twenty-four hours after the incident occurred. As a
result, district commanders could access these maps and try to discern the geographic location of
simple crime patterns as they began to emerge. Minneapolis’ CAU, unlike Newark’s, did not
produce daily printouts containing detailed information on individual crimes. Its crime analysts,
however, did prepare handouts containing updated crime numbers—crime percentages for the
week, as well as year-to-date percentages—and disseminated crime maps to the districts a day
prior to Compstat.

Data analysis and presentation

All the data displayed at each department’s regular Compstat meetings were generated
within the department’s CAU. Although their numbers varied, these units tended to include a
handful of analysts—Lowell’s CAU, for example, was composed of five full-time members and
several interns—and they all performed the same function. The CAUs were primarily
responsible for processing, organizing, aggregating, and mapping data in an attempt to identify,
and help the district commanders identify, any crime patterns or trends. Not only were these data
disseminated to the districts, they were also on display at the weekly or biweekly department
Compstat meetings. In Lowell and Newark, one district commander was responsible for
reporting on his or her district’s crime incidents, trends, and tactical responses. In addition, the
district commander, along with his executive officer, faced questions, suggestions, and
comments from audience members. Although only one district presented at each Compstat, we
noted that any district commander could be called upon to account for crime incidents in his or
her district. Although the pressure on these district commanders was not as intense as on the
presenting commander, the possibility of being called upon helped ensure that they did not slack
off. In fact, crime data on the non-presenting districts were often presented simultaneously, not
only to ensure continued accountability, but also to facilitate cross-district comparisons. In
Minneapolis, data were presented on every district during the department’s Compstat meeting,
leaving less time for in-depth discussions than in the other two sites.

The CAUs at all sites were responsible for preparing and projecting the Compstat data

onto a large screen in front of the audience. Each slide represented a separate crime category and
the type of data analysis—more specifically, the mapping of crime data and the comparison of
current with prior Compstat periods—was very similar across sites. Where they varied most was
in the time frame of their comparisons. Since the LPD did not have an RMS, it was difficult for

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

51

the CAU to compare Compstat statistics with anything other than data from the prior Compstat
meeting. In contrast, Newark and Minneapolis were able to examine crime trends over a period
of several weeks, months, or years. In Minneapolis, for example, weekly year-to-date
comparisons—for example, 11/21/2000-11/27/2000 compared to the same period in 1999 and
1998—for specific crimes and all Part I crimes were commonplace.

The CAUs in all three departments used MapInfo in their preparations for Compstat, and

the crime map was an important fixture at weekly Compstat meetings. For example, Lowell’s
Compstat meetings were oriented around district crime maps depicting individual crime
incidents for the six-week Compstat period. Interspersed with these maps were slides containing
aggregate crime statistics and useful summary data, such as the total sum of burglaries and the
total number involving the theft of jewelry. Newark’s and Minneapolis’s Compstat meetings
followed a similar format.

In contrast to the period before Compstat’s implementation, timely crime data now
featured in the organization and operation of the three police departments. Previously, the
departments had merely conducted an annual review of local Part I crime rates collected in the
UCR. The purpose of this brief examination was to provide departments with a general
indication of their overall crime-control performance for the preceding year. In contrast to the
sluggish availability of these data and their relatively narrow focus, each department’s Compstat
program played a continuous and critical role in the department-wide process of identifying
specific crimes as soon as they emerged, driving decision making, and facilitating problem-
solving strategies. As Lowell’s chief stated succinctly, if one believes in the “utility of timely
data … Compstat seems like a smart approach.” The three departments now have daily access to
police reports, as well as weekly or biweekly crime statistics. In Minneapolis, moreover, maps
were also available daily. In addition to being timely, data are routinely processed and geo-coded
to help district commanders discover underlying factors explaining or linking the occurrence of a
number of crime events. Thus, crime analysis is significantly more sophisticated than it was in
the past.

Before moving onto a more detailed discussion of how these data were used, we will
briefly explore two specific challenges associated with the data process: how data were checked
for accuracy and what data were selected, or overlooked, for Compstat. Respondents in all three
departments told us that data reliability was an important component of their Compstat programs,
but we discovered that there were few systematic checks for data accuracy, such as drawing a
random sample of records and double-checking them to see what percentage were reported
accurately. Anomalous crime spikes or downturns, which suggested errors in reporting or time
frame, might be caught during Compstat meetings, and top management in Minneapolis was
especially attentive to this. Reliability, however, was largely dependent upon the vigilance of a
handful of crime analysts who used their experience and familiarity with the data to spot errors.
Furthermore, some officers expressed concerns that Compstat privileged certain crimes over
others, increasing the likelihood that some would be invisible to the police and not be the object
of an adequate police response.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

52

Effective decision making under Compstat demands data that are not only timely but also
accurate. The accuracy of Compstat data is primarily affected by two factors: (1) officer
reporting errors, and (2) inputting errors by the CAU.

Data accuracy

Since the majority of information generated at Compstat depended upon officers’ reports,
it was crucial that this information be recorded accurately. In all three departments, the shift
supervisor was generally responsible for ensuring that careful attention was paid to writing
accurate reports, but sometimes a district commander or district/sector lieutenant would discover
and correct an error while reading a report in preparation for Compstat. During Lowell’s
Compstat meetings, the CAU sometimes highlighted reporting errors, and the chief frequently
conveyed his concern over accurate reporting by telling his command staff that they should
exhort their patrol officers to focus more intently on report writing. We heard similar concerns
expressed in the other two departments, but it was unclear how much the voicing of these
concerns had actually helped improve accurate reporting. It was impossible for either the
departments or us to confirm respondent’s more speculative remarks in the absence of a formal
procedure for systematically checking the accuracy of officer reports. It remained unclear
whether officer reports had improved because Compstat had brought attention to the importance
of crime data in decision making, as it did in Minneapolis, or whether they had barely improved
at all, as a respondent in Lowell suggested. Our survey data, however, can tell us something
about the level of praise patrol officers expected to receive for writing accurate reports. When
asked to rank ten activities according to the level of praise they might receive from their
superiors, with one signifying the least praise, as many patrol officers, approximately half,
seemed to feel that writing accurate reports was somewhat important, ranking it six through ten,
as felt it was fairly unimportant, ranking it one through five. This ambivalence would suggest
that the utility of accurate report writing, as indicated by perceptions of praise from superiors,
was unclear to many patrol officers despite the efforts and pleas of their commanding officers
and the crime analysts.

Although many members of the command staff in each department made it clear to us
that they placed a premium on reliable data, a few patrol officers commented that some incidents
were reclassified or invented in order to improve the department’s crime statistics. For example,
a sergeant in one department stated, “Numbers are changed all of the time.” He added that
burglaries were “often … coded down to thefts.” Another sergeant from the same department
expressed a similar concern and provided the following example:

We were getting killed with burglary dwellings. All of a sudden they went down
drastically. What happened? They created a new code for garage burglary. For all
those years garage burglaries were reported as burglary dwelling and now they
have their own code. It isn’t crimes going down, it is coming up with different
ways to record them. Also, we were getting killed with gas-and-go thefts. The
price of gas was through the roof. Someone decided that it wasn’t really a theft;
rather it was the breaking of a verbal contract and voila our thefts went down.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

53

We can surmise that pressure on a department to reduce crime, particularly by a specific
percentage, may cause some cooking of the books, but without any supporting evidence this
allegation remains speculative.11

None of the departments we observed performed systematic checks for the kinds of data
errors associated with manually inputting large quantities of data and transferring them from one
system to another. This did not mean, however, that they were unconcerned about minimizing
any possible errors that might occur. In Lowell and Minneapolis, three major factors contributed
to the reliability of the data: (1) Crime analysts were responsible for inputting the data and felt
personally responsible for their accuracy; (2) Data errors could be discovered as a result of the
close collaboration among members of the CAU, who were familiar with the geography and
nature of crime in their cities; and (3) District commanders would report discrepancies between
what they had read in their officers’ reports and the materials the CAU had prepared for
Compstat. These features were also common to Newark, but the department had an additional
safeguard. Since each district had its own separate database maintained by a crime analyst, crime
information could be crosschecked with the department’s CAU. When discrepancies, such as the
misclassification of a case, were noticed, they could be corrected.

Visible and invisible data

Given the high visibility within the department of the crimes displayed at regular
Compstat meetings, it is unsurprising that some respondents expressed concerns that crimes not
presented at Compstat might be ignored. For example one respondent in Lowell recalled a period
when domestic violence incidents were no longer mapped separately. Instead, they were included
at Compstat within the category of “assaults,” which mainly distinguished between simple and
aggravated assaults. This did not mean that domestic incidents were ignored, but it did make
them harder to identify. One officer described Compstat as very similar to police radar: “If
something is not on the radar, it is invisible.” It was also unlikely that a district commander
would be held to the same high level of accountability for a crime that was not on the Compstat
screen. Another officer from the same department commented, “If something is not shown at
Compstat, no one cares about it … it means that you are not paying attention to it … you are not
accountable for it.” In Minneapolis, a respondent expressed the same concern: “We only look at
the Part I numbers. We are missing part of the big picture. We do not look at simple assaults or
livability issues, and we need to move toward this.”

A similar logic could be applied to any crime-related information, and not just individual
crime incidents, that was excluded from Compstat. For example, a Minneapolis police officer
criticized Compstat for focusing on locations and not on people. He too used the example of
domestic violence: “I started in domestic assault and feel that women and kid crimes do not get
the attention they deserve. This is a sore spot for me. Units such as domestic assault need to have
mechanisms to track people. There don’t seem to be any clear-cut ways to look at things like this
on maps or through stats.”

11 For a discussion of Compstat and evidentiary questions surrounding the crime drop in New York, see Manning
2001.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

54

It may be an overstatement to argue that non-Compstat crimes are invisible to a
department, since those that are reported are usually documented and often receive some police
response. These observations, however, do raise an important issue: Constantly holding district
commanders accountable for the same crimes at Compstat may lead to potentially useful crime-
related information being overlooked. For example, proponents of “broken windows” policing
might argue that a department’s failure to use Compstat to identify and respond to physical and
social incivilities might lead it to miss opportunities to prevent more serious crimes from
occurring in the future (Wilson and Kelling 1982).

Use of data

The impressive presentation of crime data at weekly or biweekly Compstat meetings tells
us very little about how data were actually used to structure daily decision making: At what
levels in the organization were the data used and for what purposes? Results from our fieldwork
indicated that top and middle managers used crime data to make decisions, but this was not true
for first-line supervisors and rank-and-file officers, who were largely excluded from the
Compstat process. Furthermore, district commanders relied most heavily upon the information
contained in officer reports to identify and respond to emerging problems. Maps were less
important to this endeavor, and we did not observe middle managers in any of the departments
conducting an in-depth analysis of crime data to determine the underlying causes of crime
problems or to decide exactly how to mobilize for action.

District commanders

Every day Lowell’s district commanders, along with their executive officers, read all
their officers’ reports to familiarize themselves with crime in their districts. In consultation with
their executive officers, the district commanders assigned detective follow-ups. Additionally, the
executive officers were primarily responsible for notifying sergeants of their decisions; priorities
were then relayed to patrol officers. The district commanders’ diligence in reading incident
reports can be partially attributed to the absence of pre-Compstat meetings within the
department. Since district commanders did not have this opportunity to share and discuss
detailed crime information with others in their district, they were expected to take the initiative
and familiarize themselves with crime problems in their beats. Lacking the benefit of a
preparatory meeting, they felt compelled to stay on top of things: Failure to recall an incident
during Compstat would be embarrassing and could lead to censure from the chief.

This pressure was experienced less acutely in Newark, since the responsibility was
divided, albeit inequitably, between the lieutenants and first-line supervisors who attended their
district’s pre-Compstat meeting. It would appear that Minneapolis’ district commanders felt the
least pressure to read officers’ reports, since their sector lieutenants also exercised twenty-four-
hour responsibility. This does not mean, of course, that district commanders slacked off, only
that they could afford to do so, since their sector lieutenants were expected to be on top of things.
Each district’s weekly pre-Compstat meetings provided an additional opportunity for district
commanders to catch up and familiarize themselves with what had been happening in their beats.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

55

The originators of New York’s Compstat program were well aware of the importance of
crime mapping. According to Silverman, Jack Maple once commented, “No general went to war
without a map” (1999, 87). The visual presentation of crime maps was an important component
of each department’s Compstat meetings, but it played a far less important role in driving crime
strategies in the districts. In fact, in Lowell and Newark, computerized crime maps were only
made available a few days prior to Compstat to assist preparations for the upcoming meeting.
Old-fashioned pin maps, located in each district’s headquarters, were still in use in Newark, but
they were not a vital component in the development of crime strategies. When we queried district
commanders about their lack of interest in the use of daily maps to strategize, most replied that
they were sufficiently knowledgeable about the geography of their districts that they could map
crimes in their heads as they were occurring. For example, one of Newark’s district commanders
explained that he had grown up in his district, so he knew all the streets by heart and could
identify hot spots.12 We noted earlier that in Minneapolis a crime was generally mapped and
made available to the districts via the department’s mainframe within twenty-four hours of its
occurrence. This did not necessarily mean, however, that district commanders or sector
lieutenants would rely upon maps to make decisions. Certainly some might have occasionally
found the maps helpful, but we observed only one district commander who used them frequently
every day. On the whole, sector lieutenants were more likely than their district commanders to
use crime maps on a regular basis, but this was limited to a weekly, not a daily, examination.

The minimal use of maps in the districts draws attention to how data were used more

generally at the district level. Our fieldwork revealed that the level of analysis performed by each
department’s CAU in preparation for Compstat was much more sophisticated than the daily
analysis conducted in the districts. District commanders and their executive officers serially
reviewed reports to identify problems and geographic patterns, but they did not rely upon
statistics or maps. That is, they did not use the more sophisticated tools available to search for
and characterize patterns in crime; they relied instead on their ability to read a large number of
individual case files and discern larger patterns directly from the raw data. In short, district
commanders used data to react quickly to crime spikes in hot spots rather than to examine
underlying conditions and respond proactively. Even in Minneapolis, where more sophisticated
crime data in the form of maps featured more prominently in department operations than in
Lowell and Newark, district commanders limited their analysis to the time and location of
individual crime incidents. Based upon an examination of these clusters, they implemented their
response.

The reactive nature of the Compstat process, in which strategies were employed in

response to crime spikes at specific locales within each city, can probably be attributed to two
interrelated factors: (1) Middle managers received no formal training in crime analysis; and (2)
More importance was attributed to the rapid availability of data and middle managers’
knowledge of individual cases than to the quality of crime analysis.

12 For an in-depth discussion on the emergence of hot spots policing and its effectiveness, see David Weisburd and
Anthony A. Braga “Hot Spots Policing,” in Crime Prevention: New Approaches, ed. Helmut Kury and Joachim
Obergfell-Fuchs, 337-355 (Mainz, Germany: Weisser-Ring, 2003).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

56

Lack of training

The implementation of Compstat in the three departments was not accompanied by any

specialized training in crime analysis. In comparison to the LPD and NPD, the Minneapolis
Police Department made the most concerted effort to convince officers at all levels in the
organization of Compstat’s utility. Nevertheless, the focus was on selling the idea and
encouraging department-wide acceptance of the program. The workgroup and brainstorming
sessions were not designed to train personnel in data processing and statistical analysis. It was
top management’s belief that district commanders would learn crime analysis through self-
training on the job, facilitated by the type of analysis on display at the department’s weekly
Compstat meetings. Similarly, none of Lowell’s district commanders were provided any
specialized training and were left to develop crime-analysis skills on their own. The lack of
training middle managers received helps explain why they continued to rely upon incident
reports to identify crime problems and patterns. It also helps explain why they did not use crime
data to determine the causes of those problems and the specific crime strategies to pursue. This
would have required the kind of skills and knowledge that cannot be merely picked up while
working on the job. For example, a familiarity with tabulating bivariate relationships—such as
tabulating the frequency of drug crimes with homicides in a sector to see whether an increase in
homicides is possibly drug related—would be a more powerful method for detecting plausible
patterns than just relying upon information in officer reports, skimming crime maps, and
supplementing this data with personal experience and anecdotal evidence.

Rapid data availability and familiarity versus quality of analysis

The lack of intensive formal training in crime analysis also helps explain why

respondents in all three departments concentrated on the speed with which data were available
rather then the degree to which they were processed. One respondent in Minneapolis expressed
excitement about a plan to install laptop computers in officers’ patrol cars. This, he reasoned,
would provide patrol officers with instant access to incident reports. This emphasis on the speed
with which crime information was available rather than the depth and quality of analysis was
common to all sites. For instance, none of the respondents commented that they would benefit
from a more sophisticated statistical examination of crime data, such as bivariate or multivariate
analyses.

A higher level of data analysis took place in each department’s CAU where many of the
crime analysts had a master’s degree in criminal justice or a related field. However, this analysis
was still relatively limited. Our fieldwork indicated that it focused on descriptive statistics, such
as percentages, frequencies, and means. It was not the kind of in-depth research approach
advocated by Herman Goldstein (1990), and it was very far from the rigorous, analytic,
“evidence-based” research ideal advocated by Lawrence Sherman (1998). Furthermore, this
analysis was not disseminated to the districts on a daily basis. Newark’s CAU had progressed
furthest in providing district commanders with timely CAU-processed data, but this information
consisted of spreadsheets grouping incidents into crime categories rather than aggregated data.
Since the data analysis performed by the CAU was only available a short time before Compstat,
and after crime strategies had been implemented in the districts, it was used by middle managers

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

57

to corroborate their existing impressions rather than to make decisions. Similarly, pre-Compstat
meetings, attended by middle managers and some first-line supervisors, were used to review the
events of the previous week or weeks to ensure that nothing had been missed. In some respects
they resembled the cramming sessions that some students put in before the day of a mid-term
exam. One respondent specifically used this analogy when describing his preparations for
Compstat meetings: “Compstat is in some ways like being a student. As a student you study just
so that you can pass a test; just like with Compstat you just prepare in order to pass it.” Passing
did not entail analyzing data to provide penetrating insights into crime problems. Being familiar
with individual crime incidents, being able to identify obvious patterns, and having implemented
some kind of response was sufficient. Incentives and discipline associated with Compstat in
these departments endorsed this narrow focus, since middle managers were more likely to be
held accountable for not knowing something about an individual case than failing to discern
some underlying cause for a crime pattern.

Patrol officers

If middle managers used crime data to drive problem identification, how were patrol
officers involved in this process? Our fieldwork revealed that first-line supervisors and patrol
officers experienced Compstat as a series of directives instructing them what to do rather than as
a strategic and collaborative decision-making process. In Newark, patrol officers rarely attended
the department’s Compstat meetings. Minneapolis and Lowell had taken greater strides to
include officers by requiring that they attend Compstat on a regular, albeit infrequent, schedule.
But this did little to weaken the perception that Compstat belonged to the “brass.” In Lowell, 56
percent of the patrol officers we surveyed had not attended a Compstat meeting, and in Newark
the figure was even higher at 63 percent.13

In addition, officers rarely made decisions on the basis of the information that was
gathered for Compstat and presented at meetings. This could be attributed to a range of factors
that included: lack of access to Compstat via a computer terminal and dearth of interest. Access
to Compstat information was most restricted in Newark. The CAU maintained its own database,
and access was only granted following a specific request to the principal crime analyst.
Accessibility was also very limited in Lowell and Minneapolis. In Lowell, most of the
department’s computer terminals were not equipped with the software programs that would
enabled line officers to read the Compstat files on the department’s server. In Minneapolis, the
impediment was the small number of computer terminals in the districts. In all three departments,
line officers’ inadequate computer skills also helps explain their marginal role in the Compstat
process. In Minneapolis, one sergeant commented that patrol officers did not have the time to sit
down and work at the computer. He added, “The technology and mapping is over some people’s
heads.” Similarly, respondents in both Lowell and Minneapolis acknowledged that the data
component of Compstat had not “filtered” or “trickled” down to the street level.

13 A much higher percentage (67 percent) in Minneapolis responded that they had attended a Compstat meeting.
However, our survey did not distinguish between regular Compstat meetings and the initial one-day training
meetings held during Compstat’s implementation. Based upon our observations, it is likely that a much lower
percentage of line officers had actually attended a regular Compstat meeting.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

58

The most systematic means of learning about Compstat and crime-related issues was
provided through information disseminated at roll call. In Lowell, this was supplemented by the
CAU’s Daily Bulletin newsletter, which published some information about the previous day’s
events. Written directives and memos were read aloud, and district commanders in Minneapolis
provided patrol officers with specific instructions identifying directed-patrol areas. The rank and
file also learned about what happened at Compstat meetings from their fellow officers and
supervisors. The latter avenue tended to be fairly haphazard since there was no mechanism for
systematically relaying this information (see page 42). Comments from respondents suggested
that information was usually exchanged during informal conversations that patrol officers had
with captains, lieutenants, detectives, or sergeants. Following the chain of command, a district
commander or executive officer might ask a sergeant to tell his or her patrol officers to keep an
eye on a problem, or increase visibility in a problem area identified at Compstat. The limited
participation of officers at Compstat meetings and the somewhat desultory flow of information
did not imply that officers were unfamiliar with the major crime problems in their district. It only
indicates that they were not included in the Compstat process of analyzing and discussing crime
data. This contrasts sharply with community policing, which encourages patrol officers to
participate fully in identifying and assessing crime problems in their beats. Our visits to other
Compstat departments suggested that a few police departments have made more of an effort to
engage lower-ranking personnel at the district level in Compstat. In New Orleans, for example,
patrol supervisors and patrol officers regularly attended pre-Compstat meetings (Gurwitt 1998).

In sum, Compstat has increased the use of data by middle managers in identifying crime
patterns and making decisions. The importance of data was also reflected in a more general
concern that it be timely and processed accurately, although reliability was more dependent upon
individual initiative and de facto crosschecking than on a routine auditing process. In spite of this
improvement over the very limited utility of the UCR, district commanders continued to rely
upon officer reports and their own impressions of crime. Like the traditional model of policing,
the use of data to identify and assess problems was centralized among the command staff and in
department headquarters—patrol officers played a marginal role in this process.

Innovative, Problem-Solving Tactics

Police have long collected data and compiled statistics, but those data were rarely used to
make important decisions about how to solve problems (Mastrofski and Wadman 1991). Despite
its central purpose of crime reduction, Compstat’s relationship to innovative, problem-solving
tactics is probably the least developed element in the existing Compstat literature. Middle
managers are expected to select responses to crime problems that offer the best prospects of
success, not because they are “what we have always done,” but because a careful consideration
of a number of alternatives showed them to be the most effective (Sherman 1998). Innovation
and experimentation are encouraged; use of the best knowledge about practices is expected. In
this context, police are expected to look beyond their own experiences by drawing upon
knowledge gained in other departments and on innovations in theory and research about crime
control and prevention. Silverman argues, “This innovative process [Compstat] radiates
throughout the NYPD as the energizer of strikingly creative decision making at headquarters and
in the field” (1999,123-4).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

59

In our national survey, this did not seem to be the case very often in Compstat programs
across the country. When departments were asked how they decided upon a problem-solving
strategy to address “the one crime/disorder problem that used more of the department’s efforts
than any other problem in the last twelve months,” they were most likely to respond that they
relied on the department’s previous successes with an approach. Sixty-seven percent of
departments stated that this was “very important” followed by 39 percent who stated, “research
evidence suggested that this was the best approach” (Weisburd et al. 2001, 41). In other words,
departments with Compstat still tended to rely upon their own experience with traditional law
enforcement approaches, such as saturating an area with police and making arrests.

In many respects, Lowell, Minneapolis, and Newark mirrored this national trend by
mainly relying upon traditional law enforcement tactics that have long been a staple in police
agencies. It is important to recognize, however, that the Compstat approach did lead to
innovation in the way police directed their resources. Compstat facilitated the gathering and
dissemination of information that was used to identify problems and focus police efforts.
Directed patrol at hot spots, for example, represents an important innovation in policing that has
been supported by strong research evidence (Braga 2001; Sherman and Weisburd 1995;
Weisburd and Braga 2003). The departments we observed did utilize Compstat to harness
information to facilitate such concentration of police resources. Nevertheless, we observed little
innovation in the approaches used once hot spots were identified. In this sense, there was
innovation in how police focused resources but relatively little innovation in what police did
once they had identified problems.

A major factor contributing to this lack of innovation was Compstat’s heavy emphasis on
the accountability of middle managers for carefully sifting, considering, and deliberating about
crime patterns and then conducting a careful review and discussion of the benefits and
drawbacks of various approaches—the very kind of analysis that Compstat claims to encourage.
Follow-ups, when they did occur, were testimony to the powerful influence of internal
accountability, and they were limited to a determination of whether or not a district commander
had identified a crime problem and implemented a response. We did not observe any in-depth
discussions on how or why a crime strategy worked or failed to work; nor did we hear any
suggestions for its improvement. Although directed patrol was not the only recourse in the
MPD’s toolkit for responding to a crime spike, the range of alternative crime responses did
appear to be limited by the department’s focus on directed patrol as the strategy of choice, thus
inhibiting experimentation and innovation. Under Compstat, district commanders and sector
lieutenants were expected to instruct their officers to use their uncommitted time to enforce the
law aggressively in areas that showed an increase in criminal activity. Thus, the problem solving
stimulated by all three departments’ Compstat programs replicated the classic street cop’s ethos
that it is more important to act decisively—even with an incorrect diagnosis and hastily-
considered or preconceived solution—than it is to ponder the evidence and weigh options before
carefully proceeding.

Problem solving and brainstorming during Compstat meetings

In addition to brainstorming, Compstat encourages innovative thinking, according to its
supporters, by “constantly challenge [ing] precinct commanders to develop new responses to

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

60

crime problems” (Kelling and Sousa 2001, 12). Many who attended Compstat felt that there was
a great deal of brainstorming and reporting, but our observations suggested that this overstated
the case. The presentation of crime data did facilitate some sharing of information and insights
on how a problem might be tackled in a district. In other words, audience members did exchange
factual knowledge and provide suggestions based upon personal experience, but they were far
less likely to engage in in-depth discussions of the advantages and disadvantages of a broad
range of problem-solving alternatives.

The utility of the suggestions made at Compstat was severely limited by the pressures
that internal accountability placed upon district commanders. In response to this pressure district
commanders felt compelled to have a crime strategy already in place before Compstat meetings.
In addition, the onus on middle managers to come up with quick and effective solutions—ones
whose successful implementation was likely to involve the investment of precious resources—
increased the likelihood that potentially helpful criticism made during Compstat would be
regarded as threatening or unhelpful. In Minneapolis, one top manager noted that Compstat
meetings tended to be “show and tell” sessions with little collaboration among the command
staff on crime issues. She stated that it would be seen as “hostile” if other people added their
thoughts to the district commander’s presentation. At this point, the district commander had
already invested his or her reputation, as well as the district’s resources, in the chosen solution.
The likelihood that a district commander would feel maligned increased when the source was
perceived as poorly informed or inexperienced. A respondent in Lowell commented that he
sometimes regarded comments made during Compstat as “slings and arrows.” Radical changes
would either be unnecessary, since the strategy had already solved the crime problem, or
unwelcome, given the investment of resources and reputation.

In addition to the limitations imposed by internal accountability, the scheduling and
organization of pre-Compstat and Compstat meetings lends itself to little more than a review of a
district commander’s tactics. Even without the pressure of accountability, middle managers have
always had to respond quickly to crime. By the time district commanders presented at weekly or
biweekly Compstat meetings, they were trying to recall crimes committed several days, if not
weeks, earlier. One respondent in Lowell remarked trenchantly that Compstat, despite all the
hype, still promoted a traditional “reactive” response to crime. He supported his statement by
noting, “People have never comprehended technology to be proactive.” Since the information
displayed at Lowell’s Compstat was as much as six-weeks old, Compstat participants were
clearly reacting to incidents rather than implementing preventive measures. A district
commander in Minneapolis, where the data presented at Compstat was only from the prior week,
voiced a similar concern:

We are looking at old information. Part of the presentation should be focused on
what is going on right now rather than what ended last Monday. By the meeting
we have three or more days of data and I am in a different mindset already. For
example, if we had a homicide after midnight on Tuesday morning, I would be
dealing with that all day Tuesday, Wednesday, and Thursday, but come Compstat

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

61

on Thursday afternoon, the homicide is not even dealt with until the following
week.14

Follow-up

According to Silverman, one of the key features of New York’s Compstat is the
“particular emphasis on follow-up and constant monitoring” (1999, 193). This is essential for
ensuring that tactical responses are implemented and desired results are actually achieved (Walsh
2001, 355). Follow-up in our three sites was not as intensive or “relentless” as in the NYPD
model. While the level of attention paid to follow-up differed considerably among the
departments, all three generally relied upon the use of maps or crime statistics to determine
whether crime activity had stopped or diminished. If a crime pattern disappeared, it was
considered resolved. However, identifying the disappearance of a crime pattern does not provide
a careful assessment of why a particular strategy worked. A “relentless” assessment would seem
to demand an evaluation and thorough description of why it succeeded.

In contrast, those who directed New York’s Compstat program argued that follow-up

must be more intensive. This is how former police commissioner Bratton described it:

As in any problem-solving endeavor, an ongoing process of rigorous follow-up
and assessment is absolutely necessary to ensure that the desired results are being
achieved. This evaluation component also permits us to assess the viability of
particular tactical responses and to incorporate the knowledge we gain in our
subsequent tactics development efforts. By knowing how well a particular tactic
worked on a particular crime or quality-of-life problem, and by knowing which
specific elements of the tactical response worked most effectively, we are better
able to construct and implement effective responses for similar problems in the
future. The follow-up and assessment process also permits us to redeploy
resources to newly identified strategies once a problem has abated (McDonald
2001).

In the NYPD, top-level officials met after Compstat, reviewed the detailed summary

minutes of the meeting, and assigned responsibility for solutions and their implementation. In
other words, they distributed a detailed written record of whatever transpired at Compstat to top
officials in order to ensure that nothing was missed in subsequent reviews. This record frequently
recorded requests for updates at subsequent Compstat meetings and recommended strategies of
action. As Silverman writes, “Post-Compstat meetings provide an official record and ensure that
everyone is on the same page” (1999, 110). In short, the NYPD did not just rely on Compstat
maps and the presentation of crime data in the follow-up process.

Our three sites not only differed from the NYPD. They also varied considerably among

themselves in the level of attention they paid to follow-up, a difference that depended on the

14 We note that the MPD’s top leadership made concerted efforts to ensure timely crime data through discussion of
some of the most recent crime incidents at regular command staff or intelligence meetings held on the Tuesday prior
to CODEFOR. Nevertheless, as this quote demonstrates, there was concern that some crime data were not available
on a timely basis.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

62

kinds of mechanisms they had implemented to supplement their regular Compstat meetings. Our
observations suggested that follow-up was most comprehensive and systematic in Minneapolis,
where district commanders regularly monitored whether patrol officers were following orders
and engaging in directed patrol. This mechanism helped ensure that strategies were actually
being implemented, but its utility was limited by its failure to monitor the effectiveness of
directed patrol. Newark trailed closely behind Minneapolis, with the department using its
Compstat book as a general reference guide to the implementation of crime responses. Follow-up
was least systematic in Lowell, where the chief and his command staff relied upon the standard
crime maps presented at Compstat to track results.

In Minneapolis, the origins of directed patrol lay in a department study estimating that
only three hours of a patrol officer’s time, on average, was committed to answering calls for
service during a regular ten-hour shift. When a district commander had identified an emerging
pattern, he or she could then create a directed patrol sheet instructing officers to pay attention to
a specific area. The instructions were announced and posted at roll call. The district commander
could then monitor the effectiveness of this response by keeping an eye on proactive police
activities in the area, such as narcotics arrests or the stopping and tagging of cars, and any
reductions in crime. In addition to this district-level mechanism, the lieutenant in charge of the
CAU was responsible for checking crime patterns and monitoring crime responses between
Compstat meetings. He then prepared notes and follow-up questions that the officer in charge of
Compstat used to query district commanders about their crime strategies. He also summarized
what had been discussed during the meeting and disseminated this record to top management and
each district commander. Consequently, follow-up was a regular feature at both the department-
and district-level.

In Newark, follow-up was the primary domain of the officer in charge of the weekly
Compstat meetings. It was facilitated by the Compstat book, which recorded the discussions and
suggestions that occurred at the regular morning and weekly meetings. As such, it provided a
helpful tool for ensuring that relevant information was not lost or forgotten. As in Minneapolis, a
lieutenant was responsible during Compstat for taking notes that were used to remind the officer
in charge of what issues needed to be reviewed.

At Lowell, an administrator took notes during Compstat, but these cursory jottings only
consisted of a line or two on each Compstat slide and were for the chief’s personal use rather
than for review by command staff. One respondent expressed concern that there was not a more
effective “feedback loop” in the department’s Compstat program. He further noted that when a
problem came up at one meeting, there was no guarantee that it would be addressed at a
subsequent meeting. Our observations supported the assertion of this and other respondents that
there was a discontinuous relationship between Compstat meetings. In the absence of a more
systematic means of monitoring the effectiveness of crime strategies than weekly Compstat
meetings, follow-up was fairly haphazard.

To sum up, the weight of accountability and the delay between a crime’s occurrence and
its presentation at Compstat hampered innovative problem solving. District commanders were
under pressure to implement a crime strategy quickly or to have already responded to an “old”
crime prior to Compstat. Furthermore, follow-up focused on holding middle managers

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

63

accountable for having done something; they were not accustomed to thoughtfully assessing
“how interventions work, under what circumstances, with what effort, and for how long”—
norms commonly associated with problem-oriented policing (Greene 2000, 316). It is instructive
to look at a sector lieutenant’s vague response to a deputy chief’s question about why there had
been a decrease in his district’s auto thefts and thefts from motor vehicles: “Maybe the directed
patrol; they flood the areas. There are still pockets, but there is a lot of presence and they have
fliered the areas.” He added that he was not sure why the district’s crime numbers were down
from the previous year, surmising that it was “probably seasonal.”

Since there was no careful evaluation of whether different crime strategies were actually

achieving their intended results, officers generally had to resort to their impressions of what
appeared to have worked in the past. Knowing this, some middle managers actually avoided
thinking more innovatively—presumably because there was no tangible reward for doing so. We
watched one executive officer actually advise his officers, the day before their Compstat
presentation, to stop worrying about identifying and discussing hot spots. He instructed them to
focus their discussion on describing the kinds of activities—whether they were arrests or use of a
decoy car—that they had used to respond to a recent spate of thefts from motor vehicles in their
district. The utility of this tip, it appeared, lay in its capacity to divert attention from a more in-
depth discussion of crime patterns and alternative strategies. His advice was sound. During the
discussion of motor vehicle crimes, the presenters were not asked to elaborate on the
characteristics of these incidents or the possibility of a crime hot spot.

Traditional responses

Most of the problem-solving approaches identified at Compstat relied on traditional
police strategies that had been used before—in particular, asking patrol officers to identify
suspects and keep an eye on things, area saturation, stepping up traffic enforcement, knock-and-
talks, and increasing arrests. For example, in response to two unrelated incidents, an increase in
prostitution in a particular location and the constant use of a specific pay-phone by suspected
drug dealers, one of Lowell’s district commanders put extra cruisers in the area. During another
Lowell Compstat meeting 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 that they clamp down on motor vehicle violations: “You know, chief, sometimes you
just get lucky. You catch a kid and they just talk. We need to get people in to talk to them.”
Similarly, when we asked a district commander in Newark what he had done regarding a spate of
violent crimes within an area of the city, he told us, “good old police work” that included putting
extra people in the area, increasing police visibility, and sending in the drug unit.

There did, however, appear to be some small but observable differences in the capacity of
individual organizations to facilitate innovative problem solving. In Lowell, the chief strongly
encouraged his command staff to share ideas on crime strategies, and in Newark, task forces
provided a similar forum. In Minneapolis the importance accorded directed patrol, Compstat’s
focus on crime numbers rather than strategies, and the absence of task forces combined to limit
innovative thinking.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

64

Innovative responses

Many incidents that are identified at Compstat are temporary crime blips that will
disappear by the time the next Compstat meetings rolls around. Should crime analysis reveal that
there are no patterns, it is appropriate to implement a routine response, such as taking a report
and assigning follow-up. Nevertheless, there are some incidents that are clearly related and their
resolution may demand significant time, resources, and innovation. Lowell’s chief often
emphasized “thinking outside of the box” during Compstat, and we did observe an innovative
problem-solving response to the criminal activity associated with a dilapidated boarding house
occupied by vagrants, criminals, and down-and-outers. Despite increased police activity, there
were continuing incidents involving alcohol abuse, drugs, vandalism, and serious crime, such as
robbery and stabbings. As a result, the district commander pursued two more innovative
problem-solving strategies to great effect: (1) Patrol officers conducted bed checks and arrested
non-residents for trespassing; and (2) The department encouraged city administrators to secure a
order demanding that the recalcitrant landlord close the building down. These police activities,
the district commander noted, proceeded on two levels—administration and patrol—and
represented a clear example of a coordinated response to a difficult problem. Invoking laws
against trespassing, officers were able to make arrests, identify suspects, and use Compstat to
catalogue the ensuing information. The chief, along with his district commander, then presented
this evidence at City Hall to demonstrate the extent of the problem—the building had been
responsible for over three hundred calls for service over a ten-month period—and to end the
legal foot-dragging that had prevented the city from closing the building. Although we did not
have the opportunity to observe Newark’s task forces in action, they were often composed of
officers who had reputations for being experienced investigators and innovative thinkers. Task
forces could have continued to implement more traditional crime strategies, but the selection of
officers with backgrounds in brainstorming and problem solving suggests that the department
was making an effort to encourage in-depth thinking on some crime issues.

In Minneapolis, innovation was mainly composed of directed patrol and zero-tolerance
policing. Eck argues that problem-oriented policing can take one of two forms. The first is the
meticulously researched, carefully coordinated, in-depth, problem-solving response advocated by
Herman Goldstein (1990); the second is what Eck refers to as “enforcement problem-oriented
policing” (1993, 68). Directed and tactical patrols are illustrative of this approach, but
departments that use these approaches run the risk of defining problems narrowly and addressing
them with rather traditional methods, as Eck cautions. Since directed patrol in Minneapolis was
combined with zero-tolerance policing, it limited the exploration of problem-solving alternatives.
Criminal law, not the problem, was the defining characteristic of this approach. Innovative
thinking was not encouraged in task forces or at Compstat meetings, where time constraints left
no opportunity for questions or discussions until the end of meetings. The organization,
consequently, was not structured to facilitate the sharing of ideas, as became clear at one of the
department’s annual command meetings to assess Compstat. When one of the top managers
advocated the creation of a poster delineating the most effective problem-solving strategies, his
suggestion was opposed; the district commanders responded that they already knew the answers
to the problems in their districts.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

65

Unfortunately, time constraints prevented us from exploring the department’s CCP/SAFE
units in more detail. The MPD’s top leadership commented that these were important loci of in-
depth, problem-solving activities. Our own observations, though limited to brief descriptions of
CCP/SAFE activities by command officers during Compstat meetings, suggested that problem
solving may actually have occurred on a more modest scale. In response to patterns of burglaries,
thefts from motor vehicles, and smash and grabs that arose during several Compstat meetings,
CCP/SAFE units distributed crime prevention fliers, offered to initiate a “Crime Watch,” and
warned business owners to remove cash from their registers at night.

In sum, it is a serious overstatement to claim that Compstat in these three sites fostered
problem-solving innovation and experimentation, based upon brainstorming in Compstat
meetings, collaboration among districts, and a close examination of research evidence.
Compstat’s data orientation did affect what problems were identified and consequently
influenced the timing and location of responses. Compstat generally did little, however, to
stimulate a data based analysis that could drive the decision about how to respond. This is not
entirely surprising because the sort of sophisticated data analysis that might go beyond targeting
problems and hot spots was not yet part of the departments’ routines. While innovation in police
responses was not absent from the Compstat process, it was hardly the hallmark that Maple
considered it to be. Finally, in most cases middle managers were responsible for coming up with
problem-solving tactics. District commanders and their lieutenants took the lead in the decision-
making process and rarely approached the lower ranks for suggestions. In other words, some
input may have bubbled up from below, but ideas tended to originate from the top and filter
down the chain of command.

VI. DISCUSSION

Results from out national survey showed Compstat to be a policing phenomenon that was
sweeping across the nation (Weisburd et al. 2001). However, the popularity and praise afforded
Compstat by many police practitioners and scholars is fueled more heavily by rhetoric and
anecdotes than by a body of systematic research. Rather than focus on the NYPD, we decided to
contribute to a more comprehensive understanding of Compstat by analyzing how it operated in
three police agencies of different size and organization. In our discussion, we compare how each
Compstat element worked in Lowell, Minneapolis, and Newark (see Table 7 for a summary) to
illuminate the contributions and challenges of Compstat: Does Compstat function in the ways its
supporters claim? In addressing this question, we distinguish between different types of
challenges that arise in implementing Compstat: problems that are within a department’s
capacity to correct and those that stem from inconsistencies and contradictions within Compstat
as a program. Finally, we examine how Compstat worked with another highly publicized and
widely heralded program, community policing. Although much has been written about this
current transformation in American policing, we know very little about its relationship with
Compstat. Some of our field observations suggest the existence of tension or conflict between
some of Compstat’s elements and key tenets of community policing. By revealing some of the
complexities surrounding Compstat, we hope to provide police and policy makers with some
initial insights into how they might achieve desired outcomes and avoid undesired consequences.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

66

Table 7: Summary comparison of Compstat elements across sites

Lowell Minneapolis Newark
Mission
Clarification

Ambitious and visionary goal: “To be the safest city of its size
in the United States”

Message not reinforced with any other mechanisms

No specific operational tactics implemented for
accomplishment of mission

Strong acceptance of Compstat mission with some criticism of
goal for being unrealistic

Tangible and specific goal: to reduce index crime by 10
percent in the year 2000

Message reinforced through training

Directed patrol and zero-tolerance policing implemented for
accomplishment of mission

Lowest acceptance of Compstat mission. Zero-tolerance
policing not associated with accomplishment of goal and
regarded as conflicting with other important organizational
values

No mission statement that explicitly mentions crime
reduction

Message not reinforced through Compstat-related
mechanisms but more general crime-fighting ethos.

No specific operational tactics implemented for
accomplishment of mission

Strongest acceptance of Compstat mission.

Internal
Accountability

High level of accountability experienced by district
commanders due to regular Compstat meetings

No similar Compstat structure for making rank and file feel
equally accountable for their performance

Reassignment of underperforming district commanders very
rare—top leadership constrained by fear of appearing
arbitrary. Small pool of qualified candidates and limited
availability of slots for promotion imposes additional
constraints

General impression that underperforming district commanders
would not be replaced counteracted potency of Compstat
meetings among lower ranks

Decision making under Compstat experienced by rank and file
as a series of commands

High level of accountability experienced by district
commanders and sector lieutenants due to regular Compstat
meetings

District commanders scanned crime maps and officers’
reports to check that patrol officers had engaged in directed
patrol, but rank and file still experienced low level of
accountability

Reassignment of underperforming district commanders very
rare—top leadership constrained by fear of appearing
arbitrary

General impression that underperforming district
commanders would not be replaced counteracted potency of
Compstat meetings among lower ranks

Decision making under Compstat experienced by rank and
file as a series of commands. Autonomy further constrained
by focus on directed patrol and zero-tolerance policing

High level of accountability experienced by district
commanders due to regular Compstat meetings

No similar Compstat structure for making rank and file
feel equally accountable for their performance

Reassignment of underperforming district commanders
very rare—top leadership constrained by fear of
appearing arbitrary

General impression that underperforming district
commanders would not be replaced counteracted
potency of Compstat meetings among lower ranks

Decision making under Compstat experienced by rank
and file as a series of commands

Geographic
Organization
of Operational
Command

Twenty-four-hour responsibility for a geographic area
devolved down one level to district commanders

Nature of district commanders’ authority was complex. Top
management still played an important role in how decisions
were made, especially on critical issues, such as resource
allocation

Modest shift toward geographic management. Some detectives
placed under direct supervision of district commanders

Twenty-four-hour responsibility for a geographic area
devolved down two levels to district commanders and sector
lieutenants
Nature of district commanders’ authority was complex. Top
management still played an important role in how decisions
were made, especially on critical issues, such as resource
allocation
More significant shift toward geographic management. Each
district assigned own property crimes investigation unit,
community response team (CRT), and CCP/SAFE unit;
district commander ultimately in charge of these units

Twenty-four-hour responsibility for a geographic area
devolved down one level to district commanders

Nature of district commanders’ authority was complex.
Top management still played an important role in how
decisions were made, especially on critical issues, such
as resource allocation

Modest shift toward geographic management. Some
detectives placed under direct supervision of district
commanders

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation

67

Lowell Minneapolis Newark
Organizational
Flexibility

Flexibility of resource allocation achieved mainly
though ad hoc decision making. Formation of task
forces was rare, since management regarded them as
anomalous to geographic-based management. No
district-level taxi squads

Within-district flexibility limited by absence of taxi
squads. The occasional poaching of officers to form
task forces mitigated this effect but simultaneously
reduced between-district flexibility

Pressure of internal accountability fostered
competition that hindered resource sharing among
districts

Flexibility limited by traditional internal and external
constraints: lack of manpower, conventional
approaches to resource allocation, and city politics

Flexibility of resource allocation largely achieved through
district-level taxi squads. Less reliance on ad hoc decision
making and formation of task forces

Taxi squads increased within-district flexibility. Task forces
were relatively rare, thereby reducing between-district flexibility.

Pressure of internal accountability fostered competition that
hindered resource sharing among districts

Flexibility limited by traditional internal and external
constraints: lack of manpower, conventional approaches to
resource allocation, and city politics

Flexibility of resource allocation largely achieved through
formation of task forces and ad hoc decision making. No
district-level taxi squads

Frequent formation of task forces increased between-district
flexibility but simultaneously reduced within-district
flexibility. Within-district flexibility further limited by absence
of taxi squads.

Pressure of internal accountability fostered competition that
hindered resource sharing among districts

Flexibility limited by traditional internal and external
constraints: lack of manpower, conventional approaches to
resource allocation, and city politics

Data-Driven
Problem
Identification
and
Assessment

Timely crime data were a feature of organization and
operation. Daily crime reports made available to
districts. CAU did not provide daily aggregates of
crime incidents. Daily crime maps unavailable

Crime reports facilitated the identification of crime
problems and patterns at the command level

CAU provided aggregate data, descriptive statistics,
and crime maps a few days prior to Compstat

Data used to inform rather than drive decision making
at command level. Data analysis not diffused to line
officers

Timely crime data were a feature of organization and operation.
Daily crime reports made available to districts. CAU did not
provide daily aggregates of crime incidents, but crimes were
mapped immediately

Crime reports and incident mapping facilitated the identification
of crime problems at the command level. Crime maps generally
used for the identification of crime patterns on a weekly basis

CAU provided aggregate data, descriptive statistics, and crime
maps a few days prior to Compstat

Data used to inform rather than drive decision making at
command level. Data analysis not diffused to line officers

Timely crime data were a feature of organization and
operation. Daily crime reports made available to districts.
CAU provided districts with daily aggregates of crime
incidents, including crime types and location. Daily crime
maps unavailable

Crime reports facilitated the identification of crime problems
and patterns at the command level

CAU provided aggregate data, descriptive statistics, and crime
maps a few days prior to Compstat

Data used to inform rather than drive decision making at
command level. Data analysis not diffused to line officers

Innovative,
Problem-
Solving
Tactics

Continued reliance on traditional strategies to solve
problems

Leadership encouraged innovative thinking, but the
pressure of internal accountability discouraged
brainstorming and rewarded quick and decisive action
over innovation

Follow-up was unsystematic. It occurred at Compstat
but depended upon leadership remembering what had
happened at prior meetings. District commanders did
not receive any record of what was discussed during
Compstat

Some innovation through directed patrol and zero- tolerance
policing, but continued reliance on traditional law enforcement
powers of police. Emphasis on directed patrol and zero-tolerance
policing appeared to limit focus on more innovative responses

Leadership encouraged innovative thinking, but the pressure of
internal accountability discouraged brain-storming and rewarded
quick and decisive action over innovation

Follow-up was fairly comprehensive and systematic. Middle
managers monitored but did not assess why a particular strategy
worked. At Compstat, CAU recorded questions and responses
that were used at subsequent meetings. District commanders
were given a copy of what was discussed during Compstat

Continued reliance on traditional strategies to solve problems,
but task forces might have encouraged in-depth thinking on
some crime issues

Leadership encouraged innovative thinking, but the pressure of
internal accountability discouraged brainstorming and
rewarded quick and decisive action over innovation

Follow-up at Compstat was systematic. CAU recorded all the
information provided to district commanders through district
detectives at daily crime meetings and all the questions and
responses raised during Compstat in a “Compstat Book.”

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 68

Mission Clarification

Supporters of Compstat argue that one of the program’s key features is the proclamation
of a bold crime-reduction mission that fosters widespread commitment within the organization.
Our observations suggest that the implementation of Compstat does, indeed, help top
management convey a powerful message about the importance of fighting crime; the patrol
officers we surveyed strongly associated Compstat with crime fighting. Nevertheless, the level of
officer commitment was the opposite of what we had predicted. It was strongest in the NPD,
whose mission statement made no explicit mention of crime reduction, and it was weakest in
Minneapolis, where management had made the most concerted attempt to foster acceptance of
Compstat. Not only did leadership in the MPD establish a specific target for which the
organization could be held accountable, it also reinforced this message through other
mechanisms and implemented tactics designed to help the organization accomplish its goal.
Despite these efforts, Minneapolis’ rank and file appeared the most disaffected or ambivalent
about Compstat: Some failed to associate directed patrol and zero-tolerance policing with the
organization’s attempt to reduce Part I crimes, while others felt that this approach conflicted with
other equally important goals.

The reaction of Minneapolis’ patrol officers highlights some of the challenges associated
with the attempt to distill a public organization’s many, and often conflicting, goals into one
simple objective. Unlike Compstat, community policing explicitly recognizes that no
contemporary police organization is long allowed to pursue one single objective. The
organization resolves this by working within the bounds of acceptable compromises that do not
provoke crises or make the organization the target of negative publicity. Minneapolis’ patrol
officers disapproved of directed patrol because they felt it left less time for other equally
important activities. In complaining about their inability to provide quality service to other
constituents, and in articulating the harmful effects of an aggressive crime-fighting approach to
community relations, they explicitly recognized the limitations of the traditional focus on crime
reduction. Fighting crime, providing services, and solving problems are not mutually exclusive,
but Minneapolis’ emphasis on all three fostered cynicism and frustration. Although top
leadership worked hard to expand department goals well beyond crime control, officers felt the
underlying focus on crime fighting undermined the organization’s commitment to a much
broader mission.

Internal Accountability

Many people in our three sites considered Compstat’s principal element to be internal
accountability, or making a specific individual responsible for tackling and reducing crime. It
was also an element with some positive effects. Top management and district commanders were
more aware of their crime environment and more motivated to get on top of crime problems than
before.

Internal accountability, however, did present some significant challenges. Since it was
experienced most acutely at Compstat meetings, it impacted disproportionately on middle
managers, particularly district commanders. In the absence of a similarly potent mechanism for
pushing accountability further down the chain of command, first-line supervisors and patrol

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 69

officers did not feel the same kind of pressure to perform. We have argued elsewhere that the
pressure of internal accountability has the paradoxical effect of undermining the utility of two
other Compstat elements. Not only does it discourage innovative problem solving by
substituting speed for effectiveness and hindering open discussions; it also reduces
organizational flexibility by fostering competition (Willis et al. 2004b).

By enforcing internal accountability for middle managers, Compstat centralized decision-

making authority and left line officers with the primary responsibility of carrying out the district
commander’s orders. Their compliance allowed the district commander to provide an adequate
response to the chief’s inquiries at Compstat. If they neglected to follow orders, however, or
decided to pursue alternate strategies, the district commander would be vulnerable to censure
from the chief. Therefore, first-line supervisors and patrol officers experienced Compstat as a
series of directives that flowed down the chain of command rather than a system that encouraged
them to exercise discretion and solve problems. Since one of the crucial elements of community
policing is the decentralization of decision-making authority, Compstat works at cross-purposes
with this approach (Weisburd et al.2003).

Geographic Organization of Operational Command

Operational command in all three departments had been lowered to middle managers, but
it had devolved furthest in Minneapolis where district commanders and sector lieutenants
exercised twenty-four-hour responsibility for their specific beats. All middle managers exercised
considerable decision-making authority over their geographic territories, but there was
significant variation in how resources were allocated. In Lowell and Newark, top management
was responsive to each commander’s needs, but functionally specialized units remained
centralized. In Minneapolis, however, several of these units were allocated to the districts and
put under the direct control of the district commander. This reassignment of specialized units to
the districts was a significant organizational development under Compstat.

Although Compstat is supposed to decentralize decision making to the district level, it
also continues to reinforce the concept of top-down control. District commanders in all three
sites were encouraged to take initiative, but top management was willing to exercise its authority
to override their decisions. The chief and his deputies were in charge of department task forces
and were ultimately responsible for the distribution of resources among districts. Furthermore,
top management intervened in deployment decisions and arbitrated problems in coordination.

Broadly speaking, both Compstat and community policing orient operational command
around the policing of specific territories. This explains the ease with which Compstat can be
grafted onto existing community-policing programs in departments, like Lowell and
Minneapolis, where management is structured geographically. Unlike community policing,
however, Compstat revitalizes centralization of command and control—a key feature of the
traditional police bureaucratic model (Weisburd et al. 2003). According to this model, top
management establishes performance criteria for middle managers, holds them accountable, and
controls resources. Compstat, therefore, represents a challenge to departments that have
community- policing programs and are consequently attempting to reengineer their command
structures to decentralize decision making to the street level.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 70

The tension between top management and district commanders over centralized versus
decentralized decision making was nicely encapsulated in one district commander’s summation
of the chief’s leadership style as, “You can do anything you want, as long as I agree with you.”
This captures the contradictions of a program that functions as a centralized information system
for monitoring and evaluating performance and as a mechanism for decentralizing decision-
making authority. Compstat decentralizes tactical decisions to district commanders but
centralizes up-to-date statistics that top management uses to hold district commanders
accountable and to reallocate resources (Mastrofski and Ritti 2000). In short, Compstat’s
principle of internal accountability limits the extent to which geographic organization of
operational command can be realized and vice versa, lending a zero-sum quality to the allocation
of decision-making authority.

Organizational Flexibility

Compstat demands that the police organization develop flexibility to deal with emerging
or unforeseeable problems. Our field observations showed that all three sites had used a variety
of informal and formal mechanisms to increase their capacity to shift resources to where they
were needed most. Lowell’s top leadership had sought to increase flexibility by encouraging
teamwork and coordination among districts and between districts and the department’s
specialized units. For the most part, however, the strategic reallocation of resources operated on
an ad hoc basis. Unlike Lowell, Newark and Minneapolis had established specific organizational
structures that facilitated resource reallocation. Newark responded to non-routine work demands
by temporarily creating department-level task forces focused on specific crime problems.
Minneapolis had made the most concerted attempt to enhance flexibility at the district level with
the formation of permanent taxi squads, such as the CRT, and the use of directed patrol. These
units were relieved from the burden of having to respond to every call for service so that they
could focus on their district’s emerging problems.

Unfortunately, geographic organization of operational command can reduce overall

flexibility. For example, the creation of centralized task forces in Newark siphoned off resources
from the districts, thereby limiting the resources at the district commanders’ disposal and
reducing within-district flexibility. Allocating additional resources to the districts did not solve
this dilemma. Between-district flexibility was reduced in Minneapolis with the creation of
specialized units that left fewer resources available for other department priorities and problems
that crossed district lines. One Minneapolis sergeant commented: “They take people from patrol
for specialty units like CRT and SAFE, and there are very few officers left for calls for service.”
In addition to these constraints, Compstat’s emphasis on accountability combined with
geographic organization of operational command to hinder teamwork and foster competition
among districts. Since district commanders were responsible for their own beats, they were
unlikely to share resources with each other (Willis et al. 2004b). Their reluctance to use
Compstat as a mechanism for the redistribution of resources combined with their department’s
use of more traditional criteria, such as population size, to further constrain flexibility.

Our fieldwork also suggested a source of conflict between community policing and one

of the approaches to increase organizational flexibility that we have previously discussed. In
Lowell, which has a nationally recognized community-policing program, the chief explained that

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 71

the department had rejected a centralized task force approach to policing in favor of a district-
level response. He explained that task forces were less familiar than district commanders and
their staff with the complexities of beat problems and often adopted a “heavy-handed” response
that was counterproductive. Consequently, task forces were rarely used in the department, thus
providing instance in which geographic organization of operational command trumped
organizational flexibility.

Data-Driven Problem Identification and Assessment

Compstat borrows from problem-oriented policing, a recent innovation that attempts to
shift policing from a system based on response-driven calls for service to one in which police use
data to structure decision making. We found it particularly noteworthy that all three police
departments had developed sophisticated information systems and formed crime analysis units to
assist them in the collection, processing, and analysis of timely crime data. In addition, our
fieldwork showed that district commanders and their executive officers consistently used data to
actively identify problems, establish priorities, and decide when and where to mobilize
responses. As one patrol officer stated bluntly, “Concern over information is different than ten
years ago when police officers did not give ‘two craps’ about the UCR.”

Despite this increased use of information to guide decision making, middle managers
were not given any special training to help develop their skills in crime analysis; nor were they
provided with any support staff trained in research methods, statistics, or crime mapping.
Regular Compstat meetings might have helped familiarize district commanders with the use of
data, but the meetings did not help them develop the kinds of skills that might have helped them
analyze data in order to detect plausible patterns. These limitations dulled the impact of the
”cutting-edge” technology that was available through the departments’ crime analysis units.
District commanders had become significantly better at rapidly identifying the location of
problems, but they did not analyze the data to figure out the specific cause of a problem and
decide exactly what to do. Moreover, the available data drew only on incident reports rather than
multiple information sources from other agencies. District commanders relied upon a serial
review of officer reports, rather than statistics, to form their impressions of crime problems. Even
when daily maps were available, they were little used. Furthermore, the identification of patterns
using descriptive statistics only occurred a short time before Compstat. In sum, data were used
strategically but district commanders bypassed much of the analysis so that they defined
problems narrowly and responded with traditional methods. By responding to the immediate
crisis and not implementing a preventive approach, Compstat resembled a more sophisticated
model of reactive policing, not a revolution in the use of crime analysis to solve or prevent
problems.

Compstat, community-oriented policing, and problem-oriented policing all share a focus
on the use of intelligence to broaden the scope and range of police activities (Ericson and
Haggerty 1997). The existing research shows that this data-driven approach has not advanced
beyond a rudimentary level in most police organizations that claim to have adopted problem-
oriented- and community-oriented-policing (Clarke 1998; Scott 2000), and our fieldwork
supports these findings. Police officers are craft workers and pride themselves on knowing the
tricks of their trade. If a chief really wants them to change their methods, he or she must

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 72

convince them that new methods will work better and not make their work environment less
pleasant (Mastrofski 1990).

Innovative, Problem-Solving Tactics

Silverman writes: “Compstat is the forum where new problem-solving approaches are
often presented, reviewed, analyzed, reexamined, and circulated. Many of the solutions are first
developed at the precinct level” (1999,193). Our research suggested that Compstat has
encouraged some innovation, but we did not observe a process in Lowell, Minneapolis, or
Newark that resembled Silverman’s description of the NYPD. Of course, traditional law
enforcement tactics might provide the best response, but this can only be determined under
Compstat after considering an array of other possibilities, such as collaborating with other
agencies, mediation, public education, mobilizing the community for social control, altering the
physical environment, enforcing civil law, and so on (Mastrofski and Ritti 2000, 191). Based on
our observations, Compstat had only made short inroads into previous practice due to the
pressure of internal accountability and the absence of a follow-up mechanism for evaluating
crime responses.

Compstat demands that middle managers do something quickly about crime problems,
though more innovative responses would require that district commanders have sufficient time to
foster, develop, and test long-term, preventive plans. Compstat has made this process more
difficult. The pressure of accountability, coupled with any lag between a crime’s occurrence and
its presentation at Compstat, limited the utility of follow-up. Rather than acting as an opportunity
for reflection and reassessment, follow-up concentrated on whether a district commander had
implemented a crime strategy. Furthermore, district- and department-level Compstat meetings
were treated as an opportunity to review what had already been done, not as a forum for
encouraging brainstorming and in-depth discussions about problem-solving alternatives. Some
innovation did occur, but it was fostered by non-Compstat mechanisms such as a chief’s
encouragement of creative thinking, as in Lowell, or the creation of task forces, as in Newark. In
the absence of these mechanisms, it was unlikely that innovative thinking would challenge the-
more traditional responses that were commonplace in Minneapolis.

Organizational environments that are characterized by a supervisory system that is
essentially negative undermine proactive, creative police work—a signature of both community-
and problem-oriented policing (Kelling 1999, 1-2). Although its supporters claim that Compstat
does not aim to cultivate traditional, reactive police responses,15 our fieldwork indicated that it
actually did. Its emphasis on internal accountability prized a rapid response, thus discouraging
innovation. A formidable challenge that Compstat presents to police leaders is how to increase
accountability without inhibiting an environment that tolerates reasonable trial and error. Failure
is ultimately a prerequisite for success, since it is rare for any organization to be successful the
first time it tries something new. Proponents of community policing have argued that this
obstacle can be overcome if departments recognize and accept officers’ good-faith efforts to

15 Jack Maple’s oft-quoted comment could be interpreted as trying to capture this quality: “Nobody ever got in
trouble because crime numbers on their watch went up … trouble arose only if the commanders didn’t know why the
numbers were up or didn’t have a plan to address the problems”(Maple 2000, 33).

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 73

solve problems, even if they are unsuccessful. A policy that allows creative failure encourages
innovative thinking and experimentation, and it could also be beneficial for Compstat.

VII. CONCLUSION

Compstat has been hailed as a way to make profound transformations in the way that
police departments operate. Many consider New York City’s pioneering Compstat program to
have introduced a revolution in American policing. Our research at three sites that followed in
New York’s footsteps suggest that what has taken place thus far is not a transformation so much
as a graft of some elements of progressive management onto fundamentally unaltered
organizational structures. Some of those new appendages appear to have flourished, and some
have not. Others are behaving in unexpected and possibly perverse ways.

Compstat’s creators and advocates present it as a way to transform sluggish,

unresponsive police organizations into focused, efficient, and smart organizations. The basic
transformational elements are:

• Motivated employees who are guided by a focused mission, disciplined, and stimulated

by a rigorous system of direct and personal accountability,
• Organizational nimbleness, or a capacity to deliver resources to places and at times that

will nip problems in the bud,
• Organizational decisions informed by knowledge of the problems that require attention

beyond that provided by normal organizational routines and facilitated by sophisticated
electronic information management and data-analysis systems,

• Problem-solving that reflects collaboration and exchange of ideas among members of an
organization, draws on research dealing with successful practices, and acknowledges the
experience of other agencies, and

• Decision making in an atmosphere that has an elevated tolerance for risk and encourages
new approaches to persistent problems.

Together, these transformational elements paint a picture of police organizations that are

capable of reading their work environment, discerning important trends within it, and acting
quickly in ways that deal with problems effectively. The realization of this vision would
constitute a profound transformation in how local U.S. police agencies do business, but such
transformations are not easily achieved. Our fieldwork suggests that the alterations to
fundamental organizational structures that would facilitate these changes were not fully in place
in our three study departments.

First, there was a gap in each agency between the theory of a highly focused

organizational mission of crime fighting and the reality of a complex mission with a complex set
of organizational structures that remained largely at odds with the ostensible simplification of the
department’s objective. Although Minneapolis provided some relief from the calls-for-service
apparatus that dominates the organization of patrol work in most urban police departments, this
framework remained virtually untouched by Compstat reform. Indeed, in Newark, rapid response
to calls for service was a key element in top leadership’s performance agenda for the department.
This calls-for-service apparatus is the principal means by which the predominantly non-crime

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 74

aspects of police work become part of a police department’s workload (Mastrofski 1983).
Concerns expressed by many patrol officers that Compstat encouraged officers to neglect this
aspect of their work is one indication of the extent to which there was discord between Compstat
and the unspoken, but powerful, message conveyed by continuing the traditional calls-for-service
mode of generating work for the largest sworn unit in the department. For the most part, other
parts of the organizations’ missions were not eschewed; the structural reality of police work in
these cities remained as complex as before, and Compstat provided only a rhetorical facade.

The police departments we observed struggled with the mandate to create a more nimble
organization that could move resources about strategically, or “put the soldiers where the battle
is.” Sometimes flexibility was achieved by decentralizing command and by giving district
commanders control of more personnel, such as detectives. Other times, it was attained through
use of overtime. Within this framework, organizational flexibility was achieved by one or more
of the following: (1) ad hoc changes in who did what—usually without altering fundamental job
assignment routines, such as officers’ shift assignments, (2) district taxi squads not bound by
permanent shift assignments, or (3) department-level task forces. Ad hoc changes and district
taxi squads appeared to work well for relatively small-scale problems, but problems requiring
larger resource allocations ran headlong into the limitations imposed by shift assignments, the
changing of which are governed by union contracts. In this regard, police departments are very
unlike military organizations, whose virtually complete freedom to alter work routines of the
rank and file make them, at least potentially, among the most flexible of groups. Department-
level task forces did enable the department to concentrate more resources on specific problems,
but they also reduced the routine availability of resources to the districts, thereby reducing
flexibility at that level. Indeed, this creates the inevitable dilemma that arises when a department
tries to enhance flexibility without increasing resources and simultaneously tries to concentrate
command of resources under the authority of district commanders. The best that top management
can do is to try to anticipate the level at which flexibility will be most beneficial and then
organize accordingly. Finally, pressures from political interest groups forced police managers to
allocate resources to places and jobs that would not have received priority according to the
dictates of Compstat. We can say, therefore, that these Compstat programs came to grips with the
challenge of organizational flexibility but were unable or unwilling to alter the fundamental
constraints that largely determine work schedules and allocation of personnel.

The new information technology and the emphasis Compstat placed on using it did
appear to have a striking effect on key strategic decision-makers in the three agencies. District
commanders, in particular, found that Compstat produced a profound change in the nature of
their work as managers. Before Compstat, middle managers did not routinely and proactively
pore over reports and scan maps to familiarize themselves with crime problems and identify
crime trends. This became a daily imperative, due in no small part to the accountability
mechanism to which these activities were so closely tied. Based on accounts of earlier practices,
Compstat appears to have made middle managers highly sensitive to the need to know in
considerable detail what was going on in their areas. This awareness in turn sensitized them to
the location and timing of crime spikes, which in turn affected when and where they mobilized
their resources. The considerable improvements in the timeliness of these hot spot data have
made the use of data far more meaningful to these police managers. This focus on identifying hot

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 75

spots is, of course, one of the more important objectives of Compstat, which seemed fully
realized at these sites.

Middle management’s immersion in the data, however, was of a limited sort. Most of
their daily data analysis was confined to familiarizing themselves with specific cases and looking
for links or patterns with other cases. The search for patterns took place primarily in the minds of
managers who relied on “mental maps” and other specific knowledge of people and places, the
traditional tools of the police officer’s trade (Rubinstein 1973; Van Maanen 1974). With the
exception of the maps—which were used less frequently—computer analysis was narrowly
restricted to the search for hot spots. Interest in areas and problems, and the attention accorded
them, usually disappeared as soon as crime levels declined and the next hot spot arose.
“Relentless” follow up really meant focusing on a hot spot only until it was no longer hot. This,
of course, greatly weakens the capacity of an organization to learn much about the long-term
effects of any intervention. Because attention shifted with the rise and fall of crime levels in
given areas, analysis was usually episodic and highly reactive, not proactive. In other words,
police did not respond to individual calls for service, but they did react to calls for service in the
aggregate. Long-term follow-up and tracking of the life course of geographic areas, regardless of
fluctuations in crime levels, was not a part of the crime-analysis routine, so departments did not
develop systematic and verifiable knowledge explaining when and why crime might have risen
or declined. Systematic data analysis that would examined and predicted hot spots was rarely
undertaken; use of crime data to devise new solutions occurred only occasionally; and spatial
modeling did not occur at all. It is fair to note that some of the more sophisticated methods of
spatial crime analysis are of unproven utility and not yet routinely available in software that is
widely marketed to the police (Anselin et al. 2000). It may also be noted, however, that none of
the Compstat programs provided middle managers with any special training in the analysis or use
of analyzed crime data.

Thus, Compstat at these sites did markedly energize middle managers to do something
about crime, but in many respects, the pattern that evolved mimicked the reactive forms of
policing about which advocates of strategic problem solving have complained. Compstat seemed
to engender a pattern of organizational response to crime spikes in hot spots that was analogous
to the Whack-a-Mole game found at fairs and carnivals. Moles pop up randomly from holes in
the game board, and the object of the game is to whack them with a paddle before they
submerge. A premium is placed on responding quickly rather than monitoring problem holes
continuously to try to discern patterns in the eruption of moles. Successful players scan the
entire board to make sure that they can quickly identify the next mole to surface. The pressure to
act decisively and nip hot spots in the bud may be an improvement over prior practice, but it
does not conform to more ambitious notions of how police can use data effectively to ascertain
the bigger picture and act proactively to get at underlying problems. It merely transfers to the
district commander the same sort of pressure that the patrol officer experiences to respond
rapidly to assigned calls for service.

Given the ostensible importance of strategic crime analysis to these Compstat programs,

it is noteworthy that departments failed to give their CAUs enough staff, training, and support.
As a result, the CAUs could only provide police decision makers with limited, and fairly
rudimentary crime analysis. This is not to demean the CAUs’ contributions, which are quite

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 76

impressive, especially given the many challenges involved merely to process data to a state that
the software could analyze. It does, however, indicate that the accoutrements of new information
technology and data analysis were introduced into the organizations without preparing their
members to make the most effective use of them.

Another partially realized innovation stimulated by Compstat in these organizations was
providing a forum that would encourage people to exchange ideas about crime control. This was
a frequently mentioned benefit according to middle and top-level managers who routinely
attended Compstat. Again, from accounts of prior practice, Compstat introduced collaborative
planning and problem solving in instances where it had once rarely occurred. The accounts of
Compstat attendees and our observations make it clear that collaboration of this sort increased
substantially because of Compstat.

However, there were definite limits to the nature, extent, and benefits of this
collaboration. The sort of vision of brainstorming and problem solving evoked by Compstat’s
advocates is the gathering of cabinet officials and generals during the Cuban Missile Crisis,
where there was a free give-and-take of views, ideas, and information (Allison 1971). The reality
of Compstat meetings rarely approximated that vision. The pressures on managers to come to the
Compstat meeting with problems identified and solutions already implemented were tremendous.
Further, managers tended to be reluctant to volunteer “helpful” suggestions for fear that they
would create problems for the person whose feet were being held to the fire of accountability—
lest their colleagues return the favor when their turn arrived. Most of these commanders,
moreover, were not in the habit of engaging in free-flowing debate because they had not been
socialized to a problem-solving culture that exalts criticism as a means to improve the quality of
management decisions. Theirs remained a culture in which the earliest socialization experiences
onward emphasized the importance of following orders and deferring to rank. Managers in our
three sites seldom sought the input and ideas of law enforcement personnel in other agencies or
other professionals outside the department. It was even more rare for them to resort to
researchers and their studies. Thus, the solutions selected tended to be parochial—that is, they
relied on what those in the agency regarded as tried-and-true methods.

Compstat also failed to alter the low tolerance for risk that pervades the culture of police
agencies and leads to parochial decision making. While this is by no means a feature unique to
police bureaucracies, it remains a major stumbling block to the selection of effective solutions to
emerging crime problems. Organizations that aspire to technical efficiency, such as effective
crime control, need to have proven technologies to accomplish that goal. This becomes
problematic because scientifically validated knowledge about what police can do to reduce crime
is quite limited. Our capacity to predict reliably the success of even the most promising strategies
is therefore limited (Eck and Maguire 2000; Sherman 1997). Although the police can scarcely be
held responsible for the limits of scientific knowledge of this sort, they are placed in a position
where they must blindly choose among the old methods or experiment with new tactics and
strategies. Such experiments are fraught with a risk of failure that can only be tolerated if there is
a sufficient buffer of faith or good will protecting those who are responsible for accomplishing
results (Mastrofski 1998, 187). The irony here, of course, is that Compstat shrinks the size of
that buffer substantially by requiring that organizational leaders broadcast a clear and concise
statement of mission to accomplish the very thing that is so problematic when it comes to

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 77

producing results. The need to seize upon innovative, high-risk strategies may not be so great in
times of declining crime. The real test of Compstat, however, will be in times of increasing
crime, for it is then that political pressure will be greatest and the stakes perhaps higher. Under
such circumstances, one might expect it to be better to risk failure, while being seen doing
something new and different, than to cling to the hope that the old tried and true strategies will
ultimately prevail. On the other hand, the record of police chiefs and their political overseers
appears to be largely to engage in more of the traditional law enforcement activities—more
suspect stops, more arrests, and more searches—when the crime rate rises precipitously.

Some who believe in the effectiveness of Compstat may be disappointed by the findings
and implications of this study because we have found that Compstat has not operated as much of
the prior literature, both promotional and evaluative, has indicated. Perhaps we have not studied
the departments that have been most successful in implementing Compstat. We certainly must
acknowledge that results could differ in other sites, but we take some confidence in the similarity
of our small fieldwork sample to patterns in our larger national survey. Indeed, we must
emphasize that many of the limitations of Compstat that we observed are built into the very
principles of the program’s design and are therefore inevitably replicated wherever it is faithfully
attempted. That is, there are contradictions and paradoxes that naturally arise from the effort to
create a program that aspires to overcome the many challenges involved in making police
organizations effective, strategic tools for crime control. There is, for example, an inevitable
tension between promoting heightened accountability for outcomes and engaging in the
experimental risk taking required to find effective methods, between individual accountability
and teamwork, and between organizational flexibility within districts and organizational
flexibility between districts. There is a zero-sum element to these paradoxes, making it
impossible to maximize both sides. That does not mean that police organizations are doomed to
failure, but it does mean that trade-offs of this sort will be inevitable. The art is in finding the
best compromises for each organization’s circumstances. The test of that would require an
evaluation that considers which set of compromises is most likely to produce the best crime-
control outcome, a task for future research.

We have suggested that Compstat does not represent a radical transformation in the way
these departments have done business. Instead, they have transplanted some new ways of doing
business without making much change to some very fundamental structures of police
organizations. This should not be surprising, for revolutions rarely come to U.S. police
departments—or any organizations. Change is occurring, but at a much slower, evolving pace.
To learn if the promises of Compstat will be fulfilled, future researchers should pay less attention
to what police leaders say and more attention to the changes that may or may not occur in the
fundamental structures of U.S. police organizations. Since the 1960s, the reform fashion has
been to embrace complexity in the police mission—largely through the community-policing
movement—and to decrease organizational flexibility by constraining management’s power over
resource allocation. One can only speculate on the impact of the domestic war on terrorism on
these trends, but it is hard to imagine that it will reverse them. On the positive side for Compstat,
the police have taken a lead role among justice agencies in embracing and using social science to
decide what problems need solving and how to do it. As they grow more confident and more
sophisticated in their ability to do so, it is possible that social science research will rise to meet

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 78

their expectations and thus provide a firmer foundation for the sorts of data analysis and
problem-solving applications envisioned by Compstat’s advocates and implementers.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 79

Appendix:

Line Officer Survey

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 80

Officer Survey Survey# __ __ __ __ __

The Police Foundation is conducting a study of Compstat. This research project is funded by the
Department of Justice. As part of the study, we are conducting a survey of officers in your department.
The survey is completely anonymous and you will not be identified in any report. The surveys will be kept
under lock and key at the Police Foundation. The survey will take about ten minutes to complete. We
greatly appreciate your participation. Please answer as many questions as you can. You may refuse to
answer any question or stop at any time. Again, thank you for your participation.

1. Please indicate whether each of the following is very important, somewhat important, or not at all

important to the Compstat strategy in the Lowell Police Department [or NAME of other police
department].

Very

Important

Somewhat
Important

Not at all
Important

Don=t
Know

a. Reduce complaints against officers

1“

2“

3“

8“

b. Reduce violent crime in Lowell

1“

2“

3“

8“

c. Improve the quality of life in Lowell

1“

2“

3“

8“

d. Arrest people committing misdemeanor offenses

1“

2“

3“

8“

e. Hold sector captains accountable for crimes in
their sector

1“

2“

3“

8“

f. Provide timely and accurate crime data

1“

2“

3“

8“

g. Respond quickly to calls for service

1“

2“

3“

8“

h. Identify crime patterns and choose appropriate
tactics

1“

2“

3“

8“

i. Respond quickly to emerging crime problems

1“

2“

3“

8“

j. Hold officers accountable for crimes in their beats

1“

2“

3“

8“

k. Follow up to assess whether solutions were
successful

1“

2“

3“

8“

l. Make officers and equipment available to different
sectors as needed

1“

2“

3“

8“

m. Encourage officers to take responsibility for their
beat

1“

2“

3“

8“

n. Resolve disputes among different segments of
the community

1“

2“

3“

8“

o. Create and maintain open lines of communication
with the community

1“

2“

3“

8“

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 81

2. In the last five years, has your department=s effectiveness in accomplishing each of the following

increased, decreased, or stayed about the same?

Increased

Stayed

About the
Same

Decreased

Don=t
Know

a. Reduce complaints against officers

1“

2“

3“

8“

b. Reduce violent crime in Lowell

1“

2“

3“

8“

c. Improve the quality of life in Lowell

1“

2“

3“

8“

d. Arrest people committing misdemeanor offenses

1“

2“

3“

8“

e. Hold sector captains accountable for crimes in
their sector

1“

2“

3“

8“

f. Provide timely and accurate crime data

1“

2“

3“

8“

g. Respond quickly to calls for service

1“

2“

3“

8“

h. Identify crime patterns and choose appropriate
tactics

1“

2“

3“

8“

i. Respond quickly to emerging crime problems

1“

2“

3“

8“

j. Hold officers accountable for crimes in their beats

1“

2“

3“

8“

k. Follow up to assess whether solutions were
successful

1“

2“

3“

8“

l. Make officers and equipment available to different
sectors as needed

1“

2“

3“

8“

m. Encourage officers to take responsibility for their
beat

1“

2“

3“

8“

n. Resolve disputes among different segments of
the community

1“

2“

3“

8“

o. Create and maintain open lines of communication
with the community

1“

2“

3“

8“

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 82

3. Listed below are ten activities for which you might receive praise from your superiors. Please

review all of these activities and then rank them from one to ten in terms of how much praise you
might receive. Write a one by the activity which would bring you the most praise, a two by the
activity which would bring the second most praise, and so on.

Rank

a. Identifying a developing crime pattern

b. Rapid response to 911 calls

c. Assisting crime victims in obtaining services and cooperating in
prosecutions

d. On-sight misdemeanor arrests of persons known to the police

e. Working with community members to solve local crime or disorder
problems

f. Recovering a weapon

g. Writing accurate crime reports

h. Working to provide legitimate activities for potential offenders

i. Making a large drug bust

j. Overall reduction of crime in your beat

4. Have you attended a Compstat meeting? Please check one.

1 “ Yes 2 “ No [SKIP TO Q6]

5. If AYes,@ please indicate what you did at the meeting. Please check all that apply.

1 “ Observed meeting only B did not participate
2 “ Assisted someone else make a presentation but did not speak
3 “ Answered questions
4 “ Offered an opinion or information
5 “ Made a short presentation (less than two minutes)
6 “ Made a long presentation (two or more minutes)

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 83

6. How often does your supervisor discuss what has happened at the Compstat meetings? Please
check one.

1 “ Every week
2 “ About once a month
3 “ Every few months
4 “ Never

7. To the best of your knowledge, how much has Compstat changed your job responsibilities? By
job responsibilities, we mean what you are expected to do on a daily basis. Please check one.

1 “ A great deal
2 “ Somewhat
3 “ A little bit
4 “ Not at all [SKIP TO Q9]
8 “ Don=t know [SKIP TO Q9]

8. Please describe any changes.

9. Please indicate your view of each of the following statements by marking the appropriate box.

Strongly
Agree

Agree

Disagree

Strongly
Disagree

No

Opinion

a. Compstat has made supervisors place
too much emphasis on statistics

1“

2“

3“

4“

5“

b. Compstat has made it possible for
officers to get credit for doing quality work

1“

2“

3“

4“

5“

c. Compstat has kept supervisors from
spending enough time on the street

1“

2“

3“

4“

5“

d. Compstat has made me more aware of
what goes on in other parts of the
department

1“

2“

3“

4“

5“

e. Compstat has increased teamwork
between my unit and specialist units in the
department

1“

2“

3“

4“

5“

f. Compstat will be an important feature of
the department=s organization in 5 years

1“

2“

3“

4“

5“

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 84

10. Overall, how would you rate the impact of Compstat on the department=s performance in serving

the public? Please check one.

1 “ Highly beneficial
2 “ Beneficial
3 “ No effect
4 “ Detrimental
5 “ Highly detrimental
8 “ Don=t know

11. Please briefly describe your view.

12. Overall, how would you rate the impact of Compstat on the department as a good place to work

as a police officer? Please check one.

1 “ Highly beneficial
2 “ Beneficial
3 “ No effect
4 “ Detrimental
5 “ Highly detrimental
8 “ Don=t know

13. Please briefly describe your view.

14. How many years have you worked for the Lowell Police Department in a sworn position?

1 “ Fewer than 3 years
2 “ 3-5 years
3 “ 6-10 years
4 “ 11-20 years
5 “ More than 20 years

15. What is your current rank

1 “ Police officer
2 “ Rank higher than police officer

16. What hours do you work? from ___:___ am/pm to ___:___ am/pm

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 85

17. How old are you?

1 “ Under 21 5 “ 40-49
2 “ 21 – 25 6 “ 50-59
3 “ 26 – 29 7 “ 60+
4 “ 30-39

18. What is your highest level of formal education? Please check one.

1 “ Some High School 5 “ Some Graduate School /
Law School

2 “ High School Graduate or GED 6 “ Masters Degree / JD or LLB.
3 “ Some College/A.A. Degree 7 “ Ph.D.
4 “ Bachelors Degree

19. Are you male or female?

1 “ Male
2 “ Female

20. What is your racial group? Please check one.

1 “ White
2 “ African American
3 “ American Indian or Alaska Native
4 “ Asian American or Pacific Islander
5 “ Other

21. Are you Hispanic or non-Hispanic?

1 “ Hispanic
2 “ Non-Hispanic

Thank you for completing the survey!

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 86

REFERENCES

Allison, Graham T. 1971. The Essence of Decision: Explaining the Cuban Missile Crisis.

Boston: Little Brown.

Anselin, Luc, Jacqueline Cohen, David Cook, Wilpen Gorr, and George Tita. 2000. Spatial

Analyses of Crime. In Measurement and Analysis of Crime and Justice, ed. David Duffee,
David McDowall, Brian Ostrom, Robert D. Crutchfield, Stephen D. Mastrofski, and Lorraine
Green Mazerolle, 213-62. Washington, DC: U.S. Department of Justice, Office of Justice
Programs, http://www.ncjrs.org/criminal_justice2000/vol_4/04e .

Beer, Michael. 1980. Organization Change and Development: A Systems View. Santa Monica,

CA: Goodyear Publishing Company.

Bouza, Tony. 1997. NYPD Blues – Good, Lucky, or Both? Law Enforcement News (January 31):

8,10.

Braga, Anthony A. 2001. The Effects of Hot Spot Policing on Crime. Annals of the American

Academy of Political and Social Science, no. 578:104-125.

Bratton, William J. 1998. Turnaround: How America’s Top Cop Reversed the Crime Epidemic.

With Peter Knobler. New York: Random House.

Bratton, William J. 1995. Great Expectations: How Higher Expectations for Police Departments

Can Lead to a Decrease in Crime. Paper presented at a National Institute of Justice
conference on “Measuring What Matters,” Washington, DC.

Brown, Michael K. 1988. Working the Street: Police Discretion and the Dilemmas of Reform.

New York: Russell Sage Foundation.

Chan, Janet B. L. 1997. Changing Police Culture: Policing in a Multicultural Society. New

York: Cambridge University Press.

Clarke, Ronald V. 1998. Defining Police Strategies: Problem Solving, Problem-Oriented

Policing, and Community-Oriented Policing. In Problem-Oriented Policing: Crime-Specific
Patterns, Critical Issues, and Making POP Work, ed. Tara O’Connor Shelley and Anne C.
Grant, 315-329. Washington, DC: Police Executive Research Forum.

Dussault, Raymond. 1999. The Taking of New York. Government Technology,

http://www.govtech.net/magazine/gt/1999/aug/brattonfldr/bratton.phtml.

Eck, John E., and Edward R. Maguire. 2000. Have Changes in Policing Reduced Violent Crime?

An Assessment of the Evidence. In The Crime Drop in America, ed. Alfred Blumstein and
Joel Wallman, 207-265. New York: Cambridge University Press.

http://www.ncjrs.org/criminal_justice2000/vol_4/04e

http://www.govtech.net/magazine/gt/1999/aug/brattonfldr/bratton.phtml

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 87

Eck, John E. 1993. Alternative Futures for Policing. In Police Innovation and Control of the
Police, ed. David Weisburd and Craig Uchida, 9-79. New York: Springer-Verlag.

Ericson, Richard, and Kevin D. Haggerty. 1997. Policing the Risk Society. Toronto: University

of Toronto Press.

Fogelson, Robert F. 1977. Big-City Police. Cambridge, MA: Harvard University Press.

Goldstein, Herman. 1990. Problem-Oriented Policing. New York: McGraw-Hill.

Greene, Jack. R. 2000. Community Policing in America: Changing the Nature, Structure, and

Function of the Police. In Policies, Processes, and Decisions of the Criminal Justice System,
ed. Julie Horney, Washington, DC: U.S. Department of Justice, Office of Justice Programs,
http://www.ojp.usdoj.gov/nij/criminal_justice2000/vol3_2000.html.

Greene, Jack. R., and Carl B. Klockars. 1991. What Police Do. In Thinking About Police:

Contemporary Readings, 2nd ed., ed. Carl B. Klockars and Stephen D. Mastrofski, 272-284.
New York: McGraw-Hill.

Greene, Judith A. 1999. Zero Tolerance: A Case Study of Police Policies and Practices in New

York City. Crime and Delinquency 45 (2): 171-87.

Grubler, Arnulf. 1991. Diffusion and Long-Term Patterns and Discontinuities. Technological

Forecasting and Social Change 39: 159-180.

Gurwitt, Rob. 1998. The Comeback of the Cops. Governing (January): 14-19.

Kelling, George L., and William H. Sousa. 2001. Do Police Matter? An Analysis of the Impact

of New York City’s Police Reforms. Civic Report no. 22 (December 2001): Center for Civic
Innovation at the Manhattan Institute, http://www.manhattan-institute.org/cr_22 .

Kelling, George L. 1999. “Broken Windows” and Police Discretion. Washington, DC: U.S.

Department of Justice, Office of Justice Programs.

Kelling, George L., and Catherine M. Coles. 1996. Fixing Broken Windows: Restoring Order

and Reducing Crime in Our Communities. New York: Martin Kessler Books.

Kelling, George L., and William J. Bratton. 1993. Implementing Community Policing: The

Administrative Problem. Perspectives in Policing, no. 13. Washington, DC: U.S.
Department of Justice, Office of Justice Programs.

Kelling, George L., and Mark H. Moore. 1988. From Political Reform to Community: The

Evolving Strategy of Police. In Community Policing: Rhetoric or Reality, ed. Jack R. Greene
and Stephen D. Mastrofski, 3-25. New York: Praeger.

http://www.ojp.usdoj.gov/nij/criminal_justice2000/vol3_2000.html

http://www.manhattan-institute.org/cr_22

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 88

Klockars, Carl B. 1988. The Rhetoric of Community Policing. In Community Policing: Rhetoric
or Reality, ed. Jack R. Greene and Stephen D. Mastrofski, 239-258. New York: Praeger.

Lehrer. Eli. 2001. The Police Behind America’s Biggest Crime Drop. American Enterprise

(March), http://www.findarticles.com/cf_dls/m2185/2_12/71004740/p1/article.jhtml.

Manning, Peter. 2001. Theorizing Policing: The Drama and Myth of Crime Control in the

NYPD. Theoretical Criminology 5 (3): 315-344.

Manning, Peter. 1992. Information Technologies and the Police. In Modern Policing, ed.

Michael H. Tonry and Norval Morris, 349-398. Chicago: University of Chicago Press.

Maple, Jack. 1999. The Crime Fighter: Putting the Bad Guys Out of Business. With Chris

Mitchell. New York: Doubleday.

Mastrofski, Stephen D., James J. Willis, and Jeffrey B. Snipes. 2002. Styles of Patrol in a

Community Policing Context. In The Move to Community Policing: Making Things Happen,
ed. Merry Morash and J. Kevin Ford, 81-111. Thousand Oaks, CA: Sage Publications.

Mastrofski, Stephen D. 2002. The Romance of Police Leadership. In Crime & Social

Organization, ed. Elin Waring and David Weisburd, 153-196. New Brunswick: Transaction
Publishers.

Mastrofski, Stephen D. and R. Richard Ritti. 2000. Making Sense of Community Policing: A

Theory-Based Analysis. Police Practice and Research Journal 1 (2): 183-210.

Mastrofski, Stephen D. 1998. Community Policing and Police Organization Structure. In How to

Recognize Good Policing: Problems and Issues, ed. Jean-Paul Brodeur, 161-189. Thousand
Oaks, CA: Sage Publications.

Mastrofski, Stephen D., and R. Richard Ritti. 1996. Police Training and the Effects of

Organization on Drunk-Driving Enforcement. Justice Quarterly 13: 291-320.

Mastrofski, Stephen D. 1993. “Eying the Doughnut: Community Policing and Progressive

Reform. American Journal of Police 12 (4): 1-18.

Mastrofski, Stephen D., and Robert Wadman. 1991. Personnel and Agency Performance

Appraisal. In Local Government Police Management, ed. W.A. Geller, 363-397.
Washington, DC: International City Management Association.

Mastrofski, Stephen D. 1990. The Prospects of Change in Police Patrol: A Decade in Review.

American Journal of Police: An Interdisciplinary Journal of Theory and Research 9 (3): 1-
79.

Mastrofski, Stephen. D. 1988. Community Policing as Reform. In Community Policing: Rhetoric

or Reality, ed. Jack R. Greene and Stephen D. Mastrofski, 47-67. New York: Praeger.

http://www.findarticles.com/cf_dls/m2185/2_12/71004740/p1/article.jhtml

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 89

Mastrofski, Stephen D. 1983. The Police and Noncrime Services. Measuring Performance in

Criminal Justice Agencies. Vol. 18. Sage Criminal Justice System Annuals, ed. Gordon P.
Whitaker and Charles David Phillips, 33-62. Beverly Hills, CA: Sage Publications.

McDonald, Phyllis Parshall, Sheldon Greenberg, and William J. Bratton. 2001. Managing Police

Operations: Implementing the NYPD Crime Control Model Using Compstat. Belmont, CA:
Wadsworth Press.

Miller, Rowland S. 1995. Embarrassment and Social Behavior. In Self-Conscious Emotions: The

Psychology of Shame, Guilt, Embarrassment, and Pride, ed. June Price Tangney and Kurt W.
Fischer, 322-339. New York: Guilford Press.

Moore, Mark H. 1992. Problem Solving and Community Policing. In Modern Policing, ed.

Michael H. Tonry and Norval Morris, 99-158. Chicago: University of Chicago Press.

Moore, Mark H., and Darrel W. Stephens. 1991. Beyond Command and Control: The Strategic

Management of Police Departments. Washington, DC: Police Executive Research Forum.

Reiss, Albert J. Jr., 1992. Police Organization in the Twentieth Century. In Modern Policing, ed.

Michael H. Tonry and Norval Morris, 11-97. Chicago: University of Chicago Press.

Rubinstein, Jonathan. 1973. City Police. New York: Farrar, Straus and Giroux.

Scott, Michael S. 2000. Problem-Oriented Policing: Reflections on the First 20 Years.

Washington, DC: U.S. Department of Justice, Office of Community Oriented Policing
Services.

Sherman, Lawrence W., and David Weisburd. 1995. General Deterrent Effects of Police Patrol in

Crime ‘Hot Spots’: A Randomized Controlled Trial. Justice Quarterly 12: 625-48.

Sherman, Lawrence W. 1998. Evidence-Based Policing. Ideas in American Policing. The Police

Foundation: Washington, DC, http://www.policefoundation.org/pdf/Sherman .

Sherman, Lawrence, Denise Gottfredson, Doris Mackenzie, John Eck, Peter Reuter, and Shawn

Bushway. 1997. Preventing Crime: What Works, What Doesn’t, What’s Promising?
Washington DC: National Institute of Justice.

Silverman, Eli B. 1996. Mapping Change: How the New York City Police Department Re-

engineered Itself to Drive Down Crime. Law Enforcement News (December 15),
http://www.lib.jjay.cuny.edu/len/96/15dec/html/12.html.

Silverman, Eli B. 1999. NYPD Battles Crime: Innovative Strategies in Policing. Boston, MA:

Northeastern University Press.

http://www.policefoundation.org/pdf/Sherman

http://www.lib.jjay.cuny.edu/len/96/15dec/html/12.html

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 90

Simons, Robert L. 1995. Control in an Age of Empowerment. Harvard Business Review 72 (2):
80-88.

Sparrow, Malcolm K., Mark H. Moore, and David M. Kennedy. 1990. Beyond 911: A New Era

for Policing. New York: Basic Books.

Swope, Christopher. 1999. The Comstat Craze. Governing: The Magazine of States and

Localities 12 (12): 40.

Trojanowicz, Robert C., and Bonnie Bucqueroux. 1990. Community Policing: A Contemporary

Perspective. Cincinnati, OH: Anderson Publishing Company.

U.S. National Agricultural Library. 1998. Compstat: A Crime Reduction Tool,

http://www.nalusda.gov/pavnet/iag/cecompst.htm.

Van Maanen, John. 1974. Working the Street: A Developmental View of Police Behavior. In

The Potential for Reform of Criminal Justice, ed. Herbert Jacob, 83-130. Beverly Hills, CA:
Sage Publications. .

Vollmer, August. 1933. Police Progress in the Last Twenty-Five Years. Journal of Criminal

Law, Criminology, and Police Science 24 (1): 161-175.

Walker, Samuel, and Charles M. Katz. 2002. The Police in America. 4th ed. Boston, MA:

McGraw-Hill.

Walsh, William. 2001. Compstat: An Analysis of an Emerging Police Paradigm. Policing: An

International Journal of Police Strategies and Management 24 (3): 347-363.

Wasserman, Robert and Mark H. Moore. 1988. Values in Policing. Perspectives in Policing. no.

8. U.S. Department of Justice and the John F. Kennedy School of Government, Harvard
University.

Weisburd, David, Stephen D. Mastrofski, Ann Marie McNally, Rosann Greenspan, and James J.

Willis. 2003. Reforming to Preserve: Compstat and Strategic Problem Solving in American
Policing. Criminology and Public Policy 2 (3): 421-456.

Weisburd, David and Anthony A. Braga. 2003. Hot Spots Policing. In Crime Prevention: New

Approaches, ed. Helmut. Kury and Joachim Obergfell-Fuchs. Manz, Germany: Weisser Ring.

Weisburd, David, Stephen Mastrofski, Annmarie McNally, and Rosann Greenspan. 2001.

Compstat and Organizational Change: Findings from a National Survey. Report submitted to
the National Institute of Justice by the Police Foundation.

Willis, James J., Stephen D. Mastrofski, and David Weisburd. 2004a. Compstat and

Bureaucracy: A Case Study of Challenges and Opportunities for Change. Justice Quarterly
21 (3): 463-496.

http://www.nalusda.gov/pavnet/iag/cecompst.htm

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 91

Willis, James J., Stephen D. Mastrofski, David Weisburd, and Rosann Greenspan. 2004b.

Compstat and Organizational Change in the Lowell Police Department: Challenges and
Opportunities. Washington DC: The Police Foundation.

Wilson, James Q., and George L. Kelling. 1982. Broken Windows: The Police and

Neighborhood Safety. The Atlantic Monthly 249 (3): 29-38.

Wilson, James Q. 1968. Varieties of Police Behavior. The Management of Law and Order in

Eight Communities. Cambridge, MA: Harvard University Press.

Wilson, Orlando Winfield, and Roy Clinton McLaren. 1972. Police Administration. 3d ed. New

York: McGraw-Hill.

Witkin, Gordon. 1996. Making Mean Streets Nice: Computer Maps that Take the ‘Random’ Out

of Violence. U.S. News and World Report (December 30): 63.

Compstat in Practice: An In-Depth Analysis of Three Cities

Police Foundation 92

AUTHORS

James J. Willis is Assistant Professor of Sociology at the University of Massachusetts-Boston.
His research focuses on the impact of reforms and technological innovations on police
organization and practice. His other research area is punishment and penal history.

Stephen D. Mastrofski is Professor of Public and International Affairs at George Mason
University and Director of the Administration of Justice Program and the Center for Justice
Leadership and Management. His research focuses on police and police organizations.

David Weisburd is Professor of Criminology at the Hebrew University Law School in
Jerusalem, and Professor of Criminology and Criminal Justice at the University of Maryland,
College Park. He is also a Senior Fellow at the Police Foundation and Chair of its Research
Advisory Committee.

Table of Contents

Foreword

Acknowledgments

List of Tables & Figures

Table 1: Crime index in Lowell, Minneapolis, and Newark (1996-2001)

Figure 1: Crime rates in three cities

Table 2: The importance of reducing violent crime to the department's Compstat strategy

Table 3: The importance of holding district commanders accountable for crimes in their districts to the department's Compstat strategy

Table 4: The importance of holding officers accountable for crimes in their beats to the department's Compstat strategy

Table 5: In your view, Compstat has made supervisors place too much emphasis on statistics

Table 6: In your view, Compstat has increased teamwork between your unit and specialized units in the department

Table 7: Summary comparison of Compstat elements across sites

Introduction

Background on Compstat and Its Core Elements

Research Sites

Data and Methods

Analysis of Departments' Experiences with Compstat

Mission Clarification

Internal Accountability

Geographic Organization of Operational Command

Organizational structure

Decision-making authority

Coordination issues

Geographic vs. temporal organization

Organizational Flexibility

Rivalry among districts

Manpower

Traditional approaches to resource allocation

City politics

Strategies to increase flexibility

Data-Driven Problem Identification and Assessment

Data collection

Data analysis and presentation

Use of data

Innovative Problem-Solving Tactics

Problem solving and brainstorming during Compstat meetings

Follow-up

Traditional responses

Innovative responses

Discussion

Mission Clarification

Internal Accountability

Geographic Organization of Operational Command

Organizational Flexibility

Data-Driven Problem Identification and Assessment

Innovative, Problem-Solving Tactics

Conclusion

Appendix: Line Officer Survey

References

Authors