CJPS2

Policing Communities: What Works? Author(s): Lawrence W. Sherman Source: Crime and Justice, Vol. 8, Communities and Crime (1986), pp. 343-386 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/1147432 Accessed: 26-03-2017 21:27 UTC

REFERENCES Linked references are available on JSTOR for this article: http://www.jstor.org/stable/1147432?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted

digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about

JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://about.jstor.org/terms

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Crime and Justice

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Lawrence W. Sherman

Policing Communities: What Works?

ABSTRACT

Communities vary widely in the ways in which police exercise discretion; in their resources; in their mixtures of public police, private police, and voluntary citizen policing; in the nature and severity of problems requiring policing; and in the physical geography that shapes options for policing. Yet communities vary relatively little in basic police strategy. Residential neighborhoods in American metropolitan areas are policed primarily by officers driving around in unfocused patrol, largely waiting for calls for service. Public police should probably supplement the single-complaint strategy in most communities with a variety of mixed-strategy models. The specific strategies employed in any particular community should be based on the community's unique characteristics. Recent research suggests some conclusions about the relative effectiveness of various policing strategies in general, but it says little about interaction effects between strategies and neighborhoods. Informal experimentation may provide guidance for fitting strategies to the characteristics and problems of specific communities.

Policing varies substantially across communities. This variation is best

documented at the citywide level of analysis by systematically observed

police discretion (Black and Reiss 1967; Friedrich 1977; Smith 1984) and officially recorded rates of arrests per reported offense (Wilson 1968; Swanson 1978); of arrests per capita (Gardiner 1969; Wilson and Boland 1978); and of citizens killed by police (Robin 1963; Matulia 1982; Blumberg 1983; Sherman and Cohn 1986a). Such variation has also

Lawrence W. Sherman is Professor of Criminology and Director of the Center for Crime Control at the University of Maryland. Albert J. Reiss, Jr., Franklin Zimring, and Ronald V. Clarke provided helpful comments on earlier drafts.

? 1986 by The University of Chicago. All rights reserved. 0-226-80802-5/86/0008-0010$01.00

343

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

344 Lawrence Sherman

been documented in neighborhoods within cities. Fyfe (1978) has shown enormous variation across New York City precincts in the rates

at which police use deadly force. Croft (1985) has suggested substantial variation across Rochester, New York, patrol divisions in the use of nonlethal force. And Smith (in this volume) has presented systematic observation evidence on variation in arrests, use of force, and crime

reporting across neighborhoods.

This essay addresses the consequences of variations in police practice

that result from planned and purposive efforts of policymakers. The question of which policing strategies work best to control crime (or to

improve the quality of life, or to improve police-community relations, or to reduce public fear of crime) has received increased attention over

the past two decades, especially at the community level of analysis. But the question has never been formulated in terms of the difference that

community context makes in police effectiveness. Do different types of

policing produce different gains and costs in different types of com-

munities? What interaction effects, if any, exist between community characteristics and policing in shaping police effectiveness?

Unfortunately, little systematic evidence can be brought to bear on

these questions. The measurement of interaction effects requires ade-

quate sample sizes, and community-level evaluations of policing prac- tices are woefully short on sample size even for main effects. Alterna-

tively, it requires replication of experiments that test virtually identical

policing strategies in many different types of communities. These ques-

tions can be approached through a review of the literature on police effectiveness in general, focusing on the implications it may hold for

interaction effects of policing and community.

This essay defines "policing" in the means-oriented approach of Egon Bittner (1970) and Carl Klockars (1985). Policing clearly encom- passes "institutions or individuals given the general right to use coercive

force by the state within the state's domestic territory" (Klockars 1985,

p. 12). But it should also encompass persons or institutions watching for crime, threatening to act if crime is committed, or mobilizing others

to act. This definition allows this essay to focus on the question of which approaches to policing communities work best in controlling crime.

There is both more and less variation in the policing of communities than the existing literature describes. There is more variation in the mix

of organizational structures of policing arrangements than the literature suggests, which implies much greater variation across communities in

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 345

resources for policing than is apparent in the literature's concentration

on the number of public police officers or the size of the public police

budget. There is arguably less contemporary variation across com- munities in basic police strategies than the literature on discretion seems to imply.

Section I of this essay presents a model of the interaction between community characteristics and policing. Section II develops the model with an inventory of the basic choices of organizational structure, re-

sources, and strategy in policing communities. Section III reviews the recent research critique of the prevailing "single-complaint" policing strategy found in the majority of metropolitan residential communities.

Some possible alternatives to that strategy are examined from the per-

spective of public police, citizens, and private police. This essay con- cludes with some hypotheses and tentative policy recommendations on

how police should take community characteristics into account. The essay excludes many of the strategies that public police depart-

ments employ--including detective units, vice squads, hostage negoti- ation units, tactical units, and harbor police-for two reasons. First, there is almost no published evidence on the effectiveness of these units

(but see Eck 1986). Second, these units are not usually designed to operate at a neighborhood level. The value in considering the difference

that community context can make to policing is found largely in the "patrol" function, whether performed by citizens, public police, or private police.

I. Varieties of Community Policing This essay assumes that there is substantial interaction between the characteristics of each community and the way it is policed. There is enormous variation in policing across cities and within cities across neighborhoods. There is variation in the "natural" characteristics of the

community, ranging from physical geography and demographics to political power. There is also variation in the decisions made by police executives and others constrained by community characteristics about

policing in each community. Those decisions largely determine the structures, resources, and strategies of policing each community.

The consequences that strategic decision makers expect from polic- ing a specific community in a given manner may vary. The selection of options for community policing may depend heavily on the theories or research that decision makers rely on in predicting the effects of various police actions. These anticipated consequences include both costs and

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

346 Lawrence Sherman

benefits, such as judgments about how harsh a crackdown on crime is

feasible without adverse effects on community relations or pressures from specific interest groups.

A third interaction is the feedback of actual and perceived conse- quences of community policing on the specific problems it may ad- dress. What community residents and organizations may think about the effectiveness of their community's policing, or what police officials

may find in tracking crime trends or citizen complaints, or what news-

papers or citywide interest groups may say can all be important in changing or sustaining the policing practices of a specific community.

Some examples may help to clarify each of these points. Wilson (1968) has shown, for example, that some cities are far more likely than are others to make arrests for public intoxication. Sherman and Cohn

(1986b) have shown that cities vary widely in their responses to minor

domestic violence, from encouraging arrest to letting police officers ignore the problem. Both kinds of variation may be due to differences

in political pressures or political values in different cities.

Substantial differences in police practices can be found also at the neighborhood level within cities (e.g., Westley 1970, p. 97). Such vari- ation is a joint product of what police expect will be the consequences of their differing behaviors in different neighborhoods and of what the

political power of the different neighborhoods will let them get away

with. A neighborhood that objects to police behavior may or may not

be able to generate sufficient political pressure to change that behavior.

Neighborhood policing strategies aimed at solving specific problems

are often generated by the changing power bases of different neighbor-

hoods. Gentrification of previously poor areas, for example, can gener-

ate substantial pressure on police to eliminate street prostitution that may have a long neighborhood history.

Neighborhood policing strategies can also change simply because an old problem seems to be getting worse. New York City police under- took a massive effort to suppress drug dealing in the Lower East Side of Manhattan in 1984-85 after the problem received renewed media atten-

tion. Washington, D.C., police in 1985 cracked down on parking in Georgetown, which was closely linked to disorder problems caused by teenagers drinking in and around parked cars on main streets.

Some would argue that these kinds of neighborhood variations in police practices are widespread in American cities, with police officials and officers constantly adjusting their practices according to changing community characteristics. Others argue, as I do, that organizational

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 347

structures and resources may vary widely across neighborhoods but that the basic strategy of putting the bulk of police resources into rapid

response to dispatched calls for service varies little. Few would disagree

that there could be greater diversity of police strategies across neighbor-

hoods. There is ample room for more mixing of approaches and more sensitivity to community differences.

II. Basic Choices in Policing Communities Basic choices are made in each community about at least three aspects

of policing-the organizational structures of policing, the nature and the level of policing resources, and the strategies used to deploy those

resources. These choices are obviously interrelated since organizational

structure choices may limit the availability of resources, which in turn

may limit strategic possibilities. The choices may vary substantially over time. Because all these choices are not made by any centralized decision-making body, it is always difficult to measure exactly which choices any given community has made.

These choices are made for each community by public police offi- cials, private business property owners, and residents. Some commu- nities can afford a great deal of private policing; others cannot. Some

communities are public spirited enough to generate substantial volun- teer labor for policing; others are too atomized. Some communities have substantial numbers of young people or others creating disorder problems on the streets; others have little visible crime or disorder.

Some communities feature a great deal of commercial and business activity; others are exclusively residential. The physical geography of communities creates different needs and limitations for policing; popu- lation and automobile density, type of housing structure, presence or

absence of sidewalks, and street patterns vary greatly. Some com- munities are highly dense, with much foot traffic and ample opportuni-

ties for surveillance of public places from private residences. Other communities are geographically spread out, with few public or private

places under surveillance. These differences determine the range of choices that can be made for community policing and create different requirements for an appropriate mix of policing strategies.

A. Organizational Structures

The ideal mixture of policing produced by private organizations, public organizations, or citizen volunteers has long been a central pol- icy issue. The solution has important implications for police efficiency

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

348 Lawrence Sherman

and effectiveness and for the accountability of police use of force and

the character of a democratic society.

At the heart of the debate are the twin problems of what government

should do and how much tax money it should spend. In the "taxpayers'

revolt" atmosphere of the late 1970s and early 1980s, it was fashionable

to call for greater voluntary citizen participation in policing (e.g., Lav-

rakas 1985). At the same time, privately paid policing efforts have grown rapidly (Cunningham and Taylor 1985). The reality of modern America is a mixed model of citizen volunteers, privately paid guards, and full-time publicly paid police, with substantial variation across communities.

Klockars (1985) distinguishes different organizational systems for policing according to whether policing is vocational (a paid, full-time job) or avocational (unpaid and usually not full-time), whether it is done

on an obligatory basis by governmental order (such as paying taxes) or

done voluntarily, and whether it is done for fixed wages or on a fee-for-

service basis. Table 1 similarly distinguishes different "structures" of policing in terms of who does the work and why. Much of the table is

derived from the legal distinctions among the powers and obligations of the different actors performing police services. The table omits many

important differences among the models, such as the nature of their hierarchy and accountability, the organization of their intelligence about crime, and the process by which they are mobilized.

The first structure, occasional voluntary citizen policing, may vary widely among communities. Citizen's arrest is the most dramatic ex-

ample and is far from rare. A study of Part I arrests in Kansas City immediately after the offense found that 20 percent were made by

TABLE 1

Typology of Police Organizational Structures

Who Does the Work?

Why? Citizens Private Police Public Police

Voluntary: Occasional Citizen's arrest Off-duty arrest Off-duty arrest Regular Citizen patrols Take-home cars Take-home cars

Obligatory: Occasional Hue and cry Felony in presence Felony in presence Regular Watch Foreseeable crime Special request

Contract:

Public funds ... "Privatization" City police Private funds ... Private security Special details

Rewards/fees:

Public funds Informants Bounty hunters Private funds Crimestoppers Private detectives Moonlighting

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 349

citizen volunteers at the scene (Kansas City Police Department 1977, p.

29). Another study found that burglars in one jurisdiction have a better chance of being shot by their victims than they do of going to prison (Kleck [1979], as cited in Wright, Rossi, and Daly [1983], p. 139). Such "self-help" (cf. Black 1983) is central to political debates over gun con-

trol and "vigilantism."

The voluntary occasional policing of off-duty police and security guards can also be quite substantial. One-fourth of the 239 officers killed in the line of duty from 1844 to 1978 in New York City were off

duty (Margarita 1980, p. 83). Other studies found that 12-17 percent of

the citizens killed by police are killed in off-duty incidents (Milton et al.

1977, p. 27; Fyfe 1980, p. 73). A substantial regular and predictable police presence may also result when public or private police officers are permitted to take marked police cars home and use them for per- sonal business.

Regular voluntary citizen patrols in this country date back to the colonial watch system and were revitalized by civil defense work of the

two world wars. Some cities, like New York, continued those patrols in

uniform for trained auxiliary police; as of 1986, 8,700 volunteers were

certified and armed with nightsticks for regular patrol to supplement

some 25,000 paid police officers (Bird and Dunlap 1986). New York and other cities have also encouraged untrained and uncertified citizens

to patrol in plainclothes as part of the Neighborhood Watch program.

These programs seem particularly strong in middle-class or lower- middle-class, ethnically transitional neighborhoods (Marx and Archer 1971). The Guardian Angels movement of young people patrolling subways and other public places in groups (Klockars 1985, pp. 33-34) seems strongest in low-income minority neighborhoods.

Obligatory policing is now virtually gone as a matter of criminal law

but has recently been revived in the civil law. The obligation of private

citizens and even of public police to police public places has all but disappeared, but the obligations of property owners to their invitees and of public police to those who call on them have intensified. The expansion of tort law in recent years has affected the legal obligations of both private and public policing. Since entertainer Connie Francis won

over a million dollars in a lawsuit against a motel where she was raped (Garzilli v. Howard Johnson's Motor Lodges, Inc., 419 F. Supp. 1210 [1976]), many more such lawsuits have been filed and won (Sherman and Klein 1984). State supreme courts have upheld and expanded the duty of landlords to police their property adequately against foresee- able crimes committed by third parties (e.g., Isaacs v. Huntington Memo-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

350 Lawrence Sherman

rial Hospital, 695 P.2d 653 [Cal. 1985]) and have upheld the duty of public police departments to respond effectively to requests for assis- tance. Failure to meet these legal obligations to provide police protec- tion has, in recent years, produced a $2 million award against the New

York police for failing to investigate a complaint of possible child abuse

("Court of Appeals Tells City to Pay $2 Million to Girl Father Stabbed"

1985); a $2.3 million verdict against a Connecticut police department for failing to arrest a wife beater ("Officers Must Pay $2.3 Million to Wife

Maimed by Husband" 1985); and a decision against a Washington state police department that did not respond quickly enough to a 911 call to

keep the callers from being attacked (Chambers-Castanes v. King County,

669 P.2d 451[1983]). The extent of these new obligations varies from state to state and practically across communities.

"Contract" policing denotes policing as a full-time job, generally, in a

bureaucratic organization. The most common version is publicly funded and publicly operated in municipal and county police agencies.

The latest and still quite rare innovation in police structure is pub- licly funded and privately contracted policing in which public tax levies

are used to retain a private, for-profit firm to provide basic police services. Reminderville, Ohio, became one of the first communities to

employ this "privatization" arrangement with its small ($90,000 per year) police budget when the county sheriff raised the fee it would charge the city to twice that amount (Gage 1982). Although the ar- rangement was later discontinued because of persisting questions about

the legal authority of private employees acting as city police officers (Tolchin 1985a), it has been employed in several small cities in Florida (George Zoley, personal communication, 1985). In a more common variant, communities contract with other public police agencies, which

often creates competition among the candidate agencies and permits a

community to change its police force almost overnight.

Privately funded and operated contract policing personnel now out-

number publicly funded personnel by over two to one (Cunningham and Taylor 1984, p. 3). Many work behind factory walls and apart from the community, but many are employed to protect public places such as parking lots, hospitals, and hotels. Many upper-middle-class and upper-class residential communities retain such patrol services on a

collective basis, and the millions of individual home owners purchasing "central station" burglar alarms are paying, in effect, for a rapid re- sponse by private police to a signal from their home.

Less attention has been given to the privately funded public police

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 351

activities, ranging from downtown Oakland property owners commit-

ting $300,000-$400,000 annually to the police department for extra downtown patrols (Reiss 1985, p. 18) to private interests in old north-

eastern cities hiring police to work overtime at construction sites, enter-

tainment events, and late-night commercial establishments (Tolchin 1985b). In some cities, the number of public police on duty at any given

time performing private tasks may be greater than the number perform-

ing public tasks.

Although Anglo-American policing once relied heavily on publicly funded rewards and fees to support apprehension of criminals (Rad- zinowicz 1956, p. 33), the role of such incentives is greatly restricted today. The "thief taker" has been replaced with the informant, who is

generally rewarded with leniency more often than with money.

Privately funded reward policing is probably stronger now than it has been since the nineteenth century. The most visible example is the

"Crimestoppers" program organized by business people in 443 cities to

offer rewards for the solution of highly publicized crimes (Rosenbaum,

Lurigio, and Lavrakas 1985). The Crimestoppers program uses free television or radio time to describe the facts of a crime shortly after it

occurs and announces a telephone number that informants can call to arrange to provide information leading to the arrest of the guilty par-

ties; if a conviction (or, in some cases, an arrest or even just recovery of

property) results, the informant will be rewarded, with anonymity guaranteed (Rosenbaum et al. 1985). Such activity is more visible but probably smaller than the privately funded rewards or fees paid on a case-by-case basis to professional detectives.

Many of these structural choices are made at the city rather than at

the neighborhood level. But even such citywide structures as a Crime-

stoppers program may be used more by citizens in some communities

than in others. A citywide policy of allowing police to take cars home

would create more policing in some communities (where police live) than in others (where they do not). Other structural choices are made

by many individual decision makers, ranging from the operators of a shopping center who decide to hire private guards to citizens who chase a purse snatcher.

The possible combinations of policing structures present in any com- munity are numerous and complex. The combinations may also be highly transitory, although some structures are clearly more stable than

are others. There are limitations to the possible combinations set by both community resources and the public interest. Too much reliance

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

352 Lawrence Sherman

on privately funded policing would leave poor areas greatly disadvan- taged and could make people "prisoners" of well-guarded private places. As Reiss (1985, p. 17) points out, it may be more important to put our social resources into creating a safer public environment than

into creating private "fortresses." Similarly, we cannot rely too much

on voluntary citizen "self-help," given the difficulty of controlling citi-

zen use of force and the virtual absence of residents from many neigh-

borhoods during working hours. Whatever the combination of organi- zational structures, however, it will influence the level of resources

allocated to policing and the probability that crime in the community will be deterred or intercepted.

None of this touches on the variation within public police organiza-

tion. Cities vary widely in how their patrol forces are organized; some

use a primarily temporal structure, with citywide command changing every eight hours. Others use community-oriented structures of com-

mand, with area commanders given twenty-four-hour, seven-day re- sponsibility for all patrol work in their community. Both cities and communities also vary in their levels of police corruption and theft of

time (loafing or doing personal errands while on duty), which also affect

the actual level of resources brought to bear on community problems.

B. Resources

Perhaps the most important policy question about police in any city

is the number of police officers the department should have. On a per

capita basis, cities vary by as much as 300 percent: big cities range roughly from one to three police officers per 1,000 citizens (Federal Bureau of Investigation 1985). At the community level, however, the variation can be much greater. Harlem, for example, at one point had

ten times as many officers per 1,000 citizens as had Bay Ridge, a white,

middle-class section of Brooklyn. Yet Bay Ridge had many more police officers per reported crime than had Harlem (Ruth 1971).

Police managers in the 1950s developed mathematical formulas for determining how many police each community "needed" based on re- ported crime, calls for service, and other quantifiable factors (Levine and McEwen 1985). In the 1970s, more sophisticated models of patrol allocation were developed that used queuing theory to maximize the efficiency of patrol car responses to emergency calls for service (Larson 1972; Chaiken 1975). But the structures of policing presented above

should also be taken into account in measuring the level of public

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 353

resources allocated to policing in any given community. The presence of other structures may reduce the need for public police officers.

Resource levels vary by factors other than the number of police officers. The number of private police assigned to work in a community either for commercial establishments or under a contract from the home

owners is an obviously important factor. So is the number of public and

private police who reside in a community, whether they bring their cars home, and whether they are employed after hours on special details

in the community. The relative involvement of citizens in making ar-

rests, conducting patrols, and providing police with information also affects the resource levels of community policing.

Some of these alternative structures may be strongest in communities

where the public police are present in strength. The same social factors

that enhance citizen policing may also enhance community effec- tiveness in lobbying for a larger number of officers. In the Hasidic community of Borough Park in Brooklyn, for example, the citizens often mount a hue and cry to apprehend muggers or purse snatchers.

At the same time, community leaders work hard at election time to earn

political capital, much of which is spent on obtaining more police pres- ence. In the heavily Polish-American northeastern section of Min- neapolis, there is substantial volunteer citizen surveillance and little reported crime. But the police chiefs 1985 reduction in the number of

officers assigned there touched off a bloody political battle that re- sulted, among other things, in the chiefs temporary suspension by the mayor.

Nonetheless, the strong trend of recent years is toward increased resources for the organizational structures other than public police de-

partments. Private police have tripled in strength since 1950 (Cunning-

ham and Taylor 1984, p. 3). A 1982 Gallup poll reported that 17 percent of a national sample said they were aware of some sort of volunteer crime prevention effort in their neighborhood (Sherman 1983, p. 145), which seems likely to be a large rise over recent years.

The rise of the alternative structures accompanies a decline in the resources allocated to public police. This decline contrasts sharply with the steady growth in prestige and budgets throughout much of the twentieth century, especially the years after the riots of the late 1960s, when both salary levels and numbers of police officers increased dra- matically. The growth curve of public police has been flat and has even declined slightly since the late 1970s (Bureau of Justice Statistics 1986).

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

354 Lawrence Sherman

Massive layoffs in New York, Detroit, and other cities, once inconceiv-

able under any circumstances, were accomplished in the face of rising crime. This may help explain the rise of alternative structures.

C. Strategies

Policing strategies consist of four basic elements: information gather-

ing, mobilization of action, the nature of the action, and the target, or

focus, of the action. Unlike the structures and resources of policing, which vary widely across communities, these strategic elements of po-

licing have become remarkably similar in the public police agencies in

urban communities around the country. Any progress in making polic-

ing more effective may depend heavily on our ability to diversify these strategies and to make them more sensitive to the characteristics of local communities.

1. Information. Policing is essentially an information-processing en-

terprise, which produces a relatively small amount of people process- ing. Most police work consists of seeking out information about poten-

tial trouble and troublemakers, communicating information about the

potential consequences of troublemaking, recording and analyzing in-

formation about trouble that has aliready occurred, and presenting in- formation about accused and apprehended troublemakers to judicial authorities. The technology and organization of these information sys-

tems are important factors shaping the effectiveness of the police. The first public police departments at Scotland Yard and in New

York had little technology for communicating crime reports rapidly.

The geographic dispersion of large numbers of officers throughout a dense city encouraged citizens to report crimes to the police by creating

a virtual "wall-to-wall-cops" system (Reiss 1971) in which it was possi-

ble to run into the street, shout "police," and have a foot-patrol officer respond fairly quickly. The officer could then use a whistle to summon

his colleagues. The invention of the telegraph made it possible for police stations to communicate rapidly with each other, but it did little to help citizens notify police of crimes. Even the invention of the tele-

phone, in itself, failed to alter the situation much because police on patrol could not be reached by phone except at regularly appointed times when they called into the station from a police telephone box; the actual uses of that system were more for controlling police loafing on duty than for directing their work (Rubinstein 1973, chap. 1).

It was only the invention of the radio and its reduction in size to fit

into an automobile that made it possible to connect crime victims al-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 355

most instantly to an officer who could respond quickly enough to make

an apprehension. The rapid availability of that information trans- formed the way that police thought about information, changing their focus from information with long-term payoff potential to information

with very short term payoff. The short term, of course, was the oppor-

tunity to apprehend an offender red-handed. The long-term payoff was

in solving crimes where the offender gets away as well as deterring crimes by criminals who know the police are "on to them."

This technological change did not necessitate the change in informa-

tional focus, but it clearly facilitated it. With the rise of low-density

communities relying primarily on automobile transportation, the infor-

mation system of wall-to-wall police gave way to the "dial-a-cop." This

change reduced the amount of "nonemergency" information that police received as well as limiting personal relationships with local community

members. Once officers were put at the other end of requests for emer-

gency response, everything else-including preventive patrol- became secondary to waiting for that exciting call.

What exactly was that "everything else" of information gathering besides responding to emergency calls? That is a matter of some de- bate. As Lane (1980, p. 13) observes, we can learn relatively little through historical research about ordinary patrol work or routine in-

teractions with the public. Some scholars (Fogelson 1977; Walker 1977)

are skeptical that police did much at all, at least by way of police work.

But other sources, such as novelists who lived in turn-of-the-century

urban communities (e.g., Smith 1943), describe police as fairly well integrated into the life of the neighborhood. The information that police acquired through informal contacts with local residents was of-

ten essential to solving crimes or even to preventing crimes by helping to solve family or youth problems (Murphy and Plate 1977).

Whether police actually spend less time talking to neighborhood residents today than they did two decades or half a century ago is unclear. But a number of developments have clearly encouraged them to move in that direction. One is air conditioning, which led police to

drive through the streets with their car windows rolled up, maintaining a splendid and sanitary isolation from the sounds and smells of the streets. More important, perhaps, and clearly more variable across communities within cities is the availability of residential air condition-

ing. Air conditioning keeps people inside more months of the year, discouraging the sidewalk community dialogues on summer evenings that police can join in so easily. Air conditioning in communities of

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

356 Lawrence Sherman

detached homes also led to the architectural decline of the front porch,

that facilitator of informal, avocational surveillance of all aspects of public neighborhood social life. Neighborhood residents in many com- munities became less knowledgeable about each other and less able to share their insights with the police. The possibility that there may be

much more contextual knowledge to obtain in communities without air

conditioning has been lost in the process of developing a relatively uniform information-gathering strategy for all kinds of communities.

Police administrators have long been aware of these problems. As early as the 1960s, the Chicago police attempted to combine walking and riding on patrol so that police would have more informal contacts.

Police in other cities employed motorcycles, scooters, motorbikes, bicycles, and horses as forms of transportation that posed less of a barrier to police-citizen contacts. In theory, all these programs show how there is no necessary conflict between rapid police mobility and informal intelligence gathering. But virtually every evaluation of such

programs shows that they encounter enormous resistance from police

officers, who find it much easier to stay with their vehicles and talk

with other police than to park, walk, and talk with the citizenry (e.g.,

Sherman, Milton, and Kelley 1973; Sherman 1983). Police isolation may not be an inherent characteristic of mobile patrol, but it is a consequence that is difficult to reverse.

Information exchange among police officers is also important for contextual knowledge and may vary widely across communities. Com-

munities with local precinct stations may have better exchange among officers than do communities whose officers report to work in a down-

town headquarters. But even precinct stations may have minimal for-

mal exchange. Despite the television image of police roll call as an announcement session (as in "Hill Street Blues"), most actual roll calls

are confined to announcements of bureaucratic imperatives such as training sessions and pension forms. A detailed picture of recent crime

events throughout an entire precinct would take too long to relate in a

brief roll call. Written handouts are also discouraged because the mass of crime information changes so quickly. The consequence of these changes in the technology and organiza-

tion of police information is that police work has become ahistorical. Each police task is approached without information on the past or on the likely future of the parties involved. The primary unit of analysis is

the single complaint, the telephone call to which a patrol car responds. Urban police must usually enter a situation stripped of any contextual

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 357

knowledge about its causes or likely consequences. They often spend a great deal of time trying to elicit that information, although they are

frequently plagued by suspicions that the information may not be reli-

able. The same people may have encountered other police officers a few

hours ago, yet most urban police officers would have little chance of knowing that fact. The typical car-patrol officers are as unknown a quantity to the people they police as the people are to them.

2. Mobilization. The most controversial aspect of community polic- ing is the manner in which police efforts are mobilized (Reiss and Bordua 1967). The mobilization of police at their own initiative, or proactive policing, is often associated with restriction of individual freedom and privacy. The mobilization of the police at the initiative of citizens is often viewed as the most democratic and fair method of

policing a free society (Black 1973). The choice is not mutually exclu- sive, of course, since almost all societies feature both systems of mobilizing police, and totalitarian societies have a frightening capacity

to induce citizens to volunteer information to the police.

Modern American urban communities may differ very little in the

way that police are mobilized except to the extent that they are policed

by foot-patrol officers. Foot-patrol officers, historically, have been es-

pecially proactive in the maintenance of public order. Early American

police bureaucracies generated phenomenal numbers of public order arrests per officer per day (Levett 1975), which is hardly likely to have been a result of reacting to specific citizen requests. Even as late as 1966

in three inner-city communities, police in patrol cars initiated some 13

percent of all their mobilizations (Reiss 1971, p. 11). But the rise of the modern police information system described above

substantially changed the nature of proactive policing. The bulk of police patrol time and of police resources in general were drawn into single-complaint, rapid-response, reactive mobilization. Just as they rarely stop to chat with neighborhood residents, the police driving by

in air-conditioned cars may rarely keep an eye on any known suspicious people.

The evidence of variation in proactivity is substantial at the intercity level. Friedrich (1977, p. 251) found that Boston police stopped one- tenth as many motorists per tour of duty as did Chicago police. Whether it varies as much across communities within cities is unclear.

But even substantial variation would probably leave reactive mobiliza- tion with the vast majority of all contacts in all communities.

Moreover, modern proactive patrolling is based less on direct knowl-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

358 Lawrence Sherman

edge and observation of community residents and more on the use of computerized information systems. Proactive patrol is now less a mat-

ter of stopping local drunks or breaking up visible fights than it is a matter of checking license plates against a computerized list of stolen

cars or stopping suspicious people to check their identification against a computerized list of persons wanted on arrest warrants. These modern

"stranger" policing methods are often criticized for hurting police- community relations (National Advisory Commission on Civil Disor- ders 1968) and may further undermine the flow of intelligence through informal citizen contacts.

3. Action and Targets. Once police are mobilized to deal with certain

information, they can take a variety of actions. These could theoreti-

cally vary across communities, but they rarely seem to in practice. Targets include locations, perpetrators, complainants, victims, rela- tionships, generic problems (such as the mentally ill), suspected classes

(such as young men in groups), and, of course, neighborhoods. Methods include surveillance-either visible or covert, roving or sta- tionary-and street stops, roadblocks, interviews, community meet- ings, informants, traps or "stings," raids, warnings or threats to arrest,

arrest, mediation and negotiation, referral to social or health agencies,

and a myriad of other options (Goldstein 1977). But most policing in most communities is confined to the complainant as target and is lim-

ited to the actions of interviews, warnings, negotiations, and arrests.

The roving, visible surveillance is relatively unfocused as to target; patrol is rarely addressed to specific communities as distinct from the

city in general. Private police probably differ from public police in their

greater focus on the properties of their clients, just as citizen patrols are more focused on the citizens' own communities.

Other combinations of targets and methods produce familiar police strategies. A highly crime-prone physical location target approached with fixed, covert surveillance constitutes a "stakeout," for example. Traps and stings used to catch highly crime-prone perpetrators are the

common strategy of at least one police department's "Repeat Offender

Project" (Martin and Sherman 1986). Street stops of the suspicious class of young, poor males constitute a large part of the "aggressive patrol," or field interrogation, strategy. But all these strategies are rela-

tively rare. Moreover, their selection is rarely linked to community characteristics that may make them more or less effective. They are usually adopted at the citywide level to supplement the basic single- complaint, reactive strategy common to most communities.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 359

III. Critique of Single-Complaint Policing Recent research has suggested that single-complaint policing is wasteful and possibly ineffective. The critique began early (Wilson 1963, p. 237). As crime grew rapidly in the 1960s, some observers began to blame the increase on the shift in police strategy (President's Commis-

sion on Law Enforcement and Administration of Justice 1967, p. 54). The National Advisory Commission on Civil Disorders (1968, pp. 304-5) blamed poor police-community relations on the motorization of

patrol and the attendant loss of contact with the community, which adversely affects law enforcement. The commission blamed proactive policing in the absence of this contextual knowledge for causing minor-

ity perceptions of police harassment and attacked the aggressive style of proactive "stranger" patrol.

Ironically, the justification for moving to single-complaint policing

was to focus the police role on crime fighting. The abandonment of proactive order maintenance and intelligence gathering was the price of

this emphasis on crime control. But recent research has supported the

theory that single-complaint policing is less effective in controlling crime than was community policing.

1. Preventive Patrol in Cars. The Kansas City Preventive Patrol Ex- periment (Kelling et al. 1974a) tested the assumption that police driving

around streets in cars create a visible presence that prevents crime in

public places through general deterrence. The experiment varied the level of patrol presence in three groups of five beats each and compared the before and the after differences in literally hundreds of measures of

crime. The Kansas City police chief (McNamara 1974, p. vi) concluded that the lack of significant differences in the three types of beats showed

"that routine preventive patrol in marked police cars has little value in

preventing crime or making citizens feel safe." Others were more re-

served in their judgment, especially given some question about the actual extent of the differences in patrol presence across the three types

of beats (Larson 1975) and the manner in which levels of patrol were assigned to the different beats (Farrington 1982). A more basic re- search-design problem of small sample sizes used in evaluating commu- nity policing efforts also limits scientific confidence in the findings.

Despite these questions, the Kansas City findings have been widely accepted among police executives as showing that modest increases in the number of patrol cars are unlikely to reduce crime. One reason for

this gradual acceptance may be the suspicion that police managers have

that the officers are not using their patrol time very productively but

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

360 Lawrence Sherman

are just waiting for calls for service to come in. Since the experiment

found that this waiting encompassed about 60 percent of on-duty patrol

time, it raised the question of what better use police might make of that

time. But many plans to develop new tasks have run afoul of police union opposition (since the plans mean more work for police), cloaked

in the argument that such tasks would increase police response time to calls for service.

2. Rapid Response Time. Arguments are made that ready availability of

officers in cars, geographically dispersed throughout the community, is

imperative if rapid responses are to be logistically feasible. Rapid re- sponse has been justified as necessary to increase the deterrent threat of apprehension for criminals.

A second research project in Kansas City (Kansas City Police De- partment 1977), however, undermined those arguments. It began by systematically analyzing the components of "response" time, including

time of crime occurrence, time involved in reporting (by telephone) to a

police operator, time involved in sending a radio dispatch to a police car, and travel time of that car to the scene of the crime. Response time

is important, even in theory, only for "involvement" crimes, in which

the victim confronted the offender who escaped immediately thereaf-

ter. Response time is clearly not an issue for "discovery" crimes, in which the offender may have fled hours earlier. So the study examined departmental records and other data on 352 involvement crimes.

The findings showed that, by the time a police car begins its travel to

the scene of most involvement crimes, so much time has elapsed that the chances of apprehending an offender are extremely small. Of the

average fifty-minute total response time, only four and a half minutes

consisted of police travel time. On average, it took crime victims and witnesses almost forty-one minutes to report the crimes (Kansas City

Police Department 1977, p. 19). Since the availability of police waiting in cars can only reduce travel time, the citizen delay in reporting poses

a major obstacle to apprehension of fleeing criminals. Only 12 percent of 949 Part I crime calls analyzed resulted in arrests at the scene. Of these, only one-third-3 percent of all calls to police for Part I crimes- could reasonably be attributed to the rapid police response (p. 29).

Police chiefs who were skeptical that these findings applied to their cities volunteered to have the Police Executive Research Forum repli- cate the citizen-delay study. The National Institute of Justice-funded study in San Diego, California; Peoria, Illinois; Jacksonville, Florida; and Rochester, New York, found similarly long delays and confirmed

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 361

the need for police to rethink the rapid-response, single-complaint strategy (Spelman and Brown 1981). Some chiefs said that the solution

lay in public education to stimulate more rapid reporting. But others saw a need for alternative strategies to supplement-not replace-the rapid-response strategy.

3. Public Demands for Rapid Response. Many chiefs believed that, re- gardless of the ideal mix of strategies, their hands were tied because of

citizen demands for rapid response. But further research sponsored by

the National Institute of Justice falsified that argument as well. In a field test in Toledo, Ohio; Garden Grove, California; and

Greensboro, North Carolina, a differential police response (DPR) strat-

egy was implemented on a random-assignment basis. The control group received the standard rapid response for most kinds of calls for

service. The experimental groups received a variety of responses, de- pending on the nature of the call: rapid response, referral to a telephone

report unit for taking reports over the phone, a delayed mobile response

(in which callers would be told that police would not be there for thirty

to sixty minutes), referral of the call to another agency, scheduled appointments with police, walk in, or mail in (McEwen, Connors, and Cohen 1984, p. 8). The results in all three sites showed that the number

of nonemergency calls handled with a rapid response could be reduced

significantly without any reduction in citizen satisfaction (p. 17). Citi-

zens seemed willing to accept both delayed responses and alternative responses as long as the dispatchers clearly told them what to expect.

The DPR experiments suggest that it may be politically feasible to abandon rapid responses to most incidents and even any personal re- sponse to some kinds of incidents. More research is needed to be certain

about the costs and benefits of different kinds of response to the differ-

ent kinds of calls. Much will depend on the kind of information that police seek to obtain and are able to use. But the experiments provide added evidence against the use of rapid, single-complaint policing as the nearly exclusive strategy for policing residential communities. All

communities will continue to demand, and police must continue to provide, a rapid-response capacity for some calls. Depending on the nature of the community, however, much of the time now devoted to

rapid response could more usefully be devoted to other strategies. There are many examples of mixed-strategy policing tailored to the

needs of specific communities. Many of the decisions to employ such mixed strategies are politically determined responses to demands for police action on specific and highly newsworthy community problems

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

362 Lawrence Sherman

(e.g., crackdowns on prostitution in newly gentrified red-light areas).

But in the absence of such political pressures, most communities still feature little mixing of policing strategies, relying primarily on single- complaint responses.

Assuming that it is politically feasible to reallocate some of the police

resources currently devoted to rapid response, the question then be- comes, What alternatives can be pursued most productively? What strategies can public police, private police, and private citizens employ to supplement the single-complaint model in ways sensitive to the unique characteristics of each community?

IV. What Can Public Police Do?

Many alternatives to single-complaint policing have recently been tried

or experimented with. Some are widely used. These strategies have been evaluated only to a limited extent, and the evaluations all suffer

from a major problem of interpretation.

A. The Evaluation Problem

The problem is sample size, in two respects: the number of citizens that researchers must survey to estimate criminal victimization rates in

any given community and the number of communities that must be

included in any test of policing strategy at the community level. Most kinds of crime are relatively rare, in absolute numbers, within the boundaries of most patrol beats or neighborhoods. That rarity biases most statistical tests. Given the small numbers involved, even large percentage differences in crime across beats can fail to achieve statistical

significance at conventional levels, and justifiably so since such small absolute differences can be obtained by chance.

In the original Kansas City experiment, a 300 percent difference in the average before-after change in reported outside robberies each month was not statistically significant, although it came close (Kelling et al. 1974b, p. 96). The enormous percentage difference reflected the absolute difference between an increase of .936 outside robberies per month in the beats with reduced patrol and an increase of only .003 outside robberies per month in the beats with a doubled level of patrol (the beats with unchanged patrol intensity were close to the doubled beats, at .269). This pattern is all the more important because it is consistent with the hypothesis that patrol can deter and with what is

arguably the most suppressible kind of crime. Whether a larger sample size, or higher absolute levels of crime, would have produced

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 363

significant differences is impossible to say. But the problem clearly complicates interpretation of the study.

In their redesign of the Kansas City experiment, Fienberg, Larntz, and Reiss (1976) suggest abandoning public surveys as the principal method for measuring community policing effectiveness. Concluding that the sample sizes in the Kansas City experiment were much too small, they recommend using a variety of official police records to measure deterrent effects in more creative ways than simply counting

the number of recorded crimes, including data on offender modus operandi, rearrest rates and place of occurrence of rearrest, demo- graphic characteristics of arrested street criminals, ratio of property to

person crimes, ratio of first to repeat offenders arrested, and distance from the place of arrest to the arrestee's residence. While these indi-

cators would not yield estimates of prevalence of victimization in the community, they would yield more reliable estimates of indirect mea-

sures of that concept than victimization surveys are ever likely to pro- duce-and at far less cost.

The largely unheeded Fienberg et al. (1976) paper also solves the second sample-size problem. The bias toward finding no effect from an

innovation is related to the small number of patrol beats (fifteen) used in

the Kansas City study and to the even smaller numbers used in other

studies. Because expensive victimization surveys would not be used, much larger numbers of communities or patrol beats can be included in

experiments. The authors recommend a minimum of at least twenty- eight patrol beats for their sophisticated three-year, three-treatment design, in which crossovers make each beat act as its own control, and

each "primary treatment" beat is cushioned from the others by a layer of secondary beats that surround each primary beat with a common treatment.

This design is not without its drawbacks. It would, for example, be almost impossible to use with citizen participation strategies, given the

need to change strategies each year in each community-something to which few communities would be likely to agree. The three-year time

frame is also troublesome, given the political instability surrounding both police and (some) research organizations. In a three-year experi- ment in citizen-police action in Minneapolis, for example, the action program director resigned in a dispute with the city council in the first

year, and the evaluation project director was replaced in the third year. Maintaining experimental conditions and analyzing data with full con- textual knowledge of the project are difficult with such turnover. Fi-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

364 Lawrence Sherman

nally, the large number of communities in the proposed design would

create substantial practical problems of experimental control, with am-

ple opportunities for officers or others to sabotage the design by altering

the designed treatment in an area. Nonetheless, the Fienberg et al. design appears to be the only feasi-

ble way of removing the bias toward the null hypothesis in reliably estimating the effect of community policing strategies. There may be

no better alternative design for testing for interaction between policing

strategy and community characteristics, which requires even larger sample sizes. Whether any of these designs have much future in an era

of research-funding austerity seems doubtful.

In the absence of research results from such designs, we can only accept the premise that knowledge derived from less rigorous sources

still has validity for policy-making purposes (Lindblom and Cohen 1979). Both limited evaluations and practitioner intuition may thus be

usable data (until proven otherwise by more systematic tests) for assess-

ing which strategies are most promising in general and in specific kinds of communities.

Wilson (1983) has offered a useful distinction in theories of how these

various approaches can reduce crime. He defines the community- service approach as any means of making officers more familiar with

their neighborhoods in order to win the confidence and cooperation of

local residents as well as better intelligence about criminal activities. The crime-attack approach, in contrast, while not logically incompat- ible with community service, places officers as close as possible to likely

locations or perpetrators of crimes "in ways that will enable them to apprehend the criminal in the act, or at least cut short his crime almost

as soon as it begins" (Wilson 1983, p. 68). One might add efforts to deter crime by visible surveillance of potential targets (e.g., bodyguards

and off-duty police at late-night commercial establishments) as an addi-

tional element of the crime-attack approach. The evidence for both strategies is mixed, although the evidence against at least one of the crime-attack strategies ("tailing" suspects) is quite strong.

B. Community Service

The 1967 Crime Commission report stimulated many police depart- ments to undertake neighborhood team policing, one of the earliest versions of the community-service approach. These programs varied, but they generally fostered greater police knowledge of their com- munities through greater geographic focus and stability of patrol as-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 365

signments, increased communication with the community, and in- creased communication among the officers working in the small beat area (Sherman et al. 1973). The beat teams were led generally by sergeants, who were given unusual flexibility to schedule their officers

and assign them to various tasks. Like many organizational innovations,

however, a variety of organizational constraints generally blocked this

one from being implemented and ultimately discredited both the idea and its name.

One leading constraint was the unrelieved imperative to answer the calls for service on the beat. To the extent that the beat-team officers

spent time in neighborhood meetings, playing basketball with mar- ginally delinquent kids, staking out the welfare office, or doing other innovative, proactive things, the other officers in the precinct would have to come into their beat to answer the radio calls they could not handle. This created rivalry and tension across beats.

More important was the power balance between the team-leader sergeants and the middle managers above them (Sherman 1975). The demonstration projects often gave sergeants more power than lieuten-

ants and captains, whose jealousy led them to countermand and sabo- tage the orders of the team leader. While the sergeant's unusual author-

ity made programmatic sense, it made little sense from the standpoint

of the larger bureaucratic turf concerns. Thus what could have been a substantive innovation never went beyond the problems of structural

reorganization.

One consequence of all these implementation problems was that, even when the innovation was carefully evaluated, as in the Police Foundation experiment in Cincinnati (Schwartz and Clarren 1977), it was impossible to tell whether the theory of the program was successful

in controlling crime. The program failure meant that the theory was

not properly tested. Perhaps in the wake of DPR (McEwen et al. 1984) it will be possible to try again to use the beat-team approach to gather-

ing information. Failing that, the idea could still be pursued at the level of the individual officer. For example, a demonstration project in San Diego showed the feasibility (but not the effects on crime) of having individual police officers, working in the normal chain of command, develop detailed profiles of the communities they policed (Boydstun and Sherry 1975).

Other community-service strategies have been evaluated in recent years. Foot patrol, which is feasible only in dense areas (and desirable for police only in good weather), has been evaluated in several experi-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

366 Lawrence Sherman

ments. All suffered from a small number of beats available for evalua-

tion and the small numbers of crimes within each beat. Not surpris- ingly, then, a major experiment in Newark (Police Foundation 1981) found that foot patrol had no effect on crime as measured through victimization surveys. A Northeastern University project is now exam-

ining a much larger data set on changes in reported crime on beats in Boston over the course of several introductions and withdrawals of

foot-patrol officers (Bowers 1985).

Perhaps more important was the Newark finding concerning fear of

crime, which was lower in the beats receiving foot patrol. Wilson and

Kelling (1982) explained this finding as a result of the proactive order-

maintenance efforts of the foot-patrol officers. Given research findings

that fear of crime is generated by such "signs of crime" as disorderly youths, panhandlers, garbage-strewn streets, and broken windows (Skogan and Maxfield 1981), it makes sense that foot patrol aimed at controlling pedestrian disorder could reduce the fear of crime. Fear

reduction is legitimately an explicit goal for neighborhood policing, and the Newark experiment showed that it could be accomplished.

The recent experiments in police fear-reduction strategies conducted by the Police Foundation in Houston and Newark also included a community-service strategy (Wycoff et al. 1985a). In one of the Hous- ton experiments, officers were trained to focus on one beat and to

attempt to make as many contacts as possible with people who lived and worked on that beat. They ultimately visited one-third of all the households on the beat, introducing themselves and asking residents to

tell them of any concerns or neighborhood problems they could iden-

tify. The outreach effort was combined with an increased presence of police cars in the beat, although surveys showed no difference in citizen

awareness of police presence. The outcome was not only a substantial reduction in the fear of crime but also a statistically significant reduc-

tion in the percentage of households reporting that someone in the household had been victimized by crime-almost all of which was property crime (Wycoff et al. 1985a). The experiment was limited by a sample of two: the demonstration beat and an unchanged, matched- comparison beat. But it was also limited by the usual bias toward the null hypothesis, which is not what was found.

The Houston findings may conceivably result from some unique event such as the incarceration of an active neighborhood delinquent, or the neighborhood crime may simply have been displaced to other areas where police were not quite so active. But it is also possible that

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 367

the community-service approach--when done with enthusiasm by officers enjoying a Hawthorne effect and in a racially mixed but pre-

dominantly white lower-middle-class neighborhood of single-family detached houses-might just work to reduce crime.

C. Crime Attack

Somewhat less evidence is available about the effectiveness of proac-

tive crime-attack strategies. That evidence suggests that some such strategies are clearly ineffective but that others can be used to increase

arrests of serious criminals and perhaps to decrease crime. Perhaps even

more than with the community-service strategies, success may depend

on very subtle aspects of how an approach is implemented and how well it fits within a particular community.

1. Decoys. Wycoff, Susmilch, and Brown's (1981) evaluation of a Police Foundation-funded experiment in Birmingham, Alabama, dem- onstrated that crime-attack strategies cannot always be transported eas-

ily from one police department to another. The plan to test the crime-

control effectiveness of an antirobbery unit using decoy methods ran

into a number of snags, not the least of which was the police officers'

relative disinterest in dealing with crime. For example, they decided to

run their decoy operation during the "high robbery hours" of Monday

through Friday, 8:00 A.M. to 4:00 P.M. They were also reluctant to engage in any dirty or dangerous methods of decoy operation. Thus an

idea that had received much prominence in New York and elsewhere failed to work in Birmingham because it did not fit the local culture of

policing. The effectiveness of decoy operations-a classic crime-attack strat-

egy-remains unclear. Some critics argue that decoys often attract people to commit crimes who otherwise might not have committed them. Given the high risk of danger to both police and citizens, a careful evaluation of this strategy merits high priority and should con-

sider whether effects depend on the kind of neighborhood in which a

decoy operation is attempted. For example, decoys in center-city vice areas may attract predisposed assailants, while decoys in residential areas may entice people to commit crimes who otherwise would not have.

2. Covert Patrol. Patrol in disguise is another common crime-attack

strategy that remains unevaluated. This strategy obviously requires a community with a high-enough street-crime rate per square foot so that it is feasible for anyone to intercept crimes in progress. New York City,

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

368 Lawrence Sherman

or parts of it, clearly qualifies. The 5 percent (or less) of the patrol force

assigned to this strategy in the early 1970s accounted for over 18 per-

cent of all felony arrests, including over half the robbery arrests and 40

percent of the burglary and robbery arrests (Wilson 1983, p. 70). But the effect of these efforts on crime remains unclear.

3. Field Interrogations. The National Advisory Commission on Civil Disorders (1968) attacked field interrogations for breeding poor police-

community relations. It is a crime-attack strategy in which many police

have a good deal of confidence as a way to deter crime. The anecdotal evidence that this strategy can cause race riots is substantial. But such

accounts are counterbalanced by the results of a careful evaluation of an

experiment in the San Diego Police Department. In a context in which police had a long history of active use of field

interrogations, the experiment compared three patrol beats: in one, field interrogations were stopped altogether; in a second, they remained

unchanged; and in a third, police were given special training in how to conduct the street stops in a sensitive and nonabrasive fashion. The evaluation found little difference in the crime rates between the special training and the normal areas. But it did find that, where field interro-

gations were withdrawn compared with the other areas, there was a statistically significant increase over a nine-month period in the average monthly number of reported total crimes, total Part I crimes, and burglaries. Moreover, there was no difference in the results of before-

and-after surveys of community attitudes toward the police (Boydstun

1975). These results were obtained despite the bias in favor of the null

hypothesis about crime rates and despite no bias about community attitudes (which, unlike crime, are not exactly rare and more easily lead

to significant differences). The small sample size (three patrol areas) again means that some unique phenomenon may have affected the crime-rate differences. But barring that, the experiment suggests that

field interrogations can deter certain kinds of crime without annoying

the community-at least in a community that is already accustomed to

having an intrusive, proactive police department.

Such departments can generally deter robberies more effectively than can their more reactive counterparts in other cities. In a cross- sectional analysis of cities with high and low traffic-enforcement rates,

Wilson and Boland (1978) concluded that the high-traffic-enforcement

departments produced lower robbery rates. They acknowledge the difficulties in interpreting such cross-sectional data but suggest the importance of further experimentation along those lines. One of the Newark fear-reduction experiments (Pate et al. 1985a)

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 369

addressed the question of what effect intensified field interrogations can

have in the context of generally intensified proactive enforcement. The

"signs of crime" strategy in one Newark patrol beat was designed in part to test the Wilson-Kelling (1982) argument about controlling crime

through control of disorder. The program consisted of a twenty-four-

officer, specially trained task force that performed a variety of tasks

within the experimental area at least three times a week. The task force

spent over 2,500 hours in the program area over the course of nine months. Seventy percent was devoted to foot patrol, 15 percent to radar speed checks, 7.5 percent to bus checks, 4 percent to enforcement

of disorderly conduct laws, and 3 percent to conducting road checks. The disorderly conduct enforcement was particularly aggressive, con-

sisting of an order to clear the sidewalk (if four or more people were

congregating) and arresting anyone who failed to comply.

The results were mixed and hard to interpret. Before-and-after sur-

vey data were largely negative: area residents were less satisfied with the area and reported significantly higher levels of crime victimization

than did residents of a matched comparison area. Yet the residents perceived the police as becoming less aggressive. The recorded crime data showed statistically significant reductions in total Part I crimes, burglary, and personal crimes, while no significant reductions were found in the comparison area. It is possible that both measures of crime

may be correct since the survey measured the prevalence of crime (the

percentage of households suffering one or more crimes during the ex- perimental period) while the recorded crime data measured the inci- dence of crime (the total number of crimes recorded, including repeated

victimizations of the same people or households). It is also possible that

the recorded crime data were manipulated to produce an apparent program effect. But the findings are at least as strong as the San Diego experiment in showing that recorded crime is lower under conditions of

aggressive field interrogation and proactive citizen street contacts.

4. Repeat Offenders. The increasing policy attention to repeat offend-

ers has spilled over from prosecution and sentencing into police. The first evaluation of this strategy was done in Kansas City by Pate, Bow- ers, and Parks (1976), who found more of a program failure than a theory failure. The Kansas City police at that time insisted that all its tactical officers, including those assigned to covert surveillance of sus- pected active criminals, drive official (if unmarked) police cars and remain clean shaven and close cropped. Their identity was frequently recognized, and much of the surveillance was unproductive.

A similar unit, devoted to apprehending highly active offenders, was

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

370 Lawrence Sherman

implemented in Washington, D.C., a decade later, with much greater sophistication. What they found, however, was that covert surveillance of individual suspects was an extremely inefficient and frustrating way for police to spend their time. The strategy produced more "serendipi-

tous" arrests of nontargets committing crimes in the presence of the

covert officers than it did of the intended targets (Martin and Sherman

1986). Other methods of the Repeat Offender Project (ROP) paid off, however, such as searching out the whereabouts of reportedly active offenders who were already wanted on warrants, using sting-type de-

coys, and responding to hot tips from informants about the immediate

whereabouts of an active offender who had incriminating evidence on him.

The effects of these efforts on crime is unclear, but a randomized

experiment (Martin and Sherman 1986) found, predictably, that the ROP unit was five times more likely to apprehend someone they targeted than was the rest of the department in the course of its normal

operations (not knowing that such persons were designated as targets).

Moreover, the average length of the criminal history of persons arrested

by ROP officers increased substantially over those of the persons they

had been arresting in other units prior to their assignment to ROP. The

number of arrests per officer declined after their transfer to ROP, but

that is a predictable result of working in four- to six-officer teams. Such

a program, then, only makes sense to the extent that the longer records

of those apprehended indicate a higher current rate of offending than

the suspects who would have been arrested by the same officers pro- ducing a greater quantity of arrests in patrol.

Overall, ROP succeeded in focusing scarce police resources on per- sons for whom there was probable cause to believe were highly active

criminals. At the same time, ROP suffered relatively little controversy

or questions about the constitutionality of their methods. Since ROP is

a citywide venture, however, it is not clear that it is the kind of program that can be varied in its administration across communities.

5. Saturation Patrol. What most communities dream of, of course, is

having lots of police patrolling the neighborhood. Several tests of this

idea all show the same result, although with varying degrees of rigor. The first was the "Operation 25" pretest/posttest design in New York City in 1954, in which the presence of foot-patrol officers in a Manhat- tan precinct was about tripled and street crime fell dramatically (Wilson

1983, p. 62). A later analysis of a similar change in personnel resources, with two comparison precincts, showed similar if less dramatic results

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 371

(Press 1971). The most rigorous test was an experiment in four high- crime residential areas of Nashville, Tennessee, which increased patrol presence by 400 percent and increased slow patrol movement (under twenty miles per hour) by 3,000 percent for ten days (Schnelle et al. 1977). Time-series analysis showed statistically reliable reductions in nighttime, but not daytime, Part I offenses. These levels of patrol presence may be too costly to implement on a widespread basis, but the

tests do support police executives' views that saturation patrol can in- deed reduce crime.

6. Repeat Complaint Address Policing. Many crime-attack strategies re-

main unevaluated. One possible approach, for example, is to focus police efforts on those addresses that account for the highest propor- tions of calls for police service. According to an analysis of Boston citizen call data (Pierce, Spaar, and Briggs 1984), as few as 10 percent of

the addresses that police are ever dispatched to may account for up to 40 percent of all dispatches, for some types of calls. It follows that solving the underlying problems producing those repeat calls at those

most active locations would theoretically reduce the total call load by up to 40 percent.

Such problems can be addressed through a strategy of repeat com- plaint address policing (RECAP) (Sherman 1985). The strategy would consist of four steps: (1) analysis, in which the police-dispatching com-

puter would rank all addresses responded to by frequency of responses

over a fixed period; (2) diagnosis, in which the most highly ranked addresses would be studied through written reports and site visits; (3)

action, in which police officers would implement an action plan re- viewed with their supervisors, encompassing anything from threaten-

ing to arrest a repeat wife beater to assisting a bank manager in having the alarm system rewired to reduce false burglary alarms; and (4) fol-

low-up, in which the officers who implemented the action plan monitor the address regularly for the next several months to determine how well

the plan is working and whether another plan is needed.

The RECAP strategy would require highly creative officers who can

imagine how best to use their unique position as dispensers of coercive threats to cope with the problems underlying the repeat calls. It would

also require a substantial reorganization of most police departments to

allow some patrol officers to devote full time to such problem solving

and remain free of the need to answer radio calls. But it would be easy to test the RECAP strategy on a limited scale prior to any broader implementation (Sherman 1985). Simply by dividing the most active

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

372 Lawrence Sherman

addresses in half on a lottery basis, RECAP officers could work on one-

half and leave the other half alone, thus creating a controlled experi- ment. Any differences between the groups over the follow-up period in

the number of repeat calls or the seriousness of subsequent problems could then be attributed to the difference in police strategy.

Like several of the crime-attack strategies, RECAP is not especially neighborhood sensitive. The basic strategy could be pursued in almost

any neighborhood. But the creative search for solving the unique prob-

lem generating each set of repeat calls could be done much better by officers familiar with the neighborhood context and resources.

V. What Can Private Police Do?

Many, if not all, things the public police do could be done by private police. More important, private police perform many functions that public police and citizens would rarely perform because of the labor cost involved, such as controlling access to parking garages or to lobbies

of apartment buildings. The contribution of private policing to commu-

nity crime control is probably substantially underestimated.

1. Routine Patrol. The use of private guards for routine patrol in wealthier neighborhoods grew substantially in the 1970s. Residents who were dissatisfied with public police patrol banded together or worked through existing homeowners' associations to raise funding for private uniformed patrols in marked cars. In dense cities like New York, some blocks hired private foot-patrol officers. And in the grow-

ing realm of quasi-public spaces, such as shopping centers, all kinds of

parking lots, and other private property open to the public, private police often conduct routine patrol at the expense of the property owner. Much of this patrol is performed indoors, which is generally defined by public police as beyond their scope of responsibility. While

few guards are armed with guns, they usually wear enough other arma-

ment (like nightsticks) to communicate a threat that they will use force if necessary.

2. Order Maintenance. Whatever deterrent value these patrols may have, their order maintenance value is demonstrably substantial. Pri- vate police keep people from drinking from bottles, arguing loudly, running around recklessly, or playing loud music. Nonuniformed pri- vate employees such as bouncers and doormen also maintain order on the streets immediately off the premises where they are employed. And

because their job is tied to a specific location, they perform such tasks far more thoroughly and intensively than public police usually do. If,

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 373

as Wilson and Kelling (1982) suggest, the failure to maintain order can

lead to serious crime, the order-maintenance efforts of private police may help control crime as much or more than their deterrent value. At

the very least, as the Newark foot-patrol experiment suggests (Police Foundation 1981), such private policing should help reduce public fear of crime.

3. Saturation Patrol. Many operators of quasi-public places have adopted saturation patrol techniques. The defendant hospital in a land- mark California Supreme Court case on landlord liability for third- party crime (Isaacs v. Huntington Memorial Hospital, 695 P.2d 653 [Cal.

1985]) provides a good example. After a doctor was shot in 1978 in an apparent robbery attempt in a parking lot right across the street from

the emergency room, the hospital increased its outside nighttime secu-

rity patrols so that the lot would almost always be in view of a guard

(during selected high-risk hours). The hospital also constructed a guard kiosk in a central location and added a surveillance camera to monitor

the doctors' parking lot at all times. If the intensity of private policing has increased as much at many other locations where crimes occurred

during the 1970s, it could be one factor in the slight decline of stranger crime in the 1980s.

4. Access Control. One function public police perform only for officials and for virtually no one else is controlling access to secured places. Private police control access to parking garages, apartment and

office buildings, and entire residential developments. These services are more common in rich neighborhoods than in poor ones. Their role

in reducing the crime risk of those they protect is nonetheless hard to dispute.

5. Location Defense. More ubiquitous are the private police who de- fend specific locations from armed robbery. These guards, usually armed, can be found in banks and late-night retail operations from supermarkets to fast-food stores. There is a substantial debate among practitioners about whether these guards attract more violence (from

criminals looking for a challenge) than they deter (Johnson 1984). They may also displace crime from one kind of target to another-from commercial establishments to more vulnerable individuals. But their

presence also adds to the order-maintenance aspects of local policing.

VI. What Can Private Citizens Do?

One limitation of all the strategies discussed above is that none of them

involve citizens in any major way. Some proponents of such involve-

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

374 Lawrence Sherman

ment seem committed to the idea for ideological reasons, while others

have theoretical justifications (Curtis 1985). One motivation for citizen

involvement in policing, for example, may be an agenda for disman- tling the public sector in favor of a philosophically preferable voluntar- ism in many spheres (Currie 1985). It is therefore no surprise that many

police unions are suspicious of citizen-involvement programs. As much as police might like to keep such programs under their

control, community policing is increasingly done outside the auspices of the public vocational police. Foundation and even federal grants are

often made directly to community groups, many of which operate without any funding to provide patrol and other forms of surveillance.

Unfortunately, it is almost impossible to compare the relative effec- tiveness of police-directed and other programs since very little research

has been done on the nonpolice programs.

A. Police Programs for Citizens

Police programs for fostering citizen involvement consist generally of

efforts to organize voluntary efforts to watch the neighborhood for anything suspicious and to call the police when appropriate. Such pro-

grams have been hard to sustain since one meeting with a police officer

can cover the basic point of the program. What they usually lack is continuing neighborhood leadership to keep people watching the streets and a systematic plan for insuring that the neighborhood is watched at all vulnerable times-especially during business hours.

More recently, some police projects have taken a broader approach to organizing citizen and private efforts. They have also tested some strat-

egies of communication, as distinct from organizing, to establish more of a direct link to individual citizens. Recent research sheds some in-

sight on several of these efforts.

1. Police Organizing. From 1983 to 1985, the Minneapolis Police Department made a sustained effort to involve some twenty-five police

officers in a community-organizing effort led by the city's independent

community crime prevention agency. Their planned role was to serve as the "cop of the block" for several blocks each, combining the commu- nity-service model of stimulating grass-roots voluntarism. The block cop's responsibilities were to include daily visits to the block to stop and talk with residents, assistance to the block club in drawing up a plan for identifying and dealing with local quality-of-life problems (not neces- sarily criminal in nature), and communication to block residents about

recent crime developments on and around their block.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 375

The problem was that the plan called for the officers to perform these

new tasks in addition to their regular duties of answering calls. The preliminary indications were that few police officers actually played the

kind of role the program called for. No matter what the impact evalua-

tion may show about the overall program (including the independent agency's efforts to organize citizen policing), it is unlikely that the cop-

of-the-block element will contribute much to that impact. Moreover,

the police resistance may have been related to citizen resistance that was more prevalent in lower-class, transient, single-occupant, and minority neighborhoods, the areas with highest crime that needed the program most.

A niore successful organizing project was included in the Houston fear-reduction experiments. In one middle-class patrol beat, a specialist

group of officers (who were not obliged to answer radio calls while working on this task) worked for nine months to create a community

organization in a neighborhood where none had existed. They began by conducting a door-to-door survey of neighborhood residents (as distinct from the evaluation survey done by Police Foundation re- searchers) to ask them to identify neighborhood problems and potential

interest in voluntary efforts. The survey was followed by thirteen neighborhood meetings attended by twenty to sixty people each, which in turn were followed by police asking twelve residents to meet each month with the precinct captain to identify community problems and

solutions. The twelve-member task force was then encouraged to turn

itself into a community organization, which took responsibility for carrying out such programs as "safe houses" for children, a drug infor-

mation seminar, and identification of household goods with marked numbers.

The evaluation of the Houston Community Organizing Response Team program (Wycoff et al. 1985c) showed no effect on the prevalence of victimization (which was not its goal) or on the fear of crime (which

was), but it did reduce perceptions (compared with a matched compari- son area) of the level of social disorder and crime in the area. And like

other fear-reduction experiments in Houston, the evaluation suggested that the program had relatively less beneficial effect on, and was less

successful in reaching, blacks and Hispanics. A related fear-reduction experiment in Newark (Pate et al. 1985b)

examined the effect of police-organizing efforts in the context of a program of multiple strategies in one neighborhood. The program in- cluded a community police storefront center, a door-to-door police

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

376 Lawrence Sherman

contact program, a neighborhood police newsletter, intensified law-en-

forcement and order-maintenance efforts, and neighborhood cleanups.

Given the multiple treatments, it is impossible to draw any conclusions

about the effect of any single element on the community problems of

crime and fear. But this highly intensive effort, according to the results

of before and after panel surveys, did produce these statistically significant findings in relation to the comparison area: a reduction in

fear of crime and dissatisfaction with the quality of life in the area at the

same time as an increase in the prevalence of households victimized by crime.

A similar problem of multiple simultaneous treatments limits our ability to draw conclusions about the effect of police organizing efforts

in the Hartford Asylum Hill project in the early 1970s (Murray 1983).

The program was an ambitious combination of redesigning the physical

environment and of building neighborhood organizations with the in-

volvement of a local neighborhood police team. The program was not focused on particular crime-prevention activities but rather on creating

an environment for neighborhood self-control. After three years, the

evaluation showed an increase in such control, including residents' ability to recognize strangers and willingness to take action in suspi- cious circumstances. But it also found that, despite a first-year decline in burglary and robbery, crime rose thereafter and returned to the levels predicted from citywide trends.

2. Police Communicating. The idea of informing the public about crimes has great appeal not only as news but also as practical informa-

tion that people can use to protect themselves. It is consistent with the

American commitment to an informed public. It is not consistent, however, with the modern information glut. Two of the fear-reduction

experiments, one each in Houston and Newark, probably encountered this problem in attempting to use a police newsletter to address crime

and fear problems. The newsletter in both cities was tested through a

randomized experimental design with three comparison groups: house-

holds mailed no newsletter, a newsletter with only general area news about police and crime, and the same newsletter with an insert giving detailed information about all recently reported crimes in the beat area.

The Newark and Houston newsletters differed only slightly in content, format, and frequency. Before-and-after personal interviews of house- hold members in the three conditions showed relatively low levels of recall of having seen or read the newsletter although very high levels of appreciation for the police effort in producing the newsletter.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 377

The evaluation (Pate et al. 1985c) showed no effect of either version

of the newsletter (with or without crime information) on a wide range

of fear- and crime-related behavior measures. It is possible that the experiment measured the wrong things and that, at least among those who read it, the information had some beneficial effects. But it is also

possible that such a communication strategy needs to be reconcep- tualized in a more tightly theoretical fashion, such as tying crime infor-

mation to crime analysis on which citizens can act. For example, such analysis might advise citizens to stay away from certain street corners

or parts of streets during certain hours.

Encouraging communication to flow the other way (from citizens to

police) is also an old idea, with a few modern twists. In Oakland, California, for example, the special central business district police proj-

ect distributed thousands of report forms on which citizens could write

up their description of a problem or an incident that made local quality

of life suffer, for example, a drunken panhandler or a prostitute accost-

ing them, witnessing an openly visible narcotics transaction, or teen-

agers running wild. Citizens were encouraged to give these completed report-it-directly forms to police or to security guards, who would pass

them on to police. In theory, police would then analyze the information

and use it to direct police attention to those problems, and that is how

the program worked initially. But, like many of the programs reviewed

here, it encountered difficulty in keeping going. For example, many of

the reports given to security guards were not passed on to the police until long after they were received, which made it unlikely that the forms had much more value than as a catharsis for citizen anger. Fewer

cards were turned in over time, although that may have been because

foot-patrol officers obtained much of the same information through word of mouth (Reiss 1985, pp. 22-23).

A more common means of fostering two-way communication flow is

the police storefront idea. Its starting point is that police precinct sta-

tions have become backstage areas for police use rather than an on-stage

scene for relaxed and cordial citizen-police communication. The store- front provides a setting that is free of the symbolic overtones of coer- cion and administrative pressures. Depending on their location, staffing, and hours of operation, they may be more or less successful in gathering information; depending on how the mission of the storefront

is conceived, they may be more or less successful in communicating crime-related information. Most of them appear to be largely reactive in this respect.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

378 Lawrence Sherman

The Houston and Newark storefronts that were created for the fear-

reduction experiments, however, took a proactive stance toward com-

munication. It is again hard to assess the effect of the Newark store- front in the context of the multiple programs in that area. But the Houston storefront stood alone, reaching out to tell the community about the storefront's activities: monthly neighborhood meetings, a truancy program, fingerprinting of children for parents' records, a

monthly blood-pressure screening test, a ride-along program, and a program for increasing use of and reducing vandalism at a local park.

The evaluation (Wycoff et al. 1985b) found that the area served by the

Houston storefront experienced significant reductions in fear of crime

and in perceptions of the level of crime against persons. The evaluation

did not detect any effect on the prevalence of household victimization.

3. Citizens Acting Alone. One of the first evaluations of a non-police-

coordinated community policing program was done in Seattle (Cirel et

al. 1977). The program consisted of governmentally funded project personnel helping residents do three things: mark their property through operation identification, improve the physical security of their

homes by implementing the recommendations of a security inspection,

and increase their surveillance of neighborhood activities. A before- and-after survey of this program over a one-year period suggested sub-

stantial reductions in burglary rates for citizens within experimental

areas who cooperated with the program as compared with citizens who

did not and marginally significant burglary reductions in the experi-

mental areas compared with the comparison areas. While it is not clear

whether such effects can be sustained over the long term, they are nonetheless impressive.

Such a program is not strictly citizens acting alone since they were

assisted by tax-supported professionals. Nor is the Chicago program (recently evaluated by Rosenbaum, Lewis, and Grant [1985]), which was funded by the Ford Foundation to continue efforts initiated under

federal grants and without police involvement. Yet the results of that

evaluation are important and disturbing. The evaluation found that the

program succeeded in implementing its strategy of organizing residents of selected neighborhoods through door-to-door canvassing, block meetings, and neighborhood meetings, all aimed at increasing citizen surveillance of the neighborhood. The experimental areas showed sub- stantial awareness of these efforts, but the efforts did not work. Instead

they may have backfired.

Using pretest/posttest interviews of over 3,000 respondents, the

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 379

evaluation found that the large majority of comparisons revealed no

significant differential change over time between the treated and the untreated areas. But in the three treatment areas with the strongest

program implementation, there were statistically significant increases in such measured problems as fear of crime, perceptions of the severity

of the crime problem, vicarious victimization, and concern about the future of the neighborhood. The authors conclude (p. iii) that "crime prevention meetings may have exacerbated preexisting concerns about

neighborhood decline. In any event, the interventions were unable to retard these negative processes." The findings show (p. iv) "a pressing need to rethink our expectations for these popular programs," just as

there is a need for rethinking community policing in general.

VII. Fitting Policing to Communities The best answer to the question of what works in policing different kinds of communities is that we do not know. Beyond the basic minimum of having tax-supported police, the optimal combination of

public and private and vocational and avocational structures, of their information systems, and of their strategies is unclear. Yet, for all that

we do not know, the research of the past fifteen years has put us light-

years ahead of our knowledge for most of the past two centuries. Fur-

ther research may contradict the current state of that knowledge, but at

this point these tentative conclusions about community policing in gen- eral seem to be warranted:

1. The allocation of patrol cars for rapid response should be systematically varied by neighborhood according to the expected rate of calls truly requiring such a response. 2. A system of differential police response could satisfy most citizens and free enormous amounts of police time for alternative strategies of crime control. 3. Saturation patrol and field interrogations may be useful in suppressing street crimes and disorder, especially in dense high-crime areas. 4. Done properly, proactive strategies need not abuse minority rights or constitutional due process nor hinder community relations. But the difficulties of implementing such strategies are substantial, and great care is required to succeed at implementation. 5. Police efforts to involve citizens in policing may have positive effects on community quality of life, which may have long-term

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

380 Lawrence Sherman

benefits for crime control, but they have shown little evidence of short-term crime-control benefits.

6. In general, policing will probably be more effective to the extent that it relies on a mix of strategies appropriate to each neighborhood rather than relying on the single complaint strategy as the primary approach for all neighborhoods.

As for the question of what works best in different kinds of com-

munities, the research suggests these meager policy considerations:

1. Police efforts to involve citizens in policing will be more successful in stable middle-class and lower-middle-class family neighborhoods than in transient, singles, apartment-building, or poverty-stricken neighborhoods. 2. In analyzing and planning community-level policing strategies, public police should consider the residence of police officers, the employment of private police, and the propensity of citizens to render voluntary policing services as factors enhancing the possibilities of strategies available. 3. Foot patrol is obviously more efficient in high-density, pedestrian communities than in low-density, automobile communities, but door-to-door policing visits in the latter type may produce the same effects as does foot patrol in the former. 4. Community police commanders should be given the authority to conduct informal experiments with different strategies to determine the optimal fit between the unique characteristics of the community and the mixture of policing strategies.

The last recommendation could be undertaken in a quasi- experimental, trial-and-error fashion. For example, Washington, D.C.,

police in Georgetown noticed in 1985 that teenagers were causing in- creased disorder problems on the streets and were drinking in their cars. They launched an order-maintenance crackdown of intense foot-

patrol and parking enforcement, which was followed by an immediate,

though not statistically significant, reduction in police-recorded crime (Sherman 1986). This kind of analysis is obviously limited for purposes of broad generalizations (Campbell and Ross 1968), but it is well suited for guiding specific policy decisions about specific communities.

The basic question underlying all these issues is whether the commu-

nity is the most appropriate unit of analysis for policing structures and strategy. While nostalgia for small-town community life may lead our

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 381

thinking in that direction, it is possible that policing may be equally or

more effective when focused on other units of analysis or on other targets for proactive efforts. Offenders, addresses, problems, relation-

ships, and social networks may be as important targets as are com- munities, especially with modern revolutions in life-style that make many residential neighborhoods largely empty for most of the business

day. The optimum balance between focusing on communities and on other targets should be recognized as an empirical question. But, what-

ever the focus, police effectiveness will probably be enhanced by pay- ing more attention to community context and how it interacts with police strategies.

REFERENCES

Bird, David, and David Dunlap. 1986. "Day by Day: Volunteers Again." New York Times (February 22).

Bittner, Egon. 1970. The Functions of the Police in Modern Society. Chevy Chase, Md.: National Institute of Mental Health.

Black, Donald. 1973. "The Mobilization of Law."Journal of Legal Studies 2:125- 49.

- . 1983. The Manners and Customs of the Police. New York: Academic Press. Black, Donald, and Albert J. Reiss, Jr. 1967. "Patterns of Behavior in Police

and Citizen Transactions." Studies of Crime and Law Enforcement in Major Metropolitan Areas, vol. 2, Field Surveys III, Sec. I. Washington, D.C.: President's Commission on Law Enforcement and Administration ofJustice.

Blumberg, Mark. 1983. "The Use of Firearms by Police Officers." Ph.D. dissertation, State University of New York at Albany.

Bowers, William. 1985. Presentation to the Fifth Annual Crime Control The-

ory Conference, National Institute of Justice, University of Maryland, Donaldson Brown Conference Center.

Boydstun, John E. 1975. San Diego Field Interrogation: Final Report. Washing- ton, D.C.: Police Foundation.

Boydstun, John E., and Michael Sherry. 1975. San Diego Community Profile: Final Report. Washington, D.C.: Police Foundation.

Bureau of Justice Statistics. 1986. Police Employment and Expenditure Trends. Washington, D.C.: U.S. Department of Justice.

Campbell, Donald T., and H. Laurence Ross. 1968. "The Connecticut Crack- down on Speeding: Time-Series Data in Quasi-experimental Analysis." Law and Society Review 3:33-53.

Chaiken, Jan. 1975. Patrol Allocation Methodology for Police Departments. Santa Monica, Calif.: Rand.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

382 Lawrence Sherman

Cirel, Paul, Patricia Evans, Daniel McGillis, and Debra Whitcomb. 1977. An

Exemplary Project: Community Crime Prevention Program, Seattle, Washington. Washington, D.C.: U.S. Government Printing Office.

"Court of Appeals Tells City to Pay $2 Million to Girl Father Stabbed." 1985. New York Times (July 10).

Croft, Elizabeth. 1985. "Police Use of Force: An Empirical Analysis." Ph.D. dissertation, State University of New York at Albany.

Cunningham, William C., and Todd H. Taylor. 1984. The Growing Role of Private Security. Washington, D.C.: U.S. Government Printing Office.

- . 1985. Private Security and Police in America. Portland, Oreg.: Chancel- lor.

Currie, Elliott. 1985. "Crimes of Violence and Public Policy: Changing Direc- tions." In American Violence and Public Policy, edited by Lynn A. Curtis. New Haven, Conn.: Yale University Press.

Curtis, Lynn A. 1985. "Neighborhood, Family and Employment: Toward a New Public Policy against Violence." In American Violence and Public Policy, edited by Lynn A. Curtis. New Haven, Conn.: Yale University Press.

Eck, John. 1986. "The Role and Management of Criminal Investigation in Neighborhood Policing." Paper presented at the Canadian Police College Conference on Community Policing in the 1980s, Ottawa, March 18.

Farrington, David P. 1982. "Randomized Experiments on Crime and Justice." In Crime and Justice: An Annual Review of Research, vol. 4, edited by Michael Tonry and Norval Morris. Chicago: University of Chicago Press.

Federal Bureau of Investigation. 1985. Crime in the United States-1984. Wash- ington, D.C.: U.S. Government Printing Office.

Fienberg, Stephen, Kinley Larntz, and Albert J. Reiss, Jr. 1976. "Redesigning the Kansas City Preventive Patrol Experiment." Evaluation 3:124-31.

Fogelson, Robert. 1977. Big City Police. Cambridge, Mass.: Harvard University Press.

Friedrich, Robert J. 1977. "The Impact of Organizational, Individual, and Situational Factors on Police Behavior." Ph.D. dissertation, University of Michigan.

Fyfe, James. 1978. "Shots Fired: An Examination of New York City Police Firearms Discharges." Ph.D. dissertation, State University of New York at Albany.

- . 1980. "Always Prepared: Police Off-Duty Guns." Annals of the Ameri- can Academy of Political and Social Sciences 452:72-81.

Gage, Theodore. 1982. "Cops, Inc." Reason (November) pp. 23-28. Gardiner, John A. 1969. Traffic and the Police. Cambridge, Mass.: Harvard

University Press. Goldstein, Herman. 1977. Policing a Free Society. Cambridge, Mass.: Ballin-

ger.

Johnson, Elaine. 1984. "Fast Food Chains Act to Hold Down Crime and Prevent Lawsuits." Wall Street Journal (November 8).

Kansas City Police Department. 1977. "Response Time Analysis." Unpub- lished manuscript, Kansas City, Mo., Police Department.

Kelling, George, Antony M. Pate, Duane Dieckman, and Charles Brown.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 383

1974a. The Kansas City Preventive Patrol Experiment: Summary Report. Wash- ington, D.C.: Police Foundation.

--. 1974b. The Kansas City Preventive Patrol Experiment: Technical Report. Washington, D.C.: Police Foundation.

Kleck, Gary. 1979. "Guns, Homicide, and Gun Control: Some Assumptions and Some Evidence." Paper presented at the annual meeting of the Midwest Sociological Society, Minneapolis.

Klockars, Carl. 1985. The Idea of Police. Beverly Hills, Calif.: Sage. Lane, Roger. 1980. "Urban Police and Crime in Nineteenth-Century

America." In Crime and Justice: An Annual Review ofResearch, vol. 2, edited by Norval Morris and Michael Tonry. Chicago: University of Chicago Press.

Larson, Richard C. 1972. Urban Police Patrol Analysis. Cambridge, Mass.: MIT Press.

-- . 1975. "What Happened to Patrol Operations in Kansas City?" Evalua- tion 3:117-23.

Lavrakas, Paul J. 1985. "Citizen Self-Help and Neighborhood Crime Preven- tion Policy." In American Violence and Public Policy, edited by Lynn A. Cur- tis. New Haven, Conn.: Yale University Press.

Levett, Alan. 1975. "Centralization of City Police in the Nineteenth Century United States." Ph.D. dissertation, University of Michigan.

Levine, Margaret J., and J. Thomas McEwen. 1985. Patrol Deployment. Wash- ington, D.C.: U.S. Government Printing Office.

Lindblom, Charles E., and David K. Cohen. 1979. Usable Knowledge: Social Science and Social Problem Solving. New Haven, Conn.: Yale University Press.

McEwen, J. Thomas, Edward F. Connors III, and Marcia I. Cohen. 1984. Evaluation of the Differential Police Response Field Test: Executive Summary. Alex- andria, Va.: Research Management Associates.

McNamara, Joseph. 1974. "Foreword." In Kelling, Pate, Dieckman, and Brown (1974a).

Margarita, Mona. 1980. "Criminal Violence against Police." Ph.D. disserta- tion, State University of New York at Albany.

Martin, Susan E., and Lawrence W. Sherman. 1986. "Selective Apprehension: A Police Strategy for Repeat Offenders." Criminology (in press).

Marx, Gary T., and Dane Archer. 1971. "Citizen Involvement in the Law Enforcement Process: The Case of Community Police Patrols." American Behavioral Scientist 15:52-72.

Matulia, Kenneth. 1982. A Balance of Forces. Gaithersburg, Md.: International Association of Chiefs of Police.

Milton, Catherine H., Jeanne W. Halleck, James Lardner, and Gary L. Abrecht. 1977. Police Use of Deadly Force. Washington, D.C.: Police Founda- tion.

Murphy, Patrick V., and Thomas Plate. 1977. Commissioner: A View from the Top of American Law Enforcement. New York: Simon & Schuster.

Murray, Charles A. 1983. "The Physical Environment and Community Con- trol of Crime." In Crime and Public Policy, edited by James Q. Wilson. San Francisco: Institute for Contemporary Studies Press.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

384 Lawrence Sherman

National Advisory Commission on Civil Disorders. 1968. Report. New York: Bantam.

"Officers Must Pay $2.3 Million to Wife Maimed by Husband." New York Times (June 26).

Pate, Antony M., Robert A. Bowers, and Ron A. Parks. 1976. Three Ap- proaches to Criminal Apprehension in Kansas City: An Evaluation Report. Wash- ington, D.C.: Police Foundation.

Pate, Antony M., Wesley Skogan, Mary Ann Wycoff, and Lawrence W. Sherman. 1985a. Reducing the Signs of Crime: Executive Summary. Washington, D.C.: Police Foundation.

-- . 1985b. Coordinated Community Policing: Executive Summary. Washing- ton, D.C.: Police Foundation.

- . 1985c. Neighborhood Police Newsletters: Executive Summary. Washington, D.C.: Police Foundation.

Pierce, Glen L., Susan A. Spaar, and LeBaron R. Briggs IV. 1984. "The Character of Police Work: Implications for the Delivery of Services." Mim- eographed. Boston: Northeastern University, Center for Applied Social Re- search.

Police Foundation. 1981. The Newark Foot Patrol Experiment. Washington, D.C.: Police Foundation.

President's Commission on Law Enforcement and Administration of Justice. 1967. Task Force Report: The Police. Washington, D.C.: U.S. Government Printing Office.

Press, S. J. 1971. Some Effects of an Increase in Police Manpower in the 20th Precinct of New York. New York: Rand.

Radzinowicz, Leon. 1956. A History of English Criminal Law and Its Administra- tion from 1750, vol. 2. London: Stevens.

Reiss, Albert J., Jr. 1971. The Police and the Public. New Haven, Conn.: Yale University Press.

. 1985. Policing a City's Central District: The Oakland Story. Washington, D.C.: U.S. Government Printing Office.

Reiss, Albert J., Jr., and David Bordua. 1967. "Environment and Organiza- tion: A Perspective on the Police." In The Police: Six Sociological Essays, edited by David Bordua. New York: Wiley.

Robin, Gerald D. 1963. "Justifiable Homicide by Police." Journal of Criminal Law, Criminology, and Police Science 54:225-31.

Rosenbaum, Dennis P., Dan A. Lewis, and James A. Grant. 1985. "The Impact of Community Crime Prevention Programs in Chicago: Can Neigh- borhood Organizations Make a Difference?" Mimeographed. Evanston, Ill.: Northwestern University, Center for Urban Affairs.

Rosenbaum, Dennis P., Arthur J. Lurigio, and Paul J. Lavrakas. 1985. Crime Stoppers: A National Evaluation of Program Operations and Effects. Draft final report. Washington, D.C.: U.S. Government Printing Office.

Rubinstein, Jonathan. 1973. City Police. New York: Farrar, Straus & Giroux. Ruth, Henry S. 1971. Criminal Justice Plan. New York: City of New York. Schnelle, John F., Robert E. Kirchner, Jr., Joe D. Casey, Paul H. Uselton, Jr.,

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

Policing Communities: What Works? 385

and M. Patrick Mcnees. 1977. "Patrol Evaluation Research: A Multiple- Baseline Analysis of Saturation Police Patrolling during Day and Night Hours." Journal of Applied Behavior Analysis 10: 3 3-40.

Schwartz, Alfred I., and Sumner N. Clarren. 1977. The Cincinnati Team Polic-

ing Experiment: Summary Report. Washington, D.C.: Police Foundation. Sherman, Lawrence W. 1975. "Middle Management and Police Democratiza-

tion: A Reply to John E. Angell." Criminology 12:363-77. - . 1983. "Patrol Strategies for Police." In Crime and Public Policy, edited

by James Q. Wilson. San Francisco: Institute for Contemporary Studies Press.

. 1985. "Repeat Complaint Analysis Policing (RECAP)." Proposal submitted to the National Institute of Justice. Washington, D.C.: Crime Control Institute.

-- . 1986. "Uncertain Punishment: Crackdown and Backoff in Crime Con-

trol." Report submitted to the National Institute of Justice by the Center for Crime Control, University of Maryland Research Foundation.

Sherman, Lawrence, and Ellen G. Cohn. 1986a. "Citizens Killed by Big-City Police, 1970-84." Crime Control Reports, no. 2. Washington, D.C.: Crime Control Institute.

.. 1986b. "Police Policy on Domestic Violence: A National Survey." Crime Control Reports, no. 1. Washington, D.C.: Crime Control Institute.

Sherman, Lawrence, and Jody Klein. 1984. "Major Lawsuits over Crime and Security: Trends and Patterns, 1958-82." Mimeographed. College Park: University of Maryland, Institute of Criminal Justice and Criminology.

Sherman, Lawrence W., Catherine H. Milton, and Thomas Kelly. 1973. Team Policing: Seven Case Studies. Washington, D.C.: Police Foundation.

Skogan, Wesley, and Maxfield, Michael. 1981. Coping with Crime. Beverly Hills, Calif.: Sage.

Smith, Betty. 1943. A Tree Grows in Brooklyn. New York: Harper. Smith, Douglas. 1984. "The Organizational Context of Legal Control."

Criminology 22:19-38. --. In this volume. "The Neighborhood Context of Police Behavior." Spelman, William, and Dale K. Brown. 1981. Calling the Police: Citizen Report-

ing of Serious Crime. Washington, D.C.: Police Executive Research Forum. Swanson, Cheryl. 1978. "The Influence of Organization and Environment on

Arrest Policies in Major U.S. Cities." Policy Studies Journal 7:390-98. Tolchin, Martin, 1985a. "Private Guards Get New Role in Public Law En-

forcement." New York Times (November 29).

.. 1985b. "Off-Duty Officers Doubling as Private Guards, but System Draws Criticism." New York Times (December 1).

Walker, Samuel. 1977. A Critical History of Police Reform. Lexington, Mass.: Heath.

Westley, William. 1970. Violence and the Police. Cambridge, Mass.: MIT Press.

Wilson, James Q. 1968. Varieties of Police Behavior. Cambridge, Mass.: Harvard University Press.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms

386 Lawrence Sherman

. 1983. Thinking about Crime. New York: Basic. Wilson, James Q., and Barbara Boland. 1978. "The Effect of the Police on

Crime." Law and Society Review 12:367-90. Wilson, James Q., and George L. Kelling. 1982. "Broken Windows: The

Police and Neighborhood Safety." Atlantic (March), pp. 29-38. Wilson, Orlando W. 1963. Police Administration. New York: McGraw-Hill. Wright, James D., Peter H. Rossi, and Kathleen Daly. 1983. Under the Gun:

Weapons, Crime and Violence in American Life. New York: Aldine. Wycoff, Mary Ann, Wesley Skogan, Antony Pate, and Lawrence W. Sher-

man. 1985a. Citizen Contact Patrol: Executive Summary. Washington, D.C.: Police Foundation.

. 1985b. Police Community Stations: Executive Summary. Washington, D.C.: Police Foundation.

. 1985c. Police as Community Organizers: Executive Summary. Washington, D.C.: Police Foundation.

Wycoff, Mary Ann, Charles Susmilch, and Charles Brown. 1981. The Birming- ham Anti-Robbery Experiment. Draft report. Washington, D.C.: Police Foun- dation.

This content downloaded from 129.252.86.83 on Sun, 26 Mar 2017 21:27:38 UTC All use subject to http://about.jstor.org/terms